Association of Optimal Blood Pressure With Critical Cardiorenal Events and Mortality in High-Risk and Low-Risk Patients Treated With Antihypertension Medications

Key Points Question Is the optimal level of treated blood pressure (BP) associated with different rates of critical cardiorenal events and mortality in high- vs low-risk patients? Findings In this population-based cohort study of more than 1 million adults in Korea, the absolute increase in cardiorenal and mortality risk with inadequately treated BP was larger in patients with more risk factors, and the mortality threshold of treated BP was left-shifted in patients with multiple vs 1 or 0 risk factors. Meaning These findings support the need for individualized treatment of high BP considering more strict targets for higher-risk patients.


eMethods 1. National Health Information Database (NHID)
The NHID, a public database for the whole population of South Korea, was established and is being maintained by the National Health Insurance Service (NHIS). Details on the NHID are published elsewhere. 1,2 The NHID covers data from 2002 onwards, and comprises the following. (a) eligibility data: demographics, income based insurance contributions, and date of death (b) health screening records: medical history, health behavior, physical exam, and laboratory exam (c) reimbursement records of the NHIS: prescribed drugs, medical procedures, outpatient visits, hospitalizations, and lists of medical diagnosis (d) health care provider data: medical institutions, equipment, and human resources The data in each resource was assembled using de-identified join keys, which replace personal identification numbers assigned to citizens of Korea.
Nationwide health screenings are performed for citizens aged ≥40 years, generally at 2-year intervals, in hospitals or medical centers. During the health screening, information on medical history and health behaviors are obtained using standardized questionnaires. Trained medical staff perform physical examinations including blood pressure (BP) measurement. Blood and urine samples are obtained after at least an 8-h fast.
Serum creatinine and high-density lipoprotein (HDL) and low-density lipoprotein (LDL) cholesterols have been measured from 2009 health screening.

eMethods 3. Covariates
Outlier data were excluded from the health screening records. (a) systolic BP <90 mm Hg or >200 mm Hg (b) diastolic BP <30 mm Hg or >140 mmHg (c) blood glucose <30 mg/dl or >900 mg/dl (d) total cholesterol <130 mg/dl or >320 mg/dl (e) HDL cholesterol <20 mg/dl or >100 mg/dl (f) serum creatinine <0.3 mg/dl or >15.0 mg/dl (g) body mass index <10 or >50 Treated and untreated BP records were collected separately. (a) if antihypertensive prescription ≥90 days in the year of BP measurement, the BP → treated BP (b) if not, the BP → untreated BP Income levels were determined by income based insurance contributions. The amounts of alcohol consumption were calculated as the number of drinks averaged per day.
In each year of follow-up, a new average was calculated for each time period as follows. (a) primary cohort: 2002-2005, 2002-2006, 2002-2007, 2002-2008, 2002-2009, 2002-2010, 2002-2011, 2002-2012, 2002-2013, and 2002-2014 (b) secondary cohort: 2006-2009, 2006-2010, 2006-2011, 2006-2012, 2006-2013, 2006-2014, 2006-2015, and 2006-2016 Using the yearly updated values, the variables were categorized as follows. In each year of follow-up, the status of health conditions were determined. (a) diabetes mellitus: yes or no (b) hyperlipidemia: yes or no (c) proteinuria: yes or no (d) smoking: never, former, or active smoker (e) antihypertensive compliance: regular, irregular, or never use Antihypertensive compliance was determined as follows. Initiation of antihypertensive treatment was defined as the first year when antihypertensive drugs were prescribed for ≥90 days per year. Among cases that initiated treatment, regular use was considered when antihypertensives were prescribed for >half of each follow-up period (from initiation of treatment to each year of follow-up), irregular use was when the prescription was ≤half of each follow-up period, and never use when the prescription was never for ≥90 days per year.
Using baseline data, the variables were categorized as follows.

A. Determination of risk factors in both cohorts
Five risk factors (hypertension, diabetes mellitus, hyperlipidemia, proteinuria, and active smoking) were identified, using the results of health screenings and information on the prescription of drugs. Information on the prescription of drugs in the reimbursement records were captured using NHIS billing codes (eTable 2).

B-2. Calculation of WHO/ISH risk scores in the primary cohort
The scores were calculated using a single comma delimited file extracted from WHO/ISH cardiovascular risk assessment charts. 5 C. In the secondary cohort, the 10-year risk of cardiovascular disease was calculated with following variables.

C-1. Calculation of local risk scores in the secondary cohort
The scores were calculated according to a Korean prediction model, 6 which was developed on the basis of 2013 ACC/AHA risk score.

eMethods 5. Outcomes
Previous studies using the disease codes listed in NHIS reimbursement records have reported that 73%~93% of the disease codes for myocardial infarction or stroke are valid. 6,8 In addition to the disease codes, I used information on revascularization procedures, prescribed peritoneal dialysates, hemodialysis, and kidney transplantation, which were captured by NHIS billing codes, to identify outcomes more reliably.

A. Critical cardiorenal event
Critical cardiorenal event was identified through December 31, 2015 in the primary cohort and through December 31, 2107 in the secondary cohort. (a) Information on critical care unit admission were captured using NHIS code for admission (NHIS clause code, 02) to critical care unit (NHIS item code, 03). Critical care unit admission from cardiorenal diseases (ICD-10, I00-I99 and N18) was verified with the primary medical diagnosis listed in reimbursement records. (b) Information on revascularization procedures of coronary, cerebral, and carotid arteries were captured using NHIS billing codes (eTable 3). Revascularization for myocardial infarction (ICD-10, I21-I22) or stroke (ICD-10, I63-I64) was verified with the primary diagnosis listed in reimbursement records. (c) Information on hemodialysis (NHIS billing code, O7020 and O9991), prescribed peritoneal dialysates (eTable 3), and kidney transplantation (NHIS billing code, R3280; and ICD-10, Z94.0) were captured from NHIS reimbursement records. Using the information, end-stage kidney disease with dialysis for ≥90 days per year or kidney transplantation was verified.

B. All-cause death
All-cause death was confirmed through December 31, 2015 in the primary cohort and through December 31, 2017 in the secondary cohort, using information on date of death, which was included in eligibility database of the NHID.

eFigure 1. Flow Charts of Participant Selection in The Primary (A) and Secondary (B) Cohorts eFigure 5. Yearly Event Rates in Prevalent (A and C) or Recent (B and D) Antihypertensive Users
The 1-year rates were estimated by multiplying the combined hazard ratios by the combined mean of the age specific rates in the reference group (systolic BP, 120-129 mm Hg). The summary effects and 95% CIs of the primary and secondary cohorts were calculated by using the DerSimonian-Laird random-effects model. All analyses were adjusted for age, sex, family history of cardiovascular disease, income level, smoking, alcohol consumption, exercise frequency, body mass index, diabetes, hyperlipidemia, and proteinuria. The critical cardiorenal event was a composite of admission to critical care unit with cardiovascular or chronic kidney disease, revascularization for myocardial infarction or stroke, and new onset end-stage kidney disease. Error bars indicate 95% CIs. BP, blood pressure.

eFigure 7. Yearly Event Rates in Risk Categories Grouped by Risk Factors After Exclusion of Proteinuria
The risk categories were grouped by the number of risk factors present at baseline after exclusion of proteinuria: i.e., ≥3, 2, ≤1 of the four risk factors (hypertension, diabetes, hyperlipidemia, and smoking). Among a total of 487,412 primary cohort participants, 34,050 (7.0%), 110,023 (22.6%), and 343,339 (70.4%) had ≥3, 2, and ≤1 risk factors, respectively. Among a total of 915,563 secondary cohort participants, 65,631 (7.2%), 188,669 (20.6%), and 661,263 (72.2%) had ≥3, 2, and ≤1 risk factors, respectively. The 1-year rates were estimated in the primary (A and C) and secondary (B and D) cohorts, by multiplying the hazard ratios by the mean of the age specific rates in the reference group (systolic BP, 120-129). The analyses were adjusted for age, sex, family history of cardiovascular disease, income level, smoking, alcohol consumption, exercise frequency, body mass index, diabetes, hyperlipidemia, and proteinuria, and further adjusted for antihypertensive compliance (regular use, irregular use, and nonuse). The critical cardiorenal event was a composite of admission to critical care unit with cardiovascular or chronic kidney disease, revascularization for myocardial infarction or stroke, and new onset end-stage kidney disease. Error bars indicate 95% CIs. BP, blood pressure.