Oklahoma Behavioral Workforce Study Statewide 2-16-11 |
Previous | 1 of 4 | Next |
|
small (250x250 max)
medium (500x500 max)
Large
Extra Large
large ( > 500x500)
Full Resolution
|
This page
All
|
report oklahoma statewide workforce behavioral health 1 OKLAHOMA BEHAVIORAL HEALTHCARE WORKFORCE STUDY: DRAFT STATEWIDE REPORT • Separation Rates & Staff Intention to Leave • Vacancy Rates • Organizational Benefits & Staff Pay Rates • Staff Work Experience & Job Satisfaction Oklahoma Behavioral Healthcare Workforce Study: Statewide Report February 14, 2011 This research was supported by the Mental Health Transformation State Incentive Grant, 5U79SM057411 from the U.S Department of Health and Human Services, Substance Abuse and Mental Health Services Administration. The views expressed do not necessarily reflect the official policies of the Department of Health and Human Services; nor does mention of trade names, commercial practices or organizations imply endorsement by the U.S. Government. Contributors include: John Hornik, Ph.D., Jen Carpenter, Ph.D., Jeanine Hanna, Ph.D., and Nicholas Huntington, M.A., Advocates for Human Potential, Albany, NY and Karen Frensley, LMFT, David Wright, Ph.D. & Lorrie Byrum, M.A., Oklahoma Dept. of Mental Health and Substance Abuse Services, Oklahoma City, OK. 2 Oklahoma Behavioral Healthcare Workforce Study Statewide Report TABLE OF CONTENTS EXECUTIVE SUMMARY 3 INTRODUCTION 11 STAFF SEPARATIONS 16 VACANCIES AND STAFF RECRUITMENT BARRIERS 27 CURRENT AND FUTURE STAFFING NEEDS 34 BENEFITS & COMPENSATION 47 STAFF WORK EXPERIENCE AND JOB SATISFACTION 57 WORKFORCE CAPACITY 66 REPRESENTATION OF CONSUMERS AND THEIR FAMILY MEMBERS IN THE WORKFORCE 78 DISCUSSION AND RECOMMENDATIONS 85 REFERENCES 89 APPENDIX A: STAFF SEPARATIONS 90 APPENDIX B: VACANCIES AND RECRUITMENT BARRIERS 103 APPENDIX C: BENEFITS AND COMPENSATION 105 APPENDIX D: WORK EXPERIENCE AND JOB SATISFACTION 107 APPENDIX E: WORKFORCE CAPACITY 110 3 EXECUTIVE SUMMARY The Oklahoma Behavioral Healthcare Workforce Survey and associated studies were conducted by Advocates for Human Potential, Inc. (AHP) through a contract with the Oklahoma Department of Mental Health and Substance Abuse Services (ODMHSAS) to assist with evaluation activities related to Oklahoma’s behavioral health transformation initiative. The studies were developed and implemented under the guidance of an advisory group, the Workforce Study Team, which was convened through the Governor’s Transformation Advisory Board (GTAB) Workforce Committee, as part of the Transformation initiative. The primary goals of the studies were to: 1. Respond to interests of GTAB Workforce Committee convened through Oklahoma’s behavioral health transformation initiative. 2. Develop behavioral health complement to information gathered through Oklahoma Health Care Workforce Center and Oklahoma Hospital Association surveys. 3. Provide information that can be used for provider organization and state agency-level planning and advocacy. The largest of these studies was the Oklahoma Behavioral Healthcare Workforce Survey, a statewide survey that focused on staffing of agencies and programs that provide behavioral healthcare. The survey was designed with three components: an organizational survey focusing primarily on organizational accreditation and benefits as well as basic information on organizational structure; a program manager survey containing items related to program staffing, vacancy, recruitment barriers, causes of staff turnover, program and staff capacity and training needs; and a staff survey focusing on staff work experience, job satisfaction, education and training as well as demographic characteristics and status as current or prior consumers or family members of consumers. Data collection and process was structured so that the three components could be linked, and organizations were recruited in industry groups, generally according to state agency funding and oversight. While the workforce survey is the largest component of this project and is generally the focus of this report, additional resources used include: Economic Modeling Systems Inc (EMSI) data provided by the Oklahoma Department of Commerce, data drawn from a University of North Carolina (UNC) staffing needs study, and information on historical and anticipated behavioral healthcare-related degree completion rates from the Oklahoma State Regents of Higher Education. Taken together, these resources and the workforce survey are used to address the following topic areas: Staff Separations Information related to separations was gathered through program manager reports of the perceived causes of separation in their programs, program managers’ reports of their programs’ separation rate over the previous year, and staff reports of their intention to leave their position within the next year. Consistent with the findings related to recruitment barriers, the most frequently cited barrier was dissatisfaction with pay, which was cited by nearly two thirds of program managers. Excessive paperwork, emotional burnout and excessive on-the-job stress were cited by at least one third of program managers. While program and organization 4 characteristics were related to multiple perceived causes of turnover when the relationships were examined individually, generally only one or two characteristics remained significant in each logistic regression model. Organizational industry was a significant predictor of citing dissatisfaction of pay, with OPHA program managers being the least likely to cite pay as a cause of turnover. Population age and program setting were significant predictors of perceiving paperwork to be a cause of turnover, with program managers in programs serving children citing paperwork more frequently than those serving both children and adults, and program managers in outpatient settings citing paperwork more frequently than program managers in other settings. Service population also related to citation of dissatisfaction with job responsibilities, with program managers from programs serving both children and adults being less likely to cite this as a barrier than program managers from programs serving either adults or children. Program separation rates ranged from 0% to 200%, and the median of 25% was used to divide programs into two categories: low separation and high separation. These categories were related to multiple program and organizational characteristics when the relationships were examined individually, but only two characteristics remained significant in the logistic regression model. High separation programs proved to be more likely to have a high proportion of techs on staff, and less likely to be state operated. The position type results are consistent with existing literature regarding the relationship between lower staff experience/job level and higher separation rates. The vast majority (80%) of staff did not report intending to leave their positions within the 12 month period following the survey. Intention to leave was related to a range of program, organizational and staff characteristics when the relationships were examined individually, but only two remained significant in the logistic regression model. As would be expected, staff intending to leave reported lower satisfaction with their job overall. Staff age was also related to intention to leave, with the mean age for staff intending to leave being about three and a half years younger than that of staff intending to stay. Both of these findings are consistent with the literature on staff intention to leave. Vacancies and Staff Recruitment Barriers Information related to vacancies was gathered through program managers’ reports of the perceived recruitment barriers in their programs, and their reports of their programs’ current vacancies. By far the most frequently cited barrier was salary, which was cited by 57% of program managers. Lack of candidates with desired credentials or desired work experience, small applicant pool due to geographic location, and competition from other fields were all cited by more than one quarter of program managers. Program and organization characteristics that were related to multiple perceived barriers included organizational industry, state operation, organizational size, and geographic region. Salary as a perceived barrier was related to three of these characteristics when the relationships were tested individually. When examined simultaneously, salary remained significantly related to industry, with OPHA program managers being significantly less likely to cite salary as a barrier. Likewise, state operation and salary were related, with program managers in state operated organizations more likely to cite salary as a barrier. Program vacancy rates ranged from 0% to 100%, and the median of 4% was used to divide programs into two categories: low vacancy (less than 4%) and high vacancy (greater than 4%). These categories proved to be unrelated to most of the program and organizational characteristic variables. Staffing patterns offered one exception: The mean proportion of RNs in 5 low vacancy programs was slightly but significantly lower than the mean proportion of RNs in high vacancy programs, which could be in part related to the comparatively high rate of vacancies in RN positions, across programs. Current and Future Staffing Needs The purpose of this chapter is to identify unmet needs for the behavioral healthcare workforce with a focus on type of position. The first section focuses on psychiatrists and other prescribers. The second section focuses on other professional and non-professional staff. Each of these sections employs data from different sources so the methods upon which we have relied are described within each section, as well as the implications for higher education. The third section describes one underlying problem, the level of compensation currently available to the Oklahoma workforce. Among the studies that we identified was a study of the relative unmet need for professional mental health workers in the State of Washington (Morrissey, et al, 2007a), undertaken as a part of their Mental Health Transformation State Incentive Grant. The study of Washington State was a part of a larger, national study sponsored by the Health Resources Services Administration (HRSA) of the U.S. Department of Health and Human Services. This allowed Morrissey and his colleagues to develop estimates of professional shortages for every county in the U.S. These estimates for Oklahoma demonstrate an unequivocal need for more prescribing professionals in all areas of the state. The area of the State with the greatest unmet need is the Northeast quadrant, excluding Tulsa which has the smallest, relative unmet need. To examine need for other professional staff, we drew on data supplied by the Oklahoma Department of Commerce, using Economic Modeling Systems, Inc (EMSI). EMSI uses several different databases, including population projections from the census bureau, employment trends and participation rates from the Bureau of Labor Statistics (BLS), IRS income and migration data, and industry trends, legislation, and several other factors used to decipher which industries will be growing. Examining EMSI data alongside current vacancy rates and degree attainment trends, revealed that there are unmet staffing needs among both non-prescribing professionals and nonprofessionals as well, and the rates at which institutions of higher education in Oklahoma are producing new graduates with appropriate training are not sufficient to meet these needs, particularly with projected future growth of these positions. EMSI data were also used to examine salary rates across Oklahoma, and in comparison to regional and national rates. It is clear that salary rates for all positions are lower in Oklahoma than in the nation and further that Oklahomans filling these positions providing behavioral healthcare are paid less than individuals in all of the surrounding states. There is also some variation within the State. For the two position types that have the largest numbers of persons providing behavioral healthcare, MH/SA Counselors and MH/SA Techs, salaries are higher in the Oklahoma City and Tulsa areas than they are in the more rural northeast, northwest, southeast, and southwest quadrants of the state. Benefits and Compensation Information on benefits and compensation was collected through the organizational survey and the staff survey. Nearly all privately-operated organizations report providing health insurance, but the provision rate for other benefits deviates from the benefit packages provided by state- 6 operated organizations. Staff report high rates of satisfaction with paid leave, but more moderate rates of satisfaction with other benefits. Staff satisfaction with benefits varies by proportion of health insurance covered and by industry group, with industry groups composed primarily or exclusively of state-operated organizations showing higher rates of staff satisfaction with benefits. Staff reported a wide range of pay rates, but over half the responses were clustered in the lower two pay categories (less than $10.00 per hour and $10.00 - $14.99 per hour), with nearly one in five staff reporting pay of less than $10.00 per hour. Staff earning towards the upper end of the range are at roughly 185% of the 2009/2010 poverty guidelines if they have no dependents, but are under the poverty line if they have more than two dependents. Staff earning closer to minimum wage are at roughly 133% of the 2009/2010 poverty guidelines if they have no dependents, but are under the poverty line if they have any dependents (Office of the Assistant Secretary for Planning and Evaluation, 2010). Position type is strongly tied to pay rate, with techs earning an average of $11.23, less than half the average hourly wage of psychologists ($28.33) and RN’s ($26.71). While staff pay is related to a number of program and staff variables when these relationships are examined individually, only four remained significant when tested simultaneously: position type, program service type, consumer population age, and organization size. The relationship of these last three variables to pay is suspected to be caused in part by other variables, including position type. Given the key role that position type plays in staff pay rates, the remaining staff variables were tested as predictors of position type. Staff race, gender, age and highest degree obtained all predict position type, which in turn predicts staff pay. Staff Work Experience and Job Satisfaction Information on staff satisfaction and work experience was collected through two separate sets of questions in the staff survey. Most of the staff work experience items elicited positive responses from the majority of participants, with nearly all (95%) staff agreeing with the statement - I like the kind of work I do. A singe item - I recommend my organization as a good place to work - was used as an indicator of overall work experience for analysis with other variables. Of the staff and program variables considered, two proved to be significant predictors of work: Work experience was related to industry group, with the highest proportions of staff agreeing with the indicator item being those associated with the Child Guidance (89%) and Substance Abuse (86%) industries. Additionally, staff from programs serving adults were significantly less likely to endorse the indicator item than were staff in programs serving both adults and children (70% versus 80%). Staff satisfaction was measured through a separate set of items. Many of these items also received largely positive responses, with 84% of staff indicating that they were satisfied with their jobs overall, and more than 70% expressing satisfaction with their work schedules, the location of their workplaces, and their organizations overall. The lowest rates of satisfaction were related to the opportunity for advancement (41%) and pay (47%). Responses to these and other items suggest that program manager perceptions of the causes of turnover may be well founded, to the degree that staff satisfaction relates to turnover. 7 Given the importance of pay in both staff satisfaction and program manager perceptions of turnover and recruitment barriers, we examined the relationship of this item to a range of program and staff variables. Industry, service population, service type and years working in the field predicted satisfaction with pay. Staff in industries with a high proportion of state-operated organizations and with a high proportion of Masters-level staff (Child Guidance and DOC) expressed greater satisfaction with their pay, as did staff in programs serving both adults and children (as opposed to just adults, or just children), staff in programs providing substance abuse services only, and staff who reported greater tenure in the behavioral healthcare field. The job satisfaction items were then used to create a scale representing the proportion of job characteristics found satisfactory. Service type and consumer population age also proved to be a significant predictor of this score, with staff in substance abuse programs reporting satisfaction with a greater proportion of job characteristics than staff in mental health programs (74% versus 58%), and staff in programs serving both children and adults reporting satisfaction with more aspects of their jobs than did staff in programs serving only adults (66% versus 60%). Additionally, staff in both small and medium-sized organizations reported satisfaction with more aspects of their jobs than did staff from large organizations, a finding that may be related to the distribution of industry groups across organizational size. The regression also pointed to the significance of two separate demographic characteristics - Black/African American race and high school education - in predicting satisfaction with a greater proportion of job characteristics. It is important to note that these findings are not echoed in the work experience analysis, and these characteristic did not predict higher satisfaction with most of the more global scale items (organization overall, pay, and job overall). Workforce Capacity Information on workforce capacity and training needs was collected through the program manager and staff surveys. According to program manager reports, the three types of training most needed by staff are: (1) knowing about consumers’ psychiatric medications and their side effects, (2) communication skills and (3) educating consumers’ family members about subjects related to mental health or substance abuse. Bivariate analyses demonstrate that program managers from the Substance Abuse and DHS industry groups are most likely to report staff needing training related to consumers’ psychiatric medications, while 30% and 36% of staff from the OJA and OPHA industries require additional training on the role of peers as service providers. Some specific training needs also varied by program service type. In addition to basic training it is important that new professional staff have the capacity to provide evidence-based practices for adults and children. Over 65% of new professional hires are prepared to provide Cognitive Behavioral Therapy (CBT) for adults and CBT for depression, anxiety, and trauma for children. Since education about psychiatric medications was identified as one of the types of training most needed for direct care staff, it is not surprising that only 37% of new professional hires can provide the EBP, Medication Management. Staff capacity to provide the EBP, consumer-run services, was also low (35%). Data on organizational/program cultural competency were gathered based on staff perceptions of whether (1) their workplace has an attitude of acceptance of people from different cultural backgrounds; (2) their organization does a good job recruiting and retaining employees of different cultures; (3) sensitivity to diversity is an important part of supervision/team meetings; 8 (4) staff are encouraged to attend diversity training; and (5) cultural assessment is used to plan effective treatment and service delivery. The majority of staff surveyed agreed or strongly agreed with all five items. Bivariate analyses explored the relationship between the aforementioned cultural competency items and several program, organization, and staff variables. Staff perceptions of how well their organization recruits and retains employees of various cultures varied by industry, region, and staff ethnicity and highest degree earned; these relationships were significant in both the bivariate and regression analyses, when we controlled for other variables. Interestingly, staff identifying as Hispanic/Latino were more likely to report their organization recruits and retains employees of diverse cultures than non-Hispanic/Latinos. This finding may be related to another finding: staff working in the Oklahoma City metro area are most likely to agree with the aforementioned cultural competency item. This region of the state had the highest response rate and the greatest racial/ethnic diversity among staff working there. Program managers were asked to report the cultural and linguistic capacity of their programs (i.e., does it hold cultural competence training and/or provide services in Spanish or American Sign Language). Comparisons were made between program manager reports of program linguistic capacity and the self-reported fluency of staff and program managers. The vast majority of programs (78%) hold some type of cultural competency training for staff. Although 22% of program managers report that their program can provide services in Spanish, only about 3% of staff and managers reported that they are fluent in Spanish, which is less than 5%, the state average. Different interpretations of what it means to “provide services in Spanish” may account for some of the discrepancy in self-reported (staff and program managers) and program linguistic capacity. Representation of Consumers and Their Family Members in the Workforce Information on consumer and family member representation and disclosure was obtained through the staff and program manager surveys. The most important finding is that a significant proportion of the behavioral healthcare workforce that identifies themselves as adult consumers (21%) and an even larger proportion that identify themselves as family members of consumers (32%). Consumer and family member representation was generally higher among program managers than staff, and was higher for adult consumer and family members of an adult consumer categories than for former youth consumer and family of a youth consumer categories. With the exception of the youth consumer category, representation of all categories exceeded 10% for both program managers and staff, and reached a high of 37% representation of family members of adult consumers among program managers. Both representation and disclosure varied significantly by industry group. Adult consumer and family member representation was highest in the Substance Abuse and DOC industry groups, and lower in the OPHA, OJA, and Child Guidance industry groups, although Child Guidance had the greatest proportion of staff who identified as family members of youth consumers. Over three-quarters of Substance Abuse staff who identified as consumers report having disclosed this status in the workplace, compared to just over half of OPHA and DOC staff members who identified as consumers. Among staff who identified as family members, nearly three-quarters disclosed this status, while just over half of OPHA staff disclosed. 9 The analysis considered a wide range of possible predictors of both consumer status and family member status among responding staff. While many of these were initially found to be significantly related to one or both outcome variables, few remained significant when logistic regression models were used to test the relationships simultaneously. Staff working in programs serving people with substance abuse or substance abuse and mental health needs were significantly more likely to identify as consumers than were staff working in programs serving people dually diagnosed with developmental disabilities and behavioral health needs. Also, those working in outpatient programs were significantly more likely to identify as consumers than were those working in inpatient programs. Respondent education level was the only variable remaining significant in the family member representation model, with staff who reported having a Masters degree or higher being significantly more likely to identify as family members than were staff with high school diplomas or GEDs. Among staff and program managers who identify as either consumers or family members, rates of disclosure in the workplace are high. A higher proportion of program managers reported disclosing their status. For both consumer and family member status, roughly 80% of responding program managers report disclosing on the job, while roughly 66% of staff report having disclosed. The analysis also considered multiple potential predictors of staff disclosure of consumer or family member status. As with the previous analysis, many of these were related to consumer or family status in initial analysis, but did not remain related in the subsequent logistic regression models. Respondent race and type of service used proved to be significantly related to disclosure of consumer status, with White staff more likely to have disclosed than Black staff, and with staff who reported receiving both mental health and substance abuse services more likely to disclose than staff receiving either mental health or substance abuse services. It is interesting to note that while there is no significant relationship between staff member consumer status and race, among those who do identify as consumers and family members, White staff members are more likely to disclose this status in the workplace than are Black staff members. A similar pattern was noted for disclosure of family member status. Program industry group was also found to be a significant predictor of disclosure of family member status, with respondents working in the Mental Health and Substance Abuse programs significantly more likely to have disclosed their status on the job than were respondents from the OPHA programs. Discussion and Recommendations Following review of an earlier draft of this report, Workforce Study Team members were asked to make recommendations in response to the study findings. These recommendations were grouped into five topic areas: compensation, recruitment and retention, training, best practices, and future planning efforts. Regarding compensation, the Workforce Study Team recommended the prioritization of overall funding for behavioral healthcare services, pointing to the clear need for better compensation. The Team advised that current pay rates are inadequate, and that it is important for the public to become more aware of this inadequacy. Relating to recruitment and retention, the Workforce Study Team found that the report provided evidence that there is dissatisfaction with opportunities for advancement within the behavioral 10 healthcare workforce, with only 41% of staff reporting satisfaction with their opportunity to advance within their organization. The Team advised that this suggests a need for more opportunitites within agencies for positions for possible advancement, and a need to eliminate the barriers that currently make advancement difficult. Additionally, given existing reimbursement strategies, the Team noted a number of challenges in supporting staff working on achieving licensure. The Workforce Study Team’s concerns about training included the insufficient number of prescribers in the state; the need to support the development of basic behavioral health care screening, assessment, treatment, and referral skills among primary medical care providers; and the insufficient “real world” training opportunities for some professions, particularly psychologists who may be trained in settings vastly different from the public behavioral healthcare system. Resources for supporting implementation of Evidence Based Practices (EBPs), requires additional funds to train and assure model fidelity through consultation and supervision. The initial cost of training and consultation for clinicians to treat people with practices that work, should be recouped in the long run since there will be cost savings when individuals recover and no longer need services. The Workforce Study Team identified the implementation of best practices as one way to respond to the study findings related to staff paperwork burden and its relation to job satisfaction and to program manager perceptions of causes of turnover, and pointed to the difficulty in reducing documentation burden given high levels of vacancy and turnover. Additionally, the Team raised telehealth as an important best practice for implementation in Oklahoma to increase access. Finally, with respect to future efforts, Workforce Study Team members identified a need to retain the involvement and commitment of well-positioned personnel in key state agencies and within the private sector, and pointed to the importance of focusing continued work on a vision for the future of behavioral healthcare in the state. The Team recommended the formation of an advisory council to continue in-depth analysis of the state’s workforce issues. 11 INTRODUCTION In 2005, Oklahoma was one of seven states (now nine) to receive a five-year Mental Health Transformation State Incentive Grant (TSIG) from the federal Center for Mental Health Services (CMHS). The purpose of this grant was to help transform state mental health systems from “broken and fragmented” systems to systems that deliver excellent mental health care with a focus on recovery (President’s New Freedom Commission on Mental Health, 2003). A major challenge faced by all states was assuring a stable, competent workforce available to provide needed services. The Oklahoma Behavioral Healthcare Workforce Survey and associated studies were conducted by Advocates for Human Potential, Inc. (AHP) through a contract with the Oklahoma Department of Mental Health and Substance Abuse Services (ODMHSAS) to assist with evaluation activities related to Oklahoma’s behavioral health transformation initiative. The studies were developed and implemented under the guidance of an advisory group, the Workforce Study Team, which was convened through the Governor’s Transformation Advisory Board (GTAB) Workforce Committee, as part of the Transformation initiative. Purpose and Goals State mental health authorities typically do not have empirical information about the characteristics of their current workforce. In order to fill this information gap, we undertook a number of studies, as well as searched for relevant research, that provided useful information for understanding the difficulties faced by staff providing mental health services in Oklahoma. Taken together, the workforce studies were designed with three broad goals in mind: 1. Respond to interests of GTAB Workforce Committee convened through Oklahoma’s behavioral health transformation initiative. 2. Develop behavioral health complement to information gathered through Oklahoma Health Care Workforce Center and Oklahoma Hospital Association surveys. 3. Provide information that can be used for provider organization and state agency-level planning and advocacy. The largest of these studies was the Oklahoma Behavioral Healthcare Workforce Survey, a statewide survey that focused on staffing of agencies and programs that provide behavioral healthcare. The survey itself was intended to address six particular goals of the Workforce Study Team and other project stakeholders, including: 1. Estimate rates of recruitment, retention and turnover by position. 2. Determine reasons for leaving, including those related to wages and benefits (e.g., health insurance, schedule/shift, child care). 3. Analyze current representation of adult peers and family members in the workforce. 4. Describe linguistic (and cultural) competency of the workforce. 12 5. Describe capacity of state workforce to address current needs of clients and employers. 6. Describe current access to behavioral healthcare services in primary care settings and identify (types of) professionals delivering such services. Methodology Survey Measures Where possible, survey items were drawn from established measures. The two primary sources of items and item structure were: • Addition Technology Transfer Center Workforce Survey: A staff and director survey instrument was developed for the Northwest Addiction Technology Transfer Center (ATTC) and subsequently adapted for use in at least six other states. Oklahoma workforce survey items that were drawn from or based on this instrument included those relating to recruitment barriers and causes of turnover, organizational strategies for supporting staff development, and distribution of daily responsibility, as well as a number of basic demographic related items. • Federal Human Capital Survey (FHCS): The FHCS is an instrument developed by the U.S. Office of Personnel Management and used to measure employees’ job satisfaction and their perceptions of the degree to which their organization exhibits characteristics consistent with those of successful organizations. The instrument was used to survey federal employees in 2004, 2006, and 2008, with over 200,000 responses received in the 2008 use alone (United States Office of Personnel Management, n.d.). Oklahoma workforce survey items that were drawn from the FHCS include those related to staff work experience and job satisfaction. Additional items were developed and selected with the guidance of the Workforce Study Team and outside consultation when necessary. Pilot Study The pilot study involved two organizations: a residential care provider which operates congregate care facilities in locations throughout Oklahoma, and an inpatient care provider which operates a variety of behavioral healthcare programs in the Oklahoma City area. Between the two organizations, a total of 28 distinct programs participated in the pilot. These programs provided an array of services designed to respond to a variety of consumer needs and interests. Programs ranged from long-term residential care to acute detoxification, and served children, youth, adults and older adults, and supported people with needs related to mental health, substance abuse and co-occurring disorders. The pilot study took place in June and July, 2008. In August 2008, the preliminary results of the pilot were reviewed with the Workforce Study Team, as was a report of the survey process, including challenges encountered and suggestions offered by pilot participants. Based on these reports and the discussion with the Workforce Study Team, some redundant items were eliminated, the schedule and scope of organizational recruitment was scaled back, and the recruitment material packet was revised. 13 Survey Structure In order to capture the range of information desired by the Workforce Study Team and other project stakeholders, the survey was designed with three components: 1. An organizational survey focusing primarily on organizational accreditation and benefits as well as basic information on organizational structure. Organizational structure information was used to create organization-specific versions of the program manager and staff surveys described below. The organizational survey component was completed by a single member of each participating organization (typically a human resources administrator in larger organizations, or the director in smaller organizations). 2. A program manager survey containing items related to program staffing, vacancy, recruitment barriers, causes of staff turnover, program and staff capacity and training needs. Within each organization, each program manager with unique supervisory responsibilities for one or more behavioral healthcare programs was invited to complete a program manager survey. Occasionally, organizations would indicate that two or more program managers supervised a single program. In these cases, AHP worked with the organization to develop a survey plan to avoid duplication of program manager responses. 3. A staff survey focusing on staff work experience, job satisfaction, education and training as well as demographic characteristics and status as current or prior consumers or family members of consumers. All direct providers of behavioral healthcare services in participating organizations were invited to complete a staff survey. As described in the recruitment subsection below, however, staff recruitment was highly dependent on program manager assistance. Data collection and process was structured so that the three components could be linked. Staff responses could be grouped by program and organization, and linked to the appropriate program data (provided via the program manager survey) and organizational data (provided via the organizational survey). 14 Exhibit 1.1: Survey Structure ORGANIZATIONAL SURVEY PROGRAM MANAGER SURVEY PROGRAM MANAGER SURVEY PROGRAM MANAGER SURVEY STAFF SURVEY STAFF SURVEY STAFF SURVEY Recruitment & Participation Organizations were recruited in industry groups, generally according to state agency funding and oversight. The following nine industry groups were recruited: • Mental Health: Organizations providing primarily mental health services and operated under contract with or by ODMHSAS. • Oklahoma Psychiatric Hospital Association (OPHA): Psychiatric hospitals or hospitals with psychiatric units within OPHA membership. • Oklahoma Department of Human Services (OKDHS): Organizations providing a range of residential and outpatient services for children, youth and adults with a variety of service needs and operated by or under contract with OKDHS. • Oklahoma Office of Juvenile Affairs (OJA): Organizations operated by or under contract with OJA, providing services to children and youth in a range of settings. • Substance Abuse: Organizations providing primarily substance abuse services and operated under contract with or by ODMHSAS. • Oklahoma Department of Corrections (DOC): Providers employed by DOC and offering mental health services within correctional facilities across Oklahoma (substance abuse services are contracted out and were therefore not included in the survey). • Other Medicaid: A random sample of organizations that were not included in any of the above groups but that do provide behavioral healthcare services and bill Medicaid. • Federally Qualified Health Centers (FQHC): Organizations that provide behavioral healthcare services and have obtained the FQHC designation. 15 • Child Guidance: Child Guidance clinics operated by the Oklahoma State Department of Health (OSDH). The number of organizations, program managers, and staff members recruited by industry group are shown in Exhibit 1.2 on the next page. When considering the implications of the results described in this section, it may be helpful to bear in mind the degree to which the responses we received can be considered representative of the views of Oklahoma behavioral healthcare agencies, program managers, and staff. Exhibit 1.2 indicates that 63% of invited organizations responded, with participation rates by industry group ranging from 41% to 100%. We can be relatively confident that responses from agencies in high participation industry groups are representative of those industry groups, but less confident of the representativeness of responses of agencies in low participation industry groups. Similarly, among participating organizations, average program manager response rates ranged from 67% to 100%, with an overall average of 72%. Among participating programs, staff response rates ranged from 4% to 100%, with an overall average of 26%. Our confidence in program manager and staff response representativeness should also vary by industry group participation rate. Additionally, within industry groups or within the sample as a whole, we can have more confidence in the representativeness of program manager responses than we can in the representativeness of staff responses. Finally, it is important to note that, as the recruitment process was driven by state agency oversight and funding, any First Nations provider organizations that are not funded or credentialed by one or more of the above state agencies were not recruited. 16 Exhibit 1.2: Participation by Industry Group Industry Wave Date Launched Organizations Program Managers Direct Care Staff Number of Responses Response Rate Number of Responses Response Rate Number of Responses Response Rate1 Mental Health 9/30/08 27 79% 102 67% 443 21% OK Psychiatric Hospitals Association 11/04/08 12 41% 32 74% 363 26% OK Department of Human Services 1/14/09 10 83% 20 74% 150 31% OK Office of Juvenile Affairs 1/14/09 11 79% 12 86% 38 13% Substance Abuse 5/14/09 38 62% 52 74% 234 36% Department of Corrections 8/17/09 12 100% 6 100% 40 63% Other Medicaid Providers 8/19/09 11 48% 9 82% 6 4% Federally Qualified Health Centers 8/19/09 5 45% 2 67% 14 100% Child Guidance Clinics 10/26/09 12 100% 8 89% 37 73% Total: 116 63% 243 72% 1325 26% 1 Staff participation rates are based on programs for which total number of staff is known. 2 The Department of Corrections and Child Guidance Clinics are multiple service sites however due to the nature of the programs they were surveyed as one organization. 17 At the beginning of the recruitment phase for each industry, enrollment packets were mailed to the organizations that had been identified for recruitment. These packets included a cover letter from the relevant state agency administrator, describing the value of the project and encouraging the organization to participate. Following this cover letter were informational sheets from AHP about the purpose of the survey and the enrollment process. A single organizational designee completed the organizational survey component online, providing program manager names and email addresses. Organizations that did not initially respond were encouraged to do so via email, telephone, and U.S. mail reminders, which included sample reports that served as an organizational incentive. Once an organization completed the organizational component of the survey, a unique version of the program manager and staff survey was created to reflect the structure of the organization. Program managers were mailed invitational emails with recruitment letters as attachments to be distributed to staff. Regular reminders were sent to program managers, including counts of staff responses for each program, which were copied to the organizational designee and/or executive director. A variety of additional measures were employed to encourage participation at each stage of the survey. For most industries, personnel from the relevant Oklahoma state agency made additional follow-up calls. Additionally, AHP staff made in-person visits to key organizations to provide assistance in participating in the survey, or to encourage participation. Other Data Sources While the workforce survey is the largest component of this project and is generally the focus of this report, data were drawn from a variety of additional sources: • Economic Modeling Systems Inc (EMSI): The Oklahoma Department of Commerce provided average hourly wage rate norms for a range of behavioral healthcare positions at the national, regional and state level. • University of North Carolina (UNC) Staffing Needs Study: Data were drawn from a UNC study of professional staffing shortages, conducted under contract to the Health Resources and Services Administration (HRSA). • Oklahoma State Regents of Higher Education: The Regents of Higher Education provided information on the number of behavioral healthcare related degrees awarded by category and by year since 2001, as well as information on the number of degrees anticipated to be granted and anticipated to be needed. The data derived from these sources complement the data collected from the survey and provide information on subjects that could not be covered by the survey. In doing so, they allow the project to provide a more comprehensive response to the Workforce Study Team’s interests and goals. 16 STAFF SEPARATIONS Staff separation rate (turnover) is a near-universal concern in behavioral healthcare programs. High separation rates increase program costs, reduce return on investment for staff development, and impact quality of care. Anecdotal evidence of the negative impact of turnover on provider-consumer relationships abounds. Given this, it is not surprising that study stakeholders identified staff separation as a principal area for investigation. Information was gathered on staff separations through both the program manager and staff surveys. Program managers were asked to review a list of 18 possible causes of staff turnover and were then asked to indicate which of these were most relevant to their program. Managers were also asked to report on the number of separations within the last year in their program using the study’s six primary position categories. Staff members were asked to report whether they intended to leave their position within the next 12 months. This section will describe the data received from program managers and staff in response to these survey items, and the analysis conducted to investigate relationships between these items and other program, organizational and staff characteristics will be discussed. Program Manager Perceptions of Causes of Turnover Program managers were asked to identify three causes of staff turnover in their programs. The causes cited by 235 programs are shown in Exhibit 2.1, those causes cited by fewer than 10% of program managers are not shown in the exhibit.3 Percentages for this item add up to more than 100, as three causes of turnover were selected for each program. Program managers perceive dissatisfaction with salary/pay as the greatest contributor to staff separations; 63% (from all industry groups) cited dissatisfaction with pay as a significant cause of turnover in the behavioral healthcare field. Other factors contributing to turnover, cited by at least one third of the program managers, were excessive paperwork (43%), emotional burnout (36%) and excessive on-the-job stress (33%). 3The following potential causes of turnover were listed as options on the survey, but were cited by fewer than 10% of program managers: dissatisfaction with workplace location; dissatisfaction with relationship with supervisor; dissatisfaction with on-call responsibilities; difficulties with transportation; difficulties with child care; dissatisfaction with health insurance; dissatisfaction with time off; concern about on-the-job safety; and dissatisfaction with coworkers. 17 Exhibit 2.1: Program Manager Perceptions of Causes of Turnover Across Industries Data from the program manager surveys. We examined the relationships of the perceived causes of staff turnover, cited by no fewer than 10% of program managers, to seven key dimensions – industry group (Mental Health, Substance Abuse, Department of Human Services, Office of Juvenile Justice, Oklahoma Psychiatric Hospital Association, Child Guidance, Federally Qualified Health Centers, Other Medicaid and the Department of Corrections)4, region (northwest, southwest, northeast, southeast, Tulsa metro, Oklahoma City metro), service type (mental health, substance abuse, combined mental health and substance abuse, and services for people with developmental disabilities and mental health or substance abuse needs), program setting (inpatient, criminal justice, residential, or outpatient), service population (children, adults, both), organizational type (state vs. private), and organizational size (small, medium, large). The following causes of turnover were significantly different (p<.05) across at least one of the seven dimensions: (1) dissatisfaction with salary/pay (Salary), (2) dissatisfaction with career ladder, (3) excessive paperwork (Paperwork), (4) dissatisfaction with job responsibilities (Responsibilities) and (5) dissatisfaction with shift/work hours (Hours). While none of these causes of turnover varied by region or service type, there was variation across industry group, program setting, service population, organizational size, and organizational operation (state vs. private), also considered a proxy for organizational benefits. Following these findings, logistic regressions were performed to examine the relationship between the dimensions - taken together - and each of the following four causes of turnover: Salary, Responsibilities, Hours and Paperwork. Industry, service population, organizational type, program setting, and organizational size were included in this testing. Tables summarizing the results of these regressions can be found in Appendix A1. Four additional parsimonious 4 Industry group name and abbreviation: Mental Health (CMHC), Substance Abuse, Department of Human Services (DHS), Office of Juvenile Justice (OJA), Oklahoma Psychiatric Hospital Association (OPHA), Child Guidance, Federally Qualified Health Centers (FQHC), Other Medicaid (MA) and the Department of Corrections (DOC). 18 logistic regression models can be found in Appendix A1 as well, for a total of eight regression models. In summary, when controlling for other factors, program manager perceptions of causes of staff turnover suggest that: 1. The role of salary/pay in turnover varies by industry. 2. The role of excessive paperwork and dissatisfaction with job responsibilities in turnover varies by service populations. 3. The role of excessive paperwork in turnover also varies by program settings. Pay as a Perceived Cause of Turnover Exhibit 2.2 provides details of the relationships between organizational industry and pay as a perceived cause of turnover. Industries with fewer than ten program manager responses were not included in the analysis. Program managers in OJA organizations were most likely to cite pay as a cause of turnover, while those in OPHA organizations were least likely to do so. Specifically, 90% of program managers from the OJA industry group perceived staff dissatisfaction with salary/pay as one of the top reasons for staff separations while program managers from the OPHA industry group were only half as likely to name dissatisfaction with salary/pay. At least 70% of program managers from the Mental Health and DHS industry groups cited salary/pay as a cause of turnover. This relationship was upheld in the regression analyses as well with industry being a significant predictor of program manager perceptions of pay as a significant cause of turnover. Program setting was not significant in the logistic regression model. Organizational size5 and organizational operation were significant when these relationships were considered individually, but did not remain significant when multiple relationships were tested simultaneously. Exhibit 2.2: PM Perceptions of Pay as a Cause of Staff Turnover by Industry CMHC N=102 DHS N=17 OJA N=10 OPHA N=26 SA N=61 Dissatisfaction with salary/pay 76% 71% 90% 42% 53% Data are significant at the p<.05 level. ♦ Data from the program manager surveys. ♦ FQHC, DOC, Other Medicaid, and Child Guidance industries are not included in the analysis because there were fewer than ten programs in these samples. Excessive Paperwork as a Perceived Cause of Turnover Exhibit 2.3 shows the relationship between program manager perception of excessive paperwork as a cause of staff turnover and program setting. Program settings were defined as follows: Inpatient – an acute care mental health unit in a hospital, a unit in a substance abuse detoxification facility, or a residential unit within a hospital; Outpatient – a unit in a community mental health center, a day program, a psychiatric rehabilitation (PSR) program or a Program of Assertive Community Treatment (PACT)/case management program; Residential (not hospital-based) – a group home or a supported housing program; and Correctional/Criminal Justice – a 5 Organizational size – programs are the unit of analysis. Program managers were asked to identify the number of full-time staff working in each program they supervised. The number of full-time staff were aggregated for each organization. An organizational response rate was calculated and the total number of staff in each organization was divided by the organizational response rate and multiplied by 100. This yielded the total number of full-time staff in each organization (i.e., total staff) which was then divided into three groups – small, medium and large organizations – based on the overall distribution of the total staff. 19 prison or a juvenile detention facility. Excessive paperwork is cited as a cause of separations by 60% of program managers from outpatient facilities, followed by those in residential (21%), inpatient (20%) and criminal justice facilities (10%). The relationship between program setting and excessive paperwork remains when the effects of other variables are considered. Although, industry group and excessive paperwork had a strong relationship when looking at the two variables in isolation, the former is no longer a predictor of excessive paperwork when multiple relationships were tested simultaneously. On the other hand, service setting had a different effect on excessive paperwork: there was no relationship between service setting and paperwork when considered alone, but it becomes a significant predictor of paperwork when multiple relationships were tested (Model 2 of the logistic regressions). Program managers in programs serving children cite excessive paperwork as a cause of turnover more frequently than those serving both children and adults. Exhibit 2.3: PM Perceptions of Paperwork as a Cause of Staff Turnover by Program Setting Inpatient N=30 Outpatient N=119 Residential N=47 Correctional N=10 Excessive paperwork 20% 60% 21% 10% Data are significant at the p<.05 level. ♦ Data from the program manager surveys. Dissatisfaction with Job Responsibilities as a Perceived Cause of Turnover Although dissatisfaction with job responsibilities varies by service population (Exhibit 2.4), program managers supervising programs serving both children and adults are far less likely (4%) to perceive job responsibilities as one of the most important causes of staff turnover. In other words, programs serving adults only and children only are more likely to have staff dissatisfied with their job responsibilities, 21 and 22% respectively. While this relationship may not initially seem meaningful, it could be related to the relationship between service population and program setting. Eighty percent of programs serving both children and adults are categorized as outpatient programs. Compared to program managers in inpatient and residential programs, fewer outpatient program managers cite job responsibilities as a significant cause of turnover in their programs. The relationship between job responsibilities and service population is further supported by model 3 of the logistic regressions (see Appendix A1 - Factors Influencing Program Manager Perceptions of Staff Dissatisfaction with Job Responsibilities as a Cause of Turnover). Organizational size was not significant in the regression model. Although dissatisfaction with job responsibilities varied by industry, program setting, and service population, these were not significant predictors in the full regression model. Exhibit 2.4: PM Perceptions of Responsibilities as a Cause of Staff Turnover by Service Population Children/Adults Adults Only Children Only Dissatisfaction with job responsibilities 4% 21% 25% Data are significant at the p<.05 level. ♦ Data from the program manager surveys. 20 Program Manager-Reported Separation Rates Program managers were asked to report the current number of full time equivalents (FTEs) budgeted for their program and vacant in their program, as well as the number of staff separations that had occurred over the previous 12 months in their program. These items were posed in reference to each of six position categories: aids/techs/other paraprofessionals, professionals primarily holding Masters degrees (counselors/therapists/MSW-level social workers), LPNs, psychiatrists and other physicians, doctoral-level psychologists/DSW-level social workers, and RNs. This position category structure was developed based on a review of the state position classification and the U.S. Bureau of Labor Statistics Standard Occupational Code (SOC) system. Appendix A15 shows relevant SOC positions categorized according to this six-position structure. To calculate the separation rate for a given region, the number of separations was totaled across participating programs, and this sum was divided by the number of FTEs budgeted across programs. Exhibit 2.5 shows the position-specific and total separation rates statewide, and for each of the six geographic regions. It is important to note that organizations may not have included providers that are contracted with, rather than employed, in the counts that follow. Exhibit 2.5: Cross-industry Program Manager-Reported Separation Rates by Region Position NE NW OKC SE SW Tulsa Statewide Aid/tech 51% 55% 34% 38% 50% 30% 42% Masters-level professional 28% 26% 26% 27% 8% 27% 25% LPN 32% 29% 40% 50% 33% 10% 36% Psychiatrist/ physician 33% 0% 4% 44% 25% 20% 22% Psychologist 13% 0% 0% 0% 0% NA 7% RN 25% 33% 29% 56% 23% 21% 28% Total 40% 41% 31% 35% 32% 27% 34% Data from the program manager surveys. Calculating Program Separation Rate Percents in the table above were calculated by summing separations and budgeted positions across the region. In the analysis that follows, separations are calculated at the program level. Programs, rather than organizations, were chosen as the unit of analysis due to concerns that program characteristics and local program environment may vary widely within larger organizations - particularly those with programs across a wide geographic range. Program separation rates ranged from 0% to 200%. Separation rates of greater than 100% are possible because positions may turn over more than once within a year. The median separation rate was 25%, meaning that roughly half of the participating programs had a separation rate below 25%, and roughly half had a separation rate above 25%. Appendix A2 gives more information on the distribution of the program separation rates. 21 The initial analysis of relationships between separation rates and other program variables was attempted with three approaches to handling separation rates: by breaking participating programs first into two groups of equal size, then into three groups of equal size, and finally into four groups of equal size. The approaches yielded fairly similar results, with those for the two group approach being slightly more favorable than those for the alternatives. This approach involves dividing the group at the median of 25%, a rate which is consistent with a high turnover definition used in a recent, related study (Strolin-Goltzman, 2008). Relationships Between Separation Rates and Other Program Variables The relationship between separation rate and a number of program characteristics and related variables was examined. Relevant, recent literature was reviewed. The following identifying program characteristics were identified as being potentially related to separation rates: 1. Staff role clarity 2. Staff job satisfaction 3. Staff salary and benefits 4. Staff sense of personal accomplishment 5. Staff age 6. Staff intention to leave 7. Staff job level/experience 8. Staff burnout 9. Lack of alternative job options The primary source of information for items 1- 6 is the staff survey. Because of concerns about the representativeness of the staff data, these items were not considered feasible for this analysis. Most of these variables are also established predictors of staff intention to leave, and could therefore be employed in the predictive model of intention to leave (itself the strongest predictor of separation rates, Mor Barak et al., 2001). Staff job level/experience as a program characteristic was measured using the program manager reports of the FTEs budgeted for their programs. As these reports were specific to position type, we were able to create variables reflecting the proportion of each position type within each program’s staffing pattern. Masters-level counselors and techs made up by far the largest proportion of program staff. On average, Masters-level counselors made up 50% of the program staff, and techs made up 39%. The remaining four position categories ranged from a high of 6% (RNs) to a low of 1% (PhDs). Appendix A3 offers more information about the distribution of each of the six position type proportions. Staff burnout as a program characteristic was measured by program manager indication that burnout is one of the top three reasons for staff turnover within their program. We also looked for relationships between the other frequently-cited causes of turnover and separation rate. A proxy for lack of alternative job options was created using the program region code: Programs located in the Tulsa and Oklahoma City metro areas were considered to be located in areas with better alternative job options, while those in the remaining, more rural, regions were coded as being located in areas with fewer job options. 22 Finally, relationships between separation rate and each of the study dimensions described earlier (industry, region, service type, program setting, population age, state operation, and organizational size) were examined. Analysis and Results Analysis to identify relationships between separation rate and each of the variables above on an individual basis was performed. Most of these did not prove to be statistically significant: None of the frequently-cited causes of turnover were associated with program separation rate, nor was the job options proxy. Of the staffing and study dimensions variables, proportion of Masters-level counselors, proportion of techs, industry, and state operation were significantly associated with separation rate, as were two approaches at measuring benefits. Upon closer inspection, the results for the benefits items were difficult to interpret (i.e., suggesting an inconsistent or nonsensical relationship between separation rate and benefits). These items were discarded. Further information about the items and the relationships identified may be found in Appendix A4. Following this analysis, logistic regression was used to investigate whether the relationships between separation rate and program characteristics remained significant when all characteristics were considered simultaneously. Looking at the relationship between the two staffing variables (proportion of techs and proportion of Masters-level counselors) it was determined that these variables were too closely related to include in the regression model. Details of the analysis used to determine this can be found in Appendix A5. Ultimately, the model included the following program characteristics: proportion of techs, industry, and state operation. While proportion of techs and state operation remained significant in the regression model, industry became insignificant, suggesting that the relationship between industry and separation rate may have been in part due to a relationship between industry and state operation, or possibly between industry and staffing patterns. A detailed look at the results of this model can be found in Appendix A6. As shown in Exhibit 2.6, on average techs made up 31% of the staff in low separation programs, while they made up nearly half of the staff in high separation programs. This is consistent with the literature indicating that high staff experience, job level, and pay are associated with lower turnover. Exhibit 2.6: Proportion Techs in Low Separation and High Separation Programs Staff position type predictors Mean proportion low separation programs Mean proportion high separation programs Proportion Techs 31% 48% Data from the program manager surveys. A more detailed look at the relationship between position type and separation is available in Appendix A7. The distribution for programs in state vs. privately operated organizations is also as was anticipated. Half of the programs in private organizations fall into the high turnover group, 23 while less than one-third of the programs in state-operated organizations do. It is believed that this relationship is at least in part a result of the better compensation package offered by state-operated organizations. Exhibit 2.7: Proportion of Programs in High Separation Group by State/Private Operation Operation (assigned) Private (N=188) State (N=56) Total (N=244) Proportion in high turnover group 50% 29% 45% Data from the program manager surveys. While only a proportion of techs and state operation remained significant in the regression model, Appendix A8 offers details on the remaining variables that were tested. Staff Intention to Leave Staff were also asked about their plans to leave their organizations within the next year. Those who reported that they were planning on leaving were asked to indicate whether they planned to retire, find another job within the behavioral healthcare field, find a job outside the field, or pursue some other option. Exhibit 2.8 shows the percentages of program managers and staff reporting each of these plans. Exhibit 2.8: Intention to Leave Frequencies Response (N=1244) % No, don’t intend to leave within a year 80% Yes, to retire 1% Yes, to take another job in behavioral health 7% Yes, to take a job outside behavioral health 4% Yes, other 7% Data from the staff surveys. This variable was recoded into two categories by combining all categories representing any intention to leave (do intend to leave within a year: 20%, and do not intend to leave within a year: 80%) for the analysis that follows. Relationship Between Intention to Leave and Staff and Program Variables As with separation rates, predictor variables were chosen following a review of the literature. This review supported the use of the following variables: 1. Staff burnout; 2. Work-life fit; 3. Job satisfaction; 4. Empowerment; 5. Workplace incivility; 6. Staff age; 7. Job level/experience; 24 8. Professional and job commitment; and 9. Income. Staff burnout was not measured directly by the staff survey. Related items, such as My workplace is too stressful, would have appeared to provide reasonable proxies but more closely matched other predictors examined in and not supported by the literature. The same holds true for workplace incivility and empowerment. The survey did not examine work-life fit or professional/job commitment. The survey’s overall job satisfaction item was chosen as an indicator of job satisfaction. The survey’s staff age variable was transformed into a continuous variable by recoding age categories into midpoints, except for over 64 which was recoded as 69.5, the midpoint between 65 and 74. The survey’s categorical staff income variable was treated in a similar manner, with the following differences: The lowest category (<$10.00/hr) was recoded as the midpoint between $10.00 and $7.25, the minimum wage in Oklahoma. Position types and education level for respondents who checked the upper category ($50.00/hr or more) were examined, and were surprisingly found to be primarily Masters-level therapists, along with a few physicians. For this reason, we used a rate relatively close to the second-highest category, and significantly below one that might be expected for physicians: $62.50. Staff responses to the item How many years have you been in the field? were used to measure staff experience. Detailed information on the distribution of these variables is offered in Appendix A9. Gender and ethnicity were tested using the original dichotomous survey items, and race was tested by collapsing five dichotomous survey items into a single variable with up to six categories: American Indian/Alaskan Native alone, Asian alone, Black/African American alone, Native Hawaiian/Pacific Islander alone, White alone, and more than one race. Due to low Ns, the Asian alone and Native Hawaiian/Pacific Islander alone categories were eliminated from the crosstabs. In addition to the variables gathered through the staff survey, the relationship of staff intention to leave to key program variables was investigated, including the program manager-cited causes of turnover and the study dimensions described earlier (industry, region, service type, program setting, population age, state operation, and organizational size). Analysis and Results As with the analyses described earlier, relationships were examined between intention to leave and each of the variables described above on an individual basis. As with separation rate, there was no relationship between intention to leave and program manager citation of the significant causes of turnover. Of the study dimensions, only service type and region were significantly related to staff intention to leave. Staff position type, gender, ethnicity and race were not significant, but staff age, experience, pay and job satisfaction were significant. Initially, the relationship between consumer or family status and intention to leave was investigated by collapsing eight dichotomous survey items into a single four-category variable: neither, consumer only, family member only, and both consumer and family member. This variable was significantly related to intention to leave. However, the distribution was difficult to interpret as staff who identified as being consumers only seemed much less likely to intend to separate than did staff who identified as either family members or both consumers and family members (full 25 details available in Appendix A10). Given this, it seemed possible that the use of this collapsed variable could be obscuring the meaning of the relationship. A variety of alternatives were tested, including the eight original survey items (adult mental health consumer, adult substance abuse consumer, former youth mental health consumer, former youth substance abuse consumer, family member of an adult mental health consumer, family member of an adult substance abuse consumer, family member of a youth mental health consumer, family member of a youth substance abuse consumer), as well as aggregations of these items across two dimensions individually and together (adult/youth and mental health/substance abuse). Most of these tests did not yield significant results. However, family status did prove to be significantly related to intention to leave, with a higher proportion of family members than non-family members indicating that they planned to leave within the next year. When family membership was broken down further into mental health and substance abuse, the relationship between being a family member of a mental health consumer and intention to leave was significant, while that between being a family member of a substance abuse consumer and intention to leave was not significant. However, as the latter relationship showed a similar trend (higher intention to leave among family members), the combined mental health and substance abuse variable was retained for further analysis. Logistic regression was employed to determine whether the relationships noted above remained significant when considered simultaneously. We began by examining the relationship between staff age, experience, pay, and job satisfaction. While there were some relationships among these variables, none turned out to be strong enough to warrant excluding any of the variables from the regression model. The details of this analysis can be found in Appendix A11. The model tested included region, service type, population age, job satisfaction, pay, age, experience, and family member status. Of these variables, only job satisfaction and age remained significant. The mean satisfaction score for staff not intending to leave was 1.71, with 1 being very satisfied and 2 being satisfied (Exhibit 2.9). The mean for staff intending to leave was 2.59, closer to 3, or neither satisfied nor dissatisfied. Consistent with literature on the topic, staff intending to leave were younger on average than those not intending to leave (39.67 years versus 43.30 years, respectively). Complete details on the results of the regression model are shown in Appendix A12, and additional details on the relationship of job satisfaction and staff age to intention to leave are shown in Appendix A13. Exhibit 2.9: Satisfaction and Age Among Staff Intending to Stay and Intending to Leave Mean for staff staying Mean for staff leaving Staff overall job satisfaction (N=1241) 1.71 2.59 Staff age (N=1180) 43.30 39.67 Data from the staff surveys. While only these two variables remained significant in the regression model, Appendix A14 gives additional information on the other variables tested. 26 Summary Information related to separations was gathered through program manager reports of the perceived causes of separation in their programs, program managers’ reports of their programs’ separation rate over the previous year, and staff reports of their intention to leave their position within the next year. Nine causes of turnover were cited by at least 10% of program managers; those causes cited by fewer than 10% of program managers are not included in the analysis. Consistent with the findings related to recruitment barriers, the most frequently cited barrier was dissatisfaction with pay, which was cited by nearly two thirds of program managers. Excessive paperwork, emotional burnout and excessive on-the-job stress were cited by at least one third of program managers. While program and organization characteristics were related to multiple perceived causes of turnover when the relationships were examined individually, generally only one or two characteristics remained significant in each logistic regression model. Organizational industry was a significant predictor of citing dissatisfaction of pay, with OPHA program managers being the least likely to cite pay as a cause of turnover. Population age and program setting were significant predictors of perceiving paperwork to be a cause of turnover, with program managers in programs serving children citing paperwork more frequently than those serving both children and adults, and program managers in outpatient settings citing paperwork more frequently than program managers in other settings. Service population also related to citation of dissatisfaction with job responsibilities, with program managers from programs serving both children and adults being less likely to cite this as a barrier than program managers from programs serving either adults or children. Program separation rates ranged from 0% to 200%, and the median of 25% was used to divide programs into two categories: low separation and high separation. These categories were related to multiple program and organizational characteristics when the relationships were examined individually, but only two characteristics remained significant in the logistic regression model. High separation programs proved to be more likely to have a high proportion of techs on staff, and less likely to be state operated. The position type results are consistent with existing literature regarding the relationship between lower staff experience/job level and higher separation rates. The vast majority (80%) of staff did not report intending to leave their positions within the 12 month period following the survey. Intention to leave was related to a range of program, organizational and staff characteristics when the relationships were examined individually, but only two remained significant in the logistic regression model. As would be expected, staff intending to leave reported lower satisfaction with their job overall. Staff age was also related to intention to leave, with the mean age for staff intending to leave being about three and a half years younger than that of staff intending to stay. Both of these findings are consistent with the literature on staff intention to leave. 27 VACANCIES AND STAFF RECRUITMENT BARRIERS Like staff separations, position vacancies are an area of concern in many behavioral healthcare programs, and a topic of interest to the study stakeholders. We collected information on position vacancies on two issues, using the program manager survey. First, program managers were asked to review a list of 19 possible barriers to staff recruitment, and were then asked to indicate which of these were most relevant to their program. Second, program managers were asked to report on the current vacancies in their program, using the six position categories described in the separations section. This section describes the data received in response to each of these sets of items, and the analysis conducted to investigate relationships between these variables and program characteristics. Program Manager Perceptions of Recruitment Barriers Program managers were asked to identify the top three barriers to filling staff vacancies in their programs. As a result, percentages for this item add up to more than 100%. The list shown in Exhibit 3.1 does not include the barriers cited by fewer than 10% of program managers.6 The majority (57%) of program managers identified salary/pay as the greatest obstacle to filling vacancies in their programs. Lack of candidates with desired credentials or work experience, small applicant pool due to geographic location and competition from other fields were cited by 25% or more program managers as barriers to staff recruitment. Exhibit 3.1: Program Manager Perceptions of Recruitment Barriers Data from the program manager survey. 6 The following potential barriers were listed as options on the survey, but were cited by fewer than 10% of program managers: cumbersome hiring process; career ladder not attractive; childcare not offered; organizational facilities not attractive; organizational reputation; negative stereotypes of service consumers; job responsibilities not attractive; amount of training required; cost of training required; and benefits not attractive. 28 Recruitment Barriers and Program Variables The next four tables illustrate how recruitment barriers vary by industry, region, organizational size and type (those barriers cited by fewer than 10% of program managers are not included in the analyses). Other results indicate that recruitment barriers did not vary by service type (i.e., mental health, substance abuse and dual-diagnosis). The following items are those most often cited by program managers as reasons for vacancies: 1. Salary/pay not attractive; 2. No candidates with desired credentials; 3. No candidates with desired work experience; 4. Competition from other fields; 5. Problems with funding/not allowed to fill a position; and 6. Shift/work hours not attractive. Industry and Recruitment Barriers All six of the perceived barriers above show statistically significant differences between industries. Eighty percent of program managers in the OJA industry group identified salary/pay as one of the most critical barriers to filling vacancies, while only 19% of OPHA industry group program managers cited this as a recruitment barrier. Substance Abuse fell roughly in the middle of this continuum, with 49% of program managers citing pay as a barrier, while Mental Health and DHS program managers responded relatively similarly to those from OJA, with 75% and close to 60% of program managers from these industries citing pay as a recruitment barrier, respectively. No OJA program managers cited difficulty finding candidates with desired credentials, but roughly two fifths of Mental Health and Substance Abuse industry program managers perceive this to be a recruitment barrier in their programs. Competition from other fields also varies by industry. Program managers working in the OJA industry group were more likely to cite this as a barrier to staff recruitment (70%) than program managers from any other industry group. In fact, the next closest group was program managers from the Mental Health industry, with 32%. One third of program managers from the Substance Abuse industry group perceive funding or not being allowed to fill a position as one of the most pertinent causes of vacancies. OPHA and the Mental Health industries followed with 15% and 14% respectively. Only 10% to 12% of program managers in the DHS and OJA industries thought funding was an important recruitment barrier. Not surprisingly, shift/work hours is more frequently perceived as a barrier by program managers in industries with a high proportion of 24-hour programs (OPHA, OJA). Finally, while nearly one third of OJA program managers perceive the hiring process itself to be a barrier, this process was not cited as a barrier by any Substance Abuse industry program managers. 29 Exhibit 3.2: Program Manager Perceptions of Recruitment Barriers by Industry Perceived Barrier CMHC N=101 DHS N=17 OJA N=10 OPHA N=26 SA N=61 Salary/pay not attractive 74% 59% 80% 19% 49% No candidates w desired credentials 37% 24% 0% 15% 41% Competition from other fields 32% 12% 70% 31% 15% Funding/not allowed to fill position 14% 12% 10% 15% 33% Shift/work hours not attractive 17% 24% 40% 42% 15% Cumbersome hiring process 11% 18% 30% 15% 0% Data from the program manager surveys. ♦ Items cited by fewer than 10% of program managers are not included in the exhibit. ♦ All data are significant at the p<.05 level. ♦ FQHC, DOC, Other Medicaid, and Child Guidance industries are not included in the analysis due to the low number of programs responding to these items. State Operation and Recruitment Barriers The perceived barriers of salary, candidate experience, and funding vary by organizational type or operation (state operated vs. privately operated). As shown in Exhibit 3.3, nearly three-quarters of program managers from state operated organizations cite salary as a barrier, in comparison to just over half of program managers from privately operated organizations. As noted above, OPHA program managers were also significantly less likely to cite salary as a barrier. Interestingly, OPHA is the only industry group in these analyses that are made up of entirely private organizations. Program managers from state-operated organizations were significantly less likely than those from private organizations to cite lack of candidates with desired work experience as a recruitment barrier. Finally, state-operated organizations (42%) were more likely than privately-operated (17%) to cite funding as a fundamental problem to staff recruitment, and are also more likely to cite salary as a recruitment barrier. Exhibit 3.3: Program Manager Perceptions of Recruitment Barriers by Organizational Type Perceived Barrier State Operated N=53 Privately Operated N=181 Salary/pay not attractive 74% 52% No candidates with desired experience 6% 34% Funding/not allowed to fill position 42% 17% Data from the program manager surveys. ♦ Items cited by fewer than 10% of program managers are not included in the exhibit. ♦ All data are significant at the p<.05 level. Organizational Size and Recruitment Barriers Organizational size is associated with program manager perception that salary and lack of staff with desired credentials are recruitment barriers. Program managers affiliated with large organizations (those with an estimated staff size of at least 82 full time employees) cite salary/pay as a reason for staff vacancies more often than those affiliated with other organizations (67%, compared to 42% and 45% of small and medium organizations, respectively). Further bivariate analysis indicates that small organizations (those with an estimated staff size of less than 15 full-time employees) have more professional staff – requiring additional education – and are less likely to be inpatient facilities requiring a large number of 30 aides/techs who typically earn the lowest salary among direct care staff. While program managers from medium and large organizations cite lack of candidates with credentials at roughly the same rate, those from smaller organizations cite this barrier at a considerably higher rate. Exhibit 3.4: Program Manager Perceptions of Recruitment Barriers by Organizational Size Perceived Barrier Small Orgs N=33 Medium Orgs N=53 Large Orgs N=126 Salary/pay not attractive 42% 45% 67% No candidates with desired credentials 52% 30% 27% Data from the program manager and organizational surveys.♦ Items cited by fewer than 10% of program managers are not included in the exhibit. ♦ All data are significant at the p<.05 level. Region and Recruitment Barriers Finally, geographic region was significantly related to four of the perceived recruitment barriers: absence of candidates with desired work experience, small applicant pool due to geographic location, competition from other fields, and location of agency not attractive. Not surprisingly, two of these barriers are explicitly location-based, and a third (lack of candidates with desired work experience) could also be argued to be intrinsically tied to location or area. Exhibit 3.5 demonstrates that a small pool of applicants is the greatest barrier to filling vacancies (52% and 47%, respectively) in the Northeast and Southeast corridors of the state, while about half of the program managers from the Northwest indicated that competition from other fields was a problem with respect to vacancies in the behavioral healthcare field. Exhibit 3.5: Program Manager Perceptions of Recruitment Barriers by Region Perceived Barrier NE N=54 NW N=14 OK N=74 SE N=32 SW N=30 TU N=26 No candidates w desired work experience 15% 21% 30% 19% 40% 46% Small applicant pool due to geographic location 52% 43% 7% 47% 23% 4% Competition from other fields 19% 50% 34% 28% 7% 42% Location of agency not attractive 35% 14% 8% 13% 3% 0% Data from the program manager surveys.♦ Items cited by fewer than 10% of program managers are not included in the exhibit. ♦ All data are significant at the p<.05 level. Salary as a Perceived Recruitment Barrier Given that salary was the most frequently cited recruitment barrier as well as the most frequently cited cause of separations, it warranted further exploration. Logistic regression was used to determine whether the three program variables (industry, state operation, and organization size) remained significant predictors of salary as a barrier when tested simultaneously. While organization size did not remain significant, both industry and state operation were significant. The significant relationship between salary as a perceived barrier and industry can be attributed to the low proportion of OPHA program managers citing salary as a barrier. There was a significant difference between the rate of OPHA citation of salary and that of the mental health industry program managers, who were chosen as the reference group in the regression model. As suggested by the earlier analysis of the relationship between salary as a perceived barrier and 31 state operated status, the results of the regression model indicated that program managers in state operated programs were significantly more likely to cite salary as a recruitment barrier. Further details on the results of this regression model may be found in Appendix B1. Program Manager-Reported Vacancy Rates As reported in the section on separations, program managers were asked to report the current number of full time equivalents (FTEs) budgeted for their program and vacant in their program. These items were posed in reference to each of six position categories: aids/techs/other paraprofessionals, professionals primarily holding Masters degrees (counselors/therapists/MSW-level social workers), LPNs, psychiatrists and other physicians, doctoral-level psychologists/DSW-level social workers, and RNs. To calculate the vacancy rate for a given region, the number of vacancies was totaled across participating programs, and this sum was divided by the number of FTEs budgeted across programs. Exhibit 3.6 shows the position-specific and total vacancy rates statewide, and for each of the six geographic regions. It is important to note that organizations may not have included staff that they contract with (rather than employ) in the counts that follow. Exhibit 3.6: Cross-Industry Vacancies by Region Position NE NW OKC SE SW Tulsa Statewide Aid/tech 7% 13% 8% 7% 16% 8% 9% Masters-level professional 15% 9% 12% 11% 2% 36% 15% LPN 4% 14% 9% 20% 33% 0% 9% Psychiatrist/ physician 3% 0% 0% 33% 13% 10% 7% Psychologist 6% 50% 0% 0% 0% NA 7% RN 13% 22% 15% 28% 8% 7% 14% Total 10% 13% 10% 12% 10% 17% 11% Data from the program manager surveys. Calculating Program Vacancy Rate Percents in the table above were calculated by summing vacancies and budgeted positions across the region. In the analysis that follows, vacancies are calculated at the program level. As noted in the separation section, programs were chosen as the unit of analysis due to concerns that program characteristics and local program environment may vary widely within larger organizations - particularly those with programs across a wide geographic range. Program vacancy rates ranged from 0% to 100%. The median vacancy rate was 4%, meaning that roughly half of the participating programs had a vacancy rate below 4%, and roughly half had a vacancy rate above 4%. Appendix B2 gives more information on the distribution of the program vacancy rates. 32 Relationships Between Vacancy Rates and Other Program Variables We examined the relationship between vacancy rate and a number of program characteristics and related variables. Programs were categorized as either having a low vacancy rate (less than 5%) or high vacancy rate (5% or higher). We began by testing for relationships between vacancy rate and each of the frequently-cited recruitment barriers. Then, as with separation rate, we looked for a relationship between staffing patterns (e.g., proportion Masters-level counselors, proportion techs) and vacancy rate. Finally, we looked for relationships between vacancy rate and each of the study dimensions described earlier (industry, region, service type, program setting, population age, state operation, and organizational size). Analysis and Results We began by performing analysis to identify relationships between vacancy rate and each of the variables above on an individual basis, with the intention of then testing these relationships simultaneously. However, only one of the identified variables proved to be related to vacancy rates. None of the frequently cited recruitment barriers were associated with program vacancy rate, nor were any of the study dimension variables. Of the staffing patterns variables, only proportion of RNs was related to vacancy rate, with high vacancy programs having a greater proportion of RNs than low vacancy programs. As shown in Exhibit 3.7 the average proportion RNs for low vacancy programs was 4%, while the average for high vacancy programs was 7%. While this difference may appear relatively small, it was statistically significant. This finding may be related to the comparatively high rate of vacancies among RN positions overall. As noted earlier in Exhibit 3.6 the overall vacancy rate for RN positions was comparable to that for Masters-level counselors, which was the position type with the highest vacancy rate. Exhibit 3.7: Proportion RNs in Low Vacancy and High Vacancy Programs Staff position type predictors Mean proportion low vacancy programs Mean proportion high vacancy programs Proportion RNs 4% 7% Data from the program manager surveys. Because only one variable proved to be related to vacancy rate, it was not necessary to employ logistic regression to test multiple relationships simultaneously. Additional information on the (non-significant) findings for the remaining variables may be found in Appendix B3. Summary Information related to vacancies was gathered through program managers’ reports of the perceived recruitment barriers in their programs, and their reports of their programs’ current vacancies. Nine of the barriers were cited by more than 10% of program managers (those items cited by fewer than 10% of program managers were not included in the analyses). By far the most frequently cited barrier was salary, which was cited by 57% of program managers. Lack of candidates with desired credentials or desired work experience, small applicant pool due to geographic location, and competition from other fields were all cited by more than one quarter of program managers. Program and organization characteristics that were related to multiple perceived barriers included organizational industry, state operation, organizational size, and geographic region. Salary as a perceived barrier was related to three of these characteristics when the relationships were tested individually. When examined simultaneously, salary 33 remained significantly related to industry, with OPHA program managers being significantly less likely to cite salary as a barrier. Likewise, state operation and salary were related, with program managers in state operated organizations more likely to cite salary as a barrier. Program vacancy rates ranged from 0% to 100%, and the median of 4% was used to divide programs into two categories: low vacancy (less than 4%) and high vacancy (greater than 4%). These categories proved to be unrelated to most of the program and organizational characteristic variables. Staffing patterns offered one exception: The mean proportion of RNs in low vacancy programs was slightly but significantly lower than the mean proportion of RNs in high vacancy programs, which could be in part related to the comparatively high rate of vacancies in RN positions, across programs. 34 CURRENT AND FUTURE STAFFING NEEDS The purpose of this chapter is to identify unmet needs for the behavioral healthcare workforce with a focus on type of position. The first section focuses on psychiatrists and other prescribers. The second section focuses on other professional and non-professional staff. Each of these sections employs data from different sources so the methods upon which we have relied are described within each section, as well as the implications for higher education. The third section describes one underlying problem, the level of compensation currently available to the Oklahoma workforce. Need for psychiatrists and other prescribers of psychiatric medications State mental health authorities typically do not have empirical information about the characteristics of their current workforce. In order to fill this information gap, we undertook a number of studies, as well as searches for relevant data, that would provide useful information for understanding difficulties faced by staff providing mental health services in Oklahoma. Among the studies that we identified was a study of the relative unmet need for professional mental health workers in the State of Washington (Morrissey, et al, 2007a), undertaken as a part of their Mental Health Transformation State Incentive Grant. Morrissey and his colleagues employed a simple model as the foundation of their work. First, they estimated the number of adults (persons over age 18) who could be classified either as persons with serious mental illness or as persons with other mental health needs. For each of these two types of persons, they estimated the percentage that would access mental health non-inpatient services in one year and the number of units of professional services they would use. Professional services are broken down into those provided by individuals who are licensed to prescribe medications (prescribers) and individuals who are licensed to provide services other than medications (non-prescribers). These estimates then allow new estimates of the numbers of prescribers and non-prescribers needed (in full time equivalents—FTE) to serve a population within a defined geographic area. The estimates of need are then subtracted from the number of licensed professionals available to yield the shortage of professionals. They summarize their model as follows: Need = People with serious mental illness + people with other mental health needs Workforce = Prescribers + Non-prescribers Shortage = FTE available – FTE needed It is important to emphasize that these are relative not absolute measures of unmet need. This means that they are most useful in comparing the need from one area to another, but do not necessarily provide an estimate of the exact number of additional professional staff needed. Moreover, apparent surpluses produced by these estimates cannot be relied upon. The study of Washington State was a part of a larger, national study sponsored by the Health Resources Services Administration (HRSA) of the U.S. Department of Health and Human Services. This allowed Morrissey and his colleagues to develop estimates of professional shortages for every county in the U.S. We contacted them and requested estimates for 35 Oklahoma. Their findings, as well as the methods employed to arrive at their estimates, are presented here. We also discuss some of the limitations of their findings. Findings Most specialty prescribers in Oklahoma are psychiatrists, although there are a handful of advanced practice psychiatric nurses. Other physicians can and do prescribed psychiatric medications, as well. Exhibit 4.1 below presents regional and statewide estimates of counts of prescribers available to provide mental health services in Oklahoma. As previously discussed, the state is divided into six regions, as follows: Central Oklahoma (the counties in which Oklahoma City is located) and Tulsa are separately estimated. The remaining counties are grouped into four quadrants, Northeast, Northwest, Southeast, and Southwest. The table shows 278 FTE psychiatrists/prescribers. Exhibit 4.1: Available FTE Mental Health Specialty Prescribers by Licensure Group and by Oklahoma Regions Region Licensure Smoothed Total APPN PI Prescribers7 OKC 10 133 107 Northeast 2 30 78 Northwest 3 8 9 Southeast 3 13 24 Southwest 1 25 38 Tulsa 6 70 32 Total 23 278 287 For psychiatrists, full time equivalents are greater than the raw count because practice pattern data indicate that psychiatrists average more than 40 hours/week. Exhibit 4.2 below presents regional and statewide totals of FTE needed and FTE shortages for prescribers. For the prescriber group the UNC estimates produce a shortage of 410 FTE. 7 In the initial analysis, the county is used as the primary geographical unit for shortage estimation. This decision was made primarily due to the lack of accurate small-area data on mental health needs and practice locations, but also because people are likely to travel within larger areas for mental health services. Each county-level need and supply estimate was adjusted using a smoothing method that accounts for travel across county boundaries for mental health services. Within Oklahoma particularly, with its many small counties, ignoring this would lead to overestimates of need. The maximum amount of time that people can be expected to travel for mental health services is about 60 minutes (Fortney, Owen & Clothier, 1999; Fortney, Rost, Zhang et al., 1999). Therefore, for a given index county, the need and supply estimates of counties within a 60-minute radius were weighted and added to the estimates for the index county. The weighted estimates were scaled so that the national need and supply totals for prescribers and non-prescribers were unchanged by the smoothing process. In the final analysis, counties were aggregated by regions within Oklahoma. 36 Exhibit 4.2: Estimates of Shortages of Specialty Mental Health Prescribers FTE by Oklahoma Region Region Total FTE Available, Smoothed FTE Needed, Primary Care Adjusted, Smoothed Relative Shortage (FTE), Primary Care Adjusted, Smoothed OKC 107 187 -80 Northeast 78 217 -139 Northwest 9 32 -23 Southeast 24 107 -83 Southwest 38 94 -56 Tulsa 32 61 -29 Total 287 697 -410 Methods Methods are described in detail in Morrissey et al (2007b). Exhibit 4.3 below presents a brief summary of the data sources employed and how the estimates were derived. Exhibit 4.3: Data sources employed to estimate mental health workforce needs and available workforce. Variable Estimated Source of Data Oklahoma specific data Prevalence of Mental Illness (persons-in-need) National Comorbidity Survey Replication (NCSR); Medical Expenditures Panel Survey (MEPS) Yes Estimates of percent of persons-in-need using mental health services annually MEPS for non-SMI population; Assume 100% for SMI population No Estimates of average units of outpatient services used per person annually NCSR, MEPS No Estimates of visit hours per working day for prescribers Substance Abuse and Mental Health Services Administration (SAMHSA) No Estimates of need met by primary care providers Need estimate reduced by 15 percent in counties without a shortage of primary care providers (no single reference) Yes Estimates of supply of mental health professionals Various sources, generally relevant professional associations Yes Adjustments of need in rural counties that are close to larger counties Various references; assumed maximum travel time would be 60 minutes for mental health services Yes As shown above, Morrissey and his colleagues relied upon a number of data sources in order to estimate each of the variables required to determine workforce shortages. These sources are generally recognized as the most reliable sources of information available, although in several cases these may be the only sources available. 37 Discussion Prior to the completion of the work by Morrissey and his colleagues, the only available estimates of need for mental health professional services were on the website of the Health Resources and Services Administration. However, there was no explanation of the method employed to develop these estimates or references to underlying research. Thus the work described here represents the first systematic attempt to provide appropriate estimates of workforce needs and shortages. Nonetheless, there are limitations that must be recognized. Morrissey et al (2007b) acknowledge that the populations included do not extend to adults who are homeless or in institutions (e.g., inpatient, corrections) or children and adolescents. They also do not include needs for staffing of substance abuse programs. They indicate that the measure of shortage “is probably most useful when taken as an expression of relative rather than absolute unmet need.” Finally we reviewed the UNC estimates of available FTE with more recent data from Oklahoma State licensing boards. The UNC estimates are generally close, but underestimate the size of the current, licensed workforce. However, there is no data available on whether individuals who are licensed are actually engaged in clinical practice. We know anecdotally that at least some may be retired or only have a part-time practice or are working in administrative, rather than clinical positions. As we have also pointed out above, other licensed individuals are working in positions that are not counted in the need estimates (e.g., agencies serving child and youth, agencies providing adult or child inpatient care). As a result, we believe that the strategy of simply counting licensed practitioners leads to a systematic overestimate of the available supply of such professionals. Addressing the Shortage of Prescribers Oklahoma has three psychiatric residency programs which collectively produce about 13 new psychiatrists per year. Assuming that our estimate of the current need for over 400 prescribers of psychiatric medications is reasonably accurate, it would take over 30 years for these programs to fill the unmet need. This does not account for retirements during this period which will only increase the unmet need. It is unlikely that these residency programs will expand substantially or that psychiatrists will be recruited in significant numbers from elsewhere in the United States because this is a national problem. The numbers of new doctors entering psychiatric residency programs has been falling for over 20 years, and changes that would reverse this trend are not in the offing. Information about Osteopaths either training to practice psychiatry or trained to do so in Oklahoma suggests that this group also is not likely to expand the numbers of prescribers in the foreseeable future. There are no osteopathic residency training programs in psychiatry in Oklahoma, and only a few D.O.’s practice primarily psychiatry in Oklahoma (36) and even fewer are certified to do so (18). Given that psychiatrists cannot be expected to fill the need for new prescribers, what options exist? There are three that we know of: • Advanced practice psychiatric nurses can be trained to fill this need. At present there are only a handful of persons with this training in Oklahoma, but nursing schools could be encouraged to offer the necessary education. 38 • With an expansion of integrated primary care and mental health care, primary care physicians could become an expanded source of prescribers. Creating incentives to develop integrated care practices, targeting training in integrated care, and promoting continuing education in prescribing psychiatric medications, could contribute to an expansion in the numbers of competent prescribers. • Licensing Ph.D. psychologists with special additional training to prescribe medications would also expand the numbers of prescribers; two states now allow this. All three approaches may be necessary to fill the gap, which is quite substantial. If no action is taken to increase the numbers of prescribers, the problem may become worse with the retirements of older psychiatrists, who were trained in an era when psychiatry was a more attractive field. The numbers of retirements may exceed the small numbers of annual replacements. Conclusions The UNC data demonstrate an unequivocal need for more prescribing professionals in all areas of the state. The total estimate of need for 410 additional prescribers is probably an underestimate for reasons discussed above. The area of the State with the greatest unmet need is the Northeast quadrant, excluding Tulsa which has the smallest, relative unmet need. Non-prescribers Exhibit 4.4 below shows current staffing by position type for the state of Oklahoma and the six regions within the State. Exhibit 5 below shows population-based rates for behavioral healthcare positions by type in Oklahoma and the surrounding states. Oklahoma and the surrounding states are similar in most categories. The major exception is RNs which are less available in Oklahoma. LPNs and MH/SA Techs are marginally more available. There is considerable variation within Oklahoma. The Central Region (OKC) has among the highest rates of availability for all categories of positions. Tulsa is close and leads in availability of MH/SA Techs. The more rural areas of the State have significantly less availability of professionals, psychologists, MH/SA Counselors, and RNs. Exhibit 4.4: Current (2008) numbers of behavioral healthcare positions by positions type in Oklahoma Position State ODMHSAS Region Counts Tulsa OKC NE NW SE SW Psychologist 1,339 195 628 236 58 96 126 MH/SA Counselor 6,993 1,100 2,691 1,485 281 785 651 RN 26,157 5,714 10,839 3,560 962 2,766 2,316 LPN 13,463 2,411 4,163 2,062 716 2,100 2,011 MH/SA Tech 38,590 9,124 11,394 6,474 1,833 5,530 4,235 39 Exhibit 4.5: Current (2008) rates per 10,000 population of behavioral healthcare positions by position type in Oklahoma and surrounding states Position State Rate Multi- State Regional Rate* National Rate ODMHSAS Region Rates Tulsa OKC NE NW SE SW Psychologist 3.9 8.1 9.6 3.5 6.3 3.1 2.9 2.1 2.8 MH/SA Counselor 20.3 27.0 34.4 19.5 27.2 19.2 14.0 16.8 14.2 RN 75.8 87.0 92.7 101.4 109.5 46.1 47.8 59.4 50.6 LPN 39.0 35.0 27.5 42.8 42.1 26.7 35.6 45.1 43.9 MH/SA Tech 111.8 121.3 126.1 162.0 115.1 83.8 91.1 118.7 92.5 Regional rate includes the following states: Arkansas, Colorado, Kansas, Louisiana, Missouri, New Mexico, Oklahoma and Texas. Exhibit 4.6 below shows the current (as of 2008) numbers of behavioral healthcare staff by position type for Oklahoma and for the United States overall. It also shows the projected needs for staffing as of 2018. Projections are based principally upon projected population changes. The additional positions are necessary to maintain the same rates of services currently provided. Exhibit 4.6: Oklahoma and national current (2008) and projected (2018) rates of change for behavioral healthcare staffing positions Position Oklahoma National 2008 Jobs 2018 Jobs Change % Change % Change Psychologist 1,339 1,434 95 7.1% 15.8% MH/SA Counselors 6,993 8,270 1,277 18.3% 17.8% RNs 26,157 31,444 5,287 20.2% 22.3% LPNs 13,463 14,578 1,115 8.3% 12.9% Aids/Techs 38,590 46,122 7,532 19.5% 23.8% Total 86,542 101,848 15,306 17.7% 21.3% The difficulty is that these estimates include only job growth in predicting the numbers of new persons needed to fill available positions. However, current members of the behavioral healthcare workforce will be leaving their positions, either for retirement or other reasons. This also needs to be accounted for in developing estimates of persons needed to fill positions annually. Next, we outline an approach to employing appropriate data to reach this goal. 40 Job Growth We received the table below from the Department of Commerce. This means, for example, that in 2018 there will be 361 more jobs for psychologists than there were in 2008. One implication is that it is necessary to train or import 361 new psychologists into the system between 2008 and 2018. The source of these estimates is data from Economic Modeling Systems, Inc (EMSI). EMSI uses several different databases to come up with their estimates, including population projections from the Census Bureau. However, that is not the only factor, current employment trends and participation rates from the Bureau of Labor Statistics (BLS) go into the projections. Also included are IRS income and migration data that shed more light on the single employers or those that do not pay into Unemployment Insurance. The simple explanation is that it uses current employment trends by industry and certain population trends. Industry trends, legislation, and several other factors are also used to decipher which industries will be growing. Population is a key component but labor participation, county wages, migration patterns and trading patterns are also factors that influence the model. EMSI breaks down these trends to the county level, which can then be aggregated to the state level. The table below shows the rate of growth for ten years. The annual growth rate is between one and two percent, depending upon the position type. Exhibit 4.7: Ten Year Growth by Position Category Position Category State 2008 Jobs 2018 Jobs % Change Psychologist 2,738 3,099 13.2% Mental Health or Substance Abuse Counselors 9,726 11,377 17.0% Registered Nurses 26,552 32,271 21.5% Licensed Practical Nurses 13,936 15,554 11.6% Mental Health or Substance Abuse Aides/Assistants/ Technicians 44,546 54,536 22.4% Overall Total 97,498 116,837 19.8% This assumes that the persons occupying these positions in 2008 continue to be available to fill positions in 2018. However, we know that this is not the case. Some people who occupy positions in 2008 will retire or leave the field for other reasons over the next ten years. For example, it will be necessary to train or import more than 361 new psychologists into the system for these reasons. The problem is to estimate how many more psychologists will be needed to fill available positions. Accounting for persons leaving the field Let us assume that from 2008 to 2009 the growth rate for psychologists is two percent. Then the change in jobs would be an increase of 55, meaning that there would be a need for a minimum of 55 new psychologists to fill those jobs. In addition, let us assume that there is a 20 percent separation rate among psychologists during 2008 or 550 persons and further that ten percent of those separated actually leave the field. That would mean an additional 55 new psychologist 41 would be needed to fill those vacated jobs, giving a total of 110 psychologists needed to fill the new jobs and the jobs vacated by those leaving he field. If this reasoning is correct, then modeling the number of new persons needed to fill psychology jobs requires an annual estimate of the percent of persons leaving the field. As a part of our survey work, we collected information from 1349 individual staff who are currently in behavioral healthcare positions. We asked each of those individuals to indicate whether they planned to stay in their position during the next year. Twenty one percent of staff and six percent of program managers indicated that they do plan to leave their positions. The percentages of persons indicating that they planned to retire or indicating that they planned to leave behavioral healthcare are shown in the table below. Exhibit 4.8: Staff Planned Separation Rates and Program Manager Estimated Separation Rates (1) (2) (3) (4) (5) (6) (7) Position Type Persons answering survey Percent planning on separating within the next year Percent planning to retire Percent planning to leave the field Sum of columns (4) and (5) Estimated separation rates from program manager surveys Aid/tech 343 21% 0% 6% 6% 42% Masters-level professional 317 19% 1% 3% 4% 26% LPN 37 24% 3% 8% 11% 35% Psychiatrist/ physician8 - - - - - 23% Psychologist 28 21% 4% 0% 4% 11% RN 149 19% 1% 5% 6% 26% Total 874 20%9 1% 4% 5% 35% Note that, with the exception of psychologists, program manager-reported, actual separation rates exceed staff self-reported plans to separate by a wide margin. However, these data can be considered together to estimate a range of possible industry departure ranges. At the conservative end of the range is the staff self-report: an overall rate of 5%, with position-type specific rates ranging from 4% to 11% that is probably a conservative estimate of rates of person who separate leaving the field. Alternatively, the proportion of planned industry departures can be applied to the program manager-reported separation rates for a less conservative estimate. These range for a low of 11% for to a high of 35% for the Aid/Tech positions. These rates are higher than the annual growth rates projected by EMSI. This means that the growth in estimates 8 There is insufficient data for psychiatrists to provide these estimates. 9 Note that this is very slightly lower than the rate cited in the text above (21%). The information in this table is based only on responses that could be linked to a position type (N=877), while the overall number cited in the text is based on all the responses to this item that were received (N=965). 42 of persons needed to fill positions year by year will be influenced to a much greater degree by estimates of staff turnover, representing the need to replace existing members of the workforce. There is one additional consideration in estimating the numbers of new persons needed to fill positions. The “jobs” in the EMSI estimates are only filled jobs; they do not include unfilled jobs. Thus we also need to take into account vacancy rates for the appropriate position type. We do have estimates of vacancy rates for Oklahoma for each position type, as shown in the table below. Exhibit 4.9: Vacancy Rate by Position Category Position Type Percent Vacant Aid/tech 9% Masters-level professional 17% LPN 10% Psychiatrist/ physician 10% Psychologist 6% RN 14% Total 12% With this term included, for each position type the number of new persons needed to fill existing vacancies, positions vacated by persons leaving the field and by growth (from EMSI) would be given by the following equation: [new persons needed in year i+1] = [number of jobs in year i] x [growth rate + percent leaving field] x [1- vacancy rate] Findings Findings are presented for three position types, psychologists, mental health and/or substance abuse counselors, and mental health and/or substance abuse aides/techs. The latter are direct care positions that do not necessarily require professional degrees or licensure. We have not included registered nurses and licensed practical nurses because only a relatively small proportion of these positions are in behavioral healthcare and separate estimates of need have been developed by the Oklahoma Healthcare Workforce Commission. The need for psychiatrists and other prescribers is discussed earlier in this section. The “net growth” figures in the column to the right show the numbers of additional persons who must either be trained or imported over a ten year period to be sure that the estimated behavioral healthcare positions for 2018 and in the intervening years will be filled. This is 1,808 psychologists, 7,045 mental health and substance abuse counselors, and 51,625 aides or techs. 43 Exhibit 4.10: Net Growth by Position Category Position Category State 2008 Jobs 2018 Jobs 2018 Persons Needed Net Growth Psychologists 2,738 3,099 4,546 1,808 Mental Health or Substance Abuse Counselors 9,726 11,377 16,771 7,045 Mental Health or Substance Abuse Aides/Assistants/ Technicians 44,546 54,536 96,171 51,625 We consider that the estimates shown in the table above are conservative for the following reasons: ��� We employed the separation rates estimated from individual reports of intentions to leave their present positions and the field, rather than the separation rates estimated from program manager reports of persons leaving their positions. The latter are two to four times higher than the former. • The number of positions only includes those who are considered “state covered”. If we include all positions, including persons in individual or small group private practices, the numbers would also be higher. This is particularly true for psychologists who are much more present outside the public sector than inside. Exhibit 4.11 below shows the numbers of degrees awarded each year over a six year period beginning in 2001-02 and ending in 2006-07. (A detailed breakdown of degrees awarded in specific fields within each of these larger categories is provided in Appendix A15.) With the exception of psychologists, there has been an increase in each category over this time period. The two columns at the right of the table show the number of degrees expected to be awarded cumulatively from 2007-08 through 2017-18 and the need for new degree recipients to meet the demand for new staff positions. The projections of degrees awarded are based upon a simple linear trend model employing the six years of recent data available on degrees awarded. The model may be underestimating the number of psychology degrees to be awarded, in particular. 44 Exhibit 4.11: Degrees awarded by public higher education institutions 2001-02 to 2006-07 and projected to 2017-18 compared to projected need Position Category 2001-02 Degrees 2002-03 Degrees 2003-04 Degrees 2004-05 Degrees 2005-06 Degrees 2006-07 Degrees Cumulative Degrees Projected through 2017-18 Cumulative Projected New Need by 201810 Psychologist 50 64 51 41 44 44 204 1,808 Mental Health or Substance Abuse Counselors 374 375 391 409 360 421 4,478 7,045 Mental Health or Substance Abuse Aides/Assistants/ Technicians 1,122 1,090 1,129 1,203 1,208 1,262 14,913 51,625 The need for new degree recipients consistently falls short of the projections of new persons needed, calculated by combining estimates of annual position growth from EMSI and positions vacated using conservative estimates of annual separation rates. Thus it includes the number of replacements for people who will leave existing positions over the same period. Conclusions The difficulties experienced by program managers of behavioral healthcare services in recruiting staff to fill vacancies are expected to become more complicated in the coming years. For both professional and non-professional staff the numbers of new persons being trained to account for both persons leaving existing positions and the limited expansion anticipated are not keeping pace with the need, even conservatively estimated. Compensation Earlier in this report we presented survey data on the reasons why programs have high staff turnover rates and difficulties recruiting new staff. Across all position types and across almost all industry groups the single leading explanation is low salaries. We further presented some survey data from individual staff on their salaries. We were able to employ a second source of data on staff salaries by position type, as well as national comparisons, comparisons to surrounding states, and comparisons within areas of Oklahoma. The source of these data is EMSI. Findings Findings are presented for five position types, psychologists, mental health and/or substance abuse counselors, registered nurses, licensed practical nurses, and mental health and/or substance abuse techs. The latter are direct care positions that do not require professional degrees or licensure. Data for psychiatrists is not separately available in the EMSI data set. Data are further presented for the state of Oklahoma overall and for six regions within the State. The Tulsa and Central Oklahoma 10 This estimate does not include individuals needed to replace persons in existing positions who retire or leave the behavioral healthcare system. 45 (Oklahoma City) area are separated from the rest of the state. The rest of the counties are grouped into four quadrants, northeast, northwest, southeast, and southwest. Counties are grouped because data at an individual county level is often too small to provide reliable estimates. Exhibits 4.12 and 4.13 below present comparisons of wages. For all positions wage rates for Oklahoma are consistently below both national and regional averages. However, the disparity between Oklahoma and national wage rates is larger than the disparity with regional wage rates. Within Oklahoma, there is also variation among the six regions. In general, wages are among the highest in the Tulsa area for all position types except psychologists. The Central (Oklahoma City) region also tends to have higher rates than the other regions. Among the four regions with rural counties, there is no region that is consistently among the highest or the lowest. For two positions types there is considerable regional variation. Psychologists range from a high of $31.72 in the Southeast region to a low of $23.66 in Tulsa, a difference of 25%. MH/SA Counselors range from a high of $19.28 in Tulsa to a low of $13.61 in the Northwest, a difference of 33%. All other variations are less than 15%. Exhibit 4.12: Comparison of Average Hourly Wage: National, Regional and Oklahoma Norms for Behavioral Healthcare Positions by Type Rates may vary due to missing values. ♦ Regional rate includes the following states: Arkansas, Colorado, Kansas, Louisiana, Missouri, New Mexico, Oklahoma and Texas. Exhibit 4.13: Comparison of Average Hourly Wage: Oklahoma Norms for Behavioral Healthcare Positions by Type Position State Rate ODMHSAS Region Rates Tulsa OKC NE NW SE SW Psychologists $25.74 $23.66* $29.03 $30.47 $25.23* $31.72 $24.56* MH/SA Counselors $15.12 $19.28 $18.36 $15.67 $13.61* $15.57* $16.10* RNs $24.52 $25.25 $24.96 $22.39 $23.63 $21.34 $24.45 LPNs $15.55 $16.48 $16.06 $14.09 $14.73 $13.70 $14.58 MH/SA Techs $12.35 $12.74* $13.01 $11.37 $11.87* $11.64 $11.97 *Rates may vary due to missing values. Summary It is clear that salary rates for all positions are lower in Oklahoma than in the nation and further that Oklahomans filling these positions providing behavioral healthcare are paid less than Position State Rate National Rate Multi-State Regional Rate* Psychologists $25.74 $30.27 $26.76 MH/SA Counselors $15.12 $18.63 $16.43 RNs $24.52 $30.06 $26.98 LPNs $15.55 $19.51 $17.53 MH/SA Techs $12.35 $14.02 $12.94 46 individuals in all of the surrounding states. There is also some variation within the State. For the two position types that have the largest numbers of persons providing behavioral healthcare, MH/SA Counselors and MH/SA Techs, salaries are higher in the Oklahoma City and Tulsa areas than they are in the more rural northeast, northwest, southeast, and southwest quadrants of the state. Overview of current and future needs for behavioral healthcare workforce As we have indicated elsewhere in this report, behavioral healthcare programs have difficulty retaining and recruiting staff. There is a very large gap in the need for psychiatrists and other prescribers. It is estimated that there is a need for 697 prescribers and only 287 professionals available to meet the need, a difference of 410. While the unmet needs for other professionals and non-professionals are not as large proportionately, there are gaps in these position types as well. Additionally, the rates at which institutions of higher education in Oklahoma are producing new graduates with appropriate training are not sufficient to meet these needs, particularly with projected future growth of these positions. Furthermore, attracting new individuals into service or training is significantly handicapped by the fact that salaries for both professional and nonprofessional positions in Oklahoma are consistently lower than the surrounding states and the nation, as a whole. 47 BENEFITS & COMPENSATION Information on organizational benefits was collected via the organizational survey. Describe
Object Description
Description
Title | Oklahoma Behavioral Workforce Study Statewide 2-16-11 |
OkDocs Class# | M1400.3 S797b/r 2011 |
Digital Format | PDF, Adobe Reader required |
ODL electronic copy | Downloaded from agency website: www.ok.gov/.../Oklahoma%20Behavioral%20Workforce%20Study%20Statewide%202-16-... |
Rights and Permissions | This Oklahoma state government publication is provided for educational purposes under U.S. copyright law. Other usage requires permission of copyright holders. |
Language | English |
Full text | report oklahoma statewide workforce behavioral health 1 OKLAHOMA BEHAVIORAL HEALTHCARE WORKFORCE STUDY: DRAFT STATEWIDE REPORT • Separation Rates & Staff Intention to Leave • Vacancy Rates • Organizational Benefits & Staff Pay Rates • Staff Work Experience & Job Satisfaction Oklahoma Behavioral Healthcare Workforce Study: Statewide Report February 14, 2011 This research was supported by the Mental Health Transformation State Incentive Grant, 5U79SM057411 from the U.S Department of Health and Human Services, Substance Abuse and Mental Health Services Administration. The views expressed do not necessarily reflect the official policies of the Department of Health and Human Services; nor does mention of trade names, commercial practices or organizations imply endorsement by the U.S. Government. Contributors include: John Hornik, Ph.D., Jen Carpenter, Ph.D., Jeanine Hanna, Ph.D., and Nicholas Huntington, M.A., Advocates for Human Potential, Albany, NY and Karen Frensley, LMFT, David Wright, Ph.D. & Lorrie Byrum, M.A., Oklahoma Dept. of Mental Health and Substance Abuse Services, Oklahoma City, OK. 2 Oklahoma Behavioral Healthcare Workforce Study Statewide Report TABLE OF CONTENTS EXECUTIVE SUMMARY 3 INTRODUCTION 11 STAFF SEPARATIONS 16 VACANCIES AND STAFF RECRUITMENT BARRIERS 27 CURRENT AND FUTURE STAFFING NEEDS 34 BENEFITS & COMPENSATION 47 STAFF WORK EXPERIENCE AND JOB SATISFACTION 57 WORKFORCE CAPACITY 66 REPRESENTATION OF CONSUMERS AND THEIR FAMILY MEMBERS IN THE WORKFORCE 78 DISCUSSION AND RECOMMENDATIONS 85 REFERENCES 89 APPENDIX A: STAFF SEPARATIONS 90 APPENDIX B: VACANCIES AND RECRUITMENT BARRIERS 103 APPENDIX C: BENEFITS AND COMPENSATION 105 APPENDIX D: WORK EXPERIENCE AND JOB SATISFACTION 107 APPENDIX E: WORKFORCE CAPACITY 110 3 EXECUTIVE SUMMARY The Oklahoma Behavioral Healthcare Workforce Survey and associated studies were conducted by Advocates for Human Potential, Inc. (AHP) through a contract with the Oklahoma Department of Mental Health and Substance Abuse Services (ODMHSAS) to assist with evaluation activities related to Oklahoma’s behavioral health transformation initiative. The studies were developed and implemented under the guidance of an advisory group, the Workforce Study Team, which was convened through the Governor’s Transformation Advisory Board (GTAB) Workforce Committee, as part of the Transformation initiative. The primary goals of the studies were to: 1. Respond to interests of GTAB Workforce Committee convened through Oklahoma’s behavioral health transformation initiative. 2. Develop behavioral health complement to information gathered through Oklahoma Health Care Workforce Center and Oklahoma Hospital Association surveys. 3. Provide information that can be used for provider organization and state agency-level planning and advocacy. The largest of these studies was the Oklahoma Behavioral Healthcare Workforce Survey, a statewide survey that focused on staffing of agencies and programs that provide behavioral healthcare. The survey was designed with three components: an organizational survey focusing primarily on organizational accreditation and benefits as well as basic information on organizational structure; a program manager survey containing items related to program staffing, vacancy, recruitment barriers, causes of staff turnover, program and staff capacity and training needs; and a staff survey focusing on staff work experience, job satisfaction, education and training as well as demographic characteristics and status as current or prior consumers or family members of consumers. Data collection and process was structured so that the three components could be linked, and organizations were recruited in industry groups, generally according to state agency funding and oversight. While the workforce survey is the largest component of this project and is generally the focus of this report, additional resources used include: Economic Modeling Systems Inc (EMSI) data provided by the Oklahoma Department of Commerce, data drawn from a University of North Carolina (UNC) staffing needs study, and information on historical and anticipated behavioral healthcare-related degree completion rates from the Oklahoma State Regents of Higher Education. Taken together, these resources and the workforce survey are used to address the following topic areas: Staff Separations Information related to separations was gathered through program manager reports of the perceived causes of separation in their programs, program managers’ reports of their programs’ separation rate over the previous year, and staff reports of their intention to leave their position within the next year. Consistent with the findings related to recruitment barriers, the most frequently cited barrier was dissatisfaction with pay, which was cited by nearly two thirds of program managers. Excessive paperwork, emotional burnout and excessive on-the-job stress were cited by at least one third of program managers. While program and organization 4 characteristics were related to multiple perceived causes of turnover when the relationships were examined individually, generally only one or two characteristics remained significant in each logistic regression model. Organizational industry was a significant predictor of citing dissatisfaction of pay, with OPHA program managers being the least likely to cite pay as a cause of turnover. Population age and program setting were significant predictors of perceiving paperwork to be a cause of turnover, with program managers in programs serving children citing paperwork more frequently than those serving both children and adults, and program managers in outpatient settings citing paperwork more frequently than program managers in other settings. Service population also related to citation of dissatisfaction with job responsibilities, with program managers from programs serving both children and adults being less likely to cite this as a barrier than program managers from programs serving either adults or children. Program separation rates ranged from 0% to 200%, and the median of 25% was used to divide programs into two categories: low separation and high separation. These categories were related to multiple program and organizational characteristics when the relationships were examined individually, but only two characteristics remained significant in the logistic regression model. High separation programs proved to be more likely to have a high proportion of techs on staff, and less likely to be state operated. The position type results are consistent with existing literature regarding the relationship between lower staff experience/job level and higher separation rates. The vast majority (80%) of staff did not report intending to leave their positions within the 12 month period following the survey. Intention to leave was related to a range of program, organizational and staff characteristics when the relationships were examined individually, but only two remained significant in the logistic regression model. As would be expected, staff intending to leave reported lower satisfaction with their job overall. Staff age was also related to intention to leave, with the mean age for staff intending to leave being about three and a half years younger than that of staff intending to stay. Both of these findings are consistent with the literature on staff intention to leave. Vacancies and Staff Recruitment Barriers Information related to vacancies was gathered through program managers’ reports of the perceived recruitment barriers in their programs, and their reports of their programs’ current vacancies. By far the most frequently cited barrier was salary, which was cited by 57% of program managers. Lack of candidates with desired credentials or desired work experience, small applicant pool due to geographic location, and competition from other fields were all cited by more than one quarter of program managers. Program and organization characteristics that were related to multiple perceived barriers included organizational industry, state operation, organizational size, and geographic region. Salary as a perceived barrier was related to three of these characteristics when the relationships were tested individually. When examined simultaneously, salary remained significantly related to industry, with OPHA program managers being significantly less likely to cite salary as a barrier. Likewise, state operation and salary were related, with program managers in state operated organizations more likely to cite salary as a barrier. Program vacancy rates ranged from 0% to 100%, and the median of 4% was used to divide programs into two categories: low vacancy (less than 4%) and high vacancy (greater than 4%). These categories proved to be unrelated to most of the program and organizational characteristic variables. Staffing patterns offered one exception: The mean proportion of RNs in 5 low vacancy programs was slightly but significantly lower than the mean proportion of RNs in high vacancy programs, which could be in part related to the comparatively high rate of vacancies in RN positions, across programs. Current and Future Staffing Needs The purpose of this chapter is to identify unmet needs for the behavioral healthcare workforce with a focus on type of position. The first section focuses on psychiatrists and other prescribers. The second section focuses on other professional and non-professional staff. Each of these sections employs data from different sources so the methods upon which we have relied are described within each section, as well as the implications for higher education. The third section describes one underlying problem, the level of compensation currently available to the Oklahoma workforce. Among the studies that we identified was a study of the relative unmet need for professional mental health workers in the State of Washington (Morrissey, et al, 2007a), undertaken as a part of their Mental Health Transformation State Incentive Grant. The study of Washington State was a part of a larger, national study sponsored by the Health Resources Services Administration (HRSA) of the U.S. Department of Health and Human Services. This allowed Morrissey and his colleagues to develop estimates of professional shortages for every county in the U.S. These estimates for Oklahoma demonstrate an unequivocal need for more prescribing professionals in all areas of the state. The area of the State with the greatest unmet need is the Northeast quadrant, excluding Tulsa which has the smallest, relative unmet need. To examine need for other professional staff, we drew on data supplied by the Oklahoma Department of Commerce, using Economic Modeling Systems, Inc (EMSI). EMSI uses several different databases, including population projections from the census bureau, employment trends and participation rates from the Bureau of Labor Statistics (BLS), IRS income and migration data, and industry trends, legislation, and several other factors used to decipher which industries will be growing. Examining EMSI data alongside current vacancy rates and degree attainment trends, revealed that there are unmet staffing needs among both non-prescribing professionals and nonprofessionals as well, and the rates at which institutions of higher education in Oklahoma are producing new graduates with appropriate training are not sufficient to meet these needs, particularly with projected future growth of these positions. EMSI data were also used to examine salary rates across Oklahoma, and in comparison to regional and national rates. It is clear that salary rates for all positions are lower in Oklahoma than in the nation and further that Oklahomans filling these positions providing behavioral healthcare are paid less than individuals in all of the surrounding states. There is also some variation within the State. For the two position types that have the largest numbers of persons providing behavioral healthcare, MH/SA Counselors and MH/SA Techs, salaries are higher in the Oklahoma City and Tulsa areas than they are in the more rural northeast, northwest, southeast, and southwest quadrants of the state. Benefits and Compensation Information on benefits and compensation was collected through the organizational survey and the staff survey. Nearly all privately-operated organizations report providing health insurance, but the provision rate for other benefits deviates from the benefit packages provided by state- 6 operated organizations. Staff report high rates of satisfaction with paid leave, but more moderate rates of satisfaction with other benefits. Staff satisfaction with benefits varies by proportion of health insurance covered and by industry group, with industry groups composed primarily or exclusively of state-operated organizations showing higher rates of staff satisfaction with benefits. Staff reported a wide range of pay rates, but over half the responses were clustered in the lower two pay categories (less than $10.00 per hour and $10.00 - $14.99 per hour), with nearly one in five staff reporting pay of less than $10.00 per hour. Staff earning towards the upper end of the range are at roughly 185% of the 2009/2010 poverty guidelines if they have no dependents, but are under the poverty line if they have more than two dependents. Staff earning closer to minimum wage are at roughly 133% of the 2009/2010 poverty guidelines if they have no dependents, but are under the poverty line if they have any dependents (Office of the Assistant Secretary for Planning and Evaluation, 2010). Position type is strongly tied to pay rate, with techs earning an average of $11.23, less than half the average hourly wage of psychologists ($28.33) and RN’s ($26.71). While staff pay is related to a number of program and staff variables when these relationships are examined individually, only four remained significant when tested simultaneously: position type, program service type, consumer population age, and organization size. The relationship of these last three variables to pay is suspected to be caused in part by other variables, including position type. Given the key role that position type plays in staff pay rates, the remaining staff variables were tested as predictors of position type. Staff race, gender, age and highest degree obtained all predict position type, which in turn predicts staff pay. Staff Work Experience and Job Satisfaction Information on staff satisfaction and work experience was collected through two separate sets of questions in the staff survey. Most of the staff work experience items elicited positive responses from the majority of participants, with nearly all (95%) staff agreeing with the statement - I like the kind of work I do. A singe item - I recommend my organization as a good place to work - was used as an indicator of overall work experience for analysis with other variables. Of the staff and program variables considered, two proved to be significant predictors of work: Work experience was related to industry group, with the highest proportions of staff agreeing with the indicator item being those associated with the Child Guidance (89%) and Substance Abuse (86%) industries. Additionally, staff from programs serving adults were significantly less likely to endorse the indicator item than were staff in programs serving both adults and children (70% versus 80%). Staff satisfaction was measured through a separate set of items. Many of these items also received largely positive responses, with 84% of staff indicating that they were satisfied with their jobs overall, and more than 70% expressing satisfaction with their work schedules, the location of their workplaces, and their organizations overall. The lowest rates of satisfaction were related to the opportunity for advancement (41%) and pay (47%). Responses to these and other items suggest that program manager perceptions of the causes of turnover may be well founded, to the degree that staff satisfaction relates to turnover. 7 Given the importance of pay in both staff satisfaction and program manager perceptions of turnover and recruitment barriers, we examined the relationship of this item to a range of program and staff variables. Industry, service population, service type and years working in the field predicted satisfaction with pay. Staff in industries with a high proportion of state-operated organizations and with a high proportion of Masters-level staff (Child Guidance and DOC) expressed greater satisfaction with their pay, as did staff in programs serving both adults and children (as opposed to just adults, or just children), staff in programs providing substance abuse services only, and staff who reported greater tenure in the behavioral healthcare field. The job satisfaction items were then used to create a scale representing the proportion of job characteristics found satisfactory. Service type and consumer population age also proved to be a significant predictor of this score, with staff in substance abuse programs reporting satisfaction with a greater proportion of job characteristics than staff in mental health programs (74% versus 58%), and staff in programs serving both children and adults reporting satisfaction with more aspects of their jobs than did staff in programs serving only adults (66% versus 60%). Additionally, staff in both small and medium-sized organizations reported satisfaction with more aspects of their jobs than did staff from large organizations, a finding that may be related to the distribution of industry groups across organizational size. The regression also pointed to the significance of two separate demographic characteristics - Black/African American race and high school education - in predicting satisfaction with a greater proportion of job characteristics. It is important to note that these findings are not echoed in the work experience analysis, and these characteristic did not predict higher satisfaction with most of the more global scale items (organization overall, pay, and job overall). Workforce Capacity Information on workforce capacity and training needs was collected through the program manager and staff surveys. According to program manager reports, the three types of training most needed by staff are: (1) knowing about consumers’ psychiatric medications and their side effects, (2) communication skills and (3) educating consumers’ family members about subjects related to mental health or substance abuse. Bivariate analyses demonstrate that program managers from the Substance Abuse and DHS industry groups are most likely to report staff needing training related to consumers’ psychiatric medications, while 30% and 36% of staff from the OJA and OPHA industries require additional training on the role of peers as service providers. Some specific training needs also varied by program service type. In addition to basic training it is important that new professional staff have the capacity to provide evidence-based practices for adults and children. Over 65% of new professional hires are prepared to provide Cognitive Behavioral Therapy (CBT) for adults and CBT for depression, anxiety, and trauma for children. Since education about psychiatric medications was identified as one of the types of training most needed for direct care staff, it is not surprising that only 37% of new professional hires can provide the EBP, Medication Management. Staff capacity to provide the EBP, consumer-run services, was also low (35%). Data on organizational/program cultural competency were gathered based on staff perceptions of whether (1) their workplace has an attitude of acceptance of people from different cultural backgrounds; (2) their organization does a good job recruiting and retaining employees of different cultures; (3) sensitivity to diversity is an important part of supervision/team meetings; 8 (4) staff are encouraged to attend diversity training; and (5) cultural assessment is used to plan effective treatment and service delivery. The majority of staff surveyed agreed or strongly agreed with all five items. Bivariate analyses explored the relationship between the aforementioned cultural competency items and several program, organization, and staff variables. Staff perceptions of how well their organization recruits and retains employees of various cultures varied by industry, region, and staff ethnicity and highest degree earned; these relationships were significant in both the bivariate and regression analyses, when we controlled for other variables. Interestingly, staff identifying as Hispanic/Latino were more likely to report their organization recruits and retains employees of diverse cultures than non-Hispanic/Latinos. This finding may be related to another finding: staff working in the Oklahoma City metro area are most likely to agree with the aforementioned cultural competency item. This region of the state had the highest response rate and the greatest racial/ethnic diversity among staff working there. Program managers were asked to report the cultural and linguistic capacity of their programs (i.e., does it hold cultural competence training and/or provide services in Spanish or American Sign Language). Comparisons were made between program manager reports of program linguistic capacity and the self-reported fluency of staff and program managers. The vast majority of programs (78%) hold some type of cultural competency training for staff. Although 22% of program managers report that their program can provide services in Spanish, only about 3% of staff and managers reported that they are fluent in Spanish, which is less than 5%, the state average. Different interpretations of what it means to “provide services in Spanish” may account for some of the discrepancy in self-reported (staff and program managers) and program linguistic capacity. Representation of Consumers and Their Family Members in the Workforce Information on consumer and family member representation and disclosure was obtained through the staff and program manager surveys. The most important finding is that a significant proportion of the behavioral healthcare workforce that identifies themselves as adult consumers (21%) and an even larger proportion that identify themselves as family members of consumers (32%). Consumer and family member representation was generally higher among program managers than staff, and was higher for adult consumer and family members of an adult consumer categories than for former youth consumer and family of a youth consumer categories. With the exception of the youth consumer category, representation of all categories exceeded 10% for both program managers and staff, and reached a high of 37% representation of family members of adult consumers among program managers. Both representation and disclosure varied significantly by industry group. Adult consumer and family member representation was highest in the Substance Abuse and DOC industry groups, and lower in the OPHA, OJA, and Child Guidance industry groups, although Child Guidance had the greatest proportion of staff who identified as family members of youth consumers. Over three-quarters of Substance Abuse staff who identified as consumers report having disclosed this status in the workplace, compared to just over half of OPHA and DOC staff members who identified as consumers. Among staff who identified as family members, nearly three-quarters disclosed this status, while just over half of OPHA staff disclosed. 9 The analysis considered a wide range of possible predictors of both consumer status and family member status among responding staff. While many of these were initially found to be significantly related to one or both outcome variables, few remained significant when logistic regression models were used to test the relationships simultaneously. Staff working in programs serving people with substance abuse or substance abuse and mental health needs were significantly more likely to identify as consumers than were staff working in programs serving people dually diagnosed with developmental disabilities and behavioral health needs. Also, those working in outpatient programs were significantly more likely to identify as consumers than were those working in inpatient programs. Respondent education level was the only variable remaining significant in the family member representation model, with staff who reported having a Masters degree or higher being significantly more likely to identify as family members than were staff with high school diplomas or GEDs. Among staff and program managers who identify as either consumers or family members, rates of disclosure in the workplace are high. A higher proportion of program managers reported disclosing their status. For both consumer and family member status, roughly 80% of responding program managers report disclosing on the job, while roughly 66% of staff report having disclosed. The analysis also considered multiple potential predictors of staff disclosure of consumer or family member status. As with the previous analysis, many of these were related to consumer or family status in initial analysis, but did not remain related in the subsequent logistic regression models. Respondent race and type of service used proved to be significantly related to disclosure of consumer status, with White staff more likely to have disclosed than Black staff, and with staff who reported receiving both mental health and substance abuse services more likely to disclose than staff receiving either mental health or substance abuse services. It is interesting to note that while there is no significant relationship between staff member consumer status and race, among those who do identify as consumers and family members, White staff members are more likely to disclose this status in the workplace than are Black staff members. A similar pattern was noted for disclosure of family member status. Program industry group was also found to be a significant predictor of disclosure of family member status, with respondents working in the Mental Health and Substance Abuse programs significantly more likely to have disclosed their status on the job than were respondents from the OPHA programs. Discussion and Recommendations Following review of an earlier draft of this report, Workforce Study Team members were asked to make recommendations in response to the study findings. These recommendations were grouped into five topic areas: compensation, recruitment and retention, training, best practices, and future planning efforts. Regarding compensation, the Workforce Study Team recommended the prioritization of overall funding for behavioral healthcare services, pointing to the clear need for better compensation. The Team advised that current pay rates are inadequate, and that it is important for the public to become more aware of this inadequacy. Relating to recruitment and retention, the Workforce Study Team found that the report provided evidence that there is dissatisfaction with opportunities for advancement within the behavioral 10 healthcare workforce, with only 41% of staff reporting satisfaction with their opportunity to advance within their organization. The Team advised that this suggests a need for more opportunitites within agencies for positions for possible advancement, and a need to eliminate the barriers that currently make advancement difficult. Additionally, given existing reimbursement strategies, the Team noted a number of challenges in supporting staff working on achieving licensure. The Workforce Study Team’s concerns about training included the insufficient number of prescribers in the state; the need to support the development of basic behavioral health care screening, assessment, treatment, and referral skills among primary medical care providers; and the insufficient “real world” training opportunities for some professions, particularly psychologists who may be trained in settings vastly different from the public behavioral healthcare system. Resources for supporting implementation of Evidence Based Practices (EBPs), requires additional funds to train and assure model fidelity through consultation and supervision. The initial cost of training and consultation for clinicians to treat people with practices that work, should be recouped in the long run since there will be cost savings when individuals recover and no longer need services. The Workforce Study Team identified the implementation of best practices as one way to respond to the study findings related to staff paperwork burden and its relation to job satisfaction and to program manager perceptions of causes of turnover, and pointed to the difficulty in reducing documentation burden given high levels of vacancy and turnover. Additionally, the Team raised telehealth as an important best practice for implementation in Oklahoma to increase access. Finally, with respect to future efforts, Workforce Study Team members identified a need to retain the involvement and commitment of well-positioned personnel in key state agencies and within the private sector, and pointed to the importance of focusing continued work on a vision for the future of behavioral healthcare in the state. The Team recommended the formation of an advisory council to continue in-depth analysis of the state’s workforce issues. 11 INTRODUCTION In 2005, Oklahoma was one of seven states (now nine) to receive a five-year Mental Health Transformation State Incentive Grant (TSIG) from the federal Center for Mental Health Services (CMHS). The purpose of this grant was to help transform state mental health systems from “broken and fragmented” systems to systems that deliver excellent mental health care with a focus on recovery (President’s New Freedom Commission on Mental Health, 2003). A major challenge faced by all states was assuring a stable, competent workforce available to provide needed services. The Oklahoma Behavioral Healthcare Workforce Survey and associated studies were conducted by Advocates for Human Potential, Inc. (AHP) through a contract with the Oklahoma Department of Mental Health and Substance Abuse Services (ODMHSAS) to assist with evaluation activities related to Oklahoma’s behavioral health transformation initiative. The studies were developed and implemented under the guidance of an advisory group, the Workforce Study Team, which was convened through the Governor’s Transformation Advisory Board (GTAB) Workforce Committee, as part of the Transformation initiative. Purpose and Goals State mental health authorities typically do not have empirical information about the characteristics of their current workforce. In order to fill this information gap, we undertook a number of studies, as well as searched for relevant research, that provided useful information for understanding the difficulties faced by staff providing mental health services in Oklahoma. Taken together, the workforce studies were designed with three broad goals in mind: 1. Respond to interests of GTAB Workforce Committee convened through Oklahoma’s behavioral health transformation initiative. 2. Develop behavioral health complement to information gathered through Oklahoma Health Care Workforce Center and Oklahoma Hospital Association surveys. 3. Provide information that can be used for provider organization and state agency-level planning and advocacy. The largest of these studies was the Oklahoma Behavioral Healthcare Workforce Survey, a statewide survey that focused on staffing of agencies and programs that provide behavioral healthcare. The survey itself was intended to address six particular goals of the Workforce Study Team and other project stakeholders, including: 1. Estimate rates of recruitment, retention and turnover by position. 2. Determine reasons for leaving, including those related to wages and benefits (e.g., health insurance, schedule/shift, child care). 3. Analyze current representation of adult peers and family members in the workforce. 4. Describe linguistic (and cultural) competency of the workforce. 12 5. Describe capacity of state workforce to address current needs of clients and employers. 6. Describe current access to behavioral healthcare services in primary care settings and identify (types of) professionals delivering such services. Methodology Survey Measures Where possible, survey items were drawn from established measures. The two primary sources of items and item structure were: • Addition Technology Transfer Center Workforce Survey: A staff and director survey instrument was developed for the Northwest Addiction Technology Transfer Center (ATTC) and subsequently adapted for use in at least six other states. Oklahoma workforce survey items that were drawn from or based on this instrument included those relating to recruitment barriers and causes of turnover, organizational strategies for supporting staff development, and distribution of daily responsibility, as well as a number of basic demographic related items. • Federal Human Capital Survey (FHCS): The FHCS is an instrument developed by the U.S. Office of Personnel Management and used to measure employees’ job satisfaction and their perceptions of the degree to which their organization exhibits characteristics consistent with those of successful organizations. The instrument was used to survey federal employees in 2004, 2006, and 2008, with over 200,000 responses received in the 2008 use alone (United States Office of Personnel Management, n.d.). Oklahoma workforce survey items that were drawn from the FHCS include those related to staff work experience and job satisfaction. Additional items were developed and selected with the guidance of the Workforce Study Team and outside consultation when necessary. Pilot Study The pilot study involved two organizations: a residential care provider which operates congregate care facilities in locations throughout Oklahoma, and an inpatient care provider which operates a variety of behavioral healthcare programs in the Oklahoma City area. Between the two organizations, a total of 28 distinct programs participated in the pilot. These programs provided an array of services designed to respond to a variety of consumer needs and interests. Programs ranged from long-term residential care to acute detoxification, and served children, youth, adults and older adults, and supported people with needs related to mental health, substance abuse and co-occurring disorders. The pilot study took place in June and July, 2008. In August 2008, the preliminary results of the pilot were reviewed with the Workforce Study Team, as was a report of the survey process, including challenges encountered and suggestions offered by pilot participants. Based on these reports and the discussion with the Workforce Study Team, some redundant items were eliminated, the schedule and scope of organizational recruitment was scaled back, and the recruitment material packet was revised. 13 Survey Structure In order to capture the range of information desired by the Workforce Study Team and other project stakeholders, the survey was designed with three components: 1. An organizational survey focusing primarily on organizational accreditation and benefits as well as basic information on organizational structure. Organizational structure information was used to create organization-specific versions of the program manager and staff surveys described below. The organizational survey component was completed by a single member of each participating organization (typically a human resources administrator in larger organizations, or the director in smaller organizations). 2. A program manager survey containing items related to program staffing, vacancy, recruitment barriers, causes of staff turnover, program and staff capacity and training needs. Within each organization, each program manager with unique supervisory responsibilities for one or more behavioral healthcare programs was invited to complete a program manager survey. Occasionally, organizations would indicate that two or more program managers supervised a single program. In these cases, AHP worked with the organization to develop a survey plan to avoid duplication of program manager responses. 3. A staff survey focusing on staff work experience, job satisfaction, education and training as well as demographic characteristics and status as current or prior consumers or family members of consumers. All direct providers of behavioral healthcare services in participating organizations were invited to complete a staff survey. As described in the recruitment subsection below, however, staff recruitment was highly dependent on program manager assistance. Data collection and process was structured so that the three components could be linked. Staff responses could be grouped by program and organization, and linked to the appropriate program data (provided via the program manager survey) and organizational data (provided via the organizational survey). 14 Exhibit 1.1: Survey Structure ORGANIZATIONAL SURVEY PROGRAM MANAGER SURVEY PROGRAM MANAGER SURVEY PROGRAM MANAGER SURVEY STAFF SURVEY STAFF SURVEY STAFF SURVEY Recruitment & Participation Organizations were recruited in industry groups, generally according to state agency funding and oversight. The following nine industry groups were recruited: • Mental Health: Organizations providing primarily mental health services and operated under contract with or by ODMHSAS. • Oklahoma Psychiatric Hospital Association (OPHA): Psychiatric hospitals or hospitals with psychiatric units within OPHA membership. • Oklahoma Department of Human Services (OKDHS): Organizations providing a range of residential and outpatient services for children, youth and adults with a variety of service needs and operated by or under contract with OKDHS. • Oklahoma Office of Juvenile Affairs (OJA): Organizations operated by or under contract with OJA, providing services to children and youth in a range of settings. • Substance Abuse: Organizations providing primarily substance abuse services and operated under contract with or by ODMHSAS. • Oklahoma Department of Corrections (DOC): Providers employed by DOC and offering mental health services within correctional facilities across Oklahoma (substance abuse services are contracted out and were therefore not included in the survey). • Other Medicaid: A random sample of organizations that were not included in any of the above groups but that do provide behavioral healthcare services and bill Medicaid. • Federally Qualified Health Centers (FQHC): Organizations that provide behavioral healthcare services and have obtained the FQHC designation. 15 • Child Guidance: Child Guidance clinics operated by the Oklahoma State Department of Health (OSDH). The number of organizations, program managers, and staff members recruited by industry group are shown in Exhibit 1.2 on the next page. When considering the implications of the results described in this section, it may be helpful to bear in mind the degree to which the responses we received can be considered representative of the views of Oklahoma behavioral healthcare agencies, program managers, and staff. Exhibit 1.2 indicates that 63% of invited organizations responded, with participation rates by industry group ranging from 41% to 100%. We can be relatively confident that responses from agencies in high participation industry groups are representative of those industry groups, but less confident of the representativeness of responses of agencies in low participation industry groups. Similarly, among participating organizations, average program manager response rates ranged from 67% to 100%, with an overall average of 72%. Among participating programs, staff response rates ranged from 4% to 100%, with an overall average of 26%. Our confidence in program manager and staff response representativeness should also vary by industry group participation rate. Additionally, within industry groups or within the sample as a whole, we can have more confidence in the representativeness of program manager responses than we can in the representativeness of staff responses. Finally, it is important to note that, as the recruitment process was driven by state agency oversight and funding, any First Nations provider organizations that are not funded or credentialed by one or more of the above state agencies were not recruited. 16 Exhibit 1.2: Participation by Industry Group Industry Wave Date Launched Organizations Program Managers Direct Care Staff Number of Responses Response Rate Number of Responses Response Rate Number of Responses Response Rate1 Mental Health 9/30/08 27 79% 102 67% 443 21% OK Psychiatric Hospitals Association 11/04/08 12 41% 32 74% 363 26% OK Department of Human Services 1/14/09 10 83% 20 74% 150 31% OK Office of Juvenile Affairs 1/14/09 11 79% 12 86% 38 13% Substance Abuse 5/14/09 38 62% 52 74% 234 36% Department of Corrections 8/17/09 12 100% 6 100% 40 63% Other Medicaid Providers 8/19/09 11 48% 9 82% 6 4% Federally Qualified Health Centers 8/19/09 5 45% 2 67% 14 100% Child Guidance Clinics 10/26/09 12 100% 8 89% 37 73% Total: 116 63% 243 72% 1325 26% 1 Staff participation rates are based on programs for which total number of staff is known. 2 The Department of Corrections and Child Guidance Clinics are multiple service sites however due to the nature of the programs they were surveyed as one organization. 17 At the beginning of the recruitment phase for each industry, enrollment packets were mailed to the organizations that had been identified for recruitment. These packets included a cover letter from the relevant state agency administrator, describing the value of the project and encouraging the organization to participate. Following this cover letter were informational sheets from AHP about the purpose of the survey and the enrollment process. A single organizational designee completed the organizational survey component online, providing program manager names and email addresses. Organizations that did not initially respond were encouraged to do so via email, telephone, and U.S. mail reminders, which included sample reports that served as an organizational incentive. Once an organization completed the organizational component of the survey, a unique version of the program manager and staff survey was created to reflect the structure of the organization. Program managers were mailed invitational emails with recruitment letters as attachments to be distributed to staff. Regular reminders were sent to program managers, including counts of staff responses for each program, which were copied to the organizational designee and/or executive director. A variety of additional measures were employed to encourage participation at each stage of the survey. For most industries, personnel from the relevant Oklahoma state agency made additional follow-up calls. Additionally, AHP staff made in-person visits to key organizations to provide assistance in participating in the survey, or to encourage participation. Other Data Sources While the workforce survey is the largest component of this project and is generally the focus of this report, data were drawn from a variety of additional sources: • Economic Modeling Systems Inc (EMSI): The Oklahoma Department of Commerce provided average hourly wage rate norms for a range of behavioral healthcare positions at the national, regional and state level. • University of North Carolina (UNC) Staffing Needs Study: Data were drawn from a UNC study of professional staffing shortages, conducted under contract to the Health Resources and Services Administration (HRSA). • Oklahoma State Regents of Higher Education: The Regents of Higher Education provided information on the number of behavioral healthcare related degrees awarded by category and by year since 2001, as well as information on the number of degrees anticipated to be granted and anticipated to be needed. The data derived from these sources complement the data collected from the survey and provide information on subjects that could not be covered by the survey. In doing so, they allow the project to provide a more comprehensive response to the Workforce Study Team’s interests and goals. 16 STAFF SEPARATIONS Staff separation rate (turnover) is a near-universal concern in behavioral healthcare programs. High separation rates increase program costs, reduce return on investment for staff development, and impact quality of care. Anecdotal evidence of the negative impact of turnover on provider-consumer relationships abounds. Given this, it is not surprising that study stakeholders identified staff separation as a principal area for investigation. Information was gathered on staff separations through both the program manager and staff surveys. Program managers were asked to review a list of 18 possible causes of staff turnover and were then asked to indicate which of these were most relevant to their program. Managers were also asked to report on the number of separations within the last year in their program using the study’s six primary position categories. Staff members were asked to report whether they intended to leave their position within the next 12 months. This section will describe the data received from program managers and staff in response to these survey items, and the analysis conducted to investigate relationships between these items and other program, organizational and staff characteristics will be discussed. Program Manager Perceptions of Causes of Turnover Program managers were asked to identify three causes of staff turnover in their programs. The causes cited by 235 programs are shown in Exhibit 2.1, those causes cited by fewer than 10% of program managers are not shown in the exhibit.3 Percentages for this item add up to more than 100, as three causes of turnover were selected for each program. Program managers perceive dissatisfaction with salary/pay as the greatest contributor to staff separations; 63% (from all industry groups) cited dissatisfaction with pay as a significant cause of turnover in the behavioral healthcare field. Other factors contributing to turnover, cited by at least one third of the program managers, were excessive paperwork (43%), emotional burnout (36%) and excessive on-the-job stress (33%). 3The following potential causes of turnover were listed as options on the survey, but were cited by fewer than 10% of program managers: dissatisfaction with workplace location; dissatisfaction with relationship with supervisor; dissatisfaction with on-call responsibilities; difficulties with transportation; difficulties with child care; dissatisfaction with health insurance; dissatisfaction with time off; concern about on-the-job safety; and dissatisfaction with coworkers. 17 Exhibit 2.1: Program Manager Perceptions of Causes of Turnover Across Industries Data from the program manager surveys. We examined the relationships of the perceived causes of staff turnover, cited by no fewer than 10% of program managers, to seven key dimensions – industry group (Mental Health, Substance Abuse, Department of Human Services, Office of Juvenile Justice, Oklahoma Psychiatric Hospital Association, Child Guidance, Federally Qualified Health Centers, Other Medicaid and the Department of Corrections)4, region (northwest, southwest, northeast, southeast, Tulsa metro, Oklahoma City metro), service type (mental health, substance abuse, combined mental health and substance abuse, and services for people with developmental disabilities and mental health or substance abuse needs), program setting (inpatient, criminal justice, residential, or outpatient), service population (children, adults, both), organizational type (state vs. private), and organizational size (small, medium, large). The following causes of turnover were significantly different (p<.05) across at least one of the seven dimensions: (1) dissatisfaction with salary/pay (Salary), (2) dissatisfaction with career ladder, (3) excessive paperwork (Paperwork), (4) dissatisfaction with job responsibilities (Responsibilities) and (5) dissatisfaction with shift/work hours (Hours). While none of these causes of turnover varied by region or service type, there was variation across industry group, program setting, service population, organizational size, and organizational operation (state vs. private), also considered a proxy for organizational benefits. Following these findings, logistic regressions were performed to examine the relationship between the dimensions - taken together - and each of the following four causes of turnover: Salary, Responsibilities, Hours and Paperwork. Industry, service population, organizational type, program setting, and organizational size were included in this testing. Tables summarizing the results of these regressions can be found in Appendix A1. Four additional parsimonious 4 Industry group name and abbreviation: Mental Health (CMHC), Substance Abuse, Department of Human Services (DHS), Office of Juvenile Justice (OJA), Oklahoma Psychiatric Hospital Association (OPHA), Child Guidance, Federally Qualified Health Centers (FQHC), Other Medicaid (MA) and the Department of Corrections (DOC). 18 logistic regression models can be found in Appendix A1 as well, for a total of eight regression models. In summary, when controlling for other factors, program manager perceptions of causes of staff turnover suggest that: 1. The role of salary/pay in turnover varies by industry. 2. The role of excessive paperwork and dissatisfaction with job responsibilities in turnover varies by service populations. 3. The role of excessive paperwork in turnover also varies by program settings. Pay as a Perceived Cause of Turnover Exhibit 2.2 provides details of the relationships between organizational industry and pay as a perceived cause of turnover. Industries with fewer than ten program manager responses were not included in the analysis. Program managers in OJA organizations were most likely to cite pay as a cause of turnover, while those in OPHA organizations were least likely to do so. Specifically, 90% of program managers from the OJA industry group perceived staff dissatisfaction with salary/pay as one of the top reasons for staff separations while program managers from the OPHA industry group were only half as likely to name dissatisfaction with salary/pay. At least 70% of program managers from the Mental Health and DHS industry groups cited salary/pay as a cause of turnover. This relationship was upheld in the regression analyses as well with industry being a significant predictor of program manager perceptions of pay as a significant cause of turnover. Program setting was not significant in the logistic regression model. Organizational size5 and organizational operation were significant when these relationships were considered individually, but did not remain significant when multiple relationships were tested simultaneously. Exhibit 2.2: PM Perceptions of Pay as a Cause of Staff Turnover by Industry CMHC N=102 DHS N=17 OJA N=10 OPHA N=26 SA N=61 Dissatisfaction with salary/pay 76% 71% 90% 42% 53% Data are significant at the p<.05 level. ♦ Data from the program manager surveys. ♦ FQHC, DOC, Other Medicaid, and Child Guidance industries are not included in the analysis because there were fewer than ten programs in these samples. Excessive Paperwork as a Perceived Cause of Turnover Exhibit 2.3 shows the relationship between program manager perception of excessive paperwork as a cause of staff turnover and program setting. Program settings were defined as follows: Inpatient – an acute care mental health unit in a hospital, a unit in a substance abuse detoxification facility, or a residential unit within a hospital; Outpatient – a unit in a community mental health center, a day program, a psychiatric rehabilitation (PSR) program or a Program of Assertive Community Treatment (PACT)/case management program; Residential (not hospital-based) – a group home or a supported housing program; and Correctional/Criminal Justice – a 5 Organizational size – programs are the unit of analysis. Program managers were asked to identify the number of full-time staff working in each program they supervised. The number of full-time staff were aggregated for each organization. An organizational response rate was calculated and the total number of staff in each organization was divided by the organizational response rate and multiplied by 100. This yielded the total number of full-time staff in each organization (i.e., total staff) which was then divided into three groups – small, medium and large organizations – based on the overall distribution of the total staff. 19 prison or a juvenile detention facility. Excessive paperwork is cited as a cause of separations by 60% of program managers from outpatient facilities, followed by those in residential (21%), inpatient (20%) and criminal justice facilities (10%). The relationship between program setting and excessive paperwork remains when the effects of other variables are considered. Although, industry group and excessive paperwork had a strong relationship when looking at the two variables in isolation, the former is no longer a predictor of excessive paperwork when multiple relationships were tested simultaneously. On the other hand, service setting had a different effect on excessive paperwork: there was no relationship between service setting and paperwork when considered alone, but it becomes a significant predictor of paperwork when multiple relationships were tested (Model 2 of the logistic regressions). Program managers in programs serving children cite excessive paperwork as a cause of turnover more frequently than those serving both children and adults. Exhibit 2.3: PM Perceptions of Paperwork as a Cause of Staff Turnover by Program Setting Inpatient N=30 Outpatient N=119 Residential N=47 Correctional N=10 Excessive paperwork 20% 60% 21% 10% Data are significant at the p<.05 level. ♦ Data from the program manager surveys. Dissatisfaction with Job Responsibilities as a Perceived Cause of Turnover Although dissatisfaction with job responsibilities varies by service population (Exhibit 2.4), program managers supervising programs serving both children and adults are far less likely (4%) to perceive job responsibilities as one of the most important causes of staff turnover. In other words, programs serving adults only and children only are more likely to have staff dissatisfied with their job responsibilities, 21 and 22% respectively. While this relationship may not initially seem meaningful, it could be related to the relationship between service population and program setting. Eighty percent of programs serving both children and adults are categorized as outpatient programs. Compared to program managers in inpatient and residential programs, fewer outpatient program managers cite job responsibilities as a significant cause of turnover in their programs. The relationship between job responsibilities and service population is further supported by model 3 of the logistic regressions (see Appendix A1 - Factors Influencing Program Manager Perceptions of Staff Dissatisfaction with Job Responsibilities as a Cause of Turnover). Organizational size was not significant in the regression model. Although dissatisfaction with job responsibilities varied by industry, program setting, and service population, these were not significant predictors in the full regression model. Exhibit 2.4: PM Perceptions of Responsibilities as a Cause of Staff Turnover by Service Population Children/Adults Adults Only Children Only Dissatisfaction with job responsibilities 4% 21% 25% Data are significant at the p<.05 level. ♦ Data from the program manager surveys. 20 Program Manager-Reported Separation Rates Program managers were asked to report the current number of full time equivalents (FTEs) budgeted for their program and vacant in their program, as well as the number of staff separations that had occurred over the previous 12 months in their program. These items were posed in reference to each of six position categories: aids/techs/other paraprofessionals, professionals primarily holding Masters degrees (counselors/therapists/MSW-level social workers), LPNs, psychiatrists and other physicians, doctoral-level psychologists/DSW-level social workers, and RNs. This position category structure was developed based on a review of the state position classification and the U.S. Bureau of Labor Statistics Standard Occupational Code (SOC) system. Appendix A15 shows relevant SOC positions categorized according to this six-position structure. To calculate the separation rate for a given region, the number of separations was totaled across participating programs, and this sum was divided by the number of FTEs budgeted across programs. Exhibit 2.5 shows the position-specific and total separation rates statewide, and for each of the six geographic regions. It is important to note that organizations may not have included providers that are contracted with, rather than employed, in the counts that follow. Exhibit 2.5: Cross-industry Program Manager-Reported Separation Rates by Region Position NE NW OKC SE SW Tulsa Statewide Aid/tech 51% 55% 34% 38% 50% 30% 42% Masters-level professional 28% 26% 26% 27% 8% 27% 25% LPN 32% 29% 40% 50% 33% 10% 36% Psychiatrist/ physician 33% 0% 4% 44% 25% 20% 22% Psychologist 13% 0% 0% 0% 0% NA 7% RN 25% 33% 29% 56% 23% 21% 28% Total 40% 41% 31% 35% 32% 27% 34% Data from the program manager surveys. Calculating Program Separation Rate Percents in the table above were calculated by summing separations and budgeted positions across the region. In the analysis that follows, separations are calculated at the program level. Programs, rather than organizations, were chosen as the unit of analysis due to concerns that program characteristics and local program environment may vary widely within larger organizations - particularly those with programs across a wide geographic range. Program separation rates ranged from 0% to 200%. Separation rates of greater than 100% are possible because positions may turn over more than once within a year. The median separation rate was 25%, meaning that roughly half of the participating programs had a separation rate below 25%, and roughly half had a separation rate above 25%. Appendix A2 gives more information on the distribution of the program separation rates. 21 The initial analysis of relationships between separation rates and other program variables was attempted with three approaches to handling separation rates: by breaking participating programs first into two groups of equal size, then into three groups of equal size, and finally into four groups of equal size. The approaches yielded fairly similar results, with those for the two group approach being slightly more favorable than those for the alternatives. This approach involves dividing the group at the median of 25%, a rate which is consistent with a high turnover definition used in a recent, related study (Strolin-Goltzman, 2008). Relationships Between Separation Rates and Other Program Variables The relationship between separation rate and a number of program characteristics and related variables was examined. Relevant, recent literature was reviewed. The following identifying program characteristics were identified as being potentially related to separation rates: 1. Staff role clarity 2. Staff job satisfaction 3. Staff salary and benefits 4. Staff sense of personal accomplishment 5. Staff age 6. Staff intention to leave 7. Staff job level/experience 8. Staff burnout 9. Lack of alternative job options The primary source of information for items 1- 6 is the staff survey. Because of concerns about the representativeness of the staff data, these items were not considered feasible for this analysis. Most of these variables are also established predictors of staff intention to leave, and could therefore be employed in the predictive model of intention to leave (itself the strongest predictor of separation rates, Mor Barak et al., 2001). Staff job level/experience as a program characteristic was measured using the program manager reports of the FTEs budgeted for their programs. As these reports were specific to position type, we were able to create variables reflecting the proportion of each position type within each program’s staffing pattern. Masters-level counselors and techs made up by far the largest proportion of program staff. On average, Masters-level counselors made up 50% of the program staff, and techs made up 39%. The remaining four position categories ranged from a high of 6% (RNs) to a low of 1% (PhDs). Appendix A3 offers more information about the distribution of each of the six position type proportions. Staff burnout as a program characteristic was measured by program manager indication that burnout is one of the top three reasons for staff turnover within their program. We also looked for relationships between the other frequently-cited causes of turnover and separation rate. A proxy for lack of alternative job options was created using the program region code: Programs located in the Tulsa and Oklahoma City metro areas were considered to be located in areas with better alternative job options, while those in the remaining, more rural, regions were coded as being located in areas with fewer job options. 22 Finally, relationships between separation rate and each of the study dimensions described earlier (industry, region, service type, program setting, population age, state operation, and organizational size) were examined. Analysis and Results Analysis to identify relationships between separation rate and each of the variables above on an individual basis was performed. Most of these did not prove to be statistically significant: None of the frequently-cited causes of turnover were associated with program separation rate, nor was the job options proxy. Of the staffing and study dimensions variables, proportion of Masters-level counselors, proportion of techs, industry, and state operation were significantly associated with separation rate, as were two approaches at measuring benefits. Upon closer inspection, the results for the benefits items were difficult to interpret (i.e., suggesting an inconsistent or nonsensical relationship between separation rate and benefits). These items were discarded. Further information about the items and the relationships identified may be found in Appendix A4. Following this analysis, logistic regression was used to investigate whether the relationships between separation rate and program characteristics remained significant when all characteristics were considered simultaneously. Looking at the relationship between the two staffing variables (proportion of techs and proportion of Masters-level counselors) it was determined that these variables were too closely related to include in the regression model. Details of the analysis used to determine this can be found in Appendix A5. Ultimately, the model included the following program characteristics: proportion of techs, industry, and state operation. While proportion of techs and state operation remained significant in the regression model, industry became insignificant, suggesting that the relationship between industry and separation rate may have been in part due to a relationship between industry and state operation, or possibly between industry and staffing patterns. A detailed look at the results of this model can be found in Appendix A6. As shown in Exhibit 2.6, on average techs made up 31% of the staff in low separation programs, while they made up nearly half of the staff in high separation programs. This is consistent with the literature indicating that high staff experience, job level, and pay are associated with lower turnover. Exhibit 2.6: Proportion Techs in Low Separation and High Separation Programs Staff position type predictors Mean proportion low separation programs Mean proportion high separation programs Proportion Techs 31% 48% Data from the program manager surveys. A more detailed look at the relationship between position type and separation is available in Appendix A7. The distribution for programs in state vs. privately operated organizations is also as was anticipated. Half of the programs in private organizations fall into the high turnover group, 23 while less than one-third of the programs in state-operated organizations do. It is believed that this relationship is at least in part a result of the better compensation package offered by state-operated organizations. Exhibit 2.7: Proportion of Programs in High Separation Group by State/Private Operation Operation (assigned) Private (N=188) State (N=56) Total (N=244) Proportion in high turnover group 50% 29% 45% Data from the program manager surveys. While only a proportion of techs and state operation remained significant in the regression model, Appendix A8 offers details on the remaining variables that were tested. Staff Intention to Leave Staff were also asked about their plans to leave their organizations within the next year. Those who reported that they were planning on leaving were asked to indicate whether they planned to retire, find another job within the behavioral healthcare field, find a job outside the field, or pursue some other option. Exhibit 2.8 shows the percentages of program managers and staff reporting each of these plans. Exhibit 2.8: Intention to Leave Frequencies Response (N=1244) % No, don’t intend to leave within a year 80% Yes, to retire 1% Yes, to take another job in behavioral health 7% Yes, to take a job outside behavioral health 4% Yes, other 7% Data from the staff surveys. This variable was recoded into two categories by combining all categories representing any intention to leave (do intend to leave within a year: 20%, and do not intend to leave within a year: 80%) for the analysis that follows. Relationship Between Intention to Leave and Staff and Program Variables As with separation rates, predictor variables were chosen following a review of the literature. This review supported the use of the following variables: 1. Staff burnout; 2. Work-life fit; 3. Job satisfaction; 4. Empowerment; 5. Workplace incivility; 6. Staff age; 7. Job level/experience; 24 8. Professional and job commitment; and 9. Income. Staff burnout was not measured directly by the staff survey. Related items, such as My workplace is too stressful, would have appeared to provide reasonable proxies but more closely matched other predictors examined in and not supported by the literature. The same holds true for workplace incivility and empowerment. The survey did not examine work-life fit or professional/job commitment. The survey’s overall job satisfaction item was chosen as an indicator of job satisfaction. The survey’s staff age variable was transformed into a continuous variable by recoding age categories into midpoints, except for over 64 which was recoded as 69.5, the midpoint between 65 and 74. The survey’s categorical staff income variable was treated in a similar manner, with the following differences: The lowest category (<$10.00/hr) was recoded as the midpoint between $10.00 and $7.25, the minimum wage in Oklahoma. Position types and education level for respondents who checked the upper category ($50.00/hr or more) were examined, and were surprisingly found to be primarily Masters-level therapists, along with a few physicians. For this reason, we used a rate relatively close to the second-highest category, and significantly below one that might be expected for physicians: $62.50. Staff responses to the item How many years have you been in the field? were used to measure staff experience. Detailed information on the distribution of these variables is offered in Appendix A9. Gender and ethnicity were tested using the original dichotomous survey items, and race was tested by collapsing five dichotomous survey items into a single variable with up to six categories: American Indian/Alaskan Native alone, Asian alone, Black/African American alone, Native Hawaiian/Pacific Islander alone, White alone, and more than one race. Due to low Ns, the Asian alone and Native Hawaiian/Pacific Islander alone categories were eliminated from the crosstabs. In addition to the variables gathered through the staff survey, the relationship of staff intention to leave to key program variables was investigated, including the program manager-cited causes of turnover and the study dimensions described earlier (industry, region, service type, program setting, population age, state operation, and organizational size). Analysis and Results As with the analyses described earlier, relationships were examined between intention to leave and each of the variables described above on an individual basis. As with separation rate, there was no relationship between intention to leave and program manager citation of the significant causes of turnover. Of the study dimensions, only service type and region were significantly related to staff intention to leave. Staff position type, gender, ethnicity and race were not significant, but staff age, experience, pay and job satisfaction were significant. Initially, the relationship between consumer or family status and intention to leave was investigated by collapsing eight dichotomous survey items into a single four-category variable: neither, consumer only, family member only, and both consumer and family member. This variable was significantly related to intention to leave. However, the distribution was difficult to interpret as staff who identified as being consumers only seemed much less likely to intend to separate than did staff who identified as either family members or both consumers and family members (full 25 details available in Appendix A10). Given this, it seemed possible that the use of this collapsed variable could be obscuring the meaning of the relationship. A variety of alternatives were tested, including the eight original survey items (adult mental health consumer, adult substance abuse consumer, former youth mental health consumer, former youth substance abuse consumer, family member of an adult mental health consumer, family member of an adult substance abuse consumer, family member of a youth mental health consumer, family member of a youth substance abuse consumer), as well as aggregations of these items across two dimensions individually and together (adult/youth and mental health/substance abuse). Most of these tests did not yield significant results. However, family status did prove to be significantly related to intention to leave, with a higher proportion of family members than non-family members indicating that they planned to leave within the next year. When family membership was broken down further into mental health and substance abuse, the relationship between being a family member of a mental health consumer and intention to leave was significant, while that between being a family member of a substance abuse consumer and intention to leave was not significant. However, as the latter relationship showed a similar trend (higher intention to leave among family members), the combined mental health and substance abuse variable was retained for further analysis. Logistic regression was employed to determine whether the relationships noted above remained significant when considered simultaneously. We began by examining the relationship between staff age, experience, pay, and job satisfaction. While there were some relationships among these variables, none turned out to be strong enough to warrant excluding any of the variables from the regression model. The details of this analysis can be found in Appendix A11. The model tested included region, service type, population age, job satisfaction, pay, age, experience, and family member status. Of these variables, only job satisfaction and age remained significant. The mean satisfaction score for staff not intending to leave was 1.71, with 1 being very satisfied and 2 being satisfied (Exhibit 2.9). The mean for staff intending to leave was 2.59, closer to 3, or neither satisfied nor dissatisfied. Consistent with literature on the topic, staff intending to leave were younger on average than those not intending to leave (39.67 years versus 43.30 years, respectively). Complete details on the results of the regression model are shown in Appendix A12, and additional details on the relationship of job satisfaction and staff age to intention to leave are shown in Appendix A13. Exhibit 2.9: Satisfaction and Age Among Staff Intending to Stay and Intending to Leave Mean for staff staying Mean for staff leaving Staff overall job satisfaction (N=1241) 1.71 2.59 Staff age (N=1180) 43.30 39.67 Data from the staff surveys. While only these two variables remained significant in the regression model, Appendix A14 gives additional information on the other variables tested. 26 Summary Information related to separations was gathered through program manager reports of the perceived causes of separation in their programs, program managers’ reports of their programs’ separation rate over the previous year, and staff reports of their intention to leave their position within the next year. Nine causes of turnover were cited by at least 10% of program managers; those causes cited by fewer than 10% of program managers are not included in the analysis. Consistent with the findings related to recruitment barriers, the most frequently cited barrier was dissatisfaction with pay, which was cited by nearly two thirds of program managers. Excessive paperwork, emotional burnout and excessive on-the-job stress were cited by at least one third of program managers. While program and organization characteristics were related to multiple perceived causes of turnover when the relationships were examined individually, generally only one or two characteristics remained significant in each logistic regression model. Organizational industry was a significant predictor of citing dissatisfaction of pay, with OPHA program managers being the least likely to cite pay as a cause of turnover. Population age and program setting were significant predictors of perceiving paperwork to be a cause of turnover, with program managers in programs serving children citing paperwork more frequently than those serving both children and adults, and program managers in outpatient settings citing paperwork more frequently than program managers in other settings. Service population also related to citation of dissatisfaction with job responsibilities, with program managers from programs serving both children and adults being less likely to cite this as a barrier than program managers from programs serving either adults or children. Program separation rates ranged from 0% to 200%, and the median of 25% was used to divide programs into two categories: low separation and high separation. These categories were related to multiple program and organizational characteristics when the relationships were examined individually, but only two characteristics remained significant in the logistic regression model. High separation programs proved to be more likely to have a high proportion of techs on staff, and less likely to be state operated. The position type results are consistent with existing literature regarding the relationship between lower staff experience/job level and higher separation rates. The vast majority (80%) of staff did not report intending to leave their positions within the 12 month period following the survey. Intention to leave was related to a range of program, organizational and staff characteristics when the relationships were examined individually, but only two remained significant in the logistic regression model. As would be expected, staff intending to leave reported lower satisfaction with their job overall. Staff age was also related to intention to leave, with the mean age for staff intending to leave being about three and a half years younger than that of staff intending to stay. Both of these findings are consistent with the literature on staff intention to leave. 27 VACANCIES AND STAFF RECRUITMENT BARRIERS Like staff separations, position vacancies are an area of concern in many behavioral healthcare programs, and a topic of interest to the study stakeholders. We collected information on position vacancies on two issues, using the program manager survey. First, program managers were asked to review a list of 19 possible barriers to staff recruitment, and were then asked to indicate which of these were most relevant to their program. Second, program managers were asked to report on the current vacancies in their program, using the six position categories described in the separations section. This section describes the data received in response to each of these sets of items, and the analysis conducted to investigate relationships between these variables and program characteristics. Program Manager Perceptions of Recruitment Barriers Program managers were asked to identify the top three barriers to filling staff vacancies in their programs. As a result, percentages for this item add up to more than 100%. The list shown in Exhibit 3.1 does not include the barriers cited by fewer than 10% of program managers.6 The majority (57%) of program managers identified salary/pay as the greatest obstacle to filling vacancies in their programs. Lack of candidates with desired credentials or work experience, small applicant pool due to geographic location and competition from other fields were cited by 25% or more program managers as barriers to staff recruitment. Exhibit 3.1: Program Manager Perceptions of Recruitment Barriers Data from the program manager survey. 6 The following potential barriers were listed as options on the survey, but were cited by fewer than 10% of program managers: cumbersome hiring process; career ladder not attractive; childcare not offered; organizational facilities not attractive; organizational reputation; negative stereotypes of service consumers; job responsibilities not attractive; amount of training required; cost of training required; and benefits not attractive. 28 Recruitment Barriers and Program Variables The next four tables illustrate how recruitment barriers vary by industry, region, organizational size and type (those barriers cited by fewer than 10% of program managers are not included in the analyses). Other results indicate that recruitment barriers did not vary by service type (i.e., mental health, substance abuse and dual-diagnosis). The following items are those most often cited by program managers as reasons for vacancies: 1. Salary/pay not attractive; 2. No candidates with desired credentials; 3. No candidates with desired work experience; 4. Competition from other fields; 5. Problems with funding/not allowed to fill a position; and 6. Shift/work hours not attractive. Industry and Recruitment Barriers All six of the perceived barriers above show statistically significant differences between industries. Eighty percent of program managers in the OJA industry group identified salary/pay as one of the most critical barriers to filling vacancies, while only 19% of OPHA industry group program managers cited this as a recruitment barrier. Substance Abuse fell roughly in the middle of this continuum, with 49% of program managers citing pay as a barrier, while Mental Health and DHS program managers responded relatively similarly to those from OJA, with 75% and close to 60% of program managers from these industries citing pay as a recruitment barrier, respectively. No OJA program managers cited difficulty finding candidates with desired credentials, but roughly two fifths of Mental Health and Substance Abuse industry program managers perceive this to be a recruitment barrier in their programs. Competition from other fields also varies by industry. Program managers working in the OJA industry group were more likely to cite this as a barrier to staff recruitment (70%) than program managers from any other industry group. In fact, the next closest group was program managers from the Mental Health industry, with 32%. One third of program managers from the Substance Abuse industry group perceive funding or not being allowed to fill a position as one of the most pertinent causes of vacancies. OPHA and the Mental Health industries followed with 15% and 14% respectively. Only 10% to 12% of program managers in the DHS and OJA industries thought funding was an important recruitment barrier. Not surprisingly, shift/work hours is more frequently perceived as a barrier by program managers in industries with a high proportion of 24-hour programs (OPHA, OJA). Finally, while nearly one third of OJA program managers perceive the hiring process itself to be a barrier, this process was not cited as a barrier by any Substance Abuse industry program managers. 29 Exhibit 3.2: Program Manager Perceptions of Recruitment Barriers by Industry Perceived Barrier CMHC N=101 DHS N=17 OJA N=10 OPHA N=26 SA N=61 Salary/pay not attractive 74% 59% 80% 19% 49% No candidates w desired credentials 37% 24% 0% 15% 41% Competition from other fields 32% 12% 70% 31% 15% Funding/not allowed to fill position 14% 12% 10% 15% 33% Shift/work hours not attractive 17% 24% 40% 42% 15% Cumbersome hiring process 11% 18% 30% 15% 0% Data from the program manager surveys. ♦ Items cited by fewer than 10% of program managers are not included in the exhibit. ♦ All data are significant at the p<.05 level. ♦ FQHC, DOC, Other Medicaid, and Child Guidance industries are not included in the analysis due to the low number of programs responding to these items. State Operation and Recruitment Barriers The perceived barriers of salary, candidate experience, and funding vary by organizational type or operation (state operated vs. privately operated). As shown in Exhibit 3.3, nearly three-quarters of program managers from state operated organizations cite salary as a barrier, in comparison to just over half of program managers from privately operated organizations. As noted above, OPHA program managers were also significantly less likely to cite salary as a barrier. Interestingly, OPHA is the only industry group in these analyses that are made up of entirely private organizations. Program managers from state-operated organizations were significantly less likely than those from private organizations to cite lack of candidates with desired work experience as a recruitment barrier. Finally, state-operated organizations (42%) were more likely than privately-operated (17%) to cite funding as a fundamental problem to staff recruitment, and are also more likely to cite salary as a recruitment barrier. Exhibit 3.3: Program Manager Perceptions of Recruitment Barriers by Organizational Type Perceived Barrier State Operated N=53 Privately Operated N=181 Salary/pay not attractive 74% 52% No candidates with desired experience 6% 34% Funding/not allowed to fill position 42% 17% Data from the program manager surveys. ♦ Items cited by fewer than 10% of program managers are not included in the exhibit. ♦ All data are significant at the p<.05 level. Organizational Size and Recruitment Barriers Organizational size is associated with program manager perception that salary and lack of staff with desired credentials are recruitment barriers. Program managers affiliated with large organizations (those with an estimated staff size of at least 82 full time employees) cite salary/pay as a reason for staff vacancies more often than those affiliated with other organizations (67%, compared to 42% and 45% of small and medium organizations, respectively). Further bivariate analysis indicates that small organizations (those with an estimated staff size of less than 15 full-time employees) have more professional staff – requiring additional education – and are less likely to be inpatient facilities requiring a large number of 30 aides/techs who typically earn the lowest salary among direct care staff. While program managers from medium and large organizations cite lack of candidates with credentials at roughly the same rate, those from smaller organizations cite this barrier at a considerably higher rate. Exhibit 3.4: Program Manager Perceptions of Recruitment Barriers by Organizational Size Perceived Barrier Small Orgs N=33 Medium Orgs N=53 Large Orgs N=126 Salary/pay not attractive 42% 45% 67% No candidates with desired credentials 52% 30% 27% Data from the program manager and organizational surveys.♦ Items cited by fewer than 10% of program managers are not included in the exhibit. ♦ All data are significant at the p<.05 level. Region and Recruitment Barriers Finally, geographic region was significantly related to four of the perceived recruitment barriers: absence of candidates with desired work experience, small applicant pool due to geographic location, competition from other fields, and location of agency not attractive. Not surprisingly, two of these barriers are explicitly location-based, and a third (lack of candidates with desired work experience) could also be argued to be intrinsically tied to location or area. Exhibit 3.5 demonstrates that a small pool of applicants is the greatest barrier to filling vacancies (52% and 47%, respectively) in the Northeast and Southeast corridors of the state, while about half of the program managers from the Northwest indicated that competition from other fields was a problem with respect to vacancies in the behavioral healthcare field. Exhibit 3.5: Program Manager Perceptions of Recruitment Barriers by Region Perceived Barrier NE N=54 NW N=14 OK N=74 SE N=32 SW N=30 TU N=26 No candidates w desired work experience 15% 21% 30% 19% 40% 46% Small applicant pool due to geographic location 52% 43% 7% 47% 23% 4% Competition from other fields 19% 50% 34% 28% 7% 42% Location of agency not attractive 35% 14% 8% 13% 3% 0% Data from the program manager surveys.♦ Items cited by fewer than 10% of program managers are not included in the exhibit. ♦ All data are significant at the p<.05 level. Salary as a Perceived Recruitment Barrier Given that salary was the most frequently cited recruitment barrier as well as the most frequently cited cause of separations, it warranted further exploration. Logistic regression was used to determine whether the three program variables (industry, state operation, and organization size) remained significant predictors of salary as a barrier when tested simultaneously. While organization size did not remain significant, both industry and state operation were significant. The significant relationship between salary as a perceived barrier and industry can be attributed to the low proportion of OPHA program managers citing salary as a barrier. There was a significant difference between the rate of OPHA citation of salary and that of the mental health industry program managers, who were chosen as the reference group in the regression model. As suggested by the earlier analysis of the relationship between salary as a perceived barrier and 31 state operated status, the results of the regression model indicated that program managers in state operated programs were significantly more likely to cite salary as a recruitment barrier. Further details on the results of this regression model may be found in Appendix B1. Program Manager-Reported Vacancy Rates As reported in the section on separations, program managers were asked to report the current number of full time equivalents (FTEs) budgeted for their program and vacant in their program. These items were posed in reference to each of six position categories: aids/techs/other paraprofessionals, professionals primarily holding Masters degrees (counselors/therapists/MSW-level social workers), LPNs, psychiatrists and other physicians, doctoral-level psychologists/DSW-level social workers, and RNs. To calculate the vacancy rate for a given region, the number of vacancies was totaled across participating programs, and this sum was divided by the number of FTEs budgeted across programs. Exhibit 3.6 shows the position-specific and total vacancy rates statewide, and for each of the six geographic regions. It is important to note that organizations may not have included staff that they contract with (rather than employ) in the counts that follow. Exhibit 3.6: Cross-Industry Vacancies by Region Position NE NW OKC SE SW Tulsa Statewide Aid/tech 7% 13% 8% 7% 16% 8% 9% Masters-level professional 15% 9% 12% 11% 2% 36% 15% LPN 4% 14% 9% 20% 33% 0% 9% Psychiatrist/ physician 3% 0% 0% 33% 13% 10% 7% Psychologist 6% 50% 0% 0% 0% NA 7% RN 13% 22% 15% 28% 8% 7% 14% Total 10% 13% 10% 12% 10% 17% 11% Data from the program manager surveys. Calculating Program Vacancy Rate Percents in the table above were calculated by summing vacancies and budgeted positions across the region. In the analysis that follows, vacancies are calculated at the program level. As noted in the separation section, programs were chosen as the unit of analysis due to concerns that program characteristics and local program environment may vary widely within larger organizations - particularly those with programs across a wide geographic range. Program vacancy rates ranged from 0% to 100%. The median vacancy rate was 4%, meaning that roughly half of the participating programs had a vacancy rate below 4%, and roughly half had a vacancy rate above 4%. Appendix B2 gives more information on the distribution of the program vacancy rates. 32 Relationships Between Vacancy Rates and Other Program Variables We examined the relationship between vacancy rate and a number of program characteristics and related variables. Programs were categorized as either having a low vacancy rate (less than 5%) or high vacancy rate (5% or higher). We began by testing for relationships between vacancy rate and each of the frequently-cited recruitment barriers. Then, as with separation rate, we looked for a relationship between staffing patterns (e.g., proportion Masters-level counselors, proportion techs) and vacancy rate. Finally, we looked for relationships between vacancy rate and each of the study dimensions described earlier (industry, region, service type, program setting, population age, state operation, and organizational size). Analysis and Results We began by performing analysis to identify relationships between vacancy rate and each of the variables above on an individual basis, with the intention of then testing these relationships simultaneously. However, only one of the identified variables proved to be related to vacancy rates. None of the frequently cited recruitment barriers were associated with program vacancy rate, nor were any of the study dimension variables. Of the staffing patterns variables, only proportion of RNs was related to vacancy rate, with high vacancy programs having a greater proportion of RNs than low vacancy programs. As shown in Exhibit 3.7 the average proportion RNs for low vacancy programs was 4%, while the average for high vacancy programs was 7%. While this difference may appear relatively small, it was statistically significant. This finding may be related to the comparatively high rate of vacancies among RN positions overall. As noted earlier in Exhibit 3.6 the overall vacancy rate for RN positions was comparable to that for Masters-level counselors, which was the position type with the highest vacancy rate. Exhibit 3.7: Proportion RNs in Low Vacancy and High Vacancy Programs Staff position type predictors Mean proportion low vacancy programs Mean proportion high vacancy programs Proportion RNs 4% 7% Data from the program manager surveys. Because only one variable proved to be related to vacancy rate, it was not necessary to employ logistic regression to test multiple relationships simultaneously. Additional information on the (non-significant) findings for the remaining variables may be found in Appendix B3. Summary Information related to vacancies was gathered through program managers’ reports of the perceived recruitment barriers in their programs, and their reports of their programs’ current vacancies. Nine of the barriers were cited by more than 10% of program managers (those items cited by fewer than 10% of program managers were not included in the analyses). By far the most frequently cited barrier was salary, which was cited by 57% of program managers. Lack of candidates with desired credentials or desired work experience, small applicant pool due to geographic location, and competition from other fields were all cited by more than one quarter of program managers. Program and organization characteristics that were related to multiple perceived barriers included organizational industry, state operation, organizational size, and geographic region. Salary as a perceived barrier was related to three of these characteristics when the relationships were tested individually. When examined simultaneously, salary 33 remained significantly related to industry, with OPHA program managers being significantly less likely to cite salary as a barrier. Likewise, state operation and salary were related, with program managers in state operated organizations more likely to cite salary as a barrier. Program vacancy rates ranged from 0% to 100%, and the median of 4% was used to divide programs into two categories: low vacancy (less than 4%) and high vacancy (greater than 4%). These categories proved to be unrelated to most of the program and organizational characteristic variables. Staffing patterns offered one exception: The mean proportion of RNs in low vacancy programs was slightly but significantly lower than the mean proportion of RNs in high vacancy programs, which could be in part related to the comparatively high rate of vacancies in RN positions, across programs. 34 CURRENT AND FUTURE STAFFING NEEDS The purpose of this chapter is to identify unmet needs for the behavioral healthcare workforce with a focus on type of position. The first section focuses on psychiatrists and other prescribers. The second section focuses on other professional and non-professional staff. Each of these sections employs data from different sources so the methods upon which we have relied are described within each section, as well as the implications for higher education. The third section describes one underlying problem, the level of compensation currently available to the Oklahoma workforce. Need for psychiatrists and other prescribers of psychiatric medications State mental health authorities typically do not have empirical information about the characteristics of their current workforce. In order to fill this information gap, we undertook a number of studies, as well as searches for relevant data, that would provide useful information for understanding difficulties faced by staff providing mental health services in Oklahoma. Among the studies that we identified was a study of the relative unmet need for professional mental health workers in the State of Washington (Morrissey, et al, 2007a), undertaken as a part of their Mental Health Transformation State Incentive Grant. Morrissey and his colleagues employed a simple model as the foundation of their work. First, they estimated the number of adults (persons over age 18) who could be classified either as persons with serious mental illness or as persons with other mental health needs. For each of these two types of persons, they estimated the percentage that would access mental health non-inpatient services in one year and the number of units of professional services they would use. Professional services are broken down into those provided by individuals who are licensed to prescribe medications (prescribers) and individuals who are licensed to provide services other than medications (non-prescribers). These estimates then allow new estimates of the numbers of prescribers and non-prescribers needed (in full time equivalents—FTE) to serve a population within a defined geographic area. The estimates of need are then subtracted from the number of licensed professionals available to yield the shortage of professionals. They summarize their model as follows: Need = People with serious mental illness + people with other mental health needs Workforce = Prescribers + Non-prescribers Shortage = FTE available – FTE needed It is important to emphasize that these are relative not absolute measures of unmet need. This means that they are most useful in comparing the need from one area to another, but do not necessarily provide an estimate of the exact number of additional professional staff needed. Moreover, apparent surpluses produced by these estimates cannot be relied upon. The study of Washington State was a part of a larger, national study sponsored by the Health Resources Services Administration (HRSA) of the U.S. Department of Health and Human Services. This allowed Morrissey and his colleagues to develop estimates of professional shortages for every county in the U.S. We contacted them and requested estimates for 35 Oklahoma. Their findings, as well as the methods employed to arrive at their estimates, are presented here. We also discuss some of the limitations of their findings. Findings Most specialty prescribers in Oklahoma are psychiatrists, although there are a handful of advanced practice psychiatric nurses. Other physicians can and do prescribed psychiatric medications, as well. Exhibit 4.1 below presents regional and statewide estimates of counts of prescribers available to provide mental health services in Oklahoma. As previously discussed, the state is divided into six regions, as follows: Central Oklahoma (the counties in which Oklahoma City is located) and Tulsa are separately estimated. The remaining counties are grouped into four quadrants, Northeast, Northwest, Southeast, and Southwest. The table shows 278 FTE psychiatrists/prescribers. Exhibit 4.1: Available FTE Mental Health Specialty Prescribers by Licensure Group and by Oklahoma Regions Region Licensure Smoothed Total APPN PI Prescribers7 OKC 10 133 107 Northeast 2 30 78 Northwest 3 8 9 Southeast 3 13 24 Southwest 1 25 38 Tulsa 6 70 32 Total 23 278 287 For psychiatrists, full time equivalents are greater than the raw count because practice pattern data indicate that psychiatrists average more than 40 hours/week. Exhibit 4.2 below presents regional and statewide totals of FTE needed and FTE shortages for prescribers. For the prescriber group the UNC estimates produce a shortage of 410 FTE. 7 In the initial analysis, the county is used as the primary geographical unit for shortage estimation. This decision was made primarily due to the lack of accurate small-area data on mental health needs and practice locations, but also because people are likely to travel within larger areas for mental health services. Each county-level need and supply estimate was adjusted using a smoothing method that accounts for travel across county boundaries for mental health services. Within Oklahoma particularly, with its many small counties, ignoring this would lead to overestimates of need. The maximum amount of time that people can be expected to travel for mental health services is about 60 minutes (Fortney, Owen & Clothier, 1999; Fortney, Rost, Zhang et al., 1999). Therefore, for a given index county, the need and supply estimates of counties within a 60-minute radius were weighted and added to the estimates for the index county. The weighted estimates were scaled so that the national need and supply totals for prescribers and non-prescribers were unchanged by the smoothing process. In the final analysis, counties were aggregated by regions within Oklahoma. 36 Exhibit 4.2: Estimates of Shortages of Specialty Mental Health Prescribers FTE by Oklahoma Region Region Total FTE Available, Smoothed FTE Needed, Primary Care Adjusted, Smoothed Relative Shortage (FTE), Primary Care Adjusted, Smoothed OKC 107 187 -80 Northeast 78 217 -139 Northwest 9 32 -23 Southeast 24 107 -83 Southwest 38 94 -56 Tulsa 32 61 -29 Total 287 697 -410 Methods Methods are described in detail in Morrissey et al (2007b). Exhibit 4.3 below presents a brief summary of the data sources employed and how the estimates were derived. Exhibit 4.3: Data sources employed to estimate mental health workforce needs and available workforce. Variable Estimated Source of Data Oklahoma specific data Prevalence of Mental Illness (persons-in-need) National Comorbidity Survey Replication (NCSR); Medical Expenditures Panel Survey (MEPS) Yes Estimates of percent of persons-in-need using mental health services annually MEPS for non-SMI population; Assume 100% for SMI population No Estimates of average units of outpatient services used per person annually NCSR, MEPS No Estimates of visit hours per working day for prescribers Substance Abuse and Mental Health Services Administration (SAMHSA) No Estimates of need met by primary care providers Need estimate reduced by 15 percent in counties without a shortage of primary care providers (no single reference) Yes Estimates of supply of mental health professionals Various sources, generally relevant professional associations Yes Adjustments of need in rural counties that are close to larger counties Various references; assumed maximum travel time would be 60 minutes for mental health services Yes As shown above, Morrissey and his colleagues relied upon a number of data sources in order to estimate each of the variables required to determine workforce shortages. These sources are generally recognized as the most reliable sources of information available, although in several cases these may be the only sources available. 37 Discussion Prior to the completion of the work by Morrissey and his colleagues, the only available estimates of need for mental health professional services were on the website of the Health Resources and Services Administration. However, there was no explanation of the method employed to develop these estimates or references to underlying research. Thus the work described here represents the first systematic attempt to provide appropriate estimates of workforce needs and shortages. Nonetheless, there are limitations that must be recognized. Morrissey et al (2007b) acknowledge that the populations included do not extend to adults who are homeless or in institutions (e.g., inpatient, corrections) or children and adolescents. They also do not include needs for staffing of substance abuse programs. They indicate that the measure of shortage “is probably most useful when taken as an expression of relative rather than absolute unmet need.” Finally we reviewed the UNC estimates of available FTE with more recent data from Oklahoma State licensing boards. The UNC estimates are generally close, but underestimate the size of the current, licensed workforce. However, there is no data available on whether individuals who are licensed are actually engaged in clinical practice. We know anecdotally that at least some may be retired or only have a part-time practice or are working in administrative, rather than clinical positions. As we have also pointed out above, other licensed individuals are working in positions that are not counted in the need estimates (e.g., agencies serving child and youth, agencies providing adult or child inpatient care). As a result, we believe that the strategy of simply counting licensed practitioners leads to a systematic overestimate of the available supply of such professionals. Addressing the Shortage of Prescribers Oklahoma has three psychiatric residency programs which collectively produce about 13 new psychiatrists per year. Assuming that our estimate of the current need for over 400 prescribers of psychiatric medications is reasonably accurate, it would take over 30 years for these programs to fill the unmet need. This does not account for retirements during this period which will only increase the unmet need. It is unlikely that these residency programs will expand substantially or that psychiatrists will be recruited in significant numbers from elsewhere in the United States because this is a national problem. The numbers of new doctors entering psychiatric residency programs has been falling for over 20 years, and changes that would reverse this trend are not in the offing. Information about Osteopaths either training to practice psychiatry or trained to do so in Oklahoma suggests that this group also is not likely to expand the numbers of prescribers in the foreseeable future. There are no osteopathic residency training programs in psychiatry in Oklahoma, and only a few D.O.’s practice primarily psychiatry in Oklahoma (36) and even fewer are certified to do so (18). Given that psychiatrists cannot be expected to fill the need for new prescribers, what options exist? There are three that we know of: • Advanced practice psychiatric nurses can be trained to fill this need. At present there are only a handful of persons with this training in Oklahoma, but nursing schools could be encouraged to offer the necessary education. 38 • With an expansion of integrated primary care and mental health care, primary care physicians could become an expanded source of prescribers. Creating incentives to develop integrated care practices, targeting training in integrated care, and promoting continuing education in prescribing psychiatric medications, could contribute to an expansion in the numbers of competent prescribers. • Licensing Ph.D. psychologists with special additional training to prescribe medications would also expand the numbers of prescribers; two states now allow this. All three approaches may be necessary to fill the gap, which is quite substantial. If no action is taken to increase the numbers of prescribers, the problem may become worse with the retirements of older psychiatrists, who were trained in an era when psychiatry was a more attractive field. The numbers of retirements may exceed the small numbers of annual replacements. Conclusions The UNC data demonstrate an unequivocal need for more prescribing professionals in all areas of the state. The total estimate of need for 410 additional prescribers is probably an underestimate for reasons discussed above. The area of the State with the greatest unmet need is the Northeast quadrant, excluding Tulsa which has the smallest, relative unmet need. Non-prescribers Exhibit 4.4 below shows current staffing by position type for the state of Oklahoma and the six regions within the State. Exhibit 5 below shows population-based rates for behavioral healthcare positions by type in Oklahoma and the surrounding states. Oklahoma and the surrounding states are similar in most categories. The major exception is RNs which are less available in Oklahoma. LPNs and MH/SA Techs are marginally more available. There is considerable variation within Oklahoma. The Central Region (OKC) has among the highest rates of availability for all categories of positions. Tulsa is close and leads in availability of MH/SA Techs. The more rural areas of the State have significantly less availability of professionals, psychologists, MH/SA Counselors, and RNs. Exhibit 4.4: Current (2008) numbers of behavioral healthcare positions by positions type in Oklahoma Position State ODMHSAS Region Counts Tulsa OKC NE NW SE SW Psychologist 1,339 195 628 236 58 96 126 MH/SA Counselor 6,993 1,100 2,691 1,485 281 785 651 RN 26,157 5,714 10,839 3,560 962 2,766 2,316 LPN 13,463 2,411 4,163 2,062 716 2,100 2,011 MH/SA Tech 38,590 9,124 11,394 6,474 1,833 5,530 4,235 39 Exhibit 4.5: Current (2008) rates per 10,000 population of behavioral healthcare positions by position type in Oklahoma and surrounding states Position State Rate Multi- State Regional Rate* National Rate ODMHSAS Region Rates Tulsa OKC NE NW SE SW Psychologist 3.9 8.1 9.6 3.5 6.3 3.1 2.9 2.1 2.8 MH/SA Counselor 20.3 27.0 34.4 19.5 27.2 19.2 14.0 16.8 14.2 RN 75.8 87.0 92.7 101.4 109.5 46.1 47.8 59.4 50.6 LPN 39.0 35.0 27.5 42.8 42.1 26.7 35.6 45.1 43.9 MH/SA Tech 111.8 121.3 126.1 162.0 115.1 83.8 91.1 118.7 92.5 Regional rate includes the following states: Arkansas, Colorado, Kansas, Louisiana, Missouri, New Mexico, Oklahoma and Texas. Exhibit 4.6 below shows the current (as of 2008) numbers of behavioral healthcare staff by position type for Oklahoma and for the United States overall. It also shows the projected needs for staffing as of 2018. Projections are based principally upon projected population changes. The additional positions are necessary to maintain the same rates of services currently provided. Exhibit 4.6: Oklahoma and national current (2008) and projected (2018) rates of change for behavioral healthcare staffing positions Position Oklahoma National 2008 Jobs 2018 Jobs Change % Change % Change Psychologist 1,339 1,434 95 7.1% 15.8% MH/SA Counselors 6,993 8,270 1,277 18.3% 17.8% RNs 26,157 31,444 5,287 20.2% 22.3% LPNs 13,463 14,578 1,115 8.3% 12.9% Aids/Techs 38,590 46,122 7,532 19.5% 23.8% Total 86,542 101,848 15,306 17.7% 21.3% The difficulty is that these estimates include only job growth in predicting the numbers of new persons needed to fill available positions. However, current members of the behavioral healthcare workforce will be leaving their positions, either for retirement or other reasons. This also needs to be accounted for in developing estimates of persons needed to fill positions annually. Next, we outline an approach to employing appropriate data to reach this goal. 40 Job Growth We received the table below from the Department of Commerce. This means, for example, that in 2018 there will be 361 more jobs for psychologists than there were in 2008. One implication is that it is necessary to train or import 361 new psychologists into the system between 2008 and 2018. The source of these estimates is data from Economic Modeling Systems, Inc (EMSI). EMSI uses several different databases to come up with their estimates, including population projections from the Census Bureau. However, that is not the only factor, current employment trends and participation rates from the Bureau of Labor Statistics (BLS) go into the projections. Also included are IRS income and migration data that shed more light on the single employers or those that do not pay into Unemployment Insurance. The simple explanation is that it uses current employment trends by industry and certain population trends. Industry trends, legislation, and several other factors are also used to decipher which industries will be growing. Population is a key component but labor participation, county wages, migration patterns and trading patterns are also factors that influence the model. EMSI breaks down these trends to the county level, which can then be aggregated to the state level. The table below shows the rate of growth for ten years. The annual growth rate is between one and two percent, depending upon the position type. Exhibit 4.7: Ten Year Growth by Position Category Position Category State 2008 Jobs 2018 Jobs % Change Psychologist 2,738 3,099 13.2% Mental Health or Substance Abuse Counselors 9,726 11,377 17.0% Registered Nurses 26,552 32,271 21.5% Licensed Practical Nurses 13,936 15,554 11.6% Mental Health or Substance Abuse Aides/Assistants/ Technicians 44,546 54,536 22.4% Overall Total 97,498 116,837 19.8% This assumes that the persons occupying these positions in 2008 continue to be available to fill positions in 2018. However, we know that this is not the case. Some people who occupy positions in 2008 will retire or leave the field for other reasons over the next ten years. For example, it will be necessary to train or import more than 361 new psychologists into the system for these reasons. The problem is to estimate how many more psychologists will be needed to fill available positions. Accounting for persons leaving the field Let us assume that from 2008 to 2009 the growth rate for psychologists is two percent. Then the change in jobs would be an increase of 55, meaning that there would be a need for a minimum of 55 new psychologists to fill those jobs. In addition, let us assume that there is a 20 percent separation rate among psychologists during 2008 or 550 persons and further that ten percent of those separated actually leave the field. That would mean an additional 55 new psychologist 41 would be needed to fill those vacated jobs, giving a total of 110 psychologists needed to fill the new jobs and the jobs vacated by those leaving he field. If this reasoning is correct, then modeling the number of new persons needed to fill psychology jobs requires an annual estimate of the percent of persons leaving the field. As a part of our survey work, we collected information from 1349 individual staff who are currently in behavioral healthcare positions. We asked each of those individuals to indicate whether they planned to stay in their position during the next year. Twenty one percent of staff and six percent of program managers indicated that they do plan to leave their positions. The percentages of persons indicating that they planned to retire or indicating that they planned to leave behavioral healthcare are shown in the table below. Exhibit 4.8: Staff Planned Separation Rates and Program Manager Estimated Separation Rates (1) (2) (3) (4) (5) (6) (7) Position Type Persons answering survey Percent planning on separating within the next year Percent planning to retire Percent planning to leave the field Sum of columns (4) and (5) Estimated separation rates from program manager surveys Aid/tech 343 21% 0% 6% 6% 42% Masters-level professional 317 19% 1% 3% 4% 26% LPN 37 24% 3% 8% 11% 35% Psychiatrist/ physician8 - - - - - 23% Psychologist 28 21% 4% 0% 4% 11% RN 149 19% 1% 5% 6% 26% Total 874 20%9 1% 4% 5% 35% Note that, with the exception of psychologists, program manager-reported, actual separation rates exceed staff self-reported plans to separate by a wide margin. However, these data can be considered together to estimate a range of possible industry departure ranges. At the conservative end of the range is the staff self-report: an overall rate of 5%, with position-type specific rates ranging from 4% to 11% that is probably a conservative estimate of rates of person who separate leaving the field. Alternatively, the proportion of planned industry departures can be applied to the program manager-reported separation rates for a less conservative estimate. These range for a low of 11% for to a high of 35% for the Aid/Tech positions. These rates are higher than the annual growth rates projected by EMSI. This means that the growth in estimates 8 There is insufficient data for psychiatrists to provide these estimates. 9 Note that this is very slightly lower than the rate cited in the text above (21%). The information in this table is based only on responses that could be linked to a position type (N=877), while the overall number cited in the text is based on all the responses to this item that were received (N=965). 42 of persons needed to fill positions year by year will be influenced to a much greater degree by estimates of staff turnover, representing the need to replace existing members of the workforce. There is one additional consideration in estimating the numbers of new persons needed to fill positions. The “jobs” in the EMSI estimates are only filled jobs; they do not include unfilled jobs. Thus we also need to take into account vacancy rates for the appropriate position type. We do have estimates of vacancy rates for Oklahoma for each position type, as shown in the table below. Exhibit 4.9: Vacancy Rate by Position Category Position Type Percent Vacant Aid/tech 9% Masters-level professional 17% LPN 10% Psychiatrist/ physician 10% Psychologist 6% RN 14% Total 12% With this term included, for each position type the number of new persons needed to fill existing vacancies, positions vacated by persons leaving the field and by growth (from EMSI) would be given by the following equation: [new persons needed in year i+1] = [number of jobs in year i] x [growth rate + percent leaving field] x [1- vacancy rate] Findings Findings are presented for three position types, psychologists, mental health and/or substance abuse counselors, and mental health and/or substance abuse aides/techs. The latter are direct care positions that do not necessarily require professional degrees or licensure. We have not included registered nurses and licensed practical nurses because only a relatively small proportion of these positions are in behavioral healthcare and separate estimates of need have been developed by the Oklahoma Healthcare Workforce Commission. The need for psychiatrists and other prescribers is discussed earlier in this section. The “net growth” figures in the column to the right show the numbers of additional persons who must either be trained or imported over a ten year period to be sure that the estimated behavioral healthcare positions for 2018 and in the intervening years will be filled. This is 1,808 psychologists, 7,045 mental health and substance abuse counselors, and 51,625 aides or techs. 43 Exhibit 4.10: Net Growth by Position Category Position Category State 2008 Jobs 2018 Jobs 2018 Persons Needed Net Growth Psychologists 2,738 3,099 4,546 1,808 Mental Health or Substance Abuse Counselors 9,726 11,377 16,771 7,045 Mental Health or Substance Abuse Aides/Assistants/ Technicians 44,546 54,536 96,171 51,625 We consider that the estimates shown in the table above are conservative for the following reasons: ��� We employed the separation rates estimated from individual reports of intentions to leave their present positions and the field, rather than the separation rates estimated from program manager reports of persons leaving their positions. The latter are two to four times higher than the former. • The number of positions only includes those who are considered “state covered”. If we include all positions, including persons in individual or small group private practices, the numbers would also be higher. This is particularly true for psychologists who are much more present outside the public sector than inside. Exhibit 4.11 below shows the numbers of degrees awarded each year over a six year period beginning in 2001-02 and ending in 2006-07. (A detailed breakdown of degrees awarded in specific fields within each of these larger categories is provided in Appendix A15.) With the exception of psychologists, there has been an increase in each category over this time period. The two columns at the right of the table show the number of degrees expected to be awarded cumulatively from 2007-08 through 2017-18 and the need for new degree recipients to meet the demand for new staff positions. The projections of degrees awarded are based upon a simple linear trend model employing the six years of recent data available on degrees awarded. The model may be underestimating the number of psychology degrees to be awarded, in particular. 44 Exhibit 4.11: Degrees awarded by public higher education institutions 2001-02 to 2006-07 and projected to 2017-18 compared to projected need Position Category 2001-02 Degrees 2002-03 Degrees 2003-04 Degrees 2004-05 Degrees 2005-06 Degrees 2006-07 Degrees Cumulative Degrees Projected through 2017-18 Cumulative Projected New Need by 201810 Psychologist 50 64 51 41 44 44 204 1,808 Mental Health or Substance Abuse Counselors 374 375 391 409 360 421 4,478 7,045 Mental Health or Substance Abuse Aides/Assistants/ Technicians 1,122 1,090 1,129 1,203 1,208 1,262 14,913 51,625 The need for new degree recipients consistently falls short of the projections of new persons needed, calculated by combining estimates of annual position growth from EMSI and positions vacated using conservative estimates of annual separation rates. Thus it includes the number of replacements for people who will leave existing positions over the same period. Conclusions The difficulties experienced by program managers of behavioral healthcare services in recruiting staff to fill vacancies are expected to become more complicated in the coming years. For both professional and non-professional staff the numbers of new persons being trained to account for both persons leaving existing positions and the limited expansion anticipated are not keeping pace with the need, even conservatively estimated. Compensation Earlier in this report we presented survey data on the reasons why programs have high staff turnover rates and difficulties recruiting new staff. Across all position types and across almost all industry groups the single leading explanation is low salaries. We further presented some survey data from individual staff on their salaries. We were able to employ a second source of data on staff salaries by position type, as well as national comparisons, comparisons to surrounding states, and comparisons within areas of Oklahoma. The source of these data is EMSI. Findings Findings are presented for five position types, psychologists, mental health and/or substance abuse counselors, registered nurses, licensed practical nurses, and mental health and/or substance abuse techs. The latter are direct care positions that do not require professional degrees or licensure. Data for psychiatrists is not separately available in the EMSI data set. Data are further presented for the state of Oklahoma overall and for six regions within the State. The Tulsa and Central Oklahoma 10 This estimate does not include individuals needed to replace persons in existing positions who retire or leave the behavioral healthcare system. 45 (Oklahoma City) area are separated from the rest of the state. The rest of the counties are grouped into four quadrants, northeast, northwest, southeast, and southwest. Counties are grouped because data at an individual county level is often too small to provide reliable estimates. Exhibits 4.12 and 4.13 below present comparisons of wages. For all positions wage rates for Oklahoma are consistently below both national and regional averages. However, the disparity between Oklahoma and national wage rates is larger than the disparity with regional wage rates. Within Oklahoma, there is also variation among the six regions. In general, wages are among the highest in the Tulsa area for all position types except psychologists. The Central (Oklahoma City) region also tends to have higher rates than the other regions. Among the four regions with rural counties, there is no region that is consistently among the highest or the lowest. For two positions types there is considerable regional variation. Psychologists range from a high of $31.72 in the Southeast region to a low of $23.66 in Tulsa, a difference of 25%. MH/SA Counselors range from a high of $19.28 in Tulsa to a low of $13.61 in the Northwest, a difference of 33%. All other variations are less than 15%. Exhibit 4.12: Comparison of Average Hourly Wage: National, Regional and Oklahoma Norms for Behavioral Healthcare Positions by Type Rates may vary due to missing values. ♦ Regional rate includes the following states: Arkansas, Colorado, Kansas, Louisiana, Missouri, New Mexico, Oklahoma and Texas. Exhibit 4.13: Comparison of Average Hourly Wage: Oklahoma Norms for Behavioral Healthcare Positions by Type Position State Rate ODMHSAS Region Rates Tulsa OKC NE NW SE SW Psychologists $25.74 $23.66* $29.03 $30.47 $25.23* $31.72 $24.56* MH/SA Counselors $15.12 $19.28 $18.36 $15.67 $13.61* $15.57* $16.10* RNs $24.52 $25.25 $24.96 $22.39 $23.63 $21.34 $24.45 LPNs $15.55 $16.48 $16.06 $14.09 $14.73 $13.70 $14.58 MH/SA Techs $12.35 $12.74* $13.01 $11.37 $11.87* $11.64 $11.97 *Rates may vary due to missing values. Summary It is clear that salary rates for all positions are lower in Oklahoma than in the nation and further that Oklahomans filling these positions providing behavioral healthcare are paid less than Position State Rate National Rate Multi-State Regional Rate* Psychologists $25.74 $30.27 $26.76 MH/SA Counselors $15.12 $18.63 $16.43 RNs $24.52 $30.06 $26.98 LPNs $15.55 $19.51 $17.53 MH/SA Techs $12.35 $14.02 $12.94 46 individuals in all of the surrounding states. There is also some variation within the State. For the two position types that have the largest numbers of persons providing behavioral healthcare, MH/SA Counselors and MH/SA Techs, salaries are higher in the Oklahoma City and Tulsa areas than they are in the more rural northeast, northwest, southeast, and southwest quadrants of the state. Overview of current and future needs for behavioral healthcare workforce As we have indicated elsewhere in this report, behavioral healthcare programs have difficulty retaining and recruiting staff. There is a very large gap in the need for psychiatrists and other prescribers. It is estimated that there is a need for 697 prescribers and only 287 professionals available to meet the need, a difference of 410. While the unmet needs for other professionals and non-professionals are not as large proportionately, there are gaps in these position types as well. Additionally, the rates at which institutions of higher education in Oklahoma are producing new graduates with appropriate training are not sufficient to meet these needs, particularly with projected future growth of these positions. Furthermore, attracting new individuals into service or training is significantly handicapped by the fact that salaries for both professional and nonprofessional positions in Oklahoma are consistently lower than the surrounding states and the nation, as a whole. 47 BENEFITS & COMPENSATION Information on organizational benefits was collected via the organizational survey. Describe |
Date created | 2011-07-27 |
Date modified | 2011-07-27 |
Tags
Add tags for Oklahoma Behavioral Workforce Study Statewide 2-16-11