Full Text

Turn on search term navigation

Introduction

Multiple Sclerosis (MS) is a chronic neurodegenerative autoimmune disease that affects approximately 3 million people worldwide [1], with a typical disease onset between 20 and 40 years [2]. Persons with MS (PwMS) may suffer from highly individualized and very diverse symptom combinations, often including visual problems, paresthesia, neuropsychiatric disturbances, gait impairment, or fatigue. This considerable symptom burden, as well as frequent daily-life impairments and concomitant diseases (comorbidities), diminish the health-related quality of life (HRQoL), which is often lower in PwMS when compared to the general population [3–6]. Studies have identified specific MS symptoms, including fatigue, depression, gait impairment, or spasticity, as having particularly detrimental effects on HRQoL [4, 5, 7]. Furthermore, a recent investigation into predictors of HRQoL trajectories in PwMS also noted baseline associations of older age, physical impairments, or fatigue with a worse HRQoL trajectory [6].

However, the factors shaping and predicting long-term HRQoL are still poorly understood. These knowledge gaps concerning drivers of HRQoL in PwMS are consequential and hinder a more personalized, HRQoL-oriented management of MS and its symptoms. Previous studies have investigated HRQoL influences using, for example, median regression or trajectory analysis using latent growth models. These studies already offered important insights, but also faced a multitude of analytical challenges for identifying individual-level drivers of low HRQoL. For example, although simple, well validated, and frequently employed, generic HRQoL instruments such as the EuroQol-5-Dimension (EQ-5D) scale only cover a small subset of MS-related health aspects, and they are often not very responsive to short-term changes in symptom severity or disease exacerbation [8]. Also, the EQ-5D and similar non-disease-specific scales are usually normed for the general population, which limits the clinical usefulness for persons with chronic diseases. Rather, such HRQoL measures should be compared against and interpreted in the context of measurement distributions from persons with similar disease characteristics. Finally, PwMS live with their disease for decades, which leads to complex interactions between HRQoL and natural aging and disease progression that are difficult to disentangle.

To overcome these analytical challenges and to offer a novel methodological approach, we used a cross-sectional, observational study design to explore a quantile regression-based segmentation approach that aims to robustly identify individuals with low HRQoL within the PwMS population and after adjustment for age- and disease duration dependent effects on HRQoL. Disentangling age-related from disease-related health changes is complex because they are interwoven and can lead to problems of multicollinearity in statistical analyses. Over long time-horizons, HRQoL and health changes can occur both due to biological aging and disease progression, such as increasing impairment of ambulation.

By applying this segmentation analysis to the database of the Swiss Multiple Sclerosis Registry (SMSR), we sought to compare PwMS with low or high HRQoL after age and disease duration adjustment, and to identify independent disease- and symptom-related characteristics associations with low HRQoL. Our proposed analytic approach mitigates the problem of multicollinearity and uses the population of PwMS as a reference to identify persons with low or high HRQoL.

Materials and methods

Study population

This study includes SMSR study participants. The SMSR has been described extensively elsewhere [9]. The SMSR, a nationwide, longitudinal study of PwMS in Switzerland, offers two participation modes: a one-time survey with basic information for epidemiological studies, and longitudinal participation, including an extended baseline survey and invitations to digital or paper-based semi-annual follow-ups. Active since 2016, the SMSR collected one-time surveys from 2800 PwMS living or receiving care in Switzerland, with 2200 also contributing to the longitudinal study. The SMSR closely collaborates and is advertised by the Swiss Multiple Sclerosis Society and numerous neurological centers in Switzerland. PwMS aged 18 years and older can self-enroll the SMSR online or request paper-based questionnaires from the study center at the University of Zurich. Before the first questionnaire, prospective participants complete an informed consent, including opt-in agreements for medical record release or data linkage with the Swiss Multiple Sclerosis Cohort Study (SMSC) [10]. The SMSR relies on self-reports and mainly collects information on health-related quality of life and lived experience with MS. The SMSC covers high-quality clinical data and disability assessments, MRI, and biomarkers. The SMSC is conducted on-site at eight large MS centers and integrated into routine neurological care. As of December 31, 2022, approximately 1600 participants with a median follow-up period of 6.2 years have been enrolled. About 250 individuals also contribute to the SMSR and have signed informed consents for joint data linkage.

Databases

This study primarily used data from the SMSR database, including HRQoL and other self-reported measures such as sociodemographic information, MS disease characteristics, MS treatments, MS symptoms, disability status, chronic comorbidities, and familial MS risk. Additionally, the study utilized Expanded Disability Status Scale (EDSS) scores and disease duration information from the SMSC and the database for joint tasks of health insurers (SVK database) to calculate MS Severity Scores (MSSS). The SVK database collects administrative information (including EDSS) for cost approvals on behalf of specific Swiss health insurers, encompassing data from approximately 60% of all Swiss PwMS. Cost approval applications require annual re-submission by a certified neurologist. With permission of the responsible ethics committee (see below), the SMSR has been allowed to perform a probabilistic linkage between the SMSR population and the SVK database, yielding an overlap of 1021 individuals. Details on the linkage process are provided elsewhere [11].

This study included SMSR participants who participate in the longitudinal SMSR study, had at least one HRQoL self-assessment, and complete data on age and date of diagnosis (flowchart, Fig 1). The data were accessed for research purposes on September 25, 2023. Some authors had access to information that could identify participants. However, the data were anonymized for the analysis.

[Figure omitted. See PDF.]

VAS: Visual Analogue Scale; HRQoL: Health-Related Quality of Life; EDSS: Expanded Disability Status Scale; MSSS: Multiple Sclerosis Severity Score.

Ethics

The SMSR and SMSC have been approved by responsible ethics committees (PB-2016-00894; BASEC-NR 2019–01027). All participants provided written informed consent. The linkage of the SVK database with the SMSR has been approved by the Ethics Committee of the Canton of Zurich (BASEC-NR 2020–00840).

Outcome measures

The main study outcomes are HRQoL measures based on the EQ-5D survey and the visual analogue scale (VAS). The EQ-5D includes five questions on different health domains (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression), convertible into a score ranging from 0 (worst HRQoL) to 100 (best HRQoL), by utilizing French value sets, which were found to be appropriate for Switzerland in an earlier study [12]. The VAS allows participants to self-rate their HRQoL on a scale from 0 (worst imaginable quality of life) to 100 (best imaginable quality of life).

Confounders and variables of interests

In our study, we distinguished between confounders (factors that are not changeable but known to influence HRQoL) and factors of interest that are either potentially addressable in clinical care (e.g. symptoms) or potential disease course predictors (e.g. known progression risk factors, S1 Table). Pre-specified confounders handled by the quantile-based segmentation approach were age and years since diagnosis at the time of HRQoL measurement. Further analyses were also adjusted for sex and clinical MS phenotype (relapsing-remitting, primary and secondary progressive, clinically isolated syndrome). Additionally, we explored HRQoL associations with pre-specified variables of interest, including gait impairment indicator (Self-Reported Disability Status Scale (SRDSS), a proxy for the Expanded Disability Status Scale (EDSS)), self-reported current MS symptoms (dichotomous variables (yes vs. no)), six common self-reported comorbidities, specific first MS symptoms (dichotomous variables (yes vs. no), which are prognostic for a severe disease course), current DMTs, education, living situation, familial MS risk, smoking, and–where available through data linkage–MSSS, a marker for speed of MS progression, calculated according to the method of Roxburgh et al. [13]. The variables of interest were measured at the same time as HRQoL. The variables of interest and rationales for their inclusion are listed in S1 Table.

Missing information amounted to less than 3% in five categorical (clinical MS phenotype, living situation, marital status, education, smoking) and one continuous variable (Body Mass Index (BMI)), except MSSS, which was only available in 937 individuals. Categories reflecting missing or unknown information were created for the analysis of the five categorical variables. BMI was imputed using a function of age, sex, gait impairment, and self-reported diabetes or cardiovascular comorbidities. MSSS was excluded from regression model building but added to the final models in a separate analysis.

Statistical methods

Classification of HRQoL groups by quartiles.

For each included PwMS, the last complete HRQoL measurement was selected. First, distributional parameters, adjusted for age and years since diagnosis, were estimated. Specifically, we utilized quantile regression to estimate the 25^th (lowest), median, and 75^th (highest) quartiles using EQ-5D or VAS indices as outcomes and age and years since MS diagnosis at HRQoL assessment as explanatory variables. Subsequently, PwMS were classified as having low HRQoL (regarding EQ-5D or VAS) if an index measurement fell below the age- and disease duration adjusted lowest quartile or high HRQoL if a measured indicator was greater than the highest quartile. Index measurements between the 25^th and 75^th quartiles were considered the “normal range” within the population of PwMS. Thus, three groups per type of HRQoL index were created.

Exploration of factors associated with low HRQoL.

We performed descriptive analyses of pre-specified sociodemographic and disease-specific factors across the three HRQoL groups. Continuous variables were summarized as medians and interquartile ranges (IQR), and categorical data as numbers and percentages. Analyses were performed separately for the VAS and EQ-5D indices. Associations were explored using multivariable multinomial regression models, with separate models for VAS and EQ-5D indicator groups. In addition to confounder control through age-and disease-duration HRQoL categorization, three additional pre-specified confounders were included in the models (sex, clinical MS phenotype, SRDSS measure for ambulation impairment). Associations with further pre-specified variables of interest were explored by adding them incrementally to the model and retaining if the Akaike Information Criterion (AIC) was reduced by two or more points. We also examined potential collinearity between the variables of the final models but did not detect any issues to address. Our group-based comparisons of health- and sociodemographic differences across HRQoL strata have the advantage of being intuitive and easily interpretable. However, the categorization comes at the cost of reduced statistical power to detect relevant differences owing to the categorization of HRQoL measures and the associated information loss. We therefore also conducted a simultaneous quantile regression for the 25^th, 50^th, and 75^th percentile as a sensitivity analysis, which uses continuous HRQoL measures [14]. Splines for age and disease duration, as well as sex, clinical MS phenotype, and a proxy measure for gait impairment (SRDSS) were included as a-prior defined confounders, and other variables were added incrementally. Model variance was estimated via bootstrapping with 100 repetitions, and variables were selected on the basis of p-values <0.05.

Analysis of discordant VAS and EQ-5D.

We examined individuals with discordant VAS and EQ-5D assessments, namely VAS ratings markedly lower or higher than EQ-5D indices. Discordance was defined as separation by at least two quantiles. Individuals with a high EQ-5D-low VAS were identified if they had an EQ-5D in the highest quartile and a VAS in the lowest two quartiles or if they had an EQ-5D in the highest two quartiles and a VAS in the lowest quartile. Individuals with a low EQ-5D-high VAS were identified if they had an EQ-5D in the lowest quartile and a VAS in the highest two quartiles or if they had an EQ-5D in the lowest two quartiles and a VAS in the highest quartile. Descriptive and multivariable regression analyses were performed as described above.

Analyses were conducted using Stata SE Version 16.1 (Stata Corp., College Station, TX, USA).

Results

Description of study population

Our study included 1697 PwMS, of whom 73.1% were female (Table 1, column 2). The median age was 49 (IQR 38–58) years, and the median time since diagnosis was 9 (IQR 4–17) years. 68.4% of PwMS reported having relapsing-remitting MS, 4.1% a clinically isolated syndrome, a total of 26.7% a primary or secondary progressive MS, and for 0.8%, clinical MS phenotype was unclear. Median EQ-5D and VAS were 90.7 and 80 points.

[Figure omitted. See PDF.]

Factors associated with low EQ-5D

The quantile-based EQ-5D classification yielded 417 PwMS in the lowest quartile, 858 in the middle quartiles, and 422 in the highest quartile (Table 1, columns 3–5). Median EQ-5D across the groups were 61.5 (lowest quartile), 90.7 (middle quartiles), and 100 points (highest quartile). By design, the three groups were similar regarding age and time since diagnosis. The lowest quartile group included higher percentages of male PwMS (31.2% vs. 26.8%) and persons with primary and secondary progressive MS (44.4% vs. 10.2%) compared to the highest quartile group. The lowest quartile group had the highest number of symptoms (median of 8 vs. 0) and the highest percentage of individuals with severe gait impairment (27.6% with SRDSS 7 or greater vs. 0.7) compared to the highest quartile group. However, descriptive analyses revealed no marked differences in sociodemographic (e.g., education, marital status) or MS risk factors (e.g., smoking, familial MS risk).

As shown in Table 2, testing the pre-specified characteristics in a multivariable multinomial regression model identified disability benefit status (Relative Risk Ratio, RRR [95% Confidence Interval, 95%CI] compared with middle quartiles; 1.52 [1.11–2.08]), severe gait impairment measured by SRDSS of 7 or more (12.69 [7.29–22.11]), BMI (1.05 [1.03–1.08] per score increase), as well as different symptoms (depression (2.37 [1.64–3.42]), memory problems (1.46 [1.04–2.05]), pain (1.69 [1.22–2.35]), and muscle weakness (1.50 [1.07–2.08])) as independently associated factors with low EQ-5D. Based on the AIC, additional factors with less frequent accumulation in the highest quartile group, namely a number of 7 or more MS symptoms (0.16 [0.07–0.36]), past smoking (0.60 [0.41–0.87]), or the presence of fatigue (0.48 [0.33–0.70]), were retained in the model.

[Figure omitted. See PDF.]

Re-calculating the final model only including individuals with a MSSS (n = 937), the factors associated with low EQ-5D remained similar. Also, the MSSS exhibited a marked association with low EQ-5D (1.14 [1.05–1.23]).

The simultaneous quantile regression identified fewer statistically significant factors to exert a differential impact on three quantiles (S2 Table). Compared with the main analysis, SRDSS strata, symptom load, depression and memory problems were also flagged as factors with a more pronounced effect in PwMS with low HRQoL. In addition, gait impairments or spams as first symptoms were newly identified as factors associated with low HRQoL.

Factors associated with low VAS

The VAS-based classification yielded groups of 415 (lowest quartile; median VAS 40 points), 861 (middle quartiles; 80 points), and 421 (highest quartile; 95 points) persons (Table 1, columns 6–8). Like the EQ-5D analysis, the lowest VAS quartile group had the highest proportion of persons with an SRDSS of 7 or greater (16.9%), the largest median number of symptoms (8), or persons with primary or secondary progressive MS (38.8%) compared with other quartile groups.

Multivariable regression (Table 3) identified receiving disability benefits (RRR 1.86 [95% CI 1.38–2.50]), SRDSS of 7 or more (3.14 [1.91–5.16]), and the following MS symptoms: depression (1.61 [1.12–2.31]), fatigue (1.80 [1.23–2.63]), memory problems (1.77 [1.28–2.46]), pain (1.40 [1.02–1.91]), and muscle weakness (1.40 [1.02–1.92]) to be associated with the lowest VAS quartile (compared with middle quartiles). Moreover, Swiss citizenship was associated with the highest VAS quartile (1.76 [1.13–2.75]).

[Figure omitted. See PDF.]

Model re-estimation including persons with a MSSS did not substantially alter the estimates of symptoms and SRDSS, except for pain and depression, which showed somewhat lower associations with the lowest VAS quartile. MSSS nominally also showed an association with the lowest VAS quartile (1.07 [0.99–1.15]). However, the 95% CI no longer suggested statistical significance at the 5% level.

The simultaneous quantile regression identified fewer variables to exhibit differential effects across the quantile strata (S3 Table). Only SRDSS, symptom load, and memory problems were retained from the main analysis. In addition, currently working was identified as a new influential factor.

Discordant VAS and EQ-5D groupings

The VAS and EQ-5D indicators were closely correlated (Spearman’s rho = 0.74). Nevertheless, VAS and EQ-5D measure somewhat different constructs, and we further examined discordant groupings (Fig 2 and Table 4). Green dots illustrate individuals with low EQ-5D-high VAS (n = 92, median EQ-5D 76.2, median VAS 90 point), red dots show individuals with high EQ-5D-low VAS (n = 82; median EQ-5D 95.8, median VAS 66.5 points). Descriptive analyses, which are shown in Table 3, suggest that a high ED-5D and low VAS was more frequent among younger PwMS who had a diagnosis for less long, and who reported a lower symptom burden when compared to PwMS with concordant VAS and EQ-5D assessments.

[Figure omitted. See PDF.]

Green dots indicate individuals with low EQ-5D and high VAS. Red dots indicate individuals with high EQ-5D and low VAS.

[Figure omitted. See PDF.]

The multivariable regression analysis (S4 Table) identified only a limited number of associations with AIC decreases of at least two points, namely the number of symptoms and SRDSS. The high EQ-5D–low VAS group was less likely to have a SRDSS of 7 or more (0.13 [0.02–0.97]) or 7 or more symptoms (0.49 [0.25–0.95]). The low EQ-5D–high VAS group was associated with an SRDSS of 7 or more (6.78 [3.74–12.28]).

Discussion

Our analysis of 1679 PwMS illustrates the potential of distributional segmentation approaches to investigate reasons for low HRQoL. Removing the confounding effects of age and time since diagnosis, HRQoL was still predominantly associated with disease burden, characterized by gait impairment, total symptom load, or the presence of specific MS symptoms. Faster disease progression is an additional likely contributor to low HRQoL. Disease burden appeared to be more important than other characteristics, including sociodemographics, risk factors for MS progression, or comorbidities, which showed no associations in multivariable regression models.

Combined, our quantile-regression based segmentation with subsequent comparisons by HRQoL groups largely confirmed previous findings. By comparison, the sensitivity analysis based on simultaneous quantile regression identified fewer relevant factors and are more difficult to interpret for lay persons. Overall, the segmentation method offers a more intuitive approach to examine population- and health characteristics associated with low or high HRQoL. It also enables a simpler, direct comparison of two different HRQoL measures, as highlighted by our discordance analysis of EQ-5D and VAS indices.

Our main EQ-5D regression model suggests that this HRQoL measure correlates well with the EQ-5D subdomains of gait, pain, depression, and fatigue. However, the EQ-5D is generally not very responsive to small changes. By contrast, the VAS indicator does not focus on specific physical functions and has greater immediacy–VAS fluctuates on much smaller time scales and is influenced by a wide range of factors, including non-health-related conditions, such as weather [15]. It is, therefore, not entirely surprising that the VAS regression model generally exhibited weaker associations with disease burden while confirming the notable influences of depression, fatigue, pain, and muscle weakness on HRQoL. Additionally, our analysis of individual-level discordant VAS-EQ-5D assessments revealed interesting data pointing to the potential for further investigation. Notably, PwMS with a high ED-5D and low VAS tended to be younger, having a diagnosis for less long, and having a lower symptom burden than PwMS with concordant VAS and EQ-5D assessments. This finding points to a psychological component of well-being and ongoing coping and life-adjustment processes with MS [16]. This conclusion is corroborated by observations from the low EQ-5D–high VAS group, which tended to be more severely affected. EQ-5D tends to be relatively stable and less responsive to smaller changes over time or to changes in health-aspects that are not covered by the five domains. By contrast, VAS is a very general and subjective assessment of current HRQoL, that is influenced by a multitude of factors and exhibits much broader fluctuations. It has also been described previously that personal HRQoL assessments undergo baseline shifts as people grow older or increasingly cope with functional impairments [17]; that is, perceived general HRQoL as measured by VAS may increase or remain stable even though the structured, symptom-driven EQ-5D decreases due to increasing physical impairments. Combined, our observations suggest two conclusions: first, VAS assessments of persons with more severe, permanent physical impairments may undergo re-adjustments, with reference points for good HRQoL increasingly considering remaining physical functioning. Second, the analysis illustrates that low VAS assessments may not necessarily be tied to physical ailments, they can also be influenced by more immediate personal circumstances (e.g. stress) or mental well-being. Therefore, our method for identifying PwMS discordant EQ-5D–VAS self-assessments may help to identify persons in need of psychological or other support, but also highlights the need for closing gaps on health-related stressors not tied to MS symptoms or physical impairment.

Our findings generally align with existing literature on factors influencing HRQoL, including the detrimental effects of physical MS symptoms and impairment or disease progression. Contrary to other studies, however, our analysis did not indicate fatigue as a major independent contributor to low HRQoL. Other known factors, such as the role of comorbidities [4], could also not be corroborated by our study, possibly because of sample size limitations.

From a methodological viewpoint, our analysis illustrates the possible advantages of our distribution-based segmentation approach by showcasing its ease of interpretation and potential clinical applicability when compared to other quantile regression-based methods. MS management requires individuals with unexpectedly low HRQoL to be easily and reliably identified. Our method suggests a way forward but requires further optimization, for example, regarding optimal quantile boundaries. Unlike other studies investigating HRQoL trajectories, our analysis only included a cross-section. The quantile-based segmentation has the potential for longitudinal application, for example, by analyzing HRQoL change rates or repeated, intra-individual segment classifications.

Our findings also have implications for clinical practice, as factors associated with low HRQoL included symptoms like depression, fatigue, memory problems, muscle weakness, and pain, and high-level evidence supports non-pharmacological and pharmacological treatment options for symptomatic management of depression [18, 19], pain [20, 21], and fatigue [22] in MS. These “invisible” symptoms are quite frequent among persons with MS and exert a substantial burden on the lives of persons with MS [5]. They also pose significant job-related challenges and are associated with workforce drop-out [23]. Furthermore, because low HRQoL was found to be associated with depression, this indicator may also signal mental healthcare needs. Therefore, a more systematic HRQoL monitoring of health-related quality of life by means of the simple and low-effort EQ-5D may complement existing neurological assessments. Future studies should further examine the usefulness of monitoring for low HRQoL for routine care in longitudinal studies, for example, by investigating specific symptom burden changes associated with HRQoL decreases.

Our study is not without limitations. It should be emphasized that our study is unable to demonstrate (temporal) causality owing to its cross-sectional nature. Also, our study population was self-enrolled, and the majority of information is self-reported. Therefore, the effects of self-selection and reporting biases (e.g., social desirability biases) cannot be fully excluded. Furthermore, some factors (e.g., comorbidities) were only present in limited numbers, thus restricting the statistical power. Therefore, our findings require validation in different databases, as well as by including alternate measures of HRQoL, such as the 36-Item Short Form Survey (SF-36). Conversely, we consider the study population diversity, enabled by broad enrollment criteria, and the systematic approach of regression model building and co-variable specification as strengths. The distribution-based segmentation also requires no preconception of “good” or “bad” HRQoL while offering flexibility for further refinements. Furthermore, the method is easy to implement in any standard statistical software.

To summarize, our study found that low health-related quality of life was closely associated with physical impairment and specific MS symptoms, some of which are potentially mitigable. Depression, fatigue, memory problems, muscle weakness, and pain consistently emerged as symptoms with detrimental effects on HRQoL. Pharmacological and non-pharmacological MS symptom management, especially for depression, fatigue, pain, and muscle weakness, may warrant increased attention to preserve or improve HRQoL.

Supporting information

S1 Fig. Quantile-regression based HRQoL segmentation for the EQ-5D index.

https://doi.org/10.1371/journal.pone.0312486.s001

(DOCX)

S2 Fig. Correlation between age and time since MS diagnosis.

https://doi.org/10.1371/journal.pone.0312486.s002

(DOCX)

S1 Table. Selection of variables of interest and rationale for their inclusion.

https://doi.org/10.1371/journal.pone.0312486.s003

(DOCX)

S2 Table. Sensitivity analysis–simultaneous quantile regression of EQ-5D.

https://doi.org/10.1371/journal.pone.0312486.s004

(DOCX)

S3 Table. Sensitivity analysis–simultaneous quantile regression of VAS.

https://doi.org/10.1371/journal.pone.0312486.s005

(DOCX)

S4 Table. Multivariable multinomial regression to identify associations with discordant EQ-5D - VAS assessments.

https://doi.org/10.1371/journal.pone.0312486.s006

(DOCX)

Acknowledgments

The authors would like to thank the Swiss Multiple Sclerosis Society for supporting and funding the Swiss MS Registry. Moreover, we thank all participants who contributed to the Swiss MS Registry. We further thank the MS Advisory Board and SMSR employees.

MS Advisory Board Working Group “Swiss Multiple Sclerosis Registry” Members are: Claudia Baum, Pasquale Calabrese, Andrew Chan, Renaud Du Pasquier, Claudio Gobbi, Roger Häussler, Benjamin Victor Ineichen, Christian P. Kamm, Susanne Kägi, Jürg Kesselring (President), Kens Kuhle (Chair of Clinical and Laboratory Research Committee), Roland Kurmann, Christoph Lotter, Kurt Luyckx, Zina-Mary Manjaly, Patricia Monin, Stephanie Müller, Krassen Nedeltchev, Caroline Pot, Milo A. Puhan, Irene Rapold, Patrick Roth, Anke Salmen, Regula Steinlin Egli, Klaas Enno Stephan, Claude Vaney (Chair of Patient- and Population Research Committee), Viktor von Wyl (Chair of IT and Data Committee), Chiara Zecca.

References

1. 1. The Multiple Sclerosis International Federation (MSIF). Atlas of MS 3rd edition. 2020 Sep;

2. 2. Reich DS, Lucchinetti CF, Calabresi PA. Multiple Sclerosis. Longo DL, editor. N Engl J Med. 2018 Jan 11;378(2):169–80. pmid:29320652

* View Article

* PubMed/NCBI

* Google Scholar

3. 3. Marrie RA, Horwitz R, Cutter G, Tyry T. Cumulative impact of comorbidity on quality of life in MS: Comorbidity and quality of life in MS. Acta Neurologica Scandinavica. 2012 Mar;125(3):180–6.

* View Article

* Google Scholar

4. 4. Berrigan LI, Fisk JD, Patten SB, Tremlett H, Wolfson C, Warren S, et al. Health-related quality of life in multiple sclerosis: Direct and indirect effects of comorbidity. Neurology. 2016 Apr 12;86(15):1417–24. pmid:26962068

* View Article

* PubMed/NCBI

* Google Scholar

5. 5. Barin L, Salmen A, Disanto G, Babačić H, Calabrese P, Chan A, et al. The disease burden of Multiple Sclerosis from the individual and population perspective: Which symptoms matter most? Multiple Sclerosis and Related Disorders. 2018 Oct;25:112–21. pmid:30059895

* View Article

* PubMed/NCBI

* Google Scholar

6. 6. O’Mahony J, Salter A, Ciftci-Kavaklioglu B, Fox RJ, Cutter GR, Marrie RA. Physical and Mental Health–Related Quality of Life Trajectories Among People With Multiple Sclerosis. Neurology [Internet]. 2022 Oct 4 [cited 2024 Sep 27];99(14). Available from: https://www.neurology.org/doi/10.1212/WNL.0000000000200931 pmid:35948450

* View Article

* PubMed/NCBI

* Google Scholar

7. 7. LaRocca NG. Impact of Walking Impairment in Multiple Sclerosis: Perspectives of Patients and Care Partners. The Patient: Patient-Centered Outcomes Research. 2011 Sep;4(3):189–201. pmid:21766914

* View Article

* PubMed/NCBI

* Google Scholar

8. 8. Herdman M, Gudex C, Lloyd A, Janssen Mf, Kind P, Parkin D, et al. Development and preliminary testing of the new five-level version of EQ-5D (EQ-5D-5L). Qual Life Res. 2011 Dec;20(10):1727–36. pmid:21479777

* View Article

* PubMed/NCBI

* Google Scholar

9. 9. Steinemann N, Kuhle J, Calabrese P, Kesselring J, Disanto G, Merkler D, et al. The Swiss Multiple Sclerosis Registry (SMSR): study protocol of a participatory, nationwide registry to promote epidemiological and patient-centered MS research. BMC Neurol. 2018 Dec;18(1):111. pmid:30103695

* View Article

* PubMed/NCBI

* Google Scholar

10. 10. Disanto G, Benkert P, Lorscheider J, Mueller S, Vehoff J, Zecca C, et al. The Swiss Multiple Sclerosis Cohort-Study (SMSC): A Prospective Swiss Wide Investigation of Key Phases in Disease Evolution and New Treatment Options. Jacobson S, editor. PLoS ONE. 2016 Mar 31;11(3):e0152347. pmid:27032105

* View Article

* PubMed/NCBI

* Google Scholar

11. 11. Stanikić M, Braun J, Ajdacic-Gross V, Manjaly ZM, Yaldizli Ö, Ineichen BV, et al. Self-reports by persons with multiple sclerosis are an adequate surrogate for healthcare provider data on disease-modifying therapy and multiple sclerosis type. Multiple Sclerosis and Related Disorders. 2023 Dec;80:105097. pmid:39491413

* View Article

* PubMed/NCBI

* Google Scholar

12. 12. Perneger TV, Combescure C, Courvoisier DS. General Population Reference Values for the French Version of the EuroQol EQ-5D Health Utility Instrument. Value in Health. 2010 Jul;13(5):631–5. pmid:20412541

* View Article

* PubMed/NCBI

* Google Scholar

13. 13. Roxburgh RHSR, Seaman SR, Masterman T, Hensiek AE, Sawcer SJ, Vukusic S, et al. Multiple Sclerosis Severity Score: Using disability and disease duration to rate disease severity. Neurology. 2005 Apr 12;64(7):1144–51. pmid:15824338

* View Article

* PubMed/NCBI

* Google Scholar

14. 14. Aheto JMK. Simultaneous quantile regression and determinants of under-five severe chronic malnutrition in Ghana. BMC Public Health. 2020 Dec;20(1):644. pmid:32380990

* View Article

* PubMed/NCBI

* Google Scholar

15. 15. Blome C, Carlton J, Heesen C, Janssen MF, Lloyd A, Otten M, et al. How to measure fluctuating impairments in people with MS: development of an ambulatory assessment version of the EQ-5D-5L in an exploratory study. Qual Life Res. 2021 Jul;30(7):2081–96. pmid:33710593

* View Article

* PubMed/NCBI

* Google Scholar

16. 16. McCabe MP, McKern S. Quality of Life and Multiple Sclerosis: Comparison Between People with Multiple Sclerosis and People from the General Population. Journal of Clinical Psychology in Medical Settings. 2002;9(4):287–95.

* View Article

* Google Scholar

17. 17. Sprangers MAG, Schwartz CE. Integrating response shift into health-related quality of life research: a theoretical model. Social Science & Medicine. 1999 Jun;48(11):1507–15. pmid:10400253

* View Article

* PubMed/NCBI

* Google Scholar

18. 18. Fiest KM, Walker JR, Bernstein CN, Graff LA, Zarychanski R, Abou-Setta AM, et al. Systematic review and meta-analysis of interventions for depression and anxiety in persons with multiple sclerosis. Multiple Sclerosis and Related Disorders. 2016 Jan;5:12–26. pmid:26856938

* View Article

* PubMed/NCBI

* Google Scholar

19. 19. Koch MW, Glazenborg A, Uyttenboogaart M, Mostert J, De Keyser J. Pharmacologic treatment of depression in multiple sclerosis. Cochrane Multiple Sclerosis and Rare Diseases of the CNS Group, editor. Cochrane Database of Systematic Reviews [Internet]. 2011 Feb 16 [cited 2024 Sep 27]; Available from: https://doi.wiley.com/10.1002/14651858.CD007295.pub2 pmid:21328292

* View Article

* PubMed/NCBI

* Google Scholar

20. 20. Kannan P, Bello UM, Winser SJ. Physiotherapy interventions may relieve pain in individuals with central neuropathic pain: a systematic review and meta-analysis of randomised controlled trials. Therapeutic Advances in Chronic Disease. 2022 Jan;13:204062232210786. pmid:35356293

* View Article

* PubMed/NCBI

* Google Scholar

21. 21. Zhang YH, Hu HY, Xiong YC, Peng C, Hu L, Kong YZ, et al. Exercise for Neuropathic Pain: A Systematic Review and Expert Consensus. Front Med. 2021 Nov 24;8:756940. pmid:34901069

* View Article

* PubMed/NCBI

* Google Scholar

22. 22. Salarvand S, Heidari ME, Farahi K, Teymuri E, Almasian M, Bitaraf S. Effectiveness of massage therapy on fatigue and pain in patients with multiple sclerosis: A systematic review and meta-analysis. Multiple Sclerosis Journal ‐ Experimental, Translational and Clinical. 2021 Apr;7(2):205521732110227. pmid:34188950

* View Article

* PubMed/NCBI

* Google Scholar

23. 23. Coyne KS, Boscoe AN, Currie BM, Landrian AS, Wandstrat TL. Understanding Drivers of Employment Changes in a Multiple Sclerosis Population. International Journal of MS Care. 2015 Sep 1;17(5):245–52. pmid:26472946

* View Article

* PubMed/NCBI

* Google Scholar

Citation: Iaquinto S, Ineichen BV, Salmen A, Kuhle J, Benkert P, Hofer L, et al. (2024) Factors associated with low health-related quality of life in persons with multiple sclerosis: A quantile-based segmentation approach. PLoS ONE 19(11): e0312486. https://doi.org/10.1371/journal.pone.0312486

About the Authors:

Stefania Iaquinto

Roles: Methodology, Visualization, Writing – original draft, Writing – review & editing

Affiliation: Epidemiology, Biostatistics and Prevention Institute, University of Zurich, Zurich, Switzerland

ORICD: https://orcid.org/0000-0001-5808-9026

Benjamin Victor Ineichen