Change in HbA1c and weight after initiation of antipsychotic treatment in people with severe mental illness in England: an individual-level interrupted time-series analysis

Abstract

Background Second-generation antipsychotics contribute to excess cardiometabolic morbidity in people with severe mental illness, but the timing of early glycaemic change and whether it parallels weight gain remain unclear. We aimed to evaluate glycated haemoglobin A 1c (HbA 1c ) and weight trajectories before and after initiation of antipsychotic treatment. Methods We conducted an individual-level interrupted time-series analysis in England using electronic health records (EHR) from the Clinical Practice Research Datalink Aurum database between Jan 1, 2000, and June 7, 2023. Included participants were aged 18–59 years with severe mental illness (SMI) who had at least a second oral second-generation antipsychotic prescription issued within 90 days from their first antipsychotic prescription. Individuals were excluded if they had dementia, type 2 diabetes, or antipsychotic prescriptions before age 18 years. The exposure was the first recorded oral initiation of antipsychotic treatment. We used individual-level, interrupted time-series mixed models with linear splines to estimate rates of change in HbA 1c and weight in people with SMI before treatment initiation, 0–90 days after initiation, and from 91 days to 2 years after initiation. An additional outcome was time to type 2 diabetes diagnosis or antidiabetic medication prescription. Analyses were stratified by sex, antipsychotic drug, and baseline glycaemic status. People with lived experience contributed to the research question, study approach, interpretation, and dissemination. Findings Of the 96 849 eligible individuals, 53 558 (55·3%) were male and 43 291 (44·7%) were female. The mean age of participants was 37·2 years (SD 11·1; range 18–59); 62 683 (64·7%) reported their ethnicity as White, 5952 (6·1%) as south Asian, 6736 (7·0%) as Black, 3793 (3·9%) as other or mixed ethnicity, and 17 693 (18·3%) had missing ethnicity. 74 233 (76·6%) individuals had at least one recorded weight measurement and 24 084 (24·9%) had at least one HbA 1c measurement. In the pre-treatment period, HbA 1c increased at a rate of 0·01 mmol/mol per week (95% CI 0·00–0·01) and weight at a rate of 0·02 kg (0·02–0·02) per week. In the 0–90 days following antipsychotic treatment initiation, HbA 1c increased at a rate of 0·12 mmol/mol per week (0·09–0·15) and weight at 0·21 kg (0·20–0·22) per week. From day 91 to 2 years, the rate of increase of HbA 1c returned to pre-treatment levels (0·01 mmol/mol per week [0·00–0·01]) and weight increased at 0·03 kg (0·02–0·03) per week. Increases were largest for olanzapine (HbA 1c 0·18 mmol/mol per week [95% CI 0·12–0·25]; weight 0·28 kg [0·27–0·30]) and generally higher among women (HbA 1c 0·14 mmol/mol [0·09–0·18]; weight 0·22 kg [0·21–0·23]). Quetiapine was associated with weight gain (0·19 kg per week [95% CI 0·18 to 0·21]), but minimal HbA 1c change (0·02 mmol/mol [–0·03 to 0·08]). Among 918 individuals with baseline prediabetes, 126 (13·7%)developed type 2 diabetes, corresponding to an estimated 2-year cumulative incidence of 19·0%, compared with 152 (1·7%) of 9171 individuals with normal baseline HbA 1c , corresponding to an estimated 2-year cumulative incidence of 3·0%. Interpretation Antipsychotic treatment initiation was followed by rapid HbA 1c and weight increases in the first 90 days followed by slower increases up to 2 years. Imperfect concordance between HbA 1c and weight supports baseline and 12-week monitoring of both outcomes, followed by risk-stratified monitoring and intervention. Funding National Institute for Health and Care Research (NIHR) Three Research Schools Mental Health Programme, NIHR School for Primary Care Research, NIHR Applied Research Collaboration West Midlands, NIHR Dementia and Neurodegeneration Policy Research Unit, UK Research and Innovation, University College London Hospitals NIHR Biomedical Research Centre, and NIHR North Thames Applied Research Collaboration.

The authors' abstract, as published at the source. The Lancet Psychiatry, 2026 · DOI ↗