Antipsychotics and Risk of Neuroleptic Malignant Syndrome: A Population-Based Cohort and Case-Crossover Study

BACKGROUND: Neuroleptic malignant syndrome (NMS) is a rare and acute adverse drug reaction associated with antipsychotic therapy. However, few data on the risk and epidemiology of NMS are available. OBJECTIVES: The aim of this study was to ascertain the incidence risk and all-cause mortality of NMS associated with antipsychotic use, and to assess the association of recent antipsychotic exposure and NMS. METHODS: We did a population-based study using data from the Hong Kong Hospital Authority’s Clinical Data Analysis and Reporting System database. Cases had a first diagnosis of NMS between 1 January 2004 and 30 November 2017. A case-crossover analysis was used to compare antipsychotic exposure 30 days before the diagnosis of NMS (index date) and a reference period 91–120 days before the index date. To adjust for potential time trends in antipsychotic exposure, we sampled from cases to match current cases and future cases, and further adjusted for select medications and acute medical conditions. RESULTS: 297,647 patients were prescribed antipsychotics, and the incidence risk of NMS was 0.11%. Of the 336 cases included in the case-crossover analysis, 20 (6%) died within 30 days after the index date; only one case had NMS recorded as the primary cause of death. When compared with the reference period, cases were more frequently prescribed multiple antipsychotics (15.8% vs 26.8%; standardized mean difference [SMD] 0.27) and short-acting injectable antipsychotics (3.6% vs 13.7%; SMD 0.37) during the 30 days prior to the diagnosis of NMS. Odds ratios for antipsychotic exposure in the case-crossover, case-crossover adjusted for time trend, and case-crossover adjusted for time trend and potential confounders analysis were 8.00 (95% confidence interval 3.42–18.69), 5.88 (2.46–14.04), and 4.77 (1.95–11.66). CONCLUSIONS: Our results suggest that recent use of antipsychotics is associated with NMS. Although a case-only design inherently controls for confounding by time-invariant factors, residual confounding by acute medical conditions with similar presentations to NMS cannot be fully excluded