DEVELOPMENT AND VALIDATION OF AN AUDIT INSTRUMENT - THE PROSTATE OUTCOMES QUESTIONNAIRE

Objective To develop a short, patient-based questionnaire for auditing the outcomes of treatment for benign prostatic hyperplasia (BPH) and to evaluate the acceptability, reliability and validity of the measure . Patients and methods Data from a long research questionnaire used in the Oxford/North West Thames Prostatectomy Study were analysed to ide ntify the subset of items which contained the most scientifically soun d indicators of outcome. Items were selected on the basis of standard psychometric analyses to develop a short questionnaire, the Prostate O utcomes Questionnaire (POQ), a 27-item instrument covering urinary sym ptoms, complications after surgery, quality of life and patient satisf action with outcome. The POQ was field tested for acceptability, relia bility and validity in a postal survey of 125 men undergoing transuret hral resection of the prostate for BPH from seven hospitals in south-e ast England, Results The POQ was highly acceptable to patients, as ind icated by a 95% response rate, the absence of 'floor' and 'ceiling' ef fects and a low proportion of missing data. It showed excellent intern al consistency reliability (Cronbach's alpha of 0.82-0.93 for the five summary scales and a mean item-total correlation of 0.56) and test-re test reliability (test-retest correlations for items were 0.40-1.00, w ith a mean test-retest correlation of 0.85, and for summary scales of 0.91-0.95). Construct validity was confirmed by: high intercorrelation s between the five summary scales and the total score (0.88-0.95), wit h a pattern of intercorrelations among specific subscales that showed item-convergent and discriminant validity; higher scores for patients who reported an improvement after surgery than in those who were not i mproved (P < 0.001); the expected gradient of scores for patients repo rting different levels of symptom distress (P < 0.001); high correlati ons with the longer parent questionnaire (0.75-0.88); and low to moder ate correlations with the Nottingham Health Profile and Activities of Daily Living scores (0.15-0.67), with the pattern of correlations prov iding evidence of scale convergent and discriminant validity, Correlat ions with age and social class were low, suggesting that responses wer e not biased by sociodemographic factors, Conclusion The POQ is a prac tical and scientifically sound patient-based measure of outcome after treatment for BPH which can be used for routine audit. It takes < 5 mi n to complete, is feasible for routine monitoring of large numbers of patients by postal survey, and is accompanied by a Users' Manual which provides practical help in conducting a local patient survey, and a c omputer program for scoring data, Most importantly, the POQ has been s hown scientifically to perform well, having met standard psychometric criteria for reliability and validity.