Reading a report of an epidemiologic study
Suggestion: read through the paper entirely; then reread and answer these questions:
Thought questions
What is the primary objective of the study?
What is the central hypothesis?
What does the investigators' causal model look like?
What type of study design was used?
- Why do you think it was selected?
- What are the general strengths and weaknesses of the type of study conducted?
- What strengths and weaknesses can you identify in this particular study?
How were subjects identified?
- Why was this method chosen?
- Was the method different for unexposed/exposed or cases/controls? Why? In what ways?
- How successful was enrollment?
- How successful was retention?
How was the outcome of interest defined?
- What criteria were applied?
- What means were used to avoid "missing" events?
How was the exposure of interest defined?
- What criteria were applied?
- How was the exposure measured?
- What do we know about the quality of the exposure measurement?
What are the potential sources of error?
- Random error?
- Selection bias?
- Information bias?
- Confounding?
- Are these sources differential or nondifferential?
In what ways could the concerns you identified in 8 & 9 influence the findings?
- Bias towards or away from the null hypothesis?
Is there an association observed?
- How strong is it?
- How precise is the estimate?
- How would you state/interpret the results?
Where do the findings agree or diverge from those described or alluded to in the introduction and/or discussion?
How strong is the case for causality?
What are the public health implications of the study?
Checking out the numbers.
- Familiarize yourself with all figures and tables.
- Calculate the crude risk ratio or odds ratio.
- Where strata are provided, practice calculating stratified RR/OR.
- Understand what models were used in the analysis (i.e., logistic, proportional hazards, etc.)
- What variables were included? Why?
- How did adjustment affect the primary result? What does this mean?
For an examination
- Be able to succinctly define relevant methodologic terms.
- What "type(s)" of data were collected? How was the information deployed?
From "Preparing for the Final: Getting to Know the Paper", by Katherine Hartmann, M.D., Ph.D., EPID 168 lab instructor, 1997 (currently Director, North Carolina Program for Women's Health Research,
Cecil G. Sheps Center
for Health Services Research).
Minor edits and formatting by Victor Schoenbach, 3/18/2001.
3/18/2001, Victor J. Schoenbach
|