Discrimination between the ulcerous and the nonulcerous forms of interstitial cystitis by noninvasive findings.

Abstract

We determined whether the classification of interstitial cystitis cases into ulcer and nonulcer categories from cystoscopic findings could be corroborated with epidemiological data relating to demographics, risk factors, symptoms, pain and psychosocial factors. We surveyed 565 interstitial cystitis patients (111 with and 454 without ulcer), and from univariate analyses we found 10 noninvasive variables to be significantly associated with the presence of Hunner's ulcers. The 3 multivariate statistical methodologies of discriminant analysis, logistic regression and recursive partitioning were used to classify ulcer versus nonulcer cases from these variables. The retrospective and prospective performances from each methodology were also analyzed. Retrospectively, resubstitution error rates of all 3 methodologies were small, with overall misclassification rates of 19.1% with linear discriminant, 14.7% with logistic regression and 8.0% with recursive partitioning. Prospectively, the overall misclassification rates increased slightly to 20.0% with linear discriminant, 15.8% with logistic regression, and 15.9% with recursive partitioning. The classification of interstitial cystitis into distinct categories of Hunner's ulcer (classic interstitial cystitis) and nonulcer based on cystoscopic findings can be corroborated with epidemiological evidence. The 2 categories may represent different manifestations of the underlying disease pathophysiology.