1. Reverse Epidemiology: A Confusing, Confounding, and Inaccurate Term
- Author
-
Friedrich K. Port, Nathan W. Levin, Josef Coresh, George A. Kaysen, and Garry J. Handelman
- Subjects
medicine.medical_specialty ,education.field_of_study ,business.industry ,Confounding ,Population ,Disease ,Clinical trial ,Endocrinology ,Nephrology ,Internal medicine ,Epidemiology ,medicine ,Causation ,Intensive care medicine ,Epidemiologic Factors ,business ,education ,Subclinical infection - Abstract
The term "reverse epidemiology" has been proposed to address the apparent different relationship between numerous risk factors and outcomes among dialysis patients: thus, obesity, hypertension, high cholesterol, and elevated creatinine all appear to be associated with decreased risk. Since this is contrary to the general findings in otherwise healthy populations, some kind of "reversal" has been suggested, that would be contrary to classical epidemiology. The authors describe several faults to this conception. The rules of epidemiology have not been reversed in dialysis patients. In fact, the complexity of the population implies a greater need for attention to the distinction between association and causation and the importance of confounding and bias. In particular existing subclinical and clinical disease which is very common among dialysis patients can change associations so drastically that they are dominated by different causal pathways than those seen in the general population. For example, lower cholesterol is a better marker of poor health than of a healthy diet and thus is associated with different outcomes. To the extent the term reverse epidemiology implies either epidemiology or biology is different in dialysis patients it can be misleading and detrimental. The differences between risk factors in end-stage renal disease (ESRD) and other individuals are surely important, but can themselves be the basis of excellent epidemiology, applied with the classic rules developed for this discipline with the goal of uncovering causal association and hypotheses to be tested in clinical trials.
- Published
- 2007
- Full Text
- View/download PDF