Study overview and setting
We conducted a population-based, propensity score-matched cross-sectional and longitudinal study. For recruitment for the cross-sectional study, information was advertised online and through a dedicated questionnaire. For the longitudinal study, the anxiety and distress state before and during the COVID-19 pandemic were surveyed. As of February 24, 2020, a total of 135 LC patients and 165 non-lung cancer controls from two independent medical centers were included in the study. The study protocol was reviewed and approved by the Research Ethics Board of the First People’s Hospital of Yunnan Province [reference no. 20200009], and written informed consent was obtained from all participants. The reporting of this study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement (eTable 1).
Subjects and recruitment
The baseline data of primary LC patients were collected from November 2019 to December 2019 at the First People’s Hospital of Yunnan Province (Kunming City, China) and Daping Hospital (Chongqing City, China) for another study (data not shown). Inclusion criteria were 1) lung cancer diagnosis (all kinds of pathological types); 2) no psychological illness or family history of it; 3) age ≥20 years; and 4) ability to complete the questionnaire independently. The non-lung cancer subjects (NLC) were recruited from volunteers, and the exclusion criteria were 1) posttraumatic stress disorder, obsessive-compulsive disorder, or major depression; 2) current treatment with psychotropic medication; 3) age <20 years; 4) multiple chronic difficulties associated with learning and/or conduct problems; 5) any concurrent psychotherapy. When COVID-19 has broken out in China, all of the participants were recruited again through email or phone for the present study.
Propensity score matching analysis
PSM analysis was used to minimize selection bias and balance variables. Propensity scores for all patients were estimated by a logistic regression model using age, gender, religion, income, and education level as covariates, since these factors may affect the psychological characteristics of LC patients[13-15]. A one-to-one nearest-neighbor matching algorithm with a caliper of 0.2 and without replacement was used[16, 17]. PSM analysis was performed using SPSS software (v24.0,).
Instruments and measures
State Anxiety Inventory (SAI). State anxiety is the temporary and changeable feeling induced by the arousal of the autonomic nervous system (e.g., how a person is feeling at the time of a perceived threat); the SAI measures this feeling using 20 items rated on a 4-point scale (1, “none”; 2, “mild”; 3, “moderate”; 4, “severe”). Total SAI scores can range from 20 to 80. SAI scores ≥44 were considered severe anxiety. Both LC patients and NLC completed the SAI.
Chinese Distress Thermometer and problem list. The distress thermometer (DT) is a screening tool recommended by the National Comprehensive Cancer Network (NCCN). Since there are racial differences in responses to the DT, the validated Chinese version of the DT (CDT) was used to screen patients’ levels of distress ranging from 0 (no distress) to 10 (extreme distress). Patients were instructed to circle the number (0-10) that best described how much distress they had been experiencing recently. A score of 4/10 indicates clinically significant distress and warrants additional evaluation (NCCN, 2017). Analysis of the Chinese version indicated that a cut‐off score of 4/10 has a sensitivity rate of 0.80 and specificity of 0.70. Unlike the SAI, the CDT was used to evaluate distress status for LC patients only.
The original problem list was developed by a Distress Management Guideline Panel of the NCCN and consists of 39 problems commonly experienced by cancer patients. The Chinese version of the problem list was adapted from the NCCN and includes the original 39 problems plus 1 item, relationship with healthcare providers[27, 28]. Patients were asked to mark “yes” or “no”’ to indicate whether they had experienced each problem in the past week.
According to the epidemic data released by the Chinese Center for Disease Control and Prevention (http://www.chinacdc.cn/), the initial 23 days was the rapid uptrend period of the COVID-19 epidemic (cohort effects peaked around February 9); then the epidemic entered the decline period (Figure 1). Therefore, we conducted assessments on participants who were included in this study during the uptrend and decline periods (Figure 2).
The demographic and clinical characteristics, CDT score, SAI scores, and problems were described using frequencies, percentages, and measures of central tendency and dispersion. Problem totals were calculated by creating frequency scores for the number of items checked in each category. The SAI score was modeled as a count variable (ranging from 20 to 80). For within-group differences, a group× time interaction was assessed using two-way repeated measures ANOVA with the outcome as dependent variable and group (LC and NLC) and time (Baseline, Uptrend period and Decline period) as independent variables (fixed effects) . Bonferroni post hoc tests were used if interactions were detected. ESs were calculated using Hedge’s g for repeated measures. Effect size (ES) of 0.00–0.19, 0.20–0.49, 0.50–0.79, and ≥0.80 represented trivial, small, moderate, and large effects, respectively. Mean differences and their 95% confidence intervals were also calculated. Besides, we used Student’s t-test to assess the difference in SAI scores between LC patients and NLC, and a Mann-Whitney U test was performed for skew distribution. Pearson’s chi-squared test was used to determine the proportion of severe anxiety (SAI scores ≥44) and severe distress (CDT score ≥33) at different time points of the COVID-19 outbreak.
Based on the frequency results, worry (yes vs no) was the most commonly endorsed emotional problem on the problem list in the uptrend and decline periods of COVID-19; logistic regression analyses with worry as the dependent variable were performed to test whether any demographic or clinical variables were the best predictors of worry.
Statistical analyses were performed with SPSS (version 23.4, IBM Corp., Armonk, NY) and GraphPad (version 8.3), with a two-sided P<0.05 considered statistically significant for all reports.