Peanut allergy impact on productivity and quality of life (PAPRIQUA): Caregiver‐reported psychosocial impact of peanut allergy on children

Limited research has examined the impact of peanut allergy (PA) on children using validated instruments to assess psychosocial burden and the factors influencing burden.

reaction severity, or child sex and child HRQL. 10,11,14,15 Conflicting findings on the relationship between children's age and HRQL have been reported, with one study finding that older children (aged 6-12 years) had worse HRQL than children aged up to 5 years 12 and another study showing no significant association between age and HRQL. 11 In PA specifically, one study found that child anxiety and parenting stress (measured using the Parenting Stress Index) predict parent's proxy report of their child's HRQL, and child anxiety, parenting stress, length of diagnosis and AAI experience predict child's self-report of their HRQL. 16 No studies have been identified investigating the relationship between perceived disease severity or child sex and HRQL in PA specifically.
Previous research has suggested that caregivers' sex is associated with their ratings of their child's HRQL. In a study of FA, both parents scored their food-allergic child's HRQL better than the child's own assessment, whereas mother-reported child HRQL was significantly correlated with limitations in the child's social life, and father-reported child HRQL was associated with limitations in the family's social life. 14,15 In a study of families with a child with PA, mothers felt that there was a greater impact on HRQL for their peanut-allergic child, compared to that reported by siblings, fathers or the peanut-allergic children themselves. 9 Due to the limited PA-specific research exploring factors associated with child HRQL, and limited use of validated FA instruments to assess HRQL in PA, the objectives of this study were to (a) describe the caregiver-reported impact of living with PA on children's HRQL and evaluate the roles of PA severity and child sex; (b) explore how caregivers' sex and anxiety are associated with the proxy report of their child's HRQL; and (c) identify significant predictors of a child's HRQL. This paper reports data from a survey conducted in the UK with parents of children aged 4 to 15 years. Given the age range of the children, the effect of age was also considered by splitting the sample for those aged 4-10 and 11-15 years, reflecting meaningful context changes such as school environment and increased independence.

| Study design
This study used a cross-sectional design to survey caregivers of children and teenagers aged 4 to 15 years to provide reports of their child's HRQL. The study was reviewed and approved by the Freiburg Ethics Commission International (FECI) prior to participant recruitment (FECI code: 017/1938; date: 20/11/2017).

Conclusions and Clinical Relevance:
Caregivers report that children with PA experience high levels of psychosocial burden, particularly those with more severe PA and a reaction history. Interventions to decrease caregiver anxiety and reaction frequency may help reduce the child's burden. Self-report studies in children with PA would help confirm these findings.

K E Y W O R D S
food allergy, paediatrics, quality of life

| Participants
Participants were recruited through a survey recruitment panel.
Participants were eligible for the study if they were a parent or primary caregiver of a child (aged 4-15 years) with medically diagnosed PA who had experienced at least one reaction to peanuts in their day-to-day life (other than a food challenge). Participants were also asked to rate the severity of their child's PA using categories of mild, moderate or severe. In order to ensure diversity in the severity of PA in the sample, a quota was set for a maximum of 20% of participants who perceived the severity of their child's PA as mild and a minimum of 80% who perceived their child's severity as moderate to severe. At least 25% of the moderate/severe samples were required to have used an AAI, and a minimum of 10% were required to have experienced a life-threatening event.
However, in the end, the sample quotas were achieved without the need to enrich or limit any numbers. All participants came from the UK.

| Procedures and measures
Eligible participants were informed that the study involved completing an online survey, was expected to take up to 30 minutes, and would be about their and their child's experiences of living with PA.
Participants were asked to read the information provided, and if interested in participating were asked to give their consent online prior to completing the online survey. Participants received nominal reimbursement for participating in the study. The EQ-5D-Y (proxy 1) 17 is a validated generic health status measure. 18 Participants report their child's current health on five dimensions (mobility, self-care, pain and discomfort, usual activities, anxiety and depression) from no problems to extreme problems. The responses are converted into a single index value where a score of 1 represents full health and a score of 0 represents dead. As there are currently no value sets for use in children and teenagers, the UK value set was applied to score the instrument. 19 Participants also rated their child's current health on a 0-100 visual analogue scale (VAS). To the best of our knowledge, this is the first study to show proxy-reported impacts of PA on the EQ-5D-Y.
The FAQLQ-PF 20 is a 30-item questionnaire assessing three domains (emotional impact, food anxiety, and social and dietary limitations). Each item is scored on a 7-point scale ranging from 0 (no impact) to 6 (extreme impact); each domain is scored as an average of the items with the total score representing the mean of the domain scores. The instrument has been validated with internal consistency of α = 0.91 (age 6-12) for the total score and α = 0.90-0.94 for the domains (age [6][7][8][9][10][11][12]. 20 In a study of child-parent pairs, the FAQLQ-PF correlated moderately with the FAQLQ-CF (ICC = 0.57; P < .001), 21 a self-report instrument for children aged 8-12 years, which is validated for patients with PA. 22 The children's self-report scores were higher (worse) than those reported for them by their parents (P < .001). 21 The FAIM-PF 23 is usually used in conjunction with the FAQLQ and consists of eight items in total, four items related to the parent's self-reported perception of their child's likelihood of allergic reaction events (accidentally ingesting the food to which they are allergic, having a severe reaction, dying from their FA, effectively treating themselves when needed) and the same four items related to the parent-reported perception of their child's beliefs. Each item is answered on a seven-point scale, with FAIM scores ranging from 1 (low perceived disease severity) to 7 (high perceived disease severity). Total FAIM scores were significantly correlated with FAQLQ scores for corresponding age groups, and with DBPCFC scores in peanut-allergic adults and teenagers, with a trend for children. 22 The for males and females of 15.8% for anxiety and 6.9% for depression. 25 The HADS has been validated many times in different populations, with internal consistency for the anxiety scale ranging from α = 0.68-0.93 (mean: 0.83) and the depression scale ranging from α = 0.67-0.90 (mean: 0.82). 26

| Analysis
Demographic and clinical characteristics were analysed using descriptive statistics. The FAQLQ-PF and HADS were scored according to their published scoring instructions. 20,24 The EQ-5D-Y was scored using the UK EQ-5D-3L adult value set. 19 Pearson's, point biserial and Spearman's correlations were conducted as appropriate (based on scale response type and sample distribution) to explore relationships between variables. Correlations were interpreted in line with Cohen's guidelines: small, 0.10-<0.30; moderate, 0.30-<0.50; and large, 0.50. 27 To explore the impact of different variables on child burden, two-way analyses of variance (ANOVAs) and Kruskal-Wallis tests were conducted. To further explore which variables significantly predicted child burden, multivariate ordinary least squares linear regression models were conducted, with each of the FAQLQ-PF domains and total score included as dependent variables in separate models. The number of reactions in the past 12 months and number of reactions in the child's lifetime were included as categorical variables in the regression analysis: 0, 1, 2 or more reactions in the past 12 months, and 0-2, 3-5 and 6 or more reactions in their lifetime.
Results of statistical tests were considered statistically significant if P-values were equal to or below .05. All analyses were conducted using STATA (version 16.0).

| RE SULTS
The demographics and clinical characteristics of the participants and their children with PA are displayed in Table 1

| Impact of PA on children's HRQL
Summaries of the impact of PA on the HRQL of children with PA, based on proxy-reported generic (EQ-5D) and FA-specific outcome measures (FAQLQ and FAIM) for the sample as a whole and by age and severity groups, are presented in Figure 1 and Table 2 are available for the EQ-5D-Y; however, the population norm for young adults (aged 18-24 years, EQ-5D-3L) is 0.940. 28 Therefore, the proxy-reported utility for the sample as a whole (0.873) is lower than the norm for young adults.
When considering age group, Figure 1   The FAQLQ total and domain scores also followed the same pattern with regard to the proxy-reported severity of PA ( Table 2). The Similarly, there were no main effects of sex or interaction effects between sex and severity in the inferential analyses.

| Relationship between caregiver sex and anxiety and proxy report of child's HRQL
Descriptive statistics of the child HRQL outcomes by caregiver sex and HADS "probable clinical anxiety" are presented in Table 3. This shows that greater PA burden, as measured by the EQ-5D-Y, FAQLQ and FAIM, was reported amongst children whose caregivers had "probable clinical anxiety." Caregiver sex was not significantly associated with caregiver and FAQLQ-FA (F = 4.14, P < .05). There was no significant sex by anxiety interactions; the effect of caregiver anxiety on their child's proxy-reported QoL did not differ by sex.

| Predictors of HRQL in children with PA
Caregiver anxiety, caregiver confidence, PA severity, life-threatening events and number of reactions were significantly correlated with several measures of child burden (Appendix S1). Other potential predictors (AAI use, child's sex) were not significantly associated

| D ISCUSS I ON
This survey describes the caregiver-reported burden experienced by children with PA in the UK. In line with recent findings, a child's sex was not associated with proxy-reported child burden, 12 demonstrating that male and female children experience a similar level of psychosocial burden due to their PA. Increased PA severity, however, was significantly associated with increased burden among the older children; younger children appear to have a similar level of burden overall, with less variability due to severity of PA. This finding is in line with expectations, given the more uniform impact on a young person, for whom the presence of the PA may have similar effects regardless of severity, whereas an older child may be given more freedom and autonomy and be more able to evaluate a situation based on their and their caregiver's perception of their PA severity.
Caregiver anxiety was significantly correlated with almost all proxy-reported measures of child impact, partially supporting previous research finding maternal anxiety is significantly associated with child's proxy-reported HRQL in FA. 11,13 However, while caregiver sex was also significantly associated with child's PA burden for almost all FAQLQ outcomes, our findings suggest that male caregivers report higher levels of burden experienced by their child than female, whereas previous research has found mothers reported greater burden for their child than fathers reported. 9 It is possible that this novel finding is due to previous studies lacking male caregivers in their sample; alternatively, this may reflect the increase in shared parenting and greater involvement of males in childcare. Rates of anxiety were consistent across caregiver sex, so the higher burden reported by males was not due to the fact males were more anxious.
Relatedly, caregiver anxiety was associated with the child's PA reaction history, and "probable clinical" anxiety was only reported at levels above UK population norms amongst caregivers reporting their child had moderate/severe PA. Given the cross-sectional nature of this study, it is not possible to establish whether the caregiver's anxiety was caused by their child's experiences with PA, or if their anxiety led to higher levels of PA burden being reported. Given the association between caregiver anxiety and objective markers of peanut allergy severity, the former seems likely, but longitudinal research would be required to confirm this definitively.
When controlling for multiple predictors of the child's HRQL outcomes, caregiver anxiety and the experience of two or more reactions in the past 12 months were the most consistent independent predictors of child HRQL. While the descriptive results showed   reported an overall proxy-reported EQ-5D-Y utility of 0.873, compared to 0.875 reported here; likewise, their adolescent utility values also reflected better HRQL (0.91) than the child utility values (0.84). 29 Based on review of the literature, this study is the first to report caregiver-reported impacts of PA on the EQ-5D-Y, and thus may provide a benchmark for future studies using this instrument in caregivers of subjects with PA.
Some limitations should be considered when interpreting the results from this study. All data generated in this study relied upon parent-proxy reports of child HRQL. Previous studies in FA have found that parents reported significantly better HRQL for their child than the children themselves reported 14,15,30 ; therefore, the results of the current study may underestimate the burden experienced by children with PA. In addition, self-reported data are subject to recall bias and inaccuracy. As discussed above, the study also used a cross-sectional design, so it is not possible to establish cause and effect and it is unknown whether caregiver anxiety influences their proxy report of their child's HRQL or whether caregiver perception of their child's burden causes their anxiety. The FAQLQ-PF was developed for children aged 0 to 12 years; as 28 participants in this study had a child aged 13-15 years, it is unclear how this may have affected the FAQLQ-PF results. Additionally, like with many studies that collect data using online methods, this study did not have clear access to response rate data which unfortunately is a common issue. Finally, the sample is quite small, data collection was only conducted in the UK, and some of the analyses may be underpowered. Future research could extend the sample size, traceability of responses and the recruitment to other countries. The inclusion of parent-child dyads and a longitudinal design in future research could assess any potential differences between parent-proxy reports and child self-reports, and begin to unpick the relationship between caregiver anxiety and child burden in PA.
This study demonstrated the caregiver-reported burden experienced by children with PA in the UK, and to our knowledge is the first to report caregiver-reported impacts of PA on the widely used EQ-5D-Y. The results showed that male and female children experience a similar level of psychosocial burden due to their PA. For older children, increased PA severity is associated with increased burden, whereas younger children experience a consistently high level of burden unrelated to their PA severity. Caregiver anxiety and sex were significantly associated with many proxy-reported measures of child impact, with male caregivers reporting more burden experienced by their child than female caregivers, using disease-specific measures. It is clear that further research in a larger sample could explore and validate this finding. Research could also explore whether children with PA in other countries experience a similar level of burden. Caregiver anxiety was also associated with the child's PA reaction history and severity of PA, suggesting the elevated burden reported by caregivers with anxiety may be due to the child's experiences rather than their own anxiety. This issue would benefit from further research to investigate whether reducing caregiver anxiety may reduce the perceived burden of PA on the child's life.

ACK N OWLED G EM ENTS
This study was funded by Aimmune Therapeutics. The authors thank Anne Marciniak, MD, PhD, for her contributions to the study and this report. Editorial assistance was provided by The Curry Rockefeller Group, LLC, and was funded by Aimmune Therapeutics.

CO N FLI C T O F I NTE R E S T
SA, KG, JdV and RCK report consulting for Aimmune Therapeutics.
AV and RR are employees of Aimmune Therapeutics.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study captured in the current article are available from the corresponding author upon reasonable request.