Impact of imbalanced gender participation in online myeloproliferative neoplasm symptom surveys.

Abstract

e19078 Background: Patients with myeloproliferative neoplasms (MPN), including polycythemia vera (PV), essential thrombocythemia (ET), and myelofibrosis (MF), experience chronic disease-related symptoms. Surveying these patients while incorporating the MPN-Symptom Assessment Form (SAF) gains valuable insight. Independent surveys evaluating symptoms among MPN patients have seen disproportionate participation among females compared to males (upwards of 4:1). Though the general MPN patient population is roughly gender balanced, epidemiological studies show the occurrence of MPN types differ between men and women (MPN [%male]: PV [65%], ET [33%], MF [50%]. Mehta, et al. Leuk & Lymph 2014). In-clinic evaluations and international MPN surveys of disease-related symptoms suggest women report higher symptom burden on average compared to men, despite MPN type [Emanuel, et al. JCO 2012. Geyer, et al. Haematol 2017]. Further, a recent study across cancer trials showed women had a 34% increased risk of reporting severe symptoms [Unger, et al. JCO 2022]. This study aimed to evaluate gender imbalance in survey participation among MPN patient surveys and investigate its potential to overestimate symptom burden if gender is ignored. Methods: Five anonymous web-based surveys were used to assess the impact of disproportionate gender participation. The MPN-10 assessing the 10 most clinically meaningful SAF items was present in these surveys and generated the Total Symptom Score (TSS). Survey responses with no self-reported gender or ≥5 missing MPN-10 symptoms were excluded. Raking weights based on expected MPN population were applied at the survey- and MPN type-level. Relative bias of mean TSS was calculated to evaluate over- and underestimation due to this participation imbalance. Relative bias <5% is commonly acceptable. Results: There were 4962 survey participants total and 74% were female (Table). Epidemiologically expected gender frequency by MPN type was not seen (MPN [%male]: PV [31%], ET [15%], MF [37%]). The gender participation imbalance accounted for a 3.1% overestimation of overall TSS. This also resulted in a 6.8% underestimation of the mean difference in TSS between genders, across MPNs. Conclusions: Overall bias due to gender participation imbalance was relatively low. However, analyses considering gender differences were more impacted. Anonymous surveys should seek to reduce participation bias during the survey design phase. More research is needed to evaluate gender and other participation biases in surveys across cancer domains.[Table: see text]