A pragmatic clinical trial (PCT) can pose challenges during both the study design phase and the analysis phase (Cook et al 2016). This chapter discusses several important statistical considerations for the analysis of PCTs, especially PCTs that use cluster randomization. (For topics related to study design, see the Experimental Designs and Randomization Schemes chapter.)

Key to the transparency and reproducibility of the analysis of any clinical trial is a prespecified statistical analysis plan (SAP). The SAP need not be a standalone document, and it should be read together with the trial protocol (Gamble et al 2017). For a pragmatic trial, the SAP should include the standard elements of a conventional trial SAP—including the background and rationale of the trial, the study methods, the statistical principles, the trial population, and the analysis—while also addressing the unique features of the trial’s real-world setting. These features include clearly defined estimands, secondary use of real-world data sources (such as electronic health records [EHRs]), measurement of implementation fidelity, and consideration of heterogeneity of treatment effects.

For a discussion of reporting guidelines relevant to pragmatic trials, see the Reporting to the Scientific Community section of the Living Textbook. Others have described the development of guidelines for the content of SAPs for cluster randomized trials (CRTs), including those with stepped-wedge and cluster-crossover designs (Hemming et al 2025), and have reviewed the quality of reporting of SAPs for CRTs (Thompson et al 2025).

The NIH Collaboratory’s Biostatistics and Study Design Core developed a Statistical Analysis Plan Template, which provides a starting point for research teams designing a new trial. The Table below links to real-world SAPs from the NIH Collaboratory Trials, which have a variety of trial designs.

Table. SAPs From the NIH Collaboratory Trials^a

Abbreviations: CRT, cluster randomized trial; IRGT, individually randomized group treatment; SAP, statistical analysis plan.
^a The SAPs listed in the table are from NIH Collaboratory Trials for which the primary results have been published. Learn more about the NIH Collaboratory Trials.