People Diagnosed with Myasthenia Gravis have Lower health-related quality of life and Need More Medical and Caregiver Help in Comparison to the General Population: Analysis of Two Observational Studies

Two observational, digital, international studies have been conducted among members of the general population diagnosed with MG: (1) the General Population Norms study (POPUP), and (2) the MyRealWorld-MG study (MRW). Sample sizes in both studies were not based on formal calculation but driven by the aim of representativity for the general population and by feasibility for MRW.

POPUP is a multinational digital study recruiting 9000 members of the general public in eight countries (US, Canada, UK, Italy, Spain, Germany, The Netherlands, and Belgium). The main objective of this study was to document international generic HRQoL population norms, which can be used as a baseline against which outcomes of patients with a particular disease can be compared to quantify their disease burden [16]. Additionally, data specific to MG was collected to serve as a comparator on the MG measures used in MRW. In all eight countries, samples were interlaced drawn from representative panels operated by market research companies based on previously agreed variables (age, gender, education, and region). Stepped random sampling methods and propensity weighting were used to ensure that participants with a variety of profiles were enrolled in the study. Potential participants were invited via e-mail and were offered points for completing the online survey, which could later be converted into a selection of gifts. Data used in this analysis concern only the baseline data collected in January–March 2021; data from a second wave of data collection (Q1 2023) were not included.

MRW is a digital, prospective, observational, longitudinal multi-country study, conducted among 1859 adults diagnosed with MG from nine countries (US, UK, Canada, Italy, Germany, Spain, France, Denmark and Japan; data cut July 2022). The objective of this study was to provide a comprehensive real-world, long-term view of the impact of MG in a large, diverse cohort of people diagnosed with MG from their perspective. Patients were informed about the study via clinical centers and patient advocacy groups (PAGs). Patients could register themselves online, after which they entered disease characteristics (diagnosis, disease duration, antibody status, treatments received), and monthly data on their experience of living with MG over a 2-year period, using a smartphone application designed by Vitacces Ltd. More detailed information on the rationale and design of the study is available in the study protocol, published in 2019 [17]; the analysis presented here uses baseline data from when the patients joined the study.

In addition to basic demographics, such as age, gender, and country, participants were asked about their living situation (living alone, living with family, in a nursing home) and their need of help from a caregiver in their daily activities. Furthermore, participants were prompted to report any existing comorbidities and side-effects of medication they are taking, and to provide the number of health care visits during the past month (nurse, general practitioner, specialist, physiotherapist, ER visit, hospital admissions). The frequency and duration of sick leave in the past month was also documented. HRQoL data were collected using the EQ-5D-5L and the HUI 3, and the most frequently used outcome measure in MG, the MG-ADL, was also administered. Licenses for all PROMs were obtained in all appropriate languages.

The EQ-5D-5L is a generic instrument to measure HRQoL and consists of a descriptive system assessing five dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each dimension is described in a single item with five severity levels ranging from “having no problems” to “having extreme problems” [18]. The results of the EQ-5D-5L questionnaire can be converted into preference-based quality weights (utility values) on a scale anchored at 1 (full health) and 0 (for a state as bad as being dead), allowing for negative values for states considered worse than dead. Besides the descriptive system, the questionnaire includes a health thermometer, visual analogue scale (EQ VAS) ranging from 0 (worst imaginable health) to 100 (best imaginable health) on which respondents rate their overall health on the day of completion [18]. For the calculation of EQ-5D-5L utility values in POPUP, the country-specific value sets [19‐30] were applied to the respondents from each corresponding country. Utility values for the MRW sample were calculated by applying the country-specific value sets to the whole patient population, as sample sizes were insufficient in most countries.

The Health Utilities Index (HUI 3) is a generic, preference-based, comprehensive system for measuring respondents’ current health status on eight dimensions of health: vision, hearing, speech, ambulation, dexterity, emotion, cognition, and pain. Each level is described with five or six levels of ability/disability [31, 32]. HUI 3 responses can also be summarized into utility values which have good agreement with EQ-5D utilities, but are not interchangeable [33]. The Canadian value set was applied to all data from POPUP and MRW as this is the only available value set to transform the HUI 3 responses to utility values.

The MG-ADL is an MG-specific symptom scale with eight items (talking, chewing, swallowing, breathing, brushing teeth and combing hair, rising from a chair, double vision, and eyelid droop) across four domains (bulbar, respiratory, limb weakness, and ocular). The severity of each item can be indicated by assigning a score between 0 and 3, after which a total score can be calculated (ranging from 0 to 24, with higher scores implying more severe disease) indicating the total symptom burden [34, 35]. In this study, a total MG-ADL score of 0–4 was considered mild, 5–9 moderate, and 10 or higher severe. These cut-offs were established based on input from neurologists and inclusion criteria from clinical trials, where a score of 5 and above was used to classify patients as moderate-to-severe, and are currently being utilized in multiple publications and analyses [36, 37]. The MG-ADL scale was originally designed for clinical settings to be completed by a neurologist while examining his/her patient. A dedicated study on the concordance between direct self-assessment (by the patient) and the proxy-assessment (by the neurologist) indicated excellent concordance between the two assessment approaches [38]. Applying that study’s results to this research, the patient-reported MG-ADL can be considered a reliable estimation of the patient’s symptom burden.

Ethical approval was obtained in all countries for both studies from Salus IRB (US, UK, Germany, Netherlands, Denmark), CPP (France), KULeuven (Belgium), Veritas IRB (Canada), local ethical approval (Spain, Italy), and MINS-IRB (Japan) [17], and the authors received permission to access the data. Informed consent to collect data and to analyze and report anonymzed and aggregate data was obtained from all participants from both studies. The study was performed in accordance with the Declaration of Helsinki.

Data from both studies were compared using descriptive analyses. In POPUP, all data were complete and there were no missing data. In MRW, all analyses are based on all available data, no data were imputed, nor were any observations deleted from the analysis. For categorical variables, the proportions for each level were presented; for continuous variables, the mean, standard deviation (SD), and the 25% and 75% percentiles are shown. Visually, distributions were compared with a box plot, a histogram, or with Kernel densities. Statistical testing for differences between the populations or between severity groups in MG-ADL score was carried out with a generalized estimating equation model, with a normal distribution and an identity link. Testing for differences in utility values (based on the EQ-5D-5L or the HUI-s descriptive system) was carried out using a transformation of the dependent variable: the utility was transformed into a utility complement (= 1-utility) in order to obtain a right-skewed distribution which is more amenable to statistical modeling. In those models, an identity link and a normal distributions were used. Testing for differences in proportions needing a caregiver, taking time off sick, and needing medical care during the past month between the general population and people diagnosed with MG was carried out with a chi-square test.

POPUP enrolled 9000 members of the general public, whereas the MRW study enrolled 2074 people diagnosed with MG (Table 1). Among the people diagnosed with MG, 11.6% had ocular disease, 35.0% had generalized disease, and 53.4% had both ocular and generalized MG at baseline. Furthermore, 43.4% had received a thymectomy in the past, 59.0% were AChR antibody-positive, and 5.6% were MUSK antibody-positive. The mean disease duration was 9.3 years (SD 9.9 years). About 86.2% of patients were treated at baseline, and received as current treatments: 75.6% anti-chollinergics, 43.3% cortico-steroids, 19.7% azathioprine, 14.7% mycophenolate, 3.4% ciclosporine, 6.7% tacrolimus, 1.6% methotrexate, 4.4% eculizumab, 4.5% rituximab, and 3.7% plasma exchange.

In the comparison between the two populations, POPUP has a more equal gender distribution (51.2% female) than MRW (68.8%). Just over half (55.9%) of people diagnosed with MG are able to live independently, compared to 70.5% of respondents in the general population sample. Results from the EQ VAS show that people diagnosed with MG rated their own health on average 14 points lower than the general population. The most frequently occurring co-morbidities among people diagnosed with MG are thyroid problems, anxiety, depression, and respiratory problems (Supplementary Table S1). Furthermore, these patients suffered markedly more often from respiratory disease, osteoporosis, and thyroid problems/disorder than the general population.

Within the MRW patient population, 41.7% had mild MG, 39.6% had moderate disease, and 18.7% suffered from severe MG at baseline. MRW respondents had a mean total MG-ADL score of 5.8, which was strongly contrasting with the 1.2 mean score of the general population sample from POPUP (p < 0.0001, Fig. 1), in which only 3.5% of respondents approximated the symptom burden of moderate MG and 2.9% of severe MG. The POPUP respondents who had the symptom burden of moderate to severe MG were mostly younger males, living in the US and were in need of a caregiver.

Underpinning the difference in total MG-ADL score between the populations are marked differences on all 8 items. The proportion of patients in MRW reporting problems across all dimensions accumulates to 53%, whereas this was only 9% in the POPUP sample. For example, MRW patients experienced shortness of breath with exertion (51.0%) or at rest (10.7%) much more often than POPUP participants (19.9% and 2.8%, respectively) (Supplementary Table S2).

People diagnosed with MG had significantly lower EQ-5D-5L utility values compared to the general population in all countries (p < 0.001; Table 2). Mean utility values were lowest in the UK and highest in Germany in both the general and the MG population. The utility difference between these two populations was the smallest in Germany (− 0.081), while the largest difference was found in the US (− 0.142). The proportion of participants with a utility value of 1 indicating “perfect health” was 32.7% in POPUP, whereas this was only 9.9% in MRW. The proportion of patients with a negative utility value was however fairly similar between MRW patients and POPUP participants, with 1.5% and 1% respectively.

Examination of EQ-5D utilities by MG-ADL classification categories showed a statistically significant and meaningful difference between people diagnosed with MG who carry a mild symptom burden versus those who are moderately or severely affected (p < 0.001). People diagnosed with mild MG had a mean EQ-5D-5L utility of 0.872, which is comparable to, and even higher than, the average general population respondents in the POPUP study (0.843). Moderately and severely affected patients scored on average 0.165 and 0.361 lower than mildly affected patients (Fig. 2).

Figure 3 displays the left-skewed distribution of utility values in the general population, with a density peak near full health. This is in contrasts to the utility distribution among people diagnosed with moderate-to-severe MG, which displays a low density at full health and a peak around 0.7, with a markedly larger tail towards negative utility values. In Fig. 4, Kernel densities by disease severity show the distribution of EQ-5D-5L utilities among patients suffering from mild, moderate and severe disease. People with mild MG have a similar utility distribution than members of the general public, but as the disease severity worsens the distribution is shifted to the left.

Utilities based on the results of the HUI 3 survey confirmed that people diagnosed with MG experience statistically significant lower utility values (decrement of 0.253, p < 0.001) compared to the general population (Table 2). This difference was accumulated through all domains, but was especially noticeable in the comparative proportion of no problems with dexterity, ambulation, cognition, and pain (Supplementary Table S3). The proportion of participants with a utility of 1 (indicating “perfect health”) was markedly lower in MRW (1.9%) compared to POPUP (11.8%). Conversely, the proportion of participants reporting negative utilities (a health state considered worse than death) was three times higher in MRW (9.3% vs. 3.6%).

The HUI 3 utilities also showed a significantly strong association with MG severity (p < 0.001), with declining utility values for people suffering from mild (0.695), moderate (0.443), and severe (0.168) MG (Fig. 2). The histogram of HUI 3 utilities shows a peak near 0.95 among the general population, whereas the distribution is much flatter and shifted towards the left for the people suffering from MG (Fig. 3). The HUI 3 instrument is also better at differentiating utility values between members of the general population and patients with mild MG (Fig. 4). The utility distribution for people diagnosed with moderate and severe disease is also situated to the left, indicating worse utilities experienced by these patients.

The correlation between the utilities based on the Canadian HUI 3 value set, and the EQ-5D utilities based on each respondent’s country-specific dataset are strong and statistically significant: overall the correlation = 0.66 (p < 0.0001), and separately for each population the correlation = 0.65 in POPUP and correlation = 0.67 in MRW (both p < 0.00021).

Sick leave was taken by 34.4% of people diagnosed with MG in the past month, which is nearly three times more than the working or studying general population (13.2%, p < 0.0001), (Table 3). In addition, a higher proportion of respondents indicated not being able to work at all due to illness (16.8% vs. 4.8%, respectively, p < 0.0001).

One-third (34.8%) of people diagnosed with MG reported needing help from a caregiver with daily activities, in stark contrast to 8.3% in POPUP (p < 0.0001,Table 4). Caregivers were mostly family members or friends; fewer than 10% of caregivers for people diagnosed with MG were professionals. The number of hours of caregiver help did not differ between populations. The impact on the careers of caregivers of people diagnosed with MG was large however: more caregivers had to stop working, and almost two day’s work was given up among the caregivers who decided to reduce their working hours.

The proportion of people diagnosed with MG that had any type of medical care visit in the past month was twice as large in comparison with the POPUP sample (51.9% vs. 24.6%, p < 0.0001; Table 5). In addition, people diagnosed with MG paid significantly more visits to specialist (0.834 vs. 0.184 visits) and the A&E department (0.110 vs. 0.013 visits). Both the hospitalization rate and the length of stay were high among people diagnosed with MG with 0.125 hospitalizations per month and an average of 9.4 days length of stay (LOS) in the general ward and 2.3 days in the ICU ward.

In comparison with the POPUP sample, the MRW sample consisted of a somewhat younger population and relatively fewer males, which is consistent with MG being more prevalent among females [2, 43]. We sought to determine whether this difference imposed a significant effect on our findings by adjusting the POPUP sample for the age and gender distribution of the MRW sample in a separate comparison. This adjustment only had a minor effect on the results; hence, we decided not to include the adjusted data in order to maintain the national representatives of the POPUP data. Besides, age- and gender-adjusted results of the POPUP study showed a (slightly) larger difference in outcomes, making the current estimates of unmet need in people diagnosed with MG conservative.

The mean MG-ADL total scores in the MRW study is comparable to what was observed in other studies: 6.2 [41], 4.0 [12], 5.8 (for Generalized MG) [44], and is within the range of “Ever-refractory” (9.1) and “Nonrefractory” (5.1) [45].

Our study established that people diagnosed with MG had lower EQ-5D-5L utility values compared to the general population in all research countries. A recent systematic literature review [13] documenting the humanistic burden of MG included 4 studies (2009–2019) which reported EQ-5D utilities in people diagnosed with MG. Utilities observed among people diagnosed with MG were 0.626 ± 0.286 in one study [46]. Furthermore, Barnett et all (2018) [47] found in another study that the mean EQ-5D utility decreased with increasing (more severe) MGFA class: Class I, 0.89 ± 0.06, Class II, 0.78 ± 0.16, Class III, 0.58 ± 0.24, and Class IV, 0.61 ± 0.22. This is consistent with our results, which showed a declining gradient in utility values from mild to severe MG, however we observed a wider range of utilities, especially towards the lower end. Similarly to our study results, mean EQ VAS scores among people diagnosed with MG from previous studies were also found to be lower than the population norms we estimated based on POPUP: 53 ± 20 [46], 68 ± 20 [47], and 67.7 ± 19.4 [48] EQ VAS scores.

A smaller proportion of people diagnosed with MG (55.9%) reported to live independently compared to the general population (70.5%). The high level of dependency on others of people diagnosed with MG has also been established in previous research [12], reporting a high rate (86%) of patients living in a dependent partnership. Lehnerer et al. (2021) [12] investigated the lost productivity of MG, reporting that 72.6% of formerly working people diagnosed with MG experienced limitations regarding employment, of whom 45.8% had to stop working altogether. This is a markedly larger proportion than the 10.1% observed in the MRW study. This difference may partially be explained by the differences in severity distribution, in age, country of living, and perhaps in the proportion of patients with ocular problems. Our results are also lower than work productivity losses reported by Winter et al. [49], where a 40% unemployment rate due to MG was found, despite the fact that almost 90% of the people diagnosed with MG had MGFA Class I or II. A Danish cohort study investigating labor market participation and long-term sick leave among people diagnosed with MG found that MG increased the odds of unemployment with almost six times and sick leave with nine times within 2 years compared to the general population [50], a markedly larger impact than the threefold increase for both unemployment and sick leave established in our study. Furthermore, Boscoe et al. [51] did not report a significant difference between people diagnosed with refractory and nonrefractory MG regarding employment status, whereas in our study a much higher proportion of moderate-to-severe patients took sick leave or could not work in comparison to mild patients.

About one in three people diagnosed with MG reported needing a caregiver in the MRW study, which is similar to the findings of The Centre for International Economics (CIE) [52] investigating the cost of MG to patients and their community, reporting that one-third of people diagnosed with MG require their partner to be their primary caregiver.

A systematic literature review comparing EQ-5D scores in chronic diseases with the general population concluded that patients with neurological disorders have the highest reduction in HRQoL [53]. This is consistent with other studies reporting markedly lower mean EQ-5D values than in our POPUP study (0.739) for patients with ALS (0.59 [54], 0.54 [49], and 0.55 [55]), multiple sclerosis (0.6 [56], 0.31 [57], 0.59 [58], 0.59 [59], 0.68 [60], and 0.78 [61]), Duchenne muscular dystrophy (0.4 [62]), Alzheimer’s disease (self-reported: 0.77 [63] and 0.71 [64], proxy-reported: 0.6 [63], and 0.30 [64]) and Parkinson’s disease [0.59 [65], 0.71 [66], and 0.62 [67] (median)].

Although care was taken to minimize bias in the data collection, it cannot be excluded that selection bias was present in these digital data collections, as only individuals with access to the internet and a smartphone device could enroll. In MRW, the study population has a higher proportion of female participants compared to other MG-specific cohorts [40, 45, 68‐71], which might affect results as QoL in women has been documented to be lower than men in several studies. As this is a cross-sectional study, the severity distribution observed in MRW is not representative of the whole MG patient population. Indeed, potential participants of the MRW study were made aware of the study through PAGs and social media; therefore, a possible bias to more proactive individuals cannot be excluded. Furthermore, patients were required to enter their monthly data via an application on their smartphone; therefore, patients with (severe) ocular problems (common in MG) would be prevented from participating. In POPUP, the recruitment of members of the general population was carried out by different market research companies in each country, which may have caused inconsistencies in the recruitment process. In addition, potential participants were offered a small compensation, which could have encouraged more people with a lower socioeconomic status to participate [72]. Lastly, it is important to keep in mind the multidimensionality of HRQoL when interpreting our results. This study compared the outcomes of two generic instruments, an MG-specific instrument, and several variables regarding the need for a caregiver and the use of medical resources, thus not covering all aspects. For a more comprehensive picture of the HRQoL of people diagnosed with MG compared to the general population, a tandem paper has been published, identifying the burden people diagnosed with MG suffer with a focus on problems with breathing, fatigue, sleep, mental health, and pain/discomfort (Dewilde 2023, under review).