Assessment of model fit for primary efficacy endpoints

Title card for the presentation: Assessment of model fit for primary efficacy endpoints

Complete the form to view this video

You may need to disable pop-up blocker

Abstract: 'Assessment of model fit for primary efficacy endpoints'.

This study is a randomised, placebo-controlled, 12-week double blind study to assess the efficacy and safety of PAINFREE. The efficacy endpoints evaluated the mean change in the severity of muscle spasms from baseline and the mean change in the frequency of muscle spasms from baseline. Subjects recorded the number and severity of each muscle spasm via the ePRO daily diary. These primary endpoints were analysed using a mixed model repeated measures analysis of covariance. An unstructured covariance structure was used to model the within-patient errors. To assess the model fit for these endpoints, diagnostic plots were generated using Proc Mixed and interpreted.

The diagnostic plots for the severity endpoint suggested the occurrence of a ceiling effect due to the nature of the data, since severity can only take values of 1, 2 or 3. The primary frequency endpoint plots identified a strong skew and several outliers. Splitting the diagnostic plots by day/night-time found that regarding time of day there was no difference in trend. It was previously identified that there were subjects with high frequencies, so whilst assessing model fit we also wanted to investigate whether these subjects also recorded high post baselines. By plotting patient profiles of the subjects with extreme outliers we identified 'normal' baseline measurements and then a large increase in frequency in later weeks. As a result, we have followed up with the study team to determine whether the potential outlier is a data entry issue. Findings from this assessment highlight the importance of checking model fit.

Learn more about our services

We are experts in study design, statistical analysis, data science, data capture and reporting for clinical trials