Blood pressure and incidence of twelve cardiovascular diseases: lifetime risks, healthy life-years lost, and age-specific associations in 1·25 million people

Summary Background The associations of blood pressure with the different manifestations of incident cardiovascular disease in a contemporary population have not been compared. In this study, we aimed to analyse the associations of blood pressure with 12 different presentations of cardiovascular disease. Methods We used linked electronic health records from 1997 to 2010 in the CALIBER (CArdiovascular research using LInked Bespoke studies and Electronic health Records) programme to assemble a cohort of 1·25 million patients, 30 years of age or older and initially free from cardiovascular disease, a fifth of whom received blood pressure-lowering treatments. We studied the heterogeneity in the age-specific associations of clinically measured blood pressure with 12 acute and chronic cardiovascular diseases, and estimated the lifetime risks (up to 95 years of age) and cardiovascular disease-free life-years lost adjusted for other risk factors at index ages 30, 60, and 80 years. This study is registered at ClinicalTrials.gov, number NCT01164371. Findings During 5·2 years median follow-up, we recorded 83 098 initial cardiovascular disease presentations. In each age group, the lowest risk for cardiovascular disease was in people with systolic blood pressure of 90–114 mm Hg and diastolic blood pressure of 60–74 mm Hg, with no evidence of a J-shaped increased risk at lower blood pressures. The effect of high blood pressure varied by cardiovascular disease endpoint, from strongly positive to no effect. Associations with high systolic blood pressure were strongest for intracerebral haemorrhage (hazard ratio 1·44 [95% CI 1·32–1·58]), subarachnoid haemorrhage (1·43 [1·25–1·63]), and stable angina (1·41 [1·36–1·46]), and weakest for abdominal aortic aneurysm (1·08 [1·00–1·17]). Compared with diastolic blood pressure, raised systolic blood pressure had a greater effect on angina, myocardial infarction, and peripheral arterial disease, whereas raised diastolic blood pressure had a greater effect on abdominal aortic aneurysm than did raised systolic pressure. Pulse pressure associations were inverse for abdominal aortic aneurysm (HR per 10 mm Hg 0·91 [95% CI 0·86–0·98]) and strongest for peripheral arterial disease (1·23 [1·20–1·27]). People with hypertension (blood pressure ≥140/90 mm Hg or those receiving blood pressure-lowering drugs) had a lifetime risk of overall cardiovascular disease at 30 years of age of 63·3% (95% CI 62·9–63·8) compared with 46·1% (45·5–46·8) for those with normal blood pressure, and developed cardiovascular disease 5·0 years earlier (95% CI 4·8–5·2). Stable and unstable angina accounted for most (43%) of the cardiovascular disease-free years of life lost associated with hypertension from index age 30 years, whereas heart failure and stable angina accounted for the largest proportion (19% each) of years of life lost from index age 80 years. Interpretation The widely held assumptions that blood pressure has strong associations with the occurrence of all cardiovascular diseases across a wide age range, and that diastolic and systolic associations are concordant, are not supported by the findings of this high-resolution study. Despite modern treatments, the lifetime burden of hypertension is substantial. These findings emphasise the need for new blood pressure-lowering strategies, and will help to inform the design of randomised trials to assess them. Funding Medical Research Council, National Institute for Health Research, and Wellcome Trust.

Continuous variables are summarised as median (IQR) and categorical as N (%).
Abbreviations: ACEI, Angiotensin-converting enzyme inhibitors; ARBs, Alpha-adrenoreceptor blocking drugs; COPD, chronic obstructive pulmonary disease; HDL, high-density lipoprotein cholesterol.      S 4 Hazard ratios (95% CIs) by sex per 20/10 mmHg higher systolic or diastolic blood pressure adjusted for age.* *Models were fitted separately for men and women and included continuous age, quadratic age, with stratification by primary care practice. Confidence intervals are Bonferroni-corrected (2 sexes X 12 endpoints=24 tests). Figure S 5 Adjusted hazard ratios (95% CIs) per 20/10 mmHg higher systolic or diastolic blood pressure.* Abbreviations: BP meds, blood pressure lowering medications; lipids include total and HDL cholesterol. *All models included continuous age, quadratic age, with stratification by sex and primary care practice, in addition to the covariates listed above. Confidence intervals are Bonferroni-corrected (4 models X 12 endpoints=48 tests).

Figure S 6
Hazard ratios (95% CIs) per 20 mmHg higher systolic blood pressure adjusted for age and sex estimated separately in people with (N= 265,473) or without (N= 992,533) BP-lowering drugs at baseline.* *Models were fitted separately in people treated or not treated with BP medications at baseline, with adjustments for age, quadratic age, and stratification by sex and primary care practice. Confidence intervals are Bonferroni-corrected (2 groups X 12 endpoints=24 tests in total). Figure S 7 Hazard ratios (95% CIs) per 10 mmHg higher diastolic blood pressure adjusted for age and sex estimated separately in people with (N= 265,473) or without (N= 992,533) BP-lowering drugs at baseline.
*Models were fitted separately in people treated or not treated with BP medications at baseline, with adjustments for age, quadratic age, and stratification by sex and primary care practice. Confidence intervals are Bonferroni-corrected (2 groups X 12 endpoints=24 tests in total). Figure S 8 Hazard ratios (95% CIs) for the associations of different cutoffs of systolic blood pressure (reference 115 mmHg) with cardiovascular endpoints in men adjusted for age (BP was modelled as a continuous variable using splines with 3 knots). Figure S 9 Hazard ratios (95% CIs) for the associations of different cutoffs of diastolic blood pressure (reference 75 mmHg) with cardiovascular endpoints in men adjusted for age (BP was modelled as a continuous variable using splines with 3 knots). Figure S 10 Hazard ratios (95% CIs) for the associations of different cutoffs of systolic blood pressure (reference 115 mmHg) with cardiovascular endpoints in women adjusted for age (BP was modelled as a continuous variable using splines with 3 knots).

Figure S 11
Hazard ratios (95% CIs) for the associations of different cutoffs of diastolic blood pressure (reference 75 mmHg) with cardiovascular endpoints in women adjusted for age (BP was modelled as a continuous variable using splines with 3 knots).

Figure S 12
Hazard ratios (95% CIs) for the associations of different cutoffs of systolic blood pressure (reference 115 mmHg) with cardiovascular endpoints in people not receiving blood pressure medications at baseline adjusted for age (BP was modelled as a continuous variable using splines with 3 knots). Figure S 13 Hazard ratios (95% CIs) for the associations of different cutoffs of diastolic blood pressure (reference 75 mmHg) with cardiovascular endpoints in people not receiving blood pressure medications at baseline adjusted for age (BP was modelled as a continuous variable using splines with 3 knots).

Figure S 14
Hazard ratios (95% CIs) for the associations of different cutoffs of diastolic blood pressure (reference 75 mmHg) with cardiovascular endpoints in people receiving blood pressure medications at baseline adjusted for age (BP was modelled as a continuous variable using splines with 3 knots).

Figure S 15
Hazard ratios (95% CIs) for the associations of different cutoffs of systolic blood pressure (reference 115 mmHg) with cardiovascular endpoints in people receiving blood pressure medications at baseline adjusted for age (BP was modelled as a continuous variable using splines with 3 knots). *hypertension defined as systolic blood pressure ≥140 or diastolic blood pressure ≥ 90 mmHg or use of blood pressure lowering treatments or physician-recorded diagnosis at baseline.  *Estimates of fixed effects are equivalent to those reported in the main analysis (but differ slightly because unlike in the main analysis were practices with at least 1 event contributed to the estimates, here we have considered studies with 2 or more for modelling purposes). Estimates of random effects account for practice level heterogeneity; differences between fixed effects and random effects signify that the heterogeneity impacts on the overall assocaitions.

Results from sensitivity analyses
The prediction interval expresses uncertainty about the association in a randomly chosen practice. 1 Figure S 19 Hazard ratios (95% CIs) per 20/10 mmHg higher systolic or diastolic blood pressure adjusted for age and sex, using end points from different sources.
In the main analysis all 3 sources of end points are used (primary care, secondary care and death records). To compute endpoints based on secondary care and mortality (second row where applicable) patients were followed up until a hospital admission or death for the specified event, ignoring diagnoses/events recorded in primary care. Similarly, for mortality endpoints patients were followed up until death, ignoring non-fatal presentations. The majority of deaths from stroke are unclassified (not included in the mortality estimate from ischaemic stroke shown here).   Dataset including observed and imputed BP, N= 1,937,360 patients; complete cases for SBP (as in main analysis) N= 1,258,006 patients.

Multiple imputation
Multiple imputation 1 was implemented using the mice algorithm in the statistical package R. Imputation models were estimated separately for men and women and included: a) all the baseline covariates used in the main analysis (age, quadratic age, diabetes, smoking, systolic blood pressure, diastolic blood pressure, mean arterial pressure, pulse pressure, total cholesterol, HDL cholesterol, body mass index), b) prior (between 1 and 4 years before study entry) and post (between 0 and 1 year after study entry) averages of continuous main analysis covariates and other measurements not in the main analysis (white cell count, haemoglobin, creatinine, alanine transferase), c) baseline medications (statins, blood pressure medications, aspirin, oral contraceptives and hormone replacement therapy), d) coexisting medical conditions (history of depression, cancer, renal disease, liver disease and chronic obstructive pulmonary disease), e) the Nelson-Aalen hazard and the event status for each endpoint analysed in the data 2 .
Non-normally distributed variables were log-transformed for imputation and exponentiated back to their original scale for analysis. Five multiply imputed datasets were generated, and Cox models fitted to each dataset. Coefficients were combined using Rubin's rules.
We checked whether the imputations were plausible by comparing plots of the distribution of observed and imputed values of all variables. We checked whether the exclusion of patients with no blood pressure measurements (within two years of the baseline) biased the reported associations with cardiovascular diseases by comparing hazard ratios estimated based on imputed data ( Figure  S22).

Estimation of lifetime risks and years of life-lost
We estimated lifetime risks of each cardiovascular disease adjusted for the competing risk of other cardiovascular diseases and non-CVD mortality based on Cox models with age as the timescale. For a given disease q this involves a) estimating separate Cox models for events of disease q and for the competing events1,…,q-1,q+1,…,Q including the same predictor variables, b) multiplying the hazard contribution for disease q at a given age by the probability of being still at risk and not having experienced a competing event by that age, and c) computing the cumulative incidence (sum of hazards) between the baseline age (age at entry to study) and age 95 ('lifetime').
Thus, for individual i the lifetime risk of experiencing disease q from baseline age t i to attained aget i +T in the presence of the competing risks is given by: where ( ) is the baseline hazard for disease q at age t is a vector of coefficients for cause q is the vector of covariates for individual i ( ) is the hazard for disease k for individual i at age t given covariates x i Expression [1] is estimated by replacing coefficients and baseline hazards by their estimates. Because the baseline hazard is estimated as a step function, the integrals are replaced by sums over event times.
Suppose the two curves in Figure S1  where ( ) and ( ) represent risk estimates up to age u (estimated by expression [1] and adjusted for other risk factors (baseline age, sex, diabetes, smoking status, total cholesterol, HDL cholesterol) assuming normal blood pressure or hypertension respectively, averaged over individuals in the data. We estimated {2} by applying the trapezium rule in intervals of u=0.5 years, from the baseline age to age 95.