Systematic Review and Meta‐Analysis of Oral Immunotherapy Effect on Food Allergy‐Related Quality of Life

ABSTRACT

Background

Uncertainty exists regarding the health‐related quality of life (HRQL) benefits of food allergen oral immunotherapy (OIT). Up‐to‐date meta‐analyses incorporating HRQL data from recent randomised trials are lacking.

Methods

Systematic searches of MEDLINE, Embase, CENTRAL and Google Scholar were conducted for food OIT randomised trials (versus any comparator) that measured HRQL with a validated instrument (27 July 2023). Hedges' g standardised mean differences in HRQL between OIT and comparator were analysed by allergen, reporting perspective, and treatment/post‐treatment periods, and synthesised using random‐effects meta‐analysis models when results from multiple trials were available.

Results

Ten trials (nine peanut, one baked milk; N = 1330) were included. No between‐group differences were observed in parent‐reported child HRQL (g = −0.07, 95% CI: −0.19 to 0.04, 9 trials, N = 1259), self‐reported child HRQL (g = −0.23, 95% CI: −0.73 to 0.27, 5 trials, N = 435), or self‐reported teenager HRQL (g = 0.00, 95% CI: −0.41 to 0.40, 3 trials, N = 209) during peanut OIT treatment. However, 12 months post‐treatment, improved parent‐reported child HRQL was observed (g = −0.51, 95% CI: −0.84 to −0.19, 3 comparisons from 2 trials, N = 213). No between‐group differences were observed in the baked milk OIT trial in children (g = 0.118, 95% CI: −0.63 to 0.87, N = 26).

Conclusion

HRQL benefits were observed after OIT protocol completion, with limited evidence of benefit during active treatment.

1. Introduction

Food allergy is a major public health concern. It affects individuals of all ages and causes significant burdens for patients and their families [1, 2]. Food allergy management currently centres on allergen avoidance practices that contribute to increased patient anxiety and poor health‐related quality of life (HRQL) [3, 4]. Patients have indicated the outcome they desire most from food oral immunotherapy (OIT) treatment is an improved HRQL, which they have prioritised higher than a reduction in reactions or receiving a cure for their food allergy [5]. However, conflicting findings exist about whether OIT achieves a meaningful improvement to HRQL compared to placebo [6, 7, 8].

OIT is an active treatment for food allergy that involves exposure to progressively increasing doses of a target allergen to increase a patient's reaction threshold (an outcome of desensitisation), and in some cases, to redirect the underlying allergic response to achieve clinical remission (sustained unresponsiveness; SU) [9, 10]. Desensitisation reduces the likelihood and severity of allergic reactions upon accidental exposure while patients remain on maintenance dosing, while SU permits discontinuation of OIT and free consumption of the culprit allergen, eliminating reliance on allergen avoidance [11]. An OIT product (Palforzia for peanut allergy) was approved in the United States of America, United Kingdom, and Europe for desensitisation against small amounts of peanut (2–4 peanuts). However, uptake of this treatment has been low [12].

HRQL is a multidimensional construct that considers a patient's wellbeing across all aspects of their life as impacted by their health condition, and is most commonly assessed using disease‐specific or generic patient‐reported outcome measures (PROMs) [13]. The food allergy‐specific Food Allergy Quality of Life Questionnaire (FAQLQ) and generic Pediatric Quality of Life Inventory (PedsQL) are the most popular instruments used for measurement of food allergy‐related HRQL [14, 15]. Both instruments measure HRQL across physical, emotional, and social health domains that can be combined to quantify total HRQL, or considered separately to generate subscale scores for each domain. While the PedsQL measures HRQL on a scale of 0–100 points, the FAQLQ is instead inversely scaled and measures food allergy burden on a 7‐point scale with an established minimum clinically important difference of 0.45–0.50 points [15, 16]. Given food allergy often arises in childhood and can persist into adulthood, age‐appropriate versions of the FAQLQ and PedsQL are available to capture patient HRQL across different stages of the life course. For children with food allergy, HRQL is often captured via parent‐proxy report, with self‐report PROMs existing for older children, teenagers, and adults [15, 17, 18, 19].

Previous reviews that synthesised the effect of OIT on HRQL were limited by the availability of evidence at the time they were conducted and had conflicting conclusions about OIT impact [6, 7, 8]. While the availability of PROMs with different reporting perspectives is necessary to account for differences in age and reporting abilities, previous synthesis of HRQL results across multiple reporting perspectives and allergens may have contributed to conflicting opinions about the effect of OIT on HRQL due to the risk of PROM misalignment [6, 20, 21]. Additionally, no synthesis has yet accounted for potential differences in post‐treatment HRQL benefits due to the clinical outcome achieved through treatment (desensitisation, SU or persistent allergy) or due to potential influences of patient and sociodemographic characteristics, despite evidence that these can modify food allergy development and wellbeing [3, 22, 23]. An updated review and meta‐analysis of the effect of OIT on HRQL that accounts for different reporting perspectives, allergens, active treatment phase versus post‐treatment periods, and patient characteristics is crucial for informing decision‐making about more widespread OIT approval and use.

We sought to answer three research questions in this review: (1) whether OIT improved total HRQL, as captured with a validated instrument, in people with IgE‐mediated food allergy during OIT treatment protocols (active treatment) or after treatment protocol completion (referred to from this point as post‐treatment follow‐up) compared to those who did not receive OIT, (2) whether study authors identified any associations between OIT and HRQL subdomains and (3) whether study authors identified any subgroup effects, where the effect of OIT on HRQL was different between clinical, treatment‐related, demographic and/or social characteristics.

2. Methods

This review was reported consistent with PRISMA guidelines [24] and the protocol was registered on PROSPERO (#CRD42023443175).

2.1. Search Strategy and Eligibility Criteria

A systematic search of MEDLINE, Embase and CENTRAL (via the Ovid platform, inclusive of all records published from database inception to the date of search) and of Google Scholar (limited to the first 100 results due to preferential listing of search results) was conducted on 27th July 2023. The search strategy was developed in consultation with a biomedical librarian and included terms for ‘oral immunotherapy’, ‘food allergy’ and ‘quality of life’ (Table S1).

This review was limited to randomised trials that provided original data on the effect of OIT versus any comparator (including allergen avoidance) on HRQL in participants with food allergy. Other forms of interventional studies (such as single‐arm trials) and observational studies were excluded because we intended to assess a treatment effect and these study types present a risk of confounding bias. Grey literature, opinion pieces, conference abstracts, protocols and other systematic reviews were also excluded. Studies had to include a HRQL total and/or subscale score measured with a validated HRQL instrument. Trials investigating HRQL in non‐clinical populations or other allergic conditions (e.g., non‐IgE mediated food allergy, asthma, rhinitis, intolerances, food‐exercise induced anaphylaxis, coeliac disease) were excluded.

Covidence software (Veritas Health Innovation, Melbourne, Australia) was used for de‐duplication, title and abstract screening, and full text screening. All studies identified by the search strategy were independently reviewed against inclusion and exclusion criteria by two researchers (SAR, PT, ML and/or AMW). Any discrepancies during study selection were resolved through discussion with a third senior reviewer (RLP or MLKT).

2.2. Data Extraction

A bespoke data collection form was piloted and used to extract details related to study procedures, OIT and control interventions, HRQL instruments, HRQL scores with corresponding measures of uncertainty, and any subgroups assessed. Data extraction was independently performed by one author (SR) and reviewed by a second (PT). In the case where HRQL summary data relevant to answering our primary research question could not be extracted from published trial reports, trial authors were contacted to request missing information. Only missing aggregate HRQL results were sought from study authors; no participant‐level data was collected. Data extraction for all included studies was completed in October 2024.

2.3. Risk of Bias Assessment

The quality of included trials was assessed using the Risk of Bias (RoB) 2.0 tool [25], inclusive of all domains: randomisation procedures (D1), intervention assignment (D2), missing outcome data (D3), outcome measurement (D4), selective reporting (D5), and carryover effects relevant to cross over trials (DS). Risk of bias assessments were independently performed by two reviewers, and any disagreements were discussed with a third reviewer to reach consensus. All RoB 2.0 domains were assessed per trial, with additional assessment of D3–D5 (inclusive) per result for inclusion in synthesis in the cases where multiple HRQL report perspectives (parent proxy‐reports, and child, teen or adult self‐reports) were assessed in a study.

2.4. Data Synthesis and Analysis

Synthesis was performed separately for treatment and post‐treatment follow‐up periods, different reporting perspectives (parent proxy‐reports, and child, teen or adult self‐reports), and different target allergens.

For answering the first research question of interest, Hedges' g standardised mean differences comparing change‐from‐baseline HRQL total scores between OIT and control groups were calculated. Due to the inverse nature of the FAQLQ instruments, which majority of the included trials used to measure HRQL, all synthesised estimates presented were also on an inverse scale; higher scores represent poorer HRQL. Any HRQL scores derived from other instruments (such as the PedsQL) were inverted before Hedges' g estimation to ensure consistent reporting of direction of effect. When three or more comparisons were available per timeframe, reporting perspective and allergen, the Hedges' g estimates were synthesised using random‐effects meta‐analysis models with restricted maximum likelihood to estimate the between‐study variance (${\tau }^2$). These methods were selected to account for use of different HRQL instruments between trials and potentially small sample sizes in the trials [26]. Between‐trial heterogeneity was assessed using the heterogeneity standard deviation (τ), and I ² statistic. In the case where studies reported median (IQR) change‐from‐baseline HRQL scores per treatment arm, we estimated the mean (standard deviation; SD) change‐from‐baseline HRQL using methods by Wan et al., 2014 [27] prior to Hedges' g estimation or synthesis. Cochrane (version 6.5) recommended methods were used for all other transformations needed prior to meta‐analysis, including standard deviation estimation from reported means and 95% confidence intervals, combination of age‐stratified HRQL results into a single effect size estimate, and halving of the comparator group sample size to create two two‐way comparisons when including results from three‐arm trials in synthesis (this was done for the PPOIT‐001 [28, 29] and PISCES [30] trials, which both had two OIT intervention arms, with and without an adjuvant product, as well as a comparator arm) [26]. All analysis was conducted using Stata 18 (StataCorp LLC, TX). In cases where less than three comparisons were available per timeframe, reporting perspective or allergen, we reported the individual study‐level effects.

To address the second and third research questions, respectively, any subdomain HRQL scores or subgroup effects (i.e., any clinical, demographic and/or social characteristics that could have modified the effect of OIT on HRQL) reported in included trials' publications were presented as reported. Where multiple studies investigated the same subscale score or subgroup, these findings were compared in a narrative synthesis.

2.5. Sensitivity Analyses

Meta‐analyses were stratified by risk of bias assessments, comparator types, blinding statuses, OIT maintenance doses, and times of HRQL measurement to confirm design differences between included studies did not influence results. Leave‐one‐out analysis was also conducted to confirm results from a single study did not drive aggregate findings.

2.6. Ethical Considerations

Only de‐identified aggregate results collected from previously published studies were used in this review; oversight from an ethics review panel was not required.

3. Results

3.1. Included Trials

The search strategy identified 1992 records across the four databases. After removal of duplicate citations, the titles and abstracts of 1493 records were screened, the full texts of 59 records were reviewed, and 11 publications of 10 OIT trials were identified for inclusion (Figure 1).

FIGURE 1.

PRISMA flow diagram of study selection process.

Of the included trials, nine investigated the effects of peanut OIT in paediatric populations [28, 29, 30, 31, 32, 33, 34, 35, 36, 37], and one investigated baked milk OIT in children, teenagers, and adults [38] (Table 1). A wide range of target maintenance doses were used across included trials (125–5000 mg) and most compared OIT to a placebo control, with two of the peanut OIT trials including an allergen avoidance comparator group [35, 36, 37]. All trials measured child HRQL by parent‐proxy report using the FAQLQ parent form (FAQLQ‐PF) or, in one case, the PedsQL 4.0 parent‐proxy report [37]. Participant HRQL was additionally measured by self‐report in children in six trials [31, 32, 33, 37, 38] using the FAQLQ child form (FAQLQ‐CF), and in teenagers in four trials [32, 38] using the FAQLQ teen form (FAQLQ‐TF). Seven of the included trials [30, 31, 32, 33, 34, 38] reported HRQL subscale scores as well as total scores.

TABLE 1.

Characteristics of included studies.

Trial and publication year	Study design	Allergen	Participant ages at baseline (range)	Treatment(s)	Control	Maintenance dose	OIT clinical endpoint	PROM used and participant age range applied to
ARTEMIS, 2020/2022 [31, 32]	RCT (double‐blind)	Peanut	4–17 years	Peanut OIT	Placebo	300 mg	Desensitisation to cumulative dose of 2043 mg peanut protein	FAQLQ‐PF (4–17 years) FAQLQ‐CF (8–12 years) FAQLQ‐TF (13–17 years)
Blumchen et al., 2019 [33]	RCT (double‐blind)	Peanut	3–17 years	Peanut OIT	Placebo	125–250 mg	Desensitisation to a maximum dose of 4500 mg peanut protein	FAQLQ‐PF (3–12 years) FAQLQ‐CF (8–12 years)
Dantzer et al., 2022 [38]	RCT (double‐blind)	Milk	3–18 years	Baked‐milk OIT	Placebo	2000 mg	Desensitisation to cumulative maximum dose of 4044 mg	FAQLQ‐PF (3–18 years) FAQLQ‐CF (8–12 years) FAQLQ‐TF (13–17 years) FAQLQ‐AF (18+ years)
PALISADE, 2022 [32]	RCT (double‐blind)	Peanut	4–17 years	Peanut OIT	Placebo	300 mg	Desensitisation to cumulative dose of at least 1043 mg peanut protein	FAQLQ‐PF (4–17 years) FAQLQ‐CF (8–12 years) FAQLQ‐TF (13–17 years)
PISCES, 2022 [30]	Three‐arm RCT (double‐blind)	Peanut	5–10 years	Peanut OIT with adjuvant antihistamines (desloratadine and ranitidine) (OIT + AH) Peanut OIT with placebo antihistamine	Placebo	500 mg	Desensitisation to a cumulative maximum dose of 4000 mg peanut protein	FAQLQ‐PF (5–10 years)
PPOIT‐001, 2018/2021 [28, 29]	RCT (double‐blind)	Peanut	1–10 years	Peanut OIT with probiotic adjuvant (L. rhamnoses) (PPOIT)	Placebo	2000 mg	SU to two cumulative maximum doses of 4000 mg, 2–6 weeks apart.	FAQLQ‐PF (1–10 years)
PPOIT‐003, 2022 [34]	Three‐arm RCT (double‐blind)	Peanut	1–10 years	Peanut OIT with probiotic adjuvant (L. rhamnoses) (PPOIT) Peanut OIT with placebo probiotic	Placebo	2000 mg	SU to two cumulative maximum doses of 4950 mg, 8 weeks apart	FAQLQ‐PF (1–10 years)
RAMSES, 2022 [32]	RCT (double‐blind)	Peanut	4–17 years	Peanut OIT	Placebo	Safety trial; no specified maintenance dose	Safety trial; no clinical efficacy assessment	FAQLQ‐PF (4–17 years) FAQLQ‐CF (8–12 years) FAQLQ‐TF (13–17 years)
STOP II, 2014 [35, 36]	Crossover RCT (unblinded)	Peanut	7–16 years	Peanut OIT	Allergen avoidance	800 mg	Desensitised to a cumulative maximum dose of 1400 mg peanut protein	FAQLQ‐PF (7–12 years)
TAKE‐AWAY, 2019 [37]	RCT (unblinded)	Peanut	5–15 years	Peanut OIT	Allergen avoidance	5000 mg	Desensitised to cumulative maximum dose of 7500 mg peanut protein	PedsQL 4.0 parent proxy‐report, restricted to Psychosocial Functioning domains (5–15 years) PedsQL 4.0 child self‐reports, restricted to Psychosocial Functioning domains (5–15 years)

Abbreviations: AH, antihistamine adjuvant (desloratadine and ranitidine); FAQLQ‐AF, Food Allergy Quality of Life Questionnaire Adult Form; FAQLQ‐CF, Food Allergy Quality of Life Questionnaire Child Form; FAQLQ‐PF, Food Allergy Quality of Life Questionnaire Parent Form; FAQLQ‐TF, Food Allergy Quality of Life Questionnaire Teen Form; OIT, oral immunotherapy; PedsQL, Paediatric Quality of life Inventory; PPOIT, Probiotic and peanut OIT; PROM, Patient Reported Outcome Measure; RCT, randomised controlled trial; SU, remission outcome (sustained unresponsiveness).

The designs of included trials were heterogeneous (Figure 2). All trials measured HRQL at the beginning and end of treatment, and most HRQL measurements were conducted at the time of allergic status assessment via oral food challenge. One trial conducted HRQL measurement within 1 month of the exit food challenge [33] and one trial intended to assess OIT safety and did not include food challenges for efficacy assessment [32]. Two trials measured HRQL during post‐treatment follow‐up: one at 12 months post‐treatment [34], and one at 3 months, 12 months, and 4 years post‐treatment (for SU) [28, 29]. These two trials included an OIT discontinuation period within their protocols to assess OIT effect on SU outcomes, while all other treatment efficacy trials assessed OIT in terms of desensitisation through an exit food challenge. HRQL was assessed as a secondary or exploratory outcome in all included trials.

FIGURE 2.

Overview of included studies showing treatment procedures, times of HRQL measurement, and assessments of clinical allergy status. AH, antihistamine adjuvant; BMOIT, baked milk oral immunotherapy; DBPCFC, double‐blind placebo‐controlled food challenge; HRQL, health‐related quality of life; OFC, oral food challenge; OIT, oral immunotherapy; PPOIT, probiotic and peanut oral immunotherapy. *Exact timeframes needed for up‐dosing and maintenance are patient‐dependent. Here we present the expected period of these, or the minimum maintenance requirement where reported, in each included trial. **2–6 weeks elimination used. Here we present the minimum length of elimination considered for assessment of sustained unresponsiveness. ^#Safety trial; did not conduct OFCs. ^{^}Trial protocol includes an intended 36 months of maintenance (total of 48 months on treatment) for which HRQL has not been observed or published. Here we present the period of maintenance in which HRQL was observed.

3.2. Risk of Bias in Included Trials

Three of the included trials [28, 29, 33, 34] received a ‘low’ risk of bias rating, while six [30, 31, 32, 35, 36] received a ‘some’ risk of bias rating due to missing HRQL data among randomised participants (Figure 3). Missingness in HRQL data was extensive and specific to the OIT arm in the one trial considered to be at ‘high’ risk of bias [37]. Bias was also suspected in some trials due to outcome measurement in unblinded participants and potentially selective reporting of HRQL results.

FIGURE 3.

Risk of bias of included studies' HRQL results. Risk of bias was assessed with the Rob2.0 in relation to the randomisation process (D1), deviations from intended interventions (D2), missing outcome data (D3), outcome measurement (D4), selection of reported results (D5), and crossover procedures if relevant (DS). Risk of bias assessment relates only to the HRQL outcomes investigated in this review and does not necessarily reflect the quality of trials in measuring other intended outcomes.

3.3. Effects of Peanut OIT on HRQL During Treatment

In the trials evaluating peanut OIT, HRQL change between baseline and the end of treatment did not differ greatly between OIT and control groups for any of the three reporting perspectives considered (Figure 4). Trials that investigated child HRQL by parent proxy‐report observed little difference in HRQL during treatment between OIT and control groups, resulting in an overall Hedges' g standardised mean difference of −0.07 (95% CI: −0.19 to 0.04, p = 0.22, 11 trials) with no heterogeneity ($\tau$ = 0.00, I ² = 0.00%). Little HRQL benefit from OIT compared to control was seen in the meta‐analysis of child self‐reports, with uncertainty and high heterogeneity (g = −0.23, 95% CI: −0.73 to 0.27, p = 0.37, 5 trials, $\tau$ = 0.50, I ² = 80.90%). No difference between groups was observed in teenager self‐reported HRQL during treatment, although with high uncertainty (g = 0.00, 95% CI: −0.41 to 0.40, p = 0.99, 3 trials, $\tau$ = 0.24, I ² = 43.54%). No differences were seen between OIT and control groups in subscale HRQL during treatment, regardless of perspective (see Table S2).

FIGURE 4.

Meta‐analyses of trials to investigate the effect of OIT versus control on HRQL between baseline and end of treatment by reporting perspective. Synthesised result is represented by the green diamond. HRQL is inversely scored; more negative scores indicate greater improvement in HRQL. PPOIT‐003 and PISCES were three‐arm trials (2 active arms, 1 comparator arm) and results of these trials are presented as two‐way comparisons labelled with the active arm being compared to control (please refer to methods for details regarding handling of the comparator arm). (A) Parent‐proxy reported child HRQL, (B) Child self‐reported HRQL, (C) Teenager self‐reported HRQL. AH, adjuvant antihistamines; OIT, oral immunotherapy; PPOIT, peanut OIT with probiotic adjuvant; REML, random‐effects maximum likelihood.

3.4. Effects of Peanut OIT on HRQL Post‐Treatment

The peanut OIT trials that completed post‐treatment follow‐up [28, 29, 34] measured child HRQL by parent proxy‐report at 3‐months, 12‐months, and 4‐years post‐treatment. Between baseline and 3‐months post‐treatment, the PPOIT‐001 trial reported little HRQL improvement in the OIT arm (with probiotic adjuvant) (N = 24) compared to placebo arm (N = 27) with uncertainty (g = −0.23, 95% CI: −0.78 to 0.31). In contrast, meta‐analysis of 12‐month post‐treatment results from the PPOIT‐001 and PPOIT‐003 trials (Figure 5) showed a moderate benefit of OIT on HRQL compared to placebo control (g = −0.51, 95% CI: −0.84 to −0.19, 3 comparisons from 2 trials) with no heterogeneity ($\tau$ = 0.00, I ² = 0.00%). Clinically important (≥ 0.45 points for the FAQLQ instrument) improvements in 12‐month post‐treatment HRQL subscale scores related to food anxiety and social and dietary limitations were also observed in those who received peanut OIT (see Table S2). Furthermore, the PPOIT‐001 trial reported a very strong improvement in child HRQL by parent proxy‐report between baseline and 4‐years post‐treatment for those who had received peanut OIT (N = 19) compared to those who had received placebo (N = 19) (g = −0.87, 95% CI: −1.52 to −0.22).

FIGURE 5.

Meta‐analyses of studies to investigate the effect of OIT versus control on parent proxy‐reported child HRQL between baseline and 12‐months post‐treatment. Synthesised result is represented by the green diamond. HRQL is inversely scored; more negative scores indicate greater improvement in HRQL. PPOIT‐003 was a three‐arm trial (2 active arms, 1 comparator arm) and results are presented as two‐way comparisons labelled with the active arm being compared to control (please refer to methods for details regarding handling of the comparator arm). OIT, oral immunotherapy; PPOIT, peanut OIT with probiotic adjuvant; REML, random‐effects maximum likelihood.

3.5. Effects of Baked Milk OIT on HRQL

One trial [38] investigated baked milk OIT compared to placebo during active treatment only, and did not consider the post‐treatment period (Table S3). No difference in child HRQL by parent proxy‐report was observed between OIT (N = 12) and placebo (N = 14) arms (g = 0.118, 95% CI: −0.63 to 0.87). A potential worsening of child self‐reported HRQL occurred when comparing OIT (N = 5) to placebo (N = 5), though with uncertainty and limited sample size (g = 1.179, 95% CI: −0.05 to 2.41). While patients of any age were eligible for this baked milk OIT trial, insufficient teenage (N = 6) and adult participants (N = 2) were included to compare HRQL between OIT and placebo groups for these age groups.

3.6. Subgroup Effects Explored in Included Trials

Modification of OIT effect on HRQL by clinical, demographic, and/or social characteristics was explored through subgroup analyses in trials of peanut OIT only. Age (in five trials [28, 29, 31, 32, 37]) and treatment outcome (in two trials [28, 29, 34]) were the only characteristics to be examined in terms of their association with treatment impact on HRQL across multiple trials. There was limited evidence that treatment effect on HRQL differed by age during treatment, although one trial reported a clinically meaningful difference in HRQL change from baseline to 3 months post‐treatment between children aged ≥ 5 years compared to ≤ 5 years old [29]. In contrast, both trials that explored the impact of treatment outcome reported participants that achieved SU had improved HRQL compared to participants that remained allergic (Table 2). Complete details of other subgroups explored across included studies can be found in Table S4.

TABLE 2.

Findings of trials to report the effect of OIT on HRQL by treatment outcome.

Study	Timeframe	Treatment outcome	Mean change in total FAQLQ‐PF a score, p		Major findings
PPOIT‐001	Between baseline to 3‐months post‐treatment	SU	PPOIT participants (N = 23): −1.3, p = 0.01		PPOIT participants with SU had clinically and statistically significantly improved parent proxy‐reported HRQL at both 3‐months and 12‐months post‐treatment compared to PPOIT participants without SU and placebo participants
		Allergic	PPOIT participants (N = 5): 0.58, p = 0.4 Placebo participants (N = 27): −0.3 (SD = 1.1), p = 0.4
	Between baseline to 12‐months post‐treatment	SU	PPOIT participants (N = 23): −1.8, p = 0.001
		Allergic	PPOIT participants (N = 5): 0.2, p = 0.8 Placebo participants (N = 27): 0.15 (SD = 1.8), p = 0.7
PPOIT‐003	Between baseline to 12‐months post‐treatment	SU	−0.74 (SD = 1.25)	SU vs. desensitised: p = 0.014 SU vs. Allergic: p = < 0.0001	Only participants with SU had a clinically significant improvement in parent proxy‐reported HRQL, while desensitised and allergic participants did not. However, treatment outcome was not reported by treatment allocation
		Desensitised	−0.31 (SD = 1.38)
		Allergic	−0.21 (SD = 1.32)

Abbreviations: FAQLQ‐PF, Food Allergy Quality of Life Questionnaire parent form; HRQL, health‐related quality of life; PPOIT, peanut OIT with probiotic adjuvant; SU, remission outcome (sustained unresponsiveness).

^a The FAQLQ‐PF is inversely scored on a scale of 0 to 6 points; more negative scores indicate greater improvement in HRQL.

3.7. Sensitivity Analyses

Trial design characteristics examined in sensitivity analyses, including risk of bias, comparator types, blinding statuses, OIT maintenance doses, and times of HRQL measurement, did not influence meta‐analysis results, and leave‐one‐out analysis showed similar synthesised findings when each trial was removed from synthesis (Figures S1–S5).

4. Discussion

This systematic review and meta‐analysis comprises the largest and most up‐to‐date synthesis of HRQL data from randomised trials of OIT versus control that accounted for multiple reporting perspectives, time periods of follow‐up, and the role of treatment outcome and patient characteristics in modifying HRQL. We found that peanut OIT increased HRQL benefits in the longer‐term post‐treatment, after treatment outcomes (SU, desensitisation, allergic) had been realised, with 12‐month post‐treatment meta‐analysis showing a moderately strong benefit of OIT versus control (g = −0.51, 95% CI: −0.84 to −0.19, 2 studies, 3 comparisons). In contrast, there was no difference in HRQL identified between OIT and control groups during active treatment across reporting perspectives. This suggests the benefit of OIT arose after patients became aware of their post‐treatment peanut allergy status by study exit food challenge.

The high rates of desensitisation induced by currently available OIT (Palforzia), and SU with emerging OITs, are thought to produce lifestyle benefits for patients and families by reducing the need for stringent allergen management practices [12, 34, 39]. Given that the burden of allergen avoidance and fear of accidental reactions are the major contributors to poor HRQL in food allergy [3, 4, 40], it is reasonable that improvements would arise post‐treatment when the benefits of changes to allergy management practices can be realised. The impact of avoidance practices on HRQL is further supported by the observation that children with an SU outcome following peanut OIT experienced greater HRQL improvement compared to desensitised or allergic participants in the PPOIT‐001 and PPOIT‐003 trials, with those in SU consuming their former allergen ad libitum [28, 29, 34]. However, further investigation of the impact of different treatment outcomes and their contribution to HRQL is needed to confirm our observations in peanut allergy and in other foods. Additionally, the observation that OIT did not lead to an HRQL benefit until the post‐treatment period, after clinical outcomes had been assessed, presents a major implication for patient‐provider decision‐making about the use of OIT. Our findings highlight the need to emphasise that HRQL benefits of OIT are potentially delayed or more greatly associated with a clinical outcome of SU when discussing OIT in clinical practice. This will help avoid frustration and non‐adherence to treatment during lengthy OIT protocols. Further long‐term evaluation of OIT is needed to confirm this finding.

Our finding that HRQL did not improve during the treatment period is somewhat consistent with that of a prior meta‐analysis [7]. Though we were able to provide far greater clarity as to the effect of OIT across reporting perspectives (child self‐reports and parent proxy‐reports), between time periods (active treatment versus post‐treatment) and between clinical endpoints (SU, desensitisation, allergic) due to the availability of recently reported trial data. The availability of new trial data for inclusion could help explain the variation in results obtained in this review compared to previous syntheses of the effects of OIT on HRQL [6, 7, 8].

This systematic review confirmed that studies of OIT effects in adults and across a broader range of allergens (other than peanut) are lacking [10, 41]. Cross‐sectional analyses have observed very poor HRQL and high psychosocial burden in adults with food allergy, and there is a pressing need for treatment options suited to older age groups [42, 43]. A key limitation to the generalisability of our findings is that they relate largely to children and teenagers with peanut allergy. Our finding of slightly different meta‐analysis results between child self‐reported and proxy‐reported HRQL may additionally support the idea of PROM misalignment [20]. However, it should be acknowledged that these analyses included participants of different ages, with the parent proxy‐report meta‐analysis including those between 1–18 years, and the child self‐report analysis being limited to those between 8 and 15 years. Age is a potential modifier of OIT outcomes [44, 45], and so we could reasonably expect differences in results to have been influenced by both age and perspective. Additionally, while the PROMs included in this review are considered the most appropriate for measuring food allergy‐specific HRQL [46, 47, 48], they were developed to assess the HRQL of untreated allergy and could be limited in their specificity to the treatment setting. Further investigation into the inter‐rater reliability of PROMs for capturing HRQL in patients with food allergy, and into the development of clear guidelines for how to best measure and compare the HRQL of patients of different ages, and under treatment, are needed to enhance our understanding.

This review was limited by the heterogeneous designs of included trials and limited studies of long‐term outcomes, with only two trials in Australia having evaluated longer‐term post‐treatment HRQL. While subgroup and sensitivity analyses did not reveal that any meaningful bias arose due to differences in study design or patient sample characteristics (other than treatment outcome), this did mean only two trials could be included in the assessment of post‐treatment results and the meta‐analysis was limited by a small sample size. Much care should be taken when extrapolating findings to other countries and the wider food allergy population, especially teenagers and adults who were not represented among those recruited into the long‐term trials synthesised in this review. A larger number of studies and patients are needed to further confirm the presence of a long‐term benefit of OIT on HRQL and improve the generalisability of our findings. A lack of long‐term outcome data following OIT has been identified as a major limiting factor for its approval and uptake, and there is critical need for more long‐term investigations of OIT effect on HRQL [11, 49]. This is crucial for informing our understanding of when, and for which patients, OIT is best suited.

To mitigate potential sources of bias that could arise in a review such as this, we limited our inclusion criteria to randomised controlled trials to enable exchangeability between OIT and control groups and separated our meta‐analyses by similar trial periods (during treatment vs. post‐treatment) to mitigate confounding by time. We also observed low to some risk of bias across all but one of the included trials, which improves the reliability of our synthesised results. This is reassuring given HRQL was measured as a secondary or exploratory outcome in all included trials and these are often at risk of greater bias due to selective reporting [26]. The reliability of our results was also improved through the use of random‐effects meta‐analysis with restricted maximum likelihood and Hedges' g correction, which are techniques appropriate for the synthesis of a moderate number of studies, and for improving the accuracy of estimates when sample sizes are small or measured using different instruments, respectively [26].

While we chose to limit this systematic review to randomised controlled trials to better establish the presence of a relationship between OIT and HRQL in a lower confounding‐bias setting, an important area of future research could be to consider the HRQL implications of OIT beyond the RCT context. Multiple longitudinal studies that did not meet our inclusion criteria have also observed HRQL improvements following OIT [39, 50, 51]. Notably, a long‐term follow‐up study of participants who received peanut immunotherapy (OIT or sublingual immunotherapy) reported observing more improvement in parent‐reported HRQL if a child achieved SU and was eating peanut at follow‐up [50]. This complements our finding that post‐treatment clinical outcomes may be key drivers of long‐term HRQL. Future synthesis of studies beyond the RCT context would help to improve the generalisability of findings to the wider food allergy population.

5. Conclusion

The findings of this systematic review and meta‐analysis suggest the positive impact of OIT on HRQL occurs post‐treatment, with clinical outcomes (SU or desensitisation) to be likely drivers. Negligible evidence of any HRQL difference between OIT and control groups during treatment was observed. This suggests that the benefits of OIT are experienced after strict treatment protocols have been completed and changes can be made to allergy management practices. Our results are of notable relevance to clinicians, researchers, and individuals with food allergy, as the relationship between clinical outcome achieved and HRQL benefit should be considered when developing, implementing, or undertaking OIT protocols. Greater investigation of HRQL benefits associated with the different clinical outcomes of desensitisation and SU, as well as across a broader range of ages and food allergens, is needed to inform decision‐making about widespread OIT approval and use.

Author Contributions

S.A.R., M.L., R.L.P., R.W. and M.L.K.T. conceived this systematic review and designed the methodology with support from C.J.H., J.J.K. and A.K. on behalf of the National Allergy Centre of Excellence (NACE). S.A.R., M.L., P.T. and A.M.W. contributed to dual‐reviewer screening of titles, abstracts and full texts. S.A.R. conducted the data extraction, which was reviewed by P.T. S.A.R. conducted the analysis and drafted the preliminary version of the manuscript. Critical appraisal of included studies using the ROB 2.0 was performed by S.A.R. and P.T. All authors assisted with interpretation of the results and reviewed and critically appraised the manuscript before submission.

Conflicts of Interest

M.L.K.T. declares consultant fees from Pfizer and CSL Seqirus; holds shares/options in Prota Therapeutics; is a member of the Medical Advisory Board of Allergy & Anaphylaxis Australia; is a member of the Board of Directors of Asia Pacific Association of Allergy Asthma and Clinical Immunology, AllergyPal and Prota Therapeutics; is a member of expert committees of the American Academy of Allergy, Asthma & Immunology, Asia Pacific Association of Allergy Asthma and Clinical Immunology, Australasian Society of Clinical Immunology and Allergy, and World Allergy Organisation. J.J.K. and R.L.P. have received a prize from the Stallergenes Greer Foundation paid to their institution outside the submitted work and R.L.P. received speaker fees from Stallergenes Greer, outside the submitted work. M.L. has received consulting fees from the National Allergy Centre of Excellence Australia. M.L.K.T. and M.L. were investigators on the PPOIT‐001 and/or PPOIT‐003 randomised trials that were included in this review, with contributions to the conduct, analysis and publication of findings. As such, these authors were not included in the process of determining inclusion/exclusion, data extraction, or risk of bias assessments in relation to any publications related to these trials. The other authors declare no relevant conflicts of interest for this work.

Supporting information

Table S1: Search strategy used in the systematic review.

Table S2: Subscale HRQL results from included trials.

Table S3: HRQL findings from trial of baked milk OIT (Dantzer et al., 2022) during treatment by report perspective.

Table S4: Subgroup HRQL results to be explored in included studies.

Figure S1: Sensitivity analysis of peanut OIT effect on HRQL during treatment by risk of bias.

Figure S2: Sensitivity analysis of peanut OIT effect on HRQL during treatment by blinding.

Figure S3: Sensitivity analysis of peanut OIT effect on HRQL during treatment by OIT.

Figure S4: Sensitivity analysis of peanut OIT effect on HRQL during treatment by time of HRQL.

Figure S5: Leave‐one‐out sensit vity analysis for the effect of peanut OIT on HRQL.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.