Data Analysis

Demographic and survey responses were examined using frequency and percentages for categorical variables, and mean and standard deviation for continuous variables. The 95% CIs for anxiety and burnout rates were estimated using the Clopper-Pearson method. Differences between the three participant groups were examined in bivariable analyses using χ² tests. Univariable and multivariable logistic regression analyses were performed to assess the association between participant characteristics and anxiety/burnout/fears. Association of HCWs’ survey responses with the presence of burnout were assessed (controlled for the demographic variables). For questions related to fears, exploratory factor analysis was performed for the combined patient and caregiver dataset, and this showed a two-factor structure: general COVID-19 fears and COVID-19 effect on cancer fears. These two factors showed good internal consistency, with a Cronbach α of .92 and .93, respectively. The responses of questions from each factor were summed and divided by the number of questions to reflect the level of general COVID-19 fears and COVID-19 effects on cancer fears. For HCWs, questions under the factor general COVID-19 fears were analyzed in the same way (α = .92). Point estimates were reported with corresponding CIs, which were not adjusted for multiple comparisons. Proportions of missing data were reported. There was no prespecified statistical analysis plan; however, an a priori hypothesis was specified at the time of questionnaire development. Statistical analyses were performed using R, version 3.6.3.²⁵