OBJECTIVES: Blood pressure (BP) is a long-established risk factor for cardiovascular disease (CVD). SBP is used in all widely used cardiovascular risk scores for clinical decision-making. Recently, within-person BP variability has been shown to be a major predictor of CVD. We investigated whether cardiovascular risk scores could be improved by incorporating BP variability with standard risk factors. METHODS: We used cohort data on patients aged 40-74 on 1 January 2005, from English general practices contributing to the Clinical Practice Research Datalink, a research database derived from electronic health records. Data were linked to hospital episodes and mortality data. SBP variability independent of the mean was calculated across up to six clinic visits. We divided data geographically into derivation and validation data sets. In the derivation data set, we developed a reference model, incorporating risk factors used in previous scores and an index model, incorporating the same factors and BP variability. We calculated model validation statistics in the validation data set including calibration ratio and c-statistic. RESULTS: In the derivation data set, BP variability was associated with CVD, independently of other risk factors (P = 0.005). However, in the validation data set, both models had similar c-statistic (0.7415 and 0.7419, respectively), R (31.8 and 32.0, respectively) and calibration ratio (0.938 and 0.940, respectively). CONCLUSION: The association of BP variability with CVD is statistically significant in a large data set but does not substantially improve the performance of a cardiovascular risk score.
OBJECTIVES: Blood pressure (BP) is a long-established risk factor for cardiovascular disease (CVD). SBP is used in all widely used cardiovascular risk scores for clinical decision-making. Recently, within-person BP variability has been shown to be a major predictor of CVD. We investigated whether cardiovascular risk scores could be improved by incorporating BP variability with standard risk factors. METHODS: We used cohort data on patients aged 40-74 on 1 January 2005, from English general practices contributing to the Clinical Practice Research Datalink, a research database derived from electronic health records. Data were linked to hospital episodes and mortality data. SBP variability independent of the mean was calculated across up to six clinic visits. We divided data geographically into derivation and validation data sets. In the derivation data set, we developed a reference model, incorporating risk factors used in previous scores and an index model, incorporating the same factors and BP variability. We calculated model validation statistics in the validation data set including calibration ratio and c-statistic. RESULTS: In the derivation data set, BP variability was associated with CVD, independently of other risk factors (P = 0.005). However, in the validation data set, both models had similar c-statistic (0.7415 and 0.7419, respectively), R (31.8 and 32.0, respectively) and calibration ratio (0.938 and 0.940, respectively). CONCLUSION: The association of BP variability with CVD is statistically significant in a large data set but does not substantially improve the performance of a cardiovascular risk score.