Disagreement between mothers' and fathers' rating of health-related quality of life in children with cancer

Overall survival for pediatric cancer has increased dramatically, resulting in more than 94% of patients surviving acute lymphoblastic leukemia [1, 2]. As a consequence, improvements in childhood cancer care have directed attention to the entire and complex situation in which the patients and their families find themselves. Acute and chronic health conditions such as pain, fatigue, nausea, anxiety, and depression commonly develop and ultimately pose a risk for social and economic challenges that reduce quality of life during and after completion of therapy [3, 4]. Thus, a central task in clinical care is to take these aspects seriously and to shift the focus also to the emotional and psychological stress.

The gold standard for assessing patients’ experience and health-related quality of life (HRQOL) is patient-reported outcome measurements (PROMs) [5‐10]. While they are widely used in adult oncology, PROMs are still rarely present in pediatric cancer research and therapy [11]. The use of PROMs in this vulnerable group is compromised by age, developmental stage, family relationships and psychosocial challenges [12‐14], but children can reliably self-report on their health if adequate questionnaires are used [15‐17].

The health-care relationship in pediatrics is a triad, where the caregivers are involved in the management of the patient’s health conditions [18, 19]. Thus, caregiver proxy reports are often used as an alternative to the child’s self-report, particularly but not only when the child is unable to provide a self-report. The congruence between caregiver reports and child self-reports is influenced by several factors such as diagnosis, age, gender, socioeconomic status, or parent’s own HRQOL [20‐23] with the consequence that disagreement on health and well-being might result in medical mismanagement [24]. In addition, parents consistently tend to overestimate symptom burden and functional limitations compared to children’s self-report [20, 22, 23, 25, 26]. Since an overall lack of fathers’ reports is noted [12], most of the studies compared caregiver–child dyads and there are little data regarding the difference between mothers’ and fathers’ perspective on child’s HRQOL.

One recently published study obtained proxy ratings from both parents separately and suggested that paternal and maternal reports are interchangeable. However, the majority of questionnaires reported results at 3.3 years after end of treatment and, thus, were not representing the impact of acute cancer therapy [27]. In general, treatment for childhood cancer can extent over several months, but the first 2–3 months of treatment represent the most intense phase, as they are associated with urgent surgery (e.g., tumor removal, insertion of catheters), highest toxicity by chemotherapy or irradiation, and residual cancer impairment (e.g., fatigue, pain). Together with the isolation from the usual social setting, this puts the patients in the focus of major interest to compare mothers’ and fathers’ perspective on children’s HRQOL.

We recently developed a unique, web-based approach for daily child self- and parent-based proxy reporting (ePROtect) [28‐30]. Within this study, the child as well as both parents were asked to complete additionally the general and the cancer-specific version of the Pediatric Quality-of-Life Inventory (PedsQL) on a monthly basis. We aimed to compare cross-sectional as well as longitudinal agreement on HRQOL between children and adolescents newly diagnosed with cancer and their corresponding mothers and fathers in the first months of therapy.

Caregiver proxy reports are often required as an alternative to child self-reports as the integration of PROM is more challenging in pediatric oncology. This stands in strong contrast to the evidence that patients themselves are the best reporters and that caregiver proxy reports are affected by several factors such as child’s age, sex, parental educational level, social demographics, cultural background, parent’s own HRQOL, and distress [12‐14, 34‐36]. Thus, there is a strong recommendation that children with cancer should be the primary reporters of their symptoms and if they are unable to provide self-reports caregiver proxy reports should be used [12]. In addition, an overall lack of fathers as caregiver proxy reporters is noted, and only a limited number of research studies have included both parents but did not separately investigate their perspectives on child’s HRQOL during active cancer therapy [12, 20, 22, 23]. One recently published study compared 120 paternal and maternal proxy reports concerning agreement on child HRQOL, but the vast majority of children with cancer were post-treatment (mean time since diagnosis 3.3 (± 1.4) years and 87% of the patients had completed therapy). The reported study design might lead to the assumption that paternal and maternal reports are interchangeable [27]. Interestingly, agreement between mother–child and father–child dyads were more likely to differ if their child was still in active treatment. However, the authors noted that this finding should be interpreted with caution as only few children were in active treatment [27]. Our study instead shows that during active cancer therapy, mother and father proxy reports differ in the level to which they agree with children’s self-reports. Namely, mothers' proxy reports, as compared to fathers' reports, showed better agreement with the children’s reports.

Children consistently reported fewer impairments than their parents, particularly one month following cancer therapy. This seems to be the main reason for disagreement between patients and parents and is in accordance with other studies showing that caregivers have the tendency to overestimate their children’s HRQOL impairments [20, 23, 37‐39]. The PedsQL Generic for general HRQOL showed mainly moderate and good agreement shortly after diagnosis and agreement remained only on the physical domain with ongoing therapy. It could be assumed that this domain is the easiest one for evaluation, since inpatient stay and medical treatment are connected with observable physical restriction and impairment [24, 40]. The cancer-specific PedsQL is more informative and provides a comprehensive overview. These questions consider the fact that the daily routine care and, thus, closer contact between primary caregiver (mostly mothers) and child revealed many domains with moderate and good agreement for mother–child dyads throughout all assessment points. Interestingly, fathers reported more impairment of child’s HRQOL than mothers on nearly all domains and time points, thus, leading to a low number of domains with agreement. This might be influenced by their role as second caregiver who did not spend as much time with the child as the mothers did.

The strong disagreement between both parents and children in the domain communication of the PedsQL Cancer was highly surprising. The reason for the disagreement was that parents reported their children did not provide information about health status and did not ask questions of the health-care team. This stands in strong contrast to the child self-reports and the impressions of the health-care team. Ad hoc follow-up interviews with parents revealed that they misunderstood the question, i.e., by "no problems" they meant "always/often," which is the inverse sense of meaning. However, other studies using PedsQL did not notice such disagreement between children and proxies [22, 41, 42].

The main strength of this study is the approach to include the perspectives of both parents on the child’s HRQOL in a cohort of pediatric patients with diverse cancers. Moreover, the longitudinal monthly assessment during the intensive first therapy months provides an overview of the development of the parent’s perspectives, particularly the disagreement for father–child dyads assessed by the PedsQL Cancer Module. Notably, this study was performed during repetitive coronavirus disease outbreaks, which might have had negative effects on the cancer care management such as restricted visits (e.g., one single caregiver was allowed), increased fear of infection and reduced psychosocial support of patients (e.g., no service of clown doctors during lockdown). Further limitations are the monocentric study design, which includes a small sample size and restricts the description of the study group and comparison of the characteristics between patients and caregivers (e.g., cancer type, age). Since we routinely collect PRO data, we will be able to conduct more subgroup and in-depth analyses of our present results in the future. Finally, we have not collected data on the educational age of the children and, thus, were not able to analyze differences in patient–observer differences based on the children’s chronological and educational age. Despite these limitations, our findings take the first steps to characterize the perspectives of both parents on child’s HRQOL.

So far, almost all studies came to the conclusion that both patient and observer assessment of HRQOL are of key value in pediatric oncology and their use is recommended by regulatory agencies and experts in the field [20, 43‐45]. Nevertheless, no comprehensive guidance on how to deal with the well-known discrepancies between the child’s self-assessment and parent observer ratings in clinical studies is yet available. A clear guidance should offer recommendations in which cases to use patient and/or proxy ratings, present data on potential confounders for parental proxy ratings and differences in patient proxy accordance across the different age-groups, and offer recommendations on statistical approaches to overcome the bias. This would not only facilitate the use of PROs as endpoints in pediatric oncology, but also help to evolve the quality of assessed data in the field. The aim of this study was to add to the growing body of literature and, thus, to facilitate overarching recommendations on the patient–observer dilemma.