American Journal of Epidemiology Advance Access originally published online on July 11, 2007
American Journal of Epidemiology 2007 166(4):391-392; doi:10.1093/aje/kwm196
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Schootman et al. Respond to "Diabetes Causality in African Americans"
1 Department of Medicine and Pediatrics, Washington University School of Medicine, St. Louis, MO
2 North Florida/South Georgia Veterans Health System, Gainesville, FL
3 Department of Epidemiology and Biostatistics, College of Public Health and Health Professions, University of Florida, Gainesville, FL
4 Center for Research in the Implementation of Innovative Strategies in Practice, VA Iowa City Health Care System, Iowa City, IA
5 Department of Health Management and Policy, College of Public Health, The University of Iowa, Iowa City, IA
6 Department of Psychiatry, School of Medicine, St. Louis University, St. Louis, MO
7 Division of Biostatistics, Washington University School of Medicine, St. Louis, MO
8 Department of Surgery, Washington University School of Medicine, St. Louis, MO
9 Indiana University Center for Aging Research and Regenstrief Institute, Inc., Indiana University School of Medicine, Indianapolis, IN
Correspondence to Dr. Mario Schootman, Division of Health Behavior Research, Washington University, 4444 Forest Park Boulevard, Box 8504, St. Louis, MO 63108 (e-mail: mschootm{at}im.wustl.edu).
Received for publication May 29, 2007. Accepted for publication May 31, 2007.
We appreciate the thoughtful comments by Grant (1) on our paper (2). We agree that further exploration of the reasons for the observed association, using interdisciplinary research, is needed. We also agree that there are shortcomings of our pathways in that additional variables can be hypothesized to explain the association between housing conditions and diabetes development. Here, we briefly respond to the issues raised by Grant in his commentary.
Grant suggested that housing conditions may be a more sensitive measure of relative poverty. If this were true, the association between housing conditions and diabetes incidence would have been meaningfully altered by adjusting for perceived income adequacy in the logistic model, which it was not. Contrary to Grant's assertions, exercise and medication adherence were included as potential mediators in our analysis; both did not explain our findings. Clearly, additional variables (e.g., eating patterns) might have better captured the given pathways, as Grant suggested.
We agree with Grant's recommendation for greater interdisciplinary collaboration in an effort to better understand potentially mediating pathways. We further suggest not to limit this to quantitative methods but to include qualitative approaches as well. A triangulation approach, whereby quantitative and qualitative methods study the same problem, provides for a powerful means of data analysis and interpretation (3, 4).
As epidemiologists, we have often been trained to identify independent risk factors for the development of a particular disease, taking into account confounding variables. Our study used a different approach by exploring the mediating processes by which housing conditions affect diabetes development. Mediation and confounding are identical statistically and can be distinguished on only conceptual grounds (5). Moving the field of contextual effects forward will require more sophisticated approaches than controlling for confounders; it begs for examination of the causal processes by which such constructs affect disease occurrence (6). The standard analytical techniques traditionally used in observational epidemiology may not allow for full exploration of mediating pathways, especially when a more complex interplay between mediating pathways is hypothesized, as suggested by Grant.
According to Grant, one implication of our study is the potential presence of a family-level subculture. He further suggested that a reasonable next step would be to implement a home-based intervention addressing family-related attitudes in high-risk homes. Although ultimate certainty about the effect of family-related attitudes on development of diabetes can be obtained by a randomized intervention only, we think that this step would be premature since we did not measure this concept in our study. At this point, it is unclear whether this concept is the mediator that could explain the association observed.
Lastly, the conceptualization of neighborhood-based studies needs more attention. Developing ways to better investigate the adverse effects of neighborhood and housing conditions requires better understanding of confounders, mediators, and moderators/effect modification at the level of individual as well as at higher levels of aggregation.
 |
ACKNOWLEDGMENTS |
|---|
This research was supported by grants from the National Institutes of Health (AG10436, DK067172, DK20579).
Conflict of interest: none declared.
|
References |
|---|
 TOP References |
|---|
- Grant RW. Invited commentary: untangling the web of diabetes causality in African Americans. Am J Epidemiol (2007) 166:388–90.
[Abstract/Free Full Text] - Schootman M, Andresen EM, Wolinsky FD, et al. The effect of adverse housing and neighborhood conditions on the development of diabetes mellitus among middle-aged African Americans. Am J Epidemiol (2007) 166:379–87.
[Abstract/Free Full Text] - de Vries H, Weijts W, Dijkstra M, et al. The utilization of qualitative and quantitative data for health education program planning, implementation, and evaluation: a spiral approach. Health Educ Q (1992) 19:101–15.[ISI][Medline]
- Lindsay AC. Integrating quantitative and qualitative methods to assess the impact of child survival programs in developing countries: the case of a program evaluation in Ceara, Northeast Brazil. Health Educ Behav (2002) 29:570–84.
[Abstract/Free Full Text] - MacKinnon DP, Krull JL, Lockwood CM. Equivalence of the mediation, confounding and suppression effect. Prev Sci (2000) 1:173–81.[CrossRef][Medline]
- Diez Roux AV. The study of group-level factors in epidemiology: rethinking variables, study designs, and analytical approaches. Epidemiol Rev (2004) 26:104–11.
[Free Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
Related articles in Am. J. Epidemiol.:
- The Effect of Adverse Housing and Neighborhood Conditions on the Development of Diabetes Mellitus among Middle-aged African Americans
- Mario Schootman, Elena M. Andresen, Fredric D. Wolinsky, Theodore K. Malmstrom, J. Philip Miller, Yan Yan, and Douglas K. Miller
Am. J. Epidemiol. 2007 166: 379-387.[Abstract] [FREE Full Text]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||