Fear of progression in parents of childhood cancer survivors: A dyadic data analysis

Fear of progression (FoP), also referred to as fear of cancer recurrence, is gaining increasing interest in survivorship research as it constitutes a great burden for patients and relatives. However, only little is known about FoP in parents of childhood cancer survivors. The objective of this study was to investigate the impact of FoP on quality of life (QoL) in parental couples.

Fear of progression (FoP), also known as fear of cancer recurrence or fear of relapse, is gaining an increasing interest in adult survivorship research. 6 FoP describes a rationally explainable response to the real threat of a potentially life-threatening cancer disease. 7 It can be defined as the "fear that the illness will progress with all its biopsychosocial consequences, or that it will recur." 7 FoP in adult cancer patients has been researched intensively. 6 For a long time, FoP in parents of childhood cancer survivors was only implicitly described in qualitative studies. 4,[8][9][10][11] The Fear of Progression Questionnaire for the parental perspective (FoP-Q-SF/PR) was published recently and offers the opportunity to investigate parental FoP systematically. 12,13 Parental FoP seems to be associated with anxiety, depression, posttraumatic stress and a low quality of life (QoL). 13 Further, significantly negative correlations of FoP with time since diagnosis, anxiety coping skills of the parents, parental age, the child's current medical condition and the number of siblings were found. 12 Significant gender differences in FoP levels of mothers and fathers could not be found, 12,14 nor was the child's age at diagnosis associated with parental FoP. 12 According to an expert online survey in Germany, health care professionals are frequently confronted with parental FoP in clinical practice. 15 Additionally, a recent interview study identified parental FoP as an impeding factor when reintegrating into daily life after the end of treatment. 16 Overall, research on FoP of parents of childhood cancer survivors is still in its infancy. Previous studies used a broad sample of parents from children still receiving cancer treatment to long-term survivors. [12][13][14] Further, most of the parents that participated in these studies already received psychosocial support. [12][13][14] A family perspective has not been considered yet, however, according to research on FoP in adult patients, FoP seems to be a "family affair." 7 The aim of this study was to investigate how FoP impacts the QoL of affected parents using a dyadic data analysis approach. We decided for QoL as the outcome, since it is commonly used for the assessment of the multidimensional psychosocial situation of pediatric cancer patients and their families. [17][18][19] We additionally examined the association between FoP and various QoL subscales (functioning, satisfaction with the family situation, emotional distress, self-development, general well-being) to elaborate on different aspects of QoL.
Based on earlier results we examined the following hypotheses 12,13 : 1. Actor effect: FoP in mothers/fathers of childhood cancer survivors is negatively associated with their own QoL (overall QoL and QoL subscales).
2. Partner effect: FoP in mothers/fathers of childhood cancer survivors is negatively associated with their partners' QoL (overall QoL and QoL subscales).

| Design
This dyadic data analysis was performed within a prospective observational study with a longitudinal mixed-methods design. The overall study has been described in a study protocol 20 and was reviewed and approved by the Ethics Committee of the Medical Chamber of Hamburg (number: PV5277). The STROBE Statement was applied in the reporting of this article. 21

| Participants and procedure
In this study, we focused on the most frequent pediatric cancer diagnoses in Germany, leukemia and central nervous system (CNS) tumors. 2 We included biological parents and other caregivers of children (<18 years of age at diagnosis) after the end of intensive cancer treatment (eg, chemotherapy). Exclusion criteria were assessed by the health care providers in the clinics and were refusal of participation, physical and/or mental burden (applicable if the study participation would be overly burdensome), cognitive limitations and insufficient German language skills. Written informed consent was obtained. The dyadic analysis was performed with data of opposite-sex couples that both participated in the study and that were still in a couple relationship. Same-sex couples were excluded from the analysis because the statistical model requires distinguishable couples (in our study distinguishable by gender). 22 Parental couples were consecutively recruited in Germany from Parents were surveyed at the beginning of the rehabilitation measure.
A flowchart of the recruitment process can be found in Appendix A.

| Sociodemographic and medical data
Sociodemographic information was extracted from the parents' questionnaire (age and gender of parents and child, number of siblings, parent's education and employment status, relationship status of the parental couple). Depending on the recruitment path, medical information (diagnosis, time since diagnosis) was extracted from the parents' reports or the physicians' questionnaire in the rehabilitation clinic. 20

| Fear of progression
In this study, we used the 12-item Fear of Progression Questionnaire -Parent Report (FoP-Q-SF/PR). 12 Items are scored on a 5-point Likert scale ranging from never (1) to very often (5). A sum score between 12 and 60 can be calculated with higher scores indicating a higher level of fear of progression. Schepper and colleagues recommended preliminary cut-off values based on the mean value ±1 SD: 12 to 25 low FoP, 26 to 46 moderate FoP and 47 to 60 high FoP. 12 The FoP-Q-SF/PR shows good reliability and validity. 12,13

| Quality of life
Parental QoL was measured with The Ulm Quality of Life Inventory for Parents (ULQIE). 23 The ULQIE measures QoL in parents of chronically ill children with 29 items on five subscales focusing on the past week: Functioning, satisfaction with family situation, emotional distress, self-development and general well-being. 23 Items are scored on a 5-point Likert scale from never (0) to always (4). A total sum score and subscale scores can be calculated. High sum scores indicate high QoL. The ULQIE has adequate psychometric properties. 23

| Actor-partner interdependence model (APIM)
When analyzing parental couple data, it is important to note that the dyad members' responses might be dependent. 24 This would lead to a violation of the independence assumption in significance testing. The actor-partner interdependence model (APIM) directly measures nonindependence by considering the dyad in the analysis. 22,24 The APIM examines actor effects (effect of one's own predictor on one's own outcome) and partner effects (effect on one's own predictor on the partner's outcome). 22 We applied the APIM to calculate the effect of one's own FoP on one's own QoL (actor effect) and the effect of one's own FoP on the partner's QoL (partner effect). 22 We conducted the APIM using structural equation modeling (SEM) with maximum likelihood estimation. 25 The SEM was calculated with the program lavaan 26 within the app APIM_SEM. 27 The parameter k is the ratio of the partner effect to the actor effect and was calculated to detect patterns in the APIM (couple pattern, contrast pattern, actor-only pattern). 28 The child's diagnosis (CNS tumor vs leukemia), time since diagnosis and the child's age were examined as potential confounders.
Controlling for confounders is recommended if variables correlate with both the predictor (FoP) and the outcome (QoL) or with the outcome only. 27 3 | RESULTS

| Sample characteristics
Eight hundred ninety-nine families were potentially eligible for participation in the study (Appendix A). Three hundred twelve families participated in the survey. Due to the complex recruitment process, we could only indicate the reasons for the non-participation of the 60 families that were recruited at the rehabilitation clinic (n = 21 refusal of participation, n = 14 insufficient German language skills, n = 12 physical and/or mental burden, n = 3 cognitive limitations, n = 10 not specified). Five hundred twenty-seven families that were recruited via the study registries did not participate because either they did fulfil the exclusion criteria (refusal of participation, physical and/or mental burden, cognitive limitations or insufficient German language skills) or the health care providers in the clinics were not able to inform them about the study. From the 312 families that participated in the study, five families were excluded subsequently (n = 2 missing consent forms for participation, n = 2 wrong diagnosis, n = 1 incorrectly answered questionnaires because of limited German language skills). In this study, we analyzed the dyadic data of 197 parental couples (107 families recruited at the rehabilitation clinic and 90 via the study registries). Table 1

| FoP and QoL in mothers and fathers
Descriptive data on parent's FoP and QoL as well as gender differences are presented in Table 2.

| Confounders
We found significantly positive correlations between FoP and diagnosis (r = .206, p < .001), time since diagnosis (r = .194, p < .001) and child's age (r = .108, p < .05). FoP in parents of CNS tumor survivors was significantly higher than in parents of leukemia survivors (t [389] = −4.160, p < .001). Longer time since diagnosis and higher child's age were associated with higher levels of FoP. However, there was no significant difference in the QoL levels of parents of CNS tumor and leukemia survivors and no significant correlations with QoL. Therefore, we did not control for any confounders in this dyadic analysis.

| Main findings
Our results revealed a significantly negative actor effect from FoP on the overall QoL for mothers and fathers (Table 3, Figure 1). We also found significantly negative actor effects for both parents for the QoL subscales functioning, emotional distress and general well-being. A significantly negative actor effect for the subscale satisfaction with the family situation was only found for fathers.
The results revealed no significant actor effects for the subscale self-development.
No significant partner effects were found for any of the QoL scales. However, the intra-dyadic correlation of mothers and fathers of the same dyad was significantly positive for both FoP (r = .431, p < .001) and the overall QoL (r = .277, p < .001). An actor-only model (k = 0) for the effect from FoP on the overall QoL was plausible for mothers and fathers since 0 was included in both confidence intervals. 28

| DISCUSSION
The aim of this study was to investigate the impact of FoP on QoL in parents of childhood cancer survivors using a dyadic data analysis approach. Moderate to high levels of FoP were found in 81% of the parents after the end of intensive cancer treatment. FoP levels in our sample were only slightly lower than in a sample of parents of children up to 10 years after diagnosis of childhood cancer. 13 The mean QoL sum score in our sample was higher than the QoL score in a population of parents of childhood cancer patients in the first 2 weeks post diagnosis, but lower than the QoL score of healthy controls. 29 In contrast to earlier studies, we found a significant gender difference between mothers and fathers with mothers reporting higher levels of FoP. 12,14 Additionally, mothers reported a lower QoL than fathers. It is possible that mothers reported higher burden because they might be the primary carer of the child. Only a few mothers but most of the fathers were working full time. Besides, men reporting lower levels of psychosocial burden is a common finding. 30 As expected, our results display that FoP in mothers and fathers affected the overall QoL (actor effects). We also found a negative impact of FoP on parental functioning, emotional distress and general well-being. FoP also affects the father's satisfaction with the family situation. Selfdevelopment of parents was not affected by their FoP.
The results of our study also support the assumption that FoP is a "family affair." 7 FoP in mothers and fathers of the same family were highly correlated. We found a higher intra-dyadic correlation for FoP than for QoL. It is possible that FoP levels are more similar in couples because FoP describes the worry about the couple's child. Further, FoP might be highly influenced by family factors (eg, time since diagnosis) whereas QoL is a multidimensional outcome that might also be affected by various factors outside the family system (eg, job satisfaction). However, there was no association between the FoP of one parent and the QoL of the other parent. One possible reason for this result is the high intra-dyadic correlation between FoP of mothers and fathers from the same family. The partner effect may be confounded by the shared dyadic variance. 31 FoP was higher in parents of CNS tumor survivors in comparison to parents of leukemia survivors. A correlation of FoP with the actual risk of progression which is higher in CNS tumor survivors is possible. 2 FoP might be associated with a higher age of the child because parents of older children potentially perceive their child's limitations as more severe. A longer time since diagnosis was also associated with higher FoP levels in parents. From a clinical perspective, this might be caused by extending intervals between aftercare appointments. However, since 82% of the families were surveyed in the first 2 years after diagnosis, we cannot make a reliable statement on the course of FoP.

| Study limitations
For reasons of data protection, we could not contact the parents directly. Thus, we recruited the parents via a rehabilitation clinic and study registries. Our complex recruitment process did not allow for a systematic non-responder analysis. Therefore, the generalizability of the results might be limited. Furthermore, we excluded parents with particularly high levels of physical or mental burden due to ethical considerations. Hence, an underreporting of FoP in our study is possible.
This study has also several strengths. We surveyed parents of childhood cancer survivors throughout Germany and reached a large sample size. Moreover, this is the first study that investigated the association of FoP and QoL in parents of childhood cancer survivors in consideration of intra-dyadic processes. 22,24

| Clinical implications
The results of this study display that most parents show moderate to high levels of FoP even after intensive treatment has ended. Additionally, this study provides evidence on the negative impact of parental