Long-term cardiovascular events in hypertensive patients: full report of the Korean Hypertension Cohort

Study design and data collection

The KHC study enrolled 11,043 patients with hypertension who visited tertiary university hospitals for the first time and was had hypertension as a primary diagnosis from 2006 to 2011. The study design has been published previously [12]. The primary composite outcome of this study was all cause death and major cardiovascular events including heart failure, myocardial infarction, peripheral arterial disease, hemorrhagic stroke, ischemic stroke, and end stage renal disease. Moreover, sample size was calculated based on the risk reduction of SPRINT to determine the role of intensive BP lowering compared to conventional BP control, although the target BP goals were modified from those in the SPRINT [13].

The eligibility criteria included patients with hypertension who were regularly followed up at least three times out of four time points (1, 2–6, 7–12, 13–24 months) for a minimum of 2 years with laboratory evaluation and prescribed antihypertensive medications. Exclusion criteria were described in the design paper [12]. We followed up using 2-year hospital records linked with the NHIS-HEALS data from 2004 to 2018.

Furthermore, the study also included a control population without hypertension at a 1:10 ratio (100,000 persons) from the NHIS-HEALS sample cohort sex- and age-matched case-to-case [14]. Hypertension was excluded by filtering the primary and additional disease codes. Among the NHIS-HEALS sample cohort, the control population was selected according to the patients who underwent the national health examination more than twice to sufficiently compare cardiovascular risk factors. The flow diagram of the KHC is summarized in Fig. 1.

Figure 1

Flow diagram of Korean Hypertension Cohort. NHIS-HEALS, National Health Insurance System-health screening; NHE, National health examination.

The entire analysis process of the linked data sources was conducted in the NHIS data analysis office. External exportation of the data and additional extraction of personal information were not permitted. The authors are restricted from sharing the data underlying this study because The Korean NHIS owns the data. The study protocols were reviewed and approved by the Institutional Review Boards of each participating hospital, including the coordinating center (Seoul National University Hospital H-1810-056-977; Seoul National University Bundang Hospital B-1901-519-409). The dataset was anonymized to the researchers by the NHIS and the requirement for informed consent was waived.

Definition of cardiovascular diseases and risk factors

Hypertension was defined as an SBP/DBP ≥ 140/90 mmHg without antihypertensive treatment or taking antihypertensive treatment. Clinical cardiovascular diseases, subclinical organ damage, and cardiovascular risk factors were defined according to the 2018 Korean Society of Hypertension guidelines for the management of hypertension [15]. The high cardiovascular risk group was defined as patients with clinical cardiovascular disease, chronic kidney disease, or diabetes mellitus. The moderate cardiovascular risk group was defined as patients with increased vascular stiffness, left ventricular hypertrophy, or metabolic syndrome excluding diabetes mellitus. Increased vascular stiffness was defined as the presence of atherosclerotic plaques, increased pulse wave velocity, or a decreased ankle-brachial index. The low cardiovascular risk group was defined as patients who did not belong to either the high or moderate risk groups. For the NHIS claim data analysis, operational definitions of cardiovascular diseases were followed as in Choi [16].

Blood pressure measurement

BP was primarily measured by clinicians using mercury sphygmomanometers more than twice after 3 to 5 minutes of rest using standard technique. Clinic BP data were collected at baseline, and after 1, 2–6, 7–12, and 13–24 months. If the patients visited multiple times during each time interval, BP values most adjacent to the time point (1, 6, 12, 24 months) were collected. Furthermore, BP measured in the NHIS-HEALS was collected at years 3 and 5 if hypertensive patients received a national health examination. If the examination results at years 3 and 5 were unavailable, the average BP values at the 2-year interval were collected.

Statistical analysis

The baseline characteristics, including demographics, physical and laboratory findings, risk factors, and management are presented as frequencies and percentages or means and standard deviations. Clinical characteristics between the hypertensive patients and non-hypertensive controls were compared using the chi-squared test for categorical variables and Student’s t test for continuous variables. The incidence rates were estimated using the total number of outcomes during the follow-up. For the cardiovascular outcome evaluation, we used a Cox proportional risk model between-group comparisons. Hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs) were calculated from Cox proportional hazards analysis. Time to occurrence of the cardiovascular events was analyzed using Kaplan-Meier curves [17]. Comparisons were considered statistically significant if two-tailed p values were less than 0.05. Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA).

Baseline characteristics of the cohort

The baseline characteristics of the cohort population according to hypertension status are shown in Table 1. The mean age was 59 years, 50.6% of the patients were male, 30.4% received no previous hypertension treatment, and 34.8%, 16.5%, and 48.7% belonged to the high, moderate, and low cardiovascular risk groups, respectively (Supplementary Table 1). The initial BP was 148.0 ± 18.5/88.0 ± 15.4 mmHg. The average number of antihypertensive medications was 1.82 ± 0.90 tablets per day. Calcium channel blockers and angiotensin receptor blockers were prescribed to nearly 60% of the hypertensive population, and 34% were prescribed statins and aspirin. In the control group without hypertension, the initial BP was 121.0 ± 15.4/78.0 ± 10.0 mmHg.

Table 1

Baseline characteristics

Cardiovascular events according to hypertension status

A total of 1,591 cardiovascular events (14.4%) with 588 deaths (5.3%) in the population with hypertension and 7,635 cardiovascular events (6.9%) with 4,826 deaths (4.4%) in the control group without hypertension occurred during a follow-up period of 10.0 years (3,529 ± 790 days in the population with hypertension and 3,567 ± 570 days in the control group without hypertension) (Table 2). Cardiovascular event rates in the population with hypertension were more than twice those in the population without hypertension (Supplementary Table 2). The HR was the greatest within 1 year (5.61; 95% CI, 4.83 to 6.52) but was maintained at 5 years (HR, 2.75; 95% CI, 2.59 to 2.93) and 10 years (HR, 1.85; 95% CI, 1.77 to 1.93) of follow-up (Supplementary Table 3). All individual cardiovascular events invariably occurred at a higher rate in the population with hypertension (Fig. 2, Supplementary Table 4) than in the controls. Furthermore, higher rates of cardiovascular events and death were associated with a higher cardiovascular risk at baseline (Table 2), and even the low-risk population with hypertension showed a higher cardiovascular event rate than the population without hypertension. Two-year and 5-year cardiovascular event rates are presented in Supplementary Tables 5 and 6.

Table 2

Ten-year CV events according to the baseline cardiovascular risk

Figure 2

Cumulative incidence of cardiovascular events in the hypertensive group and non-hypertensive controls. (A) Heart failure. (B) Myocardial infarction. (C) Peripheral arterial disease. (D) Hemorrhagic stroke. (E) Ischemic stroke. (F) End stage renal disease.

Blood pressure changes during follow-up

To minimize the measurement error in the clinic setting, we averaged the BP measurements after 6 months to 2 years of follow-up (6 months, 1 year, and 2 years). BPs in the population with hypertension continuously decreased during 2 years of hospital data follow-up and the NHIS-HEALS data follow-up after that. Most reductions in SBP and DBP occurred in the first 6 months and maintained thereafter. SBP were decreased from 149.4 ± 18.4 to 130.6 ± 13.8 mmHg and DBP were decreased from 88.6 ± 12.5 to 78.0 ± 9.0 mmHg, significantly (Supplementary Fig. 1A). There was no significant difference in BP reduction between men and women (Supplementary Fig. 1B). However, younger aged groups showed consistently higher BP throughout the study period, although the BP difference between age groups reduced (Supplementary Fig. 1C). Remarkably, SBP was reduced in similar levels across cardiovascular risk groups. However, DBP was lower in the high-risk group compared with the lower risk groups, although the extent of DBP reduction was greater in the lower risk group than that of the high-risk group (Supplementary Fig. 1D). Overall, 79% of hypertensive patients had controlled their SBP below 140 mmHg after 6 months of the enrollment (Supplementary Table 7). After 2 years of hospital data follow-up, we observed that SBP and DBP consistently decreased in the NHIS-HEALS data (Supplementary Fig. 2). A total of 1,649 (15%) patients with hypertension received the NHIS-HEALS adjacent within 3 months of the clinic BP measurement at year 2. Therefore, we could indirectly compare the difference between the clinic BP and the NHIS-HEALS BP. Interestingly, in the BP range 130–139/80–89 mmHg, the clinic and NHIS-HEALS BP were almost identical. However, in the higher BP ranges, the clinic BP tended to be higher, and in the lower BP ranges, BP tended to be in a lower range than those recorded in the NHIS-HEALS (Supplementary Fig. 3).

Cardiovascular events according to BPs in patients with hypertension

Next, we evaluated the incidence of cardiovascular events according to clinic BPs after 6 months of enrollment. The patients with hypertension were initially divided into six SBP groups of < 120, 120–129, 130–139, 140–149, 150–159, and ≥ 160 mmHg and into DBP groups of < 60, 60–69, 70–79, 80–89, 90–99, and ≥ 100 mmHg (Supplementary Tables 7 and 8). However, the lowest and highest groups were small in number. Therefore, we merged the two lowest groups and the two highest groups creating four SBP and four DBP groups for the analysis. Interestingly, age was positively correlated with SBP but showed a negative correlation with DBP. In the unadjusted survival analysis, higher SBP ≥ 150 mmHg was associated with a higher rate of cardiovascular events. SBP of 140 to 149 mmHg showed a modestly higher incidence of combined cardiovascular events and death than the lower two SBP groups (Fig. 3A–3C). In contrast, lower DBP (< 70 and 70–79 mmHg) was significantly associated with a higher rate of cardiovascular events than the higher two DBP groups (Fig. 3D–3F). Then, we analyzed the incidence of cardiovascular events according to pulse pressure quintiles. Although the largest pulse pressure group showed higher event rates, the difference between each pulse pressure groups were not clearly demarcated compared with SBP or DBP groups (Supplementary Fig. 4). Because SBP and DBP are not independent values in each individual, we presented the cardiovascular event rates as 16 combinations of SBP and DBP groups. In subgroup analysis, the combined effect of SBP and DBP on cardiovascular events was higher in the elderly, the high-risk hypertension patients than the non-elderly, and the moderate to low-risk hypertension patients, respectively (Fig. 4).

Figure 3

The combined effect of systolic and diastolic blood pressure on cardiovascular events. (A) Major cardiovascular events or death according to systolic blood pressure (SBP) categories. (B) Major cardiovascular events according to SBP categories. (C) Death according to SBP categories. (D) Major cardiovascular events or death according to diastolic blood pressure (DBP) categories. (E) Major cardiovascular events according to DBP categories. (F) Death according to DBP categories. ^aStatistically significant difference compared with other groups (p < 0.001).

Figure 4

Combined effect of systolic and diastolic blood pressure on cardiovascular events according to age, gender, and blood pressures.

Limitations

First, the KHC is not a clinical trial but a patient cohort that used linked patients’ hospital data and NHIS-HEALS data. Therefore, patient assessments, management, and follow-up regimens were not standardized and varied between the included institutions. Notably, the BP measurement methods were not standardized but followed the values from each hospital. Moreover, although we performed the multivariate analysis to minimize the confounding factors, we cannot rule out the possibility that errors from non-random allocations may affect the analysis results, such as those regarding the optimal visit intervals. Second, we only registered patients from six national university hospitals, which are therefore tertiary clinics, because the current Korean Personal Information Protection Act permits only data linking national hospitals and NHIS-HEALS data.

Moreover, moderate to high-risk hypertensive patients were targeted for enrollment to obtain sufficient outcome data for a limited follow-up period; however, we planned to follow the patients for 10 years to the best of our ability. Therefore, the KHC did not represent the general Korean population with hypertension. As a results, the cardiovascular events rate observed in this cohort might be higher than the actual event rates in general hypertensive patients in Korea. However, the event rate was separately presented according to the patient’s cardiovascular risk level, which might be equally applicable to community clinics. Furthermore, most enrolled hypertensive patients were gradually referred back from the tertiary hospital to community clinics during follow-up. Therefore, the treatment results may be more affected by the patient’s intrinsic risk rather than differences in treatment patterns in tertiary hospitals, if present.

Conclusion and perspectives

The KHC study provides important information on the long-term outcomes according to the patients’ BP and cardiovascular risk. Our study shows that the cardiovascular event rates among Korean hypertensive patients were substantial, reaching 15% in 10 years of follow-up. Although clinic BP measurement in the real-world showed remarkable inaccuracy compared with those measured in the national health examinations, higher SBP ≥ 150 mmHg presented significantly higher cardiovascular events. Several ancillary studies using the KHC are now underway, which will ultimately help establish HT management strategies reflecting Asian ethnic characteristics.