Quality of Life in Hymenoptera Venom Allergy: Minimal Clinically Important Difference for the Vespid Quality of Life Questionnaire

Abstract

Hymenoptera venom allergy (HVA) poses significant health concerns with implications for Health-Related Quality of Life (HRQoL) [1]. However, the Vespid Quality of Life Questionnaire (VQLQ), already validated in several languages to measure HRQoL in HVA [2, 3], lacks a formal minimal clinically important difference (MCID), limiting a comprehensive assessment of HRQoL. In fact, the commonly used 0.5 cut-off [4] is extrapolated from questionnaires with different characteristics that are used in different fields [5].

The objective of this multicenter, prospective, longitudinal, cohort study was to calculate a formal MCID for the VQLQ and to use it for a deeper multi-endpoint investigation of HRQoL, through the implementation of the MCID-event frequency and the time-to-MCID endpoints, in addition to the standard VQLQ score.

Seven Italian HVA-specialised allergy centers consecutively enrolled adult patients with HVA to Yellow Jacket, eligible to venom immunotherapy (VIT), and they were stratified into the following groups: pre-VIT, on-VIT (for <1 year, ≥1 to <3 years, ≥3 to ≤5 years) and post-VIT. The VQLQ, the Expectation of Outcome questionnaire (EoO), the Psychological General Well-Being Index (PGWBI) and the Global Rating Scale (GRS) were administered at baseline and after 1 year of follow-up. The correlation between VQLQ and the other questionnaires (Pearson's r correlation coefficient) was preliminary performed to test the external validity of VQLQ and to find the best anchor for the MCID analysis. The MCID was calculated with an anchor-based approach using the area under the curve (AUC) of the receiver operating curves (ROC) analysis to find the best VQLQ score difference, after follow-up, in detecting the MCID (whole cohort method) [6]. The global rating of change was measured using EoO, PGWBI and GRS, considering an improvement of at least +1 score, after 12 months of follow-up, as a standard reference for the ROC analysis. The best MCID was chosen according to the Youden method. Two additional sensitivity analyses for MCID calculation were performed, using anchor-based and distribution-based methods (additional information about study methods is available in the following repository: https://doi.org/10.5281/zenodo.11623493). A sample size of 140 patients was estimated to detect a 0.5 effect size (i.e. MCID used in previous studies [4]) of the mean VQLQ among the pre-planned VIT groups, in the cross-sectional analysis (one-way ANOVA, alpha = 5%, power = 80%). At least 33 patients were to be in the pre-VIT group, in order to obtain an effect size of 0.5 after 12 months of VIT, in the longitudinal analysis (paired t-test, alpha = 5%, power = 80%); the other patients were planned to be equally distributed over the other VIT groups. All the statistical analyses were performed with STATA v.18 (StataCorp LLC, Texas, USA). The study was approved by the Ethics Committees of the local sites, and written informed consents were obtained from patients.

A total of 145 patients were consecutively enrolled, with a mean age of 56.6 ± 13.1 years and 69.7% of males. None of the 34 re-stung patients reported reactions, at follow-up (additional information about study findings is available in the following repository: https://doi.org/10.5281/zenodo.11623493). Both the cross-sectional and the longitudinal analyses showed a linear increase in the VQLQ score with the start and duration of VIT, and a slight decrease was observed after VIT discontinuation (Figure 1). On the contrary, PGWBI did not exhibit significant differences and showed a poor correlation with VQLQ (r = −0.16 to 0.31). This is probably due to the disease-specific nature of the VQLQ, compared to generic tools that do not capture specific aspects of HVA [7].

From the ROC analysis, a 0.1 MCID value demonstrated the best diagnostic accuracy, considering both GRS (AUC 53%, SE 0.05, 95% CI 43%–63%; sensitivity 60%, specificity 56%) and EoO (AUC = 79%, SE 0.04, 95% CI 72%–87%; sensitivity 81%, specificity 68%) as standard references. The sensitivity analyses identified an MCID ranging from 0.1 to 0.8 when using GRS and EoO, respectively (sensitivity analysis #1), and from 0.2 to 0.6 when using the distributional methods (sensitivity analysis #2).

The majority of patients reported an improved VQLQ score at follow-up (68.8%). However, not all of them reached the MCID threshold (MCID-events). According to the new 0.1 MCID, the majority of patients reached a clinically significant HRQoL improvement, at follow-up, in the pre-VIT (67.6%) and VIT <1 year groups (61.5%). The frequency of MCID-events decreased as the VIT duration increased, following a linear trend (Figure 1) that is not captured when using the 0.5 MCID, as it underestimates the frequency of MCID-events. In terms of time-to-MCID, the highest frequency of MCID-events was recorded within 1 year from VIT initiation (Figure 1), suggesting that, in addition to the cumulative positive effect of VIT duration on the HRQoL, the initiation of VIT is likely to be the factor leading to the fastest HRQoL improvement.

For the first time, we conducted a formal analysis for MCID calculation of VQLQ in a high number of patients, with a well-represented range of reaction severity, and with VIT stratification. The limitation of our study is linked to the unidimensional assessment of the VQLQ, possibly leading to overlooking individual factors (e.g. not tolerated re-stings, psychological traits) affecting the HRQoL and the MCID calculation [8]; moreover, the time-to-MCID was extrapolated from the cross-sectional analysis, as the follow-up was limited to 12 months.

In conclusion, the new 0.1 MCID showed higher diagnostic accuracy, compared to other cut-off values, and allowed to perform a multi-endpoint analysis of HRQoL, encompassing frequency of MCID-events and time-to-MCID, in addition to the VQLQ score.

Conceptualization: Matteo Martini and Maria Beatrice Bilò. Project administration: Marina Mauro. Investigation: Marina Mauro, Donatella Bignardi, Patrizia Bonadonna, Maria Chiara Braschi, Francesca Emiliani, Laura Guerra, Serena Liberati, Francesco Olivieri, Valerio Pravettoni, Donatella Preziosi, Erminia Ridolo, Federica Rivolta and Maria Beatrice Bilò. Methodology: Ilaria Baiardini and Matteo Martini. Statistical analysis: Matteo Martini. Writing original draft: Marina Mauro, Ilaria Baiardini, Matteo Martini and Maria Beatrice Bilò. Review and editing: all the authors.

The authors declare no conflicts of interest.