Background
According to the World Health Organization, the global incidence rate of childhood cancer is approximately 100–150 cases per million children and adolescents under 15 years of age [
1]. In Europe, namely in Portugal, as well as in North America, leukemias are the most common childhood cancers, followed by central nervous system tumors and lymphomas [
2].
Clinical advancements in the treatment of this condition have led to a relevant reduction in mortality rates in the last 50 years. Currently, 70 to 80 % of pediatric cancer patients can be cured in developed countries [
3] if diagnosed and treated early [
4]. This progress, however, has been achieved with aggressive medical protocols that have a significant impact on children and adolescents’ daily lives, which affect, among other aspects, their health-related quality of life (HRQoL) [
5]. As survivorship rates have increased concerns regarding patients’ HRQoL have become key aspects in cancer care.
HRQoL is a component of the more global construct of quality of life [
1] and has been defined as the patient’s perception of the impact of the disease and treatment in several domains, such as physical, mental and social [
6]. Although HRQoL is, by definition, self-referential, the assessment of HRQoL in children and adolescents has, until recently, primarily relied on parents’ reports [e.g.,
7]. However, recent studies highlight that child reports cannot be substituted with the parent reports, with the exception of patients too young, too cognitively impaired, or too ill or fatigued to complete a HRQoL assessment instrument [
8]. Hence, similar to North America, it is important to make available a reliable and valid assessment instrument in Portugal to measure children and adolescents’ self-reported HRQoL in pediatric cancer patients across different statuses of treatment [
9].
A recent systematic review [
10] identified 13 instruments for assessing quality of life, which were developed for use with children with cancer and childhood cancer survivors. Nonetheless, only two, the Pediatric Quality of Life Inventory™ Cancer Module (PedsQL™ 3.0 Cancer Module) and the Quality of Life for Children with Cancer Scale, were appropriate taking into account the following requirements: (1) self-reported measure; (2) appropriate for any type of cancer; (3) appropriate for both on- and off-treatment status; and (4) comprising versions for both children and adolescents. Nevertheless, none of them had yet been translated and validated for Portuguese. The option for the PedsQL™ 3.0 Cancer Module was based on the fact that it was currently the most widely used measure of child health [
11] and had more psychometric data published compared with other pediatric quality of life instruments [
10].
However, to our knowledge, no published studies have examined the factorial structure of the PedsQL™ 3.0 Cancer Module using confirmatory factor analysis (CFA). While exploratory factor analysis (EFA) is designed for situations where links between the observed and latent variables are unknown or uncertain, CFA enables a specific hypothesized structure to be tested [
12]. Because the PedsQL™ 3.0 Cancer Module was developed based on a theoretically driven model [
13], the use of CFA is appropriate.
Regarding the convergent validity of the self-report version, with minor exceptions, previous studies [
14‐
17] have identified significant positive associations between the scores of the PedsQL™ 3.0 Cancer Module and the PedsQL™ 4.0 Generic Core Scales [
13].
In addition, previous studies examining the reliability of the PedsQL™ 3.0 Cancer Module self-report version have demonstrated adequate values of internal consistency [
16,
17] and temporal stability [
14,
15] for the total score. Regarding the subscales, some studies found values below the established cut points for internal consistency [
18] and temporal stability [
14,
15].
Despite the growing support for the validity and reliability of the PedsQL™ 3.0 Cancer Module, to date, there is no evidence for its factorial validity, via CFA. In addition, the convergent validity and the test-retest reliability across age groups [for two exceptions: 14, 15] remain understudied.
The present research aimed at examining the psychometric properties of the self-reported European Portuguese version of the PedsQL™ 3.0 Cancer Module in children/adolescents. The specific objectives were to: (1) conduct descriptive and item analyses for the whole sample; (2a) test the factorial validity via CFA for the whole sample; (2b) evaluate convergent validity in two age groups (children 8–12 vs. adolescents 13–18); (3) assess the internal consistency, and the test-retest reliability in two age groups (children 8–12 vs. adolescents 13–18).
Discussion
The aim of this study was to assess the psychometric properties of the European Portuguese version of the PedsQL™ 3.0 Cancer Module for children and adolescents. The findings supported the factorial structure and the convergent validity. Overall, results also suggested good internal consistency and temporal stability.
The absence of missing data suggested that children and adolescents were willing and able to provide data. Similar to other versions [e.g.
14‐
16,
20] there was a tendency for ceiling effect for the subscales and items. This tendency has been reported in other studies concerning the measurement of HRQoL for children/adolescents with cancer [
40]. The inclusion of children/adolescents that were off-treatment may have potentially accentuated the ceiling effects [
20]. This study increases current knowledge by showing that total scores and all subscales of the PedsQL 3.0 Cancer Module were normally distributed, a trend that has been reported in other studies [
14,
15]. In addition, similar to the Brazilian version [
17], item-total correlations demonstrated good item discrimination.
Although the PedsQL™ 3.0 Cancer Module has been translated into 25 languages and some psychometric properties of these versions have been reported [e.g., 14, 15, 16], to our knowledge, this is the first study to analyze the factorial structure of this instrument via CFA. In accordance with the factorial structure proposed by the original authors [
13], the CFA indicated a second-order factorial model and determined that the eight first-order factors are structurally related and consistent indicators of a higher level construct, that is, the HRQoL.
The correlations between the PedsQL™ 3.0 Cancer Module and the DCGM-12, and the RCMAS-2 scores attested the convergent validity. The medium to large correlations between the two instruments, that measure the HRQoL, were similar compared to the original version [
13], the Japanese version [
15] and the Chinese version [
14] that compared the PedsQL™ 3.0 Cancer Module and the PedsQL™ 4.0 Generic Core Scales. The unique contributions of our study, compared with previous research with the PedsQL™ 3.0 Cancer Module [
13‐
17], are that the data are presented separately for each age group and that the other scale that assessed the HRQoL (DCGM-12) was not derived from the same theoretical framework. In addition, the majority of the correlations between the PedsQL™ 3.0 Cancer Module and the RCMAS-2 scores were also medium to large, with the exception of the children’s reports of Nausea and Procedural Anxiety. However, the items of the RCMAS-2 are more oriented for physiological anxiety, worry, and social anxiety [
23] and not for the physical side effects of cancer treatment (e.g., nausea, dysgeusia) or for the anxiety related to procedures (e.g., injections, blood tests, intravenous therapy).
With regard to reliability, PedsQL™ 3.0 Cancer Module exceeded the minimally acceptable internal consistency cut-point, with the exception of the Pain and Hurt, Cognitive Problems, and Perceived Physical Appearance subscales in the self-reports of children. Similar to these findings, others studies have also demonstrated poor internal consistency for one or all aforementioned subscales [
13,
17]. The low reliability of these subscales identified across countries may be related to their small number of items [
41] and also to children’s learning difficulties, which may mirror the cognitive and neuropsychological effects of the treatment and the high rate of absenteeism [
42]. Although, in Portugal, the education of youth with cancer is protected by Law (No. 71/2009 of August 6
th) [
43] with school support being available upon request at home and at the hospital, the treatment for pediatric cancer can often interrupt regular school attendance, that might be more critical for the children's group.
Finally, total score and all subscales demonstrated good to excellent test-retest reliability, with the exception of two subscales for children. Similar to Lau et al. [
14], one of the only two studies presenting the test-retest reliability for this age group, the subscales Pain and Hurt and Cognitive Problems (for children) presented moderate test-retest reliability values. However, the children/adolescents who participated in the test-retest were undergoing treatment at the time of the study. Thus, the low values in the aforementioned subscales may reflect changes over the previous week (e.g., treatment side effects, outpatient or inpatient treatment, interrupt homebound services) or impact due to a specific situation (e.g., blood test, lumbar punctures) [
44]. With few exceptions, however, the overall results suggest that the Portuguese version of the PedsQL™ 3.0 Cancer Module is stable over time.
The strengths of the PedsQL™ 3.0 Cancer Module include the sound theoretical background, the inclusion of the child-centered features, that is, important domains of a child’s life, which enables the child to assess his/her own HRQoL taking into account his/her developmental stage [
45] according to the guidelines in this field [
46, for a review]. A strength of this specific study is its large sample size, which allowed for analyzing the convergent validity and reliability of this measure separately for children and adolescents. Additionally, this was the first study to conduct the factorial validity of the PedsQL™ 3.0 Cancer Module via CFA, providing empirical support for the original model proposed by the original authors of this scale.
The current study also presents potential limitations. First, the sample collection was conducted in three of the four-oncology centers in Portugal. Although it is unlikely that regional differences may influence how children/adolescents perceive the impact of cancer on their own HRQoL, generalizations should be made with caution. Second, we did not control for the treatments and their side effects, and school support over the two assessment moments. Third, an analysis of the parents’ proxy-reports was not conducted. Future research should inspect the psychometric properties of parents’ proxy-reports and the agreement between the parents’ proxy-reports and the children/adolescents’ self-reports. Moreover, future studies with a larger sample should analyze whether the factorial structure remains valid across groups of children and adolescents and across on- and off-treatment status.
Acknowledgement
The authors would like to thank the Pediatrics Services of the Portuguese Institute of Oncology (IPO-Porto) and São João Hospital, both located in Porto, and the Pediatric Department, Centro Hospitalar e Universitário de Coimbra in Coimbra healthcare professionals, as well as all of the children/adolescents and their family caregivers who agreed to participate in this research project.
This research was supported by a PhD scholarship (SFRH/BD/80777/2011) from the Portuguese Foundation for Science and Technology to the first author. Correspondence concerning this article should be addressed to Susana Santos, Faculty of Psychology and Educational Sciences, University of Coimbra, Rua do Colégio Novo, Apartado 6153, 3001–802 Coimbra, Portugal. E-mail: susanafernandessantos@gmail.com.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SS carried out the collected sample, participated in the design of the study, participated in literature search, performed the statistical analysis and wrote all of the content. CC participated in the design of the study and coordination, performed the statistical analysis and wrote part of the content. MCC participated in its design and coordination, helped to draft the manuscript, and did critical revision of manuscript. AF, LB, DdC, and AP participated in its design and did critical revision of manuscript. All authors read and approved the final manuscript.