A cross-sectional, anonymous survey was conducted from April 22 to June 7, 2019. All patients aged 18 years or older and attending ambulatory services were eligible for inclusion. Patients could receive active treatment (chemotherapy, immunotherapy, targeted therapies, or radiotherapy) or attend as part of a follow-up visit. Patients without a cancer diagnosis were excluded. The study was conducted at the Departments of Clinical Oncology, Zealand University Hospital and the Department of Oncology, Odense University Hospital, Denmark.
The questionnaire was comprised of 91 items in total; 17 study-specific questions on sociodemographics, cancer disease, treatment, lifestyle and comorbidity, number of prescription medicines (> 5 prescription medicines was defined as polypharmacy), 7 study-specific questions on cannabis use, and the EORTC-QLQ-C30 (C30), the EORTC-CIPN20 module (CIPN20), the GAD7 and the PHQ9 questionnaires [3, 12–14]. An internal multidisciplinary group revised and validated the survey questions and the questionnaire was reviewed by a board of patient and relative representatives.
A pilot study (N=14) was conducted to evaluate the feasibility of the questionnaire as well as the validity, relevance and wording of the questions, using retrospective think-aloud interviews .
The survey was distributed within standard opening hours usually 8:30 AM to 3.00 PM, for a three-week period at each department. A serial number identified each survey. Surveys were tallied daily. Response rate was calculated by dividing the total surveys distributed with the total amount of completed surveys collected plus patient refusals. Patient refusal were not questioned, but when given freely, patients primarily refused because of time constraints and a few due to language barriers. Patients were assisted with completion of the questionnaires in cases of vision problems or paralysis/paresthesia.
Exposure and outcomes
A summary score was calculated based on items 1-18 of the CIPN20 consistent with recent studies and findings showing poor psychometric performance of the subscales [9, 16, 17]. A total score of 30 points or more on the CIPN20 score was used to divide patients into a high score and low score group [16, 17] using the high score as our definition of PN. We allowed for up to two missing items. Missing item values were imputed with the patient average score (preferentially rounded down), resulting in 34 patient score imputations.
The C30 scores, including five functioning scores, nine symptom scores, a global health score and the C30 Sum Score were calculated based on the EORTC 3rd Edition Manual .
Summary scores were calculated for the PHQ9 and GAD7. The PHQ9 and GAD7 summary scores were divided into a categorical variable of two levels using a cutoff point equal to or above 10 points [19, 20]. At this cutoff, the likelihood ratio for the presence of a major depressive disorder is 7.1 with a sensitivity and specificity of 88% , and Spitzer et al reported in a study that most patients (89%) with general anxiety disorder (GAD) had GAD-7 scores of 10 or greater, whereas most patients (82%) without GAD had scores less than 10 . We allowed for scoring of respondent total score for up to two missing items . Average imputed scores were rounded down. This procedure was completed for 34 patient GAD7 scores, and 106 patient PHQ9 scores.
We used a two-sample, unpaired Wilcoxon test to test median difference in age and Pearson’s chi squared test to test factors containing two categorical variables for proportional equality by CIPN20 high scores such as gender (male/female), active smoking (yes/no), active treatment (yes/no), cohabitation status (living with partner/living alone), presence of each comorbidity (yes/no), type of active treatment (yes/no), GAD7 (non-case/case) and PHQ9 (non-case/case). For age groups (< 30, 30-49, 50-64, 65-80, >80), BMI categories (< 18, 18-25, 25-30, > 30), education (mandatory school, upper secondary/vocational education, short higher education, medium-length higher education, long higher education), alcohol overconsumption (> 5 units at same occasion/daily, -/weekly, -/monthly, -/less than monthly, -/never), (never, less than monthly, monthly, weekly, daily) and number of prescription medicines (0-3, 4-5, > 5) we used a binomial general linear model, presenting estimates as odds ratios (OR). Q-Q plots were graphed for each regression, establishing a normal distribution of residuals for univariate linear models. The definition of polypharmacy was set at > 5 prescription medicines.
The C30 subscales differences were tested using logistic regression. Q-Q plots were graphed for each regression, establishing a normal distribution of residuals. Results from multivariate analysis were adjusted for age, gender, BMI and active treatment and cohabitation status. The calculated Variance Inflation Factor (VIF) of involved variables showed minimal collinearity.
Cohen’s d was calculated for unadjusted and adjusted mean differences defining effect sizes as negligible (< 0.25), small (> 0.25), medium (> 0.50) and large (> 0.75).
All statistical operations were done in R-Studio (ver.1.3.1093)
The study was done in compliance with the tenets of the Declaration of Helsinki. The local regional Ethical Committee reviewed the study (record no. 18-000080). The invitation letter explained the purpose of the survey, the thematic nature of the questions, and emphasized the anonymity of the respondent. The project leads, sponsoring organization and sources of funding were named.