Investigating the utility of blood pressure variability in cardiovascular risk prediction

Date of ISAC Approval: 
Lay Summary: 
Cardiovascular disease (CVD) (heart disease, heart attack and stroke) is the leading cause of death worldwide. Certain people are more likely to develop CVD than others (e.g. smokers, those who are overweight or have high blood pressure (BP)) and calculators exist to predict how likely it is that an individual will develop CVD in the future. For those at particularly high risk of CVD, doctors are encouraged to give these individuals advice about lifestyle changes and treatment that may prevent the disease. Recent research has also shown that having highly variable (instead of stable) BP may also increase the chances of developing CVD. The aim of this study is to determine if information about BP variability should be included in a new calculator to predict the chance of developing CVD. Two calculators will therefore be developed; one with and one without information about BP variability in addition to other traditional patient characteristics. The accuracy of predictions from the calculators will then be compared.
Technical Summary: 
Recent research has suggested that BP variability may be an important risk factor for CVD (over and above mean), in particular variability in visit-to-visit BP measurements. However, although the additional prognostic value of BP variability has been demonstrated through survival analysis of long-term cohort and trial data, the utility of this factor in a risk prediction model has not been assessed. This study aims to develop two risk scores to predict future risk of CVD in individuals, one including traditional risk factors alone and a second additionally including BP variability as a risk factor. Both risk scores will be developed using data from adults without prior history of CVD, with linkage to mortality and hospital episodes data (to more accurately ascertain events) and deprivation data. The risk scores will be developed in a derivation subsample using parametric survival models and fractional polynomials to model the relationship between traditional CVD risk factors and CVD outcomes. The accuracy of the two risk scores will be compared to each other in terms of discrimination, calibration and net reclassification index in a validation subsample.
Health Outcomes to be Measured: 
Cardiovascular events; - Myocardial infarction - Coronary and ischaemic heart disease - Angina - Cerebrovascular and haemorrhagic stroke events - Cause-specific mortality

Professor Richard Stevens - Chief Investigator - University Of Oxford
Professor Richard McManus - Researcher - University Of Oxford
Sarah Lay-Flurrie - Researcher - University Of Oxford

HES Admitted;ONS;Patient IMD;Practice IMD (Standard)