Background
Myasthenia gravis (MG) is a rare autoimmune disease with a prevalence of 15–20/100,000 inhabitants [
12]. First symptoms appear with an age peak around 30 and 70–80 years of age [
1]. Specific antibodies affect the neuromuscular junction and lead to fluctuating fatigability and weakness of the ocular, bulbar and skeletal muscles. In 15%, no antibodies can be detected (i.e. seronegative) [
12]. Therapy with acetylcholinesterase inhibitors, immunosuppressive agents and thymectomy, lead to a stable condition in most patients with only mild to moderate motor symptoms. Despite this, an estimated 10–20% of patients with MG do not achieve an adequate response or are intolerant to conventional treatment [
43]. These refractory cases concern more often females and are typically younger at disease onset [
29]. New therapeutic strategies like the complement-inhibitor eculizumab [
18] have been developed and several more are in the pipeline.
Apart from motor symptoms, also psychological and social well-being are limited in patients with MG [
17]. The increasing interest in health-related quality of life (HRQoL) in MG patients is reflected by a growing number of studies in this field. Tools as the widely recognized SF-36 questionnaire and the MG-specific MG-QoL15 have been used to measure the HRQoL in several MG cohorts [
2,
4‐
6,
23,
31,
37‐
39,
44,
51]. The studies show consistently that severe muscle symptoms and disability are associated with lower physical scores of HRQoL [
37,
38,
51]. Symptoms of depression frequently affect the HRQoL negatively [
23,
44,
51]. Patient characteristics, such as gender, age, education, course of disease, the use of immunosuppressive drugs, the occurrence of side effects, acceptance of disease as well as anxiety and perceived social support, have been demonstrated to be additionally associated with a poor quality of life in MG patients [
2,
5,
45]. However, none of these studies determined whether the factors influencing HRQoL are myasthenia-specific or also apply to the normal population.
The so-called global
burden of disease is a concept that was developed in the 1990s in a cooperation with the World Health Organization (WHO) to describe death and loss of health due to diseases, injuries and risk factors for all regions of the world. The gap between an ideal situation, where everyone lives free of disease and disability, and the cumulated current health status, is defined as the burden of disease [
16]. So far, the
burden of disease in MG in particular and its specific risk factors are not well defined. In treatment-refractory patients, factors like disability, drug- or surgery-associated adverse events, myasthenic crises, MG-related hospitalization, and comorbidities indicate a high burden [
6]. Further, unemployment, lower mental health and HRQoL are likely to be associated with a high burden in treatment-refractory patients [
43]. So far, the burden of disease in
non-refractory patients has not been described.
The aim of this study is to estimate the burden of MG based on a representative patient population using a multidimensional approach. In a case–control study, we matched MG patients with the general population (genP) to compare the HRQoL and to explore myasthenia-specific risk factors for a lower HRQoL.
Methods
Data collection
In May 2019, the 3262 members of the German Myasthenia Association (Deutsche Myasthenie Gesellschaft, DMG) received the study information and a questionnaire as well as a pre-stamped envelope addressed to the coordinating study centre. The study participants (SP) were instructed to return their completed questionnaire without any further identifying information to ensure the anonymity of the survey. No refund was given. Returned questionnaires were accepted within the cut-off date of 31 July 2019.
Questionnaire
The questionnaire concerned demographic data (gender, age, marital status/partnership, size of family), educational status, employment, income, and possession of a severely disabled person card (in Germany delivered at a certain degree of disability ranging from 10 (mild) to 100 (very severe)) were asked.
Regarding the medical aspects of MG, the questionnaire asked for age at symptom onset, age at medical diagnosis, subtype (ocular versus generalized), antibody (Abs) status (Acetylcholine receptor–antibody (Ach-R-Abs), muscle-specific kinase antibody (Musk-Abs), (Lipoprotein-related protein 4 antibody (LRP4-Abs), seronegative), comorbidities, thymectomy, current MG-specific medication (cholinesterase inhibitors, glucocorticoids, long-term immuno-suppressants, monoclonal antibodies, plasmapheresis (PE)/immuno-absorption (IA), intravenous immunoglobins (IVIG)) including dosage/frequency, co-medication (antidepressants, painkillers), side effects and treatment satisfaction.
Most questions were asked with a checkbox option, always specified to be answered as a single or multiple-choice option. Only few questions were asked as free-text format. The questionnaires were scanned and processed with the software TeleForm (OpenText), version 10.9.1.
Definitions
In subgroup analysis, we defined patients with generalized MG, self-rated moderate or high disease severity and any exacerbation medication use in the past (IVIG, PE, Rituximab, Eculizumab) as “treatment-refractory” in accordance with current definitions[
29].
Standardized scores
To further assess the burden of disease, standardized scores were integrated in the questionnaire, (SF-36 (Short Form Health, i.e. general HRQoL) [
33,
48], MG-Qol15 (Myasthenia gravis quality of life, i.e. MG-specific HRQoL) [
7], MG-ADL (Myasthenia gravis activities of daily living profile) [
49], CFQ-11 (Chalder Fatigue scale) [
8,
20,
30], ESSI-D (ENRICHED Social Support Inventory) [
19,
25] and HADS-D (Hospital anxiety and depression scale) [
3,
15,
52]). In the SF-36 (0–100-point scale) and the ESSI-D (5–25-point scale), the higher the score, the better is the patients’ situation. Whereas in the MG-Qol15 (0–60-point scale), the MG-ADL (0–24-point scale), the HADS-D (0–21-point scale for each sub-scale anxiety and depression) and the CFQ11 (0–33-point scale) a high score indicates a worse situation. Additional to the Likert format’, the CFQ11 offers a binary scoring where 4 points or more equate severe fatigue [
8]. In the ESSI-D, low social support is defined as a sum score of 18 or less and at least two items with 3 or less points [
19]. With an HADS-D sub-score, participants scoring 8 points or more are defined as having substantial grades of anxiety or depression [
3].
Imputation of missing values using the SF-36
Following the instructions of Morfeld et al. [
33] to calculate the subscale scores of the SF-36, missing values were replaced by the mean values of the existing items of the same subscales, if at least 50% of the items were answered. For number of missing values with and without imputation of all subscales, see Supplement 1.
Matched controls
To directly compare HRQoL to the general population (genP), we used data from participants of a German-wide representative study [
24] (German Health Interview and Examination Survey for Adults, DEGS1, 2008–2011) which was conducted by the Robert Koch Institute aimed to repeatedly collect representative data on the health status, health-related behaviour, healthcare and living conditions of adults residing in Germany who are aged 18 and over. Information on gender and age was used for the matching of cases (MG patients) and controls using exact matching by gender, and matching by age groups (18–24) (25–29) (30–39) (40–49) (50–59) (60–69) in a ratio of 1:2. Due to the low number of possible controls in the age group 70 + years, matching for this age group was conducted in a ratio of 1:1, resulting in 1649 cases assigned to 2556 controls.
Sociodemographic variables
Educational status was graded into three groups (low, medium, high) on the basis of information on the highest level of education according to the CASMIN classification [
28]. Information of net household income was based on income categories: "Less than 1000€", "Between 1000€ and 2499€", "Between 2500 and 5000€" and "More than 5000€". For comparison with the control group, currency-equivalent values were assigned to these categorical responses (750 = less than 1000€; 1750 = 1000–2499€; 3750 = 2500–5000€; 5500 = more than 5000€). Net household income was weighed according to the number of people living in the household using the new OECD-modified scale [
13].
For comparison with the DEGS1 sample, only income scores (and no data in euros) were available [
27]. The calculated income values of the myasthenia sample were therefore assigned to the corresponding scores and summarized in three income groups (low = up to 1188 euro, medium = 1189–1833 euro, high = 1834 euro and more).
Statistical analysis
The statistical calculations were performed using IBM Corp. Released 2017. IBM SPSS Statistics for Windows, Version 25.0. Armonk, NY: IBM Corp. using IBM SPSS Statistics 25 and R (version 3.5.3) [
40] software. Net diagrams were created using Excel (version 2002) from Microsoft Office 365 ProPlus.
Depending on the scale and distribution of the outcome variables, appropriate descriptive statistics (mean, standard deviation, median, interquartile range, absolute and relative frequencies) are presented. Furthermore, parametric and non-parametric measures were used to test for group differences. A two-sided significance level of
α = 0.05 was used. No adjustment for multiple testing was applied in this exploratory study. Linear mixed regression models adjusted for gender, age, educational status, income and partnership status were calculated (random intercept models, random intercept for matching ID) for the analyses of the differences between MG patients and controls in the SF-36 subdomains
physical functioning and
emotional well-being. Furthermore, interactions between disease status (MG: yes/no) and age, or sex were included. The multivariable analysis was carried out in the full analysis set including estimated values in case of missings. Multiple imputation (
m = 10 datasets) was used to estimate missing using predictive mean matching and chained equations x complete datasets were created and separately analysed. The results were then combined using Rubin’s rules [
42].
Net diagrams
To present various aspects of the burden of disease holistically in net diagrams, the different score values of MG-ADL, MG-QoL15, HADS, ESSI-D, CFQ11 and SF-36 subdomains were levelled on a unidirectional scale from zero (no complaints) to 100 points (strongest restrictions).
Data availability
Data not provided in the article because of space limitations may be shared (anonymized) at the request of any qualified investigator for purposes of replicating procedures and results.
Discussion
In this study, we demonstrate that HRQoL is markedly lower in MG patients compared with the general population (genP). The overall burden is particularly high among women, in high disease severity levels, in low-income groups and among middle-aged and older patients.
Several studies have described a lower quality of life in MG patients with MG-specific and non-specific scores [
5,
37,
38,
44]. The SF-36 has been used to compare means of patients’ values to normative values of controls [
38]. In our in-depth analysis using a matched-pair comparison to the genP in Germany, large differences in the domains
physical functioning,
physical role functioning and
vitality indicate a high individual burden for MG patients. Corresponding to our results, other studies [
2,
5,
51] describe
physical role functioning,
general health perception and
vitality as the domains with the lowest mean values. Interestingly, in our study, mean values in the domain
general health perception (67.3, SD 19.7) do not differ from the genP (67.2, SD 17.9). Similar in the domain
pain, no substantial difference to the genP was seen. Twork et al. conducted a large study with 1459 patients of the German Myasthenia Association (DMG) to explore quality of life [
44]. Compared to these results published in 2010, mean values of the single SF-36 domains have not changed remarkably apart from
pain (46.0[
44] vs. 68.4 in our cohort) and
general health perception (44.8[
44] vs. 67.3). These two categories have now reached genP values, as mentioned above. In MG patients, effects of age on domains, such as
physical functioning and
emotional well-being, are much higher than in the genP. Income and education influence HRQoL in MG patients. However, with our novel matched-pair analysis, we can demonstrate that there are no major differences of these effects compared to the genP.
Compared to other chronic diseases that directly or indirectly impair muscle activity, e.g. rheumatoid arthritis (RA), some similarities can be observed, such as lower physical functioning (SF-36) in older patients compared to genP [
32]. However, the positive association between mean age and the mental health domain described in RA cannot be observed in MG. Interestingly, comparison with other diseases that permanently impair control of muscle function also shows that performance in various SF-36 domains differs; for example, patients with Parkinson's disease and multiple sclerosis have similar limitations in the domain physical functioning as compared to MG. However, these two diseases show significantly greater differences in social role functioning and emotional well-being compared to the normal population than we observed in our cohort [
41]. The extent to which disease-specific patterns can be derived from the SF-36 profiles needs to be investigated in comparative studies. Beyond lower HRQoL, we integrated further standardised scores in our analysis to draw a comprehensive picture of the individual burden of disease, among them the scores of anxiety and depression (HADS-D), fatigue (CFQ11) and social support (ESSI-D). The frequencies of anxiety and depression in MG are remarkable. Although these psychiatric comorbidities are similarly common in other chronic neurological diseases, such as Parkinson’s disease or multiple sclerosis, they should be considered in the treatment of patients as they are known to severely affect the well-being of those affected. As known from other chronic diseases like multiple sclerosis, psychiatric abnormalities essentially change self-perceived severity of disease, as well the perception of therapy response and success [
26]. In our study, the proportion of SP with abnormal depression scale scores is highest in treatment-refractory patients. Further studies have to be conducted to evaluate the effect of depression on self-perceived severity and quality of life. Furthermore, nearly 60% reported persistent fatigue known to have a high impact on quality of life [
17] in patients with MG. Low social support was reported by more than one-fifth of our study participants. Perceived social support, however, engages a health-promoting lifestyle [
21] and in an Italian study (
n = 74) perception of support is a predictor of mental health [
39]. Therefore, low social support might increase the burden of disease. Clinical aspects, such as muscle weakness, double vision, myasthenic crisis, pain, sleep disturbances, the use of immunosuppressive drugs, and medication side effects, as well as demographic aspects, like gender, age, place of residence and medical infrastructure have been demonstrated to be additionally associated with a poor quality of life in MG patients [
2,
5,
45] and thus influencing the burden of disease.
We paid special attention to MG influence on partnership and family planning, education level, employment situation and income as we suspect these aspects to have a high impact on the perceived burden of disease. A high percentage of the patients was living in partnership (86%), which is comparable to previous findings [
6,
50]. One-third of patients require their partner to be their primary carer [
9]. In nearly one-third of study participants, MG played a role in separation or divorce from a partner and in 16.8% MG influenced family planning. A large survey on 801 women with MG revealed that over fifty percent had abstained from having children due to MG [
35], even if according to current knowledge, MG patients should not be discouraged from giving birth. Corresponding to findings in the literature [
14], the age peak around 30–40 years of our cohort concerns mostly women [
10]. This means that the first and initially often strong symptoms occur when patients still work and especially for young women this regards a period, when family planning and career building is an important topic.
A majority of our patients experienced limitations regarding employment due to MG such as incapacity of work or recurrent occupational disability. Similar, in an Italian cohort, at least two out of three MG patients suffered from changes in work and/or income [
47] and a large Japanese cross-sectional study demonstrated that MG patients often experience unemployment (27.2%), involuntary job transfers (4.1%) and a decrease in income (35.9%) [
34]. In an Australian cohort, 39.4% had stopped work due to MG and 19.4% had to change occupation [
4]. Matched Danish MG patients experienced poorer labor market experience and suffered more often from long-term sick leave [
11]. Our data demonstrate a negative influence of low income on the HRQoL in SF-36 sub-domains, such as physical functioning and emotional well-being.
So far, our study with 1660 participants is the largest conducted on this topic. Gender distribution is very comparable to our outpatient clinic (iMZ) and to other study groups from literature [
46]. However, the population of the German Myasthenia Association (DMG) might not fully represent the “average German MG-patient”; e.g., this population is slightly older than the common MG-population [
5,
6,
23,
51]. In addition, it is conceivable that more severely than mildly affected MG patients might register as members of a patient organization like the DMG. In addition, we cannot rule out selection bias: Members of the German self-help organization might have a higher educational level than MG patients who do not register themselves in a patient organisation. Because the questionnaire was written in German and asked specific aspects about the disease, MG patients whose native language is not German or less educated patients might not have returned the questionnaire. Eventually only highly motivated and less sick patients completed the questionnaire. For this reason, we offered a long response time of 4 months to catch moments when patients felt able to fill out the questionnaire. When asking for information that lies far in the past like age of symptom onset, a recall bias could have affected the results. However, the majority of questions was related to the current situation and recall bias should be small, but even relevant and an explanation for some missing answers. Another weakness of our study is that the data of the comparison group (genP) [
24] were collected 10 years ago and some answers might have changed over time. Using a questionnaire which was sent back anonymously did not allow to compare and validate the statements of patients with clinical data or to add objective data of examinations performed by a health care professional. Also, not every questionnaire was validated for use in German (but already used in research [
17,
25,
30]) and, in case of the MG-ADL [
49], not yet validated for independent completion by a patient (ongoing study in our research center). However, we know from other studies in which MG patients were both examined by a doctor in standardized tests and performed self-declarations that objective and subjective data correlate with each other [
36,
46], with the restriction that the data collection was not anonymous in these studies but was carried out by investigators. The term “treatment-refractory MG” was used in accordance to current literature [
18] even if "non-responders to standard treatments" or "high disease activity despite standard treatments" would be more appropriate. The strengths of our study are the matched-pair analysis, a comprehensive multidimensional approach, a representative cohort and a high number of participants (
n = 1660, response rate 52.5%) offering a large dataset in the real-world setting.