Article Impact Level: HIGH Data Quality: STRONG Summary of The American Journal of Cardiology, 210, 133–142. https://doi.org/10.1016/j.amjcard.2023.10.008 Krishna Gokhale et al.
Points
- The study aimed to assess the impact of incorporating repeatedly measured data for patients with type 2 diabetes mellitus (T2DM) on the predictive capability of the QRISK cardiovascular disease (CVD) risk assessment model.
- Using the IQVIA Medical Research Data United Kingdom primary care database, the research included 198,835 T2DM patients aged 25 to 85, not on statin treatment, and without pre-existing CVD.
- The analysis considered traditional and nontraditional risk factors and relevant treatments, and separate models were developed for males and females.
- The study found that including time-varying repeated measures only mildly improved CVD risk prediction for T2DM patients compared to the current practice standard.
- The research advocates further exploring additional covariates using time-varying data to enhance the accuracy of existing prediction models for CVD risk in T2DM patients.
Summary
The study aimed to assess whether incorporating abundantly available repeatedly measured data for patients with type 2 diabetes mellitus (T2DM) enhances the predictive capability of the QRISK cardiovascular disease (CVD) risk assessment model. Using the IQVIA Medical Research Data United Kingdom primary care database, the research focused on patients aged 25 to 85 with T2DM, not on statin treatment, and without pre-existing CVD. Traditional and nontraditional risk factors and relevant treatments were included in the analysis, and a Cox’s hazards model accounting for time-dependent covariates was utilized to estimate hazard rates for each risk factor and calculate a 10-year risk score. Separate models were developed for males and females, and the performance of the models was tested using validation data, with discrimination and calibration statistics calculated.
The study encompassed 198,835 patients, including 180,143 males with 11,976 outcomes and 90,466 females with 8,258 outcomes. The 10-year predicted survival probabilities for females and males were 0.87 and 0.84, respectively, with observed survival estimates closely aligning with the predictions. The Harrell’s C-index for all female and all male models was 0.71 and 0.69, respectively. Ultimately, the research concluded that including time-varying repeated measures only mildly improved CVD risk prediction for T2DM patients compared to the current practice standard. The study advocates for further research utilizing time-varying data to explore the potential enhancement of prediction model accuracy through the involvement of additional covariates.
In summary, the research underscores the limited improvement in CVD risk prediction for T2DM patients when incorporating abundantly available repeatedly measured data, emphasizing the need for continued exploration of additional covariates to enhance the accuracy of existing prediction models.
Link to the article: https://www.ajconline.org/article/S0002-9149(23)01143-8/fulltext
References
Gokhale, K. M., Chandan, J. S., Sainsbury, C., Tino, P., Tahrani, A., Toulis, K., & Nirantharakumar, K. (2024). Using Repeated Measurements to Predict Cardiovascular Risk in Patients With Type 2 Diabetes Mellitus. The American Journal of Cardiology, 210, 133–142. https://doi.org/10.1016/j.amjcard.2023.10.008
