Study design and population
This retrospective cohort study was conducted in a health check-up cohort population from the Korean National Health Insurance Service (NHIS) database. In South Korea, populations aged 40–79 years participated in the biennial national health screening programme covered by the Korean National Health Insurance cooperation. This cohort group is a simple random sample representing 10% of the 5.15 million populations who participated in the national health screening programme from 2002 to 2003.
Among the 514,844 participants, 73,404 participants aged above 65 years were excluded to reduce the confounding effects of extreme old age on the development of COPD; 3,857 were excluded owing to a diagnosis of COPD at the time of the health check-up; and 15,131 were excluded due to diagnoses of liver disease such as viral liver disease, non-alcoholic fatty liver disease, and hepatocellular carcinoma at the time of the health check-up, to limit the confounding effects of liver disease on ALT levels. Finally, 422,452 participants were included in the analysis (Figure 1).
Participants were divided into five groups, based on serum ALT levels at baseline: group 1, ALT (IU/L) < 10; group 2, 10 ≤ ALT < 20; group 3, 20 ≤ ALT < 30; group 4, 30 ≤ ALT < 40; and group 5, ALT ≥ 40. COPD development was retrospectively compared among all groups for 13 years.
Parameters included in data collection
The basic demographic data included participants’ age and sex, and insurance claim data included the dates of hospital visits, diagnostic codes encoded by the ICD-10, and prescriptions. Charlson’s comorbidity index was calculated based on comorbidities using ICD-10 codes [33]. Health check-up data provided the results of the health check-ups conducted in 2002 or 2003 and included the results of body mass index (BMI; kg/m2) and systolic and diastolic blood pressure (mmHg) measurements, as well as hemoglobin (g/dL), fasting blood glucose (mg/dL), total cholesterol (mg/dL), aspartate aminotransferase (AST; IU/L), ALT (IU/L), and gamma-glutamyl transferase (ɤGT; IU/L) levels. Moreover, health check-up data included a self-report questionnaire, which the participants answered during the health check-up, providing information regarding smoking status (never smoker, ex-smoker, and current smoker) and physical activity (0, 1–2, 3–4, 5–6 times/week, and almost every day).
Definition of COPD development
As the NHIS database does not include data from pulmonary function tests, we defined COPD development based on the diagnostic codes of the ICD-10 and prescriptions contained in the insurance claim data, as suggested by previous studies [34-36]. COPD development was determined when all of the following criteria were met: 1) age above 40 years; 2) ICD-10 diagnostic code of COPD or emphysema (J43.0x–J44.x, except J43.0, as the primary or secondary diagnosis [within the fourth position]); 3) the use of COPD medications, including a muscarinic antagonist, beta-2 agonist, inhaled corticosteroid, phosphodiesterase-4 inhibitor, or methylxanthine; and 4) first diagnosis of COPD in the follow-up period.
Definition of liver disease
Liver diseases were defined using diagnostic codes. The following ICD-10 codes were used in this study to exclude the effects of these underlying liver diseases on ALT levels: K70.x–K77.x, disease of liver; B15.x–B19.x, viral hepatitis; E83.0, Wilson disease; and C22.0, liver cell carcinoma.
Statistical analysis
To compare baseline characteristics across groups, chi-squared tests were used for categorical variables, and paired t-test or one-way analyses of variance with Bonferroni post-hoc tests were used for continuous variables. Univariable and multivariable Cox regression analyses were performed to evaluate the hazard ratio (HR) of each variable for COPD development. A subgroup analysis was performed according to sex and smoking status. Kaplan-Meier curves were drawn to evaluate the cumulative incidence of COPD among the groups according to serum ALT levels. P-values < 0.05 were considered to indicate statistical significance. Statistical analyses were conducted with R (version 3.3.3; R Foundation for Statistical Computing, Vienna, Austria) software. Furthermore, spline curves of the HRs for COPD development were drawn using the pspline R package, and correlation coefficient plot were drawn using the corrplot R package.