We matched patients on age and sex. Is that enough?

Question

A researcher cannot run a randomized trial, so instead compares patients who took a new drug with untreated patients, carefully matching each treated patient to an untreated one of the same age and sex. The treated group does better, and the researcher claims the drug works.

Why does matching on age and sex fall short of what randomization provides?

Accepted Answer

Matching only balances what you match on. By pairing patients on age and sex, the researcher removes those two factors as explanations. But health outcomes depend on much more: genetics, diet, exercise, smoking, disease severity, income, and how sick a patient looked to the doctor who decided to prescribe. If sicker or healthier patients were more likely to receive the drug, that imbalance survives the matching, because you did not match on it. These unmeasured differences are confounders hiding in plain sight. The core limitation. You can only adjust for, or match on, variables you actually …

We matched patients on age and sex. Is that enough?

Your answer

Discussion

Learn the concepts

Related questions