In cohort data analysis, which scenario best motivates using Poisson regression rather than Cox regression?

• A. Cox regression for time-to-event with censoring and hazard ratios
• B. Poisson regression for rate data with counts and exposure time
• C. Logistic regression for binary outcomes
• D. Linear regression for continuous outcomes

Answer: (B)

Poisson regression is appropriate when your data consist of event counts accumulated over person-time, so you model the rate of events per unit time and include an offset for exposure time. In a cohort study, you often have counts of events and varying follow-up times across individuals, so Poisson regression lets you estimate how the event rate changes with predictors by modeling the log rate. This yields rate ratios that reflect differences in incidence per unit of time, properly accounting for how long each person was at risk. The key distinction from Cox regression is that Cox analyzes time-to-event data with censoring and yields hazard ratios, focusing on the timing of the first event rather than the overall event rate per time at risk. Therefore, when the interest is comparing event rates across groups with differing exposure times, Poisson regression is the natural choice. The other models fit different types of outcomes—binary outcomes for logistic regression and continuous outcomes for linear regression—rather than rates derived from counts over time.