Software for Subjective Endpoints in Clinical Trials

Because many randomized clinical trials (RCTs) study several important outcome variables thatmay respond to treatment, evaluation of efficacy is often difficult. Objective algorithms that combineinformation from all relevant variables may not capture satisfactorily the total experience of patients.Therefore, a computer program that implements a flexible, subjective, and blinded ranking of patientsusing their entire medical history during the RCT has important commercial application for organizationsthat perform RCTs. This project describes a new procedure for endpoint determination. Raters, blindedto treatment assignment, independently rank outcomes of participants in the RCT. The procedure poolsranks across raters to determine a final "global outcome rank". A rank procedure, e.g., theMann-Whitney statistic, tests treatment effect. Phase I, which stresses applications to cardiovasculardisease, has three aims: (a) Investigation of feasibility of the approach by reanalyzing the SHEP (1) data.The rating assessments will incorporate different lengths of follow-up and varied severity of events.Phase I will (i) evaluate agreement between raters, (ii) study approaches for rating large numbers of RCTparticipants, and (iii) compare tests of treatment differences using subjective endpoints to standardmethods; (b) Consideration of general methodological issues related to statistical power in subjectiverankings; (c) Preliminary design of the prototype software package.