Systolic and diastolic blood pressure, mean arterial pressure and pulse pressure for prediction of cardiovascular events and mortality in a Middle Eastern population

Abstract

We compared systolic (SBP) and diastolic blood pressure (DBP), mean arterial pressure (MAP) and pulse pressure (PP) as independent predictors of cardiovascular disease (CVD), total and CVD mortality among an Iranian population. The study conducted among 5991 subjects aged ≥ 30 years without baseline CVD and antihypertensive medication. The mean of two measurements of SBP and DBP, in sitting position, was considered the subject's blood pressure. During a median follow-up of 8.7 years, 346 CVD and 157 deaths, 63 attributed to CVD, occurred. Hazard ratios (HRs) of each outcome were calculated for a one standard deviation (SD) increase in each blood pressure (BP) measures. In multivariate models, all BP measures were associated with increased risk of CVD regardless of age. In those aged < 60 years, SBP, DBP, PP and MAP were associated with total mortality (p < 0.05), but in subjects aged ≥ 60 years, only SBP and PP increased risk of total mortality significantly. In multivariate analyses, a 1SD increase in SBP, PP and MAP were associated with 35%, 31% and 28% increased risk of CVD mortality (p < 0.05). In terms of fitness and discrimination of models, DBP, PP and MAP were not superior to SBP. In conclusion, our findings provided further evidence from a Middle Eastern population, in support of SBP predictability for CVD events and CVD and all-cause mortality compared with other BP measures.

Key Words::

• blood pressure measures
• cardiovascular disease
• cardiovascular mortality
• systolic blood pressure
• total mortality

Introduction

Hypertension is an important risk factor for cardiovascular disease (CVD) and premature mortality throughout the world (1,2). In the Asia-Pacific region, it was shown that the prevalence of hypertension is between 7% and 38% in women and between 5% and 47% in men (3); in the Iranian population, the prevalence of hypertension was 25.2% in 2005, whereas in 2007 this rate was 26.6% (4).

Some previous studies reported that there were continuous and independent associations between different blood pressure (BP) components and risk of CVD and death (5,6) and compared the performance of these components for incident CVD and mortality. Recent studies on the predictive power of systolic (SBP), diastolic (DBP) and mean arterial pressure (MAP), as well as pulse pressure (PP), are relatively scarce. Some results reported that SBP is a stronger predictor of CVD than other BP components (7) or is similar to PP , defined as SBP minus DBP, for prediction of CVD or all-cause mortality (8), whereas several epidemiological studies showed that PP is stronger than SBP for prediction of CVD or mortality (9,10) apparently because of increased arterial stiffness, which increases by age and leads to elevated PP (9). It remains controversial which BP component has greater predictive ability for incident CVD or total mortality. There is lack of data on this issue in a Middle East population as to who will have the greatest increase in CVD risk factors (4,11). We aimed to compare the predictive abilities of different BP components as predictors of incident CVD, fatal CVD and total death in a Middle Eastern population.

Materials and methods

Study population

Subjects in this study were selected from among participants of the Tehran Lipid and Glucose Study (TLGS), a prospective study conducted to determine the risk factors and outcomes for non-communicable diseases (12). To summarize, 15,005 people aged 3 years and over, residents of district 13 of Tehran, underwent a baseline examination in 1999–2001. After this cross-sectional phase, subjects were categorized into the cohort and intervention groups, the latter to be educated for implementation of lifestyle modifications. In total, 8071 subjects aged ≥ 30 years were evaluated in the cross-sectional phase of the TLGS. Subjects with prevalent CVD (n = 521), those who were receiving antihypertensive medication (n = 840) at the baseline and those with missing data on blood pressure (n = 163) were excluded (Considering overlap among exclusions), leaving 6752 participants, of whom 5991 subjects (88.7%) were followed until 2009 with a median follow-up of 8.7 years (Figure 1). The Ethical Committee of the Research Institute for Endocrine Sciences approved this study. Informed written consent was obtained from all subjects.

Figure 1. Study population, *considering overlap among exclusions.

Clinical and laboratory measurements

Using a pretested questionnaire, a trained interviewer collected information, which included demographic data, past medical history and family history of CVD, consumption of antihypertensive and lipid lowering drugs and smoking behaviour. Subjects were also questioned about past history of diabetes mellitus and taking of anti-diabetic drugs. Details of the measurements of weight, height and waist circumference (WC) have been published preciously (12). Body mass index (BMI) was calculated as weight (kg) divided by square of height (m²). Two measurements of SBP and DBP were performed using a standardized mercury sphygmomanometer (calibrated by the Iranian Institute of Standards and Industrial Researches) on the right arm after a 15-min rest in a sitting position; the first and fifth Korotkoff sounds were recorded as SBP and DBP, respectively. The mean of the two measurements was considered the subject's blood pressure.

A blood sample was drawn between 07:00 and 09:00 h from all study participants after 12–14 h overnight fasting. Total cholesterol and fasting plasma glucose were measured at the TLGS research laboratory on the day of blood collection with previously reported methods (12).

CVD outcome

Details of the cardiovascular outcome collection have been published elsewhere (12). To summarize, participants were followed up for any medical event annually by phone calls. A trained nurse asked them for any medical conditions and then, a trained physician collected complementary data regarding that event during a home visit and by acquisition of data from medical files. The collected data were then evaluated by an outcome committee consisting of an internist, an endocrinologist, a cardiologist, an epidemiologist and other experts. In the present study, our events targeted were the first CVD events, including definite myocardial infarction [(MI); with diagnostic electrocardiogram (ECG) and biomarkers], probable MI (positive ECG findings plus cardiac symptoms or signs plus missing biomarkers or positive ECG findings plus equivocal biomarkers), unstable angina (new cardiac symptoms or changing symptom patterns and positive ECG findings with normal biomarkers), angiographic proven CHD, stroke (as defined by a new neurological deficit that lasted more than 24 h) and death from CVD.

Definition of terms

Family history of premature CVD reflected any prior diagnosis of CVD by a physician in any female first-degree relative, under 65 years and male first-degree relative, under 55 years old. Smoking includes past or current smoking of any kind of tobacco. Diabetes mellitus was defined as fasting plasma glucose ≥126 mg/dl or current use of anti-diabetic drugs.

Statistics analysis

Mean (SD) values or frequencies (percentage) of the baseline characteristics are expressed for individuals with and without CVD events stratified by age. Comparison of baseline characteristics between the two groups was made by Student's t-test for continuous variable and chi-square test for categorical variables.

Cox proportional hazards regression was used to estimate hazard ratio (HR) and 95% confidence intervals (CIs) for the associations of each BP measure with CVD events, all-cause mortality and CVD mortality. Follow-up duration was defined as the period between entrance to study and the end points in each analysis. End points were considered the first CVD event, death from any cause including CVD and censoring, which was defined as leaving the residence area, loss to follow-up or end of the follow-up.

Interactions between each blood pressure measure with sex and age were tested using log-likelihood ratio test of models containing first order interactions; there was no evidence of any sex-specific risk effects (p-values for interaction >0.1) but there was significant interaction between age and BP measures for incident CVD events and total mortality. Cox models stratified by age into two groups – <60 or ≥60 years – for CVD outcome and all-cause mortality. There was no interaction between age and BP measures for CVD mortality outcomes. To select the covariates to be included in the multivariate models, univariate analysis was used for each candidate covariate (sex, smoking status, family history of premature CVD, diabetes, lifestyle intervention, BMI, WC, total cholesterol); then, each covariate with a p-value less than 0.2, was selected to be included in a stepwise backward multivariate Cox regression analysis (p remove = 0.1) with three outcomes. Final selected covariates were sex, smoking status, family history of premature CVD, diabetes, lifestyle intervention group, BMI, WC and cholesterol for CVD events. For all-cause mortality, final selected covariates were sex, BMI, diabetes and smoking status. Selected covariates for CVD mortality were sex, age, diabetes, lifestyle intervention; we added BMI and smoking status to this model because they seemed to be the important confounders for incident CVD mortality. Adjusted HRs, with 95% CI, were indicated the increase risk of a 1SD increase in each BP measure. For each model, we assessed the goodness of fit of the resulting models through likelihood ratio x² and the Akaike Information Criterion (AIC). A lower value of AIC and higher value of likelihood ratio x² indicates a better model fitness. The discriminatory power of the models was calculated by the C index, which was estimated with the “somersd” STATA command that yielded CI for the Harrel's C index using jack-knife variance estimation. A value of 1 denotes perfect discrimination and a value of 0.5 is no better than chance.

The proportional hazards assumption in the Cox model was assessed graphically and with a Schoenfild residual test. All proportionality assumptions were appropriate.

Statistical analyses were performed using SPSS version 15.0 and STATA statistical software version 10. A p-value ≤0.05 was considered significant.

Results

The study sample consisted of 5991 subjects (4985 aged <60 years and 1006 aged ≥60 years with mean age of 42 and 66 years, respectively). There was no significant difference between subjects who completed the follow-up and those who did not in baseline characteristics such as age, BP parameters and cardiovascular risk factors. During a median follow-up of 8.7 years, 346 (198 persons < 60 years, 148 persons ≥ 60 years) first CVD events and 157 deaths (45 persons < 60 years, 112 persons ≥ 60 years) occurred.

Table I shows the range of the different blood pressure components among study population.

Table I. Mean and range of baseline blood pressure components in study population.

	All subjects (n = 5991)	Aged < 60 years (n = 4985)	Aged ≥ 60 years (n = 1006)
Systolic blood pressure	120 (77–220)	116 (77–219)	134 (84–203)
Diastolic blood pressure	78 (45–132)	78 (45–132)	79 (48–121)
Mean arterial pressure	92 (60–161)	91 (60–161)	97 (68–147)
Pulse pressure	41 (10–122)	39 (10–103)	54 (14–122)

Mean (minimum–maximum) of blood pressure components.

The baseline characteristics and BP parameters in subjects with and without CVD events are summarized in Table II according to age. In the group aged <60 years, in comparison with subjects without CVD, those with CVD were older and had significantly higher BMI, WC, SBP, DBP, PP, MAP, cholesterol level, smoking behaviour, family history of CVD and diabetes. Similarly, in the group aged ≥60 years, in comparison with the subjects without CVD, those with the condition had higher diabetes, smoking behaviour, WC, SBP, DBP, PP and MAP, whereas there were no differences in cholesterol level and BMI in these two groups.

Table II. Comparison of baseline characteristics between participants with and without cardiovascular disease stratified by age.

	Age < 60 years			Age ≥ 60 years
	Without CVD (n = 4787)	With CVD (n = 198)	p-value	Without CVD (n = 858)	With CVD (n = 148)	p-value
Age (years)	41.8 ± 8.18	48.94 ± 7.19	<0.001	66.17 ± 5.23	66.34 ± 5.25	0.9
Sex (%male)	41.9	62.1	<0.001	55.6	70.3	0.7
Smoking (%)	22.3	34.8	<0.001	24.5	39.5	<0.001
Family history of CVD (%)	16.9	25.8	0.002	9.8	16.9	0.01
Diabetes (%)	5.4	28.3	<0.001	15.7	31.1	<0.001
Lifestyle intervention (%)	38.2	33.3	0.18	45.21	37.2	0.07
Body mass index (kg/m^2)	27.3 ± 4.7	28.7 ± 4.5	<0.001	26.7 ± 4.3	27.0 ± 4.0	0.5
Waist circumference (cm)	88.9 ± 11.3	95.5 ± 10.8	<0.001	91.8 ± 11.2	94.6 ± 11.2	0.005
Total cholesterol (mg/dl)	211 ± 44	239 ± 50	<0.001	226 ± 45	230 ± 47	0.3
Systolic blood pressure (mmHg)	116 ± 15	127 ± 21	<0.001	132 ± 20	140 ± 22	<0.001
Diastolic blood pressure (mmHg)	78 ± 10	82 ± 13	<0.001	79 ± 11	82 ± 13	0.02
Pulse pressure (mmHg)	38 ± 10	45 ± 13	<0.001	53 ± 16	58 ± 16	<0.001
Mean arterial pressure (mmHg)	91 ± 11	97 ± 15	<0.001	97 ± 13	101 ± 14	<0.001

Mean ± SD are shown for continuous variables and p-value is calculated with t-test; % is shown for categorical variables with p-value according to chi-square; CVD: cardiovascular disease.

Table III presents HRs of a 1SD increase in each BP parameters for incident CVD event, stratified by age. In subjects aged < 60 years, there were positive associations between different BP parameters with CVD outcomes, in sex-adjusted models. After further adjustment for other risk factors, the HRs decreased but were still significant so that a 1SD increase in SBP, DBP, PP and MAP caused a 43%, 24%, 39% and 37% increase risk of CVD, respectively. Also, in those aged ≥ 60 years, all blood pressure components were significantly associated with risk of CVD events both in the sex- and multivariate-adjusted models. The models including SBP had the smallest AIC regardless of age group. There was no significant difference between the discriminatory power of SBP model with other blood pressure components (p-value >0.05 for differences in C indices), in subjects aged <60 and ≥60 years.

Table III. Adjusted hazard rates and predictive utilities for different blood pressure (BP) measures and cardiovascular disease (CVD) outcome.

		Sex-adjusted model		Multivariate-adjusted model
	SD (mmHg)	HR (95% CI)	p-value	HR (95% CI)	p-value	C index (95% CI)	AIC
<60 years
SBP	15	1.69 (1.55–1.85)	<0.001	1.43 (1.26–1.61)	<0.001	0.81 (0.74–0.87)	3029
DBP	10	1.48 (1.32–1.67)	<0.001	1.24 (1.08–1.42)	0.003	0.80 (0.73–0.86)	3052
PP	10	1.61 (1.46–1.80)	<0.001	1.39 (1.23–1.57)	<0.001	0.80 (0.74–0.87)	3031
MAP	11	1.63 (1.46–1.82)	<0.001	1.37 (1.20–1.56)	<0.001	0.80 (0.74–0.87)	3040
≤60 years
SBP	20	1.39 (1.18–1.63)	<0.001	1.38 (1.18–1.63)	<0.001	0.66 (0.56–0.76)	1861
DBP	11	1.20 (1.03–1.41)	0.02	1.20 (1.00–1.44)	0.03	0.63 (0.54–0.73)	1872
PP	16	1.33 (1.14–1.56)	<0.001	1.29 (1.10–1.51)	<0.001	0.68 (0.58–0.77)	1865
MAP	13	1.32 (1.13–1.53)	<0.001	1.33 (1.11–1.60)	<0.001	0.64 (0.54–0.74)	1865

Multivariate models adjusted with sex, smoking status, family history of cardiovascular disease, diabetes, lifestyle intervention, body mass index, waist circumference, total cholesterol. SBP, systolic blood pressure; DBP, diastolic blood pressure; PP, pulse pressure; MAP, mean arterial pressure; SD, standard deviation; CI, confidence interval; HR, hazard ratios, indicating the increase risk for a 1SD increase in different BP measures; AIC, Akaike Information Criterion.

Table IV shows the association between various blood pressure components and incident all-cause mortality. In subjects whose age were <60 years, all BP components were associated with incident total mortality both in the sex- and multivariate-adjusted models; the multivariate-adjusted HR associated with a 1SD increase in BP was higher for SBP [1.67 (1.35–2.06)] than PP [1.55 (1.24–1.93)], DBP [1.47 (1.11–1.94)] or MAP [1.62 (1.24–2.10)]. Among individuals aged ≥60 years, in the sex- and multivariate-adjusted models, the HRs of PP and SBP were significantly associated with incident death, so that a 20-mmHg increase in SBP and 16-mmHg increase in PP were associated with 28% and 29% increased risk of total mortality, respectively.

Table IV. Adjusted hazard ratios and discrimination of different blood pressure (BP) measures for predicting all-cause mortality.

		Sex-adjusted model		Multivariate adjusted model
	SD (mmHg)	HR (95% CI)	p-value	HR (95% CI)	p-value	C index (95% CI)	AIC
<60 years
SBP	15	1.64 (1.33–2.03)	<0.001	1.67 (1.35–2.06)	<0.001	0.70 (0.53–0.87)	719
DBP	10	1.47 (1.13–1.90)	0.004	1.47 (1.11–1.94)	0.006	0.76 (0.60–0.92)	728
PP	10	1.56 (1.25–1.95)	<0.001	1.55 (1.24–1.93)	<0.001	0.64 (0.44–0.83)	723
MAP	11	1.60 (1.26–2.03)	<0.001	1.62 (1.24–2.10)	<0.001	0.76 (0.61–0.92)	723
≥60 years
SBP	20	1.18 (1.00–1.38)	0.05	1.28 (1.04–1.56)	0.01	0.66 (0.55–0.77)	1430
DBP	11	0.92 (0.77–1.10)	0.41	1.05 (0.85–1.28)	0.67	0.68 (0.58–0.79)	1436
PP	16	1.29 (1.10–1.51)	0.002	1.29 (1.07–1.56)	0.005	0.66 (0.55–0.77)	1428
MAP	13	1.05 (0.88–1.26)	0.61	1.17 (0.95–1.44)	0.11	0.67 (0.56–0.78)	1433

Multivariate models adjusted with sex, smoking status, diabetes and body mass index. SBP, systolic blood pressure; DBP, diastolic blood pressure; PP, pulse pressure; MAP, mean arterial pressure; SD, standard deviation; CI, confidence interval; HR, hazard ratios, indicating the increase risk for a 1SD increase in different BP measures; AIC, Akaike Information Criterion.

Discrimination of SBP model was similar to the discriminatory powers of the other models in both age groups.

The sex-, age- and multivariate-adjusted HRs of CVD mortality for a 1SD increase in BP components are shown in Table V. The significant multivariate HRs of a 1SD increase in SBP, PP and MAP were 1.35, 1.31, and 1.28, respectively. DBP was not associated with incident CVD mortality in sex-, age-adjusted and multivariate models. According to AIC, all the three models had almost similar AIC. Discriminatory powers of the models with different BP components were similar (C index = 88%).

Table V. Adjusted hazard ratios and discrimination of different blood pressure (BP) measures for predicting cardiovascular disease (CVD) mortality.

		Sex- and age-adjusted model		Multivariate-adjusted model
	SD (mmHg)	HR (95% CI)	p-value	HR (95% CI)	p-value	C index (95% CI)	AIC
SBP	18	1.35 (1.10–1.66)	0.003	1.35 (1.10–1.66)	0.004	0.88 (0.83–0.93)	918
DBP	10	1.13 (0.92–1.39)	0.2	1.16 (0.92–1.45)	0.2	0.88 (0.83–0.93)	924
PP	13	1.31 (1.10–1.56)	0.003	1.31 (1.07–1.60)	0.006	0.88 (0.83–0.93)	919
MAP	12	1.26 (1.03–1.56)	0.03	1.28 (1.02–1.61)	0.03	0.88 (0.83–0.93)	921

Multivariate models adjusted with sex, age, smoking status, diabetes, intervention, body mass index. SBP, systolic blood pressure; DBP, diastolic blood pressure; PP, pulse pressure; MAP, mean arterial pressure; SD, standard deviation; CI, confidence interval; HR, hazard ratios, indicating the increase risk for a 1SD increase in different BP measures; AIC, Akaike Information Criterion.

Discussion

In this large prospective study of Middle Eastern residents, we examined different BP measures as predictors of CVD events, CVD mortality and all-cause mortality, using predictive performance tests. Among both younger and older ones, all of the BP measures were independent predictors of incident CVD; however, SBP and PP generally highlighted better fitness than other BP measures. With respect to total mortality, we found that among younger population all of the BP measures remained independent predictors; however, SBP, PP and MAP demonstrated better fitness than DBP. Furthermore, in the group aged ≥60 years, after adjustment for risk factors, only SBP and PP highlighted a significant risk for total mortality. The discriminatory powers of all BP measures in multivariate models for CVD events and all-cause mortality generally declined with increasing age. Concerning CVD mortality, the study highlighted SBP, PP and MAP as independent predictors, however; our goodness-of-fit findings showed no preference for either BP measures in the prediction of CVD mortality. Overall, with the exception of DBP and MAP in individuals more than 60 years and total mortality and of DBP and CVD mortality, all BP components were significantly associated with CVD, total mortality and CVD mortality.

In line with our results, in other studies SBP showed a steady linear association with CVD and mortality among both younger and older population (13,14). The current study showed that DBP did not have significant role in predicting total mortality among older population and CVD mortality among the whole population, the issues have been addressed in other relevant studies (15,16). It was hypothesized that the lower importance of DBP as a CVD mortality risk factor is related to the great influence of antihypertensive medications on DBP, which increases the position effect of SBP and PP (2). However, among the Iranian population aged < 60 years, DBP should be given simultaneous careful concern since of its tough independent association to death.

PP, as a measure of arterial stiffness, is higher after age 50 years, resulting from tendency of SBP to be higher and DBP to be lower with age (10,17,18). However, in our study population, there was no statistical evidence of superiority of PP than SBP, concerning total mortality in both young and older population and CVD mortality in whole population. In the Chicago Heart Association cohorts, Miura et al. (17) reported that PP was a weaker predictor of CHD, CVD and all-cause mortality than SBP and MAP in all age groups. Applying the same cohort as Miura et al. (17), Mosley et al. (18) concluded that SBP had greater predictive value than PP for CHD (10,17,18). Regarding CVD events, among the Iranian population, we also highlighted that SBP is similar to or slightly better than PP in both young and old population.

Of the summery measures, MAP was independently and strongly associated with CVD events in both age groups and CVD mortality among the whole population; however, it did not show any relation with total mortality among the old population. On the other hand, this measure may be complicated to include in to clinical practice, and it was no better than SBP in risk prediction in our study population and other studies (18).

In the stratified analysis according by age groups (<60 years vs ≥60 years), among the younger TLGS population, SBP and PP were stronger predictors of CVD risk than DBP or MAP; however, after the age of 60 years, SBP, PP and MAP appeared the better predictor than DBP. In contrast to our results, among adult male participants of PROCAM, no component of BP was predictor of CHD risk at age < 50 years and in the older men only PP was a significant predictor (10). In the Framingham study, among subjects aged < 50 years, DBP was a stronger predictor of CHD risk than SBP or PP; however, from age ≥ 60 years on, PP appeared the best predictor (19). The reason for the inconsistency between the mentioned studies is not clear, although part of the explanation might related to different definition of outcomes (i.e. CHD vs CVD vs mortality). The increased HRs and the higher discriminatory power of all BP measures in younger compared with older individuals among the TLGS population is expected to have some reasons. In our data analysis, CVD events occurred in 14.7% of population aged ≥ 60 years vs 4% of population aged < 60 years; hence, among older population with coexisting other comorbidities, the CVD outcomes occurred in the presence of high or even normal BP, resulting in attenuating the risk of BP components with age. Furthermore, the high probability of receiving antihypertensive medications during follow-up among the older population might decrease the risk of BP measures, as suggested by others (18,20,21).

Finally, in a population-based study of Tehranian adults, SBP and PP had similar discrimination and fitness characteristic for prediction of CVD events and total mortality among both younger and older adults, which were stronger than MAP and DBP.

There are several points that should be considered when examining the results of this study. First, BP values were based on two measures in 1 day at baseline; the use of multiple BP measurements on different days would improve accuracy and precision. Nevertheless, as highlighted here and in many population-based studies, even a single BP measurement is strongly prognostic of CVD events (7,8,10). Second, subjects with a previous history of CVD or subjects who used antihypertensive medication at baseline were excluded, but participants may have started taking antihypertensive medication during follow-up, leading to an underestimation of the associations. Finally, our population was determined in Middle East Caucasian residents in the capital city of Iran and further studies should be conducted to determine whether our findings are applicable to other populations in this region.

In conclusion, regarding the implications of our results, SBP and PP are independent predictors of CVD events, all-cause and CVD mortality in both young and old population. Use of PP may not be practical in public health practice; hence, SBP should continue to be the main focus for prevention of CVD events and total and CVD mortality among the Iranian population.

Acknowledgments

This study was supported by Grant No. 121 from the National Research Council of the Islamic Republic of Iran. We express appreciation to the participants of district 13, Tehran, for their enthusiastic support in this study. We would like to thank Ms N. Shiva for the English editing of manuscript.

Conflicts of interest: The authors declare that they have no conflicts of interest.