Strong effects of definition and nonresponse bias on prevalence rates of clinical benign prostatic hyperplasia: the Krimpen study of male urogenital tract problems and general health status

Objective To estimate the prevalence of benign prostatic hyperplasia (BPH) in the community, and study the influence of BPH definition, age and response bias on prevalence rates. Subjects and methods A community-based longitudinal study of 3924 men aged 50-75 years was conducted in a Dutch municipality (Krimpen) near Rotterdam. Data from those responding were collected using self-administered questionnaires, and during visits to the health centre and outpatient clinic of the urology department. The questionnaires included symptom scores on general well being (Inventory of Subjective Health, ISH) and lower urinary tract symptoms (International Prostate Symptom Score, IPSS). A short version of the questionnaire (including the IPSS and ISH) was sent to a random sample of those not responding. All subjects participating fully underwent a physical examination, uroflowmetry, transrectal ultrasonometry of the prostate and had their prostate specific antigen level measured. Age-specific prevalence rates of BPH were estimated using different definitions, based on one or more of symptom severity, prostate volume and maximum flow rate. The influence of response bias was estimated using the questionnaires. Results The response rate was 50% (full participants). Of those not responding, 55% completed a short version of the questionnaire (partial participants). Compared with full participants, partial participants had a lower IPSS and slightly lower ISH. The prevalence rates of clinical BPH in the study population was 9-20% (95% confidence interval, 8-11% to 22-27%) depending on the definition used. After adjusting for nonresponse bias, the age-group specific prevalences for 5-year age strata were 1.1-1.8 times lower for all BPH definitions used. Conclusions The prevalence rates of clinical BPH depend largely on the definition used and increase strongly with age. The effect of age is stronger when more variables are included in the definition. Adjustment for response bias results in substantially lower prevalence rates.