Charles Poole, University of North Carolina
The case-control study design, broadly defined, is a valuable tool for estimating effect measures more efficiently and, in some contexts, more validly than in cohort studies. For many associations that have become widely accepted as causal, all or nearly all of the epidemiologic evidence has come from case-control studies.
A method of estimating comparative rates from clinical data: applications to cancer of the lung, breast, and cervix.
Cornfield J. A method of estimating comparative rates from clinical data: applications to cancer of the lung, breast, and cervix. J Natl Cancer Inst 1951;11:1269-1275.
This paper’s main purpose was to show how to estimate risks, prevalences and rates in studies known today as case-control studies, an almost lost art. Its secondary purpose, for which it is much better known, was to show how to estimate incidence odds ratios and prevalence odds ratios if cases are excluded from the control group, and thereby to estimate risk ratios and prevalence ratios approximately if the disease is rare.
Kupper LL, McMichaels AJ, Spritas R. A hybrid epidemiologic study design useful in estimating relative risk. J Am Statist Assoc 1975;70:524-528.
This paper showed how to estimate risk and prevalence ratios in case-control studies without a rare-disease requirement by allowing cases into the control group.
Miettinen O. Estimability and estimation in case-referent studies. Am J Epidemiol 1976;103:226-236.
This paper showed how to use density control sampling to estimate rate ratios in case-control studies.
Pearce N. What does the odds ratio estimate in a case-control study? Int J Epidemiol 1993;22:1189-1192.
This paper gives a short, readable summary of the use of case-control studies to estimate risk and prevalence ratios, incidence and prevalence odds ratios, and rate ratios.
Berkson J. Limitations of the application of fourfold table analysis to hospital data. Biometrics Bull 1946;2:47-53.
This paper showed, decades ahead of its time, how hospital-based case-control studies can be severely biased, even to the point of reversing the actual association and even when the exposure does not affect the control-defining outcome(s).
Rothman KJ, Greenland S, Lash TL. Case-control studies. Chapter 8 in: Modern Epidemiology. Third edition. Lippincott Williams & Wilkins, 2008;111-128.
This chapter is an up to date compendium of theory and methods of the case-control design and many of its nuances and variants. (The important and widely misunderstood topic of matching in case-control studies is covered on pages 175-181 of the same book.)