Limitations of epidemiologically based needs assessment. The case of prostatectomy

The aim of this study was to make epidemiologically based estimates of the prevalent and incident "need" for prostatectomy for lower urinary tract symptoms, defined as the numbers of men who would both benefit from and want the operation.

The methods involved a consensus panel, a two-stage postal survey of 1,480 men aged 55 years or older from eight general practices to the northwest of London, United Kingdom, and a multistate life table.

The overall response rate was 69% (initial survey: 78%, follow-up survey: 88%). A trial-based estimate of number of candidates for prostatectomy (men with symptoms that were at least moderately severe and bothersome and who would probably or definitely want surgery) was 610 men in a population of 250,000. The corresponding incidence estimate (including men with symptoms recurring after spontaneous remission or surgery) was approximately 200 per year, including approximately 110 new cases. Consensus-based estimation, including categories of patients who have not yet been subject to a trial, gave much higher figures of approximately 3,000, 650, and 200 candidates, respectively. Adding the number of men who said they were "inclined to" choose surgery would almost double these figures.

Estimates of need were highly sensitive to choice of indications and assumptions about patients' attitudes toward surgery. Population needs assessment for specific procedures will always involve judgment as well as epidemiological data and modeling.