The use of repeated blood pressure measures for cardiovascular risk prediction: a comparison of statistical models in the ARIC study

Background: The Framingham Risk Score (FRS) is an established tool for the prediction of cardiovascular disease (CVD) risk. The established CVD risk factors age, HDL cholesterol, total cholesterol, systolic blood pressure (SBP), antihypertensive treatment, diabetes mellitus and smoking are used in the calculation of the FRS. The prevalence and distribution of these risk factors in the population have changed within the last decades and especially average levels of SBP have declined. However, the impact of this change on the risk prediction performance of the FRS has not been investigated. Hypothesis: We assessed the hypothesis that the relative contribution of SBP to CVD risk prediction within the FRS framework has changed from 1985 to 2000. Methods: We used N = 11 760 participants aged 30 - 65 years from four prospective population-based cohort studies enrolled in Southern Germany in 1985, 1990, 1995, and 2000. CVD risk was calculated by recalibrated equations of the original FRS. Predicted CVD risks using the actual SBP values were compared to predicted CVD risks using optimal (SBP < 120 mmHg) values for each participant. We assessed the relative contribution of SBP with three performance measures: First, the median difference in predicted risks with actual and optimal SBP, second, the relative positive predictive value of the FRS using actual compared to optimal SBP values and third, the population attributable risk fraction of SBP using Levin’s formula. Results: CVD events occurred in 6.3% of male participants in 1985 and 6.2% in 2000; in women, event rates were 2.4% and 2.3%, respectively. Mean SBP levels decreased from 134 mmHg (Standard Deviation: 17 mmHg) to 132 (SD: 17) mmHg in men and from 127 (SD: 19) mmHg to 121 (SD: 18) mmHg in women. The difference in median predicted risk declined from 1.21 [Interquartile range 0.52, 3.38] in 1985 to 0.93 [0.35, 2.44] in 2000 in men and from 0.26 [-0.05, 1.45] to -0.07 [-0.19, 0.89] in women. The relative positive predictive value dropped from 0.88 to 0.73 in men and from 0.61 to 0.53 in women. The population attributable risk fraction of SBP decreased from 70.2% (95% CI: 42.1, 89.6) to 29.71% (-6.4, 64.7) in men and from 85.7% (62.9, 93.1) to 57.9% (28.0, 82.0) in women. Given the results from 1990 and 1995, the declining trend was nonlinear for all three performance measures. Conclusion: In conclusion, the relative contribution of blood pressure to cardiovascular risk prediction has decreased within the last decades. This affects the future development of CVD risk prediction methods which will have to consider the changing relative importance of SBP. Furthermore, this might also influence public health policies focusing on the management of SBP and hypertension in order to effectively prevent CVD.

Download Full-text

Cardiovascular Risk Assessment in Hypertensive Patients

American Journal of Hypertension ◽

10.1093/ajh/hpab021 ◽

2021 ◽

Author(s):

Michael C Wang ◽

Donald M Lloyd-Jones

Keyword(s):

Blood Pressure ◽

Risk Assessment ◽

Cardiovascular Risk ◽

Risk Prediction ◽

Clinical Management ◽

Prognostic Significance ◽

Quantitative Risk Assessment ◽

Cardiovascular Risk Assessment ◽

Hypertensive Patients ◽

Cardiovascular Risk Prediction

Abstract Hypertension is a highly prevalent and causal risk factor for cardiovascular disease (CVD). Quantititaive cardiovascular risk assessment is a new paradigm for stratifying hypertensive patients into actionable groups for clinical management and prevention of CVD. The large heterogeneity in hypertensive patients makes this evaluation complex, but recent advances have made CV risk assessment more feasible. In this review, we first describe the prognostic significance of various levels and temporal patterns of blood pressure. We then discuss cardiovascular risk prediction equations and the rationale of taking global risk into account in hypertensive patients. Finally, we review several adjunctive biomarkers that may refine risk assessment in certain patients. We observe that, beyond individual cross-sectional measurements, both short-term and long-term blood pressure patterns are associated with incident CVD; that current cardiovascular risk prediction performs well, and its incorporation into hypertension management is associated with potential population benefit; and that adjunctive biomarkers of target organ damage show the most promise in sequential screening strategies that target biomarker measurement to patients in whom the results are most likely to change clinical management. Implementation of quantitative risk assessment for CVD has been facilitated by tools and direct electronic health record integrations that make risk estimates accessible for counseling and shared decision making for CVD prevention. However, it should be noted that treatment does not return an individual to the risk of someone who never develops hypertension, underscoring the need for primordial prevention in addition to continued innovation in risk assessment.

Download Full-text