Use of Repeated Blood Pressure and Cholesterol Measurements to Improve Cardiovascular Disease Risk Prediction: An Individual-Participant-Data Meta-Analysis

Abstract The added value of incorporating information from repeated blood pressure and cholesterol measurements to predict cardiovascular disease (CVD) risk has not been rigorously assessed. We used data on 191,445 adults from the Emerging Risk Factors Collaboration (38 cohorts from 17 countries with data encompassing 1962–2014) with more than 1 million measurements of systolic blood pressure, total cholesterol, and high-density lipoprotein cholesterol. Over a median 12 years of follow-up, 21,170 CVD events occurred. Risk prediction models using cumulative mean values of repeated measurements and summary measures from longitudinal modeling of the repeated measurements were compared with models using measurements from a single time point. Risk discrimination (C-index) and net reclassification were calculated, and changes in C-indices were meta-analyzed across studies. Compared with the single-time-point model, the cumulative means and longitudinal models increased the C-index by 0.0040 (95% confidence interval (CI): 0.0023, 0.0057) and 0.0023 (95% CI: 0.0005, 0.0042), respectively. Reclassification was also improved in both models; compared with the single-time-point model, overall net reclassification improvements were 0.0369 (95% CI: 0.0303, 0.0436) for the cumulative-means model and 0.0177 (95% CI: 0.0110, 0.0243) for the longitudinal model. In conclusion, incorporating repeated measurements of blood pressure and cholesterol into CVD risk prediction models slightly improves risk prediction.

Notes: Data from studies were split into two 50% samples, with data from the second sample censored at the prediction time. Data from both samples were used to derive the longitudinal models while data from the first sample was used to derive the Cox proportional hazards models. Risk prediction and validation were done using data from the second sample only.

Web Figure 1B. Flow-chart showing the use of the study data in model derivation and validation
No.  (9) 133.9 (20.9) 5.9 (1.

Total cholesterol
Adjusted for: sex, age, smoking status, history of diabetes, baseline systolic blood pressure, and baseline HDL cholesterol. Figure 3 continued: Regression dilution ratio graphs for systolic blood pressure, total cholesterol, and HDL cholesterol)

High-density lipoprotein (HDL) cholesterol
Adjusted for: sex, age, smoking status, history of diabetes, baseline systolic blood pressure, and baseline total cholesterol.

Web Figure 6. Changes in cardiovascular disease risk discrimination in each study
Model 1 used baseline measures of systolic blood pressure, total cholesterol and HDL cholesterol. Model 2 used cumulative means of systolic blood pressure, total cholesterol and HDL cholesterol. Model 3 used individual-level random intercepts and slopes for systolic blood pressure, total cholesterol and HDL cholesterol.    Results are weighted by the number of CVD events in each study. Interpretation: e.g. for every one unit increase in the mean age of participants in a study, the difference in risk discrimination between Model 2 and Model 1 decreased by 0.0004 (-0.0007, 0.0000). Statistically significant results in bold.

Web Figure 7. Changes in cardiovascular disease risk discrimination in each study, using study-specific survival models
Model 1 used baseline measures of systolic blood pressure, total cholesterol and HDL cholesterol. Model 2 used cumulative means of systolic blood pressure, total cholesterol and HDL cholesterol. Model 3 used individual-level random intercepts and slopes for systolic blood pressure, total cholesterol and HDL cholesterol.

Web Figure 8. Changes in cardiovascular disease risk discrimination in each study, restricted to participants with two or more repeat measures of systolic blood pressure
Model 1 used baseline measures of systolic blood pressure, total cholesterol and HDL cholesterol. Model 2 used cumulative means of systolic blood pressure, total cholesterol and HDL cholesterol. Model 3 used individual-level random intercepts and slopes for systolic blood pressure, total cholesterol and HDL cholesterol. No difference in C-index between Model 2 and Model 1, and Model 3 and Model 1, could be calculated for the studies "CHARL" and "CAPS" (all participants in the validation sample were censored) and as such, these studies were excluded from the meta-analysis of change in C-index.

Web Figure 9A. Changes in cardiovascular disease risk discrimination in each study in participants aged 40-69 years
Model 1 used baseline measures of systolic blood pressure, total cholesterol and HDL cholesterol. Model 2 used cumulative means of systolic blood pressure, total cholesterol and HDL cholesterol. Model 3 used individual-level random intercepts and slopes for systolic blood pressure, total cholesterol and HDL cholesterol. No difference in C-index between Model 2 and Model 1, and Model 3 and Model 1, were calculated for the studies "DRECE", "GOH", "ISRAEL", "PROCAM", "QUEBEC", and "WHITE2" (all participants in the validation sample were censored in one of the age groups) and as such, these studies were excluded from the meta-analysis of change in C-index. The study "DESIR" was not included in the analysis as no participants were aged 70 years or older.

Web Figure 9B. Changes in cardiovascular disease risk discrimination in each study in participants aged ≥70 years
Model 1 used baseline measures of systolic blood pressure, total cholesterol and HDL cholesterol. Model 2 used cumulative means of systolic blood pressure, total cholesterol and HDL cholesterol. Model 3 used individual-level random intercepts and slopes for systolic blood pressure, total cholesterol and HDL cholesterol. No difference in C-index between Model 2 and Model 1, and Model 3 and Model 1, were calculated for the studies "DRECE", "GOH", "ISRAEL", "PROCAM" and "WHITE2" (all participants in the validation sample were censored in one of the age groups) and as such, these studies were excluded from the meta-analysis of change in C-index. The study "DESIR" was not included in the analysis as no participants were aged 70 years or older.