BackgroundRecent disproportionality analyses and nationwide case-control studies suggested a potential association between clozapine use and hematological malignancy (HM). Nevertheless, the absolute rate difference is unclear due to the absence of cohort studies.MethodsWe extracted data from a territory-wide public healthcare database in Hong Kong to build a retrospective cohort of anonymized patients aged 18+ with a diagnosis of schizophrenia who used clozapine or olanzapine (drug comparator with highly similar chemical structure and pharmacological mechanisms) for 90+ days, with at least two prior other antipsychotic use records within both groups. Weighted by inverse probability of treatment based on propensity scores, Poisson regression was used to estimate the incidence rate ratio (IRR) of HM between clozapine and olanzapine users. The absolute rate difference was also estimated.ResultsIn total, 9,965 patients were included, with 834 clozapine users, who had a significant IRR of 2.22 (95% CI 1.52, 3.34) for HM compared to olanzapine users. Absolute rate difference was estimated to be 57.40 (95% CI 33.24, 81.55) per 100,000 person-years. Findings were consistent across sub-groups by age and sex in terms of effect size, although the IRR was non-significant for those aged 65 or older. Sensitivity analyses all supported the robustness of the results and showed good specificity to HM but no other cancers.ConclusionAbsolute rate difference in HM incidence was very small although there is a twofold elevated rate. Pharmacotherapies with clozapine may consider this potential rare risk in addition to known side effects.