Browse by author
Lookup NU author(s): Professor Azfar ZamanORCiD
This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).
© Author(s) (or their employer(s)) 2025. Re-use permitted under CC BY. Published by BMJ Group.Objective To develop prediction models for short-term outcomes following a first acute myocardial infarction (AMI) event (index) or for past AMI events (prevalent) in a national primary care cohort. Design Retrospective cohort study using logistic regression models to estimate 1-year and 5-year risks of all-cause mortality and composite cardiovascular outcomes. Setting Primary care practices in England contributing data to the Clinical Practice Research Datalink (CPRD) Aurum and CPRD GOLD databases between 2006 and 2019. Participants Patients with an incident (index) or prevalent AMI event. Models were trained on a random 80% sample of CPRD Aurum (n=1018 practices), internally validated on the remaining 20% (n=255) and externally validated using CPRD GOLD (n=248). Outcome measures Discrimination assessed using sensitivity, specificity and area under the receiver operating characteristic curve (AUC). Calibration assessed using calibration plots. Results In the index (prevalent) cohorts, 94241 (64 789) patients were included in the training and internal validation sets, and 16832 (7479) in the external validation set. For the index cohort, AUCs for 1-year [5-year] all-cause mortality were 0.802 (95% CI 0.793 to 0.812) [0.847 (0.841 to 0.853)] internally and 0.800 (0.790 to 0.810) [0.841 (0.835 to 0.847)] externally. For the primary composite outcome (stroke, heart failure and all-cause death), AUCs were 0.763 (0.756 to 0.771) [0.824 (0.818 to 0.830)] internally and 0.748 (0.739 to 0.756) [0.808 (0.801 to 0.815)] externally. Discrimination was higher in the prevalent cohort, particularly for 1-year mortality (AUC: 0.896, 95%CI 0.887 to 0.904). Models excluding treatment variables showed slightly lower but comparable performance. Calibration was acceptable across models. Conclusions These models can support clinicians in identifying patients at increased risk of short-term adverse outcomes following AMI, whether newly diagnosed or with a prior history. This can inform monitoring strategies and secondary prevention and guide patient counselling on modifiable risk factors.
Author(s): Kontopantelis E, Zghebi SS, Arsene CT, Zaman AG, Chew NWS, Wijeysundera HC, Khunti K, Ashcroft DM, Carr M, Parisi R, Mamas MA
Publication type: Article
Publication status: Published
Journal: BMJ Open
Year: 2025
Volume: 15
Issue: 12
Online publication date: 05/12/2025
Acceptance date: 03/11/2025
Date deposited: 16/12/2025
ISSN (electronic): 2044-6055
Publisher: BMJ Publishing Group
URL: https://doi.org/10.1136/bmjopen-2024-094961
DOI: 10.1136/bmjopen-2024-094961
Data Access Statement: Data may be obtained from a third party and are not publicly available. This study is based on data from the Clinical Practice Research Datalink (CPRD), obtained under licence from the UK Medicines and Healthcare products Regulatory Agency. The authors do not own the data and are not permitted to share it. Access to CPRD data is subject to approval and licensing and is not freely available. Researchers may apply for access via the CPRD website (https:// www.cprd.com)
PubMed id: 41360457
Altmetrics provided by Altmetric
