Trajectories of illness perceptions in paediatric cancer patients and their parents and associations with health‐related quality of life: Results of a prospective‐longitudinal study

In paediatric oncology, little is known about trajectories of illness perceptions and their longitudinal associations with health‐related quality of life (HRQoL). Therefore, the aim of this study was to investigate changes in illness perceptions in children and parents over a one‐year‐period and to investigate predictive value of child's and parent's illness perceptions during acute treatment for child's HRQoL 1 year later.


| INTRODUCTION
In the last decades, treatment options for childhood cancer have improved significantly, leading to increased survival rates of over 80% in high-income countries. 1However, childhood cancer is still among the leading causes of death and disability in children and adolescents. 2Also, despite improved prognosis, many paediatric patients and their families report poor health-related quality of life (HRQoL), [3][4][5][6][7][8] high fear of progression, 9 and anxiety and depressive symptoms [10][11][12][13] both during and after acute treatment.

The development of effective interventions to improve HRQoL
or other psychological health-related outcomes requires a better understanding of underlying variables and processes.5][16] Also, as those illness-and treatmentrelated variables are mostly not (psycho-)therapeutically modifiable, they are not an appropriate target for psychosocial interventions to improve psychological health-related outcomes such as HRQoL.
5][16][17][18] The Common Sense Model of Illness Representation (CSM) 19 poses that individuals create cognitive representations of a perceived health threat, for example, a cancer diagnosis.These illness perceptions are based on available information (e.g.conversations with others, information by medical staff, own experience) and launch a self-regulation process aimed to overcome the health threat. 19,20lness perceptions include perceptions on physical symptoms associated with the illness/treatment, perceived timeline (i.e.perceived chronicity or cyclicity of the illness), perceived consequences on the individual's life (e.g.restrictions concerning hobbies or school/work), perceived cause of the illness or symptoms, perceived controllability and curability, and illness-related emotions. 20

| Illness perception trajectories
Research concerning the stability of illness perceptions over time remains inconclusive, suggesting that there is no clear pattern across age groups, gender, or diagnoses: While some researchers found illness perceptions to remain relatively stable over a 6-to 12-month period in women with breast cancer, 21 young adult cancer patients, 22 and patients with colorectal cancer, 23 others found that some illness perceptions changed in adult patients with head and neck cancer and their caregivers, 24 as well as adult oesophageal cancer survivors. 25So far, to our knowledge no longitudinal studies have investigated trajectories of illness perceptions in paediatric oncology.It has to be kept in mind that in paediatrics, children's concepts of illness undergo significant changes due to their cognitive development, for example, pre-schoolers show a strong focus on present events, have limited ability to grasp the course of an illness, and have difficulty anticipating future events.These abilities typically develop in children in primary school. 26

| Associations between illness perceptions and health-related outcomes
Research has demonstrated a strong association between illness perceptions and health-related outcomes in cancer patients/survivors.For example, studies with adult cancer survivors have shown that more negative illness perceptions (e.g.perception of more negative consequences, less understanding of the illness) are associated with lower HRQoL 15,27 and higher levels of depression and anxiety. 28Studies with adult cancer patients in acute treatment have revealed similar results, with illness perceptions being associated with several dimensions of HRQoL, such as physical, emotional, and social functioning. 17Research on the association of illness perceptions with HRQoL in paediatric oncology, on the other hand, is sparse: Fonseca et al. found that several illness perceptions are significant predictors of HRQoL in 8-12-year-old cancer patients, with the perception of negative consequences being the strongest predictor. 29Also, our work group has recently contributed research on 4-18-year-old cancer patients in acute treatment and in follow-up care: Similar to results in adult cancer patients and survivors, and similar to Fonseca's results, we found that several illness perceptions contribute to the child's HRQoL, 30 namely the perception of symptoms and cyclicity of their illness.However, the aforementioned studies used cross-sectional data, which limits the causal interpretation of results.Dempster et al. conducted a longitudinal study with a sample of adult oesophageal cancer survivors and found that changes in illness perceptions over the course of 1 year contributed significantly to the variance in change of anxiety and depressive symptoms. 25  course of 6 months.Patient's perceptions of personal control and coherence, as well as caregiver's perceptions of coherence predicted patient's HRQoL. 24Results from studies with other illness populations (e.g.[33][34][35] In paediatrics, the use of patient-reported outcomes is recommended if age-appropriate assessment instruments are available. 368][39][40] As the cancer illness affects the whole family system, the assessment of parent's own illness perceptions is useful as well to investigate (dis-) similarities between child's and parent's illness perceptions 16 and reciprocal influences on each other's healthrelated well-being. 30,41As longitudinal research on illness perceptions in paediatric oncology is still limited, we will endeavour to answer the following research questions: (1) How do child's and parents' illness perceptions, and child's HRQoL change over the course of 1 year?As previous studies have shown that trajectories of illness perceptions may vary by patient characteristics, analyses will be performed in the full sample and in age-, gender-, and diagnosisspecific subsamples.( 2) To what extent do child's and parents' illness perceptions predict child's HRQoL 1 year later?

| Design
The present study is an observational prospective-longitudinal study with a 12-month-follow-up.Ethics approval was obtained from the ethics committees of the Technische Universität Dresden (EK 514112015), Universität Leipzig (366/14-ff), and the Medical University Vienna (2207/2021).The study has been pre-registered at the German Clinical Trials Register (DRKS-00022034) and the Open Science Foundation (https://osf.io/zjxaf).

| Sample
Patients and one of their parents (generally the main caregiver) were recruited in the acute wards for paediatric oncology at the university hospitals in Dresden, Leipzig (Germany), and Vienna (Austria) be- Overall, n = 98 dyads were eligible for inclusion and informed by our staff about the study, and n = 81 (response rate: 82.7%) dyads participated at baseline (T1).N = 65 (retention rate: 80.2%) dyads participated at the 1-year-follow-up (T2), while n = 6 dyads were no longer eligible due to their child transitioning to palliative care or dying.Furthermore, n = 10 dyads dropped out of the study (reasons: no time, could not be contacted by phone/mail).There were no differences in sociodemographic and illness characteristics between participants who had dropped out and those who had completed the study (Appendix 1).Socio-demographic and illness characteristics of the sample are summarised in Table 1.The flow chart on participation is shown in Figure 1.

| Measures
Illness Perception Questionnaire-Revised (IPQ-R).Illness perceptions of parents and 12-18-year-old children were assessed with the short German IPQ-R 43 or with the IPQ-R puppet interview, 44,45 which is an age-adapted interview version for 4-11-year-old children.
The IPQ-R and the puppet interview ask about the child's and parent's respective self-rated illness perceptions on seven dimensions in line with the CSM: (1) identity (symptom attribution, henceforth called symptom-dimension), (2) timeline-acute/chronic (duration), (3)   timeline-cyclical (perceived trajectory as constant or cyclical), (4)   personal control (patient's perceived influence over the illness), (5)   illness coherence (perceived comprehension of the illness), (6) consequences (perceived impact of the illness on their life), and (7)   emotional representations (emotional impact).Except for the symptom-dimension, all items ask about current perceptions, whereas the symptom-dimension retrospectively asks about symptoms that have been experienced at any point during the course of the illness.Parents are asked to self-rate their illness perceptions on their child's illness (e.g.'I get depressed when I think about my child's HERZOG ET AL. cancer illness'), while children are asked about their perceptions concerning their own illness.The symptom-dimension is assessed with 14 dichotomous items (yes/no).Assessment of the other dimensions differs between the IPQ-R and the IPQ-R puppet interview: In the IPQ-R, they are assessed with 18 items on a five-point Likertscale ranging from 1 (strongly disagree) to 5 (strongly agree).For the puppet interview, the items have been dichotomised and are presented by two hand puppets as their own opposing statements of their own illness (e.g.'I am sad because of my cancer illness'.vs. 'I am not sad…').The child then chooses the statement which corresponds best with their own perception.All puppet interviews were conducted by one of the authors (KH) following an interview guide. 44e dimension scores are calculated by summing the items after reverse scoring, 46 with a maximum of one missing item per dimension.To obtain consistent scale formats, scores of the IPQ-R ques- alpha values between 0.72 and 0.86 (except timeline-cyclical). 43Sixmonth-test-retest-reliability of the English original was good, with >0.5 for all scales (except timeline-cyclical) in a sample of n = 76 adult patients with rheumatoid arthritis. 46The puppet interview has been validated with n = 41 children in acute treatment and n = 34 children in follow-up care and shows acceptable internal consistency for the dimensions timeline-acute/chronic and personal control. 44As the puppet interview is closely based on the IPQ-R questionnaire, scores and items are comparable to each other.For more information concerning the adaptation and validation of the puppet interview, see the publication by Herzog et al. 44 KINDL-R.The KINDL-R 47 was used to obtain child's selfreported HRQoL during the past week, as well as parent's proxyreport on their child's HRQoL (e.g.'In the past week, I/my child felt anxious or insecure'.).The questionnaire consists of seven dimensions: (1) Physical well-being, (2) emotional well-being, (3) self-esteem, (4) family, (5) friends, (6) daily functioning (school or preschool/nursery school), and (7) illness.The KINDL-R offers ageadapted versions for children (Kiddy-KINDL-R for 4-6-year-olds, Kid/Kiddo-KINDL-R for 7-18-year-olds) as well as proxy-report versions for parents/carers. 48The KINDL-R dimensions are measured on five-point Likert-scales ranging from 1 (never) to 5 (always), except for the Kiddy-KINDL-R, which uses a three-point Likert scale.The sum score is calculated by summing all items (except illness-dimension) following any reverse scoring.To obtain consistent scale formats, the sum scores were normalised to values between 0 and 100, with higher scores indicating higher child's HRQoL (score 100 = ((sum score-lowest possible sum score)/(highest possible sum score-lowest possible sum score))*100) 47 and a comprehensive HRQoL score was calculated as the mean of the self-and the proxy-report.The KINDL-R shows acceptable to good internal consistency and high convergent and discriminant validity. 49,50In the present study internal consistency for the KINDL-R version for 7-18year-old children and all parents were acceptable (α children = 0.780; α parents = 0.828), while Cronbach's Alpha for the Kiddy-KINDL-R could not be calculated due to too much missing data.

| Statistical analysis
Analyses were performed using IBM SPSS Statistics 27.0.
Imputation of missing values for independent variables.We only included cases for which both T1 and T2 were available.Only 5.00% of independent variables (child's and parent's illness perceptions at T1 and T2) were missing.Therefore, no multiple imputation was performed 51 and pairwise deletion method was used for all analyses.
Research question

Research question 2. First, Kendall-Tau correlation coefficients
were calculated for all study variables.To answer the question if illness perceptions at T1 predict variance in child's HRQoL at T2, an explorative hierarchical multiple regression analysis was performed.
In the first step, child's HRQoL at T1 was entered to control for baseline HRQoL levels.Then, all child's and parent's illness perceptions that correlated significantly with HRQoL at T2 were entered.
Assumptions for hierarchical multiple regression analyses were examined before interpretation of results (Appendix 7).
Explorative post-hoc analyses showed that there were no differences between mothers' and fathers' illness perceptions at baseline (Appendix 2) and trajectories (Appendix 3).

F I G U R E 1 Flow chart on participation (n = 65).
HERZOG ET AL.
In the last decades, increasing focus has been put on understanding any underlying psychological factors for psychological well-being in patients with chronic illnesses.For example, Note: T1 = baseline, T2 = one-year-follow-up.n refers to the sample size used to calculate the t-test, that is, data from both time points were available from n participants.Dimension scores for illness perceptions (IPQ-R) differ, with the symptoms-dimension ranging from 0 to 14 and all other dimensions ranging from 0 to 100.Higher IPQ-R scores indicate more associated symptoms, more negative perception of chronicity, cyclicity, and emotional representations, and more positive perception of personal control and illness coherence.Scores for HRQoL (KINDL-R) range from 0 to 100.Bold characters indicate a significant result (p < 0.05).
Abbreviation: HRQoL, health-related quality of life.
(i.e.expectancies, illness representations)' 18; p. 318 .While the association between illness perceptions and psychological well-being has been extensively investigated in adult patients with cancer or other chronic illnesses, research in paediatric oncology-especially with younger children-is still rare.To our knowledge, no longitudinal study has previously investigated illness perceptions in paediatric cancer patients from the age of 4 years.Therefore, the aims of this study were (1) to report longitudinal trajectories of illness perceptions in paediatric cancer patients and their parents, and (2) to explore predictive value of child's and parent's illness perceptions on the child's HRQoL.
Concerning our first research question, we found age-specific results for child's illness perceptions: 4-11-year-old children reported significantly more symptoms, as well as fewer negative consequences 1 year later, whereas all illness perceptions of 12-18-yearold children years remained stable.Perception of coherence improved in the whole sample, though this effect ceased to be significant at age-level.All parent's illness perceptions except an improved perception of a cyclical trajectory remained stable.As this is the first study to investigate illness perception trajectories in paediatric cancer patients, we will have to fall back on adult cancer patients/survivors for comparison: For example, Castro et al. 22 (young adults with cancer), Johansson et al. 53 (adult colorectal cancer patients), and McCorry et al. 21(adult breast cancer patients) found that illness perceptions remained stable over six to 12 months.
Richardson et al., 24 on the other hand, found that head and neck cancer patient's and their caregiver's emotional representations and illness concern improved over time, and that caregivers perceived more symptoms over time, while their perception of treatment control improved.In adult oesophageal cancer survivors, perceptions of consequences improved and they perceived fewer symptoms, while perceptions of chronicity and treatment control got worse, 25 showing that there might be differences between diagnoses or other patient/survivor characteristics, limiting comparability of our findings with those of other study populations.
In the CSM, illness perceptions are presented as being adjustable in a dynamic process over time as new information is obtained. 19,20wever, in line with findings from other studies with cancer patients, 21,22,53 we found illness perceptions to be relatively stable over a 1-year period, excepting younger children's perceptions of symptoms and consequences.A reason for this might be that their illness perceptions may be more modifiable than those of older children and adults, because at the beginning of treatment (when T1 took place), young children do not have much information about their cancer illness and they worry that it will have a severe long-term impact on their daily lives (e.g.not being able to return to school or do sports).
One year later, on the other hand, they received psychosocial support as part of their standardised treatment (e.g. according to the S3 guidelines 42 in German-speaking countries) that includes information about symptoms, side effects, and potential long-term effects, thereby increasing their ability to identify symptoms and promoting more realistic perceptions of potential long-term consequences.Also, by T2 children have discovered that not all dreaded/expected consequences have come to pass.Illness perceptions of older children and parents, on the other hand, remained stable over the period of 1 year, potentially because they already had illness perceptions prior to diagnosis that has been informed by media or school lessons, amongst others.At the time of diagnosis, they were then able to quickly adjust their illness perceptions to their specific situation via extensive information by medical staff and retain these perceptions over time.
Concerning our second research question, after controlling for baseline HRQoL, child's perceptions of more symptoms and more negative consequences were predictive of lower HRQoL 1 year later.
Findings from adults cancer samples have similarly found these perceptions to be independent predictors of patient's HRQoL, 14,17 with other independent illness perception predictors including emotional representations, 14,17 perception of chronicity, 14,17 personal control, and coherence. 24One potential explanation for our findings may be that children who experience more symptoms as part of their illness and treatment (e.g.nausea, fatigue, pain) may be more impaired and less able to participate in daily life activities, which in turn, may therefore lead to lower HRQoL.Similarly, the child's perception that, due to the cancer illness, the child cannot do all things they want to do anymore may be an accurate apprehension of a lower ability to participate in daily life activities and therefore lead to lower HRQoL.).Also, we did not investigate parent's HRQoL and associations with their illness perceptions.More than 80% of participating parents were mothers which may have biased our findings, as some research suggests that parental experience with their child's illness may be gender-specific, 56,57 while other research found few differences. 58In our study, we did not find any differences in mothers' and fathers' baseline illness perceptions and longitudinal trajectories.Most data collection took place while there were major restrictions due to the COVID-19 pandemic.We cannot establish the potential confounding effect of the pandemic on participants' illness perceptions and HRQoL.

| Clinical implications
Studies with adult cancer patients already have demonstrated the effectiveness of changing illness perceptions to improve patient's HRQoL.To understand potential starting points for similar interventions in paediatric oncology, one must first understand longitudinal trajectories and associations with psychological healthrelated outcomes.We demonstrated that most illness perceptions of paediatric cancer patients and their parents remained stable over a 1-year period.In an exploratory multiple regression analysis, we showed that child's perceptions of symptoms and consequences were important independent predictors of their HRQoL, highlighting these dimensions as potential starting points for cognitive psychological interventions.As the cancer diagnosis affects the whole family system, parental illness perceptions should be kept in mind in a psychological intervention aimed at improving child's and parental wellbeing even though they were not predictive of child's HRQoL in our exploratory multiple regression analysis.
At this point it ought to be mentioned that there is no consensus among researchers about indications of an illness perception intervention, that is, what do 'maladaptive/negative' illness perceptions constitute?Should all 'negative' illness perceptions be improved, or may the development of realistic illness perceptions be a target of an intervention, as well?For example, decreasing perception of symptoms is not desirable in psychosocial care: It is important that patients and caregivers understand and identify symptoms and side effects of treatment correctly; therefore, the number of symptoms mentioned in the IPQ-R over time may increase.Medical symptom control goes hand in hand with activation via psychosocial and supportive therapies that aim to reduce time spent in bed and improve participation in daily life (e.g.music therapy, sport therapy, play therapy), irrespective of symptoms the child might experience.
Generally, whether an illness perception intervention should improve negative illness perceptions or promote realistic illness perceptions should be carefully considered by psychosocial staff according to the patient's current needs.

| Conclusion
Summarising, it is important to evaluate illness perceptions in paediatric cancer patients and their parents as they provide useful information for the psychosocial and medical staff.To our knowledge, this is the first longitudinal study investigating illness perception trajectories and their associations with HRQoL in paediatric oncology.The stability of illness perceptions in our sample showed that they may not necessarily be influenced by the additional knowledge gained about the illness during the treatment process.
Therefore, one future research focus may be the development and evaluation of a psychosocial intervention that focusses on illness perceptions.

tween 10 /
2019 and 11/2022.All families who met the inclusion criteria and who were in acute treatment at the time of recruitment were asked to participate.Inclusion criteria were (1) oncology patients aged 4-18 years with any oncological diagnosis and one parent, (2) first diagnosis at least 1 month ago.Exclusion criteria were (1) inability to understand questionnaire/interview (e.g.due to language barriers or cognitive impairment; assessed by psychosocial staff), (2) palliative care.All families received treatment as usual according to the S3 guideline for psychosocial care in paediatric oncology and haematology, 42 which includes interdisciplinary, resource-oriented supportive therapy for all children, parents, and any other relevant family members.After obtaining written informed consent, children and parents completed measures of illness perceptions (interview or questionnaire, explained below) and HRQoL.Information on sociodemographic and medical characteristics of the child was obtained from the parent.One year later, children and parents were contacted by phone and asked to participate for a second time.Questionnaires for families with older children (aged 12 and older) were posted to participants with a free return envelope, while interviews for families with younger children (aged 4-11 years) took place at the hospital after a scheduled control examination.Usually, children and parents completed the questionnaire/interview at the same time or within a few days of each other, with one of the authors (KH, clinical psychologist) being available to assist with any questions.
tionnaire and the puppet interview were normalised to values between 0 and 100 (score 100 = ((sum score-lowest possible sum score)/ (highest possible sum score-lowest possible sum score))*100).Higher scores indicate more associated symptoms, more negative perception of chronicity, cyclicity, consequences, and emotional representations, and more positive perception of personal control and illness coherence.The German IPQ-R has been validated with n = 817 adult rehabilitation patients and shows sufficient internal consistency with
Descriptive statistics and changes (paired-samples t-tests) in child's and parent's illness perceptions, and child's HRQoL from baseline (T1) to follow-up (T2; research question 1).
Stanton et al. concluded in his review on psychosocial interventions for adult cancer patients that 'promising classes of mediators include alterations in cognitions T A B L E 2 Prediction of HRQoL using multiple regression analysis, stepwise entry for the outcome child's HRQoL at T2 (n = 56, research question 2).HRQoL was force-entered in the first step to control for baseline levels of HRQoL.
a Child's HERZOG ET AL.