Подтверждение того, что качество жизни, связанное со здоровьем, ухудшается для мониторинга долгосрочного прогресса в борьбе с заболеваниями, связанными с MS: продольный анализ из швейцарского регистра множественного склероза.

Introduction

It remains challenging to monitor or predict the disease course of multiple sclerosis (MS). Presently, monitoring outcomes mainly focuses on relapses, the detection of brain lesions through magnetic resonance imaging (MRI), disability assessment, strongly influenced by ambulatory impairments, using the Expanded Disability Status Scale (EDSS), and, more recently, neurofilament light chain (NfL) as a biomarker for disease activity. These endpoints are closely linked to the traditional model of disease progression in relapsing–remitting MS (RRMS), suggesting that progression results from relapse‐associated worsening (RAW) due to relapse‐related inflammation and incomplete recovery [ 1 ]. However, this model has come under scrutiny with the availability of highly potent anti‐inflammatory disease‐modifying therapies (DMTs), leading to progression independent of relapse activity (PIRA) in RRMS patients [ 2 , 3 ]. Studies have shown that conventional outcome measures may fail to capture nuanced, cumulative effects of hidden symptoms such as pain, fatigue, or mental health in treated persons with PIRA, but also in untreated persons with progressive MS forms [ 4 , 5 ].

To better identify disease progression regardless of relapse activity, studies may complement EDSS measures with patient‐reported outcomes and lived experiences, including symptom assessments and health‐related quality of life (HRQoL) measures [ 5 ]. Yet, longitudinal analyses of such outcomes are relatively rare and often rely on complex analytical techniques. For example, two recent analyses from NARCOMS [ 6 , 7 ] utilized latent growth models to identify groups of persons with MS (pwMS) with similar HRQoL trajectories or symptom clusters. One study found that older age at onset, higher physical impairments, or fatigue were associated with a less favorable physical HRQoL trajectory [ 7 ]. The other study identified symptom patterns associated with less gait impairment, suggesting a more favorable trajectory [ 6 ]. Nevertheless, a gap in longitudinal investigations of patient‐reported outcomes with demonstrated added value for clinical decision‐making persists.

Our study aimed to address this gap by developing and testing an algorithm for detecting HRQoL‐based disease worsening, inspired by clinically established criteria for confirmed EDSS worsening [ 8 ]. Using longitudinal HRQoL measures from the Swiss Multiple Sclerosis Registry (SMSR), we compared the timing and frequency of HRQoL worsening against EDSS worsening and explored clinical factors and changes in disease burden associated with confirmed HRQoL worsening.

Methods

Study Population

Main Database

This study used longitudinal self‐reported data from the SMSR [ 9 , 10 ], a nationwide study that continuously enrolls adults with a confirmed diagnosis of clinically isolated syndrome (CIS) or MS receiving care in Switzerland. Launched in 2016, it currently includes 2900 pwMS overall and 2300 participants contributing to the longitudinal study (May 2025). Before the baseline questionnaire, prospective participants complete an informed consent form, including opt‐in agreements for medical record release or data linkage with the Swiss Multiple Sclerosis Cohort Study (SMSC) [ 11 ]. Variables extracted from the SMSR included sociodemographics, HRQoL assessments, MS disease characteristics, MS treatments, MS symptoms, disability status, chronic comorbidities, and familial MS history. The data were accessed on September 25, 2023, and included 2345 pwMS with at least one EuroQol‐5 Dimension (EQ‐5D) assessment and complete information on age, sex, and date of MS diagnosis. This full sample was used for the initial analysis, while specific analyses were conducted on subsamples of participants with at least three EQ‐5D self‐assessments (baseline and two follow‐ups) and those whose baseline EQ‐5D score fell within the top three quartiles (Figure  1 ).

Flowchart of the study population. EDSS, Expanded Disability Status Scale; EQ‐5D, EuroQol‐5 Dimension; SMSR, Swiss Multiple Sclerosis Registry.

Sources for EDSS Data

In addition to patient‐reported EDSS scores, our study included EDSS scores extracted from medical records by SMSR personnel and from two external sources: the SMSC and the database for joint tasks of health insurers (SVK database). The SMSC is an ongoing cohort study, integrated into routine neurological care that collects clinical data, disability assessments, MRIs, and biomarkers on‐site at eight large MS centers in Switzerland. As of January 12, 2023, approximately 1600 participants with a median follow‐up of 5.7 years have been enrolled. About 250 individuals have given signed informed consent for joint data linkage with the SMSR.

The SVK database contains administrative and clinical information, including EDSS, for cost approvals on behalf of specific Swiss health insurers, encompassing data from approximately 60% of all Swiss pwMS [ 12 ]. Certified neurologists submit cost reimbursement applications annually or more frequently if there are therapy changes. A probabilistic linkage between the SMSR population and the SVK database yielded an overlap of 1021 individuals. Details on the linkage process are provided elsewhere [ 13 ].

Ethics

The SMSR and SMSC have been approved by the 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, BASEC‐NR 2025–00159).

Outcome Measures

The main study outcomes are EQ‐5D‐5L assessments and EDSS scores. The EQ‐5D includes five questions on different health domains (mobility, self‐care, usual activities, pain/discomfort, and anxiety/depression) that convert into an index ranging from 0 (worst HRQoL) to 100 (best HRQoL) using specific value sets. In this study, we applied the French value set, which was found to be appropriate for Switzerland [ 14 ]. Following the method described in our earlier study [ 15 ], we utilized age‐ and disease‐duration‐adjusted quantile regression to estimate the 25th (lowest), median, and 75th (highest) percentiles of the EQ‐5D distribution at baseline. The use of EQ‐5D quantiles intends to mitigate three specific challenges. First, country‐specific value sets for EQ‐5D were not developed and validated for pwMS. Second, no established minimal clinically important difference (MCID) exists for changes of EQ‐5D in the context of MS progression. Third, EQ‐5D index values tend to change with increasing age (even in persons without MS), as well as with increasing disease‐duration. Our approach mitigates these challenges by taking the EQ‐5D index distribution from a population of pwMS as a reference. Furthermore, since our quantiles are calibrated for age‐ and disease‐duration, our approach is able to filter out changes that are predominantly age‐related.

The age‐ and disease‐duration adjusted percentiles were then used to classify all longitudinal EQ‐5D index measurements into quartiles. To detect confirmed worsening, individual‐level sequences of EQ‐5D index percentiles were subjected to an algorithm defined as a minimal, pre‐specified change from baseline confirmed in the following assessment. For the EQ‐5D index, confirmed worsening was determined by two consecutive EQ‐5D index follow‐up measures captured at least 180 days apart that fell within the 25th percentile (lowest quartile), with no evidence of a rebound to higher percentiles after confirmed EQ‐5D worsening. Persons with an EQ‐5D index in the 25th percentile at baseline were excluded from this analysis. In sensitivity analyses, we alternatively define worsening by a confirmed EQ‐5D index drop of 6, 8, 10, and 12 points from baseline.

EDSS worsening was determined using established algorithms, defined as a confirmed increase of 1 unit for participants with a baseline EDSS ≤ 5 or a confirmed increase of 0.5 units for participants with a baseline EDSS ≥ 5.5 [ 16 ].

Confounders and Variables of Interest

The quantile regression model for estimating EQ‐5D quartiles was adjusted for age‐ and disease‐duration. All other multivariable regression models were adjusted for age, sex (female/male), clinical MS phenotype (CIS, RRMS, primary/secondary progressive MS), and use of walking aids (cane, crutches, rollator, wheelchair) at baseline. The multivariable Cox regression models additionally included previously described disease course predictors, including specific first MS symptoms (binary variables [yes/no], prognostic for a severe disease course), education level, familial MS history, history of mononucleosis, smoking, and body mass index (BMI), all measured at baseline. Variable selection followed a manual approach based on a predefined set of confounders and explanatory variables. Additional variables were considered for inclusion using an AIC‐guided process, with a threshold of 2 score points for retention in the final model.

Symptom burden was analyzed using 12 self‐reported symptoms (balance problems, bladder issues, depression, fatigue, gait disturbances, gastrointestinal problems, speech problems, pain, sexual disturbances, spasms, tremor, and muscle weakness) captured in all follow‐ups. Participants reported whether they had experienced each symptom in the past 6 months (binary [yes/no]). These 12 symptoms were preselected from 20 symptoms based on their impact, frequency, and relevance to maintain a concise analysis [ 4 ].

Statistical Methods

We performed descriptive analyses of sociodemographic and MS‐specific variables stratified by EQ‐5D or EDSS worsening status (yes/no) at two time points: baseline (first survey with a valid EQ‐5D or EDSS measure) and either at the time of worsening confirmation for participants with confirmed worsening or at the last follow‐up for those without confirmed worsening. Continuous variables were summarized as medians and interquartile ranges (IQR), and categorical data as numbers and percentages. Time to confirmed worsening was analyzed separately for EQ‐5D and EDSS worsening status using Kaplan–Meier plots and multivariable Cox regression. In both analyses, time‐to‐event data were censored at the time of disease progression confirmation (for participants who experienced worsening) or at the last available follow‐up (for those who did not).

We analyzed the presence and changes in the frequency of 12 self‐reported symptoms stratified by EQ‐5D and EDSS worsening status. Symptom frequencies were assessed at the time of EQ‐5D or EDSS worsening for participants with worsening and at the last available assessment for those without worsening. Additionally, we examined the occurrence of new symptoms not present at baseline. Descriptive analyses were stratified by confirmed worsening status, with the results visualized in radar plots. Furthermore, we used confounder‐adjusted multivariable logistic regression to investigate associations between individual symptoms and worsening status, separately for EQ‐5D and EDSS. Each model was adjusted for the confounders described in the previous section and simultaneously included 12 symptoms for EQ‐5D and 10 symptoms for EDSS. Variance Inflation Factors (VIFs) indicated no multicollinearity issues (all VIFs for symptoms < 2.3).

We considered alternative worsening definitions for sensitivity analyses, including confirmed minimal clinically important differences (MCID) of 6, 8, 10, or 12‐point EQ‐5D index drops from baseline.

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

Results

The sample selection process is illustrated in Figure  1 . Based on 2345 persons with at least one EQ‐5D measure, we created age‐ and disease‐duration adjusted quartiles (Table  1 , column 1). For the longitudinal analysis, we identified 1087 persons with at least three eligible EQ‐5D measures (Table  1 , column 2). Interested in identifying persons with new instances of confirmed EQ‐5D worsening, we excluded 225 persons with baseline EQ‐5D in the lowest quartile. Therefore, the EQ‐5D worsening analysis included 862 pwMS (Table  1 , column 3). The EDSS analysis included 188 persons with at least three EDSS assessments (Table  1 , column 4).

TABLE 1 Characteristics of analysis samples at the time of enrollment in the Swiss MS Registry. Quantile development Full longitudinal sample Subsample excluding lowest quartile Subsample with EDSS Number 2345 (100) 1087 (100) 862 (100) 188 (100) Age, median [IQR] 47.0 [37.0; 55.0] 48.0 [38.0; 56.0] 48.0 [39.0; 56.0] 45.0 [37.5; 51.0] Years since diagnosis, median [IQR] 8.0 [2.0; 16.0] 7.0 [2.0; 16.0] 7.0 [2.0; 15.0] 7.0 [2.0; 12.0] Age at diagnosis, median [IQR] 36.0 [28.0; 45.0] 38.0 [30.0; 45.0] 38.0 [30.0; 46.0] 38.0 [29.5; 43.0] Sex Female 1717 (73.2) 798 (73.4) 640 (74.2) 144 (76.6) Male 628 (26.8) 289 (26.6) 222 (25.8) 44 (23.4) MS type Clinically Isolated Syndrome (CIS) 55 (2.3) 22 (2.0) 20 (2.3) 3 (1.6) Primary progressive MS (PPMS) 257 (11.0) 109 (10.0) 77 (8.9) 11 (5.9) Relapsing remitting MS (RRMS) 1626 (69.3) 770 (70.8) 646 (74.9) 160 (85.1) Secondary progressive MS (SPMS) 352 (15.0) 177 (16.3) 112 (13.0) 13 (6.9) Transitional phase 33 (1.4) 9 (0.8) 7 (0.8) 1 (0.5) Missing 22 (0.9) 0 (0.0) 0 (0.0) 0 (0.0) EQ‐5D assessment at baseline EQ‐5D index, median [IQR] 89.9 [80.8; 95.8] 90.7 [82.9; 96.2] 93.3 [88.2; 97.8] 94.6 [90.7; 98.0] Visual Analogue Scale, median [IQR] 80.0 [58.0; 90.0] 80.0 [60.0; 90.0] 80.0 [70.0; 93.0] 90.0 [80.0; 96.0]
Note: Values are reported as numbers and percentages unless otherwise specified. Abbreviation: IQR: Interquartile range.

Characteristics of analysis samples at the time of enrollment in the Swiss MS Registry.

Note: Values are reported as numbers and percentages unless otherwise specified.

Abbreviation: IQR: Interquartile range.

Median age (47–48 years) and disease‐duration (7–8 years) were similar across the first three subsamples, while the EDSS subsample was slightly younger (median age 45 years). The proportion of females ranged from 73.2% to 74.2% in the first three subsamples but was somewhat higher in the EDSS subsample (76.6%, Table  1 , column 4). The MS type distribution shifted with each sample selection step: the proportion of RRMS increased from 69.3% (Table  1 , column 1) to 85.1% (Table  1 , column 4), while the proportion of SPMS (from 15.0% to 6.9%) and PPMS (from 11.0% to 5.9%) decreased markedly.

Timing and Factors Associated With Confirmed Worsening

The median [IQR] follow‐up time for the EQ‐5D sample was 5.1 years [3.6; 6.5], with a median of 5 [4; 6] EQ‐5D assessments (not shown). By year 4 after baseline, 4.5% [95% confidence interval (CI) 3.2%; 6.3%] had experienced an EQ‐5D worsening event, as indicated by the Kaplan–Meier analysis (Figure  2 ).

Kaplan–Meier curve of time to specific worsening events. EDSS, Expanded Disability Status Scale; EQ‐5D, EuroQol‐5 Dimension.

The EDSS subsample had a shorter follow‐up of 3.5 [2.1; 5.4] years, with a median of 5 [4;7] assessments. By year 4, the cumulative percentage of persons with EDSS worsening was 15.8% [95% CI 10.5%; 23.5%]. Since EDSS was collected more frequently (approximately 1.5 per year) than EQ‐5D (collected only annually), we conducted a sensitivity analysis using only one annual EDSS value ( N  = 151, 20 confirmed worsening events). The Kaplan–Meier analysis showed a similar pattern to the main analysis, with EDSS worsening preceding EQ‐5D failure (Figure  S1 ) and failure by different MCID thresholds (Figure  S2 , Table  S3 ). Median [IQR] follow‐up was 3.8 [2.8; 5.5] years, with a median of 4 [3; 6] assessments. By year 4, 15.7% [95% CI 10.0%; 24.0%] had experienced an EDSS worsening event.

We examined associations of baseline characteristics and time to EQ‐5D (Table  2 ) or EDSS worsening (Table  3 ) using descriptive statistics and multivariable Cox regression. The final Cox regression model included known predictors for faster disease worsening, including age, sex, MS type, smoking, BMI, family history of MS, needing walking aids (cane, crutches, rollator, wheelchair) at baseline, and type of first symptom (visual, gait, spasms). Factors associated with a faster time to EQ‐5D worsening (Table  2 ) included needing a rollator (Hazard Ratio 2.79 [95% CI 1.11; 7.03]) or wheelchair (3.23 [1.26; 8.30]) at baseline, or a higher BMI (1.09 [1.03; 1.15] per point increase). By contrast, EDSS worsening (Table  3 ) was associated with male sex (3.94 [1.48; 10.46]), having PPMS (13.30 [2.36; 74.96]) or SPMS (8.93 [2.10; 37.91]), or having experienced spasms as a first symptom (5.47 [1.22; 24.52]).

TABLE 2 Characteristics at analysis baseline by EQ‐5D worsening status (columns 1 and 2) and association with time to worsening event in multivariable Cox regression (column 3). No worsening EQ‐5D worsening HR [95% CI]
N  = 812
N  = 50 Sociodemographic characteristics Sex Female 605 (74.5) 35 (70.0) Reference Male 207 (25.5) 15 (30.0) 1.00 [0.52; 1.91] Age, median [IQR] 48.0 [39.0; 56.0] 49.0 [38.0; 58.0] 0.94 [0.92; 0.97] Swiss Citizenship 746 (91.9) 48 (96.0) Marital status Divorced 77/795 (9.7) 5/49 (10.2) Married 433/795 (54.5) 27/49 (55.1) In registered partnership 10/795 (1.3) 3/49 (6.1) Separated 14/795 (1.8) 1/49 (2.0) Unmarried 240/795 (30.2) 13/49 (26.5) Widowed 21/795 (2.6) 0/49 (0.0) Education level Mandatory schooling completed 24 (3.0) 4 (8.0) Apprenticeship 292 (36.0) 32 (64.0) Highschool 86 (10.6) 1 (2.0) Higher professional education 119 (14.7) 2 (4.0) University degree 259 (31.9) 10 (20.0) Other education 32 (3.9) 1 (2.0) Disability benefit status Has applied for disability insurance 27/792 (3.4) 6/48 (12.5) Does not receive disability insurance 567/792 (71.6) 23/48 (47.9) Does receive disability insurance 198/792 (25.0) 19/48 (39.6) MS disease characteristics Years since diagnosis, median [IQR] 8.0 [2.0; 15.5] 7.0 [2.0; 14.0] Age at onset, median [IQR] 38 [29; 45] 38 [32; 47] MS type Clinically Isolated Syndrome (CIS) 19/806 (2.4) 1 (2.0) Reference Relapsing remitting MS (RRMS) 612/806 (75.9) 28 (56.0) Transitional phase 4/806 (0.5) 0 (0.0) Secondary progressive MS (SPMS) 102/806 (12.7) 13 (26.0) 1.92 [0.80; 4.57] Primary progressive MS (PPMS) 69/806 (8.6) 8 (16.0) 1.95 [0.77; 4.94] Health‐related quality of life measurements EQ‐5D index, median [IQR] 94.0 [88.7; 97.8] 86.8 [78.7; 92.4] Visual Analogue Scale, median [IQR] 85.0 [74.0; 95.0] 62.5 [50.0; 75.0] Symptom burden Number of MS symptoms, median [IQR] 3 [1; 6] 6 [4; 9] Ambulatory impairments Uses a wheelchair 53/758 (7.0) 16/49 (32.7) 3.23 [1.26; 8.30] Uses a cane or crutches 100/758 (13.2) 16/49 (32.7) 1.78 [0.82; 3.87] Uses a rollator 28/758 (3.7) 12/49 (24.5) 2.79 [1.11; 7.03] Risk factors for MS onset and progression Reported other MS cases in bloodline relatives 120 (14.8) 7 (14.0) 0.98 [0.43; 2.22] First symptom: Visual disturbances 316 (38.9) 20 (40.0) First symptom: Paresthesia 485 (59.7) 20 (40.0) First symptom: Gait 222 (27.3) 21 (42.0) 1.81 [0.99; 3.30] First symptom: Spasms 58 (7.1) 4 (8.0) 1.13 [0.38; 3.38] Body Mass Index, N , median [IQR] 784; 23.6 [21.3; 26.5] 49; 25.0 [21.5; 29.7] 1.09 [1.03; 1.15] Smoking status Smoking: never 384/805 (47.7) 17 (34.0) Reference Smoking: past 285/805 (35.4) 22 (44.0) 1.69 [0.86; 3.32] Smoking: current 136/805 (16.9) 11 (22.0) 1.84 [0.84; 4.03] Self‐reported comorbidities Self‐reported mononucleosis 119/809 (14.7) 7 (14.0) 1.15 [0.50; 2.62]
Note: Values are reported as numbers and percentages unless otherwise specified. The denominator is suppressed in case of no missing data. Missing data were coded as zero in time‐to‐event analysis, except for body mass index (BMI), which was imputed using sex, age, self‐reported gait disability, diabetes type 1 or 2, or cardiovascular comorbidities. Abbreviations: CI, Confidence interval; HR, Hazards Ratio; IQR, Interquartile range.

Characteristics at analysis baseline by EQ‐5D worsening status (columns 1 and 2) and association with time to worsening event in multivariable Cox regression (column 3).

Note: Values are reported as numbers and percentages unless otherwise specified. The denominator is suppressed in case of no missing data. Missing data were coded as zero in time‐to‐event analysis, except for body mass index (BMI), which was imputed using sex, age, self‐reported gait disability, diabetes type 1 or 2, or cardiovascular comorbidities.

Abbreviations: CI, Confidence interval; HR, Hazards Ratio; IQR, Interquartile range.

TABLE 3 Characteristics at analysis baseline by EDSS worsening status (columns 1 and 2) and association with time to worsening event in multivariable Cox regression (column 3). No worsening EDSS worsening HR [95% CI]
N  = 164
N  = 24 Sociodemographic characteristics Sex Female 129 (78.7) 15 (62.5) Reference Male 35 (21.3) 9 (37.5) 3.94 [1.48; 10.46] Age, median [IQR] 45.0 [36.0; 51.0] 46.5 [38.0; 53.0] 0.96 [0.91; 1.02] Swiss Citizenship 151 (92.1) 22 (91.7) Marital status Divorced 8/162 (4.9) 1 (4.2) Married 94/162 (58.0) 10 (41.7) In registered partnership 1/162 (0.6) 2 (8.3) Separated 3/162 (1.9) 0 (0.0) Unmarried 51/162 (31.5) 10 (41.7) Widowed 5/162 (3.1) 1 (4.2) Education level Mandatory schooling completed 1 (0.6) 0 (0.0) Apprenticeship 62 (37.8) 9 (37.5) Highschool 20 (12.2) 3 (12.5) Higher professional education 30 (18.3) 7 (29.2) University degree 49 (29.9) 4 (16.7) Other education 2 (1.2) 1 (4.2) Disability benefit status Has applied for disability insurance 8/163 (4.9) 1 (4.2) Does not receive disability insurance 124/163 (76.1) 19 (79.2) Does receive disability insurance 31/163 (19.0) 4 (16.7) MS disease characteristics Years since MS diagnosis, median [IQR] 6.0 [2.0; 12.0] 4.5 [1.0; 14.0] Age at onset, median [IQR] 38.0 [29.0; 42.0] 36.5 [31.0; 45.0] MS type Clinically Isolated Syndrome (CIS) 3 (1.8) 0 (0.0) Reference Relapsing remitting MS (RRMS) 142 (86.6) 17 (70.8) Transitional phase 1 (0.6) 0 (0.0) Secondary progressive MS (SPMS) 9 (5.5) 4 (16.7) 8.93 [2.10; 37.91] Primary progressive MS (PPMS) 9 (5.5) 3 (12.5) 13.30 [2.36; 74.96] Health‐related quality of life measurements EQ‐5D index; N , median [IQR] 162; 95.0 [90.7; 98.0] 23; 94.0 [86.2; 98.0] Visual Analogue Scale, N , median [IQR] 162; 90.0 [80.0; 96.0] 23; 80.0 [61.0; 85.0] Symptom burden Number of MS symptoms, median [IQR] 3.0 [1.0; 5.0] 2.0 [1.0; 6.0] Ambulatory impairments Uses a wheelchair 7/149 (4.7) 2/22 (9.1) 5.18 [0.34; 78.42] Uses a cane or crutches 12/149 (8.1) 2/22 (9.1) 0.09 [0.01; 1.19] Uses a rollator 5/149 (3.4) 1/22 (4.5) 4.71 [0.33; 66.75] Risk factors for MS onset and progression Reported other MS cases in bloodline relatives 21 (12.8) 4 (16.7) 1.85 [0.55; 6.22] First symptom: Visual disturbances 64 (39.0) 13 (54.2) First symptom: Paresthesia 111 (67.7) 9 (37.5) First symptom: Gait 45 (27.4) 7 (29.2) 0.69 [0.23; 2.04] First symptom: Spasms 5 (3.0) 3 (12.5) 5.47 [1.22; 24.52] Body Mass Index, N , median [IQR] 157; 23.3 [21.4; 26.3] 23.2 [20.3; 26.3] 0.96 [0.86; 1.07] Smoking status Smoking: never 77/162 (47.5) 12 (50.0) Reference Smoking: past 55/162 (34.0) 2 (8.3) 0.15 [0.03; 0.84] Smoking: current 30/162 (18.5) 10 (41.7) 1.54 [0.55; 4.30] Self‐reported comorbidities Self‐reported mononucleosis 25/163 (15.3) 2 (8.3) 0.17 [0.03; 0.92]
Note: Values are represented as numbers and percentages unless otherwise specified. The denominator is suppressed in case of no missing data. Missing data were coded as zero in time‐to‐event analysis, except for body mass index (BMI), which was imputed using sex, age, self‐reported gait disability, diabetes type 1 or 2, or cardiovascular comorbidities. Abbreviations: CI, Confidence interval; HR, Hazards Ratio; IQR, Interquartile range.

Characteristics at analysis baseline by EDSS worsening status (columns 1 and 2) and association with time to worsening event in multivariable Cox regression (column 3).

Note: Values are represented as numbers and percentages unless otherwise specified. The denominator is suppressed in case of no missing data. Missing data were coded as zero in time‐to‐event analysis, except for body mass index (BMI), which was imputed using sex, age, self‐reported gait disability, diabetes type 1 or 2, or cardiovascular comorbidities.

Abbreviations: CI, Confidence interval; HR, Hazards Ratio; IQR, Interquartile range.

Changes in Symptom Burden Upon Confirmed Worsening

Figure  3 illustrates the presence of the 12 self‐reported symptoms in participants without EQ‐5D worsening ( N  = 812, blue), with EQ‐5D worsening ( N  = 50, maroon), and with EDSS worsening ( N  = 24, green). Nearly all symptoms were reported more frequently in the EQ‐5D worsening group than in participants without or with EDSS worsening. However, in the logistic regression model, only depression (Odds Ratio 2.90 [95% CI 1.18; 7.14]) and pain (3.01 [1.42; 6.36]) were independently associated with EQ‐5D worsening (Table  S1 , column 3). No independent symptom associations were observed for EDSS worsening (Table  S2 , column 3). Confirmed worsening by different MCID thresholds also suggested a more frequent presence of depression, whereas associations with other symptoms were inconsistent (Table  S4 ).

Radar plots of burden of specific MS symptoms. Symptom frequency (Panel A) or report of new symptom since baseline (Panel B), both assessed at the time of last available assessment in persons without EQ‐5D worsening (blue, n  = 812) or at the time of EQ‐5D (maroon; n  = 50) or EDSS worsening (green; n  = 24).

When investigating new symptom occurrence between baseline and time of worsening or at last assessment, several symptom changes were independently associated with EQ‐5D worsening, including depression (2.99 [1.12; 8.01]), pain (2.92 [1.28; 6.66]), and gastrointestinal symptoms (2.84 [1.24; 6.51]) (Table  S1 , column 6). However, no symptom associations were found for EDSS worsening (Table  S2 , column 6). Changes in pain and depression were also consistently associated with different MCID outcomes (Table  S4 ).

Discussion

Using longitudinal EQ‐5D measurements from 862 pwMS, we developed and tested an algorithm to monitor HRQoL worsening, building on previous work that identified pwMS with low HRQoL relative to peers of similar age‐ and disease‐duration. Our algorithm applies decision rules mirroring established EDSS worsening criteria, enhancing its clinical applicability.

We found that EQ‐5D worsening occurred at lower rates and later in patient follow‐up than EDSS worsening. However, symptom burden was notably higher upon EQ‐5D worsening, with depression and pain robustly reported more frequently—a plausible finding, given that the EQ‐5D assessment includes dimensions for pain and depression/anxiety, thus supporting the validity of our worsening algorithm. Other symptoms, such as sexual disturbances or gastrointestinal problems, were only flagged in some analyses, suggesting that they may be less reliable markers of EQ‐5D worsening. The strongest baseline predictors for disease worsening included pre‐existing impairments, more severe first symptom manifestations, and MS disease type. Additionally, we observed a weak association between BMI and EQ‐5D worsening, a finding which may point to possible modifiable lifestyle influences on disease trajectories. Overall, our time‐to‐event analysis of potential risk factors for EQ‐5D or EDSS worsening confirms the difficulty of predicting the MS disease course.

Our observations align with two other studies using the SMSR database to examine the impact of symptom burden on HRQoL. Barin et al. [ 4 ] performed a cross‐sectional analysis of self‐reported symptoms and created an impact ranking based on regression coefficients, highlighting independent negative effects on EQ‐5D. Similarly, a cross‐sectional analysis by our group [ 15 ], which developed the HRQoL classification approach, also observed that pwMS with low EQ‐5D reported pain and depression more frequently than their peers with higher EQ‐5D but similar age‐ and disease‐durations. Other studies explored different analysis techniques to examine factors associated with favorable or less favorable disease trajectories [ 7 , 17 ] or specific symptoms associated with lower HRQoL [ 18 ]. Unlike these studies, our longitudinal analysis showcases the usefulness of EQ‐5D‐based indicators for disease course monitoring. Given its widespread availability and low collection burden, our methodology and algorithm can potentially complement other established clinical outcome measures, such as EDSS. While EDSS strongly focuses on gait impairments, EQ‐5D has the advantage of highlighting two “invisible” but impactful MS symptoms: depression and pain. However, our analysis also highlights that EDSS may be more responsive to health changes and occur at higher rates than EQ‐5D worsening. Furthermore, our analytical approach does not replace existing patient‐reported instruments to monitor and diagnose clinically relevant changes in pain or depression. Therefore, EQ‐5D worsening monitoring may only be useful as a complementary indicator of invisible MS symptoms over longer timescales. Our publicly available EQ‐5D index distributions (provided as raw data, Table  S5 ) support these efforts. The calibration of our EQ‐5D worsening classification on a diverse population of pwMS and its reliance on quantiles likely makes our methodology robust to changes in EQ‐5D reference sets or other HRQoL instruments.

One strength of this study is its nesting within a diverse longitudinal cohort, largely representative of pwMS in Switzerland [ 12 ]. The SMSR, in collaboration with the national MS Society, undertakes substantial community‐building and communication efforts to ensure the long‐term engagement of study participants. Consequently, study participants are well documented, with many regularly completing the semi‐annual surveys, including the validated EQ‐5D surveys. Additionally, this study enhances the longitudinal utility of EQ‐5D, a widely easily collected measure, further supporting its potential for long‐term disease monitoring.

However, a limitation of our study is that most data are based on self‐reports, therefore representing the patient's perspective rather than the clinical view. Nevertheless, a recent validation analysis from the SMSR has shown that key clinical aspects, such as treatments, are reliably reported by pwMS [ 13 ], whereas slightly larger discrepancies were noted regarding MS disease types. Another constraint is the limited availability of EDSS information. The SMSR mainly relies on self‐reports, and previous experiences indicate that only 16% of pwMS know their EDSS score (unpublished data, SMSR analysis). Therefore, we needed to compile EDSS data from multiple sources, though the sample size for the EDSS worsening analysis was limited. Nevertheless, the EDSS worsening analyses primarily served as a comparison for the EQ‐5D worsening analysis, and the overall pattern of worsening frequency and timing remained robust in the sensitivity analysis. In addition, it is important to note that the EDSS and EQ‐5D subsamples differ in their distribution of MS types. However, given that we accounted for the MS type in our analyses, we believe our main conclusions remain unaffected by these differences.

To conclude, our analysis illustrates the potential usefulness of longitudinal HRQoL measurements to monitor additional aspects of disease worsening other than those covered by the EDSS. EQ‐5D worsening does not act as an early marker of clinical deterioration but highlights complementary dimensions of disease burden, such as symptoms impacting quality of life. Our methodological approach is not confined to EQ‐5D and can potentially be applied to continuous scores from other HRQoL instruments and other patient‐related outcome measures. Given the widespread use of HRQoL measures in many longitudinal MS studies, our methodology could provide valuable further insights into MS disease trajectories with minimal additional effort. This could, for example, contribute to a better understanding of the frequency, mechanisms, and outcomes of PIRA.

Author Contributions

Stefania Iaquinto: writing – original draft, validation, methodology, investigation. Benjamin Victor Ineichen: writing – review and editing. Mina Stanikić: writing – review and editing. Jens Kuhle: writing – review and editing. Christian Philipp Kamm: writing – review and editing. Pasquale Calabrese: writing – review and editing. Patrick Roth: writing – review and editing. Claudia Baum: writing – review and editing. Chiara Zecca: writing – review and editing. Claudio Gobbi: writing – review and editing. Özgür Yaldizli: writing – review and editing. Milo Alan Puhan: writing – review and editing. Viktor von Wyl: conceptualization, methodology, investigation, writing – original draft, formal analysis, visualization, validation, funding acquisition, project administration.

Conflicts of Interest

Stefania Iaquinto, Benjamin Victor Ineichen, Mina Stanikić, Christian Philipp Kamm, Pasquale Calabrese, Claudia Baum, Milo Alan Puhan, Viktor von Wyl report no disclosures relevant to this manuscript. Jens Kuhle received speaker fees, research support, travel support, and/or served on advisory boards by Swiss MS Society, Swiss National Research Foundation (320030_212534/1), University of Basel, Progressive MS Alliance, Alnylam, Bayer, Biogen, Bristol Myers Squibb, Celgene, Immunic, Merck, Neurogenesis, Novartis, Octave Bioscience, Quanterix, Roche, Sanofi, Stata DX. Patrick Roth has received honoraria for lectures or advisory board participation from Alexion, Bristol‐Myers Squibb, Boehringer Ingelheim, Debiopharm, Galapagos, Merck Sharp and Dohme, Laminar, Midatech Pharma, Novocure, QED, Roche, Sanofi and Servier and research support from Merck Sharp and Dohme and TME Pharma. Chiara Zecca Ente Ospedaliero Cantonale (employer) received compensation for Chiara Zecca's speaking activities, consulting fees, or grants from Abbvie, Alexion, Almirall, Biogen, Bristol Meyer Squibb, Eisai, Lilly, Lundbeck, Merck, Merz, Novartis, Organon, Pfizer, Sandoz, Sanofi, Teva Pharma, Roche. Chiara Zecca is recipient of a grant for senior reseachers provided by AFRI (Area Formazione accademica, Ricerca e Innovazione), EOC. Claudio Gobbi Ente Ospedaliero Cantonale (employer) received compensation for Claudio Gobbi's speaking activities, consulting fees, or grants from Abbvie, Almirall, Biogen, Bristol Meyer Squibb, Lundbeck, Merck, Novartis, Sandoz, Sanofi, Teva Pharma, and Roche. Özgür Yaldizli received grants from ECTRIMS/MAGNIMS, University of Basel, Pro Patient Stiftung, University Hospital Basel, Free Academic Society Basel, Swiss National Science Foundation and the Swiss Multiple Sclerosis Society and advisory board fees from Biogen.

Supporting information

Data S1.

Data S2.