Leadership, Staff Empowerment, and the Retention of Nursing Assistants: Findings From a Survey of U.S. Nursing Homes

Abstract

Objectives

We examined the relationship between nursing assistant (NA) retention and a measure capturing nursing home leadership and staff empowerment.

Design

Cross-sectional study using nationally representative survey data.

Setting and Participants

Data from the Nursing Home Culture Change 2016–2017 Survey with nursing home administrator respondents (N = 1386) were merged with facility-level indicators.

Methods

The leadership and staff empowerment practice score is an index derived from responses to 23 survey items and categorized as low, medium, and high. Multinomial logistic regression weighted for sample design and to address culture-change selection bias identified factors associated with 4 categories of 1-year NA retention: 0% to 50%, 51% to 75%, 76% to 90%, and 91% to 100%.

Results

In an adjusted model, greater leadership and staff empowerment levels were consistently associated with high (76%−90% and 91%−100%) relative to low (0%−50%) NA retention. Occupancy rate, chain status, licensed practical nurse and certified nursing assistant hours per day per resident, nursing home administrator turnover, and the presence of a union were also significantly associated with higher categories of retention (vs low retention).

Conclusions and Implications

Modifiable leadership and staff empowerment practices are associated with NA retention. Associations are most significant when examining the highest practice scores and retention categories. Nursing homes seeking to improve NA retention might look to leadership and staff empowerment practice changes common to culture change.

Nursing home (NH) workforce instability, particularly for nursing assistants (NAs), is a significant and costly problem.^1,2 High NA turnover is expensive and associated with lower quality of care^2–5 and more deficiencies.⁶ Annual national NA retention rates—the percentage of staff employed on January 1 who are still employed on December 31—have been estimated at 62.5,⁷ and the American Health Care Association reports a median retention of 69.6% among its members.⁸

A number of studies have examined factors that promote NA retention and discourage turnover. Many of these focus on elements of work environments and supervision that promote job satisfaction because satisfaction is associated with lower turnover.^9,10 For example, Choi and Johantgen examined 10 National Nursing Assistants Survey items about supervision, such as “listens to [certified nursing assistant’s] CNA’s concerns about resident’s care” and “is supportive of progress in the CNA’s career,” and found a significant relationship between supportive supervision and both satisfaction (positive) and intent to leave (negative).¹¹ Bishop and colleagues also report that NA satisfaction was associated with having opportunity for teamwork and that NAs who felt respected, rewarded, and valued by their employer were less likely to be dissatisfied.¹²

The ability to retain NAs will become more critical with the growing focus on training motivated NAs to provide person-centered care.¹³ Research is increasingly focused on leadership in NHs with the recognition that leadership is key to successful innovation toward person-centered care, particularly in regard to employee socialization and achieving buy-in for implementing culture change practices,^13–15 that is, practices aimed at promoting person-centered care and deinstitutionalizing NHs.¹⁶ In an integrative review of the NH literature, Brodtkorb and colleagues found evidence that leadership plays a key role in successful innovation.¹⁷ They endorse the idea that “leadership should change from focusing on authority management toward knowledge management” (Uh-Bien 2007 as cited by Brodtkorb et al^17(p9)).¹⁸

The above approach to leadership appears to be compatible with practices targeting staff empowerment. Key to promoting person-centered care and NH deinstitutionalization is workforce redesign in which leadership practices are implemented to flatten the hierarchy and enhance the engagement and empowerment of staff.¹⁹ With the exception of increases in wages and benefits, the interventions recommended to improve the workplace and stabilize staff are similar to the leadership and staff empowerment interventions used to bring about culture change.^20,21 For example, 1 study found that NHs with consensus management and consultative autocrat leadership styles have lower turnover rates compared to those with an autocratic leadership approach.⁹

Although a number of studies have revealed significant relationships between NA turnover and various leadership and staff empowerment practices, including involving NAs in care planning^21,22 and enhancing communication with management,^10,23 others have yielded mixed results. For example, Weiner and colleagues found having a mentor in their first year of work predicted NA tenure but many other leadership and staff empowerment measures did not, such as receiving respect, working in teams, and the ability to decide how to complete the work.²⁴ Another study that measured NH culture using the Competing Values Framework found no relationships between various cultural values and NA turnover. That is, neither hierarchical cultural values (stability, continuity) nor group values (teamwork, morale, cohesion) were important for NA turnover.²⁵

Using 2009–2010 data from the first wave of a national survey of NHs, we examined the association between NA retention and staff empowerment. The 7 examined staff empowerment practices were as follows: staff working together to cover shifts, cross-training, involvement in planning social events, NA involvement in quality improvement teams, NA knowledge of resident care plan changes, rewards for training and education, and NA choice of which residents to care for.²⁶ In our ordered logistic regression model, a greater composite empowerment score was associated with higher retention. With bivariate analysis, we observed a positive relationship between retention and each of the 7 items.²⁶

Our conceptualization of staff empowerment encompasses concepts that are highly salient in the staffing stability literature. Specifically, the staff empowerment and leadership domains capture management and leadership practices that support staff independence, involvement in care planning and decision making, 2-way communication, training, and respect for staff. Therefore, the current study uses an enhanced measure of staff empowerment with 13 items and 10 new items on NH leadership (see Supplementary Table S1). We update our previous analysis using 2016–2017 data and build on it by examining the relationship between NA retention and a new measure that combines these leadership and staff empowerment items from practices considered important to NH culture change.²⁷ We hypothesize that after adjusting for NH characteristics generally associated with staff stability, NHs having higher levels of leadership and staff empowerment practices will have greater NA retention than facilities with lower levels of practices.

Methods

Study Design and Population

This cross-sectional analysis uses the second wave of a national study on culture change implementation.²⁷ We administered surveys to a stratified proportionate random sample of US NHs during the first wave in 2009–2010 using the 2008 Online Survey Certification and Reporting (OSCAR) system for the sampling fame. NH administrators received by mail a questionnaire, self-addressed return envelope, cover letter with a web-response option, a letter of support from associated professional organizations, and when relevant and available, a support letter from chain leadership. At 2 weeks, nonrespondents were called, e-mailed, or faxed. We received responses from 2215 NH administrators. The 2142 NHs that were still operational in 2016 were administered a follow-up survey in 2016–2017. Administrators had the option of completing the survey by web, mail, or telephone. A wave 2 response rate of 74% was obtained (n = 1584). Nonresponse bias was not detected in either wave.²⁷ Seventy-one NHs (4.5%) declined to fill out the full survey, and the abbreviated survey did not include all items needed for analyses and were therefore dropped. Our starting sample was 1513. Further details about the development of the survey and data collection are reported in Miller et al.²⁷

Retention Outcome Variable

NH administrators were asked “About what percent of the NURSING ASSISTANTS who were employed at your nursing home TODAY has worked at the nursing home for AT LEAST 12 MONTHS?: 0% to 50%; 51% to 75%; 76% to 90%; 91% to 100%.” The development of these 4 categories was informed by earlier cognitive-based interviews with 50 NH administrators not part of this larger survey sample, which revealed that continuous data on staff retention could not be reliably obtained.²⁸

Leadership and Staff Empowerment

This study concentrates specifically on the domains of leadership and staff empowerment. We combined responses to 13 survey items to create a staff empowerment index and 10 items to create a leadership index. The items and responses are listed in Supplementary Table S1. Response options to each of the items were never/rarely (1), sometimes (1), often (2), and almost/always (3).

The Holistic Approach to Transformational Change (HATCh) model²⁹ suggests 6 domains through which organizations can work to achieve culture change within NHs.²⁷ The domains included in the 2016–2017 survey were leadership and staff empowerment, physical environment, person-centered care, and community engagement. Items were developed based on the HATCh model with the guidance and review of an expert advisory committee including researchers and leaders from national NH and culture change organizations.²⁷ Items were identified for review and inclusion from the Commonwealth survey, Artifacts of Culture Change, and Kansas PEAK 2.0 criteria.²⁷ The literature on staff empowerment and leadership^{9,12–14,26,30–32} further informed the selection of these domain items. During survey design, items were refined and tested through structured cognitive interviewing and follow-up probing at 15 randomly selected NHs surveyed in 2009–2010 that varied by size, profit status, chain membership, and hospital affiliation.^28,33

The 10 leadership items capture 2-way communication, staff involvement in decision making, staff education and training, respect for workers, positive leadership-staff relationships, and coaching.²⁷ The 13 staff empowerment items encapsulate NA involvement and participation in care planning, independence, and practices that communicate that their care role is valued, such as consistent assignment and introducing new staff to residents.

Separate leadership and staff empowerment indexes were created by summing the responses of each item together within each domain. Missing values were imputed for up to 2 missing items within each domain by assigning the average of the nonmissing items within the domain to the missing items prior to summing. NHs that had 3 or more missing values within either domain were dropped from the study sample. Overall, 7.0% of facilities were dropped because of missing responses, reducing the study sample to 1407. As reported by Miller et al.,²⁷ the internal consistency of each index was high, with a McDonald omega of .83 for leadership and .86 for staff empowerment.^27,34 For these analyses, the individual indexes were standardized and combined to form a single composite measure of leadership and staff empowerment. The composite score was then divided into quartiles. The middle 2 quartiles were collapsed into a single category because early analyses found their associations with the study outcome were not significantly different. Therefore, our final 3-category measure reflects low, medium, and high leadership and staff empowerment practices.

Covariates

Important covariates indicated in previous research on NA turnover and retention that are included in the model come from several sources. Profit status,^26,35–37 facility size,^25,35 chain status,³⁶ and level of resources in the form of percentage of residents with Medicaid and Medicare^22,37,38 were taken from the Certification and Survey Provider Enhanced Reporting (CASPER). Other variables derived from CASPER included occupancy rate,^22,26 registered nurse, licensed practical nurse, and CNA hours per day per resident,^21,26,35 and the Herfindahl Index for NH beds in a county indicating the local competition for staff. The lower the market concentration, the more competition there should be among facilities for staff.³⁶ County unemployment,^22,24,39 home health agencies per 1000 people aged ≥65 years,²⁴ and NH beds^24,39 came from the Area Health Resource File. Whether or not the facility is a CCRC,⁴⁰ administrative turnover,^26,41 and NA union membership^21,36 were taken from the 2016–2017 survey. These covariates, their sources, and how they are coded are provided in Supplementary Table S2.

Analysis

Weights were used throughout to adjust for the sampling design, though any reported sample sizes reflect unweighted numbers. Because this was an observational study rather than a study that randomly assigned facilities into treatments (ie, varying levels of leadership and staff empowerment practices), we also used propensity score modeling to generate inverse probability weights to be used in the outcome model in order to reduce confounding. The propensity score model calculated the probability of being in the medium and high leadership and staff empowerment categories compared with the low category after controlling for a number of characteristics known to be associated with either the outcome (NA retention) or both the outcome and treatment. The goal was to further weight the sample so that variables associated with the outcome and (potentially) an NH’s selection of the treatment (ie, level of culture change practice) were more evenly distributed across the treatment groups (see Supplementary Table S3 for a list of the variables used in the propensity score models as well as their post within-treatment distributions). A multinomial logit model was used to examine retention, with the lowest category of retention (0%−50%) serving as the reference category. It included many of the same covariates that were used in the propensity score model. Finally, using model data we estimated the predicted probabilities for each of the 12 subgroups of NA retention quartiles by the 3 levels of leadership and empowerment practice.

Analyses were conducted using Stata 15. In compliance with our funder’s Resource Sharing Policy, the analytic data file used for the analyses as well as additional information about the data and methods can be found in Brown’s Digital Repository (https://repository.library.brown.edu/studio/item/bdr:956724/).

Results

The standardized composite leadership and staffing empowerment index score had a mean of 60.7 and a standard error of 0.30. Nearly 5% (4.7%) of NHs had NA retention rates of 91% to 100%, 25.7% had rates of 76% to 90%, 45.9% had rates of 51% to 75%, and 23.7% had NA retention rates of 0% to 50%. Figure 1 illustrates the relationship between low, medium, and high leadership and staff empowerment levels and NA retention rates, showing that these practices are positively associated with retention. Table 1 shows the population-weighted distribution of facility- and market-level characteristics for facilities with low, medium, and high leadership and staff empowerment levels (prior to propensity-score adjustment).

Fig. 1.

Nursing home leadership and staff empowerment practice levels by nursing assistant retention categories.

Table 1.

Nursing Home Characteristics by Leadership and Staff Empowerment Levels (n = 1407)

Nursing Home Characteristic	Leadership and Staff Empowerment Levels Weighted,* % or Mean (Standard Error)
	Low (Weighted n = 3247; Unweighted n = 366)	Medium (Weighted n = 6265; Unweighted n = 700)	High (Weighted n = 3019; Unweighted n = 341)
NH is for-profit	68.8	67.7	69.6
Small size, <80 beds	32.6	33.3	36.7
Medium size, 80–120 beds	29.4	28.1	30.6
Large size, >120 beds	37.9	38.5	32.5
Occupancy rate	80.0 (0.00)	82.1 (0.00)	82.5 (0.00)
NH is part of chain	54.6	56.7	54.9
1 NHA in the last year	46.8	55.3	65.8
2 NHAs in the last year	31.0	27.3	21.4
3+ NHAs in the last year	22.1	17.3	12.7
RN hours per day per resident	0.45 (0.02)	0.45 (0.01)	0.45 (0.01)
LPN hours per day per resident	0.81 (0.01)	0.85 (0.01)	0.80 (0.01)
CNA hours per day per resident	2.29 (0.03)	2.35 (0.02)	2.38 (0.03)
County unemployment rate	4.8 (0.07)	4.9 (0.05)	5.0 (0.08)
County home health agencies	0.29 (0.01)	0.26 (0.01)	0.27 (0.01)
Percentage of residents with Medicare	13.6 (0.70)	12.3 (0.36)	13.0 (0.57)
Percentage of residents with Medicaid	60.8 (1.14)	60.0 (0.78)	61.7 (1.04)
Continuing care retirement community	16.8	20.1	17.9
Herfindahl Index County NH beds	0.22 (0.01)	0.22 (0.00)	0.25 (0.01)
CNA union	22.7	22.1	22.6

LPN, licensed practical nurse; RN, registered nurse.

A description of each covariate and data source is listed in Supplementary Table S2.

^*These are weighted for the sampling design and do not include the inverse propensity score weights.

The relative risk ratios (RRRs) reflecting the associations between differing levels of NH leadership and empowerment practices and NA retention are shown in Table 2. Facilities with medium and high leadership and staff empowerment scores (compared to those with low scores) have a significantly greater likelihood of having higher retention (76%−90% and 91%−100%) vs lower retention (0%−50%). NHs in the medium-level category (vs those with low scores) were nearly twice as likely to report retention of 76% to 90% relative to 0% to 50% (RRR = 1.86, CI: 1.25–2.77) and 3 times as likely to report retention of 91% to 100% relative to 0% to 50% (RRR = 3.00, CI: 1.20–7.51). In NHs with a high level of leadership and staff empowerment practices (vs a low level), the corresponding RRR of retention of 76% to 90% relative to 0% to 50% retention was 2.81 (CI: 1.73–4.58) with a 6-fold increase in the likelihood of being in the highest relative to lowest retention category (RRR = 6.00, CI: 2.26–15.05). Differences in the leadership and staff empowerment index did not significantly differentiate low retention (50% or below) from NHs with retention rates from 51% to 75%.

Table 2.

Weighted Multinomial Logit Results for Nursing Assistant Retention Above 50% Compared With Retention Below 50%

	Nursing Assistant Retention, RRR (95% CI)
	51%-75%	76%-90%	91%-100%
Leadership and Staff Empowerment Index
Low (first quartile)	Ref	Ref	Ref
Medium (second and third quartiles)	1.07 (0.77–1.47)	1.86*** (1.25–2.77)	3.00** (1.20–7.51)
High (fourth quartile)	1.39 (0.91–2.13)	2.81*** (1.73–4.58)	6.00*** (2.26–15.95)
NH is for-profit	0.93 (0.65–1.33)	0.94 (0.62–1.42)	0.80 (0.39–1.66)
Number of NH beds
<80	Ref	Ref	Ref
80–120	0.93 (0.64–1.36)	0.88 (0.57–1.34)	0.42* (0.17–1.03)
≥120	0.92 (0.63–1.36)	0.86 (0.56–1.31)	0.60 (0.27–1.31)
Occupancy rate	1.04 (0.99–1.10)	1.08** (1.01–1.15)	1.02 (0.91–1.15)
NH is part of chain	1.03 (0.74–1.43)	0.83 (0.57–1.20)	0.35*** (0.18–0.71)
NH administrators in the last year
1 NHA	Ref	Ref	Ref
2 NHAs	0.79 (0.56–1.13)	0.59** (0.39–0.90)	0.40*** (0.16–0.97)
≥3 NHAs	0.62** (0.41–0.93)	0.45*** (0.28–0.75)	0.62 (0.26–1.48)
RN hours per day per resident	1.03 (0.84–1.25)	1.06 (0.86–1.30)	0.84 (0.59–1.20)
LPN hours per day per resident	0.87 (0.72–1.05)	0.88 (0.73–1.07)	0.69* (0.47–1.02)
CNA hours per day per resident	1.11 (0.93–1.33)	1.13 (0.93–1.38)	1.42*** (1.12–1.81)
County unemployment rate	1.06 (0.93–1.20)	1.10 (0.95–1.28)	0.95 (0.73–1.23)
County home health agencies	0.76 (0.45–1.28)	0.85 (0.48–1.48)	0.59 (0.21–1.67)
Percent of residents with Medicare	0.99 (0.97–1.01)	0.99 (0.97–1.01)	0.99 (0.94–1.03)
Percent of residents with Medicaid	0.99 (0.98–1.00)	1.00 (0.98–1.01)	1.01 (0.99–1.03)
Continuing care retirement communities	1.31 (0.83–2.04)	1.14 (0.68–1.91)	0.90 (0.37–2.22)
Herfindahl Index County NH beds	0.92 (0.52–1.64)	0.64 (0.33–1.28)	0.31 (0.07–1.51)
Facility CNA union status	1.13 (0.76–1.66)	1.27 (0.81–1.99)	3.85*** (1.84–8.05)

LPN, licensed practical nurse; Ref, reference group; RN, registered nurse.

^*P < .10

^**P < .05

^***P ≤ .01.

For ease of interpretation of this variable of interest, we also present the predicted probabilities of NA retention for higher vs lower leadership and staff empowerment scores, controlling for all other factors in the regression model (see Table 3). Scores at the third and fourth quartiles had consistent and statistically significantly higher predicted probabilities of higher retention than did scores at or below the median. That is, better leadership and staff empowerment practices are associated with higher probability that an NH will have retention rates higher than 75%, rather than 75% or lower.

Table 3.

Predicted Probabilities: Nursing Assistant Retention

	Mean Probability (95% CI)
	0%-50%	51%-75%	76%-90%	91%-100%
Leadership and Staff Empowerment Index
Low (first quartile)	0.28 (0.24–0.33)	0.51 (0.46–0.57)	0.18 (0.14–0.23)	0.02 (0.01–0.04)
Medium (second and third quartiles)	0.23 (0.20–0.26)	0.45 (0.41–0.48)	0.27 (0.24–0.31)	0.05 (0.03–0.06)
High (fourth quartile)	0.18 (0.13–0.22)	0.44 (0.39–0.50)	0.31 (0.26–0.36)	0.07 (0.04–0.10)

Although not our focus here, the analysis sheds light on the association with retention of the covariates included in the model. Higher NH occupancy was significantly associated with greater NA retention for the 75% to 90% category (vs 0%−50%; RRR = 1.08, CI: 1.01–1.15) but not with higher retention at 91% to 100%. Greater licensed practical nurse hours per resident day and CNA hours per resident day had significant, reverse relationships to retention at the highest compared with lowest retention categories. Higher NH administrator turnover was associated with lower NA retention. Facilities with 2 administrators in the last year had a lower likelihood of having higher (vs lower) retention (RRR for 76%−90% = 0.59, CI: 0.39–0.90, and RRR for 91%−100% = 0.40, CI: 0.16–0.97). Those with 3 or more administrators in the last year had a lower likelihood of having an NA retention rate of 51% to 75% or 76% to 90% (compared with 0%−50%; RRR = 0.62, CI: 0.41–0.93, and RRR = 0.45, CI: 0.28–0.75, respectively). Finally, the presence of an NA union is independently significantly associated with greater retention in the highest category (91%−100%) compared with the lowest (0%−50%) by a factor of 3.85 (CI: 1.84–8.05).

Although culture change research has been criticized because of its lack of control for selection bias (ie, facilities with more culture change already have higher quality), our sensitivity analysis showed very similar effects when we excluded our adjustment for selection bias (ie, excluded the inverse probability weights).

Discussion

After adjusting for covariates and for an NH’s probability of adopting the culture change–aligned practices studied, the leadership and staff empowerment score was the factor most significantly associated with NA retention. This study’s findings are consistent with others that have found a combination of managerial style, organizational culture, and staff empowerment practices to be associated with staff stability.^25,26 Together with these findings, our study findings strongly support the notion that interventions focusing on improving leadership and staff empowerment practices (see Supplementary Table S1) may be beneficial to retaining NA staff. These findings speak to the value of NA team membership, inclusion, and involvement in decision making. There are many opportunities to incorporate such practices, including involving NAs in quality improvement teams and resident care plan meetings, and providing rewards for extra training or education. The majority of nursing homes now assign new NAs to a peer mentor, but those that do not may benefit from incorporating this practice. Practices that promote shared and open decision making such as having formal processes that allow NAs to contribute ideas on improving resident care and sharing facility-wide management decision-making power with staff may also promote staffing stability.

It is important to recognize that such attempts to empower NAs and to transition away from a top-down management approach represent disruptions to the status quo. Others have noted that the relationship between these changes and retention may be complicated if experienced negatively by staff as challenging and disruptive. Culture change adoption is not easy and requires new learning and adaptive leadership.^13,14 This is a cross-sectional analysis, so it is possible that NHs experienced a flight of NAs during the implementation of new leadership and staff empowerment practices that our data do not capture.

As expected, greater turnover of NHAs was associated with lower NA retention. Also, being part of a chain was significant in distinguishing facilities with very high retention, but being a for-profit NH was not. Consistent with previous findings, we found that a higher occupancy rate was significantly associated with greater retention for facilities at 75% to 90%, compared with 0% to 50%. At higher retention levels compared with the lowest, more CNA hours per patient-day is associated with retention.

Unlike recent findings based on longitudinal work of improved retention with higher unemployment rates,³ we found no independent significant relationship between unemployment and NA retention in this cross-sectional analysis. The finding that facilities with unionized NAs are more likely to have very high retention (91%−100%) is expected based on previous work suggesting that unions improve unmeasured job factors like pay, benefits, and job security.³⁶ Our findings on retention are consistent with those of Brannon and colleagues³⁶ that a union presence is highly associated with very low turnover, though they also report a modest, nonsignificant positive relationship between union status and very high turnover. Contrary to their finding, we see no evidence of an association between union status and low retention.

Limitations

The data analyzed are cross-sectional, so the results do not measure the effect of change in any of the independent variables, most notably the leadership and empowerment index. It is possible that staff empowerment and leadership practices could be put in place or altered in tandem with other job quality improvements that we do not capture but that they could act as proxy for, such as improved compensation. There are only 63 facilities in the highest retention category of 91% to 100%; however, despite this relatively small population, the effect for the leadership and empowerment index on the probability that an NH achieves that category is estimated with sufficient precision (the confidence interval for the coefficient is narrow).

During cognitive testing, the study team found that between NH administrators and directors of nursing administrators were the better respondents,³³ yet they may still be subject to social desirability bias. We are unable to access the impact of empowerment and leadership practices from the perspectives of nursing assistants, which would strengthen our understanding of the relationship to retention. It is certainly possible that perception of the extent to which these practices are in place in an NH could differ between NH administrators and NAs.

Conclusion

This study used 23 items from a nationally representative NH survey to thoroughly examine whether culture change practices focused on leadership and empowering NAs are associated with NA retention. The leadership and staff empowerment level was the factor in our model most significantly associated with high NA retention, suggesting that these modifiable practices may support NA retention.