Data source

This study leveraged the database maintained by the Korea Disease Control and Prevention Agency (KDCA) and NHIS (Research No. KDCA-NHIS-2023-1-610). The NHIS in Korea provides coverage to 97% of the entire population and is responsible for collecting, processing, and summarizing all associated healthcare records. This dataset, primarily intended for academic research, comprises comprehensive demographics, diagnoses, prescriptions, and health examination data. The KDCA furnishes detailed COVID-19 information — including the route of infection, date of confirmation, and date of death — which is then integrated with the original data by the NHIS.

Study population and settings

This retrospective, population-based cohort study included data for all adults with psoriasis who underwent health screening between 2018 and 2021. Psoriasis was defined based on the following International Classification of Diseases, 10th revision, codes (appearing at least twice prior to the index date): L40, M07.0–M07.3, or M09.0. Propensity score matching was conducted at a 1:5 ratio to enroll and define controls. The exclusion criteria were: a SARS-CoV-2 infection route originating from abroad, partial vaccination, or infection with SARS-CoV-2 within 14 days following full vaccination¹⁰. To mitigate the risk of identifying early pandemic cases, the NHIS only provided detailed information for cases occurring after 8 October, 2020. Thus, we excluded individuals diagnosed before this date.

Operational definition of outcome

The primary outcome was the first SARS-CoV-2 infection. The secondary outcomes were severe complications associated with the infection, including death, application of invasive ventilation or oxygen therapy, and admission to the intensive care unit (ICU). The index date for the primary outcome was set as October 8, 2020, and the endpoint was determined as either the date of COVID-19 diagnosis or December 31, 2021. Regarding secondary outcomes, the starting date was the day of COVID-19 diagnosis; the endpoint was the day a severe outcome occurred, either within 60 days or 60 days after the starting date¹¹. To mitigate potential bias resulting from individuals experiencing similar outcomes unrelated to COVID-19, we excluded patients with records of such outcomes within 5 days prior to their COVID-19 diagnosis⁸. To improve the statistical power, we included additional COVID-19 cases in the secondary analysis, utilizing the expanded dataset on COVID-19 cases. The detailed information is presented in Fig. 1.

Flowchart illustrating the data collection process.

Operational definition of the treatment

Systemic treatment for psoriasis encompasses phototherapy, cyclosporine, methotrexate, acitretin, and biologics (including adalimumab, etanercept, guselkumab, infliximab, ixekizumab, secukinumab, and ustekinumab; Supplemental Table 1). Treatment thresholds (e.g., ≥ 4 prescriptions for methotrexate, ≥ 2 injections for biologics) were determined based on clinical prescribing patterns in Korea and intended to reflect sustained and therapeutically relevant exposure. Although specific cutoffs are not standardized across studies, our definitions align with those used in observational pharmacoepidemiologic research and are consistent with reimbursement and prescription frequency patterns in real-world clinical settings. Furthermore, we only included patients who sustained the same level of treatment throughout the follow-up period. Those who underwent multiple types of treatments, or who first started treatment after the index date, were excluded from the analysis. Patients who had never received any of the aforementioned systemic treatments were categorized into the nonsystemic group.

Covariates

The covariates included age, sex, insurance level, residence (categorized as metropolitan area vs. other regions), body mass index, smoking status, Charlson comorbidity index, COVID-19 vaccination status (categorized as unvaccinated, fully vaccinated, or booster shot vaccinated), and comorbidities. Comorbidities included asthma, cardiovascular disease, chronic kidney disease, chronic obstructive pulmonary disease (COPD), cerebrovascular disease, diabetes mellitus, and hypertension. Throughout the study period, metropolitan areas experienced a substantial increase in COVID-19 cases, leading to the implementation of different policies — such as social distancing measures — compared with other regions. Therefore, we categorized residences into two groups: metropolitan areas and other regions. The vaccine immunity period was defined as 6 months after full or booster shot vaccinations¹². In the secondary analysis, we incorporated additional information regarding the timing of the SARS-CoV-2 infection, considering that disease severity might be influenced by limited hospital resources during peak COVID-19 outbreaks. To address baseline imbalances and potential confounding, multivariable Cox regression models were adjusted for all listed covariates, including comorbidities. Furthermore, propensity score matching (1:5) was applied prior to analysis based on age and sex to reduce selection bias. Residual confounding due to imbalances in other characteristics was addressed through additional covariate adjustment in the multivariable models.

Statistical analysis

We evaluated the baseline characteristics of all covariates using Pearson’s chi-squared test for categorical variables, and the Student’s t-test for continuous variables. Cox proportional hazard regression was utilized to compare clinical outcomes, with the vaccine immunity period considered a time-varying covariate. The proportional hazard assumption was verified to be met.

To assess differences in the incidence of COVID-19 during the vaccine immunity period among various subgroups, we conducted an interaction analysis using the extended self-controlled case series (SCCS) method. As a case-only design, the SCCS method enabled the estimation of the relative incidence (RI) between periods with and without a vaccine effect. Given the importance of reducing selection bias in studies where the vaccine is the main factor¹³, the SCCS method was the design of choice. Additionally, considering the observed tendency to avoid vaccination right after COVID-19, a 90-day pre-exposure washout period was implemented.

All statistical analyses were performed using R software (version 4.0; R project, Vienna, Austria), SAS Enterprise Guide (version 7.15; SAS Institute Inc., Cary, NC, USA), and Rex (version 3.5.3; RexSoft, Seoul, Republic of Korea)¹⁴. Statistical significance was set at P < 0.05.

Ethical approval

This study was approved by the Institutional Review Board of the Konkuk University Medical Center (KUMC 2021-12-033), and a waiver of informed consent was granted as this study used de-identified data. All methods were performed in accordance with relevant guidelines and regulations, as outlined in the Declaration of Helsinki.