Sample
N = 230 cancer patients consented to participate in this study. N = 17 patients had to be excluded due to not filling out any of the questionnaires after registration (n = 12) or not filling out most of the questionnaires (n= 1), not fulfilling the inclusion criteria of a cancer diagnosis (n = 2), due to more than one answer in the PEACE-G items (n =1), and one participant stated after the survey that he/she could not answer the questions honestly.
Final sample size comprised N = 213 participants, which were recruited in Heidelberg (97.4%) and Leipzig (2.6%)1. 86% of the participants completed the survey online via tablet or smartphone, 14% chose to fill out the paper-pencil version. 64% of the sample were female, and the mean age was 54.24 years (SD = 11.50, range = 19-80). The mean time since diagnosis was 3.09 years (SD = 4.10 years).
Further sociodemographic characteristics of the sample are presented in Table 1, medical characteristics are presented in Table 2.
Table 1: Sociodemographic sample characteristics
Patient characteristics
|
n (%)
|
German nationalitya
|
202 (94.8)
|
Population size of residential area
100.000 20.000 – 100.000 5.000 – 20.000 < 5.000
|
38 (17.8) 36 (16.9) 69 (32.4) 70 (32.9)
|
Marital statusa
Single Married, living together Married, not living together Divorced Widowed
Participants living in a partnership
|
40 (18.9) 139 (65.6) 8 (3.8) 21 (9.9) 4 (1.9)
178 (83.6)
|
Highest school grade
Main school certificate Secondary school certificate Abiturc Other / None
|
45 (21.1) 77 (36.2) 83 (39.0) 8 (3.8)
|
Employment
Employee
Self-employed
Civil servant
Unemployed In training Retired
Other
|
113 (53.1)
20 (9.4)
13 (6.1)
5 (2.3) 1 (0.5) 46 (21.6)
15 (7.0)
|
Sick leaveb
|
92 (43.2)
|
Income (n = 207; € monthly) < 250 250 – 500 500 – 1000 1000 – 1500 1500 – 2000 2000 – 3000 3000 – 4000 4000 – 5000 > 5000 no comment
|
1 (0.5) 5 (2.3) 21 (9.9) 36 (16.9) 38 (17.8) 46 (21.6) 36 (16.9) 9 (4.2) 11 (5.2) 4 (1.9)
|
Note. N = 213.an = 212 bn = 211. cAbitur Is the highest school graduation in Germany.
Table 2: Participants’ medical characteristics
Characteristic
|
n (%)
|
Type of cancer
Breast Malignant melanoma Colorectal Pancreatic Stomach Kidney Oesophageal Head and neck
Liver Cervical Ovarian
Blood and lymph gland Neuroendocrine Penis or testicular Cholangiocarcinoma Prostate
Lung Bladder Other
|
85 (39.9) 32 (15.0) 17 (8.0) 14 (6.6) 8 (3.8) 8 (3.8) 7 (3.3) 5 (2.3) 5 (2.3) 4 (1.9) 4 (1.9) 3 (1.4) 3 (1.4) 3 (1.4) 2 (0.9) 2 (0.9) 2 (0.9) 1 (0.5) 8 (3.8)
|
First cancer diagnosis
|
164 (77.0)
|
Current treatment (multiple choice) Chemotherapy Immunotherapy OP
Radiotherapy
Metastatic cancer
|
148 (69.5) 83 (39.0) 9 (4.2) 8 (3.8)
126 (59.2)
|
Note: N=213.
Confirmatory factor analysis
Incorporating a covariance between the residuals of PC 08 and PC 10 resulted in the final model, which showed satisfactory outcomes (χ² (df) = 123.70 (52), p < .001, CFI = 0.918, RMSEA (90% CI) = 0.081 (0.062 – 0.099)), with moderate RMSEA and acceptable CFI values. An illustration of the final model is shown in Figure 1.
Please insert Figure 1
Note: Final SEM of the PEACE-G
Results of the single-sample cross-validation indicated that the integration of the covariance between the residuals of PC 08 and PC 10 led to a substantial model fit improvement in each subsample constellation (see Supplement 4).
Convergent validity
The correlation between the PA and SwI subscale was r = -.64, p < .01. Correlation coefficients of PA and SwI with symptoms of psychological distress and related constructs are depicted in Table 3. All correlation coefficients reached significance. PA showed negative associations with depression, anxiety, distress, psychological inflexibility, and positive associations with HRQoL, acceptance, resilience, and mindfulness, with small to large effect sizes. SwI showed inverse significant correlations with the constructs examined.
Table 3: Correlation between the PEACE subscales and psychological burden, HRQoL and related constructs
|
FFMQ - AOBa
|
FFMQ - MAHa
|
RS-11
|
SEK-A
|
SF-12 - physical
|
SF-12 - psych
|
PHQ-8
|
GAD-7
|
Distress
|
AAQ-IIa
|
Peaceful Acceptance
|
.38***
|
.40***
|
.37***
|
.40***
|
.14*
|
.56***
|
-.53***
|
-.56***
|
-.44***
|
-.57***
|
Struggle with Illness
|
-.33**
|
-.40**
|
-.33***
|
-.20**
|
-.30***
|
-.50***
|
.56***
|
.53***
|
.45***
|
.54***
|
Note. ***p < .001. **p < .01. *p < .05. aSpearman correlation coefficient.
Note: FFMQ-AOB = Five Facet Mindfulness Questionnaire Accept without Judgement scale. FFMQ - MAH = Five Facet Mindfulness Questionnaire Act with Awareness scale. RS-11 = German Resilience Scale; SEK-A = Acceptance scale of the Fragebogen zur Standardisierten Selbsteinschätzung Emotionaler Kompetenzen. SF-12 = Short-Form-Health-Survey-12, physical and psychological scale score. PHQ-8 = Patient-Health-Questionnaire-8; GAD-7 = Generalized Anxiety Disorder Questionnaire. Distress = assessed with NCCN distress thermometer; AAQ-II = Acceptance and Action Questionnaire-II.
Reliability
Cronbach’s α of the total scale was α = .86, the subscales showed a Cronbach’s α of .79, each, indicating good reliability.
As an additional analysis of item parameters, the item-scale correlations were calculated. Except for item 4 (‘Do you feel loved now?’, r = .26), the correlations were all between r = .43-.70. The correlations and the level of Cronbach's α when an item is excluded from the scale are depicted in Supplement 5.
Sensitivity Analysis
The PA and SwI scores had no relation with sex, nationality, marital status, children, children living at home, number of inhabitants of hometown, level of school and professional education, employment status, income level, and all medical information. Patients on sick leave reported significantly lower PA (M = 15.22, SD = 2.90) than patients not on sick leave (M = 16.53, SD = 2.82), t(207) = -3.30, p < .01, d = -0.46), and higher SwI (M = 17.57, SD = 4.39) than patients not on sick leave (M = 15.42, SD = 4.65), t(209) = 3.42, p < .01, d = 0.48). PA correlated significantly with age; r = .18, p < .05.
Results of EFA indicated three factors (see Supplement 2). In this model, the items of the SwI scale are distributed to two factors, with PC 08 and PC 10 loading remarkably high on factor 2; PC 07, PC 09, PC 11, and PC 12 loading on factor 3; and PC 06 loading almost comparably on factor 2 and 3 (see Table S2-1, Supplement 2 for a detailed description). Based on these results we conducted a CFA with three factors. The model comprised a PA factor (PC 01-05), a second factor (PC 08, PC 10), and a third factor (PC 06-09, PC 11-12). The three-factor-model yielded better fit indices than the two-factor solution (see Figure S3-1, Supplement 3). However, the slight improvement in indices comes at the cost of reduced parsimony and simplicity. Further, at least three items should load on a factor [33]. Thus, we decided to remain at the described two-factor solution.
Further sensitivity analyses indicated a substantial better model fit in the subsample of patients with non-metastatic cancer versus patients with metastatic cancer. SEMs of alternative models did not improve the model fit. Results of these models are depicted in Supplement 4.