| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
|
|
|
|||||||
|
|||||||||
Correspondence: Address correspondence to Farida K. Ejaz, Margaret Blenkner Research Institute, Benjamin Rose Institute, 11900 Fairhill Road, Suite 300, Cleveland, OH 44120-1053. E-mail: fejaz{at}benrose.org
 |
Abstract |
|---|
 TOP Abstract Conceptual ModelMethodsResultsDiscussionReferences |
|---|
Key Words: Aides in nursing homes • Home health agencies • Assisted living
Turnover among DCWs stems from a variety of factors, key among which is job dissatisfaction (Capitman, Leutz, Bishop, & Casler, 2004; Castle, Degenholtz, & Rosen, 2006; Institute of Medicine, 2001; National Commission on Nursing Workforce for Long Term Care, 2005; Nursing workforce: Recruitment, 2001; Parsons, Simmons, Penn, & Furlough, 2003; Wiener, 2003). Given the extent of DCW turnover, gaining a better understanding of the factors affecting job satisfaction is critical in order to develop and implement workplace interventions to enhance job satisfaction. Prior research studies on the determinants of job satisfaction among DCWs are sparse. Furthermore, they are limited by the general absence of a conceptual framework to guide the research or one that is narrow in scope, a failure to consider both DCW-level and organizational-level factors to simultaneously address the issue of nested data, and the lack of inclusion of DCWs from different types of LTC settings (Castle et al., 2006). We designed this study to address these limitations. It was guided by the LTC stress and support model, and we collected data from DCWs in three major sectors of the LTC industry and from study site liaisons on organizational issues. Furthermore, we used hierarchical linear modeling to identify organizational effects on the study outcome while controlling for data obtained at the individual level from DCWs.
|
Conceptual Model |
|---|
|
TOPAbstractConceptual ModelMethodsResultsDiscussionReferences |
|---|
We did not use other conceptual models in the study because of their limitations. One study used a conceptual framework organized around staff demographic characteristics, job characteristics, and work environments to predict employee job satisfaction, job commitment, and turnover intent in a single type of LTC setting (Karsh, Booske, & Sainfort, 2005). Another study on factors distinguishing nursing home facilities with high and low DCW turnover used job, organizational, and environmental factors at the organizational level but did not include any individual-level DCW characteristics (Brannon, Zinn, Mor, & Davis, 2002). Thus, the proposed research responded to the need for a more inclusive conceptual framework to empirically test how individual-level DCW characteristics and stressors as well as organizational characteristics and management issues influence job satisfaction.
Elements of the Conceptual Model
At the DCW level, variables in the model (see Figure 1) include the background characteristics of the DCWs (A), their personal and job-related sources of stress (B), and workplace support (C). At the organizational level (D), the model includes characteristics of the organizations (e.g., type of LTC setting) and management issues (e.g., high DCW turnover). The various sources of stress and support at both the DCW and organizational levels influence the study outcome (E), that is, DCW job satisfaction.
|
Personal Stressors
Because DCWs are the lowest paid workers in the health care labor force, financial and family worries are likely to be personal sources of stress that can carry over to the work setting (Stone & Weiner, 2001; Tellis-Nayak & Tellis-Nayak, 1989). In addition, because the care DCWs give to the frail elderly is physically and emotionally demanding, over time it is likely to negatively affect their physical and emotional health (Mutaner et al., 2004; Noelker & Harel, 2000; Stone & Weiner, 2001).
Job-Related Stressors
Stressors on the job include job design features such as the frequency of scheduling changes (e.g., coming in early, staying late, being called to work on a day off) and lack of permanent assignments. Scheduling changes are likely to disrupt the practice of permanent assignment to residents or clients and are indicative of short staffing and turnover problems, which in turn impact overall job satisfaction. Another important component of job-related stress is inadequate on-the-job training. Such training includes orientation to the job, peer mentoring, and continuing education. Other job-related stressors investigated here are poor pay and lack of benefits such as health insurance (U.S. Department of Health and Human Services, 2003, 2004; U.S. General Accounting Office, 2001).
Support in the Workplace
Research has provided evidence that positive support can attenuate the negative effects of stress (Bass, Noelker, & Rechlin, 1996), whereas negative support can exacerbate stress (Krause, 1995). For example, DCWs often hear racist and ethnic remarks made about them by clients, families, and other staff (Berdes & Eckert, 2001; 2007; Foner, 1994; Tellis-Nayak & Tellis-Nayak, 1989). Using the LTC stress and support model, we investigated the effects of DCW perceptions of workplace racism and other negative and positive interactions on job satisfaction.
Organizational-Level Predictors of DCW Job Satisfaction
At the organizational level, a key background characteristic is type of LTC setting. Examining differences by type of LTC setting is important because settings can significantly differ in their structure and operation (Ejaz et al., 2006). For example, DCWs in nursing homes work in a more institutional setting compared to home care aides, who provide care in a client's home. Research has also shown that staffing patterns and working conditions in for-profit facilities place greater demands on nursing staff (Anderson, Issel, & McDaniel, 1997).
Other organizational features have the potential to influence worker outcomes, including the availability of resources that are affected by reimbursement streams (i.e., Medicare, Medicaid, and private pay). A study by Mor, Zinn, Angelelli, Teno, and Miller (2004) indicated that organizations serving primarily Medicaid and minority residents have fewer resources (e.g., lower reimbursement rates under public programs), which can negatively impact staffing levels, rate of pay, amount of staff training, and availability of equipment and supplies. Another key organizational characteristic is the difficulty an organization faces because of high DCW turnover rates (Harris-Kojetin et al., 2004). Based on these prior studies, the LTC stress and support model also included the following organizational-level predictors: the percentage of minority and nonminority clients served; the percentage of reimbursement under Medicare, Medicaid, and private pay; and the difficulty the organization reports with DCWs quitting or being fired.
|
Methods |
|---|
|
TOPAbstractConceptual ModelMethodsResultsDiscussionReferences |
|---|
Site Selection
We developed a database of all nursing homes, assisted living facilities, and home health agencies in a five-county area of northeast Ohio by obtaining information from various sources (e.g., the state's Department of Health, the local area agency on aging, and the state's assisted living association). The comprehensive list included 75 home health agencies, 143 nursing homes, and 101 assisted living facilities, for a total of 319 agencies. Because the list of LTC facilities was obtained from state agencies, it excluded organizations that were not certified or licensed (e.g., private home health agencies). Following the creation of a database of the 319 facilities, we contacted these organizations to determine if they met the criteria for inclusion in the study: (a) facilities with more than 10 DCWs and (b) facilities that employed their own DCWs and did not use only agency staff or outside contractors. Of the 319 agencies in the study area, we were able to obtain information from 217 (68%). Based on the information supplied by the 217 organizations, a total of 161 (29 home health agencies, 40 assisted living facilities, and 92 nursing homes) met the study criteria.
We determined that, in order to effectively address Research Question 2, we would need a minimum of 30 organizations to investigate the effects of organizational-level variables while controlling for DCW-level variables (Kraemer & Thiemann, 1988). We used proportionate random sampling techniques (Cochran, 1977) to select a larger number of sites than the minimum number required. This sampling strategy allowed us to randomly select 8 home care agencies, 13 assisted living facilities, and 25 nursing homes, for a total of 46 sites.
In view of the project's restricted time frame for site and sample recruitment and data collection, we assumed a 100% refusal rate to ensure that the replacement pool contained enough sites. Thus, we contacted a total of 90 sites (see Table 1), and 41% of those refused to participate. Refusals included active and passive refusals, the latter referring to cases in which facility administrators agreed to participate but did not follow up with the research team. Refusal rates differed by type of LTC setting. Almost half (49%) of the nursing homes contacted were refusals, whereas only one third (33%) of home care agencies were. In the end, 46 facilities participated in the study. Ultimately, we added three additional sites to boost the overall number of DCWs in the sample (see "Respondent Selection"). Comparisons between participating and nonparticipating sites revealed that sites did not differ on key variables such as type of LTC setting, number of clients/residents served, or the number of DCWs that they employed.
|
Because of the limited time frame for data collection and in anticipation of a high DCW refusal rate, we contacted a total of 1,058 DCWs (see Table 1). Of these, 39% refused to participate. We completed a total of 644 interviews, which was more than the minimum number of DCWs required to address Research Question 1. Challenges encountered in contacting and interviewing DCWs included disconnected and wrong phone numbers and DCWs' failure to keep interview appointments, despite the fact that interviewers were willing to conduct interviews at odd hours and in any place and time convenient to the DCW.
Data Collection Procedures
We collected two types of data: (a) DCW-level data based on either in-person or telephone interviews with workers and (b) organizational-level data based on a mailed survey completed by the site liaison (usually a human resources director or the administrator). In cases where the organizational survey was incomplete, we contacted site liaisons to complete parts of the instrument over the phone. All 49 organizations in the study completed the survey.
Consent Procedures
Prior to starting the interview, we obtained informed consent from DCWs. The institutional review board of the research organization approved the consent procedures. The average amount of time taken to complete an interview was 50 min (SD = 34.58), and respondents received $20.
Measures
Outcome Variable
Job satisfaction was an overall measure of satisfaction with various aspects of a DCW job. The scale was originally developed for use with DCWs in nursing homes (Kiefer et al., 2005). Items are scored using a 4-point Likert scale, with higher scores indicating higher job satisfaction levels. We assessed the psychometrics of the scale in two steps. First, we dropped 2 of the original 18 items because of insufficient variance in responses (i.e., the criterion for exclusion was having less than 80%–20% response variance on an item). Second, we entered the remaining 16 items into a factor analysis using principal axis factoring and Varimax rotation. It revealed a one-factor solution. Thus, we used all 16 items (factor loadings.58–.80) to develop the scale (Cronbach's 
=.94). (See the Appendix for the scale items.)
|
DCW Personal Stressors
We used five measures to assess DCW personal stressors. One measure of perceived financial adequacy included six items covering whether the DCW had sufficient money to meet regular and unexpected obligations (Cronbach's =.81). The second measure is an additive index of four items assessing worries about family while at work (Cronbach's =.75). The health-related personal stressors included the Center for Epidemiologic Studies–Depression scale (CES-D; Radloff, 1977) and two single-item measures of perceived change in emotional health and physical health since becoming a DCW, scored as much worse (0), worse (1), about the same (2), better (3), or much better (4).
DCW Job-Related Stressors
We examined three components of job-related stress: (a) job design features, (b) training issues, and (c) pay and benefits. Job design features included a measure of scheduling changes (how often in the past 2 months the DCW had been asked to come in early, stay late, or work on a scheduled day off) and a measure of the frequency with which the DCW was permanently assigned to clients. Three single-item measures on training assessed perceived adequacy of job orientation to the facility, perceived adequacy of continuing education at the facility, and usefulness of having a mentor at the time of hire. With respect to pay and benefits, we included a one-item measure of being fairly compensated for the job along with four single-item measures of whether the organization provided the DCW with sick leave, paid holidays, a retirement or pension plan, and paid health insurance.
DCW Job-Related Support
We included three measures of job-related support in the analysis. We asked DCWs whether or not certain positive and negative reactions were elicited when the DCW interacted with other DCWs and with residents. Positive items focused on feelings of respect, affection, and reassurance, whereas negative items queried about feelings of anger, frustration, and guilt. However, we did not use positive items in the analysis because of a lack of variance. We created one additive index of total negative interactions on the job (regardless of with whom the negative interactions occurred) based on how many negative interactions were reported. In addition, DCWs reported the frequency of hearing racial or ethnic remarks made by residents or other staff members. We also used these two items on the frequency of racial or ethnic remarks by residents or other staff as independent measures of negative support.
Organizational Variables
Background Characteristics
We created several descriptive organizational variables for use in the analyses. The variables included type of LTC setting (assisted living/home health agency = 0; nursing home = 1); profit status (not for profit = 0; for profit = 1); percentage of minority residents/clients served; and percentage of resident/client services reimbursed through (a) Medicaid, (b) Medicare, or (c) private pay.
Organizational Issues
Based on the literature, we included selected measures of organizational issues. For example, the organizational survey included information on the extent to which the organization faced difficulties with turnover (4-point scale ranging from no difficulty to a great deal of difficulty) because of DCWs quitting or being fired, and the minimum starting hourly rate of pay for DCWs.
Data Analyses
Data analysis required several phases. In the first phase, we used a correlation matrix of all DCW independent variables in the model and the dependent variable to check for multicollinearity. Although an intercorrelation of.80 or higher is an example of a cutoff value at which to begin excluding predictors as being too similar (Stevens, 2002, p. 93), given the large number of predictors being considered, we used a more conservative cutoff of.50 for deciding to keep or exclude variables. The correlation of the outcome with the predictors retained in the analyses are included in Table 2.
|
In the third phase, we addressed Research Question 2. In this phase, we used a two-level hierarchical linear model (HLM) to determine the extent to which organizational factors could predict DCW job satisfaction after we controlled for the effects of individual-level DCW predictors. We considered HLM the most appropriate statistical technique to use because study data were nested, in that a number of DCWs from each study site participated in the interviews.
We entered the significant predictor variables that we had identified earlier (Research Question 1) in Level 1 (individual- or DCW-level data) of the HLM analyses. In Level 2 (organizational-level data), we entered key organizational characteristics and management issues in the HLM analyses. We selected organizational variables based on their importance in the literature and used an iterative process to select only those that emerged as consistently significant. Due to the relatively small number of organizations (49), we entered no more than four organizational characteristics and management issues (using a ratio of 1 variable to 10 cases/organizations) at any one time in Level 2 of the HLM analyses.
|
Results |
|---|
|
TOPAbstractConceptual ModelMethodsResultsDiscussionReferences |
|---|
|
On average, DCWs had been asked to make changes to their work schedule five times within the 2 months prior to their interview. Also, DCWs tended to perceive their job orientation and continuing education as fairly useful. Almost 75% of DCWs received paid holidays off, but only 49% received fully paid health insurance. On average, a DCW experienced one or two negative interactions with other staff or residents out of a possible three negative interactions. DCWs reported hearing racial/ethnic remarks more frequently from residents than from other staff.
Regarding organizational characteristics, Table 4 includes descriptive data on the three types of facilities. There was good representation of for-profit and nonprofit organizations (28 and 21, respectively). The organizations served, on average, 21% minority clients, although they ranged from serving no minority clients/residents to more than 90% minority clients/residents. Reimbursement sources also varied, with the average organization receiving about 40% of its reimbursement from Medicaid.
|
.001), the job-related stressors ( R2 change =.20, p .001), and workplace support variables (R2 change =.05, p .001). Thus, each step of the hierarchical regression analysis resulted in significant changes in the R2 value.
|
Among the job-related stressors that were entered next, we identified significant predictors in relation to scheduling changes, training issues, and pay and benefits. Specifically, DCWs who perceived they had better on-the-job training in terms of the usefulness of continuing education and job orientation had higher job satisfaction. In terms of pay and benefits, DCWs who reported being fairly compensated for their job, having a retirement/pension plan, and having paid health insurance had higher job satisfaction.
With respect to workplace support variables that were entered last, racism and negative interactions were significant predictors of DCW job satisfaction. When DCWs reported hearing fewer racial or ethnic remarks from other staff and experienced fewer negative interactions at work, their job satisfaction was higher.
Organizational-Level Variables
We included significant DCW-level predictors that we identified in Phase 1 in Level 1 of the two-level HLM model. The y-intercept (β0j) estimated from the Level 1 model represents the adjusted average job satisfaction scores of the jth LTC facility. Because the Level 1 predictors were group mean centered, the adjusted average represented the job satisfaction score for a typical DCW in a specific facility. We treated this index of job satisfaction as the outcome variable in Level 2 of the HLM analysis.
In Level 1 of the HLM analyses, all of the individual-level predictors identified in the multiple regression, except for race of DCW, emerged as significant predictors of the adjusted average job satisfaction level in a facility. With respect to organizational factors identified in Level 2 of the HLM analyses, the following were significant: type of LTC facility (i.e., nursing home vs home health agency and assisted living facility), difficulty with turnover of DCWs, and minimum hourly rate of pay for DCWs (see Table 6). Nursing homes had lower adjusted average job satisfaction scores compared to home health agencies and assisted living facilities. Greater difficulty with DCWs quitting or being fired had a negative relationship, whereas higher rates of starting wages had a positive relationship with the LTC facility's adjusted average DCW job satisfaction score. The percentage of minority clients served was not a significant predictor of the adjusted average DCW job satisfaction score.
|
|
Discussion |
|---|
|
TOPAbstractConceptual ModelMethodsResultsDiscussionReferences |
|---|
Among the personal stressors, physical and emotional health changes since working as a DCW and depression were significant. From a practice perspective, LTC organizations need to be sensitive to perceived changes in the health of DCWs. Offering workplace health screening, employee assistance, and health promotion programs may be efficacious in addressing DCWs' personal stress in order to enhance their job satisfaction.
In addition to personal stressors, a variety of job-related stressors were significant predictors of job dissatisfaction. They fell into the following categories: issues related to scheduling changes, training, and pay and benefits. Organizations can alleviate some job-related stress by offering better job orientation and continuing education programs. In fact, DCW-related findings from our study indicate that DCWs want better training in specific content areas and in the methods for providing training (Menne, Ejaz, Noelker, & Jones, 2007).
Although LTC organizations are not always amenable to offering better pay and benefits, it is clear that feeling fairly compensated for the job and receiving key benefits are significantly related to DCW job satisfaction. Of all of the benefits included in our study, two were significant: health insurance and a retirement/pension plan. Other benefits such as paid holidays and paid sick leave did not emerge as significant. Because studies have suggested that job satisfaction is related to turnover (Castle et al., 2006), further research needs to be conducted examining the business case for offering better wages and benefits in relation to the cost of turnover.
In addition to addressing sources of job-related stress, organizations can also focus on improving communication patterns and relationships in the workplace. Our findings suggest that perceptions of racism by other staff and negative interactions with staff and residents are likely to generate greater job dissatisfaction. With respect to racism, our findings support those of other studies in that DCWs were more likely to condone racist remarks from residents and clients because they believed that the elders were too impaired to be maliciously racist (Berdes & Eckert, 2001). Hence, racism from staff was a significant predictor of job satisfaction but racism from residents was not, even though 70% of the DCWs in our sample had heard residents and clients make racist remarks. Thus, having sensitivity training on racial and cultural differences; promoting effective communication between residents, families, and staff via newsletters, brochures, and discussion groups; and having a no-tolerance policy on discrimination are likely to enhance workplace support.
In terms of differences between organizations, it is interesting to find that nursing homes had lower DCW job satisfaction levels compared to home health agencies and assisted living facilities. Additionally, organizations providing a higher minimum starting rate of pay for DCWs and those not struggling with turnover issues were more likely to have higher levels of DCW job satisfaction. In a low-income population of working women, the issue of pay is likely to be linked to the alarming rates of DCW turnover in the industry (Harris-Kojetin et al., 2004; U.S. Department of Health and Human Services, 2003, 2004; U.S. General Accounting Office, 2001).
The limitations of this study include the following: (a) the inclusion of sites from a restricted geographical area (five-county area in Ohio), (b) the inability to obtain a random sample of DCWs due to the challenges involved in contacting and completing interviews with DCWs who volunteered to participate, and (c) the lack of longitudinal data to examine cause and effect between variables in the model. Thus, the findings from this study need to be viewed with caution, particularly those regarding the differences between nursing homes and other LTC organizations. We could not conduct a more in-depth examination of the differences between the three types of LTC settings in our sample because there were fewer home health agencies and assisted living facilities than nursing homes due to the proportionate sampling techniques used in the study. To examine the effect of organizational variables while controlling for DCW individual-level factors by type of LTC setting, HLM would require a minimum of 30 organizations for each of the three types of settings. Thus, future studies using larger samples of organizations are required to effectively address differences by type of setting.
Similarly, although DCWs in Ohio are likely to have characteristics similar to other samples of DCWs in terms of gender and income level, the study questions and the conceptual model need to be tested with larger, more generalizable samples.
One of the major strengths of the study lies in the use of a conceptual model that supported using two levels of data (i.e., DCW and organizational levels) to predict job satisfaction. Other strengths include the preliminary examination of three types of LTC settings and the large sample of 644 DCWs who completed in-depth interviews. Future research with larger, more representative samples and a longitudinal design can lead to further refinements in the LTC stress and support model for examining the relationship between job satisfaction and actual turnover among DCWs.
|
Footnotes |
|---|
1 Margaret Blenkner Research Institute, Benjamin Rose Institute, Cleveland, OH.
2 College of Education and Human Services, Cleveland State University, OH.
|
References |
|---|
|
TOPAbstractConceptual ModelMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| All GSA journals | Journals of Gerontology Series A: Biological Sciences and Medical Sciences | Journals of Gerontology Series B: Psychological Sciences and Social Sciences |
