Criterion-related validity of a diarrhea questionnaire in HIV-infected patients.

Clinical trials evaluating HIV-related diarrhea have used varied unidimensional end points to assess diarrhea severity. We hypothesized that a self-reported measure of diarrhea that assesses stool form, stool frequency, and diarrhea morbidity would accurately portray the severity of HIV-related diarrhea. During a clinical trial for HIV-related diarrhea, we evaluated the instrument among 17 patients, comparing survey results with objective measures of diarrhea morbidity recorded concurrently. The survey scores demonstrated consistently high Spearman correlations with nursing assessment of stool form (0.6693), observed stool frequency (0.7023), and cumulative stool weight (0.8216), all recorded over six days of intensive inpatient observation (P < 0.01 for each). Of the three components of the survey, only the stool form assessment, which uses pictorial representations of stool consistency, correlated significantly across all three objective measures (0.8069-0.8792). In demonstrating the concurrent, criterion-related validity of this survey, we found it helpful for evaluating HIV-related diarrhea and suggest its utility for HIV-seronegative subjects as well.