A Multilevel Workforce Study on Drivers of Turnover and Training Needs in State Health Departments: Do Leadership and Staff Agree?

Supplemental Digital Content is Available in the Text.

Abstract

Objectives:

To characterize agreement between senior governmental public health staff and their subordinates concerning drivers for staff turnover, and skill importance and ability.

Design:

Data were combined from 2 national surveys conducted in 2017; one was a nationally representative, individual-level survey of public health workers, and one was an individual-level survey of their leadership.

Setting:

State health agencies.

Participants:

Respondents who held scientific, nonsupervisory positions at state health agency central offices (n = 3606) were matched with leadership (n = 193) who provided programmatic area oversight.

Main Outcome Measures:

Drivers of turnover and training needs are the primary outcomes examined in this article.

Results:

Leaders and their staff agreed on the main 2 drivers of turnover (low salary and lack of opportunities for advancement), but discordance was observed for other major drivers of turnover. Substantial discordance was observed between leaders and their staff in terms of perceived staff proficiency with selected skills.

Conclusions:

This multilevel assessment of workplace perceptions offers evidence around training needs and drivers of turnover in state health agencies. Although staff and leaders agree on some major drivers of turnover, other potential reasons for leaving cited by staff, and the difference in perceptions of skills, can help target job satisfaction, training, and retention efforts in state health agencies.

Workforce studies in public health tend to focus on a single organizational level for its unit of analysis (eg, leadership perceptions of drivers of turnover, or intention to leave from the perspectives of staff).1,2 Rarely are true multilevel organizational studies conducted among public health personnel. However, multilevel analyses have been a mainstay of organizational research in other fields for at least 25 years.3 Studying supervisors and their employees as a cohesive unit allows for a greater understanding of the research aim in question.4 Multilevel analyses can, for instance, quantify the significant influence of employees' satisfaction with their leadership and with their compensation at multiple levels on turnover intentention.5,6 A multilevel study of turnover intention in the private sector reported that the significant variability in employee's turnover intention was explained by both individual-level characteristics and organizational-level characteristics, including leadership style, compensation levels, and average tenure of peers. It also confirmed a negative correlation between job satisfaction and turnover intention across levels.5 Multilevel studies further elucidate supervisor-supervisee or leadership-employee relationships, including analyses revealing that supervisor and leader satisfaction is strongly related to job satisfaction and turnover intention of the employee.7–9

The Public Health Workforce Interests and Needs Survey (PH WINS), first conducted in 2014 and again in 2017, was the first nationally representative survey of individual public health workers. Previously, the field had relied on smaller surveys or organizational-level profiles to develop conclusions about the workforce. PH WINS helped address pressing questions associated with turnover intention, job satisfaction, workplace environment, and training needs.10,11 However, it raised new interests about the employee's relatively high self-assessments of skills and abilities, given that employees are likely to rate their performance more favorably than their supervisors and leaders.12–16 In addition, PH WINS indicated substantial turnover intention, particularly among persons of color and those with shorter tenure in one's position, but without a clear indication of reasons.17,18

Both addressing drivers of turnover and training needs are regarded as primary needs in workforce development broadly19 and are potential areas of substantial disagreement. Characterizing concordance of these 2 domains may aid in retention efforts and prioritization of workforce development in general. The primary aim in this study was to examine concordance and discordance between public health leadership and their staff concerning 2 key areas: drivers of turnover and training needs.

Methods

To consider the perspectives of state health agency (SHA) leadership and their nonsupervisory staff, data were combined and matched by agency and programmatic area oversight across 2 surveys—the PH WINS and the Directors' Assessment of Workforce Needs Survey (DAWNS). Both were fielded in 2017 and analyzed in 2018.

The staff perspective

The first data source used in this study is the 2017 Public Health Workforce Interests and Needs Survey.10 In 2017, PH WINS was fielded to approximately 102 000 staff members at state and local health departments in the United States. Recruitment and definitions for PH WINS are described elsewhere.20 Approximately 17 000 of 47 000 respondents (35%) were from the SHA-Central Office frame, where respondents worked at an SHA's central office, instead of local or regional offices for that agency. In PH WINS, staff were asked which programmatic areas they worked in and whether they were nonsupervisory, scientific staff. Although PH WINS has 4 primary domains and many questions asked of respondents, 2 particular sets of items were used in this study. First, staff were asked whether they were intending to leave their agency within the next year. Of those who responded “yes,” reasons for leaving were ascertained. The second set of items used in this study relates to training skill gaps. Staff were asked both about the importance of certain skills and abilities to their day-to-day work, and their proficiency level in each. When an item was rated as somewhat or very important as well as unable to perform or beginner in terms of proficiency, a training skill gap was identified. The full items are available in the Supplemental Digital Content Appendix (available at http://links.lww.com/JPHMP/A613).

The leadership perspective

The second data source is the Directors' Assessment of Workforce Needs Survey. DAWNS was conducted in 2017 to complement PH WINS, with the idea that PH WINS provided staff perspectives on multiple areas of interest but that leadership perspectives were also needed to assess and prioritize workforce development needs in SHAs. DAWNS was fielded to 574 SHA leaders across 12 different programmatic areas (eg, maternal and child health, tobacco control, and public health laboratories (see the full list in the Supplemental Digital Content Appendix, available at http://links.lww.com/JPHMP/A613). DAWNS methodology is described elsewhere.21 SHA leaders were asked to answer questions only for a subset of their staff, specifically those who were nonsupervisory and also were scientific staff (excluding administrative and social services or other staff). Respondents were also asked to identify which programmatic areas they oversaw within their agency. Across 50 states and the District of Columbia, 250 leaders responded (response rate: 43%), of whom 193 oversaw 1 or more staff members. Respondents from US territories or freely associated states and those who did not oversee staff were excluded from analysis. DAWNS focused mainly on perceived drivers of turnover, training needs, recruitment, and staff readiness to perform activities independently (not examined in this article). Leaders were asked what drivers of turnover they saw as substantial among their own staff. Similarly, with respect to training skill gaps, leaders were asked how important particular skills and abilities were in their staff's day-to-day work. In addition, leaders were asked what percentage of their staff were proficient or expert in a given skill or ability. Language for the items around turnover and specified training needs were comparable across the 2 surveys.

Alignment and analytic approach

Unlike other forms of multilevel organizational research, PH WINS did not ask staff to identify their supervisors or leadership because of privacy and logistical concerns.20 Functionally, then, this analysis is not a supervisor-nonsupervisor multilevel study but an executive-nonsupervisor multilevel study. To examine concordance and discordance between staff and leadership perspectives, PH WINS and DAWNS data were aligned on the basis of respondents' supervisory status, identification as scientific staff, health agency, and programmatic area within each agency. Staff who worked in program areas outside DAWNS respondents' purview were excluded, as were any staff members who work outside of SHA central offices—that is, local and regional health department staff. Characteristics of respondents and multiple forms of concordance or discordance between leadership and staff perspectives are reported in the analyses. Leadership perceptions of drivers of turnover were compared with reasons cited by nonsupervisory staff actively considering leaving. The percentage of nonsupervisory staff who said that they were proficient or expert for a given task was compared with the leadership perspectives. In addition, 6 strategic skills were examined across the 2 surveys. These include the following: effectively target communications to different audiences (effective communications); collect valid data for use in decision making (data collection); identify evidence-based approaches to address public health issues (evidence-based public health); deliver socially, culturally, and linguistically appropriate programs and services (cultural competency); participate in quality improvement (QI) processes for agency programs and services (QI); and engage community assets and resources to improve health in a community (community engagement), with staff asked to self-rate their ability and leader asked to indicate the perceived percentage of their staff proficiency in each area.

Inferential comparisons between staff and leaders were conducted using Pearson's χ² test. Analytic weights were applied to leadership perceptions, weighted to the number of staff they oversee. This allowed for aggregate comparisons across agencies. Sensitivity analyses are presented in the Supplemental Digital Content Appendix (available at http://links.lww.com/JPHMP/A613) for alternative means of PH WINS <—> DAWNS alignment. The Centers for Disease Control and Prevention reviewed this study for human subjects protection and deemed it to be nonresearch.

Results

Overall, 4521 SHA staff members from PH WINS were identified as nonsupervisors and scientific staff. Of those, 3606 positively matched by agency and program area with 193 responding leaders from the DAWNS study. The 3606 matching PH WINS nonsupervisory scientific staff respondents (staff) and 193 DAWNS respondents (leaders) collectively constitute our analytic sample. In total, 91% of staff and 100% of leaders had a bachelor's degree or higher (P < .001) (Table 1). Seventy-one percent of staff were female compared with 62% of responding leaders (P = .009). As a group, leaders had more experience and were older than their staff. Sixty-seven percent of leaders were 46 years or older compared with 45% of staff (P < .001). Sixty-one percent of leaders had been in practice for 16 or more years compared with 29% of staff (P < .001). Approximately 55% of leaders had been in their position for 5 years or less compared with 70% of staff (P < .001). Leaders managed 17 staff members on median (interquartile range = 6-40).

TABLE 1. Characteristics of Leaders (DAWNS) and Staff^a (PH WINS) Respondents in State Health Agency Central Offices.

	Leaders—DAWNS (n = 193)	Staff—PH WINS (n = 3606)
Sex
Male	38%	28%
Female	62%	71%
Nonbinary or other	0%	1%
Age, y
≤30	1%	19%
31-45	32%	36%
46-60	51%	32%
≥61	16%	13%
Tenure in current position, y
0-5	55%	70%
6-10	24%	14%
11-15	13%	8%
16-20	6%	4%
≥21	3%	5%
Tenure in agency, y
0-5	26%	53%
6-10	16%	18%
11-15	16%	12%
16-20	20%	7%
≥21	21%	11%
Tenure in PH practice, y
0-5	11%	36%
6-10	10%	20%
11-15	18%	14%
16-20	19%	11%
≥21	42%	18%
	Leaders—DAWNS (n = 193)	Staff—PH WINS (n = 3306)
Highest level of education
No college degree	0%	3%
Associate degree	0%	6%
Bachelor's degree	20%	39%
Master's degree	53%	43%
Doctoral degree	27%	9%
Race/Ethnicity
American Indian or Alaskan Native	1%	1%
Asian	5%	6%
Black or African American	7%	9%
Hispanic or Latino	2%	7%
Native Hawaiian or other Pacific Islander	1%	0%
≥2 races	2%	5%
White	81%	72%
Health and Human Services region
Region 1	12%	11%
Region 2	4%	6%
Region 3	10%	7%
Region 4	13%	16%
Region 5	13%	18%
Region 6	8%	12%
Region 7	9%	6%
Region 8	12%	7%
Region 9	8%	10%
Region 10	10%	6%

Abbreviations: DAWNS, Directors' Assessment of Workforce Needs Survey; PH, public health; PH WINS, Public Health Workforce Interests and Needs Survey.

^aStaff respondents are nonsupervisory scientific staff (ie, not administrative staff or social services or other staff).

Drivers of turnover

Among staff, 1287 (36%) indicated they were considering leaving within the next year. Of those, 96% indicated that they were considering leaving for reasons other than retirement, and 4% said that they were planning to retire. Staff were asked why they were considering leaving. Low salaries and lack of opportunities for professional advancement were selected most frequently, with 49% of staff selecting each as a reason they were considering leaving. Job satisfaction (32%) and lack of acknowledgement or recognition (27%) were the next most common reasons among staff. Nationally, from a leadership perspective, 3 reasons were cited substantially more often as main drivers of turnover, including pay (76%), lack of opportunities for advancement (74%), and other opportunities outside the agency (68%). In addition, work overload or burnout (43%) and job satisfaction (24%) were also cited by leaders as main drivers of turnover.

The reasons for considering leaving were consistent across almost all states. The Figure shows the reasons identified as drivers of turnover, from both the leadership and staff perspectives, across states. Pay was among the most frequently identified driver of turnover reported by leaders in 44 states and by staff in 39 states. Leaders and staff also agreed that lack of opportunities for professional advancement was a main driver of turnover (Table 2). However, in many of the other identified drivers, differences in responses were observed. Staff selected “other opportunities outside agency” as a main reason for considering leaving in 7 states compared with the leaders' perspective in 37 states that it was among a main reason for turnover. Conversely, in 24 states, job satisfaction was identified as a main reason for turnover by staff but only in 10 states was this the case for leaders (Figure).

TABLE 2. Drivers of Turnover—Perspective From Leaders (DAWNS) and Staff (PH WINS) Respondents in State Health Agency Central Offices^a.

	Leaders Perspective—DAWNS	Staff Perspective—PH WINS
What Drives Turnover?	(n = 193)	(n = 1287)	P
Pay	76%	49%	<.001
Lack of opportunities for advancement	74%	49%	<.001
Other opportunities outside agency	68%	21%	<.001
Work overload or burnout	43%	21%	<.001
Stress	37%	21%	<.001
Retirement	34%	3%	<.001
Job satisfaction	24%	32%	.025
Weakening of benefits	18%	18%	1
Satisfaction with your supervisor	16%	21%	.108
Lack of acknowledgement or recognition	14%	27%	.001
Lack of flexibility	13%	17%	.163
Leadership changeover	12%	14%	.452
Lack of training	5%	16%	.001

Abbreviations: DAWNS, Directors' Assessment of Workforce Needs Survey; PH WINS, Public Health Workforce Interests and Needs Survey.

^aData are shown as DAWNS versus PH WINS. Interpret as follows: In total, 37% of leaders identified stress as a driver of turnover and 21% of staff who are considering leaving selected it as a reason they are considering leaving. In total, 193 DAWNS and 1287 PH WINS respondents answered to these questions. PH WINS respondents only were presented with these items if they indicated they were considering leaving their organization in the next year.

FIGURE.

Drivers of Turnover by Number of States—Perspective From Leaders (DAWNS) and Staff (PH WINS) in State Health Agencies^a

Abbreviations: DAWNS, Directors' Assessment of Workforce Needs Survey; PH WINS, Public Health Workforce Interests and Needs Survey.

^aDAWNS bar represents leaders' perceptions of large/very large drivers of turnover in their staff. PH WINS bar represents self-reported reasons for considering leaving among staff who are considering leaving their organization in the next year for any reason. Bars represent the number of states that have a given driver of turnover in the top 3 most highly cited drivers of turnover. (Interpret as: In 24 states in PH WINS, job satisfaction was in the top 3 drivers of turnover; in 10 states in DAWNS, job satisfaction was in the top 3 drivers of turnover.)

Training skill gaps

Beyond drivers of turnover, both PH WINS and DAWNS assessed various training skill gaps. As Supplemental Digital Content Appendix Table 1 (available at http://links.lww.com/JPHMP/A613) shows, substantial agreement exists that each of the 6 strategic skills examined are important to the public health workforce's day-to-day work, both from the staff and leadership perspectives. Overall, 94% of leaders and 88% of staff said that effective communication was somewhat or very important. Data collection was somewhat or very important to 100% of leaders and 94% of staff; evidence-based public health was similarly important for 95% of leaders and 84% of staff; cultural competency was somewhat or very important to 98% of leaders and 76% of staff. Eighty-five percent of leaders and 71% of staff rated QI as somewhat or very important; and 96% of leaders and 79% of staff said that community engagement was somewhat or very important.

Substantial discordance was observed between leaders and staff in terms of staff proficiency with the selected skills (Table 3). Overall, leaders considered that 62% of their staff were proficient or expert in effectively targeting communications to different audiences compared with 75% of staff respondents (P = .001). Similarly, leaders rated an average of 69% of their staff as proficient or expert in collecting valid data for decision making compared with 82% of staff (P < .001). Leaders and staff generally agreed on ability related to evidence-based public health (65% vs 69%, P = .24), and leaders considered that their staff were more capable in 3 areas compared with staff: cultural competency, QI, and community engagement. Leaders said that 69% of their staff, on average, were proficient or expert in cultural competency compared with 60% of their staff (P = .01). Leaders reported that 54% of their staff were proficient or expert in QI compared with 41% of staff (P = .004). Finally, leaders felt that 58% of their staff, on average, were proficient or expert at community engagement compared with 50% of staff indicating proficiency or expertise in this area (P = .03).

TABLE 3. Perceived Skill Proficiency of Strategic Public Health Skills—Perspective From Leadership (DAWNS) and Respondents (PH WINS)^a.

Perceived Skill	Leadership Perspective—DAWNS	Staff Perspective—PH WINS	P
Collect valid data for use in decision making	69%	82%	<.001
Deliver socially, culturally, and linguistically appropriate programs and services	69%	60%	<.05
Identify evidence-based approaches to address public health issues	65%	69%	.24
Effectively target communications to different audiences	62%	75%	<.01
Engage community assets and resources to improve health in a community	58%	50%	<.05
Participate in quality improvement processes for agency programs and services	54%	41%	<.01

Abbreviations: DAWNS, Directors' Assessment of Workforce Needs Survey; PH WINS, Public Health Workforce Interests and Needs Survey.

^aInterpret as follows: In total, leaders thought 62% of their staff were able to “effectively target communications to different audiences” at a proficient or expert level, on average. Comparatively, 75% of staff thought they were able to “effectively target communications to different audiences” at a proficient or expert level. There were 193 DAWNS respondents and 3528 PH WINS respondents to these questions.

Discussion

Although the multilevel organizational research approach is common in other fields,4,14 this is the first similar large-scale study conducted in governmental public health. This approach provides value similar to that of 360-degree reviews commonly employed in leadership development programs, with the multirater feedback becoming the gold standard of such assessments.3,17 With the advent of PH WINS as the first major nationally representative workforce survey of public health practitioners in 2014, the opportunity to examine perspectives among staff became a possibility on a national scale for the first time. However, because the scope of PH WINS (>42 000 respondents during the 2017 fielding) and concerns about privacy and identifiability, a true linked supervisor-subordinate study was impractical within PH WINS alone. Our study combining 2 data sources is the first one to derive an alternative mode of linking public health staff to agency leadership by merging 2 national workforce surveys in support of the multilevel questions that had arisen from PH WINS.

Evidence presented in this article highlights 2 major domains of interest. First, our findings indicate that low salaries and lack of opportunities for advancement are considered the top 2 drivers of turnover for public health. Generally, across all groups and regions, staff and their leaders agreed that these were major drivers. These major drivers have been previously identified in the literature.22,23 However, our study indicates that other reasons that staff were considering leaving their organization were not necessarily well-represented among leadership perspectives captured in this study. For instance, relatively few leaders reported that job satisfaction and lack of acknowledgement were major drivers of turnover. Those drivers were frequently cited as potential reasons for leaving by staff who were actively considering leaving their agency at the time of completing the survey. While the fact that salaries matter is reflected by both groups, leaders need not think that only salaries (something often out of their control) or opportunities for professional advancement (also likely not within their full control) are driving employees to leave their jobs.

Our analysis did identify drivers of turnover that may be within the control of leaders to act upon and can reduce voluntary turnover.24 Job satisfaction and perceived organizational support on the part of supervisors and managers are correlated with better relationships with their subordinates and have been identified as part of the strategies to decrease the expected turnover among governmental public health workers.25–27 Specifically, leaders can work to increase perceptions of organizational support through implementing staff-supportive policies and increase expectations for open communication between staff, supervisors, and managers. Public health leaders could implement nonfinancial motivators including work schedules that consider telework opportunities, delegation of the lead role on certain projects, and targeted praise and recognition, among others, to facilitate opportunities for job satisfaction, growth, and advancement. In addition, identifying projects or activities that involve and increase employee engagement can help developing the skills and potential of the workforce. Furthermore, regularly evaluating job satisfaction can offer evidence needed for the identification of problem areas needing improvement and the implementation of an improvement plan and ongoing QI.

The next major area we explored in this study was related to concordance in the perceived importance and ability of 6 particular strategic skills needed to complement specialized skills. Although there was general agreement among leaders and staff that these areas were universally important in their day-to-day work, our study identified disagreement in how skilled staff were in certain areas. Compared with leadership perceptions, staff tended to consider themselves more competent in 2 areas, effective communication and data for decision making. Conversely, there were 3 skills where leaders felt that their staff were more capable than self-assessments by staff, including culturally competent service delivery, participating in QI, and community engagement. Agency leaders can help narrow the gap between the perceived ability and the level of importance by facilitating targeted training and developmental opportunities in the areas perceived as critical where ability is not high, facilitating the use of those skills into day-to-day work, and introducing job performance measures that can guide ongoing performance improvement and nonfinancial incentives for employees demonstrating the use of those skills in the workplace. National membership organizations, public health training centers and funders, and decision makers within SHAs can continue to enhance training availability in these strategic skill areas.

Implications for Policy & Practice

• The public health workforce is experiencing turnover both due to high levels of retirement and due to staff considering leaving their position for other reasons.

• This article assesses how well leaders and their staff agree on the drivers of turnover. While pay and lack of opportunities are generally agreed upon, staff also cite several workplace and quality-of-life factors in their reasons for considering leaving.

• Increasing pay to increase retention may not be practical, but leaders can look to improve other areas of job satisfaction and engagement to increase retention.

• Leadership and staff tend to agree on importance of skills, but there is poor concordance in ability levels. Staff tend to overestimate ability compared with their leaders.

• More trainings can be offered in the highest need areas to address on-the-job training needs for the governmental public health workforce.

Limitations

This study has limitations. First, and most substantively, our development of an alternative means of linking leaders and their staff is a proxy and not an exact link between supervisors and supervisees that is more common in multilevel studies. Although this approach is superior to general cross-sectional comparisons between all leaders and all responding staff members, it is imprecise as it focuses only on executives and all their subordinates, beyond merely direct reports.6 In addition, although DAWNS had a strong response rate and was fielded nationally to more than 500 public health leaders, certain programmatic areas were not well-represented in the final response pool (specifically around clinical care provision, vital statistics, and emergency medical services). The full list of participant groups is available in Supplemental Digital Content Appendix Table 2 (available at http://links.lww.com/JPHMP/A613). There is also a potential for measurement error. While both studies were piloted, some areas are open for interpretation, which is common on self-reported surveys.

Conclusions

We developed a multilevel assessment of workplace perceptions around training needs and drivers of turnover, 2 of the most critical areas in workforce development.28 Data show that staff and leadership are in agreement on some, but not all, major drivers of turnover. Leadership can do more to improve job satisfaction and recognize their staff's accomplishments—2 mutually reinforcing efforts that can improve retention. Given a limited ability to raise salaries or create new management-level jobs, these concrete foci are the most likely paths for limiting voluntary turnover—a significant concern in the coming years.29