Epidemiology and Biostatistics: Measures, Design, and Inference
This domain tests the logic of study design and the interpretation of data. Know the measures of frequency and association, the strengths and weaknesses of each design, bias, confounding, and the meaning of confidence intervals and p-values.
Concepts to master
- Measures: incidence, prevalence, relative risk, odds ratio, attributable risk, and number needed to treat.
- Study designs: cohort, case-control, cross-sectional, randomized trial; when each is appropriate.
- Bias types: selection, information, recall, and confounding; how randomization and matching address them.
- Statistical inference: p-values, confidence intervals, type I and II errors, power.
- Diagnostic test statistics: sensitivity, specificity, likelihood ratios, and ROC curves.
Traps
Questions often mislabel an odds ratio as a relative risk or confuse incidence with prevalence. Another trap is treating statistical significance as clinical importance. Be careful with case-control studies: they estimate odds ratios, not incidence. Watch for confounding presented as a causal finding, and for questions that ask which bias is present rather than how to fix it.
Worked rule of thumb
To choose a measure: cohort or trial → relative risk and attributable risk; case-control → odds ratio; cross-sectional → prevalence. To interpret a confidence interval, ask whether it includes the null value (1 for ratios, 0 for differences) and how wide it is. A wide interval means imprecision, not absence of effect.