1993-95 PRIORITIES AND PROGRESS UNDER THE GREAT LAKES WATER QUALITY AGREEMENT

Chapter Two:Great Lakes Science Advisory Board

2.2 Human Health in Ecosystem Health: Issues of Meaning and Measurement

2.2.3 Environmental Burden of Illness

Criteria for the Evaluation of Epidemiological Studies Linking Environmental Toxicant Exposures and Health Effects

1. Basic design of study
   (a) What type of study was used (cohort, case-control, ecologic)?
   -strengths
   -weaknesses
   

2. Exposure assessment
   (a)Is the nature of the suspected exposure known?
   (b)Is the overall dose known?
   -timing and duration of exposure
   -route of exposure
   -body burden
   (c)Is a dose gradient known? How accurate is (are) the exposure category(ies)?
   (d)Were controls used? How accurate is the non-exposed (or non-diseased) classification?

3. Outcome assessment (measurement of health effect)
   (a)How appropriate to the particular exposure in question is the outcome being studied?
   -Does other human or animal evidence relate the health effect to suspected exposure? How strong is it?
   -Is the outcome assessment appropriately timed (latency period considered)?
   -Is the health effect examined validated as adversely affecting human health?
   (b) How accurate is the outcome assessment?
   -completeness (few false negatives)
   -correctness (few false positives)
   (c)Is there possible bias in the ascertainment of the health outcome for the various exposure category(ies) and controls?
   

4. Control for other factors influencing outcome
   (a)Are the exposed category(ies) - or cases, in a case control study - and controls comparable (except for exposure)?
   -nature of underlying populations
   -sampling bias
   (b)How great is the problem of confounders likely to be?
   -specificity of health outcome studied for the particular exposure
   (c)How successfully were possible confounders controlled?
   - adequacy of matching or adjustment of all possible confounders (age, sex, socio-economic status, ethnicity, other exposures to toxicants, access to medical care, secular time trends)

5. Strength of association between exposure and outcome (relative risk)
   (a) Does the relative risk have clinical or practical significance?
   (b) Does the relative risk have statistical significance?
   (c) Was a clearcut dose-response gradient demonstrated?
   (d) If no statistically significant relative risk exposure was found, was the statistical power of the study adequate to find a risk of practical importance if it existed?

6. Evaluation of final conclusion
   (a)If the result is positive, could it be a false positive association?
   (b)If the result is negative, could it be a false negative association?
   (c)Is the result consistent with other well-conducted studies of the same association and/or related epidemiological knowledge on the distribution and dynamics of the health outcome or condition in question?