Printer Friendly

Reply to Rohrer Commentary.

We appreciate Dr. Rohrer's interest in our article and thank him for his letter. Dr. Rohrer points out several limitations of utilization studies that have used the behavioral model: measurement error in the dependent variable, low explained variance, low response rates, exclusion of psychosocial variables, and complex sampling schemes.

Our response to his discussion of limitations is simple: we agree. These limitations in utilization studies and the application of the behavioral model have been discussed by various researchers over the past 20 years as well as by the originators of the model (Bass and Noelker 1987; Becker and Maiman 1983; Mechanic 1979; Pescosolido and Kronenfeld 1995; Rundall 1981; Tanner, Cockerham, and Spaeth 1983; Wolinsky 1978; Andersen 1995; Aday and Awe 1997).

Our article identifies new dimensions that have significant importance and relevance for studies of healthcare utilization in the future: the organizational and environmental context of healthcare utilization. Our primary disagreement with Dr. Rohrer is with his proposed solutions--to both the old and newly identified limitations of the model. Although we heartily encourage any efforts to improve the understanding of utilization behavior, we are afraid that the solutions proposed by Dr. Rohrer only touch the surface.

First, the suggestion to classify subjects as high users (versus not high users) or as nonusers (versus users) will only deal with recall bias if it can be assumed that recall is more accurate for high use versus low use and if users can be accurately grouped into dichotomous categories. Although this approach may be appropriate for some research questions, many research questions of interest do require specific information on the amount and timing of use so that dichotomous classification of the dependent variable would be inappropriate. Extensive methodological work has been conducted that applies the principles of cognitive psychology to survey questionnaire design to minimize recall bias (Bradburn, Rips, and Shevell 1987; Herrmann 1994; Jobe and Mingay 1989; Jobe, Tourangeau, and Smith 1993; Jobe, White, Kelley, et al. 1990; Sudman, Bradburn, and Schwarz 1995). Efforts to improve the accuracy of survey self-reports should be instituted at the beginning of the study, when the survey questions are being de veloped and tested, in order to avoid analytic compromises at the end in addressing the investigators' primary and intended research aims.

Further, using logistic regression does not make "R-square" a non-issue. The important issue is how well the model fits the data--and simply using logistic regression rather than linear regression does not absolve researchers from examining how well their model fits the data. (For a discussion of diagnostic approaches for logistic regression see Hosmer and Lemeshow 1989.) The use of R-square is only one approach to examining model fit, and its limitations have been well documented (Kennedy 1985).

Second, we agree that conducting utilization studies at the level of a single primary care market would hold constant some contextual variables. However, this statement misses the whole point of our article: that some important research questions require the understanding of how utilization varies across different systems, environments, and communities. Furthermore, there is still likely to be variation across health plans and providers within a single market--and it is this type of variation that is of great interest in the current healthcare environment. New analytic methods are required to fully capture these added and important institutional and environment levels of effect on individuals' patterns of healthcare utilization.

Third, we do not believe that our article states a "need to develop perfectly formulated models" and "exact representations of reality." We stated in the article that the behavior model "does not dictate (italics added) the precise variables and methods that must be used," and that "the appropriateness of the inclusion of environmental or provider-related variables and the use of more complex methods will vary (italics added) depending on the extent of prior research, the research question, the purpose of the study, and the availability of data." Perhaps there is some confusion here about the difference between precision in regression models--the goal of obtaining accurate estimates--and misspecification--the goal of including the appropriate variables in the model. It is the latter that is the focus of our article, because if the variables that are appropriate to the research question are not included in analyses, then precision becomes a moot point.

Dr. Rohrer's fourth point, that "the health behavior literature has almost forgotten the behavioral model because its users have failed to incorporate psychological variables," does not adequately acknowledge the extensive research that has been conducted using the model, incorporating psychosocial correlates, such as stress and social support, among others (Bass and Noelker 1987; Coulton and Frost 1982; Counte and Glandon 1991; Freedman 1993; Mutran and Ferraro 1988). The model accommodates and, in fact, some applications of it require the inclusion of these and related variables as, for example, in studies of healthcare for vulnerable populations, such as elderly people with chronic illness.

In response to Dr. Rohrer's last comment, we are not sure why he reads our article as promoting rigor at the expense of relevance. We believe that our article promotes both relevance--by encouraging studies that examine variables that are of interest from a programmatic and policy perspective--and rigor--by encouraging the use of previously underutilized methods and more appropriate analytic approaches.

To close, we are reminded of the old East Indian legend about the blind men and their first encounter with an elephant (John Godfrey Saxe). Each felt a different part of the elephant and proclaimed that he alone knew what the creature was. Our article examined previously unexplored limitations of the behavioral model of healthcare utilization and offered new ways of responding to them. Dr. Rohrer focused on previous limitations of the model and challenged our ways of addressing the newly revealed ones. Perhaps the truth lies, as it did with the blind men and the elephant: that in examining and dissecting the parts, we must not lose sight of the whole--the importance of critically examining the relevance and rigor of a long-standing model of healthcare utilization in a changing healthcare environment.

Kathryn A. Phillips, Ph.D. is Associate Professor of Health Economics and Health Services, University of California, San Francisco, School of Pharmacy and Institute for Health Policy Studies; Kathleen R. Morrison, M.P.H. is a Doctoral Student, Dept. of Health Services, School of Public Health, University of California, Los Angeles; Ronald Andersen, Ph.D. is Wasserman Professor, Dept. of Health Services, School of Public Health, University of California, Los Angeles; and Lu Ann Aday, Ph.D. is Professor of Behavioral Sciences and Management and Policy Sciences, School of Public Health, University of Texas, Houston.


Aday, L. A., and W. C. Awe. 1997. "Health Services Utilization Models." In Handbook of Health Behavior Research, Vol. I, Personal and Social Determinants, edited by D. S. Gochman. New York: Plenum.

Andersen R. 1995. "Revisiting the Behavioral Model and Access to Medical Care: Does it Matter?" Journal of Health and Social Behavior 36(1): 1-10.

Bass, D. M., and L. S. Noelker. 1987. "The Influence of Family Caregivers on Elder's Use of In-home Services: An Expanded Conceptual Framework." Journal of Health and Social Behavior 28(2): 184-96.

Becker, M., and L. Maiman. 1983. "Models of Health-related Behavior." In Handbook of Health, Health Care and the Health Professions. New York: Free Press.

Bradburn, N. M., L.J. Rips, and S. K. Shevell. 1987. "Answering Autobiographical Questions: The Impact of Memory and Inference on Surveys." Science 236 (4798): 157-61.

Coulton, C., and A. K. Frost. 1982. "Use of Social and Health Services by the Elderly." Journal of Health and Social Behavior 23 (4): 330-39.

Counte, M. A., and G. L. Glandon. 1991. "A Panel Study of Life Stress, Social Support, and the Health Services Utilization of Older Persons." Medical Care 29 (4): 348-61.

Freedman, V. A. 1993. "Kin and Nursing Home Lengths of Stay: A Backward Recurrence Time Approach." Journal of Health and Social Behavior 34 (2): 138-52.

Herrmann, D. 1994. The Contributions of the NCHS Collaborative Research Program to Memory Research. Report No. 14, Cognitive Methods Staff Working Paper Series. National Center for Health Statistics, Hyattsville, MD.

Hosmer, D. W., and S. Lemeshow. 1989. Applied Logistic Regression. New York: Wiley.

Jobe, J.B., and D.J. Mingay. 1989. "Cognitive Research Improves Questionnaires." American Journal of Public Health 79 (8): 1053-55.

Jobe, J. B., R. Tourangeau, and A. F. Smith. 1993. "Contributions of Survey Research to the Understanding of Memory." Applied Cognitive Psychology 7 (7): 567-84.

Jobe, J. B., A. A. White, C. L. Kelley, D.J. Mingay, M.J. Sanchez, and E. F. Loftus. 1990. "Recall Strategies and Memory for Healthcare Visits." The Milbank Quarterly 68 (2): 171-89.

Kennedy P. 1985. A Guide to Econometrics. Cambridge, MA: The MIT Press.

Mechanic, D. 1979. "Correlates of Physician Utilization: Why Do Multivariate Studies of Physician Utilization Find Trivial Psychosocial and Organizational Effects?" Journal of Health and Social Behavior 20 (4): 387-96.

Mutran, E., and K. Ferrarro. 1988. "Medical Need and Use of Services Among Older Men and Women." Journal of Gerontology 43 (Supplement): S162-S171.

Pescosolido, B. A., and J. J. Kronenfeld. 1995. "Health, Illness, and Healing in an Uncertain Era: Challenges from and for Medical Sociology." Journal of Health and Social Behavior (Extra Issue): 5-33.

Rundall, T. G. 1981. "A Suggestion for Improving the Behavioral Model of Physician Utilization." Journal of Health and Social Behavior 22 (1): 103-104.

Sudman, S., N. M. Bradburn, and N. Schwarz. 1995. Thinking About Answers: The Application of Cognitive Processes to Survey Methodology. San Francisco: Jossey-Bass.

Tanner, J. L., W. C. Cockerham, and J. L. Spaeth. 1983. "Predicting Physician Utilization." Medical Care 21(3): 360-69.

Wolinsky, F. D. 1978. "Assessing the Effects of Predisposing, Enabling, and Illness-Morbidity Characteristics on Health Service Utilization." Journal of Health and Social Behavior 19 (4): 384-96.
COPYRIGHT 2000 Health Research and Educational Trust
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2000 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Phillips, Kathryn A.; Morrison, Kathleen R.; Andersen, Ronald; Aday, Lu Ann
Publication:Health Services Research
Date:Feb 1, 2000
Previous Article:Rigor at the Expense of Relevance Equals Rigidity: Where to Next in the Study of Medical Care Utilization?
Next Article:The Impact of Utilization Management on Readmissions Among Patients with Cardiovascular Disease.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters