Which strategy reduces confounding bias in a study?

• A. Randomization, stratification, and multivariable adjustment
• B. Increasing sample size alone
• C. Measuring exposure more accurately without adjusting
• D. Ignoring potential confounders

Answer: (A)

Confounding bias happens when another variable influences both the exposure and the outcome, making the observed relationship between exposure and outcome misleading. The strongest way to reduce this bias is to combine design and analysis approaches: randomization in studies assigns participants to groups by chance, so both known and unknown confounders tend to be balanced between groups; stratification divides participants into subgroups where the confounder is held constant, allowing comparisons that are not distorted by that confounder; and multivariable adjustment uses statistical models to control for the influence of measured confounders, estimating the exposure effect while holding those confounders constant. Increasing sample size alone improves precision but doesn’t ensure confounders are balanced. Measuring exposure more accurately helps with misclassification but doesn’t address confounding if other variables differ between groups. Ignoring potential confounders inherently leads to biased results. The combination of randomization, stratification, and multivariable adjustment most effectively reduces confounding bias.