- Split View
-
Views
-
Cite
Cite
Laura Pekkarinen, Timo Sinervo, Marja-Leena Perälä, Marko Elovainio, Work Stressors and the Quality of Life in Long-Term Care Units, The Gerontologist, Volume 44, Issue 5, October 2004, Pages 633–643, https://doi.org/10.1093/geront/44.5.633
- Share Icon Share
Abstract
Purpose. The purpose of this work was to examine how structural factors, residents' needs for physical and psychosocial assistance, and the work stressors experienced by employees are related to the quality of life of elderly residents in long-term care. Design and Methods. Cross-sectional survey data were collected from 1,194 employees and 1,079 relatives of residents in 107 residential-home units and health-center bed wards. Data were analyzed using multilevel modeling. Results. The majority of differences in both employees' and relatives' perceptions of residents' quality of life across units could be explained by work stressors such as time pressure. Large unit size was related to both increased time pressure among employees and reduced quality of life of residents. Implications. Long-term care units are encouraged to review their practices so that employee well-being is supported. Attention also should be focused on unit size, as small units appear better able to help employees cope with work stress, resulting in better quality of life for residents.
A growing body of evidence suggests that work stressors adversely affect health-care staff job performance and thus influence the quality of care received by or the quality of life of the patients (Cohen-Mansfield, 1995; Garman, Corrigan, & Morris, 2002; Hannan, Norman, & Redfern, 2001; Jamal, 1984; Jones, Barge, Steffy, Fay, Kunz, & Wuebker, 1988; Motowidlo, Packard, & Manning, 1986; Packard & Motowidlo, 1987). Leveck and Jones (1996) examined 50 nursing units of four acute-care hospitals and found that those units with lower levels of work stress exhibited better care processes. Similar results were found in a study by Leiter, Harvie, and Frizzell (1998) on 16 inpatient units; in those units where employees felt exhausted, the patients were less satisfied with various components of their care.
Prior stress research on the relationships between work stressors and quality of care in health-care settings has mostly focused on acute care. Few investigations have examined the ways in which work stressors affect employee job performance in long-term care (Hannan et al., 2001; Schaefer & Moos, 1996). Hannan and colleagues (2001) reviewed recent research conducted in long-term care settings and, despite the inconsistencies between studies, found empirical support for the relationships between structural factors (such as staffing level and unit size), work stressors, and both the quality of care and the well-being of residents. In caregiving, the characteristics of elderly residents, such as their dependency on assistance in performing activities of daily living, also have been found to affect employee stress (Chappell & Novak, 1992; Chappell & Reid, 2002; Pearlin, Mullan, Semple, & Skaff, 1990).
Structural factors may in turn affect employee work stressors and job performance (Gray-Toft & Anderson, 1981; Hannan et al., 2001). Hannan and colleagues concluded in their review that improvements in organizations' staffing levels are related to reduced work stress among employees and to increased resident satisfaction with care. Evidence for the effect of unit size on quality of care is far less conclusive (Davis, 1991).
Three of the most widely studied work stressors in care for elderly residents are time pressure (or work overload), role ambiguity, and resident-related stressors (Chappell & Novak, 1992; Cohen-Mansfield, 1995; Elovainio & Sinervo, 1997; Hannan et al., 2001; Schaefer & Moos, 1996; Sheridan, White, & Fairchild, 1992; Sinervo, 2000). Numerous studies indicate that work overload has a negative influence on health-care staff job performance and on the quality of care (Gray-Toft & Anderson, 1981; Hannan et al., 2001; Schaefer & Moos, 1996). Role ambiguity refers to a lack of clear goals and clarity in the behavioral requirements of one's job (Rizzo, House, & Lirtzman, 1970). This has been shown to unfavorably affect job performance and patient care in hospital settings (Gray-Toft & Anderson, 1981; Jamal, 1984). Finally, work in long-term care settings consists of caring for chronically ill residents, which creates occupation-specific stressors related to the intensive interaction between employees and residents. When these interactions involve negative affect from residents, this stressor is likely to result in employee distress (Schaefer & Moos, 1993, 1996; Sinervo, 2000). However, it remains unclear how this resident-related stressor affects employee job performance.
Employees' job performance is often studied in terms of the quality of the care they provide. Quality of care has been described as a multidimensional concept that has proved to be difficult to conceptualize (Davis, 1991; Zimmerman, 2003). Donabedian's now-classic framework of structure, process, and outcome (1980, 1988) introduced the importance of assessing the care process, that is to say, “what is actually done in giving and receiving care” (Donabedian, 1988), as an antecedent of patient outcomes and thus the quality of care. Systematic assessments of clinical quality have produced quality indicators that capture both the processes and the outcomes of care (Zimmerman, 2003). However, the outcomes related to residents' quality of life are often minimized in these assessments (Kane, 2001; Kane & Kane, 2000).
Quality of life in long-term care is suggested to consist of domains such as autonomy, dignity, individuality, privacy, enjoyment, meaningful activity, relationships, security/safety, comfort, spiritual well-being, and functional competence (Kane & Kane, 2000). These domains can be considered to stem from the interpersonal processes that arise between health-care staff and residents in meeting individual needs (Donabedian, 1988). Assessment of processes in caregiving is especially important in settings where residents cannot themselves give feedback, as their quality of life can be influenced through practices of institutions (Frytak, 2000; Kane, 2003). In this study, residents' quality of life was defined in terms of the interpersonal processes that occur in carrying out the tasks of daily life. Both employees' and relatives' perceptions were used in investigating how sufficiently employees help residents with their activities of daily living and support residents' individuality.
The activities of living and the support for individuality have been introduced by Roper, Logan, and Tierney (2002) in their model of nursing. The concept of activities of living encompasses quality-of-life domains such as the residents' functional competence, comfort, safety, meaningful activity, relationships, and enjoyment. The model also emphasizes the concept of individualized nursing that supports the residents' individuality, autonomy, and dignity.
Research results show a mainly negative linear relationship between work stress and employee job performance (Jamal, 1984, 1985; Muse, Harris, & Feild, 2003; Westman & Eden, 1996). However, some results suggest only a weak relationship or no relationship between stress and job performance (Jamal, 1984; Sullivan & Bhagat, 1992). Existing findings in occupational stress research have the following two methodologic limitations: (a) reliance solely on self-reports as reflections of objective stressful job conditions and (b) a lack of a multilevel perspective that views organizations as multilevel systems in which the social context influences individual perception and performance.
In addition to self-reports, objective data on job conditions should be collected (Spector, Dwyer, & Jex, 1988; Sullivan & Bhagat, 1992). In this study, data were collected on both residents' needs and structural factors of units. These can be considered as explanatory variables of employee work stressors, although research results have shown inconsistent effects of residents' dependency on caregiver stress (Cohen-Mansfield, 1995; Pearlin et al., 1990; Schaefer & Moos, 1993, 1996).
A drawback in most studies concerns the lack of a multilevel perspective that recognizes employees rarely work alone but are instead usually embedded within “work groups” in the organization (Bliese & Jex, 1999, 2002). There is some dependence between how individuals and how their colleagues within the work group perceive and react to shared job conditions (Bliese & Jex, 1999; Kozlowski & Klein, 2000). In other words, interaction among work colleagues is likely to influence which job conditions individuals consider as stressors and how individuals respond to these stressors (Bliese & Jex, 2002).
A work group's shared perceptions of job conditions can be measured as aggregate variables (typically as group means); these aggregated variables may represent the actual job conditions more reliably than do individual employees' self-reports (Bliese & Jex, 1999). Health-care employees are nested within units that form rather homogeneous natural groupings with their own resources, supervisors, and objectives (Klein, Tosi, & Cannella, 1999). For example, Garman and colleagues (2002) confirmed that employee burnout does exist as a meaningful group-level construct in health-care teams. They further discovered significant negative relationships between employee burnout and resident satisfaction with care at the group level.
However, the group-level analyses provide little insight into how individual employees within work groups differ in their perceptions or reactions. Researchers have been encouraged to apply multilevel models that recognize the hierarchical structure of the kind of data that they typically examine (Bliese & Jex, 1999, 2002; Elovainio, Kivimäki, Steen, & Kalliomäki-Levanto, 2000). Ignoring this hierarchical structure can have serious consequences such as the inducement of high collinearity among explanatory variables and biased standard errors for the estimates (Goldstein, 1995). Multilevel models incorporate both individual and group levels of analysis and provide statistically efficient regression coefficient estimates and covariates at different levels of hierarchy (Goldstein, 1995). In the current study, we had the possibility to take into account the hierarchical structure of the data by using a proper statistical method for a large sample of units with both employee and relative respondents.
Based on previous research results, we hypothesized the following: (a) Residents' needs for assistance in their daily activities and unit structural factors (unit size and staffing level) are related to employee work stressors; (b) in addition to variance in individual perceptions of residents' quality of life, there are differences in employees' and relatives' perceptions of quality between units; (c) differences in these quality perceptions between units can be explained by residents' needs, unit structural factors, and employee work stressors in that increased levels of employee work stressors are related to reduced quality of life for residents; and (d) subjective work stressors act as mediators through which the objective job conditions affect employee job performance. The research model is presented in Figure 1.
Methods
Participants and Procedure
The data used here were drawn from a study on the quality of long-term institutional care for the elderly in Finland in which a representative sample of elderly residents in long-term care facilities was formed by means of cluster sampling. First, a stratified sample of facilities was selected from the Finnish Municipal Database for Social and Health Statistics according to the facility type. A random sample of facilities was then selected from each facility type, and all the residents and employees in the selected facilities formed the sample of the study. The data were collected with three surveys conducted in long-term care institutions in early 1999. Data were drawn from self-report questionnaires circulated among the employees, of whom 63% returned the questionnaires. Questionnaires for the relatives were available in the units for a 2-week period. Participating relatives represented one-third of the residents who were cared for in the units. In addition to these, head nurses provided information on unit structural factors (number of residents and the staffing ratio per residents in units).
The study sample consisted of 107 units (34 residential-home units and 73 health-center bed wards) and included 1,194 employees and 1,079 relatives of residents. Units with less than three responding staff members or relatives were eliminated from this sample, which led to the selection of fewer residential-home units in the sample. The mean number of residents per unit was 34 (SD = 8.6, range = 12–50), and the mean number of employees per unit was 20 (SD = 7.0, range = 5–41). There was a mean of 0.6 employee per resident (SD = 0.2, range = 0.23–0.97). Of the 1,194 employees, 53% were licensed practical nurses, 28% were registered nurses, 14% were nursing assistants, and 5% were head nurses. The majority (92%) worked full-time: 70% rotated between three shifts, 23% rotated between two shifts, and 4% worked the day shift. The employees' mean age was 43 (SD = 9.0) years, and most (65%) had worked at their current jobs for >2 years.
The relatives (n = 1,079) were mostly women (68%), and 64% were offspring of the residents. The mean age of the relatives was 58 (SD = 11.6) years. Nearly half (42%) were relatives of residents who had been treated in the unit for ≥2 years. According to head nurses, 35% of the residents were completely dependent in their daily activities and 52% were classified as suffering from dementia.
In Finland, the local municipalities are legally responsible for providing long-term care services for their residents (Hermanson, Aro, & Bennett, 1994). Services are provided at two types of institutional long-term care facilities: health-center hospitals and residential homes. Public health-center bed wards are traditionally administered through health-care services, and they mainly take care of the most severely ill long-term patients, although acute care is also provided. Residential homes are the most common form of institutional care, and most of these are publicly run by social service agencies. Typically, in residential homes, care is provided for elderly people with fewer care needs (Noro & Aro, 1997). On average, staff in health-center hospitals are more highly trained than those in residential homes: Registered nurses represent one-third of the staffing in health-center bed wards, and nursing assistants account for only 10%. The average proportion of registered nurses in residential homes is 20% and of nursing assistants 25%. In both facilities, licensed practical nurses form the largest occupational group, representing about half of the staff (Vaarama, Kainulainen, Perälä, & Sinervo, 1999).
Measures
Residents' need for assistance
The scales developed to measure residents' needs for assistance in physical and psychosocial activities were based on the classification of human needs by Roper and colleagues (2002). These measurement scales have been shown to have adequate structural validity (Perälä & Räikkönen, 2000). Employees were asked to assess the proportion of residents in the unit needing assistance with nine different physical activities on a 4-point rating scale (ranging from “none” to “most”). These activities included eating and drinking, eliminating, personal cleansing, dressing, mobilizing, changing position for bedridden patients, sleeping, controlling body temperature, and pain relief. Assistance with seven psychosocial activities also was assessed: recreation, maintaining a safe environment, contact with relatives, expressing sexuality, grief work, dying, and remembering. These item ratings were used to calculate mean scores ranging from 1 to 4, where larger values indicate that most of the residents in the unit needed assistance with these activities. Cronbach's reliability α for the scale measuring the proportion of residents in need of assistance with physical activities was.83 and that for the scale measuring assistance with psychosocial activities was.82.
Relatives were asked to assess residents' needs for physical and psychosocial assistance using the same item inventories as the employees. However, they assessed individual residents' needs rather than the proportion of residents who needed assistance. Instead of using the 4-point rating scale, relatives used divergent 2-point rating scales (0 = resident does not need assistance; 1 = resident needs assistance with the activity). Ratings for these two groups of activities were summed up as measurement scales: one for the need for assistance with physical activities (Cronbach's α =.79), which could have scores ranging from 0 to 9, and one for the need for assistance with psychosocial activities (Cronbach's α =.76), which could have scores ranging from 0 to 7. For both scales, large values indicate that a resident needs assistance with most of these activities.
Residents' quality of life
Due to the considerable number of residents in the sample who had difficulties in giving feedback themselves, proxy reports of both employees and relatives of residents were used to evaluate the residents' quality of life in units. Three measures that have adequate structural validity were used in assessing quality of life (Perälä & Räikkönen, 2000). Employees were asked to assess on a 3-point rating scale (1 = totally insufficient; 3 = sufficient) the degree of sufficiency of help that residents in the unit have been receiving for their physical and psychosocial needs. The items covering residents' needs for assistance with physical and psychosocial activities form the basis of a nine-item scale on received physical help (Cronbach's α =.84) and a seven-item scale on received psychosocial help (Cronbach's α =.83) with scores that range from 1 to 3, where large values indicate sufficient help received. The measurement scale for the client-centered practices consists of four items for which employees were asked to indicate on a 4-point rating scale (1 = totally disagree; 4 = totally agree) whether they agree that the practices in the unit include components such as kindness, individually tailored care, and autonomy. These item ratings were used to calculate scale scores ranging from 1 to 4, where larger values indicate that employees agree that the practices in the unit are client centered. Cronbach's reliability α was.72 for this scale.
Relatives were asked to assess the quality of life of individual residents with measures consisting of items identical to those in the measures for employees, but relatives used divergent rating scales ranging from 1 to 4 (1 = totally insufficient or totally disagree; 4 = totally sufficient or totally agree). The measurement scale reliability coefficients (Cronbach's α) were.91 for sufficiency of received physical help,.93 for sufficiency of received psychosocial help, and.84 for client-centered practices.
Work stressors
Work stressors were assessed using three scales: (a) time pressure (five items, Cronbach's α =.89), (b) interaction with troublesome residents (three items, Cronbach's α =.64), and (c) role ambiguity (three items, Cronbach's α =.83). The first two measures were developed based on prior research conducted among nurses and other health-care staff, which has shown that the measures have adequate internal reliability and validity (Elovainio & Sinervo, 1994, 1997; Kivimäki & Lindström, 1992). The time pressure scale measured stress due to scheduling problems and time shortages at work, and the resident-related stressor scale asked respondents to assess how often they had experienced stressful incidents associated with caring for complaining or passive residents. The measure of role ambiguity was derived from the six-item role ambiguity scale developed by Rizzo and colleagues (1970) that has gone through careful psychometric testing (Sawyer, 1992). The three-item role ambiguity measure asked respondents to indicate how often they were uncertain about tasks, responsibilities, and requirements of their job. For all the items in the three scales, respondents were asked to indicate on 5-point rating scales (1 = never; 5 = very often) how frequently they had experienced the investigated work situation as stressful.
Data Analysis
The study sample consisted of 1,194 individual employees nested within 107 elder-care units and 1,079 relatives of the residents cared for in these units. Multilevel modeling allows analyses of these types of hierarchically structured data (Goldstein, 1995). The basic structure of a multilevel model is an extension of ordinary least squares regression where the total variability in the outcome variables is decomposed into variance within and between units. The utility of multilevel models is well documented, and the methods have been applied in health settings (Bravo, Wals, Dubois, & Charpentier, 1999; Rice & Leyland, 1996). Here multilevel models were estimated using iterative generalized least squares carried out using the statistical software package LISREL 8.52 (Jöreskog, Sörbom, du Toit, & du Toit, 2000).
In Step 1, the means, standard deviations, and Pearson product–moment correlations were calculated for the relations among physical and psychosocial needs of residents, unit size, staffing ratio, work stressors, and quality-of-life measures in both the employee and the relative samples. Additionally, correlations were calculated between employees' and relatives' perceptions of quality at the unit level. Variables for each analysis were then standardized to have equal means and variances, which facilitated the interpretation of multilevel models. The employee sample was analyzed in Step 2, beginning with the calculation of the intraclass correlation (ICC) of the variables, which measures the degree of dependency in observations within units.
Three multilevel models were constructed for three outcome variables: sufficiency of received physical help, sufficiency of received psychosocial help, and client-centered practices. Null models provided estimates of random effect variances of the outcome variables, which indicated the extent to which the quality measures varied across individuals (level 1 variance) and units (level 2 variance). The effects of explanatory variables on the outcomes were then estimated by including residents' needs for physical and psychosocial assistance measured at the individual level and unit size and staffing ratio per resident as unit-level structural factors in Model A and Model B and by further including time pressure, resident-related stressor, and role ambiguity as individual-level work stressors in Model B. The statistical significance of the improvements in model fit was tested based on the reduction in the log likelihood statistics (deviance).
Step 3 involved standardized data from the sample of relatives. Two multilevel models were constructed on each of the three outcome variables. Null models were fitted first, followed by the inclusion in full models of the explanatory variables of residents' needs for physical and psychosocial assistance as individual-level measures, unit size and staffing ratio per resident as unit-level structural factors, and unit-level means of employee work stressors. Only work stressors showing significant ICC in Step 2 were chosen for use in the models.
Results
The results of the Step 1 analyses are presented in Tables 2 and 3. In the employee sample, there was a moderate positive correlation between residents' needs for physical assistance and the time pressure and the resident-related stressor, whereas the psychosocial needs of residents were not related to any of the work stressors. Large unit size correlated strongly with increased time pressure. The three quality-of-life measures were negatively related to unit size, but neither residents' needs for physical assistance nor their needs for psychosocial assistance correlated with the quality-of-life measures. Of the work stressors, time pressure had strong negative correlations with the sufficiency of received physical and psychosocial help and a moderate negative correlation with client-centered practices. Other measured work stressors were moderately related to the quality of life.
As among employees, relatives' perceptions of the psychosocial needs of residents were not related to work stressors, and their perceptions of residents' physical needs correlated positively with the resident-related stressor but not with time pressure. Unit size yielded a strong positive correlation with time pressure. Unlike among employees, in the sample of relatives, strong negative correlations were found between the three quality-of-life measures and residents' needs for physical and psychosocial assistance, whereas large unit size correlated only slightly with reduced quality of life. Increased time pressure among employees was related to decreased sufficiency of received physical and psychosocial help. The three quality-of-life measures correlated highly with each other in the relatives' sample. Additional analyses indicated, however, that employees' and relatives' perceptions of quality of life correlated with each other only moderately: Pearson correlation coefficients were.29 (p =.00) for received physical help,.21 (p =.03) for received psychosocial help, and.02 (p =.80) for client-centered practices.
Step 2 of the analyses was conducted for the employee sample to determine the effects of residents' needs, structural factors, and work stressors on quality of life. ICC analyses indicated a lack of independence in observations within units for certain variables. The ICC for employees' perceptions of residents' needs for physical assistance (.30) was significant and that for their perceptions of residents' needs for psychosocial assistance was moderate (0.13). Of the work stressors, time pressure showed at least a moderate ICC (.22). The ICC estimates for the resident-related stressor (.09) and for role ambiguity (.05) were small.
Tables 4–6 present the three models constructed for each of the outcome variables: sufficiency of received physical help, sufficiency of received psychosocial help, and client-centered practices. The null models indicate that the received psychosocial help varied most between units, with the level 2 intercept variance being 22% of the total variance. The sufficiency of received physical help and client-centered practices also showed significant variance between units (the level 2 intercept variance was 19 and 12% of the total variance, respectively). The considerable amount of variance between units in outcome variables indicates that the employee data have a hierarchical structure and that the variation in quality of life arises from something other than differences between individuals. It was thus appropriate to proceed with multilevel modeling.
Multilevel models (A models) indicated that the fixed effect of residents' needs for physical assistance was statistically significant in explaining the sufficiency of received physical help. The residents' needs for psychosocial assistance had a slight negative effect on the psychosocial help received but a slight positive effect on client-centered practices. Unit size had significant negative effects on all three quality-of-life measures. Time pressure, resident-related stressor, and role ambiguity were included in the final models (B models), which reduced the magnitude of the effect of unit size on all three quality-of-life measures. The inclusion of work stressors also diminished the effect that residents' physical needs had on the sufficiency of received physical help in Model A but had very little impact on the effects of residents' psychosocial needs on the two other quality-of-life measures.
Time pressure had the largest negative effect on received physical and psychosocial help, whereas the work stressor role ambiguity was closely associated with client-centered practices. The resident-related stressor had a slight negative effect on the sufficiency of received physical help and on client-centered practices, but the coefficient was statistically nonsignificant in the model relating to the sufficiency of psychosocial help received.
As shown in the final models (B models), work stressors accounted for most of the variance between units in all quality-of-life measures: When the work stressors were included in the analysis, the level 2 intercept variances of received physical help, of received psychosocial help, and of client-centered practices fell by 63, 52, and 25%, respectively. The final models explained the variance between units so that R2B =.68 in the model for the sufficiency of received physical help, R2B =.54 for the sufficiency of received psychosocial help, and R2B =.35 for the client-centered practices. The overall reduction in the variance between individuals was modest, being 12% of the level 1 variance of received physical help (R2W =.12), 9% of that of received psychosocial help (R2W =.09) and 7% of that of client-centered practices (R2W =.07). The final models fit the data significantly better than did the null models, with only 7 df lost. The difference in deviance (i.e., likelihood ratios) for the received physical help models was 208.55 (p =.000), that for the received psychosocial help models was 148.63 (p =.000), and that for the models for client-centered practices was 95.38 (p =.000).
In Step 3, relatives' estimates of quality of life were explained in two models that are shown in Table 7. Slightly different patterns can be found in these models compared with those constructed from employee data. First, the null models indicate that the relatives' quality-of-life perceptions vary less between units than do the employees'. The amount of level 2 variance out of the total variance was 7% for received physical help, 8% for received psychosocial help, and 10% for client-centered practices. Second, in the final models (full models), the residents' needs for assistance with physical and psychosocial activities had significant fixed effects on all quality-of-life measures, whereas the coefficient for unit size was statistically nonsignificant. The coefficient for time pressure was statistically significant or almost significant in all analyses (full models).
The models based on data from relatives accounted for much less of the variance in quality of life between units than did those based on the employee data. Compared with the null models, the level 2 intercept variance of received physical help fell by 49% (R2B =.49), that of received psychosocial help fell by 31% (R2B =.31), and that of client-centered practices fell by 13% (R2B =.13) in the final models where all the explanatory variables were included. The corresponding reduction in variance between individuals was only 6% of received physical help (R2W =.06), 6% of received psychosocial help (R2W =.06), and 3% of client-centered practices (R2W =.03). The final models fit the data significantly better than the null models, with 5 df lost. The differences in deviance (i.e., likelihood ratios) were 73.85 (p =.000) for models of sufficiency of received physical help, 72.69 (p =.000) for models of sufficiency of received psychosocial help, and 35.89 (p =.000) for models of client-centered practices.
Discussion
The purpose of this study was to examine the relationships between residents' needs, structural factors, employee work stressors, and the quality of life of the elderly residents in long-term care units. As hypothesized, residents' needs for extensive assistance with daily physical activities and large unit size were related to increased levels of work stressors experienced by employees. Employees suffered from increased time pressure particularly in larger units. Residents' psychosocial needs were not related to work stressors. Both employees' and relatives' perceptions of the quality of life of elderly residents differed between caregiving units, and the differences were to a large extent explained by the structural factors of units and by work stressors experienced by employees. The needs of residents and structural factors alone explained only a fraction of the differences in quality of life between units. However, unit size in particular appeared to be indirectly related to quality of life through work stressors occurring in units. The larger the unit, the more work stressors present and the poorer the quality of life. The resident-related stressors, role ambiguity, and especially time pressure were related to reduced quality of life in units. Support for the relationships between high work stress and poor quality of life was found upon examination of both employees' and relatives' ratings of residents' quality of life.
Similar to prior research results, only modest links were found between structural factors alone and the quality of care (Davis, 1991; Kruzich, Clinton, & Kelber, 1992). Here an effort was made to control also for residents' dependency and the mediating effect of employee stress (Hannan et al., 2001; Kruzich et al., 1992). In addition, the hierarchical structure of the data was acknowledged by using a multilevel methodology, an approach that has rarely been used when investigating the factors affecting the quality of institutional care (Bravo et al., 1999). Data from a sample of >100 elder-care units were used, a sample that is substantially larger than in most previous studies. Multilevel modeling showed that 10–20% of the variation in quality of life lies at the unit level. As prior research on health-care staff and quality of care has been mostly based on small samples with few organizational units, the variation occurring between units simply may not have been large enough for the relationships to be found (Garman et al., 2002; Hannan et al., 2001; Kruzich et al., 1992; Leveck & Jones, 1996).
The results of multilevel analyses indicated that employees and relatives share perceptions of residents' quality of life within units to some extent. These perceptions were influenced by structural factors and especially by work stressors experienced by employees in units. However, in the future, multiple perceptions of different quality-of-life domains should be used to capture the nuances of quality in long-term care. Many of the differences between units in employees' perceptions of how sufficiently residents were helped could be explained by structural factors and work stressors, whereas their perceptions of the client-centered practices were not associated with these factors. Additionally, employees' and relatives' perceptions of quality of life in units were somewhat different. Kane (2003) has referenced similar results and suggested that relatives' perceptions may be inaccurate if they visit the residents infrequently. Unlike among employees, the relatives' perceptions of quality of life were mostly affected by individual residents' needs for physical and psychosocial assistance. The more residents were in need of assistance, the poorer the quality of life reported. This may be one reason why relatives' perceptions of quality differed only modestly between units. Some of the heterogeneity in relatives' perceptions within units may also be explained with the fact that relatives and employees used somewhat different measurement tools to assess residents' needs and quality of life. Relatives assessed needs and quality of life related to individual residents, whereas employees rated residents' needs and quality of life at the unit level.
The use of both objective data and self-reports on work stressors, together with a multilevel analytical approach, offered an efficient way to examine the nature of work stressors. Although work stress is regarded as the experience of an individual and most of the variance in stress was found between individual employees, units also appeared to differ in how work stressors were perceived by employees, and work stressors were related to structural factors such as the unit size. This implies that employees working together share perceptions of the unit's job conditions that are associated with stress. Work stressors were in turn strongly related to reduced quality of life of residents. The results indicate that it is not only the objective job conditions (such as staffing level or unit size) but rather stressors experienced by employees that are important in determining the stress-related outcomes (Spector et al., 1988). This has been widely shown in stress research on employee health (Arnold, Cooper, & Robertson, 1995), but the current study included the job stressor effect on performance outcomes.
There are some limitations that should be borne in mind when considering the results of this study. One is the reliance on cross-sectional data, which prevents strong causal conclusions from being made about the relationships between residents' needs, structural factors, work stressors, and quality of life. Another is the conceptualization and measurement of residents' quality of life applied in this study. Residents' quality of life was defined as the caregiving processes that occur between employees and residents in the activities of daily living and as support of the individuality of the residents. These processes encompass many of the suggested domains of the quality of life of long-term residents, but our operational definition of quality of life is by no means comprehensive (Kane & Kane, 2000).
The use of proxies may also have introduced bias into the measures of residents' quality of life (Frytak, 2000; Kane, 2003). Clearly, the quality-of-life measures should include residents' own experiences of their quality of life. Here validity on measuring residents' quality of life was sought by using both employees and relatives as raters of quality. These ratings were not fully comparable because different scale formats were applied for employees and relatives. In addition, we suggest that no inferences can be made about the clinical quality of care on the basis of the current data. Multiple measures, both self-assessed and clinical, are needed to achieve content validity. Clinical measures should also be used in collecting data on the characteristics of elderly residents. Finally, it should be noted that many of the differences in the quality of life of the residents still remain unexplained because most of the variation occurs in individual employees' and relatives' perceptions.
Some practical implications can be drawn from this study. To begin with, employee work stressors must be taken into account when improving the quality of life of long-term care residents. Many of the differences in residents' quality of life between units could be explained as resulting from work stressors. On the basis of this study, some solutions can also be offered as to how stress can be alleviated in long-term care settings, yet many questions remain unanswered. Clearly, stress is not directly a matter of staffing levels. However, small units or modules appear to help employees to cope better with stress and to give better care that supports residents' quality of life (Anderson & Hughes, 1993). Employees may be able to organize work in an individualized way, which may break up routine patterns, increase control over the job, and make work less scheduled. It is not only the structural factors such as staffing level or unit size or the residents' needs for extensive assistance that influence how employees experience stressors (Sinervo, 2000). There are issues of staff training, work methods, and management (Elovainio, Kivimäki, & Helkama, 2001). Improving the quality of life in long-term care requires both organizational interventions and the development of human relation practices in units.
This research project was supported by the Academy of Finland and the Finnish Work Environment Fund.
The authors acknowledge Drs. Hans-Werner Wahl and Alison McCallum for their helpful comments.
STAKES, National Research and Development Centre for Welfare and Health, Helsinki, Finland.
Decision Editor: Linda S. Noelker, PhD
. | Summary of Measures . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | Employee Sample . | . | Relative Sample . | . | ||||
Variable . | No. Items . | Range . | No. Items . | Range . | ||||
Explanatory | ||||||||
Residents' needs for assistance with daily activitiesa | ||||||||
Need for physical assistance | 9 | 1–4 | 9 | 0–9 | ||||
Need for psychosocial assistance | 7 | 1–4 | 7 | 0–7 | ||||
Structural factorsb | ||||||||
Unit size | 1 | 12–50 | 1 | 12–50 | ||||
Staffing ratio per resident | 1 | 0.23–0.97 | 1 | 0.23–0.97 | ||||
Work stressors | ||||||||
Time pressurec | 5 | 1–5 | Unit meane | 2.5–4.8 | ||||
Resident-related stressorc | 3 | 1–5 | Unit meane | 2.1–3.5 | ||||
Role ambiguityd | 3 | 1–5 | Unit meane | 1.3–2.7 | ||||
Outcome | ||||||||
Residents' quality of lifea | ||||||||
Sufficiency of received physical help | 9 | 1–3 | 9 | 1–4 | ||||
Sufficiency of received psychosocial help | 7 | 1–3 | 7 | 1–4 | ||||
Client-centered practices | 4 | 1–4 | 4 | 1–4 |
. | Summary of Measures . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | Employee Sample . | . | Relative Sample . | . | ||||
Variable . | No. Items . | Range . | No. Items . | Range . | ||||
Explanatory | ||||||||
Residents' needs for assistance with daily activitiesa | ||||||||
Need for physical assistance | 9 | 1–4 | 9 | 0–9 | ||||
Need for psychosocial assistance | 7 | 1–4 | 7 | 0–7 | ||||
Structural factorsb | ||||||||
Unit size | 1 | 12–50 | 1 | 12–50 | ||||
Staffing ratio per resident | 1 | 0.23–0.97 | 1 | 0.23–0.97 | ||||
Work stressors | ||||||||
Time pressurec | 5 | 1–5 | Unit meane | 2.5–4.8 | ||||
Resident-related stressorc | 3 | 1–5 | Unit meane | 2.1–3.5 | ||||
Role ambiguityd | 3 | 1–5 | Unit meane | 1.3–2.7 | ||||
Outcome | ||||||||
Residents' quality of lifea | ||||||||
Sufficiency of received physical help | 9 | 1–3 | 9 | 1–4 | ||||
Sufficiency of received psychosocial help | 7 | 1–3 | 7 | 1–4 | ||||
Client-centered practices | 4 | 1–4 | 4 | 1–4 |
aScales were developed for the study based on Roper, Logan, & Tierney's (2002) model of nursing.
bInformation provided by the head nurses of units.
cScales were based on prior research conducted among health care staff (Elovainio & Sinervo, 1994, 1997; Kivimäki & Lindström, 1992).
dScale was derived from Rizzo, House, & Lirtzman's (1970) role ambiguity scale.
eMean scores of work stressor scales were calculated at the unit level from the employee sample.
. | Summary of Measures . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | Employee Sample . | . | Relative Sample . | . | ||||
Variable . | No. Items . | Range . | No. Items . | Range . | ||||
Explanatory | ||||||||
Residents' needs for assistance with daily activitiesa | ||||||||
Need for physical assistance | 9 | 1–4 | 9 | 0–9 | ||||
Need for psychosocial assistance | 7 | 1–4 | 7 | 0–7 | ||||
Structural factorsb | ||||||||
Unit size | 1 | 12–50 | 1 | 12–50 | ||||
Staffing ratio per resident | 1 | 0.23–0.97 | 1 | 0.23–0.97 | ||||
Work stressors | ||||||||
Time pressurec | 5 | 1–5 | Unit meane | 2.5–4.8 | ||||
Resident-related stressorc | 3 | 1–5 | Unit meane | 2.1–3.5 | ||||
Role ambiguityd | 3 | 1–5 | Unit meane | 1.3–2.7 | ||||
Outcome | ||||||||
Residents' quality of lifea | ||||||||
Sufficiency of received physical help | 9 | 1–3 | 9 | 1–4 | ||||
Sufficiency of received psychosocial help | 7 | 1–3 | 7 | 1–4 | ||||
Client-centered practices | 4 | 1–4 | 4 | 1–4 |
. | Summary of Measures . | . | . | . | ||||
---|---|---|---|---|---|---|---|---|
. | Employee Sample . | . | Relative Sample . | . | ||||
Variable . | No. Items . | Range . | No. Items . | Range . | ||||
Explanatory | ||||||||
Residents' needs for assistance with daily activitiesa | ||||||||
Need for physical assistance | 9 | 1–4 | 9 | 0–9 | ||||
Need for psychosocial assistance | 7 | 1–4 | 7 | 0–7 | ||||
Structural factorsb | ||||||||
Unit size | 1 | 12–50 | 1 | 12–50 | ||||
Staffing ratio per resident | 1 | 0.23–0.97 | 1 | 0.23–0.97 | ||||
Work stressors | ||||||||
Time pressurec | 5 | 1–5 | Unit meane | 2.5–4.8 | ||||
Resident-related stressorc | 3 | 1–5 | Unit meane | 2.1–3.5 | ||||
Role ambiguityd | 3 | 1–5 | Unit meane | 1.3–2.7 | ||||
Outcome | ||||||||
Residents' quality of lifea | ||||||||
Sufficiency of received physical help | 9 | 1–3 | 9 | 1–4 | ||||
Sufficiency of received psychosocial help | 7 | 1–3 | 7 | 1–4 | ||||
Client-centered practices | 4 | 1–4 | 4 | 1–4 |
aScales were developed for the study based on Roper, Logan, & Tierney's (2002) model of nursing.
bInformation provided by the head nurses of units.
cScales were based on prior research conducted among health care staff (Elovainio & Sinervo, 1994, 1997; Kivimäki & Lindström, 1992).
dScale was derived from Rizzo, House, & Lirtzman's (1970) role ambiguity scale.
eMean scores of work stressor scales were calculated at the unit level from the employee sample.
Variable . | M . | SD . | 1 . | 2 . | 3 . | 4 . | 5 . | 6 . | 7 . | 8 . | 9 . | 10 . |
---|---|---|---|---|---|---|---|---|---|---|---|---|
1. Residents' physical needs | 3.43 | 0.46 | — | |||||||||
2. Residents' psychosocial needs | 3.10 | 0.62 | .50 | — | ||||||||
3. Unit size | 33.90 | 8.52 | .10 | −.02 | — | |||||||
4. Staffing ratio per resident | 0.60 | 0.15 | .06 | −.08 | −.10 | — | ||||||
5. Time pressure | 3.93 | 0.77 | .10 | .03 | .19 | .03 | — | |||||
6. Resident-related stressor | 3.01 | 0.63 | .10 | .05 | .09 | .00 | .31 | — | ||||
7. Role ambiguity | 1.92 | 0.68 | .00 | .00 | −.03 | −.08 | .28 | .26 | — | |||
8. Received physical help | 2.55 | 0.36 | −.05 | .01 | −.19 | −.02 | −.44 | −.23 | −.24 | — | ||
9. Received psychosocial help | 2.19 | 0.45 | −.03 | −.05 | −.16 | .00 | −.41 | −.18 | −.17 | .64 | — | |
10. Client-centered practices | 3.41 | 0.44 | .00 | −.08 | −.10 | .05 | −.20 | −.18 | −.24 | .39 | .37 | — |
Variable . | M . | SD . | 1 . | 2 . | 3 . | 4 . | 5 . | 6 . | 7 . | 8 . | 9 . | 10 . |
---|---|---|---|---|---|---|---|---|---|---|---|---|
1. Residents' physical needs | 3.43 | 0.46 | — | |||||||||
2. Residents' psychosocial needs | 3.10 | 0.62 | .50 | — | ||||||||
3. Unit size | 33.90 | 8.52 | .10 | −.02 | — | |||||||
4. Staffing ratio per resident | 0.60 | 0.15 | .06 | −.08 | −.10 | — | ||||||
5. Time pressure | 3.93 | 0.77 | .10 | .03 | .19 | .03 | — | |||||
6. Resident-related stressor | 3.01 | 0.63 | .10 | .05 | .09 | .00 | .31 | — | ||||
7. Role ambiguity | 1.92 | 0.68 | .00 | .00 | −.03 | −.08 | .28 | .26 | — | |||
8. Received physical help | 2.55 | 0.36 | −.05 | .01 | −.19 | −.02 | −.44 | −.23 | −.24 | — | ||
9. Received psychosocial help | 2.19 | 0.45 | −.03 | −.05 | −.16 | .00 | −.41 | −.18 | −.17 | .64 | — | |
10. Client-centered practices | 3.41 | 0.44 | .00 | −.08 | −.10 | .05 | −.20 | −.18 | −.24 | .39 | .37 | — |
Notes: All correlations above 0.10 are statistically significant (p <.001).
Variable . | M . | SD . | 1 . | 2 . | 3 . | 4 . | 5 . | 6 . | 7 . | 8 . | 9 . | 10 . |
---|---|---|---|---|---|---|---|---|---|---|---|---|
1. Residents' physical needs | 3.43 | 0.46 | — | |||||||||
2. Residents' psychosocial needs | 3.10 | 0.62 | .50 | — | ||||||||
3. Unit size | 33.90 | 8.52 | .10 | −.02 | — | |||||||
4. Staffing ratio per resident | 0.60 | 0.15 | .06 | −.08 | −.10 | — | ||||||
5. Time pressure | 3.93 | 0.77 | .10 | .03 | .19 | .03 | — | |||||
6. Resident-related stressor | 3.01 | 0.63 | .10 | .05 | .09 | .00 | .31 | — | ||||
7. Role ambiguity | 1.92 | 0.68 | .00 | .00 | −.03 | −.08 | .28 | .26 | — | |||
8. Received physical help | 2.55 | 0.36 | −.05 | .01 | −.19 | −.02 | −.44 | −.23 | −.24 | — | ||
9. Received psychosocial help | 2.19 | 0.45 | −.03 | −.05 | −.16 | .00 | −.41 | −.18 | −.17 | .64 | — | |
10. Client-centered practices | 3.41 | 0.44 | .00 | −.08 | −.10 | .05 | −.20 | −.18 | −.24 | .39 | .37 | — |
Variable . | M . | SD . | 1 . | 2 . | 3 . | 4 . | 5 . | 6 . | 7 . | 8 . | 9 . | 10 . |
---|---|---|---|---|---|---|---|---|---|---|---|---|
1. Residents' physical needs | 3.43 | 0.46 | — | |||||||||
2. Residents' psychosocial needs | 3.10 | 0.62 | .50 | — | ||||||||
3. Unit size | 33.90 | 8.52 | .10 | −.02 | — | |||||||
4. Staffing ratio per resident | 0.60 | 0.15 | .06 | −.08 | −.10 | — | ||||||
5. Time pressure | 3.93 | 0.77 | .10 | .03 | .19 | .03 | — | |||||
6. Resident-related stressor | 3.01 | 0.63 | .10 | .05 | .09 | .00 | .31 | — | ||||
7. Role ambiguity | 1.92 | 0.68 | .00 | .00 | −.03 | −.08 | .28 | .26 | — | |||
8. Received physical help | 2.55 | 0.36 | −.05 | .01 | −.19 | −.02 | −.44 | −.23 | −.24 | — | ||
9. Received psychosocial help | 2.19 | 0.45 | −.03 | −.05 | −.16 | .00 | −.41 | −.18 | −.17 | .64 | — | |
10. Client-centered practices | 3.41 | 0.44 | .00 | −.08 | −.10 | .05 | −.20 | −.18 | −.24 | .39 | .37 | — |
Notes: All correlations above 0.10 are statistically significant (p <.001).
Variable . | M . | SD . | 1 . | 2 . | 3 . | 4 . | 5 . | 6 . | 7 . | 8 . | 9 . | 10 . |
---|---|---|---|---|---|---|---|---|---|---|---|---|
1. Residents' physical needs | 7.06 | 2.13 | — | |||||||||
2. Residents' psychosocial needs | 5.08 | 1.91 | .37 | — | ||||||||
3. Unit size | 33.90 | 8.52 | .05 | −.01 | — | |||||||
4. Staffing ratio per resident | 0.60 | 0.15 | .18 | −.01 | −.10 | — | ||||||
5. Time pressurea | 3.89 | 0.42 | .09 | .05 | .27 | .12 | — | |||||
6. Resident-related stressora | 2.99 | 0.27 | .15 | .01 | .21 | −.05 | .32 | — | ||||
7. Role ambiguitya | 1.92 | 0.26 | .02 | .03 | −.10 | −.14 | .32 | .24 | — | |||
8. Received physical help | 3.39 | 0.56 | −.23 | −.21 | −.07 | −.08 | −.12 | −.10 | −.01 | — | ||
9. Received psychosocial help | 3.20 | 0.73 | −.20 | −.23 | −.05 | −.07 | −.11 | −.10 | −.03 | .77 | — | |
10. Client-centered practices | 3.41 | 0.59 | −.14 | −.15 | −.09 | −.04 | −.07 | −.14 | −.07 | .60 | .61 | — |
Variable . | M . | SD . | 1 . | 2 . | 3 . | 4 . | 5 . | 6 . | 7 . | 8 . | 9 . | 10 . |
---|---|---|---|---|---|---|---|---|---|---|---|---|
1. Residents' physical needs | 7.06 | 2.13 | — | |||||||||
2. Residents' psychosocial needs | 5.08 | 1.91 | .37 | — | ||||||||
3. Unit size | 33.90 | 8.52 | .05 | −.01 | — | |||||||
4. Staffing ratio per resident | 0.60 | 0.15 | .18 | −.01 | −.10 | — | ||||||
5. Time pressurea | 3.89 | 0.42 | .09 | .05 | .27 | .12 | — | |||||
6. Resident-related stressora | 2.99 | 0.27 | .15 | .01 | .21 | −.05 | .32 | — | ||||
7. Role ambiguitya | 1.92 | 0.26 | .02 | .03 | −.10 | −.14 | .32 | .24 | — | |||
8. Received physical help | 3.39 | 0.56 | −.23 | −.21 | −.07 | −.08 | −.12 | −.10 | −.01 | — | ||
9. Received psychosocial help | 3.20 | 0.73 | −.20 | −.23 | −.05 | −.07 | −.11 | −.10 | −.03 | .77 | — | |
10. Client-centered practices | 3.41 | 0.59 | −.14 | −.15 | −.09 | −.04 | −.07 | −.14 | −.07 | .60 | .61 | — |
Notes: All correlations above 0.11 are statistically significant (p <.001).
aValues for the work stressors are unit means calculated from the employee sample.
Variable . | M . | SD . | 1 . | 2 . | 3 . | 4 . | 5 . | 6 . | 7 . | 8 . | 9 . | 10 . |
---|---|---|---|---|---|---|---|---|---|---|---|---|
1. Residents' physical needs | 7.06 | 2.13 | — | |||||||||
2. Residents' psychosocial needs | 5.08 | 1.91 | .37 | — | ||||||||
3. Unit size | 33.90 | 8.52 | .05 | −.01 | — | |||||||
4. Staffing ratio per resident | 0.60 | 0.15 | .18 | −.01 | −.10 | — | ||||||
5. Time pressurea | 3.89 | 0.42 | .09 | .05 | .27 | .12 | — | |||||
6. Resident-related stressora | 2.99 | 0.27 | .15 | .01 | .21 | −.05 | .32 | — | ||||
7. Role ambiguitya | 1.92 | 0.26 | .02 | .03 | −.10 | −.14 | .32 | .24 | — | |||
8. Received physical help | 3.39 | 0.56 | −.23 | −.21 | −.07 | −.08 | −.12 | −.10 | −.01 | — | ||
9. Received psychosocial help | 3.20 | 0.73 | −.20 | −.23 | −.05 | −.07 | −.11 | −.10 | −.03 | .77 | — | |
10. Client-centered practices | 3.41 | 0.59 | −.14 | −.15 | −.09 | −.04 | −.07 | −.14 | −.07 | .60 | .61 | — |
Variable . | M . | SD . | 1 . | 2 . | 3 . | 4 . | 5 . | 6 . | 7 . | 8 . | 9 . | 10 . |
---|---|---|---|---|---|---|---|---|---|---|---|---|
1. Residents' physical needs | 7.06 | 2.13 | — | |||||||||
2. Residents' psychosocial needs | 5.08 | 1.91 | .37 | — | ||||||||
3. Unit size | 33.90 | 8.52 | .05 | −.01 | — | |||||||
4. Staffing ratio per resident | 0.60 | 0.15 | .18 | −.01 | −.10 | — | ||||||
5. Time pressurea | 3.89 | 0.42 | .09 | .05 | .27 | .12 | — | |||||
6. Resident-related stressora | 2.99 | 0.27 | .15 | .01 | .21 | −.05 | .32 | — | ||||
7. Role ambiguitya | 1.92 | 0.26 | .02 | .03 | −.10 | −.14 | .32 | .24 | — | |||
8. Received physical help | 3.39 | 0.56 | −.23 | −.21 | −.07 | −.08 | −.12 | −.10 | −.01 | — | ||
9. Received psychosocial help | 3.20 | 0.73 | −.20 | −.23 | −.05 | −.07 | −.11 | −.10 | −.03 | .77 | — | |
10. Client-centered practices | 3.41 | 0.59 | −.14 | −.15 | −.09 | −.04 | −.07 | −.14 | −.07 | .60 | .61 | — |
Notes: All correlations above 0.11 are statistically significant (p <.001).
aValues for the work stressors are unit means calculated from the employee sample.
. | . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
Variable . | Null Model . | A . | B . | |||
Fixed effect | ||||||
Intercept | .021 (.052) | .001 (.049) | .005 (.036) | |||
Residents' physical needs | −.072 (.035) | −.020 (.032) | ||||
Residents' psychosocial needs | −.015 (.033) | .006 (.031) | ||||
Unit size | −.188 (.049) | −.136 (.037) | ||||
Staffing ratio per resident | −.070 (.049) | −.054 (.036) | ||||
Time pressure | −.311 (.030) | |||||
Resident-related stressor | −.072 (.028) | |||||
Role ambiguity | −.127 (.027) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .193 (.039) | .167 (.035) | .062 (.018) | |||
Level 1 intercept variance | .804 (.035) | .791 (.035) | .706 (.031) | |||
Log likelihood | 3,103.396 | 3,080.369 | 2,894.842 |
. | . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
Variable . | Null Model . | A . | B . | |||
Fixed effect | ||||||
Intercept | .021 (.052) | .001 (.049) | .005 (.036) | |||
Residents' physical needs | −.072 (.035) | −.020 (.032) | ||||
Residents' psychosocial needs | −.015 (.033) | .006 (.031) | ||||
Unit size | −.188 (.049) | −.136 (.037) | ||||
Staffing ratio per resident | −.070 (.049) | −.054 (.036) | ||||
Time pressure | −.311 (.030) | |||||
Resident-related stressor | −.072 (.028) | |||||
Role ambiguity | −.127 (.027) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .193 (.039) | .167 (.035) | .062 (.018) | |||
Level 1 intercept variance | .804 (.035) | .791 (.035) | .706 (.031) | |||
Log likelihood | 3,103.396 | 3,080.369 | 2,894.842 |
Notes: Values are regression coefficients; standard errors are given parenthetically. Statistically significant (p <. 05) regression coefficient estimates are bolded.
. | . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
Variable . | Null Model . | A . | B . | |||
Fixed effect | ||||||
Intercept | .021 (.052) | .001 (.049) | .005 (.036) | |||
Residents' physical needs | −.072 (.035) | −.020 (.032) | ||||
Residents' psychosocial needs | −.015 (.033) | .006 (.031) | ||||
Unit size | −.188 (.049) | −.136 (.037) | ||||
Staffing ratio per resident | −.070 (.049) | −.054 (.036) | ||||
Time pressure | −.311 (.030) | |||||
Resident-related stressor | −.072 (.028) | |||||
Role ambiguity | −.127 (.027) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .193 (.039) | .167 (.035) | .062 (.018) | |||
Level 1 intercept variance | .804 (.035) | .791 (.035) | .706 (.031) | |||
Log likelihood | 3,103.396 | 3,080.369 | 2,894.842 |
. | . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
Variable . | Null Model . | A . | B . | |||
Fixed effect | ||||||
Intercept | .021 (.052) | .001 (.049) | .005 (.036) | |||
Residents' physical needs | −.072 (.035) | −.020 (.032) | ||||
Residents' psychosocial needs | −.015 (.033) | .006 (.031) | ||||
Unit size | −.188 (.049) | −.136 (.037) | ||||
Staffing ratio per resident | −.070 (.049) | −.054 (.036) | ||||
Time pressure | −.311 (.030) | |||||
Resident-related stressor | −.072 (.028) | |||||
Role ambiguity | −.127 (.027) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .193 (.039) | .167 (.035) | .062 (.018) | |||
Level 1 intercept variance | .804 (.035) | .791 (.035) | .706 (.031) | |||
Log likelihood | 3,103.396 | 3,080.369 | 2,894.842 |
Notes: Values are regression coefficients; standard errors are given parenthetically. Statistically significant (p <. 05) regression coefficient estimates are bolded.
Variable . | Null Model . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
A . | B . | |||||
Fixed effect | ||||||
Intercept | .034 (.055) | .012 (.054) | .015 (.042) | |||
Residents' physical needs | −.022 (.035) | .023 (.033) | ||||
Residents' psychosocial needs | −.098 (.032) | −.076 (.031) | ||||
Unit size | −.172 (.054) | −.119 (.042) | ||||
Staffing ratio per resident | −.071 (.053) | −.047 (.042) | ||||
Time pressure | −.298 (.031) | |||||
Resident-related stressor | −.046 (.029) | |||||
Role ambiguity | −.049 (.028) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .227 (.043) | .217 (.042) | .104 (.025) | |||
Level 1 intercept variance | .790 (.035) | .773 (.034) | .722 (.032) | |||
Log likelihood | 3,096.692 | 3,044.476 | 2,948.058 |
Variable . | Null Model . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
A . | B . | |||||
Fixed effect | ||||||
Intercept | .034 (.055) | .012 (.054) | .015 (.042) | |||
Residents' physical needs | −.022 (.035) | .023 (.033) | ||||
Residents' psychosocial needs | −.098 (.032) | −.076 (.031) | ||||
Unit size | −.172 (.054) | −.119 (.042) | ||||
Staffing ratio per resident | −.071 (.053) | −.047 (.042) | ||||
Time pressure | −.298 (.031) | |||||
Resident-related stressor | −.046 (.029) | |||||
Role ambiguity | −.049 (.028) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .227 (.043) | .217 (.042) | .104 (.025) | |||
Level 1 intercept variance | .790 (.035) | .773 (.034) | .722 (.032) | |||
Log likelihood | 3,096.692 | 3,044.476 | 2,948.058 |
Notes: Values are regression coefficients; standard errors are given parenthetically. Statistically significant (p <.05) regression coefficient estimates are bolded.
Variable . | Null Model . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
A . | B . | |||||
Fixed effect | ||||||
Intercept | .034 (.055) | .012 (.054) | .015 (.042) | |||
Residents' physical needs | −.022 (.035) | .023 (.033) | ||||
Residents' psychosocial needs | −.098 (.032) | −.076 (.031) | ||||
Unit size | −.172 (.054) | −.119 (.042) | ||||
Staffing ratio per resident | −.071 (.053) | −.047 (.042) | ||||
Time pressure | −.298 (.031) | |||||
Resident-related stressor | −.046 (.029) | |||||
Role ambiguity | −.049 (.028) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .227 (.043) | .217 (.042) | .104 (.025) | |||
Level 1 intercept variance | .790 (.035) | .773 (.034) | .722 (.032) | |||
Log likelihood | 3,096.692 | 3,044.476 | 2,948.058 |
Variable . | Null Model . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
A . | B . | |||||
Fixed effect | ||||||
Intercept | .034 (.055) | .012 (.054) | .015 (.042) | |||
Residents' physical needs | −.022 (.035) | .023 (.033) | ||||
Residents' psychosocial needs | −.098 (.032) | −.076 (.031) | ||||
Unit size | −.172 (.054) | −.119 (.042) | ||||
Staffing ratio per resident | −.071 (.053) | −.047 (.042) | ||||
Time pressure | −.298 (.031) | |||||
Resident-related stressor | −.046 (.029) | |||||
Role ambiguity | −.049 (.028) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .227 (.043) | .217 (.042) | .104 (.025) | |||
Level 1 intercept variance | .790 (.035) | .773 (.034) | .722 (.032) | |||
Log likelihood | 3,096.692 | 3,044.476 | 2,948.058 |
Notes: Values are regression coefficients; standard errors are given parenthetically. Statistically significant (p <.05) regression coefficient estimates are bolded.
Variable . | Null Model . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
A . | B . | |||||
Fixed effect | ||||||
Intercept | .009 (.045) | .003 (.044) | .003 (.040) | |||
Residents' physical needs | −.044 (.036) | −.023 (.034) | ||||
Residents' psychosocial needs | .084 (.034) | .091 (.033) | ||||
Unit size | −.094 (.044) | −.078 (.041) | ||||
Staffing ratio per resident | .031 (.044) | −.027 (.040) | ||||
Time pressure | −.088 (.032) | |||||
Resident-related stressor | −.094 (.030) | |||||
Role ambiguity | −.178 (.030) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .121 (.029) | .106 (.027) | .079 (.022) | |||
Level 1 intercept variance | .873 (.038) | .869 (.038) | .816 (.036) | |||
Log likelihood | 3,160.625 | 3,148.970 | 3,065.247 |
Variable . | Null Model . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
A . | B . | |||||
Fixed effect | ||||||
Intercept | .009 (.045) | .003 (.044) | .003 (.040) | |||
Residents' physical needs | −.044 (.036) | −.023 (.034) | ||||
Residents' psychosocial needs | .084 (.034) | .091 (.033) | ||||
Unit size | −.094 (.044) | −.078 (.041) | ||||
Staffing ratio per resident | .031 (.044) | −.027 (.040) | ||||
Time pressure | −.088 (.032) | |||||
Resident-related stressor | −.094 (.030) | |||||
Role ambiguity | −.178 (.030) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .121 (.029) | .106 (.027) | .079 (.022) | |||
Level 1 intercept variance | .873 (.038) | .869 (.038) | .816 (.036) | |||
Log likelihood | 3,160.625 | 3,148.970 | 3,065.247 |
Notes: Values are regression coefficients; standard errors are given parenthetically. Statistically significant (p <.05) regression coefficient estimates are bolded.
Variable . | Null Model . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
A . | B . | |||||
Fixed effect | ||||||
Intercept | .009 (.045) | .003 (.044) | .003 (.040) | |||
Residents' physical needs | −.044 (.036) | −.023 (.034) | ||||
Residents' psychosocial needs | .084 (.034) | .091 (.033) | ||||
Unit size | −.094 (.044) | −.078 (.041) | ||||
Staffing ratio per resident | .031 (.044) | −.027 (.040) | ||||
Time pressure | −.088 (.032) | |||||
Resident-related stressor | −.094 (.030) | |||||
Role ambiguity | −.178 (.030) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .121 (.029) | .106 (.027) | .079 (.022) | |||
Level 1 intercept variance | .873 (.038) | .869 (.038) | .816 (.036) | |||
Log likelihood | 3,160.625 | 3,148.970 | 3,065.247 |
Variable . | Null Model . | Random Intercept Model . | . | |||
---|---|---|---|---|---|---|
A . | B . | |||||
Fixed effect | ||||||
Intercept | .009 (.045) | .003 (.044) | .003 (.040) | |||
Residents' physical needs | −.044 (.036) | −.023 (.034) | ||||
Residents' psychosocial needs | .084 (.034) | .091 (.033) | ||||
Unit size | −.094 (.044) | −.078 (.041) | ||||
Staffing ratio per resident | .031 (.044) | −.027 (.040) | ||||
Time pressure | −.088 (.032) | |||||
Resident-related stressor | −.094 (.030) | |||||
Role ambiguity | −.178 (.030) | |||||
Random effect variances | ||||||
Level 2 intercept variance | .121 (.029) | .106 (.027) | .079 (.022) | |||
Level 1 intercept variance | .873 (.038) | .869 (.038) | .816 (.036) | |||
Log likelihood | 3,160.625 | 3,148.970 | 3,065.247 |
Notes: Values are regression coefficients; standard errors are given parenthetically. Statistically significant (p <.05) regression coefficient estimates are bolded.
. | Physical Help Received . | . | Psychosocial Help Received . | . | Client-Centered Practices . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Variable . | Null Model . | Full Model . | Null Model . | Full Model . | Null Model . | Full Model . | ||||||
Fixed effect | ||||||||||||
Intercept | −.028 (.042) | −.009 (.037) | −.026 (.043) | −.002 (.039) | −.030 (.045) | .019 (.043) | ||||||
Residents' physical needs | −.158 (.036) | −.114 (.035) | −.093 (.036) | |||||||||
Residents' psychosocial needs | −.151 (.035) | −.192 (.035) | −.120 (.035) | |||||||||
Unit size | −.054 (.040) | −.026 (.042) | −.075 (.046) | |||||||||
Staffing ratio per resident | −.045 (.038) | −.050 (.040) | −.032 (.044) | |||||||||
Time pressure | −.092 (.039) | −.090 (.041) | −.056 (.045) | |||||||||
Random effect variance | ||||||||||||
Level 2 intercept variance | .070 (.026) | .036 (.019) | .078 (.026) | .054 (.022) | .097 (.029) | .084 (.027) | ||||||
Level 1 intercept variance | .934 (.046) | .882 (.043) | .896 (.044) | .839 (.041) | .894 (.044) | .865 (.042) | ||||||
Log likelihood | 2,660.053 | 2,586.201 | 2,627.724 | 2,555.038 | 2,635.679 | 2,599.789 |
. | Physical Help Received . | . | Psychosocial Help Received . | . | Client-Centered Practices . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Variable . | Null Model . | Full Model . | Null Model . | Full Model . | Null Model . | Full Model . | ||||||
Fixed effect | ||||||||||||
Intercept | −.028 (.042) | −.009 (.037) | −.026 (.043) | −.002 (.039) | −.030 (.045) | .019 (.043) | ||||||
Residents' physical needs | −.158 (.036) | −.114 (.035) | −.093 (.036) | |||||||||
Residents' psychosocial needs | −.151 (.035) | −.192 (.035) | −.120 (.035) | |||||||||
Unit size | −.054 (.040) | −.026 (.042) | −.075 (.046) | |||||||||
Staffing ratio per resident | −.045 (.038) | −.050 (.040) | −.032 (.044) | |||||||||
Time pressure | −.092 (.039) | −.090 (.041) | −.056 (.045) | |||||||||
Random effect variance | ||||||||||||
Level 2 intercept variance | .070 (.026) | .036 (.019) | .078 (.026) | .054 (.022) | .097 (.029) | .084 (.027) | ||||||
Level 1 intercept variance | .934 (.046) | .882 (.043) | .896 (.044) | .839 (.041) | .894 (.044) | .865 (.042) | ||||||
Log likelihood | 2,660.053 | 2,586.201 | 2,627.724 | 2,555.038 | 2,635.679 | 2,599.789 |
Notes: Random intercept model was used. Values are regression coefficients; standard errors are given parenthetically. Statistically significant (p <.05) regression coefficient estimates are bolded.
. | Physical Help Received . | . | Psychosocial Help Received . | . | Client-Centered Practices . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Variable . | Null Model . | Full Model . | Null Model . | Full Model . | Null Model . | Full Model . | ||||||
Fixed effect | ||||||||||||
Intercept | −.028 (.042) | −.009 (.037) | −.026 (.043) | −.002 (.039) | −.030 (.045) | .019 (.043) | ||||||
Residents' physical needs | −.158 (.036) | −.114 (.035) | −.093 (.036) | |||||||||
Residents' psychosocial needs | −.151 (.035) | −.192 (.035) | −.120 (.035) | |||||||||
Unit size | −.054 (.040) | −.026 (.042) | −.075 (.046) | |||||||||
Staffing ratio per resident | −.045 (.038) | −.050 (.040) | −.032 (.044) | |||||||||
Time pressure | −.092 (.039) | −.090 (.041) | −.056 (.045) | |||||||||
Random effect variance | ||||||||||||
Level 2 intercept variance | .070 (.026) | .036 (.019) | .078 (.026) | .054 (.022) | .097 (.029) | .084 (.027) | ||||||
Level 1 intercept variance | .934 (.046) | .882 (.043) | .896 (.044) | .839 (.041) | .894 (.044) | .865 (.042) | ||||||
Log likelihood | 2,660.053 | 2,586.201 | 2,627.724 | 2,555.038 | 2,635.679 | 2,599.789 |
. | Physical Help Received . | . | Psychosocial Help Received . | . | Client-Centered Practices . | . | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Variable . | Null Model . | Full Model . | Null Model . | Full Model . | Null Model . | Full Model . | ||||||
Fixed effect | ||||||||||||
Intercept | −.028 (.042) | −.009 (.037) | −.026 (.043) | −.002 (.039) | −.030 (.045) | .019 (.043) | ||||||
Residents' physical needs | −.158 (.036) | −.114 (.035) | −.093 (.036) | |||||||||
Residents' psychosocial needs | −.151 (.035) | −.192 (.035) | −.120 (.035) | |||||||||
Unit size | −.054 (.040) | −.026 (.042) | −.075 (.046) | |||||||||
Staffing ratio per resident | −.045 (.038) | −.050 (.040) | −.032 (.044) | |||||||||
Time pressure | −.092 (.039) | −.090 (.041) | −.056 (.045) | |||||||||
Random effect variance | ||||||||||||
Level 2 intercept variance | .070 (.026) | .036 (.019) | .078 (.026) | .054 (.022) | .097 (.029) | .084 (.027) | ||||||
Level 1 intercept variance | .934 (.046) | .882 (.043) | .896 (.044) | .839 (.041) | .894 (.044) | .865 (.042) | ||||||
Log likelihood | 2,660.053 | 2,586.201 | 2,627.724 | 2,555.038 | 2,635.679 | 2,599.789 |
Notes: Random intercept model was used. Values are regression coefficients; standard errors are given parenthetically. Statistically significant (p <.05) regression coefficient estimates are bolded.
References
Anderson, C. L., & Hughes, E., (
Arnold, J., Cooper, C. L., & Robertson, I. T., (
Bliese, P. D., & Jex, S. M., (
Bliese, P. D., & Jex, S. M., (
Bravo, G., Wals, P. D., Dubois, M.-F., & Charpentier, M., (
Chappell, N. L., & Novak, M., (
Chappell, N. L., & Reid, R. C., (
Cohen-Mansfield, J., (
Davis, M. A., (
Donabedian, A., (
Donabedian, A., (
Elovainio, M., Kivimäki, M., & Helkama, K., (
Elovainio, M., Kivimäki, M., Steen, N., & Kalliomäki-Levanto, T., (
Elovainio, M., & Sinervo, T., (
Elovainio, M., & Sinervo, T., (
Frytak, J. R., (
Garman, A. N., Corrigan, P. W., & Morris, S., (
Goldstein, H., (
Gray-Toft, P., & erson, J. G., (
Hannan, S., Norman, I. J., & Redfern, S. J., (
Hermanson, T., Aro, S., & Bennett, C. S., (
Jamal, M., (
Jamal, M., (
Jones, J. W., Barge, B. N., Steffy, B. D., Fay, L. M., Kunz, L. K., & Wuebker, L. J., (
Jöreskog, K., Sörbom, D., du Toit, S., & du Toit, M., (
Kane, R. A., (
Kane, R. A., (
Kane, R. L., & Kane, R. A., (Eds.) (
Kivimäki, M., & Lindström, K., (
Klein, K. J., Tosi, H., & Cannella, A. A. J., (
Kozlowski, S. W. J., & Klein, K. J., (
Kruzich, J. M., Clinton, J. F., & Kelber, S. T., (
Leiter, M. P., Harvie, P., & Frizzell, C., (
Leveck, M. L., & Jones, C. B., (
Motowidlo, S. J., Packard, J. S., & Manning, M. R., (
Muse, L. A., Harris, S. G., & Feild, H. S., (
Noro, A., & Aro, S., (
Packard, J. S., & Motowidlo, S. J., (
Pearlin, L. I., Mullan, J. T., Semple, S. J., & Skaff, M. M., (
Perälä, M.-L., & Räikkönen, O., (
Rice, N., & Leyland, A., (
Rizzo, J. R., House, R. J., & Lirtzman, S. I., (
Roper, N., Logan, W., & Tierney, A., (
Sawyer, J. E., (
Schaefer, J. A., & Moos, R. H., (
Schaefer, J. A., & Moos, R. H., (
Sheridan, J. E., White, J., & Fairchild, T. J., (
Sinervo, T., (
Spector, P. E., Dwyer, D. J., & Jex, S. M., (
Sullivan, S. E., & Bhagat, R. S., (
Vaarama, M., Kainulainen, S., Perälä, M.-L., & Sinervo, T., (
Westman, M., & Eden, D., (