- Split View
-
Views
-
Cite
Cite
Nicholas G. Castle, John Engberg, Further Examination of the Influence of Caregiver Staffing Levels on Nursing Home Quality, The Gerontologist, Volume 48, Issue 4, August 2008, Pages 464–476, https://doi.org/10.1093/geront/48.4.464
- Share Icon Share
Abstract
Purpose: Weak empirical evidence exists showing that nursing home staffing levels influence quality of care. We propose that weak findings have resulted in many prior analyses because research models have underspecified the labor composition needed to influence care processes that, in turn, influence quality of care. In this analysis, we specified the nursing home labor composition by using staff stability, use of agency staff, and professional staff mix, in addition to staffing levels. Design and Methods: Data used in this investigation came from surveys of nursing home administrators (N = 6,005); Nursing Home Compare; the Online Survey, Certification and Reporting data; and the Area Resource File. Staffing characteristics, quality indicators, facility, and market information from these data sources were all measured in 2004. Results: The regression analyses showed that staffing levels alone were weakly associated with the six quality measures examined. However, when the regression models were more fully specified (by including agency staff, stability, and professional staff mix), staffing levels were generally associated with the quality measures (i.e., 15 of the 18 staffing coefficients were significant). Implications: Simply adding more staff may be a necessary but not sufficient means of improving nursing home quality. Some accounting for agency staff, stability, and professional staff mix is also needed.
Despite the generally low and often variable levels of staffing found in nursing homes, only weak empirical evidence exists that staffing levels influence quality of care (Arling, Kane, Mueller, Bershadsky, & Degenholtz, 2007; Bates-Jensen, Schnelle, Alessi, Al-Samarrai, & Levy-Storms, 2004; Bostick, 2004; Dellefield, 2000; Hunt & Hagen, 1998; Kane, 2004; Sullivan-Marx, Strumpf, Evans, Baumgarten, & Maislin, 1999). We propose that staffing levels are a proxy for care processes, and weak findings have resulted in many prior analyses because research models examining staffing levels have underspecified the labor composition needed to influence care processes that, in turn, influence resident outcomes (i.e., quality of care). Thus, in this research, we further examine this staffing–quality relationship. As described further in “Conceptual Model,” we more adequately specify the labor composition (and therefore care processes) in this investigation examining nursing home quality by using measures of agency staff, professional staff mix, and staff stability in addition to staffing levels. By agency staff we mean temporary workers hired by facilities to fill staff positions, by professional staff mix we mean the ratio of registered nurses (RNs) to other caregivers (i.e., licensed practical nurses [LPNs] and nurse aides [NAs]), and by staff stability we mean the number of staff that have worked at the facility for 5 years or more.
The aim of this research was not to disprove the notion that more staffing is better, but rather to show the limitations of using this simple headcount (i.e., using staffing levels) when examining quality. We aimed to add to the staffing–quality equation the composition of staff. Few previous analyses have examined nursing home staffing in this way.
Nursing home residents have increasingly more complex care needs, and facilities are using increasingly complex technology. Care is also dependent not only on how much is done (i.e., quantity of care), but upon consistency of care, coordination, and care practices. Thus, simply adding more staff may be a necessary but not sufficient means of improving quality. As Bowers, Esmond, and Jacobson (2000, p. 63) noted, “Staffing is more complex than common measures might suggest … assessment of staffing levels requires more than counting the number of bodies reporting to work each shift.” Kane (2004, p. 251) stated that staffing is “more complex than simply a body count.” More recently, Arling and associates (2007) also advocated that staffing research move beyond simply measuring the amount of care available. Examining care processes more adequately is important and may lead to broader policy debate over staffing issues rather than staffing levels in nursing homes.
Prior Literature
As part of our literature review, we sought to summarize the findings of studies examining staffing levels and quality. The studies identified were diverse in many respects (i.e., sample size, quality indicators examined, and staffing measures used). However, from the 59 studies examined (published from 1991–2006), we found 120 of the 302 quality indicators they used (or 40%) to be significantly associated with staffing levels (review tables available upon request).
Recent studies identified with predominantly nonsignificant findings include those by Moseley and Jones (2003). Using 28 homes in Nevada, these authors identified 8 of 10 quality indicators as not associated with staffing levels. In a larger study using data from 14,642 nursing homes, Wan (2003) found four of six quality indicators to be not associated with staffing levels. Bostick (2004), using 413 nursing homes, found 12 of 17 quality indicators to be not associated with staffing levels. We do note, however, that other studies have identified predominantly significant findings. For example, Dorr, Horn, and Smout (2005) found all three quality indicators examined from 82 nursing homes to be associated with staffing levels.
Very few studies have examined additional components of the labor composition of nursing homes in addition to staffing levels. As Kane (2004, p. 251) described, researchers have an “incomplete story” in this area. Notable exceptions include studies by Harrington and Swan (2003), Barry, Brannon, and Mor (2005), and Castle and Engberg (2005, 2007a). After accounting for turnover rates, Harrington and Swan recently identified RN staffing levels to be associated with residents' activity of daily living dependency. Barry and colleagues examined pressure ulcer incidence and social engagement, using NA turnover and retention as controls. They used RN-to-NA ratio as a staffing variable, and it was nonsignificant in both analyses. Castle and Engberg (2005) found consistently significant findings between NA plus LPN turnover and quality, and between RN turnover and quality, but staffing levels included in these analyses were not significant (in this case, they used measures of full-time equivalent [FTE] RNs, FTE LPNs, and FTE NAs).
Castle and Engberg (2007a) recently examined the influence staffing levels, turnover, worker stability, and agency staff had on nursing home quality. They used one index of quality, and data were from approximately 1,000 facilities. The quality index was influenced, to some degree, by all of these staffing characteristics. However, the estimated interaction effects also indicated that achieving the highest quality was dependent on having more than one favorable staffing characteristic. To our knowledge, this research is the first to empirically show that multiple staffing characteristics rather than staffing levels alone influence nursing home quality.
The research presented by Castle and Engberg (2007a) was limited to a relatively small sample size and one quality index using data from 2002. Significantly, they presented little conceptual basis for the modeling used for the findings. In this research, we use a substantially larger sample size, more recent data, and individual quality measures from Nursing Home Compare. We also present a comprehensive conceptual model that represents a first attempt at explaining the association between different staffing characteristics and quality of care. The research we present also has a different focus from that of Castle and Engberg (2007a). We use multiple staffing characteristics to help explain why many empirical findings from prior studies of staffing show only weak relationships with quality indicators.
Thus, in summary, our review of the literature in this area showed only weak evidence for the notion that staffing levels influence quality. This provided the motivation for this study. Recent literature has suggested that limitations of prior studies include suspect staffing variables (e.g., Arling et al., 2007; Castle & Engberg, 2007a; Kane, 2004) and underspecified analytic models (Castle & Engberg, 2007a). We address both of these limitations in our analyses. Following the analytic specification guidelines from Castle and Engberg (2007a), we show the limitations of simply using staffing levels when examining quality.
Conceptual Model
As we stated above, resident outcomes are dependent on quantity of care, coordination, consistency of care, and care practices. With respect to quantity of care, we define this as the amount of hands-on care given to each resident.
We define coordination as “the extent to which the various interdependent parts of an organization [or staff] function each according to the needs and requirements of the other parts and of the total system” (Georgopoulos & Mann, 1962, p. 273). The organizational behavior literature has examined mechanisms to achieve coordination (ranging from direct supervision and rules, to organizational culture). Most of this literature has shown better coordination to be associated with more favorable outcomes (Kreitner & Kinicki, 1991).
We define consistency as “an understandable sequence of steps that can be followed in doing the work” (Price & Mueller, 1986, p. 241). Research has addressed consistency of recordkeeping, and consistency in the different tasks that are done as part of the job (Price & Mueller, 1986). Consistency of nurse and health care technician jobs, for example, is beneficial (Price & Mueller, 1986). Conversely, inconsistency predisposes workers to learn new processes each time they perform a task. Pavitt (1985) detailed that when this occurs, even simple processes may have to be substantially modified, take more time, and can lead to errors.
We define care practices as “the degree to which the norms of an organization are performed” when giving care (Price & Mueller, 1986, p. 137). In many organizations these are reduced to written form, but this is not necessary for all care practices to become the norm. Other related concepts in the organizational behavior literature include specification of procedures, degree of structure, and job codification (Price & Mueller, 1986). Research by Brannon and colleagues examining NAs' responses to job redesign reinforces the salience of consistent care practices for enhancing caregiving performance (e.g., Brannon, Streit, & Smyer, 1992; Smyer, Brannon, & Cohn, 1992).
For our analyses, we did not have direct measures of quantity of care, consistency of care, coordination of care, or care practices. However, as described below, we propose that the staffing characteristics of staffing levels, stability, professional staff mix, and agency use influence quantity of care, consistency of care, coordination of care, and care practices.
Our conceptual model in Figure 1 shows the proposed relationships, with the influence of staffing levels, staff stability, professional staff mix, and agency staff denoted by +, 0, and − (with + indicating a positive hypothesized influence on care processes, 0 no influence, and – a negative hypothesized influence on care processes). We also propose that some relationships may be stronger or weaker, and we show this by the use of more or less +'s or –'s. Differences in care processes lead to differences in resident outcomes. We use resident outcomes as representative of quality of care, and this could include outcomes as diverse as resident satisfaction, physical restraint use, and pressure ulcers.
Clearly, staffing levels should be associated with quantity of care, because more staff are likely providing more care for residents. Staffing levels may also influence other care process characteristics (i.e., care practices and consistency). With more available time, more comprehensive care practices can be used. With higher staffing levels, care may also be more consistent. That is, with higher staffing levels, resident–caregiver assignments may be more stable. Higher staffing levels do not necessarily ensure that coordination has occurred, such that the most appropriate caregiver is attending to the residents' needs.
Staff stability imparts a degree of experience and exposure to organizational norms and requirements. Thus, highly stable employees are likely to deliver consistent care and to have a greater appreciation for the care practices that are used (e.g., when resident care should be delivered). However, depending upon which staff are stable, this may have a negligible impact on quantity of care and coordination of care.
Professional nurses (e.g., RNs) are primarily responsible for providing the most complex resident care and for directing the activities of LPNs and NAs. Thus, a higher professional-to-nonprofessional nursing ratio should impart more coordination over care processes. As others have noted, the staff mix may be “as important as the number of nurses to staff” (Bates-Jensen, 2005, p. 13). RNs also teach, direct, and implement resident care practices. Thus, a higher professional-to-nonprofessional nursing ratio should have a positive influence on care practices. In addition, more RNs likely improve the consistency of care by facilitating more stable resident assignments and oversight of care. However, the impact of professional staff mix on the quantity of care delivered may be minimal.
Agency staff serve to increase FTE levels and should influence the quantity of care given in the facility. Nevertheless, agency staff are invariably unfamiliar with at least some facility practices (thereby weakening care practices) and are invariably unfamiliar with residents (thereby weakening consistency of care). Given that agency staff are themselves subsequently replaced by other workers (i.e., other agency staff or permanent staff), this also serves to diminish the consistency of care delivered. As Bowers and associates (2000) described, “Staff who do not have relationships with residents can be a hindrance as well as a help in the accomplishment of care tasks” (p. 63). When staff are new and unfamiliar with care practices, the coordination challenge also increases.
The conceptual model shows the proposed influence of staffing levels, staff stability, professional staff mix, and agency staff on care processes (i.e., +, 0, and −). As we described above, this shows that we believe the staffing characteristics may positively influence some areas and negatively influence others. For example, use of agency staff can increase caregiver time with residents but decrease consistency of care. Thus, the ultimate influence each staffing characteristic has on care processes (and resident outcomes) will depend on the total relative mix of the staffing characteristics (i.e., staffing levels, stability, staffing ratios, and agency staff).
In addition, in some cases the effect of each of the staffing characteristics may depend on the level of that characteristic (i.e., a nonlinear relationship) and of the other characteristics (i.e., an interaction). Thus, the conceptual model includes staffing nonlinearities and staffing interactions. However, our knowledge base is incomplete on the influence of these terms, and there is a potentially large number of possible nonlinear and interaction relationships. Thus, we do not present proposed relationships with quality for these factors.
This conceptual model (see Figure 1) also includes facility characteristics (e.g., bed size, ownership, chain membership, occupancy, and Medicaid occupancy) and market characteristics (i.e., competition and unemployment rates), as these factors may also influence the context of care delivery and quality of care. Researchers frequently use this general approach of including facility and market characteristics in empirical studies examining nursing home quality, including those with a focus on staffing (e.g., Harrington & Swan, 2003; Schnelle et al., 2004).
Small facility size (i.e., fewer beds) generally has a positive association with quality (e.g., Castle, 2002; Harrington & Swan, 2003). This may be because small facilities are more able to cater to individual resident needs due to the familiarity and bonds that can form between residents and staff (Castle & Engberg, 2007b). For-profit ownership generally has a negative association with quality (e.g., see review by Hillmer, Wodchis, Gill, Anderson, & Rochon, 2005). For-profit nursing homes have greater incentives than not-for-profit nursing homes to reduce resource use (as part of the profit motive), which may lead to low quality of care (Hillmer et al., 2005). Several studies have shown chain ownership to be associated with low quality of care (e.g., Castle & Engberg, 2005; Hughes, Lapane, & Mor, 2000). This may be because chain ownership leads to standardization that does not account for local operating realities (Banaszak-Holl, Berta, Bowman, Baum, & Mitchell, 2001). High occupancy rates may result in low-quality care (e.g., Hughes et al., 2000). When facilities have high occupancy, resources may be stretched and oversight may be reduced, which may lead to low quality of care (Castle & Engberg, 2005; Hughes et al., 2000). There is also some indication that quality is low in facilities with a high Medicaid census (e.g., Bourbonniere et al., 2006; Decker, 2006; Hughes et al., 2000; O'Neill, Harrington, Kitchener, & Saliba, 2003). Medicaid provides lower reimbursement to nursing homes than other sources of payment, so it may be difficult for facilities serving high numbers of these beneficiaries to provide adequate services (O'Neill et al., 2003).
With respect to market characteristics, competition (often measured using a Herfindahl Index) is often associated with quality (e.g., Bourbonniere et al., 2006; Castle & Engberg, 2005). That is, in more competitive environments, nursing homes share a limited pool of resources, and survival/profitability depends on high-quality care (Mukamel, Spector, & Bajorska, 2005). Researchers also often include unemployment rates as a control for market conditions when examining staffing (e.g., Castle & Engberg, 2005; Harrington & Swan, 2003). This is because in areas with high unemployment, fewer positions will be vacant, and voluntary caregiver turnover is likely to be lower (Castle & Engberg, 2005). Also, more caregivers will be available in the employment pool, which could raise staffing levels.
Methods
Source of Data
Data used in this investigation came from five sources: (a) from a survey of nursing home administrators conducted March through June 2005, (b) from a survey of nursing home administrators conducted January through March 2006, (c) from the 2004 Nursing Home Compare, (d) from the 2005 OSCAR (i.e., Online Survey, Certification and Reporting), and (e) from the 2006 Area Resource File (ARF). The information regarding staffing factors came from the two administrator surveys, quality measures came from Nursing Home Compare, characteristics of the nursing home came from the OSCAR, and characteristics of the market came from the ARF. The staffing characteristic, quality indicator, facility, and market information were all measured in 2004.
Nursing Home Administrator Surveys
We used information from two surveys of nursing home administrators. These two surveys each had identical questions asking about staffing. We used the data from both surveys because they provided a more representative picture of nursing home staffing than those from either individual survey.
We used this primary data collection because some staffing information (e.g., stability) either was not found in commonly used secondary sources of nursing home information, or was not considered to be reliable if reported in secondary sources (e.g., agency use and staffing levels) such as the OSCAR (Straker, 1999). Both surveys collected information for staffing characteristics of NAs, LPNs, and RNs for 2004, including data on the use of agency staff, stability, and staffing levels.
We sent the first survey to 4,000 nursing home administrators, and 2,946 were returned (response rate = 74%). We sent the second survey to 6,000 nursing homes, with 3,939 returned (response rate = 66%). We created the mailing samples by using information from the OSCAR data (described further below). For each survey, we retained the OSCAR facility identification number so that we could subsequently examine facility characteristics of the sample. We excluded small nursing homes (<30 beds) from both survey samples because these smaller facilities have staffing characteristics with low signal-to-noise ratios.
We developed the survey items based on our experience with three prior large-scale nursing home questionnaires, including interviews with nursing home administrators. However, having well-formulated questions does not mean administrators will answer appropriately or accurately. Therefore, we conducted face-to-face interviews with 89 directors of nursing and 87 nursing home administrators to determine how they had completed our survey items. For staffing levels, we were able to check the construct validity of our measures by cross-checking our results with payroll records in 152 facilities. The resulting kappa statistic was extremely high at.95, indicating that our staffing information was extremely reliable. All of the top managers interviewed (N = 176) indicated that agency staff use information was routinely tracked. When we cross-checked our results with payroll records for agency staff in 152 facilities, the resulting kappa statistic was extremely high at.97 (indicating that this information was also extremely reliable). In addition, all of the top managers interviewed indicated that our questions made sense operationally and that they would be able to provide highly accurate answers to all questions. Still, we acknowledge that even though we conducted these interviews after the surveys were completed, we used a convenience sample of facilities in close proximity to us. The possibility exists that directors of nursing and nursing home administrators who were willing to be interviewed were also most likely to provide the most accurate answers on our surveys.
OSCAR
The OSCAR data came from the Medicare and/or Medicaid certification process conducted by state licensure and certification agencies on a yearly basis. These data included almost all nursing homes in the United States. In 2005, approximately 17,000 nursing homes were included in the data, including all of the facilities used in this analysis. Many characteristics reported in the data, especially organizational characteristics, are considered reliable (Harrington et al., 2000). In this study, we used information from the OSCAR on the following organizational characteristics: bed size, ownership, chain membership, occupancy, and Medicaid occupancy.
Nursing Home Compare
As quality indicators, we used the quality measures reported on the Nursing Home Compare Web site (www.Medicare.gov/NHCompare/home.asp). Nursing Home Compare is a Web-based report card providing information for all Medicare and/or Medicaid certified nursing homes. The quality measures reported are subject to extensive testing, are derived from the Minimum Data Set, are readily available online, represent measures relevant to both consumers and providers (General Accounting Office, 2002), and are becoming commonly used in research (e.g., Castle & Engberg, 2007a). For these reasons, we used the quality measures in this analysis. The development of the quality measures, and Nursing Home Compare in general, is described in numerous publications (Abt Associates, 2004; General Accounting Office, 2002).
The quality measures included 11 measures for long-stay residents (defined as those with a quarterly Minimum Data Set assessment) and 3 measures for short-stay residents (defined as those with only a 14-day Minimum Data Set assessment). The 11 quality measures for long-stay residents were the percentage of (a) residents with increased need for help with daily activities, (b) high-risk residents with pressure ulcers, (c) low-risk residents with pressure ulcers, (d) residents with physical restraint use, (e) low-risk residents with loss of bladder or bowel control, (f) residents who had a catheter inserted and left in bladder, (g) residents who spent most of their time in bed or in a chair, (h) residents with urinary tract infection, (i) residents with moderate to severe pain, (j) residents more depressed or anxious, and (k) residents with decreasing ability to move in/around room. The three quality measures for short-stay residents were the percentage of residents (a) with delirium, (b) who had moderate to severe pain, and (c) with pressure sores.
Given that little prior work exists in this area (and given the large number of quality measures), we did not examine every quality measure. Following our conceptual model, it would seem reasonable to believe that quality measures that are sensitive to care processes are restraints, catheter use, inadequate pain management, and pressure ulcers. That is, these quality measures can develop quickly when care processes are inadequate, and they were thus the quality measures we examined in this investigation. The other quality measures may also be influenced by care processes but may develop over a longer period of time (e.g., need for help with daily activities). Our data structure (i.e., cross-sectional measures of staffing) was likely not sensitive enough to capture these changes. Studies in the hospital literature have used a similar approach to choosing quality indicators by examining Outcomes Sensitive to Nursing (Needleman, Beurhaus, Mattke, Stewart, & Zelevinsky, 2001).
ARF
ARF data were a compilation of data from various sources. These data sources included the American Hospital Association annual hospital survey, the U.S. Census of Population and Housing, the Centers for Disease Control and Prevention, and the National Center for Health Statistics. These data were aggregated at the county level. Extensive details regarding these data are on the World Wide Web (http://wonder.cdc.gov/wonder/sci_data/census/arf). In this research, the ARF was a minor data source that we used to examine the level of nursing home competition and unemployment rates in each county.
Model Specification and Operationalization
We used negative binomial regression in multivariate analyses to examine the association of staffing levels with the quality measures. The quality measures were counts of specific negative events per nursing home, each divided by the number of residents at risk for that negative event. For many facilities, these counts were low or zero. Negative binomial regression is based on a generalization of the Poisson distribution that can account for the skewed nature of data. This allows for more unmeasured heterogeneity among the observations in the sample, which can be manifested when several observations have low or zero events. Given that larger nursing homes have more residents for whom the negative outcomes could occur, the negative binomial regression used the number of residents at risk for each measure as the exposure level for that measure. In order to account for possible correlation of outcomes within markets, which can bias the standard errors of the estimates, we used the Huber-White sandwich estimator (i.e., robust standard errors) clustered by county for all of the multivariate analyses. In these analyses, higher values reflect lower (i.e., poor) quality.
We defined staffing levels as the number of FTE staff per 100 residents for NAs, LPNs, and RNs. This included full-time and part-time workers. We defined stability as the percentage of NAs, LPNs, and RNs who had worked for the facility 5 years or more. We defined the professional staff mix as the number of RNs divided by the number of LPNs plus NAs. We defined agency staff as the number of NA, LPN, and RN agency staff used in the past year expressed as a percentage of regular staff.
We know from other nursing home studies that organizational factors (in addition to staffing characteristics) have a strong impact on quality indicators. Therefore, we included the organizational factors size, ownership, chain membership, overall occupancy, and Medicaid occupancy as controls. Likewise, market factors can influence quality indicators. We included market competition from other nursing homes and unemployment rates. Table 1 gives further details of the variables used in the analyses.
Analyses
We examined the level of collinearity among the independent variables and multicollinearity by using the variance inflation factor test. We were especially concerned with the potential correlation between the staffing level variables, but the correlation between these variables was low.
We included eight characteristics of the facilities and the markets in which they operated. The first specification included only these facility and market characteristics. The second specification included the staffing levels of caregivers (plus facility and market controls). Again, following the work of Castle and Engberg (2007a), we also included terms to capture nonlinear and interaction effects (described further below). Our main interest was the effect of the additional three staffing factors (agency staff, professional staff mix, and stability) for each of the three types of nursing staff on quality of care, so we included variables added sequentially to capture these effects of interest. Thus, the third specification included agency staffing factors (plus staffing levels and facility and market controls). The fourth specification included professional staff mix (plus staffing levels, agency staffing factors, facility controls, and market controls). The fifth specification included stability (plus staffing levels, professional staff mix, agency staffing factors, facility controls, and market controls).
Castle and Engberg (2007a) showed that in some cases the effect of each of the staffing characteristics depends on the level of that characteristic and of the other characteristics. To capture this variation in effect, Castle and Engberg (2007a) included a set of squared terms and interactions in their analyses. The squared terms capture nonlinear relationships between staffing characteristics and quality, and the interactions capture whether the effect of a particular staffing characteristic on quality is influenced by the level of a different staffing characteristic. Given the number of staffing variables of interest in our analyses, the potential number of possible squared terms and interactions that could be included in the model specifications was large. Therefore, for parsimony our model specifications included the same squared terms and interactions previously included by Castle and Engberg (2007a). To aid interpretation of these terms, we subtracted out the mean of each staffing characteristic variable before creating the squared and interaction terms. This allows the coefficients on the linear staffing characteristic terms to be interpreted as the effect of the staffing characteristic on quality at the mean value of the staffing characteristic. For each staffing variable we used the natural logarithm of each of the values (after shifting to remove zeros) in order to reduce the skewness of the distributions.
Results
Staffing information came from two different surveys, and we identified duplicate responses from 832 facilities. In these duplicate cases, we randomly chose information from one survey to be included in the analyses. All staffing items from these duplicate surveys had levels of agreement exceeding 96% (80% is an often used minimum reliability standard; Porell, Caro, Silva, & Monane, 1998). In general, most items on the questionnaire were answered. After accounting for these few missing items (n = 48) and duplicate facilities, the analytic file consisted of information from 6,005 nursing homes.
Table 1 presents descriptive statistics for the variables used in the analysis. For the staffing characteristics of interest, the average staffing levels were 11.7, 15.6, and 31.4 FTEs per 100 residents for RNs, LPNs, and NAs, respectively. The average professional staff mix (i.e., RNs / [NAs + LPNs]) was 0.25. RN, LPN, and NA 5-year stability rates were 19.9%, 17.9%, and 14.3%, respectively. Agency use (as a percentage of FTE positions) averaged 8.7%, 10.2%, and 11.1% for RNs, LPNs, and NAs, respectively. These staffing characteristics were not highly correlated; the highest correlation (−.56) was between NA staffing and NA agency use (not shown).
Table 2 presents the coefficient estimates for the negative binomial regressions of the quality measure for physical restraint use (for long-stay residents). Standard errors, adjusted for clustering within markets, are in parentheses below the coefficient estimates. We present the results for the five analytic specifications for this quality measure because they appeared to be generally representative of all of the other quality measures examined (however, Table 3 subsequently shows the fully specified models for all of the quality measures).
We first examined the effects of the facility and market control variables (shown in Column 1 of Table 2). As would be expected, based on prior research and the conceptual model, several of these variables were significant. For example, for-profit ownership was associated with higher levels of restraint use (p <.01). Column 2 adds the staffing variables (along with their squared terms). For these staffing levels, we did not find any significant association with the restraint use quality measure. We next added agency staffing variables (along with their squared terms and interaction terms) to the model specification (see Column 3). High NA agency staffing and high RN agency staffing levels were associated with high levels of restraint use (p <.05). For the stability measures, added next to the model (along with their squared terms and interaction terms), we found that high rates of RN stability were associated with low levels of restraint use (p <.01), and high rates of NA stability were associated with low restraint use (p <.01; see Column 4). Finally, we added professional staff mix to the model specification (along with the squared terms and interaction terms; see Column 5). High professional staff mix was associated with low restraint use (p <.01).
Continuing to examine the findings displayed in Column 5 of Table 2, we found that the staffing variables reached significance in Column 5 when the model was fully specified. That is, high NA staffing was associated with low restraint use (p <.05), high LPN staffing was associated with low restraint use (p <.05), and high RN staffing levels were associated with low restraint use (p <.05). Thus, we found support for our proposition that a more adequately specified labor composition may be advantageous when examining the quality–staffing relationship.
Table 3 shows the fully specified regression results for all six quality measures of interest. Column 1 presents the results for percent physical restraint use (for long-stay residents; this is a repeat of the findings previously presented in Column 5 of Table 2). Subsequent columns present the findings for the other quality measures. Of note, of the 18 staffing level variables of interest (i.e., 6 quality measures and NA, LPN, and RN staffing), 15 coefficients were statistically significant. All six of the NA staffing coefficients were statistically significant in the hypothesized direction (i.e., more staffing was associated with better quality), four of the six LPN staffing coefficients were statistically significant in the hypothesized direction, and five of the six RN staffing coefficients were statistically significant in the hypothesized direction.
Discussion
Several reasons likely exist for the generally weak relationship seen in the literature between nursing home caregiver staffing levels and quality. We primarily address poor-quality staffing data and underspecified models (i.e., omitted variable bias) in this research. Moreover, with primary data coming from more than 6,000 nursing homes, our analyses represent a significant advance over prior works in this area.
The results of our analyses indicate that simple analytic specifications including only facility and market control variables, along with staffing level variables, produce few significant findings at a p level of 1% (from all six quality measures, 2 of 18 staffing level variables were significant; results not shown). Thus, following the conclusion of our literature review, our empirical findings show only a weak association between staffing levels and quality.
When we used a more complete model specification (including agency staffing, stability, and professional staff mix), we consistently identified significant findings for all three staffing level variables. Thus, following our conceptual model, we assert that resident care is dependent not only on how much is done (represented primarily by staffing levels), but also upon consistency of care, coordination, and care practices. This finding has implications for both researchers and policy makers.
For researchers, our results show that future examinations of staffing level–quality relationships need to be more inclusive than those of the past. Similar to some other recent research (i.e., Arling et al., 2007; Castle & Engberg, 2007a), we propose that researchers need to move beyond simply examining staffing levels. Examining care processes more adequately is important and may lead to broader policy debate over staffing issues rather than staffing levels in nursing homes.
Despite the established weak staffing level–quality relationship, some state policy makers have mandated minimum staffing levels. Our results show that use of agency staff and professional staff mix are significant influences on quality. These may also represent a legislative opportunity to improve the quality of care in nursing homes. For example, experts could specify minimum ratios of RNs (although some states already do this; Mueller et al., 2006) and maximum levels of agency staff. However, to make such recommendations, experts would need subsequent analyses to identify the functional form for the RN–quality and agency–quality relationships.
Policy makers may be hard-pressed to directly regulate staff stability; nevertheless, indirect approaches designed to improve the job satisfaction of caregivers may also represent a legislative opportunity. Examples of this indirect approach include both the Better Jobs Better Care demonstrations (www.bjbc.org) and federal demonstration grants (www.cms.hhs.gov/newfreedom). Both of these initiatives focus on improving working conditions, and thereby improving job satisfaction and retention. This could increase average stability levels in facilities. Still, these demonstration projects have yet to show their “downstream” influences on staff retention.
We also found several facility and market variables to be significant. This is consonant with the theoretical model used and with prior literature. The results for the facility characteristics bed size and for-profit ownership, and the market characteristic competition, were particularly robust in all of the model specifications.
As we noted previously, much of our analyses followed the prior work of Castle and Engberg (2007a). This prior analysis determined that quality of care was influenced by all of these staffing characteristics examined, but achieving the highest quality was dependent on having more than one favorable staffing characteristic. Our findings also seem to generally support this prior work. However, in comparing our findings with this prior analysis, some differences are evident. Castle and Engberg (2007a) found few significant associations for LPNs. We identified fewer significant findings for LPNs than for NAs and RNs, but, nevertheless, some significant findings for LPNs were evident. Castle and Engberg (2007a) also found that low RN stability was associated with high quality (contrary to their expectations). We found only two of the six RN stability coefficients to be significant, but in the expected direction (i.e., high RN stability was associated with better quality).
Limitations and Suggestions for Future Research
We used the quality measures from Nursing Home Compare as quality indicators. As discussed previously, these measures have numerous advantages. Nevertheless, the quality measures have limitations, the most notable of which is the numerous facilities with missing information (General Accounting Office, 2002).
The full model specification used in the analyses (see Table 3) accounted for some omitted variable bias. However, in future analyses the inclusion of other variables may be appropriate. For example, our study (like most in this area) was cross-sectional. As such, unobserved nursing home traits could have biased the findings. This represents a further form of potential omitted variable bias that investigators should address by using longitudinal analyses.
We chose to include interactions among the staffing characteristics based on prior analyses. Other researchers could examine alternative specifications in the type and number of interactions. For example, future analyses could examine interactions among the types of caregivers for each staffing characteristic. Such additional analyses could further researchers' understanding of the complex relationships between staffing characteristics and quality.
Prior analyses have used a composite quality factor as a dependent variable in the analyses (e.g., Castle & Engberg, 2007a). This has the advantage of streamlining the analyses, but at the cost of creating difficulties in interpretation. Therefore, we used individual quality measures in our analyses. However, we caution that the analysis of individual quality measures in this way represents a first attempt at this type of analysis. Our findings may not hold with other quality indicators.
We also note that, despite statistical significance, the strength of the staffing level findings was generally not great. Nevertheless, the practical significance of these findings may still be considerable. The practical significance of staffing lies in the cumulative influence across multiple quality measures (which our analyses clearly show is occurring).
We believe the findings generally correspond very well with the proposed conceptual model (with some exceptions that are noted below). Nevertheless, future analyses could add other influences. For example, top management likely influences care processes and staffing decisions (Anderson, Issel, & McDaniel, 2003).
Some inconsistencies between the findings and our proposed conceptual model do exist. For example, for some staffing characteristics we did not identify significant findings. This was the case for RN stability, with only two of the six coefficients significant. This, of course, was in contrast to NA stability (with five of the six coefficients significant). We speculate that our stability measure could be improved, as our use of 5 years or more tenure at a facility was a relatively insensitive metric. Alternatively, these findings may also highlight the need to further refine the conceptual model to be more specific for different types of caregivers. A second inconsistency between our proposed conceptual model and the findings occurred for RN agency staffing. In this case, for three quality measures high RN agency staffing was associated with poor quality, whereas for a further two high RN agency staffing was associated with better quality. Very little is known about agency staff use in nursing homes, but we speculate that for some quality measures agency RNs may bring an accumulation of knowledge that can actually improve care (which seems to be a reasonable explanation for our findings for the prevention of pressure ulcers and catheter use). A further explanation for these inconsistencies comes from the known orthogonal nature of quality indicators (Mor et al., 2003). That is, a facility's undertaking of appropriate care processes in one area does not necessarily generalize to other areas of care. As such, some inconsistent findings between the quality measures may be expected.
Conclusion
The results of our literature review showed that, in general, only a weak association exists between nursing home staffing levels and quality. We showed in empirical analyses that when we examined staffing level variables alone, we found a weak relationship with quality. However, when we used a more complete model specification (including agency staffing, stability, and professional staff mix, in addition to staffing level), we identified consistently significant findings for all three staffing level variables. Thus, we assert that resident care is dependent not only on how much is done (represented primarily by staffing levels), but also on consistency of care, coordination, and care practices. We advocate that future analyses examining staffing levels include more fully specified analytic approaches.
This study was supported in part by a grant from the Agency for Healthcare Research and Quality, 1 R01 HS016808-01 Staffing Characteristics of Nursing Homes and Quality (PI: Castle).
School of Public Health, University of Pittsburgh, PA.
RAND Corporation, Pittsburgh, PA.
Decision Editor: William J. McAuley, PhD
Variable . | Definition . | M or % . | SD . | |||
---|---|---|---|---|---|---|
Dependent variablesa | ||||||
Pain | Percent with moderate to severe pain (LSR) | 4.82 | 5.20 | |||
Pressure sores (low risk) | Percent low-risk residents with pressure sores (LSR) | 5.22 | 3.51 | |||
Pressure sores (high risk) | Percent high-risk residents with pressure sores (LSR) | 5.30 | 5.16 | |||
Physical restraint | Percent physical restraint use (LSR) | 13.11 | 8.57 | |||
Catheterized | Percent had a catheter inserted and left in bladder (LSR) | 6.47 | 7.08 | |||
Pain | Percent with moderate to severe pain (SSR) | 20.49 | 8.19 | |||
Independent variables | ||||||
Staffing characteristicsb | ||||||
Staffing levels | ||||||
RN staffing | FTE RNs per 100 residents | 11.70 | 9.54 | |||
LPN staffing | FTE LPNs per 100 residents | 15.63 | 8.55 | |||
NA staffing | FTE NAs per 100 residents | 31.41 | 9.87 | |||
Professional staff mix | ||||||
Staff mix | Ratio of RNs to NAs plus LPNs (i.e., RNs / [NAs + LPNs]) | 0.25 | 0.42 | |||
Stability | ||||||
RN stability | Percent RNs with 5 or more years tenure at the facility | 19.91% | 13.07 | |||
LPN stability | Percent LPNs with 5 or more years tenure at the facility | 17.93% | 15.18 | |||
NA stability | Percent NAs with 5 or more years tenure at the facility | 14.33% | 8.88 | |||
Agency staffing | ||||||
RN agency | Percent FTE positions filled by agency RNs in past year | 8.72% | 3.10 | |||
LPN agency | Percent FTE positions filled by agency LPNs in past year | 10.20% | 3.38 | |||
NA agency | Percent FTE positions filled by agency NAs in past year | 11.13% | 4.56 | |||
Organizational characteristicsc | ||||||
Organizational size | Number of beds | 133.21 | 90.16 | |||
Ownership | For profit or not for profit | 57.40% | — | |||
Chain membership | Whether member of a nursing home chain or not | 52.03% | — | |||
Occupancy | Average daily occupancy rate | 87.69% | 13.77 | |||
Medicaid occupancy | Average daily Medicaid occupancy rate | 50.28% | 24.10 | |||
Market characteristicsd | ||||||
Competition | Herfindahl index. The sum of each facility's squared percentage share of beds in the county for all facilities in the county (0–1). Higher values indicate a less competitive market. | 0.21 | 0.20 | |||
Unemployment rate | Percent workers in county unemployed | 4.23% | 1.71 |
Variable . | Definition . | M or % . | SD . | |||
---|---|---|---|---|---|---|
Dependent variablesa | ||||||
Pain | Percent with moderate to severe pain (LSR) | 4.82 | 5.20 | |||
Pressure sores (low risk) | Percent low-risk residents with pressure sores (LSR) | 5.22 | 3.51 | |||
Pressure sores (high risk) | Percent high-risk residents with pressure sores (LSR) | 5.30 | 5.16 | |||
Physical restraint | Percent physical restraint use (LSR) | 13.11 | 8.57 | |||
Catheterized | Percent had a catheter inserted and left in bladder (LSR) | 6.47 | 7.08 | |||
Pain | Percent with moderate to severe pain (SSR) | 20.49 | 8.19 | |||
Independent variables | ||||||
Staffing characteristicsb | ||||||
Staffing levels | ||||||
RN staffing | FTE RNs per 100 residents | 11.70 | 9.54 | |||
LPN staffing | FTE LPNs per 100 residents | 15.63 | 8.55 | |||
NA staffing | FTE NAs per 100 residents | 31.41 | 9.87 | |||
Professional staff mix | ||||||
Staff mix | Ratio of RNs to NAs plus LPNs (i.e., RNs / [NAs + LPNs]) | 0.25 | 0.42 | |||
Stability | ||||||
RN stability | Percent RNs with 5 or more years tenure at the facility | 19.91% | 13.07 | |||
LPN stability | Percent LPNs with 5 or more years tenure at the facility | 17.93% | 15.18 | |||
NA stability | Percent NAs with 5 or more years tenure at the facility | 14.33% | 8.88 | |||
Agency staffing | ||||||
RN agency | Percent FTE positions filled by agency RNs in past year | 8.72% | 3.10 | |||
LPN agency | Percent FTE positions filled by agency LPNs in past year | 10.20% | 3.38 | |||
NA agency | Percent FTE positions filled by agency NAs in past year | 11.13% | 4.56 | |||
Organizational characteristicsc | ||||||
Organizational size | Number of beds | 133.21 | 90.16 | |||
Ownership | For profit or not for profit | 57.40% | — | |||
Chain membership | Whether member of a nursing home chain or not | 52.03% | — | |||
Occupancy | Average daily occupancy rate | 87.69% | 13.77 | |||
Medicaid occupancy | Average daily Medicaid occupancy rate | 50.28% | 24.10 | |||
Market characteristicsd | ||||||
Competition | Herfindahl index. The sum of each facility's squared percentage share of beds in the county for all facilities in the county (0–1). Higher values indicate a less competitive market. | 0.21 | 0.20 | |||
Unemployment rate | Percent workers in county unemployed | 4.23% | 1.71 |
Notes: LSR = long-stay residents; SSR = short-stay residents; RN = registered nurse; FTE = full-time equivalent; LPN = licensed practical nurse; NA = nurse aide.
aVariables taken from Nursing Home Compare. Variables are not significantly different (p >.05) than the national averages reported in Nursing Home Compare for 2004.
bVariables taken from primary data collection. Statistics presented come from the analytic file consisting of 6,005 facilities and 1,551 markets.
cVariables taken from the Online Survey, Certification and Reporting (OSCAR) data. Variables are not significantly different (p >.05) than the national averages reported in OSCAR data for 2004.
dVariables taken from the Area Resource File (ARF). Variables are not significantly different (p >.05) than the national averages reported in the ARF data for 2004.
Variable . | Definition . | M or % . | SD . | |||
---|---|---|---|---|---|---|
Dependent variablesa | ||||||
Pain | Percent with moderate to severe pain (LSR) | 4.82 | 5.20 | |||
Pressure sores (low risk) | Percent low-risk residents with pressure sores (LSR) | 5.22 | 3.51 | |||
Pressure sores (high risk) | Percent high-risk residents with pressure sores (LSR) | 5.30 | 5.16 | |||
Physical restraint | Percent physical restraint use (LSR) | 13.11 | 8.57 | |||
Catheterized | Percent had a catheter inserted and left in bladder (LSR) | 6.47 | 7.08 | |||
Pain | Percent with moderate to severe pain (SSR) | 20.49 | 8.19 | |||
Independent variables | ||||||
Staffing characteristicsb | ||||||
Staffing levels | ||||||
RN staffing | FTE RNs per 100 residents | 11.70 | 9.54 | |||
LPN staffing | FTE LPNs per 100 residents | 15.63 | 8.55 | |||
NA staffing | FTE NAs per 100 residents | 31.41 | 9.87 | |||
Professional staff mix | ||||||
Staff mix | Ratio of RNs to NAs plus LPNs (i.e., RNs / [NAs + LPNs]) | 0.25 | 0.42 | |||
Stability | ||||||
RN stability | Percent RNs with 5 or more years tenure at the facility | 19.91% | 13.07 | |||
LPN stability | Percent LPNs with 5 or more years tenure at the facility | 17.93% | 15.18 | |||
NA stability | Percent NAs with 5 or more years tenure at the facility | 14.33% | 8.88 | |||
Agency staffing | ||||||
RN agency | Percent FTE positions filled by agency RNs in past year | 8.72% | 3.10 | |||
LPN agency | Percent FTE positions filled by agency LPNs in past year | 10.20% | 3.38 | |||
NA agency | Percent FTE positions filled by agency NAs in past year | 11.13% | 4.56 | |||
Organizational characteristicsc | ||||||
Organizational size | Number of beds | 133.21 | 90.16 | |||
Ownership | For profit or not for profit | 57.40% | — | |||
Chain membership | Whether member of a nursing home chain or not | 52.03% | — | |||
Occupancy | Average daily occupancy rate | 87.69% | 13.77 | |||
Medicaid occupancy | Average daily Medicaid occupancy rate | 50.28% | 24.10 | |||
Market characteristicsd | ||||||
Competition | Herfindahl index. The sum of each facility's squared percentage share of beds in the county for all facilities in the county (0–1). Higher values indicate a less competitive market. | 0.21 | 0.20 | |||
Unemployment rate | Percent workers in county unemployed | 4.23% | 1.71 |
Variable . | Definition . | M or % . | SD . | |||
---|---|---|---|---|---|---|
Dependent variablesa | ||||||
Pain | Percent with moderate to severe pain (LSR) | 4.82 | 5.20 | |||
Pressure sores (low risk) | Percent low-risk residents with pressure sores (LSR) | 5.22 | 3.51 | |||
Pressure sores (high risk) | Percent high-risk residents with pressure sores (LSR) | 5.30 | 5.16 | |||
Physical restraint | Percent physical restraint use (LSR) | 13.11 | 8.57 | |||
Catheterized | Percent had a catheter inserted and left in bladder (LSR) | 6.47 | 7.08 | |||
Pain | Percent with moderate to severe pain (SSR) | 20.49 | 8.19 | |||
Independent variables | ||||||
Staffing characteristicsb | ||||||
Staffing levels | ||||||
RN staffing | FTE RNs per 100 residents | 11.70 | 9.54 | |||
LPN staffing | FTE LPNs per 100 residents | 15.63 | 8.55 | |||
NA staffing | FTE NAs per 100 residents | 31.41 | 9.87 | |||
Professional staff mix | ||||||
Staff mix | Ratio of RNs to NAs plus LPNs (i.e., RNs / [NAs + LPNs]) | 0.25 | 0.42 | |||
Stability | ||||||
RN stability | Percent RNs with 5 or more years tenure at the facility | 19.91% | 13.07 | |||
LPN stability | Percent LPNs with 5 or more years tenure at the facility | 17.93% | 15.18 | |||
NA stability | Percent NAs with 5 or more years tenure at the facility | 14.33% | 8.88 | |||
Agency staffing | ||||||
RN agency | Percent FTE positions filled by agency RNs in past year | 8.72% | 3.10 | |||
LPN agency | Percent FTE positions filled by agency LPNs in past year | 10.20% | 3.38 | |||
NA agency | Percent FTE positions filled by agency NAs in past year | 11.13% | 4.56 | |||
Organizational characteristicsc | ||||||
Organizational size | Number of beds | 133.21 | 90.16 | |||
Ownership | For profit or not for profit | 57.40% | — | |||
Chain membership | Whether member of a nursing home chain or not | 52.03% | — | |||
Occupancy | Average daily occupancy rate | 87.69% | 13.77 | |||
Medicaid occupancy | Average daily Medicaid occupancy rate | 50.28% | 24.10 | |||
Market characteristicsd | ||||||
Competition | Herfindahl index. The sum of each facility's squared percentage share of beds in the county for all facilities in the county (0–1). Higher values indicate a less competitive market. | 0.21 | 0.20 | |||
Unemployment rate | Percent workers in county unemployed | 4.23% | 1.71 |
Notes: LSR = long-stay residents; SSR = short-stay residents; RN = registered nurse; FTE = full-time equivalent; LPN = licensed practical nurse; NA = nurse aide.
aVariables taken from Nursing Home Compare. Variables are not significantly different (p >.05) than the national averages reported in Nursing Home Compare for 2004.
bVariables taken from primary data collection. Statistics presented come from the analytic file consisting of 6,005 facilities and 1,551 markets.
cVariables taken from the Online Survey, Certification and Reporting (OSCAR) data. Variables are not significantly different (p >.05) than the national averages reported in OSCAR data for 2004.
dVariables taken from the Area Resource File (ARF). Variables are not significantly different (p >.05) than the national averages reported in the ARF data for 2004.
Variable . | (1) Facility and Market Controls . | (2) Plus Staffing Levels . | (3) Plus Agency Staffing . | (4) Plus Staff Stability . | (5) Plus Professional Staff Mix . |
---|---|---|---|---|---|
Organizational size (beds) | 1.088** (0.038) | 1.085** (0.037) | 1.083** (0.037) | 1.086** (0.033) | 1.087** (0.039) |
Ownership (for profit = 1) | 1.228* (0.148) | 1.222* (0.142) | 1.221* (0.140) | 1.219* (0.142) | 1.220* (0.142) |
Chain membership (chain member = 1) | 0.952 (0.134) | 0.950 (0.136) | 0.945 (0.132) | 0.947 (0.130) | 0.922 (0.136) |
Occupancy | 0.925 (0.059) | 0.935 (0.062) | 0.920 (0.069) | 0.940 (0.072) | 0.931 (0.065) |
Medicaid occupancy | 1.114** (0.056) | 1.121** (0.055) | 1.122** (0.052) | 1.131** (0.057) | 1.119** (0.053) |
Competition | 0.929* (0.038) | 0.932* (0.039) | 0.941* (0.040) | 0.936* (0.036) | 0.933* (0.037) |
Unemployment rate | 0.966 (0.044) | 0.961 (0.043) | 0.963 (0.040) | 0.967 (0.045) | 0.955 (0.047) |
Log NA staffing | 0.921 (0.039) | 0.937* (0.040) | 0.932* (0.039) | 0.935* (0.037) | |
Log LPN staffing | 0.815 (0.042) | 0.810 (0.047) | 0.837* (0.036) | 0.833* (0.035) | |
Log RN staffing | 0.650 (0.038) | 0.655 (0.036) | 0.640 (0.038) | 0.682* (0.025) | |
Log NA staffing squared | 0.824 (0.159) | 0.831 (0.156) | 0.844 (0.145) | 0.835 (0.162) | |
Log LPN staffing squared | 0.991 (0.033) | 0.995 (0.036) | 0.992 (0.038) | 0.987 (0.043) | |
Log RN staffing squared | 0.987 (0.031) | 0.985 (0.030) | 0.974 (0.036) | 0.969 (0.041) | |
Log NA agency staffing | 1.078* (0.044) | 1.088** (0.034) | 1.097** (0.033) | ||
Log LPN agency staffing | 1.014 (0.021) | 1.015 (0.025) | 1.021 (0.028) | ||
Log RN agency staffing | 1.087*** (0.030) | 1.098*** (0.029) | 1.101*** (0.026) | ||
Log NA agency staffing squared | 0.886** (0.051) | 0.882** (0.042) | 0.870** (0.040) | ||
Log LPN agency staffing squared | 1.049 (0.127) | 1.042 (0.131) | 1.041 (0.137) | ||
Log RN agency staffing squared | 1.059** (0.029) | 1.063** (0.025) | 1.069** (0.020) | ||
Log NA Agency Staffing × NA Staffing | 1.037 (0.230) | 1.047 (0.233) | 1.065 (0.245) | ||
Log LPN Agency Staffing × LPN Staffing | 1.025 (0.064) | 1.032 (0.063) | 1.048 (0.060) | ||
Log RN Agency Staffing × RN Staffing | 0.850** (0.054) | 0.842** (0.053) | 0.859** (0.050) | ||
Log NA stability | 0.935** (0.028) | 0.914** (0.025) | |||
Log LPN stability | 1.032 (0.016) | 1.030 (0.019) | |||
Log RN stability | 0.912 (0.061) | 0.915 (0.064) | |||
Log NA stability squared | 0.936** (0.031) | 0.920** (0.029) | |||
Log LPN stability squared | 0.750 (0.240) | 0.744 (0.223) | |||
Log RN stability squared | 0.910 (0.225) | 0.917 (0.229) | |||
Log NA Stability × NA Staffing | 0.853* (0.080) | 0.861* (0.074) | |||
Log LPN Stability × LPN Staffing | 1.055 (0.084) | 1.031 (0.092) | |||
Log RN Stability × RN Staffing | 0.813** (0.072) | 0.818** (0.075) | |||
Log NA Stability × Agency | 0.917*** (0.028) | 0.921*** (0.026) | |||
Log LPN Stability × Agency | 0.951 (0.115) | 0.975 (0.122) | |||
Log RN Stability × Agency | 0.963 (0.044) | 0.956 (0.061) | |||
Log professional staff mix | 0.889** (0.048) | ||||
Log professional staff mix squared | 0.981 (0.043) | ||||
Log Professional Staff Mix × Agency | 0.826* (0.035) | ||||
Log Professional Staff Mix × Stability | 0.923* (0.044) | ||||
Pseudo R2 | 0.14 | 0.15 | 0.17 | 0.21 | 0.24 |
Variable . | (1) Facility and Market Controls . | (2) Plus Staffing Levels . | (3) Plus Agency Staffing . | (4) Plus Staff Stability . | (5) Plus Professional Staff Mix . |
---|---|---|---|---|---|
Organizational size (beds) | 1.088** (0.038) | 1.085** (0.037) | 1.083** (0.037) | 1.086** (0.033) | 1.087** (0.039) |
Ownership (for profit = 1) | 1.228* (0.148) | 1.222* (0.142) | 1.221* (0.140) | 1.219* (0.142) | 1.220* (0.142) |
Chain membership (chain member = 1) | 0.952 (0.134) | 0.950 (0.136) | 0.945 (0.132) | 0.947 (0.130) | 0.922 (0.136) |
Occupancy | 0.925 (0.059) | 0.935 (0.062) | 0.920 (0.069) | 0.940 (0.072) | 0.931 (0.065) |
Medicaid occupancy | 1.114** (0.056) | 1.121** (0.055) | 1.122** (0.052) | 1.131** (0.057) | 1.119** (0.053) |
Competition | 0.929* (0.038) | 0.932* (0.039) | 0.941* (0.040) | 0.936* (0.036) | 0.933* (0.037) |
Unemployment rate | 0.966 (0.044) | 0.961 (0.043) | 0.963 (0.040) | 0.967 (0.045) | 0.955 (0.047) |
Log NA staffing | 0.921 (0.039) | 0.937* (0.040) | 0.932* (0.039) | 0.935* (0.037) | |
Log LPN staffing | 0.815 (0.042) | 0.810 (0.047) | 0.837* (0.036) | 0.833* (0.035) | |
Log RN staffing | 0.650 (0.038) | 0.655 (0.036) | 0.640 (0.038) | 0.682* (0.025) | |
Log NA staffing squared | 0.824 (0.159) | 0.831 (0.156) | 0.844 (0.145) | 0.835 (0.162) | |
Log LPN staffing squared | 0.991 (0.033) | 0.995 (0.036) | 0.992 (0.038) | 0.987 (0.043) | |
Log RN staffing squared | 0.987 (0.031) | 0.985 (0.030) | 0.974 (0.036) | 0.969 (0.041) | |
Log NA agency staffing | 1.078* (0.044) | 1.088** (0.034) | 1.097** (0.033) | ||
Log LPN agency staffing | 1.014 (0.021) | 1.015 (0.025) | 1.021 (0.028) | ||
Log RN agency staffing | 1.087*** (0.030) | 1.098*** (0.029) | 1.101*** (0.026) | ||
Log NA agency staffing squared | 0.886** (0.051) | 0.882** (0.042) | 0.870** (0.040) | ||
Log LPN agency staffing squared | 1.049 (0.127) | 1.042 (0.131) | 1.041 (0.137) | ||
Log RN agency staffing squared | 1.059** (0.029) | 1.063** (0.025) | 1.069** (0.020) | ||
Log NA Agency Staffing × NA Staffing | 1.037 (0.230) | 1.047 (0.233) | 1.065 (0.245) | ||
Log LPN Agency Staffing × LPN Staffing | 1.025 (0.064) | 1.032 (0.063) | 1.048 (0.060) | ||
Log RN Agency Staffing × RN Staffing | 0.850** (0.054) | 0.842** (0.053) | 0.859** (0.050) | ||
Log NA stability | 0.935** (0.028) | 0.914** (0.025) | |||
Log LPN stability | 1.032 (0.016) | 1.030 (0.019) | |||
Log RN stability | 0.912 (0.061) | 0.915 (0.064) | |||
Log NA stability squared | 0.936** (0.031) | 0.920** (0.029) | |||
Log LPN stability squared | 0.750 (0.240) | 0.744 (0.223) | |||
Log RN stability squared | 0.910 (0.225) | 0.917 (0.229) | |||
Log NA Stability × NA Staffing | 0.853* (0.080) | 0.861* (0.074) | |||
Log LPN Stability × LPN Staffing | 1.055 (0.084) | 1.031 (0.092) | |||
Log RN Stability × RN Staffing | 0.813** (0.072) | 0.818** (0.075) | |||
Log NA Stability × Agency | 0.917*** (0.028) | 0.921*** (0.026) | |||
Log LPN Stability × Agency | 0.951 (0.115) | 0.975 (0.122) | |||
Log RN Stability × Agency | 0.963 (0.044) | 0.956 (0.061) | |||
Log professional staff mix | 0.889** (0.048) | ||||
Log professional staff mix squared | 0.981 (0.043) | ||||
Log Professional Staff Mix × Agency | 0.826* (0.035) | ||||
Log Professional Staff Mix × Stability | 0.923* (0.044) | ||||
Pseudo R2 | 0.14 | 0.15 | 0.17 | 0.21 | 0.24 |
Notes: Robust standard errors in parentheses. Note that dummy variable indicating source of primary data was included and was not significant. Bold print is used to highlight the staffing level variables. NA = nurse aide; LPN = licensed practical nurse; RN = registered nurse.
*p <.05; **p <.01; ***p <.001.
Variable . | (1) Facility and Market Controls . | (2) Plus Staffing Levels . | (3) Plus Agency Staffing . | (4) Plus Staff Stability . | (5) Plus Professional Staff Mix . |
---|---|---|---|---|---|
Organizational size (beds) | 1.088** (0.038) | 1.085** (0.037) | 1.083** (0.037) | 1.086** (0.033) | 1.087** (0.039) |
Ownership (for profit = 1) | 1.228* (0.148) | 1.222* (0.142) | 1.221* (0.140) | 1.219* (0.142) | 1.220* (0.142) |
Chain membership (chain member = 1) | 0.952 (0.134) | 0.950 (0.136) | 0.945 (0.132) | 0.947 (0.130) | 0.922 (0.136) |
Occupancy | 0.925 (0.059) | 0.935 (0.062) | 0.920 (0.069) | 0.940 (0.072) | 0.931 (0.065) |
Medicaid occupancy | 1.114** (0.056) | 1.121** (0.055) | 1.122** (0.052) | 1.131** (0.057) | 1.119** (0.053) |
Competition | 0.929* (0.038) | 0.932* (0.039) | 0.941* (0.040) | 0.936* (0.036) | 0.933* (0.037) |
Unemployment rate | 0.966 (0.044) | 0.961 (0.043) | 0.963 (0.040) | 0.967 (0.045) | 0.955 (0.047) |
Log NA staffing | 0.921 (0.039) | 0.937* (0.040) | 0.932* (0.039) | 0.935* (0.037) | |
Log LPN staffing | 0.815 (0.042) | 0.810 (0.047) | 0.837* (0.036) | 0.833* (0.035) | |
Log RN staffing | 0.650 (0.038) | 0.655 (0.036) | 0.640 (0.038) | 0.682* (0.025) | |
Log NA staffing squared | 0.824 (0.159) | 0.831 (0.156) | 0.844 (0.145) | 0.835 (0.162) | |
Log LPN staffing squared | 0.991 (0.033) | 0.995 (0.036) | 0.992 (0.038) | 0.987 (0.043) | |
Log RN staffing squared | 0.987 (0.031) | 0.985 (0.030) | 0.974 (0.036) | 0.969 (0.041) | |
Log NA agency staffing | 1.078* (0.044) | 1.088** (0.034) | 1.097** (0.033) | ||
Log LPN agency staffing | 1.014 (0.021) | 1.015 (0.025) | 1.021 (0.028) | ||
Log RN agency staffing | 1.087*** (0.030) | 1.098*** (0.029) | 1.101*** (0.026) | ||
Log NA agency staffing squared | 0.886** (0.051) | 0.882** (0.042) | 0.870** (0.040) | ||
Log LPN agency staffing squared | 1.049 (0.127) | 1.042 (0.131) | 1.041 (0.137) | ||
Log RN agency staffing squared | 1.059** (0.029) | 1.063** (0.025) | 1.069** (0.020) | ||
Log NA Agency Staffing × NA Staffing | 1.037 (0.230) | 1.047 (0.233) | 1.065 (0.245) | ||
Log LPN Agency Staffing × LPN Staffing | 1.025 (0.064) | 1.032 (0.063) | 1.048 (0.060) | ||
Log RN Agency Staffing × RN Staffing | 0.850** (0.054) | 0.842** (0.053) | 0.859** (0.050) | ||
Log NA stability | 0.935** (0.028) | 0.914** (0.025) | |||
Log LPN stability | 1.032 (0.016) | 1.030 (0.019) | |||
Log RN stability | 0.912 (0.061) | 0.915 (0.064) | |||
Log NA stability squared | 0.936** (0.031) | 0.920** (0.029) | |||
Log LPN stability squared | 0.750 (0.240) | 0.744 (0.223) | |||
Log RN stability squared | 0.910 (0.225) | 0.917 (0.229) | |||
Log NA Stability × NA Staffing | 0.853* (0.080) | 0.861* (0.074) | |||
Log LPN Stability × LPN Staffing | 1.055 (0.084) | 1.031 (0.092) | |||
Log RN Stability × RN Staffing | 0.813** (0.072) | 0.818** (0.075) | |||
Log NA Stability × Agency | 0.917*** (0.028) | 0.921*** (0.026) | |||
Log LPN Stability × Agency | 0.951 (0.115) | 0.975 (0.122) | |||
Log RN Stability × Agency | 0.963 (0.044) | 0.956 (0.061) | |||
Log professional staff mix | 0.889** (0.048) | ||||
Log professional staff mix squared | 0.981 (0.043) | ||||
Log Professional Staff Mix × Agency | 0.826* (0.035) | ||||
Log Professional Staff Mix × Stability | 0.923* (0.044) | ||||
Pseudo R2 | 0.14 | 0.15 | 0.17 | 0.21 | 0.24 |
Variable . | (1) Facility and Market Controls . | (2) Plus Staffing Levels . | (3) Plus Agency Staffing . | (4) Plus Staff Stability . | (5) Plus Professional Staff Mix . |
---|---|---|---|---|---|
Organizational size (beds) | 1.088** (0.038) | 1.085** (0.037) | 1.083** (0.037) | 1.086** (0.033) | 1.087** (0.039) |
Ownership (for profit = 1) | 1.228* (0.148) | 1.222* (0.142) | 1.221* (0.140) | 1.219* (0.142) | 1.220* (0.142) |
Chain membership (chain member = 1) | 0.952 (0.134) | 0.950 (0.136) | 0.945 (0.132) | 0.947 (0.130) | 0.922 (0.136) |
Occupancy | 0.925 (0.059) | 0.935 (0.062) | 0.920 (0.069) | 0.940 (0.072) | 0.931 (0.065) |
Medicaid occupancy | 1.114** (0.056) | 1.121** (0.055) | 1.122** (0.052) | 1.131** (0.057) | 1.119** (0.053) |
Competition | 0.929* (0.038) | 0.932* (0.039) | 0.941* (0.040) | 0.936* (0.036) | 0.933* (0.037) |
Unemployment rate | 0.966 (0.044) | 0.961 (0.043) | 0.963 (0.040) | 0.967 (0.045) | 0.955 (0.047) |
Log NA staffing | 0.921 (0.039) | 0.937* (0.040) | 0.932* (0.039) | 0.935* (0.037) | |
Log LPN staffing | 0.815 (0.042) | 0.810 (0.047) | 0.837* (0.036) | 0.833* (0.035) | |
Log RN staffing | 0.650 (0.038) | 0.655 (0.036) | 0.640 (0.038) | 0.682* (0.025) | |
Log NA staffing squared | 0.824 (0.159) | 0.831 (0.156) | 0.844 (0.145) | 0.835 (0.162) | |
Log LPN staffing squared | 0.991 (0.033) | 0.995 (0.036) | 0.992 (0.038) | 0.987 (0.043) | |
Log RN staffing squared | 0.987 (0.031) | 0.985 (0.030) | 0.974 (0.036) | 0.969 (0.041) | |
Log NA agency staffing | 1.078* (0.044) | 1.088** (0.034) | 1.097** (0.033) | ||
Log LPN agency staffing | 1.014 (0.021) | 1.015 (0.025) | 1.021 (0.028) | ||
Log RN agency staffing | 1.087*** (0.030) | 1.098*** (0.029) | 1.101*** (0.026) | ||
Log NA agency staffing squared | 0.886** (0.051) | 0.882** (0.042) | 0.870** (0.040) | ||
Log LPN agency staffing squared | 1.049 (0.127) | 1.042 (0.131) | 1.041 (0.137) | ||
Log RN agency staffing squared | 1.059** (0.029) | 1.063** (0.025) | 1.069** (0.020) | ||
Log NA Agency Staffing × NA Staffing | 1.037 (0.230) | 1.047 (0.233) | 1.065 (0.245) | ||
Log LPN Agency Staffing × LPN Staffing | 1.025 (0.064) | 1.032 (0.063) | 1.048 (0.060) | ||
Log RN Agency Staffing × RN Staffing | 0.850** (0.054) | 0.842** (0.053) | 0.859** (0.050) | ||
Log NA stability | 0.935** (0.028) | 0.914** (0.025) | |||
Log LPN stability | 1.032 (0.016) | 1.030 (0.019) | |||
Log RN stability | 0.912 (0.061) | 0.915 (0.064) | |||
Log NA stability squared | 0.936** (0.031) | 0.920** (0.029) | |||
Log LPN stability squared | 0.750 (0.240) | 0.744 (0.223) | |||
Log RN stability squared | 0.910 (0.225) | 0.917 (0.229) | |||
Log NA Stability × NA Staffing | 0.853* (0.080) | 0.861* (0.074) | |||
Log LPN Stability × LPN Staffing | 1.055 (0.084) | 1.031 (0.092) | |||
Log RN Stability × RN Staffing | 0.813** (0.072) | 0.818** (0.075) | |||
Log NA Stability × Agency | 0.917*** (0.028) | 0.921*** (0.026) | |||
Log LPN Stability × Agency | 0.951 (0.115) | 0.975 (0.122) | |||
Log RN Stability × Agency | 0.963 (0.044) | 0.956 (0.061) | |||
Log professional staff mix | 0.889** (0.048) | ||||
Log professional staff mix squared | 0.981 (0.043) | ||||
Log Professional Staff Mix × Agency | 0.826* (0.035) | ||||
Log Professional Staff Mix × Stability | 0.923* (0.044) | ||||
Pseudo R2 | 0.14 | 0.15 | 0.17 | 0.21 | 0.24 |
Notes: Robust standard errors in parentheses. Note that dummy variable indicating source of primary data was included and was not significant. Bold print is used to highlight the staffing level variables. NA = nurse aide; LPN = licensed practical nurse; RN = registered nurse.
*p <.05; **p <.01; ***p <.001.
Variable . | (1) Percent Physical Restraint Use (LSR) . | (2) Percent With Moderate to Severe Pain (LSR) . | (3) Percent Low-Risk Residents With Pressure Sores (LSR) . | (4) Percent High-Risk Residents With Pressure Sores (LSR) . | (5) Percent Had a Catheter Inserted and Left in Bladder (LSR) . | (6) Percent With Moderate to Severe Pain (SSR) . |
---|---|---|---|---|---|---|
Organizational size (beds) | 1.087** (0.039) | 1.020* (0.011) | 1.010* (0.006) | 1.016 (0.010) | 1.051* (0.032) | 1.049 (0.056) |
Ownership (for profit = 1) | 1.220* (0.142) | 1.008 (0.029) | 1.039** (0.019) | 1.030*** (0.010) | 1.026 (0.017) | 1.032** (0.016) |
Chain membership (chain member = 1) | 0.922 (0.136) | 1.021* (0.013) | 1.092*** (0.031) | 1.013 (0.010) | 1.006 (0.007) | 0.888** (0.047) |
Occupancy | 0.931 (0.065) | 0.934 (0.116) | 0.998 (0.077) | 0.996 (0.068) | 0.852** (0.057) | 0.924 (0.089) |
Medicaid occupancy | 1.119** (0.053) | 1.216* (0.120) | 1.483*** (0.153) | 1.284** (0.150) | 1.419*** (0.132) | 1.228*** (0.090) |
Competition | 0.933* (0.037) | 0.940*** (0.017) | 0.913*** (0.030) | 0.893*** (0.025) | 0.967** (0.016) | 1.000 (0.016) |
Unemployment rate | 0.955 (0.047) | 0.978 (0.021) | 1.062 (0.042) | 1.015 (0.032) | 1.010 (0.045) | 0.999 (0.059) |
Log NA staffing | 0.935* (0.037) | 0.954*** (0.012) | 0.932*** (0.024) | 0.940*** (0.017) | 0.913*** (0.030) | 0.478*** (0.106) |
Log LPN staffing | 0.833* (0.035) | 1.121 (0.099) | 0.944 (0.042) | 0.856*** (0.043) | 0.925** (0.033) | 0.867** (0.049) |
Log RN staffing | 0.682* (0.025) | 0.771** (0.083) | 0.836* (0.082) | 0.878** (0.051) | 0.960 (0.063) | 0.844*** (0.052) |
Log NA staffing squared | 0.835 (0.162) | 0.842*** (0.053) | 0.988 (0.022) | 0.975 (0.062) | 0.929* (0.038) | 0.950 (0.075) |
Log LPN staffing squared | 0.987 (0.043) | 1.014 (0.021) | 0.986 (0.018) | 0.998 (0.030) | 0.950** (0.022) | 0.984 (0.012) |
Log RN staffing squared | 0.969 (0.041) | 0.939 (0.106) | 0.869** (0.054) | 0.878 (0.100) | 0.888 (0.099) | 0.740 (0.102) |
Log NA agency staffing | 1.097** (0.033) | 1.081*** (0.030) | 1.052*** (0.016) | 1.075*** (0.016) | 1.054*** (0.010) | 1.055*** (0.021) |
Log LPN agency staffing | 1.021 (0.028) | 0.920 (0.149) | 1.231 (0.283) | 1.005 (0.281) | 1.434** (0.235) | 1.019 (0.192) |
Log RN agency staffing | 1.101*** (0.026) | 1.184* (0.114) | 0.970* (0.015) | 1.028** (0.014) | 0.732** (0.021) | 1.012 (0.030) |
Log NA agency staffing squared | 0.870** (0.040) | 0.998 (0.030) | 0.950** (0.022) | 0.984 (0.012) | 0.989 (0.014) | 1.006 (0.013) |
Log LPN agency staffing squared | 1.041 (0.137) | 1.058 (0.048) | 1.104 (0.093) | 1.050 (0.084) | 1.066 (0.043) | 1.029 (0.040) |
Log RN agency staffing squared | 1.069** (0.021) | 0.976 (0.045) | 1.050** (0.024) | 1.078** (0.035) | 1.058** (0.027) | 1.003 (0.035) |
Log NA Agency Staffing × NA Staffing | 1.065 (0.245) | 0.978 (0.091) | 1.027 (0.034) | 0.961 (0.084) | 1.144** (0.078) | 0.993 (0.063) |
Log LPN Agency Staffing × LPN Staffing | 1.048 (0.062) | 1.167 (0.139) | 1.187 (0.323) | 1.118 (0.103) | 1.027 (0.230) | 1.222 (0.201) |
Log RN Agency Staffing × RN Staffing | 0.859** (0.051) | 0.913*** (0.031) | 0.893*** (0.025) | 0.966** (0.016) | 0.969* (0.018) | 1.014 (0.018) |
Log NA stability | 0.914** (0.025) | 1.026 (0.100) | 0.876** (0.050) | 0.885** (0.048) | 0.891** (0.040) | 0.867** (0.062) |
Log LPN stability | 1.030 (0.019) | 0.978 (0.038) | 1.012 (0.013) | 0.975 (0.026) | 0.992 (0.023) | 0.965 (0.036) |
Log RN stability | 0.915 (0.064) | 0.917* (0.039) | 0.947 (0.038) | 0.922* (0.045) | 0.913 (0.051) | 1.038 (0.066) |
Log NA stability squared | 0.920** (0.029) | 0.940 (0.114) | 0.998 (0.070) | 0.997 (0.069) | 0.853** (0.057) | 0.926 (0.090) |
Log LPN stability squared | 0.744 (0.223) | 0.996 (0.129) | 0.986 (0.117) | 1.070 (0.118) | 0.986 (0.115) | 1.005 (0.118) |
Log RN stability squared | 0.917 (0.229) | 1.021 (0.114) | 0.998 (0.070) | 0.999 (0.067) | 0.853** (0.057) | 0.928 (0.089) |
Log NA Stability × NA Staffing | 0.861* (0.074) | 1.026 (0.100) | 0.862** (0.041) | 0.887** (0.048) | 0.891** (0.040) | 0.867** (0.062) |
Log LPN Stability × LPN Staffing | 1.031 (0.092) | 1.022 (0.046) | 0.995 (0.020) | 0.990 (0.025) | 1.025 (0.021) | 0.950* (0.028) |
Log RN Stability × RN Staffing | 0.818** (0.075) | 0.976* (0.014) | 0.991 (0.026) | 0.877** (0.050) | 0.853** (0.057) | 0.928 (0.089) |
Log NA Stability × Agency | 0.920*** (0.026) | 0.842*** (0.053) | 0.988 (0.022) | 0.975 (0.062) | 0.929* (0.038) | 0.950 (0.055) |
Log LPN Stability × Agency | 0.975 (0.122) | 0.957 (0.088) | 1.049 (0.060) | 0.915 (0.100) | 1.017 (0.097) | 1.009 (0.053) |
Log RN Stability Agency | 0.956 (0.061) | 0.939 (0.101) | 0.869** (0.059) | 0.878 (0.102) | 0.888 (0.098) | 0.769*** (0.058) |
Log professional staff mix | 0.889** (0.048) | 0.773** (0.082) | 0.836* (0.086) | 0.878** (0.053) | 1.060 (0.061) | 0.958 (0.040) |
Log professional staff mix squared | 0.981 (0.043) | 0.998 (0.034) | 0.957 (0.088) | 0.949 (0.060) | 0.915 (0.100) | 1.017 (0.097) |
Log Professional Staff Mix × Agency | 0.826* (0.035) | 0.881** (0.050) | 0.980 (0.055) | 1.028 (0.085) | 0.984 (0.064) | 0.950* (0.028) |
Log Professional Staff Mix × Stability | 0.923* (0.044) | 0.932*** (0.024) | 0.940*** (0.017) | 0.913*** (0.030) | 0.893*** (0.025) | 0.968** (0.016) |
Pseudo R2 | 0.24 | 0.26 | 0.27 | 0.29 | 0.25 | 0.21 |
Variable . | (1) Percent Physical Restraint Use (LSR) . | (2) Percent With Moderate to Severe Pain (LSR) . | (3) Percent Low-Risk Residents With Pressure Sores (LSR) . | (4) Percent High-Risk Residents With Pressure Sores (LSR) . | (5) Percent Had a Catheter Inserted and Left in Bladder (LSR) . | (6) Percent With Moderate to Severe Pain (SSR) . |
---|---|---|---|---|---|---|
Organizational size (beds) | 1.087** (0.039) | 1.020* (0.011) | 1.010* (0.006) | 1.016 (0.010) | 1.051* (0.032) | 1.049 (0.056) |
Ownership (for profit = 1) | 1.220* (0.142) | 1.008 (0.029) | 1.039** (0.019) | 1.030*** (0.010) | 1.026 (0.017) | 1.032** (0.016) |
Chain membership (chain member = 1) | 0.922 (0.136) | 1.021* (0.013) | 1.092*** (0.031) | 1.013 (0.010) | 1.006 (0.007) | 0.888** (0.047) |
Occupancy | 0.931 (0.065) | 0.934 (0.116) | 0.998 (0.077) | 0.996 (0.068) | 0.852** (0.057) | 0.924 (0.089) |
Medicaid occupancy | 1.119** (0.053) | 1.216* (0.120) | 1.483*** (0.153) | 1.284** (0.150) | 1.419*** (0.132) | 1.228*** (0.090) |
Competition | 0.933* (0.037) | 0.940*** (0.017) | 0.913*** (0.030) | 0.893*** (0.025) | 0.967** (0.016) | 1.000 (0.016) |
Unemployment rate | 0.955 (0.047) | 0.978 (0.021) | 1.062 (0.042) | 1.015 (0.032) | 1.010 (0.045) | 0.999 (0.059) |
Log NA staffing | 0.935* (0.037) | 0.954*** (0.012) | 0.932*** (0.024) | 0.940*** (0.017) | 0.913*** (0.030) | 0.478*** (0.106) |
Log LPN staffing | 0.833* (0.035) | 1.121 (0.099) | 0.944 (0.042) | 0.856*** (0.043) | 0.925** (0.033) | 0.867** (0.049) |
Log RN staffing | 0.682* (0.025) | 0.771** (0.083) | 0.836* (0.082) | 0.878** (0.051) | 0.960 (0.063) | 0.844*** (0.052) |
Log NA staffing squared | 0.835 (0.162) | 0.842*** (0.053) | 0.988 (0.022) | 0.975 (0.062) | 0.929* (0.038) | 0.950 (0.075) |
Log LPN staffing squared | 0.987 (0.043) | 1.014 (0.021) | 0.986 (0.018) | 0.998 (0.030) | 0.950** (0.022) | 0.984 (0.012) |
Log RN staffing squared | 0.969 (0.041) | 0.939 (0.106) | 0.869** (0.054) | 0.878 (0.100) | 0.888 (0.099) | 0.740 (0.102) |
Log NA agency staffing | 1.097** (0.033) | 1.081*** (0.030) | 1.052*** (0.016) | 1.075*** (0.016) | 1.054*** (0.010) | 1.055*** (0.021) |
Log LPN agency staffing | 1.021 (0.028) | 0.920 (0.149) | 1.231 (0.283) | 1.005 (0.281) | 1.434** (0.235) | 1.019 (0.192) |
Log RN agency staffing | 1.101*** (0.026) | 1.184* (0.114) | 0.970* (0.015) | 1.028** (0.014) | 0.732** (0.021) | 1.012 (0.030) |
Log NA agency staffing squared | 0.870** (0.040) | 0.998 (0.030) | 0.950** (0.022) | 0.984 (0.012) | 0.989 (0.014) | 1.006 (0.013) |
Log LPN agency staffing squared | 1.041 (0.137) | 1.058 (0.048) | 1.104 (0.093) | 1.050 (0.084) | 1.066 (0.043) | 1.029 (0.040) |
Log RN agency staffing squared | 1.069** (0.021) | 0.976 (0.045) | 1.050** (0.024) | 1.078** (0.035) | 1.058** (0.027) | 1.003 (0.035) |
Log NA Agency Staffing × NA Staffing | 1.065 (0.245) | 0.978 (0.091) | 1.027 (0.034) | 0.961 (0.084) | 1.144** (0.078) | 0.993 (0.063) |
Log LPN Agency Staffing × LPN Staffing | 1.048 (0.062) | 1.167 (0.139) | 1.187 (0.323) | 1.118 (0.103) | 1.027 (0.230) | 1.222 (0.201) |
Log RN Agency Staffing × RN Staffing | 0.859** (0.051) | 0.913*** (0.031) | 0.893*** (0.025) | 0.966** (0.016) | 0.969* (0.018) | 1.014 (0.018) |
Log NA stability | 0.914** (0.025) | 1.026 (0.100) | 0.876** (0.050) | 0.885** (0.048) | 0.891** (0.040) | 0.867** (0.062) |
Log LPN stability | 1.030 (0.019) | 0.978 (0.038) | 1.012 (0.013) | 0.975 (0.026) | 0.992 (0.023) | 0.965 (0.036) |
Log RN stability | 0.915 (0.064) | 0.917* (0.039) | 0.947 (0.038) | 0.922* (0.045) | 0.913 (0.051) | 1.038 (0.066) |
Log NA stability squared | 0.920** (0.029) | 0.940 (0.114) | 0.998 (0.070) | 0.997 (0.069) | 0.853** (0.057) | 0.926 (0.090) |
Log LPN stability squared | 0.744 (0.223) | 0.996 (0.129) | 0.986 (0.117) | 1.070 (0.118) | 0.986 (0.115) | 1.005 (0.118) |
Log RN stability squared | 0.917 (0.229) | 1.021 (0.114) | 0.998 (0.070) | 0.999 (0.067) | 0.853** (0.057) | 0.928 (0.089) |
Log NA Stability × NA Staffing | 0.861* (0.074) | 1.026 (0.100) | 0.862** (0.041) | 0.887** (0.048) | 0.891** (0.040) | 0.867** (0.062) |
Log LPN Stability × LPN Staffing | 1.031 (0.092) | 1.022 (0.046) | 0.995 (0.020) | 0.990 (0.025) | 1.025 (0.021) | 0.950* (0.028) |
Log RN Stability × RN Staffing | 0.818** (0.075) | 0.976* (0.014) | 0.991 (0.026) | 0.877** (0.050) | 0.853** (0.057) | 0.928 (0.089) |
Log NA Stability × Agency | 0.920*** (0.026) | 0.842*** (0.053) | 0.988 (0.022) | 0.975 (0.062) | 0.929* (0.038) | 0.950 (0.055) |
Log LPN Stability × Agency | 0.975 (0.122) | 0.957 (0.088) | 1.049 (0.060) | 0.915 (0.100) | 1.017 (0.097) | 1.009 (0.053) |
Log RN Stability Agency | 0.956 (0.061) | 0.939 (0.101) | 0.869** (0.059) | 0.878 (0.102) | 0.888 (0.098) | 0.769*** (0.058) |
Log professional staff mix | 0.889** (0.048) | 0.773** (0.082) | 0.836* (0.086) | 0.878** (0.053) | 1.060 (0.061) | 0.958 (0.040) |
Log professional staff mix squared | 0.981 (0.043) | 0.998 (0.034) | 0.957 (0.088) | 0.949 (0.060) | 0.915 (0.100) | 1.017 (0.097) |
Log Professional Staff Mix × Agency | 0.826* (0.035) | 0.881** (0.050) | 0.980 (0.055) | 1.028 (0.085) | 0.984 (0.064) | 0.950* (0.028) |
Log Professional Staff Mix × Stability | 0.923* (0.044) | 0.932*** (0.024) | 0.940*** (0.017) | 0.913*** (0.030) | 0.893*** (0.025) | 0.968** (0.016) |
Pseudo R2 | 0.24 | 0.26 | 0.27 | 0.29 | 0.25 | 0.21 |
Notes: Robust standard errors in parentheses. Note that dummy variable indicating source of primary data was included and was not significant. Bold print is used to highlight the staffing level variables. LSR = long-stay residents; SSR = short-stay residents; NA = nurse aide; LPN = licensed practical nurse; RN = registered nurse.
*p <.05; **p <.01; ***p <.001.
Variable . | (1) Percent Physical Restraint Use (LSR) . | (2) Percent With Moderate to Severe Pain (LSR) . | (3) Percent Low-Risk Residents With Pressure Sores (LSR) . | (4) Percent High-Risk Residents With Pressure Sores (LSR) . | (5) Percent Had a Catheter Inserted and Left in Bladder (LSR) . | (6) Percent With Moderate to Severe Pain (SSR) . |
---|---|---|---|---|---|---|
Organizational size (beds) | 1.087** (0.039) | 1.020* (0.011) | 1.010* (0.006) | 1.016 (0.010) | 1.051* (0.032) | 1.049 (0.056) |
Ownership (for profit = 1) | 1.220* (0.142) | 1.008 (0.029) | 1.039** (0.019) | 1.030*** (0.010) | 1.026 (0.017) | 1.032** (0.016) |
Chain membership (chain member = 1) | 0.922 (0.136) | 1.021* (0.013) | 1.092*** (0.031) | 1.013 (0.010) | 1.006 (0.007) | 0.888** (0.047) |
Occupancy | 0.931 (0.065) | 0.934 (0.116) | 0.998 (0.077) | 0.996 (0.068) | 0.852** (0.057) | 0.924 (0.089) |
Medicaid occupancy | 1.119** (0.053) | 1.216* (0.120) | 1.483*** (0.153) | 1.284** (0.150) | 1.419*** (0.132) | 1.228*** (0.090) |
Competition | 0.933* (0.037) | 0.940*** (0.017) | 0.913*** (0.030) | 0.893*** (0.025) | 0.967** (0.016) | 1.000 (0.016) |
Unemployment rate | 0.955 (0.047) | 0.978 (0.021) | 1.062 (0.042) | 1.015 (0.032) | 1.010 (0.045) | 0.999 (0.059) |
Log NA staffing | 0.935* (0.037) | 0.954*** (0.012) | 0.932*** (0.024) | 0.940*** (0.017) | 0.913*** (0.030) | 0.478*** (0.106) |
Log LPN staffing | 0.833* (0.035) | 1.121 (0.099) | 0.944 (0.042) | 0.856*** (0.043) | 0.925** (0.033) | 0.867** (0.049) |
Log RN staffing | 0.682* (0.025) | 0.771** (0.083) | 0.836* (0.082) | 0.878** (0.051) | 0.960 (0.063) | 0.844*** (0.052) |
Log NA staffing squared | 0.835 (0.162) | 0.842*** (0.053) | 0.988 (0.022) | 0.975 (0.062) | 0.929* (0.038) | 0.950 (0.075) |
Log LPN staffing squared | 0.987 (0.043) | 1.014 (0.021) | 0.986 (0.018) | 0.998 (0.030) | 0.950** (0.022) | 0.984 (0.012) |
Log RN staffing squared | 0.969 (0.041) | 0.939 (0.106) | 0.869** (0.054) | 0.878 (0.100) | 0.888 (0.099) | 0.740 (0.102) |
Log NA agency staffing | 1.097** (0.033) | 1.081*** (0.030) | 1.052*** (0.016) | 1.075*** (0.016) | 1.054*** (0.010) | 1.055*** (0.021) |
Log LPN agency staffing | 1.021 (0.028) | 0.920 (0.149) | 1.231 (0.283) | 1.005 (0.281) | 1.434** (0.235) | 1.019 (0.192) |
Log RN agency staffing | 1.101*** (0.026) | 1.184* (0.114) | 0.970* (0.015) | 1.028** (0.014) | 0.732** (0.021) | 1.012 (0.030) |
Log NA agency staffing squared | 0.870** (0.040) | 0.998 (0.030) | 0.950** (0.022) | 0.984 (0.012) | 0.989 (0.014) | 1.006 (0.013) |
Log LPN agency staffing squared | 1.041 (0.137) | 1.058 (0.048) | 1.104 (0.093) | 1.050 (0.084) | 1.066 (0.043) | 1.029 (0.040) |
Log RN agency staffing squared | 1.069** (0.021) | 0.976 (0.045) | 1.050** (0.024) | 1.078** (0.035) | 1.058** (0.027) | 1.003 (0.035) |
Log NA Agency Staffing × NA Staffing | 1.065 (0.245) | 0.978 (0.091) | 1.027 (0.034) | 0.961 (0.084) | 1.144** (0.078) | 0.993 (0.063) |
Log LPN Agency Staffing × LPN Staffing | 1.048 (0.062) | 1.167 (0.139) | 1.187 (0.323) | 1.118 (0.103) | 1.027 (0.230) | 1.222 (0.201) |
Log RN Agency Staffing × RN Staffing | 0.859** (0.051) | 0.913*** (0.031) | 0.893*** (0.025) | 0.966** (0.016) | 0.969* (0.018) | 1.014 (0.018) |
Log NA stability | 0.914** (0.025) | 1.026 (0.100) | 0.876** (0.050) | 0.885** (0.048) | 0.891** (0.040) | 0.867** (0.062) |
Log LPN stability | 1.030 (0.019) | 0.978 (0.038) | 1.012 (0.013) | 0.975 (0.026) | 0.992 (0.023) | 0.965 (0.036) |
Log RN stability | 0.915 (0.064) | 0.917* (0.039) | 0.947 (0.038) | 0.922* (0.045) | 0.913 (0.051) | 1.038 (0.066) |
Log NA stability squared | 0.920** (0.029) | 0.940 (0.114) | 0.998 (0.070) | 0.997 (0.069) | 0.853** (0.057) | 0.926 (0.090) |
Log LPN stability squared | 0.744 (0.223) | 0.996 (0.129) | 0.986 (0.117) | 1.070 (0.118) | 0.986 (0.115) | 1.005 (0.118) |
Log RN stability squared | 0.917 (0.229) | 1.021 (0.114) | 0.998 (0.070) | 0.999 (0.067) | 0.853** (0.057) | 0.928 (0.089) |
Log NA Stability × NA Staffing | 0.861* (0.074) | 1.026 (0.100) | 0.862** (0.041) | 0.887** (0.048) | 0.891** (0.040) | 0.867** (0.062) |
Log LPN Stability × LPN Staffing | 1.031 (0.092) | 1.022 (0.046) | 0.995 (0.020) | 0.990 (0.025) | 1.025 (0.021) | 0.950* (0.028) |
Log RN Stability × RN Staffing | 0.818** (0.075) | 0.976* (0.014) | 0.991 (0.026) | 0.877** (0.050) | 0.853** (0.057) | 0.928 (0.089) |
Log NA Stability × Agency | 0.920*** (0.026) | 0.842*** (0.053) | 0.988 (0.022) | 0.975 (0.062) | 0.929* (0.038) | 0.950 (0.055) |
Log LPN Stability × Agency | 0.975 (0.122) | 0.957 (0.088) | 1.049 (0.060) | 0.915 (0.100) | 1.017 (0.097) | 1.009 (0.053) |
Log RN Stability Agency | 0.956 (0.061) | 0.939 (0.101) | 0.869** (0.059) | 0.878 (0.102) | 0.888 (0.098) | 0.769*** (0.058) |
Log professional staff mix | 0.889** (0.048) | 0.773** (0.082) | 0.836* (0.086) | 0.878** (0.053) | 1.060 (0.061) | 0.958 (0.040) |
Log professional staff mix squared | 0.981 (0.043) | 0.998 (0.034) | 0.957 (0.088) | 0.949 (0.060) | 0.915 (0.100) | 1.017 (0.097) |
Log Professional Staff Mix × Agency | 0.826* (0.035) | 0.881** (0.050) | 0.980 (0.055) | 1.028 (0.085) | 0.984 (0.064) | 0.950* (0.028) |
Log Professional Staff Mix × Stability | 0.923* (0.044) | 0.932*** (0.024) | 0.940*** (0.017) | 0.913*** (0.030) | 0.893*** (0.025) | 0.968** (0.016) |
Pseudo R2 | 0.24 | 0.26 | 0.27 | 0.29 | 0.25 | 0.21 |
Variable . | (1) Percent Physical Restraint Use (LSR) . | (2) Percent With Moderate to Severe Pain (LSR) . | (3) Percent Low-Risk Residents With Pressure Sores (LSR) . | (4) Percent High-Risk Residents With Pressure Sores (LSR) . | (5) Percent Had a Catheter Inserted and Left in Bladder (LSR) . | (6) Percent With Moderate to Severe Pain (SSR) . |
---|---|---|---|---|---|---|
Organizational size (beds) | 1.087** (0.039) | 1.020* (0.011) | 1.010* (0.006) | 1.016 (0.010) | 1.051* (0.032) | 1.049 (0.056) |
Ownership (for profit = 1) | 1.220* (0.142) | 1.008 (0.029) | 1.039** (0.019) | 1.030*** (0.010) | 1.026 (0.017) | 1.032** (0.016) |
Chain membership (chain member = 1) | 0.922 (0.136) | 1.021* (0.013) | 1.092*** (0.031) | 1.013 (0.010) | 1.006 (0.007) | 0.888** (0.047) |
Occupancy | 0.931 (0.065) | 0.934 (0.116) | 0.998 (0.077) | 0.996 (0.068) | 0.852** (0.057) | 0.924 (0.089) |
Medicaid occupancy | 1.119** (0.053) | 1.216* (0.120) | 1.483*** (0.153) | 1.284** (0.150) | 1.419*** (0.132) | 1.228*** (0.090) |
Competition | 0.933* (0.037) | 0.940*** (0.017) | 0.913*** (0.030) | 0.893*** (0.025) | 0.967** (0.016) | 1.000 (0.016) |
Unemployment rate | 0.955 (0.047) | 0.978 (0.021) | 1.062 (0.042) | 1.015 (0.032) | 1.010 (0.045) | 0.999 (0.059) |
Log NA staffing | 0.935* (0.037) | 0.954*** (0.012) | 0.932*** (0.024) | 0.940*** (0.017) | 0.913*** (0.030) | 0.478*** (0.106) |
Log LPN staffing | 0.833* (0.035) | 1.121 (0.099) | 0.944 (0.042) | 0.856*** (0.043) | 0.925** (0.033) | 0.867** (0.049) |
Log RN staffing | 0.682* (0.025) | 0.771** (0.083) | 0.836* (0.082) | 0.878** (0.051) | 0.960 (0.063) | 0.844*** (0.052) |
Log NA staffing squared | 0.835 (0.162) | 0.842*** (0.053) | 0.988 (0.022) | 0.975 (0.062) | 0.929* (0.038) | 0.950 (0.075) |
Log LPN staffing squared | 0.987 (0.043) | 1.014 (0.021) | 0.986 (0.018) | 0.998 (0.030) | 0.950** (0.022) | 0.984 (0.012) |
Log RN staffing squared | 0.969 (0.041) | 0.939 (0.106) | 0.869** (0.054) | 0.878 (0.100) | 0.888 (0.099) | 0.740 (0.102) |
Log NA agency staffing | 1.097** (0.033) | 1.081*** (0.030) | 1.052*** (0.016) | 1.075*** (0.016) | 1.054*** (0.010) | 1.055*** (0.021) |
Log LPN agency staffing | 1.021 (0.028) | 0.920 (0.149) | 1.231 (0.283) | 1.005 (0.281) | 1.434** (0.235) | 1.019 (0.192) |
Log RN agency staffing | 1.101*** (0.026) | 1.184* (0.114) | 0.970* (0.015) | 1.028** (0.014) | 0.732** (0.021) | 1.012 (0.030) |
Log NA agency staffing squared | 0.870** (0.040) | 0.998 (0.030) | 0.950** (0.022) | 0.984 (0.012) | 0.989 (0.014) | 1.006 (0.013) |
Log LPN agency staffing squared | 1.041 (0.137) | 1.058 (0.048) | 1.104 (0.093) | 1.050 (0.084) | 1.066 (0.043) | 1.029 (0.040) |
Log RN agency staffing squared | 1.069** (0.021) | 0.976 (0.045) | 1.050** (0.024) | 1.078** (0.035) | 1.058** (0.027) | 1.003 (0.035) |
Log NA Agency Staffing × NA Staffing | 1.065 (0.245) | 0.978 (0.091) | 1.027 (0.034) | 0.961 (0.084) | 1.144** (0.078) | 0.993 (0.063) |
Log LPN Agency Staffing × LPN Staffing | 1.048 (0.062) | 1.167 (0.139) | 1.187 (0.323) | 1.118 (0.103) | 1.027 (0.230) | 1.222 (0.201) |
Log RN Agency Staffing × RN Staffing | 0.859** (0.051) | 0.913*** (0.031) | 0.893*** (0.025) | 0.966** (0.016) | 0.969* (0.018) | 1.014 (0.018) |
Log NA stability | 0.914** (0.025) | 1.026 (0.100) | 0.876** (0.050) | 0.885** (0.048) | 0.891** (0.040) | 0.867** (0.062) |
Log LPN stability | 1.030 (0.019) | 0.978 (0.038) | 1.012 (0.013) | 0.975 (0.026) | 0.992 (0.023) | 0.965 (0.036) |
Log RN stability | 0.915 (0.064) | 0.917* (0.039) | 0.947 (0.038) | 0.922* (0.045) | 0.913 (0.051) | 1.038 (0.066) |
Log NA stability squared | 0.920** (0.029) | 0.940 (0.114) | 0.998 (0.070) | 0.997 (0.069) | 0.853** (0.057) | 0.926 (0.090) |
Log LPN stability squared | 0.744 (0.223) | 0.996 (0.129) | 0.986 (0.117) | 1.070 (0.118) | 0.986 (0.115) | 1.005 (0.118) |
Log RN stability squared | 0.917 (0.229) | 1.021 (0.114) | 0.998 (0.070) | 0.999 (0.067) | 0.853** (0.057) | 0.928 (0.089) |
Log NA Stability × NA Staffing | 0.861* (0.074) | 1.026 (0.100) | 0.862** (0.041) | 0.887** (0.048) | 0.891** (0.040) | 0.867** (0.062) |
Log LPN Stability × LPN Staffing | 1.031 (0.092) | 1.022 (0.046) | 0.995 (0.020) | 0.990 (0.025) | 1.025 (0.021) | 0.950* (0.028) |
Log RN Stability × RN Staffing | 0.818** (0.075) | 0.976* (0.014) | 0.991 (0.026) | 0.877** (0.050) | 0.853** (0.057) | 0.928 (0.089) |
Log NA Stability × Agency | 0.920*** (0.026) | 0.842*** (0.053) | 0.988 (0.022) | 0.975 (0.062) | 0.929* (0.038) | 0.950 (0.055) |
Log LPN Stability × Agency | 0.975 (0.122) | 0.957 (0.088) | 1.049 (0.060) | 0.915 (0.100) | 1.017 (0.097) | 1.009 (0.053) |
Log RN Stability Agency | 0.956 (0.061) | 0.939 (0.101) | 0.869** (0.059) | 0.878 (0.102) | 0.888 (0.098) | 0.769*** (0.058) |
Log professional staff mix | 0.889** (0.048) | 0.773** (0.082) | 0.836* (0.086) | 0.878** (0.053) | 1.060 (0.061) | 0.958 (0.040) |
Log professional staff mix squared | 0.981 (0.043) | 0.998 (0.034) | 0.957 (0.088) | 0.949 (0.060) | 0.915 (0.100) | 1.017 (0.097) |
Log Professional Staff Mix × Agency | 0.826* (0.035) | 0.881** (0.050) | 0.980 (0.055) | 1.028 (0.085) | 0.984 (0.064) | 0.950* (0.028) |
Log Professional Staff Mix × Stability | 0.923* (0.044) | 0.932*** (0.024) | 0.940*** (0.017) | 0.913*** (0.030) | 0.893*** (0.025) | 0.968** (0.016) |
Pseudo R2 | 0.24 | 0.26 | 0.27 | 0.29 | 0.25 | 0.21 |
Notes: Robust standard errors in parentheses. Note that dummy variable indicating source of primary data was included and was not significant. Bold print is used to highlight the staffing level variables. LSR = long-stay residents; SSR = short-stay residents; NA = nurse aide; LPN = licensed practical nurse; RN = registered nurse.
*p <.05; **p <.01; ***p <.001.
References
Abt Associates. (
Anderson, R. A., Issel, L. M., & McDaniel, R. R., (
Arling, G., Kane, R. L., Mueller, C., Bershadsky, J., & Degenholtz, H. B., (
Banaszak-Holl, J., Berta, W., Bowman, D., Baum, J., & Mitchell, W., (
Barry, T., Brannon, D., & Mor, V., (
Bates-Jensen, B., (
Bates-Jensen, B. M., Schnelle, J. F., Alessi, C. A., Al-Samarrai, N. R., & Levy-Storms, L., (
Bostick, J. E., (
Bourbonniere, M., Intrator, O., Mor, V., Feng, Z., Angelelli, J., & Zinn, J., (
Bowers, B. J., Esmond, S., & Jacobson, N., (
Brannon, D., Streit, A., & Smyer, M., (
Castle, N. G., (
Castle, N. G, & Engberg, J., (
Castle, N. G., & Engberg, J., (
Castle, N. G., & Engberg, J., (
Decker, F. H., (
Dellefield, M. E., (
Dorr, D. A., Horn, S. D., & Smout, R. J., (
General Accounting Office. (
Harrington, C., Kovner, C., Kayser-Jones, J., Burger, S., Mohler, M., & Burke, R., et al (
Harrington, C., & Swan, J. H., (
Hillmer, M. P., Wodchis, W. P., Gill, S. S., Anderson, G. M., & Rochon, P. A., (
Hughes, C. M., Lapane, K. L., & Mor, V., (
Hunt, J., & Hagen, S., (
Kane, R. L., (
Mor, V., Berg, K., Angelelli, J., Gifford, D., Morris, J., & Moore, T., (
Moseley, C. B., & Jones, L., (
Mueller, C., Arling, G., Kane, R., Bershadsky, J., Holland, D., & Joy, A., (
Mukamel, D. B., Spector, W. D., & Bajorska, A., (
Needleman, J., Beurhaus, P. I., Mattke, S., Stewart, M., & Zelevinsky, K., (
O'Neill, C., Harrington, C., Kitchener, M., & Saliba, D., (
Porell, F., Caro, F. G., Silva, A., & Monane, M., (
Price, J. L., & Mueller, C. W., (
Schnelle, J. F., Simmons, S. F., Harrington, C., Cadogan, M., Garcia, E., & Bates-Jenson, B., (
Smyer, M., Brannon, D., & Cohn, M., (
Straker, J., (
Sullivan-Marx, E. M., Strumpf, N. E., Evans, L. K., Baumgarten, M., & Maislin, G., (