Changes in the Risk of Reaching Multiple Sclerosis Disability Milestones In Recent Decades: A Nationwide Population-Based Cohort Study in Sweden | Demyelinating Disorders | JAMA Neurology | JAMA Network
[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure.  Censoring and Truncation in Longitudinal Analysis
Censoring and Truncation in Longitudinal Analysis

Possible scenarios and pitfalls of patient follow-up are shown in a time-to-event analysis. In time-to-event settings, risks (hazard) of reaching an outcome are compared between groups of patients over follow-up time. A distinguishing feature of this analysis is its ability to incorporate the effect of censoring or loss to follow-up in the estimation of hazard. Censoring occurs when information about an individual’s follow-up time is incomplete. Left censoring, which happens when the outcome occurs before start of follow-up, is a common problem in studies of multiple sclerosis as a proportion of individuals are excluded owing to the occurrence of outcome before the start of follow-up time. The main problem in our data is left-censored observation that indicates disability milestone had already happened.

Table 1.  Clinical and Demographic Characteristics of Included Patients
Clinical and Demographic Characteristics of Included Patients
Table 2.  Results of Weibull Model for Patients With ROMS and POMS by Disability Milestones of Interest
Results of Weibull Model for Patients With ROMS and POMS by Disability Milestones of Interest
Table 3.  Hazard Ratios and 95% CIs of Year of Diagnosis for Patients With Relapsing-Onset MS Adjusted for Time of Treatment
Hazard Ratios and 95% CIs of Year of Diagnosis for Patients With Relapsing-Onset MS Adjusted for Time of Treatment
1.
Steinvorth  SM, Rover  C, Schneider  S, Nicholas  R, Straube  S, Friede  T.  Explaining temporal trends in annualised relapse rates in placebo groups of randomised controlled trials in relapsing multiple sclerosis: systematic review and meta-regression.  Mult Scler. 2013;19(12):1580-1586. doi:10.1177/1352458513481009Google Scholar
2.
Röver  C, Nicholas  R, Straube  S, Friede  T.  Changing EDSS progression in placebo cohorts in relapsing MS: a systematic review and meta-regression.  PLoS One. 2015;10(9):e0137052. doi:10.1371/journal.pone.0137052PubMedGoogle Scholar
3.
Tremlett  H, Zhao  Y, Rieckmann  P, Hutchinson  M.  New perspectives in the natural history of multiple sclerosis.  Neurology. 2010;74(24):2004-2015. doi:10.1212/WNL.0b013e3181e3973fPubMedGoogle Scholar
4.
Weinshenker  BG, Bass  B, Rice  GP,  et al.  The natural history of multiple sclerosis: a geographically based study: I: clinical course and disability.  Brain. 1989;112(pt 1):133-146. doi:10.1093/brain/112.1.133PubMedGoogle Scholar
5.
Runmarker  B, Andersen  O.  Prognostic factors in a multiple sclerosis incidence cohort with twenty-five years of follow-up.  Brain. 1993;116(pt 1):117-134. doi:10.1093/brain/116.1.117PubMedGoogle Scholar
6.
Confavreux  C, Vukusic  S, Moreau  T, Adeleine  P.  Relapses and progression of disability in multiple sclerosis.  N Engl J Med. 2000;343(20):1430-1438. doi:10.1056/NEJM200011163432001PubMedGoogle Scholar
7.
Pittock  SJ, Mayr  WT, McClelland  RL,  et al.  Change in MS-related disability in a population-based cohort: a 10-year follow-up study.  Neurology. 2004;62(1):51-59. doi:10.1212/01.WNL.0000101724.93433.00PubMedGoogle Scholar
8.
Tremlett  H, Paty  D, Devonshire  V.  Disability progression in multiple sclerosis is slower than previously reported.  Neurology. 2006;66(2):172-177. doi:10.1212/01.wnl.0000194259.90286.fePubMedGoogle Scholar
9.
Brown  M, Andreou  P, Kirby  S, Murray  J. EDSS natural history atlas: MS disability progression in Nova Scotia 1979-2004. Presented at: ECTRIMS Conference; September 17-20, 2008; Montreal, Quebec, Canada.
10.
Kister  I, Chamot  E, Cutter  G,  et al; MSBase Investigators.  Increasing age at disability milestones among MS patients in the MSBase Registry.  J Neurol Sci. 2012;318(1-2):94-99. doi:10.1016/j.jns.2012.03.017PubMedGoogle Scholar
11.
Shirani  A, Zhao  Y, Kingwell  E, Rieckmann  P, Tremlett  H.  Temporal trends of disability progression in multiple sclerosis: findings from British Columbia, Canada (1975-2009).  Mult Scler. 2012;18(4):442-450. doi:10.1177/1352458511422097Google Scholar
12.
Pittock  SJ, Mayr  WT, McClelland  RL,  et al.  Disability profile of MS did not change over 10 years in a population-based prevalence cohort.  Neurology. 2004;62(4):601-606. doi:10.1212/WNL.62.4.601PubMedGoogle Scholar
13.
Manouchehrinia  A, Beiki  O, Hillert  J.  Clinical course of multiple sclerosis: a nationwide cohort study.  Mult Scler. 2017;23(11):1488-1495. doi:10.1177/1352458516681197Google Scholar
14.
Freilich  J, Manouchehrinia  A, Trusheim  M,  et al.  Characterization of annual disease progression of multiple sclerosis patients: a population-based study.  Mult Scler. 2018;24(6):786-794. doi:10.1177/1352458517706252Google Scholar
15.
Burkill  S, Montgomery  S, Hajiebrahimi  M, Hillert  J, Olsson  T, Bahmanyar  S.  Mortality trends for multiple sclerosis patients in Sweden from 1968 to 2012.  Neurology. 2017;89(6):555-562. doi:10.1212/WNL.0000000000004216PubMedGoogle Scholar
16.
Ahlgren  C, Odén  A, Lycke  J.  High nationwide prevalence of multiple sclerosis in Sweden.  Mult Scler. 2011;17(8):901-908. doi:10.1177/1352458511403794Google Scholar
17.
Ahlgren  C, Odén  A, Lycke  J.  High nationwide incidence of multiple sclerosis in Sweden.  PLoS One. 2014;9(9):e108599. doi:10.1371/journal.pone.0108599PubMedGoogle Scholar
18.
Hillert  J, Stawiarz  L.  The Swedish MS registry: clinical support tool and scientific resource.  Acta Neurol Scand. 2015;132(199):11-19. doi:10.1111/ane.12425PubMedGoogle Scholar
19.
Andersen  O.  From the Gothenburg cohort to the Swedish multiple sclerosis registry.  Acta Neurol Scand Suppl. 2012;(195):13-19. doi:10.1111/ane.12023PubMedGoogle Scholar
20.
Kurtzke  JF.  Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS).  Neurology. 1983;33(11):1444-1452. doi:10.1212/WNL.33.11.1444PubMedGoogle Scholar
21.
Collett  D.  Modelling Survival Data in Medical Research. 3rd ed. Boca Raton, Florida: Chapman and Hall/CRC; 2015.
22.
Kalbfleisch  JD, Prentice  RL.  The Statistical Analysis of Failure Time Data. 2nd ed. Hoboken, New Jersey: Wiley-Interscience; 2002.
23.
Zhang  Z, Sun  J.  Interval censoring.  Stat Methods Med Res. 2010;19(1):53-70. doi:10.1177/0962280209105023PubMedGoogle Scholar
24.
McAlpine  D, Compston  A. McAlpine's Multiple Sclerosis. 4th ed. London, United Kingdom: Churchill Livingstone; 2005.
25.
Kappos  L, Traboulsee  A, Constantinescu  C,  et al.  Long-term subcutaneous interferon beta-1a therapy in patients with relapsing-remitting MS.  Neurology. 2006;67(6):944-953. doi:10.1212/01.wnl.0000237994.95410.cePubMedGoogle Scholar
26.
Ford  C, Goodman  AD, Johnson  K,  et al.  Continuous long-term immunomodulatory therapy in relapsing multiple sclerosis: results from the 15-year analysis of the US prospective open-label study of glatiramer acetate.  Mult Scler. 2010;16(3):342-350. doi:10.1177/1352458509358088Google Scholar
27.
Ebers  G, Traboulsee  A, Langdon  D,  et al. The final results of the interferon beta-1b 16-year long-term follow-up study. https://ora.ox.ac.uk/objects/uuid:4f0c52f2-64a4-486e-a074-a22498bd236b. Accessed February 12, 2019.
28.
Trojano  M, Pellegrini  F, Paolicelli  D,  et al; Italian Multiple Sclerosis Database Network (MSDN) Group.  Real-life impact of early interferon beta therapy in relapsing multiple sclerosis.  Ann Neurol. 2009;66(4):513-520. doi:10.1002/ana.21757PubMedGoogle Scholar
29.
Veugelers  PJ, Fisk  JD, Brown  MG,  et al.  Disease progression among multiple sclerosis patients before and during a disease-modifying drug program: a longitudinal population-based evaluation.  Mult Scler. 2009;15(11):1286-1294. doi:10.1177/1352458509350307PubMedGoogle Scholar
30.
Kister  I, Chamot  E, Bacon  JH, Cutter  G, Herbert  J; New York State Multiple Sclerosis Consortium.  Trend for decreasing Multiple Sclerosis Severity Scores (MSSS) with increasing calendar year of enrollment into the New York State Multiple Sclerosis Consortium.  Mult Scler. 2011;17(6):725-733. doi:10.1177/1352458510396269Google Scholar
31.
Odell  PM, Anderson  KM, D’Agostino  RB.  Maximum likelihood estimation for interval-censored data using a Weibull-based accelerated failure time model.  Biometrics. 1992;48(3):951-959. doi:10.2307/2532360PubMedGoogle Scholar
32.
Coria  VH, Maximov  S, Rivas-Dávalos  F, Melchor-Hernández  CL.  Perturbative method for maximum likelihood estimation of the Weibull distribution parameters.  Springerplus. 2016;5(1):1802. doi:10.1186/s40064-016-3500-yPubMedGoogle Scholar
33.
Westerlind  H, Stawiarz  L, Fink  K, Hillert  J, Manouchehrinia  A. A significant decrease in diagnosis of primary progressive multiple sclerosis: a cohort study.  Mult Scler. 2016;22(8):1071-1079. doi:10.1177/1352458516643394Google Scholar
34.
D'Hooghe  MB, Nagels  G, Bissay  V, De Keyser  J.  Modifiable factors influencing relapses and disability in multiple sclerosis.  Mult Scler. 2010;16(7):773-785. doi:10.1177/1352458510367721Google Scholar
35.
Zivadinov  R, Zorzon  M, Weinstock-Guttman  B,  et al.  Epstein-Barr virus is associated with grey matter atrophy in multiple sclerosis.  J Neurol Neurosurg Psychiatry. 2009;80(6):620-625. doi:10.1136/jnnp.2008.154906PubMedGoogle Scholar
36.
Ginde  AA, Liu  MC, Camargo  CA  Jr.  Demographic differences and trends of vitamin D insufficiency in the US population, 1988-2004.  Arch Intern Med. 2009;169(6):626-632. doi:10.1001/archinternmed.2008.604PubMedGoogle Scholar
37.
Gyllensten  H, Wiberg  M, Alexanderson  K, Friberg  E, Hillert  J, Tinghög  P.  Comparing costs of illness of multiple sclerosis in three different years: a population-based study.  Mult Scler. 2018;24(4):520-528. doi:10.1177/1352458517702549Google Scholar
38.
Hobart  J, Freeman  J, Thompson  A.  Kurtzke scales revisited: the application of psychometric methods to clinical intuition.  Brain. 2000;123(Pt 5):1027-1040. doi:10.1093/brain/123.5.1027PubMedGoogle Scholar
39.
Kragt  JJ, Nielsen  JM, van der Linden  FA, Polman  CH, Uitdehaag  BM.  Disease progression in multiple sclerosis: combining physicians' and patients' perspectives?  Mult Scler. 2011;17(2):234-240. doi:10.1177/1352458510385505Google Scholar
40.
Ontaneda  D, Cohen  JA.  EDSS improvement: recovery of function or noise?  Mult Scler. 2012;18(11):1520-1521. doi:10.1177/1352458512441689Google Scholar
41.
Ebers  GC, Heigenhauser  L, Daumer  M, Lederer  C, Noseworthy  JH.  Disability as an outcome in MS clinical trials.  Neurology. 2008;71(9):624-631. doi:10.1212/01.wnl.0000313034.46883.16PubMedGoogle Scholar
Original Investigation
March 18, 2019

Changes in the Risk of Reaching Multiple Sclerosis Disability Milestones In Recent Decades: A Nationwide Population-Based Cohort Study in Sweden

Author Affiliations
  • 1Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden
  • 2Cognizant Technology Solutions, Stockholm, Sweden
  • 3Kermanshah University of Medical Sciences, Kermanshah, Iran
  • 4Institute of Environmental Medicine, Unit of Biostatistics, Karolinska Institutet, Stockholm, Sweden
JAMA Neurol. 2019;76(6):665-671. doi:10.1001/jamaneurol.2019.0330
Key Points

Question  Has the risk of reaching disability milestones in multiple sclerosis changed over the last decade?

Findings  In this nationwide population-based cohort study of 7331 patients with multiple sclerosis diagnosed between 1995 and 2010, a significant 3% decrease per calendar year of diagnosis for the risk of sustained Expanded Disability Status Scale score (EDSS) 3.0, a significant 6% decrease for the risk of EDSS 4.0, and a significant 7% decrease for the risk of EDSS 6.0 among patients with relapsing-onset multiple sclerosis was found.

Meaning  The risk of reaching disability milestones decreased significantly over the last decade in patients with relapsing-onset multiple sclerosis in Sweden.

Abstract

Importance  Clinicians’ experience and findings from recent natural history studies suggest that multiple sclerosis (MS) may now be running a more slowly progressing course than before.

Objective  To investigate whether the risk of reaching MS disability milestones has changed over the last decade in Sweden.

Design, Setting, and Participants  A nationwide population-based retrospective cohort study. By April 2017, 12 512 patients with available information on demographics, MS phenotype, and date of MS onset and diagnosis were registered in the Swedish MS Registry of which 7331 patients with at least 2 recorded Expanded Disability Status Scale scores (EDSS) and diagnosed between January 1995 and December 2010 were included. No further exclusion criteria were applied. Patients were followed up until December 2016 with a median duration follow-up of 8.5 (interquartile range, 4.7-13.8) years. Statistical analysis began in April 2017.

Main Outcomes and Measures  Patients were followed up from MS onset date to the date of sustained EDSS 3.0, 4.0, and 6.0. To handle interval-censored observations, a Weibull model was fit, and the change in the risk of EDSS 3.0, 4.0, and 6.0 over calendar years was estimated and hazard ratios (HRs) with corresponding CIs were calculated.

Results  Of 7331 patients, 5196 (70.9%) were women, and the mean (SD) age at diagnosis was 38.3 (11.7) years. Adjusting for sex, number of clinic visits, diagnostic delay, and onset age, a 3% decrease per calendar year of diagnosis for the risk of sustained EDSS 3.0 (HR, 0.97; 95% CI, 0.96-0.97), a 6% decrease for the risk of EDSS 4.0 (HR, 0.94; 95% CI, 0.93-0.95), and a 7% decrease for the risk of EDSS 6.0 (HR, 0.93; 95% CI, 0.91-0.94) among patients with relapsing-onset MS was found. The trends were not significant for patients with progressive-onset MS (EDSS 3.0: HR, 1.01; 95% CI, 0.98-1.03; EDSS 4.0: HR, 1.00; 95% CI, 0.98-1.02; EDSS 6.0: HR, 1.00; 95% CI, 0.98-1.02).

Conclusions and Relevance  Risk of reaching major disability milestones has significantly decreased over the last decade in patients with relapsing-onset MS in Sweden. Several factors could potentially be responsible for this observation. However, given that no change was seen in disability accrual of patients with progressive-onset MS and the absence of efficacious treatment option in this group, increased use of more efficacious disease-modifying treatments could be a possible driver of this change.

Introduction

Clinicians’ experience from practice, findings from placebo arms in randomized clinical trials (RCTs) in relapsing-onset multiple sclerosis (ROMS),1 and natural history studies suggest that the progression of multiple sclerosis (MS) may be slower than previously described.

A systematic review of RCTs in ROMS revealed a decrease in placebo annualized relapse rates during the past 2 decades.1 Researchers concluded that the observed decline may result, apart from increasing patient age and duration of illness, from decreasing pretrial annualized relapse rates and a shorter period over which pretrial annualized relapse rates were calculated. Disability outcomes in MS are a key component that regulators have identified as the principal target for an increasing range of therapies targeting the underlying disease process. Quiz Ref IDThe most commonly used method of measuring MS disability in trials is the Extended Disability Status Scale score (EDSS). A review of published RCT data suggested that disease progression, as measured by the EDSS, exhibits similar behavior as the annualized relapse rate and confirmed that in ROMS trial placebo groups, the rates of disease progression has slowed in recent years.2

Findings from natural history studies have been less consistent, perhaps owing to heterogeneity in methodological approach, study population, and period.3 Some recent natural history studies reported a trend of increasing time to disability milestones. The median time from onset to requiring a cane has increased from 15 to 20 years in studies published between 1989 and 20004-6 to around 30 years in later reports published between 2004 and 2008.7-9 A 2012 study showed a significant increase in patient age at enrollment in each mild to moderate disability strata in 20 countries contributing to the MSBase registry.10

Two natural history studies found no evidence of shifting trajectory of disability over time. In a study of 2236 patients with ROMS in British Columbia, Canada, the average rate of disease progression remained stable between 1975 and 2009, leading to a conclusion that differences in disease progression findings between natural history studies may be associated with factors other than time period.11 Similarly, in Minnesota, the distribution of MS disability remained largely unaltered between 1991 and 2000.12

Using population-based data from Sweden, we (1) found that the mean age at disability milestones is higher than previously described in clinic-based and regional-based samples,13 (2) confirmed established factors associated with MS disease worsening in time to disease milestones,14 and (3) found that there has been a gradual reduction in mortality risk over time among patients with MS.15 In this study, we aimed to investigate whether the risk of reaching disability milestones has changed over the last decade.

Methods
Patient Population

Sweden has a high prevalence (189 per 100 000 individuals) and incidence rate (10.2 per 100 000 person-years) of MS.16,17 We analyzed prospectively collected data from the Swedish MS Registry (SMSreg),18 a nationwide and population-based register that collates information from every neurology clinic in Sweden. The SMSreg officially started in 2000 as a national quality registry and contains clinical (including date of disease onset and diagnosis, disease course, and disability required for this study) and demographic data on about 18 500 patients with MS (as of April 2017). The SMSreg has been designed in cooperation with neurologists in the decentralized Swedish health care system and aims to assure quality health care for patients with MS. For many years, the SMSreg is used in all Swedish neurology clinics and covers around 80% of all prevalent cases of MS in the country.19

Through the SMSreg, we selected a nationwide population-based retrospective cohort of patients with at least 2 recorded EDSS scores of registered patients who were diagnosed between January 1995 and December 2010. By April 2017, 12 512 patients with available information on demographics, MS phenotype, and date of MS onset and diagnosis were registered in the SMSreg, of which 7331 patients with at least 2 recorded EDSS scores and registered patients who were diagnosed between January 1995 and December 2010 were included. No further exclusion criteria were applied. Analysis began in April 2017. This study was approved by the Swedish regional ethical review board in Stockholm. Consent to the sharing of data within the SMSreg is voluntary for patients and clinicians; written informed consent was received from all participating patients.18

Outcomes: EDSS Milestones 3.0, 4.0, and 6.0

The EDSS is the most widely used method of quantifying disability in MS. It ranges from 0 (no disability) to 10 (death due to MS) in 0.5-unit increments. Quiz Ref IDThe EDSS is scored prospectively by a neurologist and is based on the examination of 8 functional systems.20 We used sustained EDSS milestones 3.0 (moderate disability but no impairment of walking), 4.0 (significant disability but able to walk without aid or rest for 500 m), and 6.0 (requires unilateral assistance to walk about 100 m with or without resting) as our outcomes of interest. In the SMSreg, clinicians indicate whether an EDSS is measured during relapse or not. Using this variable, we excluded all relapse-related EDSS from our analyses.

Statistical Analysis

We followed up patients from the date of MS onset to the date of first sustained EDSS 3.0, 4.0, and 6.0, death, or last clinic visit date, whichever occurred first. The last day of follow-up was December 2016, which allowed at least 6 years of follow-up for patients diagnosed in 2010 (median [interquartile range] duration of follow-up: 8.5 [4.7-13.8] years). We used a Weibull model21,22 to evaluate the change in risk of reaching EDSS 3.0, 4.0, and 6.0 over time. We calculated hazard ratios (HRs) and corresponding 95% CIs by calendar year of diagnosis. The Weibull model is similar to a Cox model in which the baseline hazard, instead of being nonparametric, is assumed to be that of a Weibull distribution. Using Weibull models permitted handling interval-censored observations (Figure). All models were adjusted for sex, age at onset, and total number of clinic visits. Owing to changes in the diagnostic criteria and the availability of magnetic resonance imaging, clinicians diagnose more patients, establish diagnosis earlier, and diagnose more patients with benign cases. To account for this, all models were also adjusted for the time gap between MS onset and diagnosis.

A benefit of the Weibull model is that it can handle interval-censored observations. Interval censoring occurs frequently in RCTs and time-to-event analysis. Interval censoring is the situation when the exact timing of an outcome is unknown, but the time interval during which the outcome occurred is known (Figure).23 For instance, when the first recorded EDSS score is equal to or higher than the disability milestone of interest, the exact time of the event cannot be determined. However, the interval in which it occurred is known, some time between the date of first symptom and the date of the first EDSS measurement. This is particularly problematic for short-term outcomes like EDSS 3.0 and 4.0 and is less problematic for EDSS 6.0 since a substantial fraction of patients have already reached the milestone (eTable 1 in the Supplement).

It is well recognized that the conventional approach (right censoring) of analyzing interval censored data can lead to biased estimates, as individuals with highest risk of the event are excluded from the analysis in the conventional approach. The Weibull model allowed us to include persons who had reached the disability milestone on their first clinic visit, thereby minimizing this potential bias.

Owing to changes in the diagnostic criteria and the availability of magnetic resonance imaging, it is possible that more benign cases of MS are diagnosed in recent years. Quiz Ref IDTo account for this, we performed a supplementary analysis adjusting the models for the first recorded EDSS. Only patients with an EDSS recorded within their first 3 years from MS onset were included. To assess the severity of MS at presentation, we also compared early-course relapse rate as well as characteristics of the presenting attack, ie, type of attack and degree of recovery from the first attack, between 2 groups of patients, 1 consisting of those diagnosed between 1995 and 2000 and the other of those diagnosed between 2005 and 2010.

To determine the association of disease-modifying therapies (DMTs) with estimates, we performed further supplementary analyses by stratifying patients with ROMS into 2 groups based on the timing of their treatment initiation (early treatment [≤3 years from onset] or late treatment [>3 years from onset]) and adjusted all models for time receiving first-generation (interferon beta-1a, interferon beta-1b, and glatiramer acetate) and second-generation (teriflunomide, fingolimod, alemtuzumab, rituximab, natalizumab, and dimethyl fumarate) DMTs. Linearity in all models was tested by the likelihood ratio test. We used R software version 3.4.1 (R Foundation for Statistical Computing) for all statistical analyses.

Results

In total, 7331 patients of 12 512 registered patients in SMSreg were diagnosed between January 1995 and December 2010 and were included in the analyses. Characteristics of the cohort are shown in Table 1. Most patients were women (5196 [70.9%]), and 6707 (91.5%) had ROMS. The mean (SD) age at diagnosis was 38.3 (11.7) years, the mean (SD) age at MS onset was 34.3 (10.5) years, and the mean (SD) number of recorded EDSS scores was 8.2 (5.9). Individuals who reached EDSS 6.0 were older at onset, older at diagnosis, and more likely to have progressive-onset MS (POMS), compared with those who reached only to EDSS 3.0 and/or 4.0. More details on patients’ characteristics are found in eTable 1 in the Supplement. Those who were diagnosed recently (2005-2010) had on average more annual relapses during the first 2 and 5 years from MS onset and a lower rate of recovery from the presenting attack compared with those diagnosed between 1995 and 2000 (eTable 2 in the Supplement). Concerning symptoms of the presenting attack, 19% (n = 281) of patients with MS from the recent cohort showed motor symptoms, while the corresponding number for the early cohort was 14% (n = 169). The proportion of those with isolated optic neuritis was also significantly higher in the early cohort compared with the recent cohort (197 [27%] vs 195 [19%]). However, it is worth noting that at least 28% (n = 456) of patients in the early cohort and 47% (n = 1198) of patients in the recent cohort had missing data concerning relapse characteristics, which may in some way influence results.

The results of the Weibull models are shown in Table 2. After adjusting for sex, number of visits, diagnostic delay, and onset age, we found a 3% decrease per calendar year of diagnosis for the risk of reaching a sustained EDSS 3.0 (HR, 0.97; 95% CI, 0.96-0.97), a 6% decrease per calendar year of diagnosis for the risk of reaching a sustained EDSS 4.0 (HR, 0.94; 95% CI, 0.93-0.95), and a 7% decrease per calendar year of diagnosis for the risk of reaching a sustained EDSS 6.0 (HR, 0.93; 95% CI, 0.91-0.94) among patients with ROMS. For example, a patient diagnosed in 2010 has 14%, 26%, and 30% lower adjusted risk of EDSS milestones 3.0, 4.0, and 6.0 compared with a patient diagnosed in 2005, respectively. The trend was not statistically significant for patients with POMS (EDSS 3.0: HR, 1.01; 95% CI, 0.98-1.03; EDSS 4.0: HR, 1.00; 95% CI, 0.98-1.02; EDSS 6.0: HR, 1.00; 95% CI, 0.98-1.02).

Quiz Ref IDAmong ROMS, the adjusted risk of reaching any of the disability milestones of interest was significantly higher for men, those with more clinic visits, and those with an older age at onset. Only older age at onset was found to have such influence among those with POMS.

To determine if the initial EDSS score influenced the risk estimates, we included the first EDSS within 3 years from MS onset in the models. This supplementary analysis included 35% (n = 2299) and 20% (n = 130) of the patients with ROMS and POMS, respectively. Corresponding HRs for EDSS 3.0, 4.0, and 6.0 were 0.99 (95% CI, 0.98-1.01), 0.96 (95% CI, 0.94-0.98), and 0.93 (95% CI, 0.90-0.96) among patients with ROMS and 0.96 (95% CI, 0.87-1.06), 0.97 (95% CI, 0.90-1.05), and 1.02 (95% CI, 0.95-1.10) among patients with POMS, respectively. In fact, HRs did not change substantially and only a little wider CIs were observed indicating that confounding effect of the first EDSS cannot explain the decreasing risks that were observed for recent years.

Table 3 shows the results of our supplementary analyses, in which patients with ROMS were stratified by early (≤3 years since onset) or late (>3 years since onset) treatment status and simultaneously adjusted for time with first-generation and second-generation DMTs. Adjusting for time with first- and second-generation DMTs did not significantly impact the original estimates. Hazard ratios for the association of time with first- and second-generation DMTs on reaching EDSS 6.0 were 0.94 (95% CI, 0.92-0.95) and 0.96 (95% CI, 0.94-0.98), respectively. Corresponding HRs for EDSS 4.0 were 0.95 (95% CI, 0.94-0.96) and 1.00 (95% CI, 0.98-1.02) and for EDSS 3.0 were 0.96 (95% CI, 0.95-0.96) and 1.03 (95% CI, 1.02-1.05), respectively. On the other hand, those who were treated early (≤3 years from MS onset) were at a nonsignificantly lower risk of reaching EDSS score 6.0 compared with those who treated after 3 years since onset (data not shown).

Discussion

This nationwide cohort study suggests that MS is a more slowly progressing disease than previously thought. Using real-world data from a high-quality MS registry and robust statistical methodology, we found a significant increase in the time to reach disability milestones during the last decade among patients with ROMS but not among those with POMS in Sweden.

The natural history of MS has been studied extensively,24 but there are little data on whether the trajectory of the disease has changed over time, as new diagnostic and therapeutics for MS have come into clinical use. The 2 natural history studies that reported no changes in disability profile among patients with MS over time were conducted on an essentially untreated population.11,12 Conversely, several extension studies of the pivotal trials of DMTs suggest that DMT use may minimize disability.25-27 A few observational studies are broadly consistent with the hypothesis that DMTs may have a positive effect on the long-term course of the disease in a real-life setting.28-30 Before our study, no comparably large-scale study has examined trends in disease severity in a mostly treated MS population, to our knowledge.

A common statistical problem in the real-world clinical setting is the difficulty in dealing with interval censoring, which can spuriously suggest better outcomes in recent years. To handle this, we applied a model that can effectively deal with such a problem.31,32 In addition, the lack of such a signal in POMS serves as a negative control supporting the robustness of our approach.

Reasons for the observed reduction in disability risk over time are not known, but we have several hypotheses: the most obvious being the increase in DMT use over time in ROMS. Our finding of a lower risk of progression among ROMS, but not POMS, over time is compatible with availability of DMTs for ROMS but not POMS. However, the supplementary analyses showed similar estimates when adjusting for time with first- and second-generation DMTs. We also observed an independent effect of treatment from calendar year on disability milestones, indicating that treatment cannot fully explain the lower risk of reaching disability milestones in recent years. It should be noted that since DMT variables were only intended for model adjustment, the coefficient obtained from these variables cannot be used to interpret the drug effectiveness. Owing to the complexity of analyses for treatment effect, such a study requires a different design and a control group.

Another explanation for better prognosis of patients with more recent MS can be earlier treatment of MS, apart from enhanced efficacy or earlier diagnosis, with more efficient medications in recent years for ROMS. Naturally, the increasing use of DMTs for MS makes the access to unbiased untreated cohorts progressively more challenging. The association of DMTs with disease progression is an important question that needs to be answered using well-designed pharmaco-epidemiological methods including those capable of handling indication bias.

Another possible explanation for the discrepancy between ROMS and POMS is the new diagnostic criteria. In a previous study, we found that the diagnosis of primary progressive MS has significantly decreased in Sweden, specifically after the introduction of DMTs.33 Thus, the smaller proportion of patients with POMS in recent years may have rendered the analysis underpowered to detect a difference over time in this cohort. Further, the newer diagnostic criteria may have led to more benign cases of MS in the cohort in later years, which would reduce the average disability progression over time. Adjusting for the initial EDSS score did not alter our findings substantially, suggesting that this alone cannot explain our findings.

Another explanation for our finding could be that a more complete coverage of the MS population, or application of new MS criteria or better access to magnetic resonance imaging, would include more patients with milder disease in recent years. To rule out that our findings of decreased progression rates were caused by a change of the natural course of MS of whatever reason, we therefore compared the early course characteristics of patients diagnosed at the beginning of our observation period (1995-2000) with those diagnosed at the end of our observation period (2005-2010). Surprisingly, patients of the more recent period more often had characteristics of more active disease rather than the opposite. Indeed, we found the opposite. Patients in the more recent epoch evidenced greater and not reduced disease activity compared with the earlier cohort. This may be considered to be in conflict with observations of decreased relapse rates in recent placebo arms of RCTs. However, we reason that the reduced relapse rate seen in recent RCTs may be influenced by many factors including that investigators are likely to avoid offering highly active patients participation in placebo-controlled clinical trials, rather than reflecting a change in MS phenotype. In addition, the inclusion rate of new patients in the MS registry has been strikingly similar since 2002 with a virtually linear increase in the number of patients.

Other potential explanations for the improvement in disability outcomes include changes in health behaviors or environmental exposures (eg, smoking rate, vitamin D deficiency, the Epstein-Barr virus),34-36 improvements in MS care, improvements in the recognition and management of symptoms and complications of MS, as well as of infections and other coexisting medical conditions. In a previous study, we found the inflation-adjusted cost of illness of patients with MS was lower in 2012 than in 2006, despite the increase in the cost of treatments, which may also indicate a change in the course of MS.37 A potential gene-environment interaction mechanism is another complex explanation worth consideration. We thus confirm an apparent increase in age at disability observed among mostly treated cohorts in other studies such as MSBase, reflecting a true slowing in the rate of disability accumulation at least in Sweden.10

While Kurtzke EDSS20 has several well-recognized limitations as an outcome measure, including poor intra- and interrater reliability,38 it is still the most widely used disability scale in observational and interventional studies.39 The ability of the EDSS to measure improvement in neurological function are less well studied,40 but EDSS 6.0 has been an important and clinically relevant disability milestone for many studies so far.3 We used sustained EDSS milestones as our outcomes to take into account reported fluctuations in EDSS after confirmed progression.41 We also followed our patients for at least 7 years, providing sufficient time to assess the long-term disease progression. Furthermore, results were adjusted for the potential association of a gap between onset and diagnosis to lower the association of longer referral or diagnostic delay on our results.

Strengths and Limitations

Other important strengths of this study include the population-based nature of our cohort and the use of robust statistical analyses, which allowed for the inclusion of patients who had already reached the milestones of interest at the time of their first clinic visit. Some natural history studies have dealt with this issue by attempting to retrospectively estimate the time of reaching disability milestone (mainly EDSS 6.0) through medical record review or patient interview or by limiting their analysis to the proportion of patients actually reaching EDSS 6.0 at 15 years.11 However, it is not clear as to what extent these approaches have been successful in eliminating the bias.

Our study also has limitations. An estimated 20% of prevalent cases of MS in Sweden are not registered in the SMSreg. We do not know whether this same pattern would have emerged for this subpopulation of patients not followed in a neurology clinic. Furthermore, we could not explore changes in disease progression prior to 1995. Finally, we cannot formally exclude associations of changes in ascertainment or disease monitoring over time that may have influenced our findings. However, the consistency of the results across analyses suggest that these findings are authentic. Future studies should endeavor to identify the underlying reasons for these changes. Our findings are compatible with a medium-term and a long-term beneficial effect of the widely used MS DMTs, but we cannot fully rule out the possibility that other factors influence our observation. Because significant societal resources are presently being invested in MS treatments, our findings, suggesting that these medications may affect the long-term course of disease, are encouraging.

Conclusions

Quiz Ref IDThe risk of reaching major disability milestones has significantly decreased during the last decade in patients with ROMS in Sweden. Several factors (including increase in availability and use of more efficient DMTs, earlier diagnosis, earlier treatment, change in health behaviors, or environmental exposures) could potentially be responsible for this observation. However, given that no change was seen in disability accrual of patients with POMS and the absence of efficacious treatment option in this group, increased use of more efficacious DMTs could be a possible driver of this change.

Back to top
Article Information

Corresponding author: Jan Hillert, MD, PhD, Department of Clinical Neuroscience, Karolinska Institutet, Widerströmska huset, Tomtebodavägen 18, 171 77 Solna, Stockholm, Sweden (jan.hillert@ki.se).

Accepted for Publication: December 18, 2018.

Published Online: March 18, 2019. doi:10.1001/jamaneurol.2019.0330

Author Contributions: Drs Beiki and Manouchehrinia had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Manouchehrinia and Hillert contributed equally as co–last author.

Concept and design: Beiki, Manouchehrinia, Hillert.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Beiki, Manouchehrinia.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: All authors.

Obtained funding: Hillert.

Supervision: Hillert.

Conflict of Interest Disclosures: Dr Beiki has received salary for epidemiological consultation for pharmaceutical companies from Cognizant Technology Solutions. Dr Hillert has received honoraria for serving on advisory boards for Biogen, Sanofi Genzyme, and Novartis, speaker’s fees from Biogen, Novartis, Merck Serono, Bayer Schering, Teva Pharmaceutical Industries, and Sanofi Genzyme; and has served as principal investigator for projects or has received unrestricted research support from Biogen Idec, Merck Serono, Teva Pharmaceutical Industries, Sanofi Genzyme, and Bayer Schering. No other disclosures were reported.

Funding/Support: This research was funded by the Swedish Research Council and the Swedish Brain Foundation.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

References
1.
Steinvorth  SM, Rover  C, Schneider  S, Nicholas  R, Straube  S, Friede  T.  Explaining temporal trends in annualised relapse rates in placebo groups of randomised controlled trials in relapsing multiple sclerosis: systematic review and meta-regression.  Mult Scler. 2013;19(12):1580-1586. doi:10.1177/1352458513481009Google Scholar
2.
Röver  C, Nicholas  R, Straube  S, Friede  T.  Changing EDSS progression in placebo cohorts in relapsing MS: a systematic review and meta-regression.  PLoS One. 2015;10(9):e0137052. doi:10.1371/journal.pone.0137052PubMedGoogle Scholar
3.
Tremlett  H, Zhao  Y, Rieckmann  P, Hutchinson  M.  New perspectives in the natural history of multiple sclerosis.  Neurology. 2010;74(24):2004-2015. doi:10.1212/WNL.0b013e3181e3973fPubMedGoogle Scholar
4.
Weinshenker  BG, Bass  B, Rice  GP,  et al.  The natural history of multiple sclerosis: a geographically based study: I: clinical course and disability.  Brain. 1989;112(pt 1):133-146. doi:10.1093/brain/112.1.133PubMedGoogle Scholar
5.
Runmarker  B, Andersen  O.  Prognostic factors in a multiple sclerosis incidence cohort with twenty-five years of follow-up.  Brain. 1993;116(pt 1):117-134. doi:10.1093/brain/116.1.117PubMedGoogle Scholar
6.
Confavreux  C, Vukusic  S, Moreau  T, Adeleine  P.  Relapses and progression of disability in multiple sclerosis.  N Engl J Med. 2000;343(20):1430-1438. doi:10.1056/NEJM200011163432001PubMedGoogle Scholar
7.
Pittock  SJ, Mayr  WT, McClelland  RL,  et al.  Change in MS-related disability in a population-based cohort: a 10-year follow-up study.  Neurology. 2004;62(1):51-59. doi:10.1212/01.WNL.0000101724.93433.00PubMedGoogle Scholar
8.
Tremlett  H, Paty  D, Devonshire  V.  Disability progression in multiple sclerosis is slower than previously reported.  Neurology. 2006;66(2):172-177. doi:10.1212/01.wnl.0000194259.90286.fePubMedGoogle Scholar
9.
Brown  M, Andreou  P, Kirby  S, Murray  J. EDSS natural history atlas: MS disability progression in Nova Scotia 1979-2004. Presented at: ECTRIMS Conference; September 17-20, 2008; Montreal, Quebec, Canada.
10.
Kister  I, Chamot  E, Cutter  G,  et al; MSBase Investigators.  Increasing age at disability milestones among MS patients in the MSBase Registry.  J Neurol Sci. 2012;318(1-2):94-99. doi:10.1016/j.jns.2012.03.017PubMedGoogle Scholar
11.
Shirani  A, Zhao  Y, Kingwell  E, Rieckmann  P, Tremlett  H.  Temporal trends of disability progression in multiple sclerosis: findings from British Columbia, Canada (1975-2009).  Mult Scler. 2012;18(4):442-450. doi:10.1177/1352458511422097Google Scholar
12.
Pittock  SJ, Mayr  WT, McClelland  RL,  et al.  Disability profile of MS did not change over 10 years in a population-based prevalence cohort.  Neurology. 2004;62(4):601-606. doi:10.1212/WNL.62.4.601PubMedGoogle Scholar
13.
Manouchehrinia  A, Beiki  O, Hillert  J.  Clinical course of multiple sclerosis: a nationwide cohort study.  Mult Scler. 2017;23(11):1488-1495. doi:10.1177/1352458516681197Google Scholar
14.
Freilich  J, Manouchehrinia  A, Trusheim  M,  et al.  Characterization of annual disease progression of multiple sclerosis patients: a population-based study.  Mult Scler. 2018;24(6):786-794. doi:10.1177/1352458517706252Google Scholar
15.
Burkill  S, Montgomery  S, Hajiebrahimi  M, Hillert  J, Olsson  T, Bahmanyar  S.  Mortality trends for multiple sclerosis patients in Sweden from 1968 to 2012.  Neurology. 2017;89(6):555-562. doi:10.1212/WNL.0000000000004216PubMedGoogle Scholar
16.
Ahlgren  C, Odén  A, Lycke  J.  High nationwide prevalence of multiple sclerosis in Sweden.  Mult Scler. 2011;17(8):901-908. doi:10.1177/1352458511403794Google Scholar
17.
Ahlgren  C, Odén  A, Lycke  J.  High nationwide incidence of multiple sclerosis in Sweden.  PLoS One. 2014;9(9):e108599. doi:10.1371/journal.pone.0108599PubMedGoogle Scholar
18.
Hillert  J, Stawiarz  L.  The Swedish MS registry: clinical support tool and scientific resource.  Acta Neurol Scand. 2015;132(199):11-19. doi:10.1111/ane.12425PubMedGoogle Scholar
19.
Andersen  O.  From the Gothenburg cohort to the Swedish multiple sclerosis registry.  Acta Neurol Scand Suppl. 2012;(195):13-19. doi:10.1111/ane.12023PubMedGoogle Scholar
20.
Kurtzke  JF.  Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS).  Neurology. 1983;33(11):1444-1452. doi:10.1212/WNL.33.11.1444PubMedGoogle Scholar
21.
Collett  D.  Modelling Survival Data in Medical Research. 3rd ed. Boca Raton, Florida: Chapman and Hall/CRC; 2015.
22.
Kalbfleisch  JD, Prentice  RL.  The Statistical Analysis of Failure Time Data. 2nd ed. Hoboken, New Jersey: Wiley-Interscience; 2002.
23.
Zhang  Z, Sun  J.  Interval censoring.  Stat Methods Med Res. 2010;19(1):53-70. doi:10.1177/0962280209105023PubMedGoogle Scholar
24.
McAlpine  D, Compston  A. McAlpine's Multiple Sclerosis. 4th ed. London, United Kingdom: Churchill Livingstone; 2005.
25.
Kappos  L, Traboulsee  A, Constantinescu  C,  et al.  Long-term subcutaneous interferon beta-1a therapy in patients with relapsing-remitting MS.  Neurology. 2006;67(6):944-953. doi:10.1212/01.wnl.0000237994.95410.cePubMedGoogle Scholar
26.
Ford  C, Goodman  AD, Johnson  K,  et al.  Continuous long-term immunomodulatory therapy in relapsing multiple sclerosis: results from the 15-year analysis of the US prospective open-label study of glatiramer acetate.  Mult Scler. 2010;16(3):342-350. doi:10.1177/1352458509358088Google Scholar
27.
Ebers  G, Traboulsee  A, Langdon  D,  et al. The final results of the interferon beta-1b 16-year long-term follow-up study. https://ora.ox.ac.uk/objects/uuid:4f0c52f2-64a4-486e-a074-a22498bd236b. Accessed February 12, 2019.
28.
Trojano  M, Pellegrini  F, Paolicelli  D,  et al; Italian Multiple Sclerosis Database Network (MSDN) Group.  Real-life impact of early interferon beta therapy in relapsing multiple sclerosis.  Ann Neurol. 2009;66(4):513-520. doi:10.1002/ana.21757PubMedGoogle Scholar
29.
Veugelers  PJ, Fisk  JD, Brown  MG,  et al.  Disease progression among multiple sclerosis patients before and during a disease-modifying drug program: a longitudinal population-based evaluation.  Mult Scler. 2009;15(11):1286-1294. doi:10.1177/1352458509350307PubMedGoogle Scholar
30.
Kister  I, Chamot  E, Bacon  JH, Cutter  G, Herbert  J; New York State Multiple Sclerosis Consortium.  Trend for decreasing Multiple Sclerosis Severity Scores (MSSS) with increasing calendar year of enrollment into the New York State Multiple Sclerosis Consortium.  Mult Scler. 2011;17(6):725-733. doi:10.1177/1352458510396269Google Scholar
31.
Odell  PM, Anderson  KM, D’Agostino  RB.  Maximum likelihood estimation for interval-censored data using a Weibull-based accelerated failure time model.  Biometrics. 1992;48(3):951-959. doi:10.2307/2532360PubMedGoogle Scholar
32.
Coria  VH, Maximov  S, Rivas-Dávalos  F, Melchor-Hernández  CL.  Perturbative method for maximum likelihood estimation of the Weibull distribution parameters.  Springerplus. 2016;5(1):1802. doi:10.1186/s40064-016-3500-yPubMedGoogle Scholar
33.
Westerlind  H, Stawiarz  L, Fink  K, Hillert  J, Manouchehrinia  A. A significant decrease in diagnosis of primary progressive multiple sclerosis: a cohort study.  Mult Scler. 2016;22(8):1071-1079. doi:10.1177/1352458516643394Google Scholar
34.
D'Hooghe  MB, Nagels  G, Bissay  V, De Keyser  J.  Modifiable factors influencing relapses and disability in multiple sclerosis.  Mult Scler. 2010;16(7):773-785. doi:10.1177/1352458510367721Google Scholar
35.
Zivadinov  R, Zorzon  M, Weinstock-Guttman  B,  et al.  Epstein-Barr virus is associated with grey matter atrophy in multiple sclerosis.  J Neurol Neurosurg Psychiatry. 2009;80(6):620-625. doi:10.1136/jnnp.2008.154906PubMedGoogle Scholar
36.
Ginde  AA, Liu  MC, Camargo  CA  Jr.  Demographic differences and trends of vitamin D insufficiency in the US population, 1988-2004.  Arch Intern Med. 2009;169(6):626-632. doi:10.1001/archinternmed.2008.604PubMedGoogle Scholar
37.
Gyllensten  H, Wiberg  M, Alexanderson  K, Friberg  E, Hillert  J, Tinghög  P.  Comparing costs of illness of multiple sclerosis in three different years: a population-based study.  Mult Scler. 2018;24(4):520-528. doi:10.1177/1352458517702549Google Scholar
38.
Hobart  J, Freeman  J, Thompson  A.  Kurtzke scales revisited: the application of psychometric methods to clinical intuition.  Brain. 2000;123(Pt 5):1027-1040. doi:10.1093/brain/123.5.1027PubMedGoogle Scholar
39.
Kragt  JJ, Nielsen  JM, van der Linden  FA, Polman  CH, Uitdehaag  BM.  Disease progression in multiple sclerosis: combining physicians' and patients' perspectives?  Mult Scler. 2011;17(2):234-240. doi:10.1177/1352458510385505Google Scholar
40.
Ontaneda  D, Cohen  JA.  EDSS improvement: recovery of function or noise?  Mult Scler. 2012;18(11):1520-1521. doi:10.1177/1352458512441689Google Scholar
41.
Ebers  GC, Heigenhauser  L, Daumer  M, Lederer  C, Noseworthy  JH.  Disability as an outcome in MS clinical trials.  Neurology. 2008;71(9):624-631. doi:10.1212/01.wnl.0000313034.46883.16PubMedGoogle Scholar
×