Epidemiology & Prevention USMLE Step 1 Practice Question
A 15-year prospective cohort study enrolls 5,000 asymptomatic adults aged 40-60 years and follows them for cardiovascular disease (CVD) outcomes. Baseline measurements include total cholesterol, LDL, HDL, blood pressure, smoking status, physical activity, and alcohol consumption. During analysis, researchers examine the association between baseline LDL cholesterol and incident myocardial infarction while adjusting only for age and sex. They plan to analyze each risk factor independently without multivariable adjustment. Which of the following biases is most likely to distort their estimate of the LDL-CVD association?
Answer choices
- ADetection bias, because more intensive cardiac monitoring in high-cholesterol participants may increase CVD ascertainment
- BInformation bias, because self-reported smoking status at baseline may be inaccurately recalled over 15 years
- CConfounding by unmeasured or unadjusted variables, such as diet quality, family history, and inflammatory markers that correlate with both LDL and CVD riskCorrect answer
- DBerkson's paradox, because inclusion of asymptomatic baseline participants creates a collider stratification problem
- EReverse causality, because early subclinical CVD may lower cholesterol through metabolic effects
- FSelection bias, because healthier participants are more likely to complete the 15-year follow-up
See the full explanation
Get the correct-answer rationale, why each distractor is wrong, the underlying mechanism, and high-yield associations — plus adaptive practice that targets your weak areas — with a free MedBoardPRO account.