Ad-hoc or post-hoc analyses are typically performed after database lock, after one has already seen the planned analyses. These additional analyses are not part of the statistical analysis plan, and are usually driven by the observed unblinded study results. As such, they could be biased. Their value is not as strong as the planned analyses, but serve a more exploratory purpose. They can be very useful to help the study team gain a deeper understanding of the data. If used in the study report, such analyses should be clearly marked as “ad-hoc” or “post-hoc” so that the reader knows it is something that was added at a later moment.
Examples of such exploratory analyses are: subgroup analyses, correlations between endpoints, additional categorizations of already available scores, or trying other statistical techniques than the ones specified in the protocol or SAP. The Emtex statistician can help you identify any analyses of interest, provide the additional analysis results, and discuss the meaning of these results.
It is actually a good idea to already discuss any additional analyses of interest when the SAP is created. The SAP Outline document discussions are the ideal moment to identify such additional analyses.