MyRealWorld-MG is a digital, observational study that began gathering real-world evidence on the burden of MG through patient-reported outcome data in December 2019. The study recruited 2,424 adult patients diagnosed with MG through clinical centers and/or patient advocacy groups across ten countries (US, UK, Canada, Italy, Germany, Spain, France, Belgium, Denmark, Japan). Using a smartphone application, patients provided data on demographics, disease duration, sick leave and caregiver help. They also completed the Myasthenia Gravis Activities of Daily Living (MG-ADL) and Myasthenia Gravis Foundation of America (MGFA) questionnaires (14). It should be noted that our analysis solely utilized data from MyRealWorld-MG entered at baseline. All patients were eligible for inclusion in the study. There were no exclusion criteria based on disease severity. The sole exception was that patients above the age of 65 were excluded from all analyses since the intention was to examine productivity losses related to paid work time, and it was assumed that patients above this age were more likely to be retired.

The MG-ADL consists of 8 items (talking, chewing, swallowing, breathing, brushing teeth and combing hair, rising from a chair, double vision and eyelid droop) across 4 domains (bulbar, respiratory, limb weakness and ocular). The 8 items are scored on a scale of 0 to 3 with the final cumulative score ranging between 0 and 24. Higher scores indicate the patient is suffering from more severe symptoms (15). We used the MG-ADL total score as a continuous variable in our analysis to estimate the association between symptom severity and reduced work productivity, and to categorize patients as having mild (0–4), moderate (5–9), or severe (> = 10) MG. This categorization matched previous MG-related investigations, followed the advice of neurologists, and mirrored the inclusion criteria from clinical trials, where a score of 5 or higher was used to classify patients as moderate to severe (14, 16, 17). The MGFA consists of 5 classes and several subdivisions that are used to group patients with similar clinical features or disease severity. The classes progress in severity, with Class I denoting exclusively ocular MG, while Class V refers to intubation with/without mechanical ventilation (18).

ADAPT (clinicaltrials.gov: NCT03669688) was a multi-center, double-blind, randomized controlled trial evaluating the efficacy of efgartigimod in adult patients with generalized MG (16). The primary outcome measure in this study was the MG-ADL score, which was measured at baseline and after 4 weeks of treatment. Unlike the MyRealWorld-MG study, which recruited patients from all severity levels, ADAPT only enrolled patients with moderate to severe generalized MG. To apply the relationship between MG-ADL and productivity losses observed in the MyRealWorld-MG dataset in a cost analysis, we have utilized the change in MG-ADL between baseline and 4 weeks of treatment with efgartigimod or standard of care (SoC).

We evaluated productivity losses based on the latest available data cut from MyRealWorld-MG study at the time of analysis (July 2023). In the base case analysis, we included data from all countries. Productivity losses were firstly analyzed and reported by symptom severity, using the MG-ADL score. Next, reductions in productivity were examined by disease duration to understand changes in types of productivity losses over time. Thirdly, logistic regression models were developed to evaluate the impact of patient characteristics on work productivity for key outcomes of interest: whether sick leave was taken in the past month (yes/no), whether patients needed regular help from a caregiver (yes/no), and whether caregivers stopped paid work or reduced their working hours (yes/no). In all three equations, the explanatory variables considered were based on determinants found in published literature. These variables included MG type, patient age, sex, geographical region (Europe, Japan, US + Canada) and disease duration. p-values from a Wald Chi-Square test and Type-3 p-values are reported, with the former testing the significance of individual response categories against the reference category, and the latter if the inclusion of the predictor variable in the model significantly improves the fit compared to a model without that predictor. Data were analyzed using SAS version 9.4 (2020), proc genmod with a binary distribution and a logit link.

Our base case cost analysis was based on the whole MyRealWorld-MG dataset and took a UK perspective. We estimated productivity losses by multiplying the number of working hours lost per month (sick leave, reducing hours or stopping work) by the average wage per hour in the United Kingdom (UK) (19), calculated to be £20.44/h using 2023 data. The number of hours per lost workday were calculated to be 7.28. Mean monthly productivity losses were extrapolated over 1 year to estimate annual productivity losses, separately according to each MG-ADL score (0–24) and overall. In sensitivity analysis, productivity loss estimates for three countries were estimated in a comparable fashion using their country-specific MyRealWorld-MG data on wages, worktime and reductions in work productivity, combined with OECD average wage estimates for the country in question (20–22). The selected countries were Italy, the United States and Spain, as they had, in that order, the highest number of patients, jointly comprising 69.2% (726 patients) of the total sample.

As an example of how these productivity loss estimates can be applied, we combined these data with clinical efficacy data to estimate the change in productivity losses associated with treatment. In this study, we used the distribution of patients by MG-ADL at study baseline and week 4 from the ADAPT trial to explore the impact efgartigimod may have on work productivity losses (16).

Of the 2,424 patients from the MyRealWorld-MG study, 1,049—those with sick leave and MG-ADL data and/or caregiver and MG-ADL data—were included for this analysis (Table 1). These patients were predominantly women (73.4%), with the majority (52.3%) aged between 35 and 54 years. They had been diagnosed with MG on average 8 (Standard Deviation (SD) 9.8) years previously. The sample included patients from 10 countries, with most being from Italy (32.8%), the United States (24.3%) and Spain (12.1%). The same trends were observed in the total MyRealWorld-MG population (Supplementary Table S1).

MyRealWorld-MG data indicated that patients’ capacity to work was severely affected by MG (Table 2). Absenteeism data indicated that 36.5% of patients had taken sick leave in the past month, with an average of 14.5 days lost per month. In addition, 11.4% of patients indicated they had not taken sick leave because the severity of their illness prevented them entirely from working, or they had retired early. The study did not include questions on presenteeism for MG patients and thus yielded no data on this front.

In MyRealWorld-MG, the percentage of patients requiring assistance from a caregiver when carrying out their regular activities was 36.0% (Table 3). In almost all cases, this caregiver was a family member or partner. On average, a caregiver dedicated 1.7 (SD 1) h a day to caregiving, which affected the amount of time they could dedicate to their work. In total, 14.6% of caregivers reported cutting working hours, with an average of 13 h cut per week, while 15.5% had to stop working altogether.

Higher MG-ADL scores (i.e., more severe disease) were associated with a higher proportion of patients taking sick leave, an increase in the number of sick days taken as well as a greater proportion of patients needing assistance in carrying out daily activities (Figures 1–3). Higher MG-ADL scores also were associated with the inability to work. The proportion of caregivers reducing working hours or stopping paid work increased slightly as a result of higher MG-ADL scores (Figure 4).

Patients were most likely to take time off from work in the form of sick leave within the first 2 years after their diagnosis (Table 4). As disease duration increased, the proportion of patients taking sick leave decreased, while the proportion of patients going on disability leave or retiring early increased. The highest proportions of patients no longer being able to work due to their disease were observed in the groups that had lived 5–10 and 11–20 years with MG (16.8 and 15.3% respectively). Caregivers also increasingly stopped working or retired early as the duration of MG increased and they had to assist one third or more of patients throughout the follow-up.

All three multivariable regression equations indicated that increases in the MG-ADL score significantly raised the odds that patients would take sick leave, that they would need help from a caregiver, and that the caregiver would need to reduce working hours (Supplementary Tables S2–S4). Multivariable regression analyses also revealed that time since diagnosis, patient age and region were associated with taking sick leave: older patients were less likely to take sick leave, while recently diagnosed patients and patients from Japan were more likely to take sick leave. Similarly, patient age and region were associated with needing a caregiver: patients between 30–39 years-old were more likely to require caregiving, while patients from Japan were less likely to seek the assistance of a caregiver (Supplementary Tables S2, S3). It is noted that these analyses utilize pooled data from different countries/regions, which could have affected the resulting p-values. Sensitivity analyses regarding MG type (ocular vs. generalized) were also conducted, but no statistically significant associations with taking sick leave, needing a caregiver or caregivers reducing working hours were found.

Table 5 shows that total productivity losses based on all patients in the MyRealWorld-MG cohort and on UK wage estimates amounted to £16,630, of which £13,891 (84%) was incurred by patients and £2,739 (16%) was incurred by caregivers. The majority of productivity losses were the result of taking sick leave, while caregivers primarily incurred productivity losses as a result of retiring early. A clear association between MG symptom severity and productivity losses emerged from our analysis of this dataset: the total productivity losses per year on average for patients with mild MG was £7,884 compared to £19,897 for patients with moderate MG and £30,016 for patients with severe MG. Patients with severe MG incurred 3.2 times more productivity losses compared to patients with mild disease, while the caregivers of patients with severe MG incurred 11.9 times more productivity losses than the caregivers of patients with mild MG.

Furthermore, the proportion of caregiver productivity losses (as part of the total losses) increased with more severe disease: it was around 7.2% for patients with mild disease compared to 22.4% of total productivity losses for caregivers of severe MG patients.

Whilst the base case analysis included data from all 10 countries in MyRealWorld-MG, this sensitivity analysis reports country-specific data from the countries with sufficient data to analyze results by MG-ADL, namely Italy, the US and Spain (Supplementary Table S5). Reductions in patient and caregiver productivity in Italy and Spain followed a similar trend as the overall data from MyRealWorld-MG: approximately a third of patients reported taking sick leave in the last month and approximately a third reported requiring caregiver support. In contrast, productivity reductions in the US were reported to be higher than other countries with approximately 45% of patients reporting taking sick leave in the last month and 49% requiring caregiver support. In all three countries, a large proportion of caregivers had to reduce working hours or stop working altogether, although very few patients were found to retire early. It is noted that in absolute value the productivity losses will be quite different between these countries, due to a different structure of productivity losses, determined by the country’s labor laws, social security and access to healthcare, and a different absolute prevalence of patients.

Our analyses also showed a clear association between MG symptom severity and productivity costs in Italy, Spain and the US (Table 6), although estimates showed considerable between-country variability, attributable to wage differences on the one hand, and variability in the proportions of sick leave and early retirement on the other hand. Notably, productivity losses for caregivers of severe MG patients in Italy were markedly high compared to the other countries, comprising up to 42% of the total productivity losses experienced by severe MG patients and their caregivers. Spain showed the highest total productivity losses for MG patients per year out of the three countries examined. In the United States, total productivity losses per year correlated the least with MG symptom severity.

Table 7 shows an example cost analysis that utilizes the MG-ADL distribution reported in the ADAPT study at baseline. As the goal of the study was to evaluate the efficacy of efgartigimod, only patients with moderate-to-severe MG were enrolled, hence the lack of patients with mild disease severity at baseline. When the disease severity distribution after 4 weeks in the SoC arm is compared with that of the treatment arm, a clear treatment effect can be observed: a little over a quarter of patients in the SoC arm reported mild disease severity (27%), compared to over half of patients in the treatment arm (57%). Using data on productivity losses (previously reported in Table 5) and the severity distributions from ADAPT, we calculated the impact of efgartigimod on productivity losses with a weighted average. Productivity losses were estimated to fall from £23,139 to £13,836 for efgartigimod treated patients (a 40.2% reduction in costs), compared to £18,212 for SoC patients (a 21.3% reduction in costs). Our analysis suggests that efgartigimod treatment could reduce the number of work days lost by patients and caregivers, mitigating productivity losses by £4,376 compared to SoC and £9,303 versus baseline (Figures 5, 6). This difference in productivity losses of £4,376 could be considered a gain in productivity.