| Neeta V. Shenvi |
The SAS procedure GENMOD is often used to fit classical linear models to analyze longitudinal data such as patient-reported outcomes (PRO) measurements overtime. We used SAS procedure GENMOD to estimate responder rate (defined for each visit as 25- point improvement) over 48 weeks. The responder (yes/no) endpoint evaluated at one year post-baseline visit was analyzed for binary outcome. The model included terms for sex, age group, 4 treatment groups, two -way and three-way interactions. We show use of SAS template language to create a publication grade 4 panel plot to visualize model-based responder rates and 95% confidence interval. |