Study design and setting
This prospective cohort study was in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology statement. This study was conducted at a tertiary referral hospital in Japan (Hyogo Prefectural Amagasaki General Medical Center). The study protocol and other results have been published previously [14, 15]. The study was conducted in accordance with the Declaration of Helsinki. All patients provided written informed consent before participation.
Participants and variables
The eligibility criteria were as follows: (i) patients who were pathologically diagnosed for lung cancer from April 2018 to March 2020, (ii) stage IIIA or higher based on the 8th edition of the American Joint Commission on Cancer and Union for Cancer Control tumor node metastasis (TNM) classification stage [16], (iii) patients who were considered unsuitable for radical surgery (to avoid omission of data with surgery), (iv) patients who had not been treated for lung cancer in the past two months (since there were some cases in which urgent treatment was started before the patient was judged for inclusion, we included the patients within two months after the start of treatment), and (v) written informed consent was obtained from the patient for participation. The exclusion criteria were as follows: (i) inability to complete the questionnaire due to dementia or psychiatric illness, and (ii) those considered unsuitable for the study by the physician.
Exposures were defined as social isolation and loneliness. Social isolation was assessed using the Japanese version of the Lubben Social Network Scale (LSNS-6) [17]. Since the Japanese version did not have a threshold, we used the cut-off value of the English version (12 points out of 30) [17]. Loneliness was assessed using the third Japanese version of the University of California, Los Angeles (UCLA) Loneliness Scale (3-12 points) [18]. The UCLA scale was revised and converted into three questions to facilitate patient responses [19]. Since the scale did not have a clear cutoff point, participants were categorized into quartiles (low to high loneliness) to create a relatively homogeneous group.
The primary and secondary outcomes were OS and death at home, respectively. Potential confounders included sex, age, smoking status, and insurance status (welfare or other). Education status, employment status, dementia, Eastern Cooperative Oncology Group performance status, histology, TNM classification stage [16], targetable driver gene mutation, programmed death-ligand 1 (PD-L1) tumor proportion score, and initial treatment. Dementia was assessed using the Life Function Evaluation for Care Provision [20].
Statistical analysis
The protocol for this study required the enrollment of 300 patients. However, due to inadequate enrollment, recruitment was withdrawn two years later. To reduce information bias, the clinical research coordinator conducted the data collection.
The Kaplan-Meier method was used to estimate OS, and the differences were evaluated using the log-rank test. Cox proportional-hazards models were used for the univariate and multivariate analyses of the prognostic factors. Odds ratios and confidence intervals with death at home were calculated using a logistic regression model. We conducted a complete case analysis and multiple imputation to analyze the missing data for exposure and outcome variables and combined 20 imputed datasets using Rubin's rule [21]. Statistical significance was set at P <0.05.