We invited 100 children managed by the department of PPC of a tertiary care hospital between June 2021 and August 2023 participate in the study, and 58 of them (58%) responded. The flowchart of the assessed paediatric patients (Appendix A of the supplemental material) presents the reasons for not participating. The inclusion criteria were: (1) age >8 years; (2) fluency in Spanish;(3) having undergone an initial multidisciplinary assessment; (4) approval of participation by referring psychologist, and (5) informed consent/assent signed by legal guardians and patient. We excluded patients with moderate or severe neurological impairment or at risk of imminent death.

• a)

  Sociodemographic variables: we collected data on age, sex and nationality from the health records.

• b)

  Disease-related variables: including the diagnosis, treatment withdrawal or withholding, clinical exacerbations, congenital disease and time elapsed from diagnosis. We collected this information from the health records and by asking specific questions from health professionals, who answered within a maximum of 3 days from the receipt of the child's response, thereby avoiding potential sources of bias.

• c)

  Psychological variables through the following instruments:

  • none-

    Evaluation of Suffering in Children and Adolescents with Palliative Needs scale (qESNA): it comprises 16 items, all in visual analogue scale format, which the children completed on an electronic device through a digital platform (Aimentia Health), accompanied by the psychologist of the research team during the interview. The total score is obtained by adding the individual item scores for a maximum possible score of 160 points (range, 0–160). Five of the items are reverse-scored to control for response bias.

  • none-

    Emotion regulation: we used the emotion regulation (mood repair) subscale of the Spanish version of the Trait Meta-Mood Scale-24 (TMMS-24).13,14 It consists of 8 self-report items and it has been adapted and applied in children and adolescents,13 with a Cronbach α of 0.82. Items are rated on a 5-point Likert scale. In our study, we found an omega coefficient of 0.85.

  • none-

    Cognitive strategies for emotion regulation: we used a short version in Spanish of the Cognitive Emotion Regulation Questionnaire-Kids (CERQ-k).15,16 It comprises 18 items in 9 dimensions (self-blame, acceptance, rumination, positive refocusing, refocus on planning, positive reappraisal, catastrophizing, putting into perspective, and other-blame), each with 2 items rated on a 5-point Likert scale. Higher scores indicate greater use of regulation strategies. This self-report scale has been adapted for use in children and adolescents,15,17 with a Cronbach α of 0.80. In our study, we found an omega coefficient of 0.66.

  • none-

    Psychosocial functioning: the Strengths and Difficulties Questionnaire (SDQ)18 comprises a total of 25 items in 5 scales: emotional symptoms, conduct problems, hyperactivity problems, peer problems and prosocial behaviour. In every scale except for prosocial behaviour, higher scores indicate more severe problems, and the questionnaire achieved a Cronbach α of 0.73.19 In our study, we found an omega coefficient of 0.62.

  • none-

    Depression and anxiety symptoms: we used two questionnaires, each with 2 items, to screen for depression (Patient Health Questionnaire-2 [PHQ-2]) and for anxiety (Generalized Anxiety Disorder 2-item [GAD-2]),20 with answers given in a 4-point Likert scale (0–3) and a cut-off point of 3 (positive ≥3). The PHQ-2 (Cronbach α = 0.81) and GAD-2 (Cronbach α = 0.77) have been used previously in the paediatric population.21 In our study, we found an omega coefficient of 0.82.

Once the legal guardians and the patients had given informed consent, the psychologists of the department of PPC collected the data by interviewing the children, reviewing the health records and asking questions from the referring providers. We carried out a pilot study in 8 children (aged 8–18 years), which confirmed the feasibility of the use of these scales.

We used the software R, version 4.3.1, going through the following steps: (a) basic description of the data; (b) construct validity: exploratory factor analysis (EFA) with the Pearson correlation matrix, Bartlett test of sphericity and Kaiser-Meyer-Olkin (KMO) index, and confirmatory factor analysis (CFA). The goodness of fit was assessed by means of the χ2 test, comparative fit index (CFI), Tucker-Lewis index (TLI), standardised root mean square residual (SRMR) and root mean square error of approximation (RMSEA); (c) Reliability: Cronbach α, omega correlation coefficient, average variance extracted (AVE), composite reliability (CR) and test-retest correlation coefficient; (d) convergent validity with the Pearson correlation coefficient; and (e) analysis of specificity and sensitivity to determine the optimal cut-off point. We defined statistical significance as P < .05.

The qESNA scale was completed by 58 children, 19 (32.8%) female and 39 (67.2%) male, with a mean age of 15.6 years (standard deviation [SD], 4.50; range, 8.57–23.96), and 40 (69.0%) had oncological/haematological disease. Table 2 presents additional information on the characteristics of the sample.

Item 4 did not meet the criterion of a KMO greater than 0.5 (KMO = 0.47), and was consequently eliminated from the scale. After analysing the factor loading matrices and the corresponding graph, we found that item 11, with a factor loading of less than 0.3, was not associated with any of the common factors, so it was also eliminated. With the 14 remaining items, we carried out the Bartlett test again (Bartlett = 305.266; P < .001) and the KMO index, and found that all values were greater than 0.5.

We performed the EFA and, using the parallel analysis scree plots (Appendix A, Supplemental Material, Parallel Analysis Scree Plots Figure), we found 3 eigenvalues above the thresholds proposed by Horn,22 so we propose a structure consisting of one eigenvector and 3 factors.

Fig. 1 presents the linear correlation matrix for the 11 items remaining in the new version of the qESNA scale.

Figure 1.

Linear correlation matrix for the qESNA scale.

(0.57MB).

One scale with 14 items and 3 factors (qESNA) exhibited a good fit for the data (CFI = 0.93; TLI = 0.91 and RMSEA = 0.07). Fig. 2 presents the standardised loadings for the scale.

Figure 2.

Standardised loadings for the qESNA scale.

(0.12MB).

Together, these 3 factors explained 45% of the variance: the first factor explained 10% of the variance, encompassed items 9 and 13 and was labelled “Frustration and wish for change”; the second factor, labelled “Overall wellbeing”, explained 23% of the variance and encompassed items 1–4, 6, 8, 11 and 12; and the third one, labelled “Worry”, explained 12% of the variance and encompassed items 5, 7 and 10. The SS loadings for each factor were greater than 1 (factor 1: 1.49; factor 2: 1.86; factor 3: 3.38). Table 3 presents the factor loadings for the estimated model.

We assessed internal consistency with the Cronbach α (0.85; 95% CI, 0.80−0.91), Guttman’s lambda-6 (0.90) and the omega coefficient (0.89), which were indicative of a high internal consistency. The Cronbach α values for individual items ranged from 0.83 to 0.86. The AVE values were less than 0.5 (0.126; 0.269; 0.479). The CR values (0.223; 0.741; 0.726) were greater than 0.7 for subscales 2 and 3. Following the criterion proposed by Fornell and Larcker,23 the latter result indicates that the validity of the scale is still acceptable, even if the explained variance is small.

We assessed temporal stability using data for 7 (12.1%) of the 58 children, with an average of 39.6 days elapsed between the two tests. The intraclass correlation coefficient was 0.66, indicative of an adequate stability.

The mean suffering score in the sample was 95.53 (SD, 19.96; range, 50.00–134.00), with a median of 97.00. Very few patients expressed discomfort (5.17%); 77.59% stated they did not find the interview long, and 79.31% stated that they believed it could help other children like them. Table 4 presents the results of the descriptive analysis of the psychological variables under study.

Table 5 shows the significant Pearson correlation coefficients for the association between the total score and the different factors of the qESNA with the other variables under study. We found strong correlations (>0.60) between the total score and the risk of anxiety (−0.60) and of depression (–0,69) and between factor 2—overall wellbeing—and the risk of depression (−0.67).

Based on the existing literature,12 we created a criterion variable for the assessment of suffering that was a composite of 3 criteria1: presence of emotional or peer problems based on the SDQ scale2; risk of anxiety or depression based on the PHQ-43; scoring above the mean in the use of the rumination and catastrophizing cognitive regulation strategies (>12). Fig. 3 shows the ROC curve obtained in the analysis, with an area under the curve of 0.914 (P < 0.001) and a 95% CI of 0.810−0.971. This proved the predictive/diagnostic capacity of the qESNA, and we established a cut-off point of 81, which was the value that offered the best possible combination of sensitivity (83.33%) and specificity (93,48%) values. A score of less than 81 would be indicative of suffering in children with a significant impact on the emotional, interpersonal and psychological domains. The total score can range from 0 to 140. Based on the establish cutoff, 13 children (22.41%) had suffering.

Figure 3.

ROC curve for the qESNA scale.

(0.14MB).

There are some limitations in this study in relation to the interpretation and generalisability of the results. One of the main limitations of the study is that the sample size was smaller than recommended for an adequate psychometric analysis. The hypothesised factor pattern was confirmed by the analysis of the errors of the model; however, these errors encompassed a substantial portion of the variance that was not explained by the factors, and therefore the reliability analysis of the subscales did not fulfil all the criteria to corroborate the reliability of the 3 factors. Another limitation is the wide age range of the sample under study, a factor that may have affected participants’ ability to understand and respond based on their stage of development. On the other hand, the internal consistency of the used scales was acceptable, with the exception of the scale assessing strengths and skills. (SDQ). In any case, we consider the study has also strengths, including achieving a large sample from the population of interest and the fact that the scale has good psychometric properties, is easy to administer and has a multifactorial structure fitting the multidimensional perspective of suffering. The research team believes it is an advance in the field and a starting point for further research. New studies in larger samples should be conducted to corroborate these findings and making a more detailed analysis.