Study Design

This was a retrospective cohort study conducted using the DPC database. Data for patients meeting the inclusion criteria were extracted from the database. Data for patients meeting the exclusion criteria were then excluded. Patients were categorized into either the experimental (ABCP) cohort or the control (BCP) cohort. Differences between the cohorts in terms of the incidence of AEs were examined and further analyzed using survival analysis methods.

This study was conducted in compliance with the principles of the Declaration of Helsinki and the Ethical Guidelines for Medical and Health Research Involving Human Subjects (Notification No. 1 of the Ministry of Education, Culture, Sports, Science and Technology/Ministry of Health, Labour and Welfare, dated February 28, 2017). The study was approved by the Ethical Committee of the Non-Profit Organization MINS Institutional Review Board (200220) and is registered in the UMIN Clinical Trials Registry (ID: UMIN000041507). Informed consent was not required as this is a database study using anonymized data that are devoid of any patient-identifying information per the Japanese ethical guidelines.

Inclusion Criteria

Patients were required to have an entry for NSQ-NSCLC in their medical records. NSQ-NSCLC was defined using the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10) codes. Patients with disease codes in category C34 (malignant neoplasm of bronchus and lung), excluding those considered not to be NSQ-NSCLC, were defined as eligible for this study. The list of excluded diseases is shown in Supplementary Materials Table S1. Each patient was required to have a medical record before the start of the observation period.

Exclusion Criteria

Of the patients who met the inclusion criteria, those who met any of the following criteria were excluded: (1) patients who had been diagnosed with a malignant tumor other than NSQ-NSCLC during the time period in which data were extracted, and (2) patients who had undergone chemotherapy (except with epidermal growth factor receptor tyrosine kinase inhibitors) for NSQ-NSCLC during the baseline period.

Data Source

A healthcare database owned by Medical Data Vision Co., Ltd. (MDV; Tokyo, Japan) was used as the data source. This database contains information from medical institutions using the DPC payment system. Approximately 28,440,000 patients from 385 medical institutions were registered in this database at the time of the study.

Definitions

The ABCP and BCP cohorts were defined based on the drugs prescribed in the month of the data extraction and the inclusion of NSQ-NSCLC in the records during this month as defined in the inclusion criteria. Drugs were defined using the Anatomical Therapeutic Chemical (ATC) classification system. The ATC codes for each drug are as follows: ABCP cohort: L01XC32, atezolizumab; L01XC07, bevacizumab; L01XA02, carboplatin; and L01CD01, paclitaxel. BCP cohort: L01XC07, bevacizumab; L01XA02, carboplatin; and L01CD01, paclitaxel.

Data Extraction Period

For the ABCP cohort, data from December 2018 (the month in which ABCP treatment was approved in Japan for unresectable advanced or recurrent NSQ-NSCLC) to March 2020 (the most recent data available for use) were used. For the BCP cohort, data from December 2015 to March 2020 were used to allow for a sufficient number of patients to be included in the study. However, no specific period was determined for extracting the baseline data. All data from before the start date of the observation period were used to extract the covariate information for the patients. Patients without a baseline period were also included in this study because it was assumed that after the initial NSQ-NSCLC diagnosis in routine clinical practice, treatment would have been started relatively quickly for most patients following their referral to the medical institutions included in the DPC database. If an imbalance was observed in the duration of the baseline period between cohorts, the duration of the period before the start of the observation period was checked in each cohort, as this could have resulted in a bias or confounding of the results.

First Day of the Observation Period

For each patient included in the study, the first day of the observation period was defined as the date of the earliest medical institution visit on which ABCP or BCP treatment was prescribed and a disease corresponding to NSQ-NSCLC was recorded.

Assessment of AEs

The incidence of the following 19 AEs was evaluated: interstitial lung disease (ILD), hepatic dysfunction, colitis/diarrhea, pancreatitis, type 1 diabetes mellitus, thyroid disease, adrenal disorder, hypopituitarism, meningitis/encephalitis, neuropathy, Guillain–Barré syndrome, myasthenia gravis, dermatopathy, renal dysfunction, myositis/rhabdomyolysis, cardiac disorder, febrile neutropenia, arthritis, and eye disorder.

The above AEs were defined based on the ICD-10 codes corresponding to diseases in the health insurance claims extracted from the MDV database. Similarly, the treatments (i.e., ABCP, BCP, or other) were defined based on the ATC codes. The ICD-10 and ATC codes are provided in Supplementary Materials Tables S2 and S3.

Statistical Analyses

The incidence of AEs was evaluated using incidence rates (case per person-month), and differences between the ABCP and BCP cohorts were compared using incidence rate ratios.

The following covariates were adjusted using the inverse probability weighting (IPW) method: age (< 75 years or ≥ 75 years), sex, body mass index (BMI), presence of ILD before treatment, steroid use, activities of daily living score (100 or not), TNM classification 1 (T4 or not), and TNM classification 2 (M1 or not). Propensity scores were estimated using a logistic regression model with these covariates as the explanatory variables, and the weighted population was created using the normalized stabilized weights calculated based on the propensity scores.

In addition to the adjusted incidence rates, adjusted Kaplan–Meier curves were generated and the adjusted restricted mean survival times (RMSTs), with 12 months as the truncation time, were derived based on the Kaplan–Meier curves. Results from two measures (i.e., incidence rate ratio and RMST difference) ensured robustness of the results.

Clinically Significant AEs

For each of the 19 AE categories, clinically significant adverse events (CSAEs) were extracted and analyzed. Two definitions were used for the analysis of CSAEs: (1) if the patient, after starting treatment for the AE, was not prescribed ABCP or BCP by the planned date of prescription (43 days after the day on which ABCP or BCP was prescribed, immediately before the AE occurrence), then the corresponding AE was considered a CSAE, and (2) of the AEs meeting definition 1, if hospitalization was required after the occurrence of the AE, with the main reason for the hospitalization being the AE, then the corresponding AE was considered a CSAE. Separate analyses were performed using each of the two definitions.

Treatment of AEs

The number and proportion of patients who experienced AEs and received treatment for the AEs were analyzed. The analyzed treatments included drugs prescribed for the treatment of the AE, drugs in the L04 (immunosuppressants) category of the ATC classification, and other drugs and medical procedures of clinical significance. Of these, treatments administered in the same month as when the AE was documented were used for the analyses.