Long-term air pollution exposure is associated with higher incidence of ST-elevation myocardial infarction and in-hospital cardiogenic shock

Study protocols and population

The study subjects were enrolled in the Korea AMI registry (KAMIR) and KAMIR-National Institutes of Health (NIH). The KAMIR study protocol has been introduced previously¹⁴. KAMIR and KAMIR-NIH are nationwide prospective multicenter registration study series that aim to establish treatment guidelines and derive risk factors through the analysis of various clinical characteristics and follow-up of Korean AMI patients since October 2005 onwards. A flowchart of the study is shown in Fig. 1. A total of 50,130 patients with AMI were enrolled in the KAMIR and KAMIR-NIH between January 2006 and December 2015. The exclusion criteria were as follows: (1) date of symptom onset before 2006, (2) missing date of symptom onset, (3) age < 18 years, and (4) no final diagnosis of myocardial infarction (MI) at discharge.

Ethical approval

This study was approved by the Institutional Review Board (IRB) of Korea University Guro Hospital (KUGH, #2016GR0740) and was conducted in accordance with the principles of the Declaration of Helsinki. Prior to giving written consent to participate, the participants or their legal guardians received a thorough and detailed explanation of the study procedures, both in written and verbal form.

AP measurement

Hourly AP concentrations were provided by the Korean Ministry of Environment (http://www.airkorea.or.kr). In 2001, 329 monitoring stations nationwide began measuring the concentration of air pollutants. Measurement of air pollutants involved the β-ray absorption method for particulate matter (PM) 10 µm or less in diameter (PM₁₀), the non-dispersive infrared method for carbon monoxide (CO), the pulse ultraviolet fluorescence method for sulfur dioxide (SO₂), the chemiluminescence method for nitrogen dioxide (NO₂), and the ultraviolet photometric method for ozone (O₃). The concentration measurement of PM 2.5 µm or less in diameter (PM_2.5) began in January 2015; therefore, annual average concentration values were not available during the patient enrollment period (2006–2015) and was excluded.

We transformed collected data into the daily average value, and then, the annual average value of air pollutants before the symptom day was calculated the way previous research was performed¹³. Each monitoring station was matched by the closest distance in a straight line to 68 hospitals registered in KAMIR to measure individual exposure concentration of air pollutants. Monitoring stations were selected based on hospital admission addresses for the following reasons: (1) Patient addresses were not included in the multicenter registry data. (2) As AMI is an emergency, it is assumed that the patient was admitted to an emergency room close to the workplace and residence at the time of symptom onset. If a pollutant measurement was missed due to a connection error with a monitoring station, the measurement of the next-nearest monitoring station was assigned. Symptom date was defined as the first occurrence of MI-related symptoms such as chest pain or dyspnea.

Study definitions

The diagnosis of AMI was defined as an elevation in cardiac biomarkers (creatinine kinase-MB, and troponin I, or T) with typical changes on 12 leads electrocardiogram (ECG) or clinical symptoms. STEMI was diagnosed as a new ST-elevation segment measuring ≥ 1 mm from ≥ 2 contiguous leads on ECG. Patients with positive cardiac biomarkers but without ECG findings of STEMI were defined as NSTEMI. Cardiogenic shock was defined as a systolic blood pressure < 90 mmHg for > 30 min, the need for supportive management to maintain systolic blood pressure > 90 mmHg, and clinical signs of pulmonary congestion. A complication of cardiogenic shock is defined as its new onset after admission.

Individual cardiovascular risk factors, including hypertension (HTN), dyslipidemia (DL), diabetes mellitus (DM), prior cardiovascular disease, heart failure (HF), prior cerebrovascular disease (CVA), family history of CAD, and smoking history, were based on self reports by the patient.

Statistical analysis

All statistical analyses were performed using R version 4.1.2. (R Core Team, 2021; R: Language and Environment for Statistical Computing; R Foundation for Statistical Computing, Vienna, Austria, URL: https://www.R-project.org/).

We compared the clinical and angiographic characteristics using a X² test or Fisher’s exact test for categorical variables and Student’s t-test or Mann–Whitney rank test for continuous variables. In our analysis, X² tests were used for categorical variables with expected cell frequencies of five or more; otherwise, Fisher’s exact test was applied. Continuous variables were analyzed with Student’s t-test if data were normally distributed (assessed by the Kolmogorov–Smirnov test) and with the Mann–Whitney rank test for non-normal distributions. Categorical data were expressed as percentages, and continuous variables were described as mean ± standard deviation.

We used generalized logistic mixed effect models with a random effect term for hospitals to examine the associations of each air pollutant with the incidence rate of STEMI and cardiogenic shock complication rates, and to account for hospital and regional effects such as accessibility and treatment plans. All variables used in the models for the incidence of STEMI and cardiogenic shock complications are presented in Table 1.

Using a multivariable model, we adjusted for potential confounding factors for STEMI incidence, including age, sex, body mass index (BMI), smoking status, HTN, DM, DL, stroke, HF, previous IHD, and family history of CAD. To analyze the incidence of cardiogenic shock complications, we considered the factors previously mentioned, in addition to STEMI status, percutaneous coronary intervention (PCI), and left ventricular ejection fraction (LVEF).

To assess and mitigate the risk of collinearity, we conducted correlation analyses and variance inflation factor (VIF) assessments among the included air pollutants. The VIF values obtained were below the commonly used threshold of 4, indicating that collinearity was unlikely to significantly impact the results of our regression analyses.

In the subgroup analysis, we conducted several stratified analyses using interaction terms for each specified group. For the STEMI group analysis, we included the following terms respectively: age, sex, HTN, DM, DL, CVA, HF, prior IHD, smoking, family history of CAD. In the cardiogenic shock group, STEMI status, PCI, and LVEF were added. The results were presented as adjusted odds ratios (OR) for logistic regression with corresponding 95% confidence intervals (CI). Statistical significance was defined as a p-value < 0.05.