Objectives To compare partial likelihood and pseudolikelihood approaches in a nested case-control study under countermatching design in the presence of time varying-covariates.

Methods Within a prospective cohort of 17603 French steel workers, cases were newly diagnosed with a histologically-confirmed carcinoma of the bladder in 2006–2012. Three controls per case matched on age at diagnosis were randomly selected following a countermatched sampling scheme using a four-strata surrogate time-varying exposure to metalworking fluid (MWF) mists covariate, assessed by a job-exposure matrix. Cases (n=84) and controls (n=251) provided information during questionnaire interviews concerning smoking history and occupational exposures history that were compiled by experts to assess occupational exposure to straight MWFs, soluble MWFs, synthetic MWFs and polycyclic aromatic hydrocarbons (PAHs) not included in MWFs. Three quantitative time-varying metrics were used in the models: the duration, the frequency-weighted duration and the cumulative exposure index. Cox models were fitted with the standard partial likelihood approach using only the 3 countermatched controls for each case as well as by maximizing the pseudolikelihood which uses all the controls sampled for each case with calibrated weights.

Results Compared to the standard method, in the pseudolikelihood approach there is a reduction in the variance of the estimates for straight MWFs, synthetic MWFs and smoking but an increase for soluble MWFs and PAH. The hazard ratios that were > 1 with the standard method were attenuated when considering the pseudolikelihood approach. The hazard ratios for the straight MWFs in the pseudolikelihood approach were 1.10 (95% CI: 1.01 – 1.19) per year of exposure and 1.33 (95% CI: 1.05 – 1.7) per full-time equivalent year of exposure.

Conclusion Nested case-control study under countermatching design would benefit from pseudolikelihood approach. Results from the current study suggest that occupational exposure to straight MWFs increases the risk of bladder cancer.