Psychosocial functioning of adult siblings of Dutch very long‐term survivors of childhood cancer: DCCSS‐LATER 2 psycho‐oncology study

To describe psychosocial outcomes among adult siblings of very long‐term childhood cancer survivors (CCS), to compare these outcomes to reference populations and to identify factors associated with siblings' psychosocial outcomes.


| INTRODUCTION
Siblings of children with cancer are impacted by their brother or sister's disease. During treatment, siblings may experience disruptions of daily and academic life, changes in family relations and feelings of worry, loneliness and neglect. [1][2][3][4] During this period and even after treatment, siblings may need psychosocial support. 5 The Integrative Trajectory Model of Pediatric Medical Traumatic Stress describes that although most families recover over time after the diagnosis of a pediatric illness, a small proportion continues to experience problems, even after months or years. 6 Most literature on siblings' wellbeing focuses on children below the age of 18. A systematic review shows that emotional and behavioral functioning of adult siblings was similar to references and states that methodological limitations hamper conclusions about health related quality of life (HRQoL). 7 A more recent study shows lower mental HRQoL in adult siblings than in peers. 8 Regarding post-traumatic stress, adult siblings do not experience more symptoms than controls. 9 Besides generic outcomes such as HRQoL, disease-specific outcomes such as benefits and burden of having a brother of sister with cancer may be relevant. Literature on siblings' burden of and positive experiences with cancer is scarce, 10 but post-traumatic growth seems to be higher in young-adult siblings than in controls that were asked about a major stressful event. 9 Minor siblings report higher self-esteem as a positive effect. 11 Previous research into factors associated with psychosocial functioning of adult siblings points out that females and older siblings are at risk, 9 as well as adult siblings who were older at diagnosis and older than the childhood cancer survivor (CCS). 12 Further risk factors are lower income, lower education, nonwhite race, lower perceived social support and health problems, whereas cancer-related factors typically are not associated with wellbeing of siblings. 7 It is especially relevant to study the wellbeing of adult siblings of very long-term CCS since knowledge on very long-term psychosocial functioning in siblings is limited and it was found that siblings' risk for mental health problems starts to diverge from controls from approximately 15 years after diagnosis, 12 while the follow-up period of most studies is shorter. The aim of the current study is to describe HRQoL, anxiety, depression, post-traumatic stress, benefit and burden and self-esteem among adult siblings of very long-term CCS and to compare these outcomes to reference populations. Furthermore, we aimed to determine factors associated with siblings' psychosocial outcomes. and result in a total scale and three subscales: re-experiencing, avoidance and hyper arousal. Presence of at least one symptom in all three scales is considered subclinical PTSD 19 and presence of at least one re-experiencing, three avoidance and two hyper arousal symptoms is considered symptomatic PTSD. 20,21 Psychometric properties are adequate. 20 Cronbach's α for the total scale in the current study was 0.70.

| Participants
Self-esteem was measured with the Rosenberg self-esteem questionnaire, which consists of 10 items (4-point Likert scale). A higher score indicates higher self-esteem (range 10-40). Psychometric properties are good. 22 Cronbach's α in the current study was 0.90. Reference data from the general Dutch population are available. 23 Benefit finding and disease-related burden of having a sibling CCS were measured with the Benefit and Burden Scale for Children (BBSC 24 ), minimally adapted for the use in adults with approval of the original author. The questionnaire consists of two scales with 10 items (see Table 1) on a Likert scale ( [1] 'Not at all true for me', [2] 'a little bit', [3] 'somewhat', [4] 'quite a bit', [5] 'Very much true for me'): benefit finding (Cronbach's α 0.90) and disease-related burden (Cronbach's α 0.77) of childhood cancer. Mean item scale scores were calculated. Scores of siblings <4 years old at the time of diagnosis of the CCS were not used.
Associated factors consisted of demographic characteristics (siblings' age at participation and at diagnosis of CCS, sex, level of education, number of siblings, survivor's age) and cancer-related characteristics of the CCS (primary childhood cancer diagnosis, metastasis and recurrence of primary tumor). Demographic factors were obtained from the DCCSS-LATER 1 or 2 study. 25 Cancerrelated factors were obtained from the DCCSS-LATER registry. 25

| Statistical analyses
Differences between characteristics of participants and non-participants/LATER cohort were tested with independent t-tests and chisquare tests, having Cohen's d and Cramer's V as effect sizes. Outcomes were analyzed descriptively. In addition, HRQoL, anxiety, depression and self-esteem were compared with references of comparable age for males and females separately, using Mann Whitney-U tests with effect size r, and chi-square tests.
Mixed model analyses were conducted to determine factors associated with psychosocial outcomes, controlled for age and sex.
Random intercepts were included to account for dependency of outcomes of siblings of the same CCS.
p-values ≤0.05 were considered statistically significant, except for comparisons between siblings and references, where a Bonferroni correction was applied for the number of scales within the outcomes.

| Participants
In total, 505 of 1479 siblings (34% of siblings invited through their CCS, 64% female, mean age 37.5 years, mean time since diagnosis 29.5 years) of 412 CCS gave informed consent and completed at least one questionnaire. Participants and non-participants differed in sex and age, but not in education ( Table 2). CCS' diagnosis differed from the LATER cohort. 25 All effect sizes are small to moderate.

| Outcomes
Several small differences (r = 0.08−0.18) between siblings and references were found on psychosocial outcomes (Table 3)  and (sub)clinical depression than references, while no differences were found on anxiety. Self-esteem was higher for females compared to references, but not for males. The percentage of siblings reporting subclinical and symptomatic PTSD because of the CCS's disease was respectively 4.4% and 0.4% for females and 4.3% and 0.6% for males.
Mean scale scores for benefit and burden were 2.3 ("a little bit" to "somewhat") and 1.1 ("none at all' to "a little bit") respectively, indicating less burden than benefit (Table 1). For benefit, the items with the highest score were "I have learned what is really important in life" (mean 3.2), "I have learned to be happy and to enjoy good things" (mean 3.0) and "I became a stronger person" (mean 2.7). For burden the items with the highest score were "I am afraid to make other T A B L E 2 Sociodemographic characteristics of siblings and CCS cancer-related characteristics.

Participants
Non-participants a  people sad or upset" (mean 1.5), "I can't enjoy like I used to anymore" (mean 1.3) and "I am less hopeful about life" (mean 1.3). Tables 4 and 5 show the associations of sociodemographic and cancer-related factors with psychosocial outcomes. Age group 12+ at diagnosis had better sleep (compared to 0-4 years) and more perceived benefit (compared to 4-12 years). Having >1 sibling was associated with less pain compared to having 1 sibling. Being

1408
but <1% seems very low compared to minor siblings shorter after diagnosis, of whom 22% had PTSD symptoms related to the childhood cancer of their brother or sister. 31 These percentages are based on the DSM-IV criteria, since no measurement based on the DSM V criteria was available at the time of inclusion. Self-esteem was comparable to references for males and slightly higher for females.
Higher self-esteem of siblings was also found in minor siblings around 9 years after diagnosis. 11 Siblings reported more benefit than burden of their CCS disease.
Items that had the highest means correspond with previously identi-

| Associated factors
No clear trend is visible in the studied associated factors. Effect sizes are small to moderate and no factor impacts more than one outcome, other than being 12+ years old at diagnosis, which is related to better sleep and more perceived benefit. In previous research, being older at diagnosis was only associated with worse health behaviors such as use of tobacco. 32 Furthermore, we found only small effects of birth order, in worse cognitive functioning for younger siblings and worse sleep for older siblings. Previous studies into birth order found that having an older CCS is a risk factor for higher rates of reaching out for mental health support 12 and in siblings of children with a chronic disease, the older siblings seem to be more impacted. 33 Literature about adult siblings is lacking. Concerning cancer-related factors, we expected higher risk for siblings of CCS with CNS tumors, because of worse outcomes for CCS with CNS themselves. However, having a child with a CNS tumor was also not related to psychosocial functioning in parents of long-term CCS. 34 Apart from an association of small to medium size between siblings of CNS CCS and pain, we found no associations between the outcomes and cancer-related factors, which is substantiated in literature. 8,11,35 Following these results, factors that impact long-term psychosocial functioning of siblings of CCS might not differ from factors that impact the functioning of the general population.

| Clinical implications
Literature suggests siblings are at risk for problems in psychosocial functioning closer to diagnosis and the need for sibling support is stressed in the standards of care. 5 The PAT could be used for psychosocial risk screening, which may help providing early intervention to prevent long-term consequences for families at risk. 36 Besides, age adequate information about the diagnosis and treatment is recommended and it is now commonly part of psychosocial support which will further empower the siblings of now diagnosed patients who will be the CCS in the future. Nonetheless, the current findings are reassuring and can be used in psycho educating families and potentially in comforting families.

| Limitations
This study is one of the first to report on very long-term psychosocial functioning of siblings of CCS in a nationwide cohort. Besides generic outcomes such as HRQoL we looked into benefit and related burden of siblings. Our sample is large, which is of added value to the shortage of research on big samples. 7 To date, research on long-term functioning has been limited to approximately 20 years after diagnosis, whereas our sample's mean time since diagnosis is 29.5 years.
Some limitations of the study should be taken into account in the interpretation of the results. Information on CCS health impairment, family functioning or psychosocial support that siblings received was not collected, where it could have impacted the current wellbeing. 11,35,37 Further research could investigate these factors to identify siblings at-risk. Another limitation is that siblings were compared to the norm populations, so that siblings' functioning might not be accurately contextualized and which makes comparison of results between studies difficult. Concerning methodology, one of the multivariate models could not be adjusted for dependency of siblings of the same CCS, but because the intra-class coefficient was not significant, a non-adjusted model could be used instead. 38 The internal consistency of two HRQOL scale scores were questionable (0.63 for aggressive emotions and 0.68 for pain), which means results should be interpreted with caution. Siblings were only eligible after the CCS gave consent, which may have caused selection bias. Besides, bereaved siblings were outside the scope of the current research question. Findings are mixed in terms of vulnerability of that specific group. 12,36 The reported response rate (34%) is a reflection of the invited siblings after the CCS gave consent and not of the entire sibling cohort. The exact response rate of the entire sibling cohort is unknown but will be lower than 34%, which may have decreased the generalizability of the results, even though the differences between participants and non-participants are small.

| CONCLUSION
On the very long-term, siblings in our sample do not have impaired psychosocial functioning compared to references. Factors related to the disease of their brother or sister seem not to impact siblings' psychosocial functioning. JOOSTEN ET AL.