The present study was performed as part of a multicenter self-report questionnaire survey to develop measurements for evaluating eating-related distress experienced by patients with advanced cancer and their family members. The survey consisted of a development phase and validation phase. The development phase was conducted at 5 hospitals between July and September in 2020, and the validation phase at 11 hospitals between January and July in 2021. All participating hospitals were required to obtain data up to the designated number of subjects according to the size of the institution. The same dataset was used in this analysis.
Consecutive eligible patients were enrolled during the study period. Inclusion criteria were (1) patients newly referred to palliative care, (2) adult patients (≥20 years old), (3) patients with locally advanced or metastatic cancer (hematological neoplasms were included), (4) patients with awareness of the diagnosis of malignancy, and (5) patients with the ability to reply to a self-reported questionnaire. Exclusion criteria were (1) patients forbidden to eat by the physician for medical reasons, e.g., dysphagia or malignant bowel obstruction, and (2) psychological issues recognized in an interview with the primary or palliative care physician. If subjects did not want to participate, we requested them to return the questionnaire with ‘no participation’ indicated. The completion and return of the questionnaire were regarded as consent to participate in this study. Ethical approval for the present study was granted by the Institutional Review Board of each hospital.
The questionnaire for the present study was developed by the authors based on a previous survey of bereaved family members  and discussions among the authors. The face validity of the questionnaire was confirmed by a pilot test with five medical personnel, five palliative care physicians, and three palliative care nurses.
We initially asked about patient characteristics, such as sex, age, the primary cancer site, Eastern Cooperative Oncology Group Performance Status (ECOG PS), setting of care, and treatment status. We also asked patients to report on dietary intakes with the ingesta-Verbal/Visual Analogue Scale (ingesta-VVAS), using 10-point analogue scales to estimate dietary intake in patients with cancer (high scores indicate better dietary intakes) .
We subsequently requested patients to report anthropometric measurements, i.e., height, current body weight, and previous body weight, to calculate body mass index (BMI) and % weight loss (WL) in 6 months.
We finally asked patients to answer 15 items regarding their beliefs and perceptions about PNH using the following seven-point Likert scale: (1) absolutely agree, (2) agree, (3) somewhat agree, (4) not either, (5) somewhat disagree, (6) disagree, and (7) absolutely disagree. In the questionnaire, we explained PNH as ‘supplying nutrition and hydration through an intravenous drip’ in easy Japanese.
Patient characteristics were presented as n (%) or medians (interquartile ranges) where appropriate.
BMI was calculated by dividing current body weight (kg) by height (m)2. %WL was calculated as follows: (current body weight [kg] - previous body weight [kg]) / previous body weight (kg) × 100. Cachexia was %WL in 6 months ≥5% or BMI <20 kg/m2 + %WL in 6 months ≥2%. Patients above or below these cut-off values were grouped as follows: the non-cachexia group and cachexia group .
The proportions of patients with ‘absolutely agree’, ‘agree’, or ‘somewhat agree’ were calculated with a 95% confidence interval (CI) regarding the 15 items about their beliefs and perceptions of PNH. Comparisons of the scores for the 15 items between the non-cachexia and cachexia groups were also performed using the Mann-Whitney U test.
We then conducted an explanatory factor analysis using the principle method with a promax rotation. Based on the results of the factor analysis, attributes with factor loadings less than 0.4 (standardized regression coefficient) were not excluded. We also calculated Cronbach’s alpha coefficients to assess the internal consistency of a set of items in each of the conceptual groups extracted.
A multiple logistic regression analysis was performed to identify the independent factors affecting cancer cachexia stages using patient characteristics and the mean scores for items in each concept of patients’ beliefs and perceptions, which were dichotomized with <4 (absolutely agree, agree, and somewhat agree) or ≥4 (not either, somewhat disagree, disagree, and absolutely disagree). A multivariate model was adjusted for sex, age, the primary cancer site, and ECOG performance status.
All results were considered to be significant when the p-value was less than 0.05. All analyses were performed using SPSS software version 27.0.