Study population and data collection
The data for this retrospective cohort study were obtained from a single regional emergency medical center: Hanyang University Guri Hospital, Guri, Gyeonggi-do, Korea. The study design and primary results were previously published4. The medical records of 172,105 patients who visited the ED between January 2016 and December 2019 were reviewed; 16,404 patients who had elevated initial SBP ≥ 180 mmHg or DBP ≥ 100 mmHg were enrolled in this study (Fig. 1). Of these, patients aged < 18 years who presented with acute trauma, certificates, or who visited the ED multiple times were excluded from this study. Among these 10,219 patients with acute severe hypertension, 4867 had available data regarding BMI. The National Emergency Department Information System (NEDIS) data were used to identify eligible patients who visited the ED. The system collects data, including initial vital signs and baseline clinical characteristics. This information on all patients visiting the ED is automatically transmitted from each hospital to a central government server called NEDIS. The patients were classified into five groups according to BMI level as follows; underweight group whose BMI < 18.5 kg/m2, normal group whose BMI between 18.5 kg/m2 and 22.9 kg/m2, overweight group whose BMI kg/m2 between 23 and 24.9 kg/m2, obese class I group whose BMI between 25 kg/m2 and 29.9 kg/m2, obese class II + III group whose BMI ≥ 30 kg/m2. The study was conducted in accordance with the Declaration of Helsinki, and was reviewed and approved by the Institutional Review Board of Hanyang University Guri hospital (GURI 2020-01-028). The institutional review board of Hanyang University Guri hospital waived the requirement for written informed consent.
Data were collected using electronic medical records by experienced data collectors under principal investigator supervision. The following demographic and clinical characteristics were extracted: age, sex, initial BP at the ED, and traditional cardiovascular risk factors, including a history of hypertension, diabetes, dyslipidemia, chronic kidney disease (CKD), end-stage renal disease, smoking and alcohol consumption status. History of heart failure, ischemic stroke, hemorrhagic stroke, coronary artery disease, and peripheral artery disease was also extracted.
The following laboratory data were extracted: estimated glomerular filtration rate (eGFR) (mL/min/1.73 m2), B-type natriuretic peptide (BNP), and proteinuria in a dipstick urinalysis. In addition, diagnostic test findings to determine the presence of acute hypertension-mediated organ damage (HMOD) were obtained.
The primary outcome was the incidence of all-cause mortality during the 3-year follow-up (until March 15, 2021) according to the BMI level. The secondary outcomes were all-cause mortality within 1-month, 3-months and 1-year follow-up, hospital admission, and death at the ED according to BMI level. The incidence of mortality and its timing were extracted from the National Health Insurance Service in South Korea, and other outcome data were extracted from local medical records.
Obesity and overweight were defined according to the WHO guidelines for the Asia Pacific region and the Korean Society for the Study of Obesity Guideline for the management of obesity12. Acute HMOD was defined as hypertensive encephalopathy, cerebral infarction, intracerebral hemorrhage, retinopathy, acute heart failure, acute coronary syndrome (ACS), acute renal failure, and aortic dissection13. Hypertensive encephalopathy was defined as a severe increase in BP associated with other unexplained lethargy, seizures, cortical blindness, and coma13. Cerebral infarction and intracerebral hemorrhage were defined on the basis of neurological symptoms and brain imaging at the ED. Fundus examination confirmed retinopathy based on the presence of flame-like hemorrhages, cotton wool spots, or papilledema. Proteinuria was defined as a dipstick urinalysis result of ≥ 1 +14. BP was measured in the ED above the brachial artery using an automated BP machine, Spot Vital Signs LXi (Welch Allyn, Skaneateles Falls, NY, USA).
All categorical data are presented as numbers and percentages, while statistics for continuous variables are presented as means and standard deviations. The Cochran–Mantel–Haenszel test was used to extract the trend of clinical outcomes, including mortality, hospital admission, and death at the ED, and categorical data of baseline characteristics according to BMI level. Linear regression was used to show the distribution of the means and the trend of continuous variables of baseline characteristics according to BMI. Kaplan–Meier survival analyses and log-rank tests were used to compare the cumulative survival probability according to BMI level. The restricted cubic spline curve showed a continuous adjusted association between BMI and the risk of all-cause mortality in patients with acute severe hypertension. Hazard ratio (HR) and respective 95% confidence intervals (CIs) for 3-year all-cause mortality in the underweight, overweight, and obese groups compared with the normal BMI group were calculated using multivariable Cox proportional hazards regression analyses. We investigated the association between BMI level and 3-year all-cause mortality, with and without adjustment for the selected confounders. Three models were used. In Model 1, age and sex were considered as possible confounders. Model 2 included factors which were thought to be clinically relevant (age, sex, and social history) and medical history with p < 0.05 at baseline, as follows: age, sex, smoking, alcohol consumption, and medical history of hypertension, diabetes mellitus, dyslipidemia, ischemic stroke, hemorrhagic stroke, chronic kidney disease, and end-stage renal disease. Model 3 included variables with p < 0.2 in the univariable Cox proportional hazards regression analyses, as follows: age, sex, smoking, alcohol consumption, and medical history of hypertension, diabetes mellitus, dyslipidemia, ischemic stroke, hemorrhagic stroke, coronary artery disease, heart failure, chronic kidney disease, and end-stage renal disease. To evaluate the risk factors for 3-year all-cause mortality of each BMI category, we performed the multivariable Cox proportional hazards regression analyses using the variables with p < 0.2 in the univariable analysis of each group. Additionally, we performed a subgroup analysis with the multivariable Cox proportional hazards regression analyses according to the age group (< 50, 50–59, 60–69, and ≥ 70 years), presence or absence of a history of diabetes mellitus, and patients without chronic disease, including heart failure, ischemic stroke, hemorrhagic stroke, and end-stage renal disease. Among the candidate variables, social history and medical history were missing for some patients. History of alcohol consumption and cigarette smoking was missing for 62 (1.3%) and 65 patients (1.3%), respectively, and end-stage renal disease, which had the highest missing data in the medical history, was missing for 31 patients (0.6%). We conducted the Cox proportional hazard regression with complete-case analysis, which involves restricting the analysis to individuals with no missing data. We tested the proportional hazards assumption for each variable in the multivariable models using the supremum test. All variables in the models were found to satisfy the proportional hazards assumption. All tests were two-tailed, and statistical significance was set at p < 0.05. All analyses were performed using the Statistical Analysis Software (version 9.4; SAS Institute, Cary, NC, USA).