CHAPTER 20 Getting the Hint from Epidemiologic Inference 297

Imagine making a model for the study of asbestos workers, cigarette smoking,

and lung cancer. The variable asbestos is coded 1 for workers exposed to asbestos

and 0 for workers not exposed to asbestos, and the variable smoker is coded 1 for

cigarette smokers and 0 for nonsmokers. The final model would already have

asbestos and smoker in it, so the interaction model would add the additional covari-

ate asbestos × smoker, which is called the higher order interaction term. For indi-

viduals who have a 0 for either asbestos or smoker or both, this term falls out of

their individual predicted probability (because 1 × 0 = 0, and 0 × 0 = 0). Therefore,

if this term is statistically significant, then individuals who qualify to include this

term in their individual predicted probability have a statistically significantly

greater risk of the outcome, and the interaction term should be kept in the model.

Getting Casual about Cause

Chapter 7 explains epidemiologic study designs and presents them in a pyramid

format. The closer to the top of the pyramid, the better the study design is at pro-

viding evidence for causal inference, meaning providing evidence of a causal

association between the exposure with the outcome (or in the case of a clinical

trial of an intervention, the intervention and the outcome). At the top of the

pyramid are systematic review and meta-analysis, where the results of similar

studies are combined and interpreted. Because systematic reviews and meta-

analyses combine results from other high-quality studies, they are at the very

top of the pyramid — meaning they provide the strongest evidence of a causal

association between the exposure or intervention and outcome.

An international organization called the Cochrane Collaboration organizes the

production of systematic reviews and meta-analyses to help guide clinicians.

Their reviews are internationally renowned for being high-quality and are avail-

able at www.cochrane.org.

The study designs on the evidence-based pyramid that could be answered with a

regression model include clinical trial, cohort study, case-control study, and

cross-sectional study. If in your final model your exposure is statistically signifi-

cantly associated with your outcome, you now have to see how much evidence you

have that the exposure caused the outcome. This section provides two methods by

which to evaluate the significant exposure and outcome relationship in your

regression: Rothman’s causal pie and Bradford Hill’s criteria of causality.

Rothman’s causal pie

Kenneth Rothman described how causes of an outcome are not determinate. In

other words, two people can have the same values of covariates and one will get