An epidemiologist is planning a study to investigate the association between maternal folic acid supplementation and neural tube defects (NTD) in a low-resource setting in rural India. She has identified 300 women who delivered infants with NTD and 300 women who delivered unaffected infants, and will collect data on folic acid use during pregnancy. However, recall bias is a concern because mothers of affected infants may over-report or under-report supplementation differently than mothers of unaffected infants. What is the most appropriate next step to minimize this bias?

Explanation

## Study Design and Bias Identification **Key Point:** This is a **case-control study** with a major threat: **differential recall bias** — mothers of affected infants (cases) may recall or report exposure differently than mothers of unaffected infants (controls). This is a **design-stage intervention** question. ## Why Objective Records Are the Best Solution **High-Yield:** Using objective, independent records (antenatal clinic cards, pharmacy records) to verify self-reported data is the **gold standard** for minimizing recall bias in case-control studies. These records are collected prospectively and are not influenced by the outcome (NTD status). **Clinical Pearl:** In maternal–fetal epidemiology, differential recall bias is particularly problematic because mothers of affected infants are more motivated to remember details and may unconsciously distort their recall. Objective verification bypasses this psychological bias entirely. ## Comparison of Bias-Reduction Strategies | Strategy | Mechanism | Effectiveness | |----------|-----------|----------------| | **Objective records** | Eliminates recall; uses prospectively-collected data independent of outcome | **Highest** — gold standard | | **Increase sample size** | Does not reduce bias; only increases precision around a biased estimate | **None** — bias persists | | **Switch to cohort** | Eliminates recall bias but is more expensive, time-consuming, and impractical in low-resource settings | High, but **not feasible** here | | **Exclude poor recallers** | Introduces **selection bias** — systematically excludes a subset of the population | **Harmful** | ## Why Other Options Fail **Warning:** A common misconception is that increasing sample size reduces bias. **Sample size affects precision (confidence intervals), NOT accuracy (bias).** A large biased study is still biased. **Tip:** When recall bias is identified in a case-control study, the next step is always to **validate self-report with objective records** — not to change the study design or exclude participants. [cite:Park 26e Ch 10]