Hematological Profiles During Long-Term Clozapine Treatment: A Retrospective Real-World Comparison

Background: Clozapine is effective in treatment-resistant schizophrenia, but hematological concerns contribute to underuse. Recent profiles among long-term continuers are poorly described. We compared repeated routine hematological measurements in long-term clozapine and other antipsychotic patients who had continued the same primary antipsychotic regimen for at least 5 years.Methods: This retrospective cross-sectional comparison included adults with schizophrenia-spectrum disorders who had remained on the same primary antipsychotic regimen for at least 5 years. We averaged each participant's three most recent routine hematological measurements obtained 1–3 months apart. Between-group comparisons included Hodges–Lehmann location-shift estimates with 95% confidence intervals and rank-based effect sizes. WBC and ANC were additionally examined using median regression adjusted for age, sex, and current smoking. White blood cell count (WBC) and absolute neutrophil count (ANC) were designated primary outcomes; other hematological indices, neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR) were secondary outcomes.Results: The analysis included 183 patients: 27 receiving clozapine and 156 receiving other antipsychotics. Median WBC was 7.52 ×10³/µL with clozapine and 7.70 ×10³/µL with other antipsychotics; the Hodges– Lehmann difference was 0.05 ×10³/µL (95% CI, −0.77 to 0.92; Cliff's δ = 0.014). Median ANC was 4.75 and 4.88 ×10³/µL, respectively, with a Hodges–Lehmann difference of −0.01 ×10³/µL (95% CI, −0.65 to 0.71; Cliff's δ = −0.005). After adjustment for age, sex, and current smoking, the estimated clozapine-group differences remained small for WBC (0.15 ×10³/µL; 95% CI, −1.00 to 1.31) and ANC (−0.09 ×10³/µL; 95% CI, −0.92 to 0.74).Conclusion: Among patients who had successfully continued an unchanged antipsychotic regimen for at least 5 years, recent WBC and ANC distributions were similar between treatment groups, with negligible point estimates, although the confidence intervals did not exclude small-to-moderate differences.