A five-year risk prediction model of cardiovascular disease in individuals with bipolar disorder: a nationwide register study from Sweden
Open Access
- 19 December 2025
- journal article
- research article
- Published by Springer Nature in Molecular Psychiatry
- Vol. 31 (5) , 2489-2497
Abstract
Cardiovascular disease (CVD) risk prediction models for the general population may not provide accurate predictions in individuals with bipolar disorder (BD) who have elevated risks of cardiometabolic conditions and premature mortality. Therefore, we aimed to: 1) develop a five-year CVD risk prediction model in this population by using nationwide register data from Sweden, 2) investigate whether the performance improved when we considered additional risk factors, including psychiatric comorbidity, psychotropic medication, and socio-demographic variables, compared to using established CVD risk factors only, and 3) whether machine learning approach provided improvements compared to standard logistic regression models. We followed 33,933 persons with BD aged 30–82 years old, without previous CVD, from the date of BD diagnosis registered between 2007–2014, for up to five years. The logistic regression model containing only established risk factors yielded an area under the receiver operating characteristic curve (AUC) of 0.76 (95% confidence interval 0.74–0.78) in the test dataset, while the logistic regression model and the best performing machine learning model including additional predictors yielded similar results (AUC was 0.77 (0.75, 0.79) in both models). The performance of logistic regression models slightly improved with additional predictors when continuous risk scores were used. In conclusion, standard logistic regression and established CVD risk factors may be sufficient to predict CVD in individuals with BD when using population register-based data from Sweden. External validation across diverse healthcare settings and rigorous assessment of clinical impact will be crucial next steps before implementing these models in clinical practice.Keywords
This publication has 68 references indexed in Scilit:
- Diagnosis and Treatment of Bipolar DisorderJAMA, 2023
- The Akaike information criterion: Background, derivation, properties, application, interpretation, and refinementsWIREs Computational Statistics, 2019
- Cardiovascular Risk Prediction Models for People With Severe Mental IllnessJAMA Psychiatry, 2015
- Prevalence and Correlates of Bipolar Spectrum Disorder in the World Mental Health Survey InitiativeArchives of General Psychiatry, 2011
- Coronary Artery Calcium Score and Risk Classification for Coronary Heart Disease PredictionJAMA, 2010
- Multivariable Model‐Building2008
- Evaluating the added predictive ability of a new marker: From area under the ROC curve to reclassification and beyondStatistics in Medicine, 2007
- The new Swedish Prescribed Drug Register—Opportunities for pharmacoepidemiological research and experience from the first six monthsPharmacoepidemiology and Drug Safety, 2006
- MULTIVARIABLE PROGNOSTIC MODELS: ISSUES IN DEVELOPING MODELS, EVALUATING ASSUMPTIONS AND ADEQUACY, AND MEASURING AND REDUCING ERRORSStatistics in Medicine, 1996
- Index for rating diagnostic testsCancer, 1950