A public health researcher in Delhi conducts a case-control study on the association between oral contraceptive use and venous thromboembolism (VTE) in women aged 20–45 years. Among 150 women with VTE, 60 reported prior OCP use; among 300 controls without VTE, 40 reported prior OCP use. The researcher calculates an odds ratio of 3.0 (95% CI: 1.8–5.0, p < 0.001). Before presenting these findings to the state health authority, what is the most appropriate next step?

Explanation

## Study Design & Measure Interpretation The researcher has conducted a **case-control study** and calculated an **odds ratio (OR)**. Before translating this finding into public health action, critical appraisal is essential. ### Why Sensitivity Analysis & Confounder Assessment? **Key Point:** In case-control studies, the odds ratio estimates the relative risk only under specific conditions (rare disease assumption). However, confounding variables—such as age, smoking status, family history of thrombosis, and duration of OCP use—can inflate or deflate the observed association. **High-Yield:** The next step in any observational study showing a statistically significant association is **not** to act on it immediately, but to: 1. Identify potential confounders from the literature 2. Perform stratified analysis or multivariable adjustment 3. Conduct sensitivity analysis (e.g., exclude smokers, adjust for age groups) 4. Assess for effect modification 5. Only then draw causal inference ### Why Not the Other Options? | Option | Problem | |--------|----------| | **Report OR as RR directly** | OR ≠ RR in case-control studies unless the disease is rare. VTE prevalence in this age group is ~1–2 per 1000, so the rare disease assumption *may* hold, but this must be verified. Reporting without adjustment risks policy error. | | **Assume OR ≈ RR because CI excludes 1** | Statistical significance (p < 0.001) does NOT mean the association is causal or unconfounded. Confounding can produce a statistically significant spurious association. | | **Switch to cross-sectional design** | This is not a "next step"—it would require a new study. Case-control studies are valid for rare outcomes and are the appropriate design here. | ### Clinical Pearl **Confounding bias** is the most common threat to validity in observational studies. A classic example: the association between OCP use and VTE may be confounded by smoking (smokers may be more likely to use OCPs in some populations, and smoking independently increases VTE risk). Failing to adjust can lead to overestimation of the OCP–VTE causal effect. **Mnemonic: STROBE** — Strengthening the Reporting of Observational Studies in Epidemiology. Before acting on observational findings, check for: **S**election bias, **T**emporal sequence, **R**esidual confounding, **O**bservational (not experimental), **B**ias assessment, **E**ffect modification. **Tip:** In NEET PG / INI-CET, whenever you see an observational study result (especially case-control or cohort), the next step is almost always "assess for confounders" or "perform sensitivity analysis," not "implement the finding immediately."