Nurse Aide Empowerment Strategies and Staff Stability: Effects on Nursing Home Resident Outcomes

The Gerontologist Vol. 45, No. 3, 309 317 Copyright 2005 by The Gerontological Society of America Nurse Aide Empowerment Strategies and Staff Stability: Effects on Nursing Home Resident Outcomes Theresa Teta Barry, PhD, 1 Diane Brannon, PhD, 1 and Vincent Mor, PhD 2 Purpose: This study examines the moderating effect of staff stability on the relationship between management practices used to empower nurse aides and resident outcomes in a multistate sample of nursing homes. An adaptation of Kanter s theory of structural power in organizations guided the framework for the model used in this study. Design and Methods: Management practices and nurse aide staff stability measures were taken from a survey of directors of nursing (n = 156) and day-shift charge nurses (n = 430) in a stratified random sample of nursing facilities in Maine, Mississippi, New York, and Ohio (n = 156). Facility risk-adjusted pressure ulcer incidence rates and social engagement scores, taken from the Minimum Data Set, served as resident outcome measures. A hierarchical linear regression model was used to test the moderating effect of staff stability. Results: The situation in which higher numbers of rewards were given to nurse aide staff was associated with lower incidence of pressure ulcers, whereas the situation in which nurse aides had more influence in resident care decisions was associated with higher aggregate social engagement scores. The inclusion of nurse aide staff stability in the regression model provided additional explanatory information for the relationship between management practices and resident outcomes. Facilities experiencing low turnover and high retention were associated with lower pressure ulcer incidence, whereas facilities with high turnover and high retention were associated This research was supported by Grant AG11624-01A1 from the National Institute on Aging to Vincent Mor, Principal Investigator. We acknowledge Dr. Joe Vasey and Chris Bostrup-Jensen for their assistance in preparing the data for analysis. We also thank the anonymous reviewers for their contributions to improving this manuscript. Address correspondence to Theresa Teta Barry, PhD, 116 Henderson Building, The Pennsylvania State University, University Park, PA 16802. E-mail: txb13@psu.edu 1 Department of Health Policy and Administration, The Pennsylvania State University, University Park, PA. 2 Center for Gerontology and Health Care Research, Brown University, Providence, RI. with higher social engagement scores. Implications: This study suggests that certain management practices used to empower nurse aides can influence resident outcomes. Further, effects of nurse aide staff stability vary with respect to the physical versus psychosocial nature of the outcome. Key Words: Managerial effectiveness, Long-term care, Nursing facility performance, Rosabeth Kanter In this article we explore the moderating effect of staff stability on the relationship between management practices used to empower nurse aides in nursing homes and facility performance measured in terms of resident outcomes. Nurse aides serve as essential providers of care and constitute a major portion of the staff in nursing facilities. Although the significance of the nurse aide role in the provision of care is widely acknowledged by both researchers and health care providers in the industry, these workers experience stressful working conditions and little career mobility, and they are among the lowest paid workers in the health care field (Stone & Salsberg, 2001). It is not surprising that, given these circumstances, the nursing home industry experiences high nurse aide turnover rates and faces considerable difficulty in recruiting and retaining a quality nurse aide workforce. With the impending demand for these workers, coupled with the documented relationship between nurse aide turnover and quality of care (Anderson, Issel, & McDaniel, 2003; Atchison, 1998; Burnfeind & O Connor, 1992), industry leaders continue to recognize the importance of investigating practicable solutions to the problem (Stone, 2001). A wage increase has been deemed as one possible solution to the nurse aide turnover problem in longterm-care facilities (Schnelle, McNees, Simmons, Agnew, & Crooks, 1993). However, cost constraints and low reimbursement levels that confront the nursing home industry render this solution unlikely in today s health care environment. Moreover, it appears that dissatisfaction among nurse aides Vol. 45, No. 3, 2005 309

involves more than just poor wages; these workers desire more respect, opportunities for advancement, and autonomy (Leon, Marainen, & Marcotte, 2001; Monahan & McCarthy, 1992). In light of this, attempts to identify and implement solutions to the turnover problem have relied on the use of management strategies other than financial incentives. For example, the Wellspring model and the Eden Alternative, considered to be leading models of long-term care, have initiated a culture change that recognizes the importance of nurse aides as part of the care team. One of the basic tenets of these models that is believed to show promise in retaining nurse aides is staff empowerment. These models shift from a hierarchical to a more lateral decision-making structure, in which nurse aides have the authority to make more autonomous decisions when appropriate (Ransom, 2000; Stone et al., 2002). Similarly, these models alter the decision-making structure to allow for more discretion among nurse aides, serving as another example of how organizations can make changes in management practices to empower directcare staff. Other efforts to empower direct-care staff have included career ladders, mentoring, and reward systems (Leon et al.; Ransom). Although empowerment can be measured at the social, individual, and organizational levels (Fulton, 1997; Thomas & Velthouse, 1990; Zimmerman, Israel, Schultz, & Checkoway, 1992), in this study we consider empowerment at the organizational level, assuming that empowerment tools provided by the organization become internalized by staff members at the individual level. In the context of the present research, we consider empowerment as a set of strategies provided by the organization to foster a sense of individual empowerment among workers. Individuals in management can influence organizational effectiveness, in this case resident outcomes, by focusing on the factors within their realm of control (Scott & Shortell, 1988). There is evidence in the literature that managerial practices in health care are associated with better patient outcomes (Anderson et al., 2003; Davies & Nolan, 1998; Morris et al., 2003; Shortell et al., 1994). However, a systematic evaluation of the effectiveness of culture-change models (Wellspring and Eden Alternative) on resident outcomes in nursing facilities has met with mixed results. For example, although the Wellspring evaluation revealed qualitative improvements in the quality of resident life and interactions between residents and staff, improvements in quantitative clinical outcomes were not documented (Stone et al., 2002). In addition, although an initial evaluation of the Eden Alternative reported promising results regarding resident clinical outcomes (Ransom, 2000), a more recent study showed no beneficial effects for residents on several clinical outcomes such as functional status and rate of infection (Coleman et al., 2002). Absent from these culture-change evaluations was an investigation of the possible effect that the stability of the nursing staff had on the relationship between changing management practices and resident outcomes. Although the use of empowerment strategies in these models positively affected turnover and retention of nurse aide staff in participating facilities (Ransom, 2000; Stone et al., 2002), it is possible that, despite improvements, the nurse aide workforce remained relatively unstable. As reported by Mentes and Tripp-Reimer (2002), one of the major barriers to effective nursing home intervention research is the instability of nursing home staff. In other words, the effectiveness of implementing empowerment practices in nursing homes is likely compromised if the workforce is unstable. Therefore, an evaluation of the effect of management practices on patient or resident outcomes should consider the stability (or instability) of the staff. To further develop and explain the relationship between management practices and organizational effectiveness in the context of worker empowerment and resident outcomes, in this investigation we adapted Rosabeth Kanter s (1977) theory of structural power in organizations as a theoretical framework. Kanter s theory posits that powerlessness is associated with structural job characteristics such as routinization and the lack of opportunity to exercise discretion or to take risks. Kanter believes that access to certain job-related empowerment structures affects employees in ways that enhance work outcomes, for example, by increasing organizational commitment and job satisfaction. The empowerment structures involve opportunity structures, power structures, and the structure of proportions. Opportunity structures include chances for advancement or growth in the organization and can be accessed through increases in knowledge and skill, as well as the ability to participate in task forces or committees. Kanter also considers an organization s awareness of worker contributions as an additional aspect of opportunity, including methods that impart a sense of value to employees. The three indicators of power, or lines of power structures according to Kanter, include access to resources, information, and support. Resources may include the ability to mobilize human, financial, or material inputs needed to complete the job. Access to information involves awareness of organizational decisions and policy changes, whereas support refers to the ability to take risks and engage in extraordinary activities without having to go through many channels of authority (e.g., autonomy). Kanter believes that access to these power structures promotes the empowerment of individuals by providing an environment where workers have a sense of control over their working conditions. This sense of control encourages productivity and leads to improved organizational effectiveness. Kanter s proportions structure refers to the social composition of the organization, especially the male-to-female ratio. As the gender composition of nurse aides in nursing facilities is largely homogenous, we did not address the structure of proportions in this study. 310 The Gerontologist

Few studies outside of the nursing literature utilized Kanter s theory as an organizing framework. However, studies of nurse empowerment using Kanter s model have revealed positive relationships between empowerment of hospital nurses and employee impact such as job satisfaction, organizational commitment, and self-efficacy (Laschinger, Finegan, & Shamian, 2001; Laschinger & Shamian, 1994) as well as work effectiveness (Laschinger & Wong, 1999; Laschinger, Wong, McMahon, & Kaufmann, 1999). Empowerment research using this theory in the long-term-care industry has been limited to organizational factors affecting job satisfaction and organizational commitment of licensed nurses (Beaulieu, Shamian, Donner, & Pringle, 1997; Mullins, Nelson, Busciglio, & Weiner, 1988). Although evidence in the nursing literature supports the roles of Kanter s opportunity and power structures as influential to nurses work effectiveness, this model of empowerment has not been tested within the context of nurse aides work in nursing homes. As nurse aide empowerment gains increased attention in the long-term-care industry, empirical tests of its effect on resident care are needed. Thus, in this study we adapt Kanter s concepts of empowerment structures (opportunity and power), employee impact, and work effectiveness to investigate the relationship between management practices used to empower nurse aides and resident outcomes in nursing homes. In this study, Kanter s opportunity structures are represented by indicators of organizational structures that facilitate nurse aide empowerment, including reward systems, career ladders, use of job-enhancement techniques, and committee participation. Empowerment strategies representing Kanter s power structures include measures of delegation to nurse aides and the amount of influence aides have on resident-care decisions. A measure of nurse aide staff stability (e.g., turnover and retention) serves as a proxy for employee impact where it is posited that empowerment strategies used in an organization influence whether an employee will leave or stay. Finally, select indicators of quality, namely, pressure ulcer incidence and resident social engagement, represent work effectiveness in the form of resident physical and psychosocial health outcomes. We hypothesized that nursing homes that employ jobrelated nurse aide empowerment strategies would experience better aggregate resident outcomes. We also expected that the stability of the nurse aide staff would moderate or influence the strength of the relationship between nurse aide empowerment strategies and resident outcomes. Methods Cross-Sectional Sample of Nursing Homes In this study we utilized data from a project designed to examine the effects of organizational structure and care processes on nursing home resident outcomes. We drew a stratified random sample of non-hospital-based nursing homes from the universe of all Medicare- or Medicaid-certified nursing homes in eight states (Maine, Missouri, Mississippi, Nebraska, New York, Ohio, Washington, and Wisconsin) reported in the Online Survey, Certification, and Reporting System (OSCAR) from mid-1995. The sample design ensured proper representation of homes with special care units and Medicare populations and was based on strata defined by urban versus rural location, ownership status, chain affiliation, and size (over and under 100 beds). Of the 477 contacted facilities, 64% agreed to participate, resulting in a final sample of 307 homes. Instruments We took the independent variables of interest and the moderator variable (nurse aide staff stability) from a facility survey. In 1995 and early 1996, directors of nursing in sampled homes were interviewed by telephone to characterize the management structure and processes in place regarding care planning and service delivery (N = 307). In addition, day-shift charge nurses within each facility (one per unit) were administered questionnaires to characterize unit-level organizational and care processes related to six domains of resident care: pressure ulcers, psychosocial well-being, limb contractures, cognition, pain, and depression. The questionnaire asked the charge nurses about their supervisory style in the form of specific questions about activities that affect the identification and management of the targeted domains by nurse aides. For example, charge nurses were asked, How do you give your nurse aides their daily work assignments they must perform for the prevention of pressure ulcers? and How much influence do nurse aides have in deciding how resident care related to psychosocial well-being is performed on your unit? A total of 857 unit-level questionnaires were distributed among the 307 homes; among these homes, 273 returned a total of 739 charge nurse questionnaires, resulting in a unit-level response rate of 86%. Outcome variables were extracted from the Minimum Data Set (MDS), Version 2.0. MDS data were not available from Missouri, Nebraska, Washington, or Wisconsin, resulting in a total of 156 facilities reporting MDS data for this study. Approximately 50% of the homes in this study were located in Ohio, 35% in New York, 7% in Mississippi, and 9% in Maine. Within this sample of 156 homes, 430 charge nurse questionnaires were completed from an initial mailing of 484, resulting in a unit-level response rate of 89%. Comparisons of the 156 facilities included in the study to the original stratified sample of 307 revealed no statistically significant differences with regard to the 16 strata. Residual analyses revealed that, although the study sample is not a nationally Vol. 45, No. 3, 2005 311

representative sample of nursing homes, it adequately represents nursing facilities located in Ohio, New York, Mississippi, and Maine. We took several control variables from the 1995 OSCAR and the Area Resource File. We used variables from these sources to control for the effects of the following factors identified from previous research: facility size (Aaronson, Zinn, & Rosko, 1994), staffing levels (Harrington, Zimmerman, Karon, Robinson, & Beutel, 2000), and economic conditions (Banaszak-Holl & Hines, 1996) on the outcomes of interest. We also included nurse aide unionization, taken from the facility survey, to control for possible effects of unions on nurse aide turnover as evidenced in the literature (Brannon, Zinn, Mor, & Davis, 2002). Finally, we used the state in which the facility was located to control for variation in regulatory and financing environments on the outcomes of interest. Independent Variables We took measures of empowerment strategies from the director of nursing and charge nurse questionnaires. In this study, we define delegation in terms of how nurse aides are assigned their daily resident care tasks, and higher delegation scores reflect a situation in which workers are more empowered because a sense of autonomy and decision making have been promoted. Charge nurses were asked to identify how they assign work to nurse aides on the unit in both physical and psychosocial domains. We consider a facility to delegate work if the charge nurse assigns nurse aides to residents in a general manner (score of 2) rather than assigning specific tasks to be completed in a specified order (score of 0). The final facility-level delegation score reflects the average of unit delegation scores within a facility, adjusted for the size of the unit. Having influence over resident care decisions increases power by promoting a sense of value to the workers and acknowledging the relevance of the job to the overall functioning of the nursing home. Unit charge nurses rated the amount of influence nurse aides have in care-related decisions. That power was rated by supervisors rather than by the nurse aides themselves suggests that this measure captures the aides access to power as opposed to self-perceived empowerment; thus, it represents an empowerment strategy offered by the organization. Charge nurses responded to two questions by rating the amount of influence that nurse aides have concerning care provision to residents on a scale ranging from 1 (no influence) to5(very much influence). We took a unitlevel mean score from these questions to determine the average level of influence that nurse aides maintain on the unit regarding resident care. We aggregated scores to the facility level by weighting the mean score for the amount of influence by unit size, or the proportion of total facility beds on the unit. Opportunity structures included the number of types of rewards offered to nurse aides (range 0 4), whether an advanced nurse aide position exists in the facility (0,1), whether nurse aides participate on committees (0,1), and whether the facility uses jobenhancement techniques such as training and orienting new nurse aides (0,1). We took these variables from the survey of directors of nursing. Moderator Variable We used a measure of nurse aide staff stability as a proxy for employee impact in the context of Kanter s model, similar to other measures used in the nursing literature such as organizational commitment. Although turnover rates have been widely used in the literature to assess the effects of interventions on employee retention and to establish a link between the nurse aide workforce and care quality, interpreting turnover rates in the absence of staff retention measures may be incomplete. For example, a facility that experiences high nurse aide turnover also may maintain a stable core of experienced staff. It is possible that this facility s turnover rate is measuring turnover within relatively few nurse aide positions or full-time equivalents, which can be suggestive of successfully weeding out inappropriate hires. Therefore, turnover within this type of facility is unlikely to have a large effect on resident care or the success of a managerial intervention. Alternatively, a facility with a high turnover rate as well as an unstable staff is, perhaps, experiencing high turnover that is qualitatively different from that of the first facility. There may be internal organizational conflict or ineffective management practices affecting staff retention. In this case, high turnover may, indeed, influence the effect of an intervention on resident care. Although it has been argued that some turnover in health care organizations can be beneficial in terms of workforce renewal (Stryker, 1982), staff retention also has been shown to have positive effects on resident-care outcomes (Cohen-Mansfield, 1997). Additionally, a recent study by Brannon and colleagues (2002) suggests that modeling turnover categorically may provide clearer explanations and recommendations for practice. We created dummy variables to measure staff stability represented by four categories of turnover and retention: high turnover high retention, high turnover low retention, low turnover high retention, and low turnover low retention. We measured turnover as the number of nurse aide resignations or terminations relative to the number of nurse aides employed in the facility for the 6 months prior to the interview with the director of nursing. We measured retention as the proportion of nurse aides employed at the facility for 2 years or longer. We took both measures from the survey of the directors of nursing. We used median values for both turnover (28%) and retention (52%) to categorize a facility s nurse aide staff stability. 312 The Gerontologist

Dependent Variables We include two resident-outcome measures in the analysis, resident social engagement and pressure ulcer incidence. We chose these measures as indicators of worker effectiveness because of their reliance on nurse aide involvement. The Social Engagement Scale is taken from the MDS and is considered a marker for quality of life in nursing homes (Mor et al., 1995). Pressure ulcer incidence measures the presence of newly acquired decubitis ulcers in nursing home residents and is a widely accepted measure of nursing home quality. The development of these measures, along with reliability estimates, has been reported elsewhere (Hawes et al., 1995; Mor et al., 2003). We aggregated each of these two measures to the facility level and adjusted them for risk, with final scores reflecting the ratio of the observed score to a predicted score given a set of risk factors. For social engagement, an adjusted score of greater than 1 denotes that the level of resident engagement in the life of the facility was higher than expected, whereas a risk-adjusted pressure ulcer incidence score of greater than 1 indicates that the incidence was worse than predicted given a set of covariates. We took the MDS data used for the calculation of the Social Engagement Scale and pressure-ulcerincidence scores from quarterly, annual, and significant change assessments in each facility that were conducted within 6 months of the director-of-nursing interview. We did not include assessments conducted within 60 days of a prior assessment in the calculation. For the Social Engagement Scale, we calculated the observed score for each facility as the sum of the items from section F1 of the MDS divided by the number of valid cases, defined as residents who had a score on the MDS Cognitive Performance Scale of less than or equal to 3, needed limited assistance in transferring, and had full valid data for social engagement. We modeled the predicted score in a regression equation by using the following variables: history of mental illness, mental retardation or developmental disability, conditions related to mental retardation or developmental disability manifested before age 22 and likely to continue indefinitely, and diagnosis of Alzheimer s disease or other dementia. We aggregated the predicted score to the facility level as a grand mean. We measured pressure ulcer incidence as the occurrence of any Stage 2 or more severe pressure ulcers at a follow-up assessment among resident assessments indicating no worse than Stage 1 ulcers at baseline. We calculated the observed score as a count of the number of times an assessment showed a Stage 2 or more severe pressure ulcer at a follow-up assessment divided by the number of assessments indicating Stage 1 or no pressure ulcer in a facility. We calculated the predicted score as the probability that a pressure ulcer would develop or worsen given the following set of covariates: age, gender, cognitive status, ability to hear, ability to be understood, ability to understand, and diagnosis of stroke. The sum of these predicted probabilities divided by the number of assessments indicating the presence of a Stage 1 or no pressure ulcer was used as the predicted pressure ulcer score for a facility. Organizational control variables included facility size (total number of beds), profit status (1 = for profit, 0 = not for profit), registered nurse-to-nurse aide ratio (registered nurse full-time equivalents, or FTEs, to nurse aide FTEs) and nurse aide unionization (1 = union). We included the county-level unemployment rate to control for local economic conditions. Finally, we used a dichotomous variable comparing New York to the other three states in the sample to control for possible effects of the location of the facility. Analytic Methods We conducted a factor analysis on a set of variables representing Kanter s power and opportunity structures (see Appendix A), although the results of reliability analyses did not support the use of scales or factor scores to measure these constructs. Instead, we entered variables into a hierarchical regression equation to determine the unique contribution of the independent variable after we controlled for other variables theoretically and empirically related to the outcomes. Results Table 1 includes descriptive statistics for the variables representing nursing home management empowerment strategies, control variables, and the outcome variables. Descriptive statistics reveal that the average amount of influence nurse aides have in resident-related decisions is rated as quite high (x = 4.16, SD = 0.6). The number of rewards offered and the level of delegation to nurse aides are moderate. On average, nurse aides may receive two out of four possible reward systems and are delegated to in one out of a possible two care domains. A greater proportion of all facilities indicated using jobenhancement techniques and having nurse aides serve on committees, whereas a smaller proportion of facilities reported having an advanced nurse aide position. Facilities in the sample had an average ratio of registered nurse-to-nurse aide FTEs of 0.17, meaning that, for every 10 nurse aide FTEs, there were 1.7 registered nurse FTEs. This ratio is comparable with that found in prior research using OSCAR data (Banaszak-Holl & Hines, 1996). The average county unemployment rate was almost 8% and the average number of beds (123) in this sample was somewhat lower than the average number of beds (127) in the four sampled states. A majority of the facilities were run for profit, whereas a minority had unionized nurse aide staff. Finally, 34.6% of facilities were located in New York. Vol. 45, No. 3, 2005 313

Description Variable Table 1. Sample Descriptive Data Range Minimum Maximum Empowerment variables No. of rewards offered to nurse aides 0 4 1.83 1.23 Does facility have an advanced nurse aide position 0 1.2792 0.4501 Do nurse aides serve on committees 0 1.7597 0.4286 Does facility use job enhancement techniques 0 1.8312 0.3758 How much influence nurse aides have in resident care 1 5 4.16 0.60 Amount of charge nurse delegation to nurse aides 0 2 1.1 0.50 Control variables Registered nurse-to-nurse aide ratio 0 0.65 0.17 0.09 Unemployment rate 4.2 12.9 7.8 1.73 Facility size 25 520 123.13 75.4 For-profit facility 0 1.6410 0.4812 Union presence 0 1.2310 0.4226 New York 0 1.3462 0.4772 Dependent variables Pressure ulcer incidence 0 2.70 1.0 0.54 Social engagement 0.39 1.69 1.0 0.22 Table 2. Distribution of Dummy Variables for NA Staff Stability N %of Total Low turnover low retention (LOW LOW) 18 11.5 Low turnover high retention (LOW HIGH) 56 35.9 High turnover low retention (HIGH LOW) 50 32.1 High turnover high retention (HIGH HIGH) 32 20.5 M SD Table 3. Hierarchical Regression Results for Pressure Ulcer Incidence and Social Engagement Variables in the Equation Pressure Ulcer Incidence Social Engagement R 2 R 2 b R 2 R 2 b Step 1.03.04 Facility size 0.007 0.0006** For-profit facility 0.105 0.021 Register nurse-tonurse aide ratio 0.074 0.198 Unemployment rate 0.028 0.0007 Union presence 0.078 0.051 New York 0.105 0.002 Step 2.08.05 a *.12.08 b *** Amount of delegation 0.087 0.025 Amount of influence 0.043 0.085*** No. of rewards 0.070* 0.014 Advanced nurse aide position 0.182* 0.043 Nurse aides serve on committees 0.020 0.014 Facility uses job enhancement techniques 0.150 0.075 Step 3.10.02.15.03 Low turnover low retention 0.038 0.039 Low turnover high retention 0.229* 0.091* High turnover low retention 0.155 0.106** a F statistic for R 2 = 1.836; df (6,143). b F statistic for R 2 = 3.965; df (6,143). *p,.10; **p,.05; ***p,.01. The adjusted pressure ulcer incidence for this sample was M = 1.0, with SD = 0.5. Although slightly skewed to the left, the distribution suggests that the average facility s raw pressure ulcer incidence rates were equivalent to predicted rates. Eight facilities, or 5% of the sample, had adjusted incidence rates of 0, meaning that the facilities experienced no new or worsening ulcers during the 12-month time period. The mean social engagement score for the sample was 1.0, with SD = 0.2. This distribution approximates the normal distribution curve, but is still slightly skewed to the left. As with the pressure ulcer rate, the distribution of the adjusted social engagement score suggests that the average facility in the sample experienced social engagement outcomes that were equivalent to what we predicted by using the given set of covariates. Table 2 includes the distribution of staff stability across the four discrete categories created to test for moderating effects of staff stability. We express these as dummy variables and use the high retention high turnover category as the reference group. We used hierarchical regression to test for the unique contribution of the empowerment strategy variables as well as for the moderating effect of staff stability. Table 3 summarizes the results of the regression analysis. Although the hierarchical model predicting pressure ulcer incidence was not significant, the addition of the empowerment strategy variables in Step 2 of the model appeared to explain additional variance in pressure ulcer incidence. Contrary to what we expected, having an advanced nurse aide position slightly increases the incidence of Stage 2 or greater pressure ulcers in these facilities. However, the 314 The Gerontologist

finding that pressure ulcer incidence decreases as the number of rewards increases supports our theoretical expectations. A summary of the findings from the Social Engagement model also is presented. As with the pressure ulcer outcome variable, the model did not significantly predict a facility s risk-adjusted social engagement score. A facility s size was slightly negatively associated with the social engagement outcome variable. However, the addition of the power and opportunity variables in Step 2 increased the proportion of explained variance in the social engagement outcome significantly. In addition, the power variable (amount of influence nurse aides have in resident care decisions) had a significant, positive relationship to the outcome variable. We entered the dummy variables expressing staff stability into Step 3 of the hierarchical regressions to further explain the relationships between the independent and dependent variables. Although the addition of the dummy variables does not contribute significantly to the amount of explained variance, the findings provide some additional information regarding the relationship between nurse aide staff stability and resident outcomes. Facilities with low turnover and high retention experienced lower pressure ulcer incidence rates relative to facilities with high turnover and high retention when empowerment strategy variables were in the equation. Social engagement scores were lower in facilities experiencing either high turnover and low retention or low turnover and high retention relative to facilities where both turnover and retention were high. Discussion This study offers evidence that some management practices used to empower nurse aides are related to a facility s pressure ulcer incidence rate and psychosocial well-being scores. Although reliability estimates did not permit the use of scales to measure empowerment strategies as the multidimensional constructs described by Kanter, the addition of the group of variables representing these empowerment strategies in the regression models explained additional variance in the outcome measures above that which was explained by the structural control variables. Namely, opportunity structures affected the physical outcome of incidence of pressure ulcers, whereas power structures influenced the psychosocial outcome of social engagement. In a recent study, Zinn, Brannon, Mor, and Barry (2003) tested the proposition that the structure of work in nursing homes is differentiated according to the nature of the clinical domain of care. They found that, for physical care, work is structured more mechanistically and is characterized by higher task standardization and formalization whereas psychosocial care is more organically structured with greater task autonomy and delegation. The results of this study lend support to this notion that management strategies may be most effective when the type of clinical care is considered. This finding also adds support to the earlier finding that nurse aide involvement in care planning is related to the psychological well-being of residents (Mor, 1995). Although the finding by Mor was based on nurse aide attendance at care-planning meetings, results from this study take this one step further. The amount of influence was assessed by the nurse aides supervisors, suggesting that charge nurses in these facilities truly value nurse aide opinions beyond mere attendance at the care-plan meetings, where involvement of aides may be superficial. The relationship between the amount of nurse aide influence and level of resident engagement in the life of the facility is not surprising in that nurse aides, who provide most of the hands-on care, gain an understanding of the residents likes, dislikes, and idiosyncrasies. This knowledge can readily be passed on to other staff who can prepare appropriate activities and interventions for the resident. Analyses showed a positive relationship between the opportunity structure represented by an advanced nurse aide position and the incidence of pressure ulcers. Indeed, this may be indicative of reverse causation, in which facilities are reacting to poor care quality by implementing an opportunity structure to empower aides. Nurse aide career ladders are relatively new in the industry, and it may be some time before the effects of this strategy are realized. The cross-sectional nature of this study makes it difficult for us to fully understand this finding, and it suggests the use of longitudinal studies to determine the effect of career ladders in improving organizational effectiveness. Though moderating effects of the staff stability variables were not supported, perhaps because of the strong relationships among the empowerment-strategy variables and staff stability, the hierarchical analyses revealed some interesting direct effects of staff stability on the outcome variables. Findings suggest that pressure ulcer incidence rates are lower in facilities with lower staff turnover and higher retention relative to facilities with higher turnover and higher retention. Although both types of facilities reported that more than 50% of their nurse aide staff had tenure of 2 or more years, outcomes were poorer for those with higher turnover. One possible explanation for this is that the time and effort required to continually train new nurse aides in the high-turnover homes, where on-the-job training may involve peer mentoring, could be invested in direct resident care by experienced nurse aides. This supports the finding of Bowers, Esmond, and Jacobson (2000), in which experienced nurse aides reported that working with new staff parallels that of working short staffed, often compromising resident care. In contrast, homes with high turnover Vol. 45, No. 3, 2005 315

and high retention fared better with the psychosocial outcome. Although we expected these homes to experience higher social engagement scores compared with the high turnover low retention group, the finding that they also perform better on this outcome relative to the low turnover high retention group is curious. However, it is possible that nurse aides who are not suited for a caregiving role are appropriately weeded out, leaving a better-qualified pool of workers who feel comfortable interacting with and motivating the nursing home population. In contrast, nurse aides in the low turnover high retention homes may be less effective at interpersonal relationships yet remain in the job. As nurse aides are a difficult group to retain, managers in these homes may be willing to overlook deficiencies in interpersonal skills in favor of higher levels of retention. Further, as Bowers and colleagues (2000) described, aides who remain successfully on the job may find ways of bundling routine care tasks to the exclusion of personalized care, which may result in poorer psychosocial outcomes for the residents. That effects varied with respect to the nature of the outcome lends support to earlier work that suggests a relationship between staff stability and quality of care (Bowers et al., 2000). Moreover, it offers additional information to managers who maintain a stable core of nurse aide employees yet experience different levels of turnover. For managers in facilities where retention is high and turnover is low, results suggest that more attention may have to be paid to improving the level of social engagement of residents in the facility. If, indeed, these managers are overlooking deficits in interpersonal skills in order to retain nurse aide staff, investing in improving communication skills may lead to higher quality interactions between aides and residents, which could allow for improvements in social engagement for the residents. Likewise, results of this study suggest that, in facilities where both retention and turnover are high, improvements can be made in pressure ulcer prevention methods. This suggests a need for improved training of nurse aides in resident care. In that these facilities do maintain a stable core of nurse aides, the effectiveness of enhanced training seems promising as they do not face the barrier of instability described by Mentes and Tripp-Reimer (2002). The findings of this study contain several important implications for future research. In response to the staffing challenges currently faced by nursing facilities, managers and industry leaders have begun to realize and emphasize the critical role that nurse aides play in organizational effectiveness. Moreover, the culture-change movement that is being introduced in the long-term-care industry suggests a managerial paradigm shift from one that is authoritarian to one that is more participative and empowering for both staff and residents. Given this paradigm shift, research concerning the nurse aide workforce should attempt to delineate the relationships among the concepts explored in this study. Although this study provides a preliminary examination of the links between nurse aide empowerment strategies, aide staff stability, and resident outcomes, future research can build on the findings here and address some of the measurement challenges inherent in these data. For example, expanding the sample to include evening and night charge nurses may more completely capture an organization s management practices. Additionally, the underlying assumption in this study is that organizational-level empowerment strategies as perceived by the nurse aides supervisors translate into a more empowered workforce. The inability in this study to include aides perceptions of empowerment strategies limits the depiction of true relationships among the constructs of interest, as it is has been shown that nurse aide perceptions of the implementation of management strategies do not always parallel those of their supervisors (Corazzini- Gomez, Anderson, & McDaniel, 2002). Studies that build on the findings here should test this assumption to determine if empowerment strategies offered by the organization do, indeed, translate into a more empowered nurse aide workforce. There is clearly a need to further validate this assumption as nursing facilities begin to implement management strategies associated with the culture-change initiative. In the absence of a valid relationship between empowerment strategies and individual nurse aide empowerment, resources invested in developing new management strategies may be wasted. In an era of cost containment and low profit margins, the systematic study and knowledge of effective interventions is highly important. Although nurse aide empowerment and its effects on quality of care remain of great interest to consumers, providers, and policy makers, there is very little solid research available from which to build a theoretical model. This study, one of the first to explore these concepts by using a theoretical framework, identifies promising components that can be used in future studies to model these relationships. As this study uncovered significant relationships despite the use of secondary data, these findings can be used as a springboard in developing models that are less rudimentary. In addition, since these data were gathered, more research has been conducted on the MDS and its uses, including the relative instability of quality indicators over short periods of time (Mor et al., 2003). Future studies using quality indicators as measures of organizational effectiveness may reveal more significant patterns if data are taken from a time period of longer than 6 months as in the present study. References Aaronson, W. E., Zinn, J. S., & Rosko, M. D. (1994). Do for-profit and notfor-profit nursing homes behave differently? The Gerontologist, 34, 775 786. Anderson, R., Issel, L. M., & McDaniel, R. R. (2003). Nursing homes as complex adaptive systems: Relationship between management practice and resident outcomes. Nursing Research, 52, 12 21. 316 The Gerontologist

Atchison, J. H. (1998). Perceived job satisfaction factors of nursing assistants employed in Midwest nursing homes. Geriatric Nursing, 19, 135 137. Banaszak-Holl, J., & Hines, M. A. (1996). Factors associated with nursing home staff turnover. The Gerontologist, 36, 512 517. Beaulieu, R., Shamian, J., Donner, G., & Pringle, D. (1997). Empowerment and commitment of nurses in long-term care. Nursing Economics, 15, 32 41. Bowers, B. J., Esmond, S., & Jacobson, N. (2000). The relationship between staffing and quality in long-term care: Exploring the views of nurse aides. Journal of Nursing Care Quality, 14(4), 55 64. Brannon, D., Zinn, J., Mor, V., & Davis, J. (2002). An exploration of job, organizational and environmental factors associated with high and low nursing assistant turnover. The Gerontologist, 42, 159 168. Burnfeind, J. D., & O Connor, S. J. (1992, August). Employee turnover and retention rates as predictors of nursing home code violations. Paper presented at the annual meeting of the Academy of Management, Las Vegas, NV. Cohen-Mansfield, J. (1997). Turnover among nursing home staff. Nursing Management, 28(5), 59 64. Coleman, M., Looney, S., O Brien, J., Ziegler, C., Pastorino, C., & Turner, C. (2002). The Eden Alternative: Findings after one year of implementation. Journal of Gerontology: Medical Sciences, 57A, M422 M427. Corazzini-Gomez, K., Anderson, R., & McDaniel, R. (2002, October). Quality of care and nurse aide participation in decision making about resident care. Paper presented at the 55th Annual Scientific Meeting of the Gerontological Society of America, Washington, DC. Davies, S., & Nolan, M. (1998). Educating nursing home staff in the reduction of pressure sores. British Journal of Nursing, 7(3), 146 151. Fulton, Y. (1997). Nurses views on empowerment: A critical social theory perspective. Journal of Advanced Nursing, 26, 529 536. Harrington, C., Zimmerman, D., Karon, S. L., Robinson, J., & Beutel, P. (2000). Nursing home staffing and its relationship to deficiencies. Journal of Gerontology: Social Sciences, 55B, S278 S287. Hawes, C., Morris, J., Phillips, C., Mor, V., Fries, B., & Nonemaker, S. (1995). Reliability estimates for the Minimum Data Set for nursing home resident assessment and screening (MDS): Results from the final field testing. The Gerontologist, 35, 172 178. Kanter, R. M. (1977). Men and women of the corporation. New York: Basic Books. Laschinger, H., Finegan, J., & Shamian, J. (2001). The impact of workplace empowerment and organizational trust on staff nurses work satisfaction and organizational commitment. Health Care Management Review, 26(3), 7 23. Laschinger, H., & Shamian, J. (1994). Staff nurses and nurse managers perceptions of job-related empowerment and managerial self-efficacy. Journal of Nursing Administration, 24, 30 35. Laschinger, H., & Wong, C. (1999). Staff nurse empowerment and collective accountability: Effect on perceived productivity and self-rated work effectiveness. Nursing Economics, 17, 308 316. Laschinger, H., Wong, C., McMahon, L., & Kaufmann, C. (1999). Leader behavior impact on staff nurse empowerment, job tension, and work effectiveness. Journal of Nursing Administration, 29, 28 39. Leon, J., Marainen, J., & Marcotte, J. (2001). Pennsylvania s frontline workers in long-term care: The organization perspective. Jenkintown, PA: Polisher Research Institute. Mentes, J. C. & Tripp-Reimer, T. (2002). Barriers and facilitators in nursing home intervention research. Western Journal of Nursing Research, 24, 918 936. Monahan, R. S., & McCarthy, S. (1992). Nursing home employment: The nurse s aide s perspective. Journal of Gerontological Nursing, 18(2), 13 16. Mor, V. (1995). Invest in your frontline worker: Commentary. The Brown University Long-Term Care Quality Newsletter, 7(1), 4 5. Mor, V., Angelelli, J., Jones, R., Roy, J., Moore, T., & Morris, J. (2003). Inter-rater reliability of nursing home quality indicators in the U.S. BMC. Health Services Research, 3(20), 1 13. Retrieved April 1, 2004 from www.biomedcentral.com/1472-6963/3/20 Mor, V., Branco, K., Fleishman, J., Hawes, C., Phillips, C., Morris, J., et al. (1995). The structure of social engagement among nursing home residents. Journal of Gerontology: Psychological Sciences, 50B, P1 P8. Morris, J., Moore, T., Jones, R., Mor, V., Angelelli, J., Berg, K., et al. (2003). Validation of long-term and post-acute care quality indicators: Final report. Cambridge, MA: Abt Associates. Mullins, L. C., Nelson, C. E., Busciglio, H., & Weiner, H. (1988). Job satisfaction among nursing home personnel: The impact of organizational structure and supervisory power. Journal of Long-Term Care Administration, 16(1), 12 18. Ransom, S. (2000). Eden Alternative: The Texas project. San Marcos, TX: Institute for Quality Improvement in Long Term Health Care. Schnelle, J. F., McNees, M. P., Simmons, S. F., Agnew, M. E., & Crooks, V. C. (1993). Managing nurse aides to promote quality of care in the nursing home. In L. Rubenstein & D. Wieland (Eds.), Improving care in the nursing home: Comprehensive reviews of clinical research (pp. 314 331). Newbury Park, CA: Sage. Scott, W. R., & Shortell, S. M. (1988). Organizational performance: Managing for efficiency and effectiveness. In S. Shortell & A. Kaluzny (Eds.), Healthcare management: A text in organization theory and behavior (pp. 418 457). New York: Wiley. Shortell, S. M., Zimmerman, J., Rousseau, D. M., Gillies, R., Wagner, D., Draper, E., et al. (1994). The performance of intensive care units: Does good management make a difference? Medical Care, 32, 508 524. Stone, R. (2001). Research on frontline workers in long-term care. Generations, 25(1), 49 57. Stone, R., Reinhard, S., Bowers, B., Zimmerman, D., Phillips, C., Hawes, C., et al. (2002). Evaluation of the Wellspring Model for improving nursing home quality. Washington, DC: American Association of Homes and Services for the Aging. Stone, R., & Salsberg, E. (2001, February). Building a high-quality longterm care paraprofessional workforce. Legislative workshop presented to state and local policymakers, Dallas, TX. Stryker, R. (1982). The effect of managerial interventions on high personnel turnover in nursing homes. Journal of Long-Term Care Administration, 10(2), 21 33. Thomas, K. W., & Velthouse, B. A. (1990). Cognitive elements of empowerment. Academy of Management Review, 15, 666 681. Zimmerman, M. A., Israel, B. A., Schulz, A., & Checkoway, B. (1992). Further explorations of empowerment theory: An empirical analysis of psychological empowerment. American Journal of Community Psychology, 20, 707 727. Zinn, J., Brannon, D., Mor, V., & Barry, T. (2003). A structure-technology contingency analysis of caregiving in nursing facilities. Health Care Management Review, 28, 293 306. Received April 5, 2004 Accepted December 6, 2004 Decision Editor: Linda S. Noelker, PhD Appendix A Table A-1. Two-Factor Solution of Empowerment Indicators Factor Eigenvalue Difference Proportion of Variance Cumulative Variance 1 1.71 0.41 0.28 0.28 2 1.30 0.47 0.22 0.50 3 0.82 0.03 0.14 0.64 4 0.79 0.09 0.13 0.77 5 0.71 0.04 0.12 0.89 6 0.66 0.11 1.00 Table A-2. Summary of Factor Loadings (Oblique Rotation) Variable Factor Loadings Factor 1 Factor 2 Opportunity Nurse aides serve on committees 0.868 0.028 Advanced nurse aide position 1.000 0.000 Facility uses job enhancement techniques 0.654 0.127 No. of rewards 0.868 0.029 Power Amount of delegation 0.035 1.000 Amount of influence 0.002 0.868 Note: Bolded values indicate significant factor loadings. Vol. 45, No. 3, 2005 317