Associations of Systolic Blood Pressure and Diastolic Blood Pressure With the Incidence of Coronary Artery Disease or Cerebrovascular Disease According to Glucose Status

Cardiovascular disease, which consists of coronary artery disease (CAD) and cerebrovascular disease (CVD), is a major cause of death worldwide, including Japan. High blood pressure (BP) is a well-known strong risk factor for cardiovascular disease. A previous Japanese cohort study, NIPPON DATA 80, showed a dose-response relation between systolic BP (SBP) levels and cardiovascular risks: cardiovascular mortality increased with higher quintiles of SBP (<120, 120–139, 140–159, 160–179, ≥180 mmHg) and diastolic BP (DBP) (<80, 80–84, 85–89, 90–99, ≥100 mmHg) (1). However, that study did not examine the relationship according to glucose status (i.e., normoglycemia, borderline glycemia, and diabetes).

The target of optimal BP is different between hypertensive patients with and without diabetes, and it was established that treatment of hypertension reduces cardiovascular mortality (2–6). A previous meta-analysis (4) indicated that achieving SBP <140 mmHg reduces cardiovascular events both in the presence and absence of diabetes, and further benefit was seen in the absence of diabetes at SBP <130 mmHg. In particular, the Systolic Blood Pressure Intervention Trial (SPRINT) study indicated that intensive treatment with a target SBP of <120 mmHg was associated with a 25% reduction in major cardiovascular events compared with standard treatment with an SBP target of <140 mmHg (2). However, the meta-analysis indicated little benefit in lowering SBP <130 mmHg in those with diabetes (6). Also, the BP target is entirely unknown in patients with hypertension and borderline glycemia (7,8). Those findings suggest that the association of BP with cardiovascular risk should be examined by glucose status.

Furthermore, the association of DBP with cardiovascular risks according to glucose status is entirely unknown, although, unlike SBP, DBP was not an independent risk factor for cardiovascular disease in an Asian population (1). Therefore, our cohort study aimed to determine the associations of SBP and DBP with the incidence of CAD and CVD according to glucose status.

The current study analyzed data from a nationwide claims-based database that included information on 805,992 people enrolled with a health insurance provider for company employees and their dependents in Japan (9). Details of the claims data and classifications were described elsewhere (10,11). Individuals aged 18–64 years who were monitored for at least 3 years, from 1 April 2008 to 31 July 2016, were included in this analysis and continued to be monitored until 31 August 2019. Excluded were those with CAD or CVD at baseline, with type 1 diabetes, and with no health examination data including those on blood tests. Finally, this study included 589,501 individuals with no prior CAD and 593,196 individuals with no prior CVD (380,812/382,332 normoglycemia; 172,151/173,376 borderline glycemia; 36,538/37,488 diabetes). The Niigata University Ethics Committee approved this study.

Participants were classified as having normoglycemia, borderline glycemia, or diabetes defined by fasting plasma glucose (FPG), HbA_1c, and claims database data as follows: normoglycemia, both FPG <5.6 mmol/L and HbA_1c <5.7% (39 mmol/mol) and no antidiabetic drug prescription; borderline glycemia defined by FPG 5.6–6.9 mmol/L or HbA_1c 5.7–6.4 (39–46 mmol/mol), or both, and no antidiabetic drug prescription; and diabetes, FPG ≥7.0 mmol/L or HbA_1c ≥6.5% (47 mmol/mol), or both, or with an antidiabetic drug prescription regardless of FPG or HbA_1c (12). The value for HbA_1c is that used by the NGSP (13).

All facilities measured BP in accordance with the guidelines of the Japanese Society of Hypertension (14). For medical checkups, these guidelines recommended measurement of BP twice by the oscillometric method and averaging the results. Current smoking information was obtained from the questionnaire. The presence of CAD was determined according to claims using the ICD-10 codes for cardiac events but excluding heart failure and procedure codes for medical interventions, such as percutaneous coronary intervention (PCI) and coronary artery bypass grafting after 1 month of follow-up. The presence of CVD was determined according to claims using the ICD-10 codes for cerebrovascular events and procedure codes for medical interventions, such as thrombolytic therapy and endovascular recanalization after a 1-month follow-up.

Categorical variables are expressed as numbers and percentages and were compared with χ² tests. Continuous variables are expressed as median (interquartile range) and were compared using the unpaired Student t test if the normal distribution could be assumed by the Kolmogorov-Smirnov test; otherwise, the Mann-Whitney U test was used.

Unadjusted overall time to the incidence of CAD or CVD was indicated by Kaplan-Meier analysis with log-rank testing. A Cox proportional hazards regression model identified the associations of SBP with the incidence of CAD or CVD according to glucose status. Covariates included age, sex, current smoking (yes or no), BMI, LDL-cholesterol, HDL-cholesterol, and antihypertensive drug therapy (yes or no). Data were compared among 15 groups of participants divided according to combinations of glucose status and five stratified levels of SBP (i.e., ≤119, 120–129, 130–139, 140–149, and ≥150 mmHg) or DBP (i.e., ≤74, 75–79, 80–84, 85–89, and ≥90 mmHg). Analyses were performed using SPSS 19 software (IBM, Armonk, NY). Statistical significance was initially set at two-sided P = 0.05 and adjusted for multiple comparison by applying the Bonferroni comparison.

Baseline characteristics of study participants with and without CAD/CVD according to glycemic status are shown in Tables 1 and 2. Among 589,501/593,196 participants with no prior CAD/CVD, 380,812/382,332, 172,151/173,376, and 36,538/37,488 had normoglycemia, borderline glycemia, and diabetes, respectively. The proportion of participants with SBP <130 mmHg was 84% for normoglycemia, 71% for borderline glycemia, and 53% for diabetes. Of participants with SBP <120 mmHg, 61% had normoglycemia, 45% had borderline glycemia, and 26% had diabetes. The worse the glucose tolerance, the lower the proportion of achievement of the BP control target (Supplementary Fig. 2). The medium follow-up period was 5.2 years.

During the study period, 2,240 CAD and 3,207 CVD events occurred in the overall study population. Among participants with normoglycemia, borderline glycemia, and diabetes, 733, 796, and 711 CAD events and 1,478, 1,160, and 569 CVD events occurred, respectively. The incidence of CAD/CVD per 1,000 person-years was 0.34/0.69 for normoglycemia, 0.83/1.19 for borderline glycemia, and 3.52/2.74 for diabetes. The incidence of CAD/CVD per 1,000 person-years was 0.27/0.48 for individuals with SBP ≤119 mmHg, 0.80/1.02 for those with SBP 120–129 mmHg, 1.18/1.50 for those with SBP 130–139 mmHg, 1.78/2.26 for those with SBP 140–149 mmHg, and 2.49/4.04 for those with SBP ≥150 mmHg. The incidence of CAD/CVD per 1,000 person-years was 0.28/0.49 for participants with DBP ≤74 mmHg, 0.74/0.90 for those with DBP 75–79 mmHg, 1.02/1.26 for those with DBP 80–84 mmHg, 1.32/1.70 for those with DPB 85–89 mmHg, and 1.98/2.96 for those with DBP ≥90 mmHg.

Fig. 1A (left) and B (left) shows Kaplan-Meier curves for CAD/CVD according to five stratified SBP values and glycemic status. A linear relationship was observed between cumulative incidence rates and SBP categories across all glucose tolerance status designations using SBP ≤119 mmHg as the reference. Table 3 shows the adjusted hazard ratios (HRs) for the incidence of CAD/CVD according to five stratified SBP categories and all glycemic status designations. Generally, CAD/CVD risks were elevated with increases in BP categories regardless of glycemic status. Compared with SBP ≤119 mmHg, the relative risks (95% CIs) for CAD/CVD in the other 4 quintiles (120–129, 130–139, 140–149, and ≥150 mmHg) were 2.10 (1.73–2.56)/1.46 (1.27–1.68), 2.35 (1.89–2.92)/2.22 (1.91–2.58), 3.01 (2.31–3.93)/2.98 (2.46–3.60), and 3.21 (2.37–4.34)/4.76 (3.94–5.75) for normoglycemia, 1.39 (1.14– 1.69)/1.70 (1.44–2.10), 1.78 (1.45–2.18)/1.89 (1.58–2.25), 1.89 (1.48–2.43)/2.41 (1.95–2.96), and 2.52 (1.95–3.26)/4.12 (3.38–5.02) for borderline glycemia, and 1.50 (1.19–1.90)/1.72 (1.31–2.26), 1.46 (1.14–1.86)/1.56 (1.17–2.07), 1.83 (1.41–2.38)/1.99 (1.47–2.71), and 2.52 (1.95–3.26)/3.54 (2.66–4.70) for diabetes, respectively.

The cumulative incidences of CAD/CVD according to five stratified DBP categories and glycemic status are shown in Fig. 1A (right) and B (right). Table 3 shows adjusted HRs for the incidence of CAD/CVD according to five stratified DBP categories and glycemic status. Similar to the relationship between SBP and CAD/CVD, the CAD/CVD risk concurrently increased with higher DBP categories. Compared with DBP ≤74 mmHg, the relative risks (95% CIs) for CAD/CVD in the other 4 quintiles (75–79, 80–84, 85–89, and ≥90 mmHg) were 1.59 (1.25–2.01)/1.40 (1.19–1.66), 1.91 (1.53–2.38)/1.76 (1.50–2.06), 2.56 (2.02–3.24)/2.49 (2.10–2.96), and 2.99 (2.39–3.74)/4.03 (3.46–4.69) for normoglycemia, 1.30 (1.03–1.63)/1.25 (1.02–1.54), 1.43 (1.15–1.77)/1.80 (1.51–2.15), 1.46 (1.15–1.85)/2.01 (1.66–2.45), and 2.20 (1.79–2.71)/3.27 (2.76–3.87) for borderline glycemia, and 1.33 (1.04–1.71)/1.34 (1.02–1.77), 1.38 (1.09–1.74)/1.19 (0.91–1.55), 1.42 (1.10–1.82)/1.35 (1.01–1.79), and 1.79 (1.43–2.25)/2.25 (1.76–2.86) for diabetes, respectively.

Considering that using metformin and/or a sodium–glucose cotransporter 2 inhibitor (SGLT-2i) improves the cardiovascular prognosis in patients with diabetes (15–17), the association of SBP and DBP with CAD/CVD events was examined excluding individuals with diabetes who took metformin and/or a SGLT-2i at baseline (Supplementary Table 1). However, the exclusion had little influence on the above results. Taking into consideration the characteristics of hypertension in the elderly with high SBP but normal DBP values due to arterial stiffness (18), we also performed a sensitivity analysis limiting participants to those ≥65 years old. Generally, neither SBP nor DBP was associated with CAD/CVD risk regardless of glycemic status (Supplementary Table 2).

Supplementary Fig. 1 shows the incidence of CAD/CVD according to five stratified SBP and DBP values and three designations of glycemic status. The glycemic status and BP values had a synergistic effect on the incidence of CAD/CVD. Supplementary Tables 2, 3, and 4 show the adjusted risks for CAD/CVD in five groups categorized according to SBP or DBP values and three designations of glycemic status using Cox regression models. The HRs (95% CIs) for participants in the highest SBP category and with borderline glycemia and in the highest SBP category (≥150 mmHg) and with diabetes compared with those with SBP ≤119 mmHg and with normoglycemia were 3.89 (3.01–5.02) and 8.39 (6.62–10.63) for CAD, respectively, and 4.44 (3.71–5.32) and 5.59 (4.54–6.87) for CVD, respectively (Supplementary Table 3). The HRs (95% CIs) for those in the highest DBP category (≥90 mmHg) and with borderline glycemia and in the highest DBP category (≥90 mmHg) and with diabetes compared with those with a DBP ≤74 mmHg and normoglycemia were 3.37 (2.74–4.15)/6.55 (5.26–8.14) for CAD and 3.73 (3.21–4.34)/4.77 (3.95–5.77) for CVD (Supplementary Table 3).

The main finding of the current cohort study was a linear relationship between SBP and DBP and the risk of CAD or CVD regardless of glucose status, although with one exception: individuals with diabetes and SBP of 120–129 mmHg had a higher risk of CAD/CVD than those with SPB of 130–139 mmHg. In particular, the CAD/CVD risk was higher in participants with SBP of 120–129 mmHg (i.e., lower than SBP targets for hypertension in major global guidelines [i.e., 130 mmHg] [19,20]) compared with those with SBP <120 mmHg. A plausible explanation for this finding is that an increase in the presence of coronary artery calcium was in parallel with an increase in SBP levels beginning from 90 mmHg (21).

The novelty of our study findings was that cardiovascular risks were gradually elevated in accordance with increases in both SBP and DBP values and that a linear relationship between BP and cardiovascular risks was observed not only in the general population but also when the population was designated as having normoglycemia, borderline glycemia, and diabetes. For example, NIPPON DATA 80 showed such linearity only in the general population (1), and a Swedish cohort study showed it only in patients with diabetes (22).

Our results did not support a previous cohort study (23). That Chinese population-based study showed that BP of 130–139/80–89 mmHg was significantly associated with a higher cardiovascular risk compared with BP of <130/80 mmHg in individuals with diabetes but not with normoglycemia or borderline glycemia. However, it did not assess cardiovascular risk in participants with prehypertension (i.e., 120–139 mmHg) compared with normal BP (≤120 mmHg). It is possible that the heterogeneity in cardiovascular risk among participants with SBP ≤130 mmHg resulted in the inconsistency in the findings of that cohort study and ours.

The current results suggested that strict BP control is necessary to prevent cardiovascular diseases in patients with borderline glycemia and diabetes as well as in normoglycemic individuals. However, that has not been recommended for patients with diabetes by previous trials (6,24,25). Although it may be difficult to compare findings from cohort studies with those from interventional trials, the reasons for inconsistencies are of value for discussion. Most previous trials targeted patients at particularly high cardiovascular risk among those with diabetes (3,26). It is speculated that such patients would develop cardiovascular diseases due to risk factors other than BP (e.g., poor glycemic control, obesity), even if strict BP control had been achieved. It was suggested that randomized controlled trials should target those frequently encountered patients with borderline glycemia or diabetes and hypertension who are not considered at high risk for cardiovascular disease to confirm the findings of our cohort study and apply them to clinical practice.

Our study strengths included the large sample size and accurate definitions of CAD and CVD based on data from medical practices, thereby allowing precise identification of almost all patients with incident CAD and CVD during the follow-up.

However, several limitations should be addressed. First, unfortunately, this database did not have data on several confounders. As examples of confounders without which cardiovascular risks could be overestimated, we should address the estimated glomerular filtration rate and urinary protein excretion because these values are associated with both high BP and cardiovascular events (27,28). Although we adjusted the risks for antihypertensive drugs using dichotomous data on BP treatment, the intensity of BP interventions was not considered. For example, the current study showed that in patients with diabetes, the incidence of CAD and CVD was more frequent with SBP of 120–129 mmHg than in those with SBP of 130–139 mmHg. This is not consistent with the linear relationship between BP values and cardiovascular risks observed among normoglycemia and borderline glycemia conditions. It is possible that the intervention for BP may have been enhanced in the 120–129 mmHg group based on guidelines for the treatment of diabetes. However, our study was incapable of adjusting for this reversal phenomenon. In addition, we did not distinguish nonsmokers from past smokers although they had different cardiovascular risks, depending on length of time after smoking cessation (29). However, the effect of combining nonsmokers and past smokers on the current result is unpredictable.

Second, as explanatory variables, we only used baseline values and did not consider subsequent changes in BP or details of treatment for various metabolic abnormalities. In particular, the failure to consider BP changes during the follow-up period could underestimate the true cardiovascular risk in relation to high BP values.

Third, there could be misclassification bias in determining glucose status for two main reasons: 1) we did not have data on oral glucose tolerance testing, meaning some individuals with normoglycemia might have been classified as borderline glycemia or diabetes, and some with borderline glycemia would have been classified as having diabetes; 2) the values for HbA_1c had been changed from Japanese Diabetes Society to NGSP during the study period, which would mean that the diagnosis of normoglycemia, borderline glycemia, or diabetes depended on the time of measurement.

Lastly, because our study was composed of people enrolled with a health insurance provider for company employees and their dependents, a limited number of elderly persons was recruited. The number of elderly (aged ≥65 years) was 11,401 (1.9% of total participants). The numbers of CAD and CVD events in the elderly group were 151 (only 6.3% of CAD events in the entire group) and 228 (only 6.6% of CVD events in the entire group), respectively. In contrast with the NIPPONDATA 80 (1) showing that SBP but not DBP was associated with cardiovascular risk in the elderly (≥75 years), we could indicate the association of neither SBP nor DBP with CAD/CVD risk in participants aged ≥65 years (Supplementary Table 2). It is possible that the number of data were insufficient to evaluate the significance of BP in the elderly. Further studies are needed to examine the effect of age on the association between BP and cardiovascular risks throughout a range of glycemic status.

In conclusion, the current cohort study indicated that cardiovascular risks gradually increased with increases in both SBP and DBP values regardless of the glucose abnormality. Further trials to examine strict BP interventions for preventing cardiovascular diseases should focus on patients with borderline glycemia and diabetes to confirm findings from the current cohort study and apply them to clinical practice.

Acknowledgments. The authors thank Ms. Mami Haga, Niigata University Faculty of Medicine, for excellent secretarial assistance, and Mitsuru Hashiramoto, Mitsuru Clinic, for warm encouragement.

Funding. This work is supported by JMDC Inc., the Japan Society for Promotion of Science(JSPS), and the Ministry of Health, Labour and Welfare.

The sponsor had no role in the design and conduct of the study.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. M.H.Y. and K.F. developed the study design, researched the data, contributed to discussions, and wrote, reviewed, and edited the manuscript. S.K., T.S., T.O., Y.Y., M.Y., M.K., Y.M., T.Y., H.Se., and W.O., and researched the data, contributed to discussions, and wrote, reviewed, and edited the manuscript. H.So. developed the study design, contributed to discussions, and reviewed and edited the manuscript. M.H.Y. and K.F. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 80th Scientific Sessions of the American Diabetes Association, 12–16 June 2020.