Data sources
We accessed a one million population-based sample data recorded by the Health Insurance Review and Assessment (HIRA) service of South Korea in 2018. This claim database holds all health care information from both inpatients and outpatients using codes from the Korean Standard Classification of Diseases (KCD), 7th revision with a few changes related specifically to Korea based on the International Classification of Diseases (ICD), 10th revision. The novelty, as well as detailed information regarding the HIRA database, has been provided in our previous study.1,2
The current nationwide cohort study complied with the Declaration of Helsinki and was approved by the Institutional Review Board of the Myongji Hospital, South Korea (IRB No. 2020-10-040). The requirement for informed consent was waived by the Institutional Review Board of the Myongji Hospital, South Korea due to the retrospective design of the study and the anonymous nature of the data.
Participants and sample selection
All patients who had possible AC (registration code H101 - H103) during the 1-year study period and had a continuous enrollment of two independent visits were included in our estimates. All enrollees with these registration codes were confirmed and registered by an ophthalmologist to ensure AC diagnosis. Furthermore, those who had a history of keratoconjunctivitis (H1620), keratoconjunctivitis sicca (H1621), neurotrophic keratoconjunctivitis (H1622), phlyctenular keratoconjunctivitis (H1624), vernal keratoconjunctivitis (H1625), and any other or unspecified keratoconjunctivitis (H1628-9) were excluded. Subjects with AC who were finally enrolled in the study were referred to a combination of diagnosis code and simultaneous prescription of eye drops for AC.
To help exclude conditions that can affect the development of hordeolum, those who had any history of acne (L700-719), blepharitis (H011-019), atopic/seborrheic skin disease (L208-219), leishmaniasis (B551, B559, Q909), or any other infectious conditions, including herpes zoster (B023), tuberculosis (A185), diphtheria (A368), meningococcus (A398), gonococcus (A543), trachoma (A71) chlamydia (A74, H131), measles (B05), acanthamoebiasis (B60), and any other bacterial/viral conjunctivitis (B300-309) were excluded.
For the prevalence estimates, the date of the earliest claim along with their registration code was defined as the index date, and the patient was considered as an incident case for that year. Moreover, most of the patients in this database were Asian. Characteristics such as age, sex, medical visit records, date of diagnosis, and comorbidities were obtained from the HIRA database.
Analyses
Prevalence
We categorized the pediatric and adolescent patients into eight groups based on four strata of age as well as sex. Considering the degree of cooperation of patients who can perform the slit lamp examination, patients under 3 years of age were excluded. The pediatric and adolescent patients were classified into strata of age between 3-5, 6-9, 10-12, and 13-19 which had been established by HIRA system. Those diagnosed with AC within a year of the study were also included in the prevalence estimates. The monthly distribution of AC diagnosis, as well as incision surgeries, were estimated and noted.
Establishing cohort and ascertainment of outcome
The current cohort data was established to evaluate the risk of incision surgery in patients with AC. Based on the exposure of AC diagnosis, the AC cohort and non-AC cohort were established for the final analysis. All incidental AC cases were enrolled in the AC cohort, and subjects who met clinicians due to diseases other than AC were obtained from the same period and included in the non-AC cohort, and exclusive diseases were excluded. To establish a one-year cohort and remove potential preexisting cases of surgery, the first two months (January and February 2018) were set as a wash-out period, with identified cases of incision surgery before the diagnosis of AC excluded.
To identify possible episodes of incision surgery, the patients were linked to the outpatient and inpatient records using an encrypted personal identification key. In the current cohort, patients who newly underwent incision surgery (S5400, S5250) for hordeolum or chalazion after the index date of AC diagnosis were determined as incident cases. Moreover, we also censored the follow-up time at the end of this study for anonymity. (December 31, 2018).
Statistics
Data handling and statistical analyses were performed by an independent data analyst (J. L.) specially trained by the HIRA institute for their 2017 HIRA big data. A comparison of continuous variables between the groups was performed using the paired t-test, and a comparison of the proportion of each variable between the groups was analyzed using the Chi-squared test. The stratified log-rank test was used to compare the incidence rates of surgery between the AC and non-AC groups. Cox proportional hazards regression with a cluster effect was used to compute an adjusted hazard ratio for the two groups. A confidence level of 95% was used for this analysis, and all results are presented as mean ± standard deviation (SD). P-value less than 0.05 indicated statistical significance. SAS Enterprise Guide version 6.1 software (SAS Inc., Cary, NC, USA) was used for all analyses.
Data availability
The raw data used in this study can be requested from any qualified investigator through the national HIRA system.