Key result
Recalibrating CV risk models improves discrimination, achieving C-statistics up to ~0.80 in rural Chinese cohorts.
Why the study?
The performance and applicability of various cardiovascular disease mortality risk prediction models in Chinese rural populations were uncertain and required assessment and recalibration.
Cohort (n=10,338)
No
Effect estimate: AUC 0.776 to 0.795 in men, 0.753 to 0.760 in women
May aid risk assessment in rural Chinese adults; leaves open effects on outcomes and broader applicability.
This research aims to assess application of different cardiovascular disease (CVD) mortality risk prediction models in Chinese rural population. Data was collected from a 6-year follow-up survey in rural area of Henan Province, China. 10338 participants aged 40 to 65 years were included. Baseline study was conducted between 2007 and 2008, and followed up from 2013 to 2014. Seven models: general Framingham risk score (general-FRS), simplified-FRS, Systematic Coronary Risk Evaluation for high (SCORE-high), SCORE-low, Chinese ischemic CVD (CN-ICVD), Pooled Cohort Risk Equation for white (PCE-white) and for African-American (PCE-AA) were assessed and recalibrated. The model performance was evaluated by C-statistics and modified Nam-D’Agostino test. 168 CVD deaths occurred during follow-up. All seven models showed moderate C-statics ranging from 0.727 to 0.744. Following recalibration, general-FRS, simplified-FRS, CN-ICVD, PCE-white and PCE-AA had improved C-statistics of 0.776, 0.795, 0.793, 0.779, and 0.776 for men and 0.756, 0.753, 0.755, 0.758 and 0.760 for women, respectively. Calibrations χ2 of general-FRS, simplified-FRS, SCORE-high, CN-ICVD and PCE-AA model for men, and general-FRS, CN-ICVD and PCE-white model for women were statistically acceptable, indicating these models predicts CVD mortality risk more accurately than others and could be recommended in Chinese rural population.
No takes yet. Share an insight, caveat, or question.
Sun et al. (2017) conducted a cohort in Cardiovascular disease risk (n=10,338). Recalibrated cardiovascular disease mortality risk prediction models vs. Original uncalibrated risk prediction models was evaluated on Discriminative ability (C-statistic/AUC) for 5-year cardiovascular mortality (AUC 0.776 to 0.795 in men, 0.753 to 0.760 in women). Recalibration of cardiovascular risk prediction models improved their discrimination, achieving C-statistics of 0.776-0.795 in men and 0.753-0.760 in women in a rural Chinese population.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: