Impact of Uncontrolled Symptoms on the Health-Related Quality of Life (EQ-5D-5L) of Patients With Allergic Rhinitis: A MASK-air Study

The impact of allergic rhinitis (AR) on patients' quality of life (QoL) may vary with disease control and comorbidities [1]. Previously, we quantified utilities for different levels of AR control but did not assess the impact of specific symptoms on QoL [2]. In this study, we used EQ-5D-5L to assess the impact of uncontrolled individual symptoms on the QoL of patients with AR.

A full description of the Methods is available electronically (https://doi.org/10.6084/m9.figshare.25253323.v2). We assessed data from European users of the mHealth app MASK-air [3] between May 2015 and December 2022. These users had self-reported AR, were aged between 16 (or 15 in countries with a lower age of digital consent [4]) and 74 years and had filled-in the full EQ-5D-5L questionnaire and/or the EQ-5D visual analogue scale (VAS) alone.

MASK-air comprises a daily monitoring questionnaire consisting of (i) VASs assessing the daily impact of ocular, nasal, asthma and global allergy symptoms (0–100 scale, a higher score corresponds to a higher impact of symptoms) and (ii) the EQ-5D VAS (0–100 scale, the higher the value the better the patient is feeling on that day). Additionally, users may opt to respond to the EQ-5D-5L questionnaire, which allows for the computation of utilities [5].

We computed Spearman correlation coefficients between the EQ-5D utility index score or the EQ-5D VAS and symptom VASs (VASs on eye, nose and asthma symptoms).

We then categorised each symptom VAS into ‘good’, ‘partial’ and ‘poor’ control [6]. We first studied the association between each symptom VAS and QoL by building mixed-effects linear regression models for each symptom individually. Additionally, to measure which isolated symptoms have the greatest impact on QoL (removing the effect of the remaining symptoms), we performed similar regression analyses restricted to observations with ‘good’ control of the two remaining symptom VASs (e.g., to assess the impact of poor versus good control on VAS Eye, we considered only the observations in which there was a simultaneously good control of VAS Nose and VAS Asthma). We performed this stratified analysis (instead of adjusting for the remaining symptoms in regression models) due to multicollinearity between allergy symptoms and the need to account for interactions in multivariable models (which would render regression coefficients difficult to interpret). Separate analyses were performed considering observations from all patients with AR, patients with AR only or patients with AR+asthma [7].

We analysed 4008 days (reported by 2424 users) with information on utilities and 82,737 days (reported by 7905 users) with information on the EQ-5D VAS (Table S1).

We found moderate correlations between utilities and symptom VASs (coefficients from −0.38 to −0.41) or between the EQ-5D VAS and symptom VAS (coefficients from −0.38 to −0.51). Similar results were found for patients reporting both VAS and full EQ-5D data. VAS Eye showed the weakest correlations with QoL measures (Table S2).

In multivariable analyses unadjusted for the remaining symptoms, poor control of ocular, nasal or asthma symptoms was consistently associated with lower EQ-5D utilities or VAS (Table S3).

To better assess the impact on QoL of each symptom separately, we restricted the analyses to observations in which only one of the symptom domains was uncontrolled. When considering all observations from patients with AR, poorly controlled eye and nose symptoms were associated with decreased utilities (−0.07 [95% CI = −0.09; −0.04] to −0.03 [95% CI = −0.05; −0.02] units) and EQ-5D VAS (−11.7 [95% CI = −12.0; −11.4] to −10.5 [95% CI = −11.0; −9.9]) (Table 1). A similar impact was found for patients with AR alone or AR+asthma. In patients with AR+asthma, poorly controlled asthma symptoms were associated with decreased utilities (−0.07 [95% CI = −0.10; −0.04]) and EQ-5D VAS (−13.9 [95% CI = −14.4; −13.0]) (Table 1).

In all performed analyses, poor symptom control was associated with a greater impact on QoL than partial symptom control (Tables 1 and S3). That is, there were larger differences in QoL when comparing days of good vs. poor control or even days of partial versus poor control than when comparing days of good versus partial control.

Overall, we found that poor symptom control was associated with decreased QoL. Poor control of eye and nose symptoms seems to have a similar impact on QoL in patients with AR alone and AR+asthma, underscoring the importance of controlling both ocular and nasal symptoms in patients with AR, although considering patients' preferences [8]. The largest differences in QoL concerned comparisons between poor versus other levels of control for all symptoms, particularly for the EQ-5D VAS. Importantly, the EQ-5D VAS is a quantitative measure that reflects the patients' own judgement of their health, whereas utilities are computed based on EQ-5D health states and better reflect health-related QoL [9]. Nevertheless, we observed a consistent pattern in utilities and the EQ-5D VAS, with both decreasing when AR symptom control worsened. This study's limitations include the fact that mHealth users may not be representative of the general patient population. Another limitation is that AR is characterised by variable control levels across and within the same individual. To minimise this potential effect modification of AR control, analyses were performed with stratification by symptom control levels. Finally, utilities were estimated from EQ-5D-5L health profiles, instead of the standard gamble method.

In conclusion, each poorly controlled symptom in AR was consistently associated with decreased health-related QoL. In patients with AR+asthma, asthma symptoms were those associated with the greatest decrease in QoL. Poor control of ocular and nasal symptoms was found to impact the QoL similarly for patients with AR alone or AR+asthma (suggesting that having asthma does not impact the effect of ocular and nose symptoms on QoL).

Rafael José Vieira: conceptualisation, data analysis, writing–original draft. Lucas Leemann: data analysis, writing–review and editing. Holger J. Schünemann, Luís Filipe Azevedo, João A. Fonseca and Jean Bousquet: conceptualisation, writing–review and editing. Bernardo Sousa-Pinto: conceptualisation, data analysis, writing–review and editing.

J.B. reports personal fees from Cipla, Menarini, Mylan, Novartis, Purina, Sanofi-Aventis, Teva, Uriach, other from KYomed-Innov, other from MASK-air SAS, outside the submitted work.

H.S. reports and developed guidelines on Allergic Rhinitis and its Impact on Asthma (ARIA) and his academic institution received research funding for it.

The other authors have no conflicts of interest to disclose, outside the submitted work.