Ten studies met the inclusion criteria (2,657,365 patients, range 386 (RCT) to 1,491,060 (cohort study)). Four of the included studies were RCTs, two of which were open-label. Three RCTs were considered to be at high risk of bias (two due to substantial differential losses to follow-up and one due to insufficient reporting of methods and early termination). Six studies were observational, five were cohort studies and one was a case-control study. The case-control study was considered to be at high risk of bias due to inadequately defined cases and unrepresentative controls. Where reported, mean length of follow-up across all studies ranged from 2.4 to six years.
A total of 3,643 patients had newly diagnosed bladder cancer (overall incidence 53.1 per 100,000 person-years).
From three of the RCTs, incidence of bladder cancer was 101.0 per 100,000 person-years among those who used a thiazolidinedione and 65.5 per 100,000 person-years among those who did not. There appeared to be a non-significant increase in the risk of bladder cancer with exposure to any thiazolidinediones (RR 1.45, 95% CI 0.75 to 2.83; Ι²=2%; four RCTs). There was no association between use of pioglitazone (RR 2.36, 95% CI 0.91 to 6.13; one RCT) or rosiglitazone (RR 0.87, 95% CI 0.34 to 2.23; two RCTs) and bladder cancer.
The cohort studies showed significant associations with bladder cancer with any thiazolidinedione (RR 1.15, 95% CI 1.04 to 1.26; Ι²=0%; six studies) and for pioglitazone specifically (RR 1.22, 95% CI 1.07 to 1.39; Ι²=0%; three studies). Studies that evaluated dose-response relationships reported contradictory results.