Question

Confounding and Effect Modification 30. A case-control study with 500 cases and 250 controls was conducted to look at the relationship between blood clots and oral contraceptives. Researchers suspect that smoking cigarettes may confound the relationship between blood clots and birth control pills. The crude odds ratio was 3.25. You stratify your 2x2 tables by smoking status and find that the stratum-specific odds ratio for smokers is 5.25, and the stratum-specific odds ratio for non- smokers is .67. Based on these results, was smoking a confounder, an effect modifier or neither? How do you know? How would you report the results?

Answer 1

In epidemiology, a variable is a confounder if it is associated with both the exposure and the outcome and if it distorts the apparent relationship between the exposure and the outcome. An effect modifier, on the other hand, is a variable that modifies the effect of the exposure on the outcome, meaning that the relationship between the exposure and the outcome differs depending on the level of the effect modifier.

Given the information provided, smoking appears to be an effect modifier rather than a confounder. Here's why:

1. Confounding: If smoking was a confounder, we would expect the crude odds ratio (which does not adjust for smoking) to be noticeably different from both the stratum-specific odds ratios for smokers and non-smokers. However, the crude odds ratio (3.25) is not a simple average or within the range of the stratified odds ratios for smokers (5.25) and non-smokers (0.67). This suggests that smoking does not confound the relationship between oral contraceptive use and blood clots.

2. Effect Modification: The stratum-specific odds ratios (5.25 for smokers and 0.67 for non-smokers) are dramatically different. This suggests that the effect of oral contraceptives on the risk of blood clots differs depending on whether the individual is a smoker or a non-smoker. In other words, smoking modifies the effect of oral contraceptives on the risk of blood clots, which is the definition of an effect modifier.

In reporting these results, it would be important to clearly state that smoking appears to modify the effect of oral contraceptives on the risk of blood clots. Specifically, among smokers, oral contraceptive use seems to substantially increase the risk of blood clots (with an odds ratio of 5.25), while among non-smokers, oral contraceptive use may actually decrease the risk (with an odds ratio of 0.67).