Simultaneous derivation, validation, and comparison of predictor hazard ratios for cardiovascular risk prediction equations in patients with diabetes from high versus non-high income countries: cohort study

Abstract

Abstract Objective To directly test the generalisability of cardiovascular disease (CVD) risk predictor effects across populations in a high income versus non-high income country, derive comparable risk equations, and identify more effective strategies for prediction across populations. Design Population based cohort study. Setting Primary prevention cohorts in New Zealand and China. Participants Equivalent New Zealand and Chinese cohorts of people aged 30-74 years with diabetes and no CVD based on electronic health records. Main outcome measures Country specific five year equations were derived using standardised methods to predict first cardiovascular events; predictor hazard ratios were compared. New Zealand equations were applied to the Chinese cohort, with performance assessed before and after standard recalibration, followed by the updating of predictor hazard ratios that differed statistically significantly. Results Between 2004 and 2018, 47 958 New Zealand participants had 5622 (11.7%) first CVD events. Between 2010 and 2018, 46 558 Chinese participants had 3650 (7.8%) first CVD events. Sex specific equations included 16 predictors. Hazard ratios for most predictors were similar between New Zealand and Chinese equations but differed markedly for age (eg, women: 1.61, 95% confidence interval 1.51 to 1.71 per 10 years in New Zealand, 2.51 (2.32 to 2.72) per 10 years in China). Standard recalibration of New Zealand equations did not improve calibration in the Chinese cohort (expected to observed ratio: 0.938 in men, 0.809 in women). Replacing New Zealand age coefficients with Chinese coefficients noticeably improved calibration (expected to observed ratio 1.027 in men, 1.025 in women). Conclusions Applying a high income country derived CVD risk prediction equation to a population from a non-high income country with diabetes may introduce systematic errors, primarily driven by differences in the effect of age. Moving beyond simple recalibration to updating key predictor coefficients, particularly age, should be considered for more accurate risk stratification. This study highlights potential limitations of high income country equation updating practices and proposes a feasible approach to improve their accuracy when used in populations from non-high income countries.