The data sources used were MEDLINE, Excerpta Medica, and International
Pharmaceutical Abstracts databases, and the manufacturer. Bibliographies of articles obtained were reviewed. The keywords used were pentosanpolysulfate, pentosanpolysulfate
sodium, and pentosan. Inclusion criteria were blinded selection of English language, prospective, randomized, placebo-controlled comparative trials; > or = 8 weeks' duration; > or = 300 mg daily; adult humans with > or = 1 symptoms including
pain, urgency, frequency, and
nocturia; symptoms for > or = 12 months; normal urinalysis; negative findings for urine culture and cytology. Exclusion criteria were
hemorrhagic cystitis;
drug-, microbial-, or radiation-induced
cystitis;
carcinoma in situ; other influencing diseases. The outcome of success was defined as a > or = 50% decrease in
pain, urgency, frequency, and
nocturia. The number of successes was extracted by blinded investigators, treating withdrawals as failures. The percentage difference in success rates of
pentosan polysulfate and placebo, and the number needed to treat (NNT) were determined for each variable; P values and 95% confidence intervals (CIs) were determined for combined data. Homogeneity of effect was determined by calculating Q (chi-squared). Article quality was assessed using the Chalmers scale to determine if quality affected outcome. Effective inter-rater reliability was determined using Rosenthal's method. Significance was set at P < 0.05.
RESULTS: Four studies were included. Data were extracted from all four studies for
pain (n = 398), three for urgency (n = 306), two for frequency (n = 160), and one study for
nocturia (n = 106). The differences (95% confidence limits) were
pain: 16.6% (95% CI 8%, 25%), NNT = 7; urgency: 13.0% (1.0%, 25%), NNT = 7.5; frequency: 16.7% (2.3%, 31.1%), NNT = 6;
nocturia: -1% (-19.8%, 21.8%). P values from homogeneity tests were not significant. Mean quality scores were 63.8%, 48.1%, 50.4%, and 65.6%, respectively, in the four studies; the effective inter-rater reliability was 0.96. Results did not differ when weighted by quality score.
CONCLUSIONS: