| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||
a Public Health Sciences, Wichita State University, KS
Correspondence: James H. Swan, PhD, Wichita State University, Public Health Sciences, Box 152, 1845 N. Fairmont Street, Wichita, KS 67260-0152. E-mail: swan{at}chp.twsu.edu.
The topic of staff, particularly aide, turnover is of highest importance and currently the subject of intense interest. In this issue of The Gerontologist, Brannon and colleagues (Brannon, Zinn, Mor, and Davis 2002
) make some interesting points and report some important results of exploratory research on the topic. The authors suggest a new approach to the issue, and their results appear to validate this divergence. However, I have some issues with the article in two areas: (a) choice of and interpretation of cut-off points in the spectrum of staff turnover, and (b) issues of factors in the authors' model that are possibly endogenous with each other.
Brannon and colleagues are to be commended for considering possible qualitative differences represented by differing levels of turnover, but I am troubled by the ways in which they deal with these issues. The authors note that there are no benchmarks for appropriate and inappropriate turnover. Nevertheless, a purely statistical approach to setting cutting points does not promise a solution. Issues for further research in this area could be: (a) seeking expert consensus on the meanings of differing turnover rates, (b) seeking facility administrator opinion on turnover rates, (c) analyzing the financial implications for facilities of varying turnover rates, (d) analyzing the relation of turnover rates to indicators of quality. Analysis based purely on statistically driven cutting points (as used by Brannon and colleagues) is indeed exploratory and preliminary, pending the setting of benchmarks for staff turnover rates.
Brannon and colleagues make the interesting argument that some turnover levels may be too low, just as many are too high. This is based on cogent points that some level of turnover may be healthy by eliminating unqualified or abusive staff and that organizational gains may ensue from such tactics as substitution of cheaper for more costly labor. The authors argue that this will result in some level of turnover. The case of an unhealthily low rate of turnover is not really made, however. Despite involuntary factors, hiring decisions are still the responsibility of the facility, and better staff selection is clearly preferable to enhanced elimination of problematic staff. Moreover, given the low pay and low morale in nursing facilities, replacing higher-paid with lower-paid staff can more correctly be seen as a tragedy than as a legitimate organizational tactic.
It is not that no turnover rate can be too low, but it is rather difficult to determine what might be too low for different facilities. The authors argue that "the optimal rate for a given organization ... is almost always greater than zero" (p. 160). If true, it also follows that optimal rates differ among facilities. The authors cite "unfavorable selection ratios" (p. 161) in the labor market as a reason for necessary turnover, but they do not follow up by trying to account for such ratios. Might not the healthy rate of turnover be associated with measures of the availability of labor? This certainly suggests issues for further research. Moreover, the cutting point used here is way too high to generally represent a level that is too low. Although a 10th percentile rule does provide for a level assuring that this group of facilities "was unequivocally composed of low-turnover nursing facilities" (p. 164), I suggest that a 6-month turnover rate of 5% to 7% may be considered quite healthy, not bordering on unhealthily low. Furthermore, whatever level of turnover may be healthy for facilities, the overwhelming evidence is that the vast preponderance of facilities suffer from rates that are too high, not too low. More attention needs to be given to defining levels that are unhealthily high.
However, although there are problems with establishing a cutting point for low turnover, the findings certainly establish Brannon and colleagues' contention that the predictors at the high end of the turnover scale may differ from those lower on the scale. Future research will need to address the issue of nonlinear and even nonmonotonic relationships.
The other major issue is that some factors may be endogenous. A variety of measures are employed to explain turnover, some of which may be endogenous with it. For example, the unionization measure may be endogenous, in that both prevailing turnover rates and the conditions that spawn them may influence the likelihood of unionization, as well as reflect its existence. With low turnover, aide staff may know one another quite well and so band together more tightly. With very high turnover, aide staff may unionize to obtain job security and/or to fight conditions that may cause the high turnover. Likewise, the emergency-room (ER) admission rate may also be endogenous with turnover. Since the relationship is positive with high turnover, this raises the possibility that high aide turnover leads to more ER admissions. The authors acknowledge this and note that this analysis is preliminary. However, the issue remains. It is one that calls for further research and, in the meantime, necessitates restraint in interpreting results involving predictors that may also be influenced by turnover. Another complication is that registered nurse and aide turnover may be endogenous with one another. Staffing measures for different types of caregivers tend to be highly intercorrelated, presenting major challenges for modeling and analysis.
So, what does Brannon and colleagues' article contribute? Of greatest importance are findings that the predictors of low turnover may differ considerably from those of high turnover. One can appreciate the importance of a nonmonotonic response of turnover to predictors without accepting that there may be such a thing as turnover "too low in terms of quality of care and financial outcomes" (p. 168). Certainly this article shows the need for defining what is high and what is low. Moreover, although factors such as unionization (linked to low turnover but not shown to be linked to high turnover) may be endogenous and greatly complicate interpretation, they do not reduce the import of the finding.
Future research might: (a) consider nonmonotonic responses of aide turnover to predictors, (b) develop and refine benchmarks for high and low turnover, and (c) model possibly endogenous relationships among factors important to the analysis. Brannon and colleagues are correct that simple linear-regression techniques may be inadequate to the exploration of the important issue of nursing facility turnover. What is also apparent is that detailed modeling of the relationships involved in such turnover, as well as better understanding of what is acceptable turnover, will be essential. Solutions to nursing-facility staffing problems, particularly turnover, are essential for the industry, regulators and payers, and researchers alike. Brannon and colleagues' article demonstrates that such solutions are likely to be at least as complex as the problems they address.
The Forum
Book Reviews
Practice Concepts
 |
References |
|---|
 TOP References |
|---|
| ||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
