Study designs — MCQs

Question 1: A research team has data from three completed studies on statin use and Alzheimer's disease: Study A (case-control, OR=0.6, n=500), Study B (retrospective cohort, RR=0.7, n=10,000), and Study C (RCT with cognitive decline as secondary endpoint, RR=0.9, n=2,000). The case-control study used prevalent cases, the cohort study had significant loss to follow-up in the unexposed group, and the RCT was underpowered for cognitive outcomes. Synthesize the evidence to determine the most reliable conclusion about the association.

• A. The RCT provides the strongest evidence despite being underpowered
• B. The retrospective cohort study offers the best balance of validity and precision (Correct Answer)
• C. Case-control study is most reliable due to efficient rare outcome assessment
• D. Evidence is contradictory and no conclusion can be drawn
• E. Meta-analysis of all three studies provides the most accurate estimate

Explanation: ***The retrospective cohort study offers the best balance of validity and precision*** - Despite **differential loss to follow-up** in the unexposed, the very large **sample size (n=10,000)** provides high **statistical power** and precision compared to the other studies. - Cohort studies establish a **temporal sequence** (statin use before disease onset), which is superior to cross-sectional or prevalent case-control designs for causal inference. *The RCT provides the strongest evidence despite being underpowered* - While **Randomized Controlled Trials (RCTs)** are higher in the evidence hierarchy, an **underpowered** study lacks the precision to detect a true effect, increasing the risk of a **Type II error**. - Since cognitive decline was only a **secondary endpoint**, the study may not have been designed or followed long enough to assess Alzheimer's disease progression accurately. *Case-control study is most reliable due to efficient rare outcome assessment* - The use of **prevalent cases** instead of incident cases introduces **Neyman bias (survival bias)**, as it only includes patients who survived long enough to be studied. - Case-control studies are prone to **recall bias** and cannot definitively prove that statin use preceded the onset of cognitive decline. *Evidence is contradictory and no conclusion can be drawn* - While the results vary, researchers can still synthesize evidence by weighing studies based on their **methodological quality**, **sample size**, and **bias profile**. - A conclusion can be drawn by acknowledging the **conservative estimate** (bias toward the null) in the cohort study which still suggested a protective effect. *Meta-analysis of all three studies provides the most accurate estimate* - **Meta-analysis** of studies with fundamentally different designs (RCT, Cohort, Case-Control) and significant **methodological limitations** can lead to inaccurate pooled results ("garbage in, garbage out"). - Combining studies with different measures of association (OR vs RR) and distinct biases like **survival bias** and **differential attrition** increases **heterogeneity**.