pharmacoepidemiology

To avoid immortal time bias in observational studies, align time zero for all cohort members and use an incident-user design. This skill provides a structured framework to identify and correct such common pitfalls in real-world data analysis.

Target trial emulation is a causal inference framework that maps observational data to a hypothetical randomized experiment. This skill uses it to ensure drug safety and effectiveness studies are scientifically defensible by explicitly defining eligibility, treatment strategies, and outcomes.

Propensity score matching balances cohorts by pairing treated and untreated subjects with similar covariate distributions. This skill supports propensity score modeling alongside weighting and stratification to achieve standardized balance diagnostics for valid causal effect estimation.

Yes, evaluating active comparators is supported to minimize confounding by indication when comparing two treatments. This skill helps critique proposed study designs to ensure they correctly account for time-varying confounders before running any analysis code.

The best way to address unmeasured confounding is applying rigorous sensitivity analysis frameworks to test how robust effect estimates are to hidden biases. This skill provides a structured methodological framework to evaluate these biases in observational real-world data.