Data source
The HIRA database examined herein includes information on healthcare services for approximately 50 million people (i.e., more than 98% of the Korean population). This database is the primary data source through which healthcare insurance costs are claimed and reimbursed and includes detailed information regarding patients demographics, the treating medical institutions, disease codes, treatment history, and prescription drugs. To protect personal privacy, all data are encrypted and only variables approved by the HIRA are granted limited distribution to registered investigators either at the designated analysis center or via prespecified static Internet Protocol addresses.
Study design and population
This retrospective study assessed prescription patterns as well as safety and efficacy of MA prescribed to patients with metastatic cancer registered in the HIRA database. Patients newly diagnosed with gastric, colorectal, and pancreatobiliary cancer between July 1, 2014, and December 31, 2015 (i.e., the index period) were followed for up to 36 months (i.e., the follow-up period) after their date of diagnosis (i.e., the index date) (Online Resource 1). The following inclusion and exclusion criteria were applied to select patients with metastatic cancer: 1) patients with diagnostic codes for gastric (C16), colorectal (C18, C19, C20), biliary (C23 and C24), and pancreatic (C25) cancer, classified according to the International Classification of Diseases (tenth revision; ICD-10), as well as the supplementary code of the copayment decreasing policy (V193), which records data on nearly all cancer patients in Korea; 2) no prior diagnostic code of cancer during the last 12 months prior to the index date (i.e., the baseline period) to reduce the risk of misclassifying secondary metastatic tumors are primary cancers and to exclude complicated presentations and comorbidities; and 3) patients with a diagnostic code of metastatic cancer (C76-C79.9) received within 90 days after the index date. Patients for whom MA was prescribed during the baseline period, those with another diagnostic code of cancer recorded during the follow-up period, and those aged less than 18 years were excluded from this study. Additional exclusion criteria were used to identify patients with metastatic cancer more accurately, as follows: 1) patients treated with curative surgery within 180 days after the index date (Online Resource 2), as well as patients receiving concurrent chemoradiotherapy (CCRT; operationally defined as more than 10 fractions of radiotherapy concurrently administered during one or more days of chemotherapy within 35 days following the first date of radiotherapy); 2) patients who survived for more than two years after the index date who did not receive any active anticancer therapies, including surgery, radiotherapy, and chemotherapy (these cases might be misclassified in the database because it is rare that patients with metastatic cancer survive long-term without active anticancer therapy). This study was approved by the Institutional Review Board of Gyeongsang National University Changwon Hospital (GNUCH 2019-11-027) and was conducted in accordance with the principles of the Declaration of Helsinki and its later amendments. Informed consent was waived due to the retrospective nature of the study.
Definitions and assessments
The prescription patterns for MA evaluated herein included the presence of MA prescriptions (any vs. none), the time from the index date to the first prescription, the total days of the prescriptions and the average daily dose, and prescription continuity. Patients who were prescribed MA at least once and those who were never prescribed MA during the follow-up period were grouped into MA prescription and non-MA prescription groups, respectively. The average daily dose of MA was calculated as the total number of prescriptions divided by total days of prescription. The continuity of prescription was assessed using the medication possession ratio (MPR), which was calculated as the total number of days of prescription divided by the days elapsing between the first and last prescription date. MPR values ≥0.8 and <0.8 were defined as continuous and intermittent prescriptions, respectively. Patient characteristics including demographic information, insurance and hospital type, modified Charlson Comorbidity Index (mCCI) values, and treatment status were abstracted and compared according to MA prescription pattern status.
VTE was defined as a case diagnosed with deep vein thrombosis (DVT, ICD-10 code I80 and related codes) or pulmonary thromboembolism (PTE, I26 and related codes). If the patient was diagnosed with VTE, treated with VTE-associated procedures, and/or received anticoagulation during the baseline period, this case was excluded from the VTE assessment. The incidence of VTE was compared based on the prescription patterns for MA, with adjustment for covariates. Death was indicated if there was an insurance claim in which a death-related code was recorded during the follow-up period, and the death date was determined as the expiration date within the national health insurance system. In this study, we evaluated the impact of MA prescription patterns on survival.
Statistical analysis
Categorical and continuous variables were compared using chi-square and Fisher’s exact tests and independent t-test or Wilcoxon rank-sum tests as appropriate (i.e., according to the distribution of the data). Logistic regression was performed with adjustment for baseline characteristics to elucidate the factors affecting MA prescriptions and VTE occurrence. Survival curves were plotted using the Kaplan-Meier method and compared using the log-rank test. Time-dependent Cox regression was performed to assess hazard ratios (HR) and associated 95% confidence intervals (CI) for death-associated covariates. Missing data were excluded from this analysis. A two-sided p-value of less than .01 was considered statistically significant given the large sample size. All statistical analyses were performed using SAS statistical software (version 9.4; SAS Institute, Cary, NC, USA).