Advertisement

Hospitalization is associated with subsequent disability in multiple sclerosis

  • Allan Garland
    Affiliations
    Department of Internal Medicine, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Canada

    Department of Community Health Sciences, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Canada

    Manitoba Centre for Health Policy, Winnipeg, Canada
    Search for articles by this author
  • Luanne M. Metz
    Affiliations
    Department of Clinical Neurosciences, University of Calgary, Calgary, Alberta, Canada

    Hotchkiss Brain Institute, University of Calgary, Calgary, Alberta, Canada
    Search for articles by this author
  • Charles N. Bernstein
    Affiliations
    Department of Internal Medicine, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Canada

    IBD Clinical and Research Centre, University of Manitoba, Winnipeg, Canada
    Search for articles by this author
  • Christine A. Peschken
    Affiliations
    Department of Internal Medicine, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Canada

    Department of Community Health Sciences, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Canada
    Search for articles by this author
  • Carol A. Hitchon
    Affiliations
    Department of Internal Medicine, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Canada
    Search for articles by this author
  • Ruth Ann Marrie
    Correspondence
    Correspondence to: Health Sciences Centre, GF 543- 820 Sherbrook Street, Winnipeg, MB, Canada R3A 1R9.
    Affiliations
    Department of Internal Medicine, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Canada

    Department of Community Health Sciences, Max Rady College of Medicine, Rady Faculty of Health Sciences, University of Manitoba, Canada
    Search for articles by this author
Open AccessPublished:March 24, 2017DOI:https://doi.org/10.1016/j.msard.2017.03.009

      Highlights

      • Over five years, one-fifth of persons with MS were hospitalized.
      • Acute illness-related hospitalization was associated with a step increase in disability.
      • Acute illness-related hospitalizations did not change the rate of disability progression.

      Abstract

      Background

      Although an increasing amount of research has evaluated interactions between MS and comorbid chronic disease, few data exist regarding the interactions between MS and acute illness. As compared to age and sex-matched persons without MS, persons with MS experience higher rates of hospitalization and critical illness, and higher mortality rates and health care utilization following critical illness. We aimed to determine whether acute illness requiring hospitalization is associated with progression of multiple sclerosis (MS).

      Methods

      We conducted this population-based, retrospective cohort study by linking data from the regional MS Clinic in Calgary, Canada with the Canadian Discharge Abstract Database to identify non-obstetric hospitalizations. We included individuals with a confirmed diagnosis of MS, at least one recorded Expanded Disability Status Scale (EDSS) measurement, and known age of symptom onset of age 10 years or older. Using data from 2009 to 2014, we used generalized linear models with generalized estimating equations to establish the association within individuals between hospitalization and the time course of MS-related disability (as measured by the EDSS), adjusting for sex, age, disease course at onset, and use of disease-modifying therapies.

      Results

      We included 2104 individuals with MS in the analysis, who had a median of 4 EDSS measurements each. Of these 491 (23.3%) had at least one hospitalization. Most subjects were female, with a relapsing disease course at onset, and a mean (SD) age at symptom onset of 33.0 (10.0) years. The underlying rate of disability progression averaged 0.9 EDSS points per decade. Following hospitalization, there was a step increase in EDSS, averaging 0.23 points, equivalent to 2.5 years of time-related disease progression. Hospitalization did not alter the subsequent temporal rate of disability progression. The findings did not differ in those hospitalized for MS versus other reasons.

      Conclusions

      Acute illness requiring hospitalization is associated with a worsening of MS-related disability.

      Keywords

      1. Introduction

      Many people with a given chronic illness have comorbid chronic conditions (
      • van den Akker M.
      • Buntinx F.
      • Metsemakers J.F.
      • Roos S.
      • Knottnerus J.A.
      Multimorbidity in general practice: prevalence, incidence, and determinants of co-occurring chronic and recurrent diseases.
      ;
      • Broemeling A.
      • Watson D.E.
      • Prebtani F.
      • on behalf of the Councillors of the Health Outcomes Steering Committee of the Health Council of Canada
      Population patterns of chronic health conditions, co-morbidity and health care use in Canada: implications for policy and practice.
      ). In multiple sclerosis (MS), comorbidities are now recognized to influence the length of the delay between MS onset and diagnosis, disability progression, use of disease-modifying therapy, health care utilization and mortality (
      • Marrie R.A.
      • Miller A.
      • Sormani M.P.
      • Thompson A.
      • Waubant E.
      • Trojano M.
      • et al.
      Recommendations for observational studies of comorbidity in multiple sclerosis.
      ;
      • Zhang T.
      • Tremlett H.
      • Leung S.
      • Zhu F.
      • Kingwell E.
      • Fisk J.D.
      • et al.
      Examining the effects of comorbidities on disease-modifying therapy use in multiple sclerosis.
      ). Although an increasing amount of research has evaluated interactions between MS and comorbid chronic disease few data exist regarding the interactions between MS and acute illness. As compared to age and sex-matched persons without MS, persons with MS experience higher rates of hospitalization and critical illness, and higher mortality rates and health care utilization following critical illness (
      • Marrie R.A.
      • Bernstein C.N.
      • Peschken C.A.
      • Hitchon C.A.
      • Chen H.
      • Fransoo R.
      • et al.
      Intensive care unit admission in multiple sclerosis: increased incidence and Increased mortality.
      ,
      • Marrie R.A.
      • Elliott L.
      • Marriott J.
      • Cossoy M.
      • Blanchard J.
      • Tennakoon A.
      • et al.
      Dramatically changing rates and reasons for hospitalization in multiple sclerosis.
      ,
      • Marrie R.A.
      • Bernstein C.N.
      • Peschken C.A.
      • Hitchon C.A.
      • Chen H.
      • Fransoo R.
      • et al.
      Health care utilization before and after intensive care unit admission in multiple sclerosis.
      ,
      • Marrie R.A.
      • Elliott L.
      • Marriott J.
      • Cossoy M.
      • Tennakoon A.
      • Yu N.
      Comorbidity increases the risk of hospitalizations in multiple sclerosis.
      ;
      • Karamyan A.
      • Dünser M.W.
      • Wiebe D.J.
      • Pilz G.
      • Wipfler P.
      • Chroust V.
      • et al.
      Critical illness in patients with multiple sclerosis: a matched case-control study.
      ).
      However, the interactions between chronic and acute illness are likely to be bidirectional. For example, in older adults cognitive impairment increases the risk of pneumonia, and pneumonia accelerates the progression to dementia (
      • Shah F.A.
      • Pike F.
      • Alvarez K.
      • Angus D.
      • Newman A.B.
      • Lopez O.
      • et al.
      Bidirectional relationship between cognitive function and pneumonia.
      ). The effects of intercurrent acute illness on MS are poorly understood. Therefore, we aimed to evaluate the association between acute illnesses requiring hospitalization and the subsequent progression of MS-related disability. We hypothesized that such acute illnesses would result in both acquisition of disability following hospitalization, and that the subsequent progression of disability would be greater than before hospitalization.

      2. Materials and methods

      This was a retrospective cohort study conducted using two population-based data sources from the Calgary Health Region of the Canadian province of Alberta, population 1.54 million. We linked two existing databases, both obtained from the Canadian Institute for Health Information (CIHI), the Calgary MS Clinic Database (CMSD) and the Canadian Discharge Abstract Database (DAD). The University of Manitoba Health Research Ethics Board approved the study.

      2.1 Calgary MS clinic database

      The Calgary MS Clinic provides care to 98% of persons with MS in the surrounding region. Diagnoses of MS are confirmed by the treating neurologist based on the prevailing diagnostic criteria at the time of diagnosis (
      • Poser C.M.
      • Paty D.W.
      • Scheinberg L.
      • McDonald W.I.
      • Davis F.A.
      • Ebers G.C.
      • et al.
      New diagnostic criteria for multiple sclerosis: guidelines for research protocols.
      ;
      • McDonald W.I.
      • Compston A.
      • Edan G.
      • Goodkin D.
      • Hartung H.-P.
      • Lublin F.D.
      • et al.
      Recommended diagnostic criteria for multiple sclerosis: guidelines from the international panel on the diagnosis of multiple sclerosis.
      ;
      • Polman C.H.
      • Reingold S.C.
      • Edan G.
      • Filippi M.
      • Hartung H.P.
      • Kappos L.
      • et al.
      Diagnostic criteria for multiple sclerosis: 2005 revisions to the McDonald criteria.
      ,
      • Polman C.H.
      • Reingold S.C.
      • Banwell B.
      • Clanet M.
      • Cohen J.A.
      • Filippi M.
      • et al.
      Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria.
      ). The Calgary MS Clinic Database (CMSD) captures standardized demographic and clinical information regarding all persons with MS seen in the clinic including a unique patient identifier, date of birth, sex, date of MS symptom onset, and disease course at onset (relapsing, progressive or unknown). The date of each clinic visit is recorded as well as disability status as measured using the Expanded Disability Status Scale (EDSS), and the use of disease-modifying therapies. The EDSS is a physician-scored measure used as the gold standard for disability progression in MS clinical trials. It is scored from 0 (no disability) to 10 (death due to MS). A score of 6 indicates the need for unilateral assistance (e.g. cane), while a score of 6.5 indicates the need for bilateral assistance (e.g. crutches), and a score of 7.0 indicates wheelchair use (
      • Kurtzke J.F.
      Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS).
      ). The EDSS scores are recorded by experienced MS physicians and trainees using a standardized scoring form. The MS physicians are formally trained to do the EDSS and certified. If a trainee is involved in assessing the individual with MS, this is verified by the attending neurologist. Data were available for clinic visits from September 21, 2009 to March 31, 2014.

      2.2 Discharge abstract database

      The DAD contains information about every hospitalization in Canada, excluding those in Quebec (

      Canadian Institute for Health Information, 2014a. Data Quality Documentation: Discharge Abstract Database 2013–2014.

      ;

      Canadian Institute for Health Information, 2014b. Discharge Abstract Database (DAD) Metadata.

      ). Data are collected by rigorously trained personnel in each hospital, using uniform definitions and collection methods, regardless of the site of care or reason for admission. Further validation and data cleaning are performed by the provincial health departments, and subsequently by CIHI. We obtained DAD records of all discharges from acute care hospitals for the period April 1, 2005, to March 31, 2015, for all the individuals in the CMSD. This allowed us to identify hospitalizations among participants in the CMSD even if they occurred outside Alberta. Variables include a unique patient identifier, sex, date of birth, postal code, dates of admission and discharge, whether the admission was elective vs. emergent/urgent, up to 25 hospital diagnoses in International Classification of Disease (ICD)-10-CA format, whether the hospitalization included any time in an intensive care unit (ICU), and post-hospital disposition. The DAD includes separate records when an individual is directly transferred between hospitals, therefore we combined such records into episodes of hospital care as described elsewhere, allowing a maximum difference in admission dates of 24 h (
      • Fransoo R.
      • Yogendran M.
      • Olafson K.
      • Ramsey C.
      • McGowan K.L.
      • Garland A.
      Constructing episodes of inpatient care: data infrastructure for population-based research.
      ).

      2.3 Study population and period

      We merged the two datasets deterministically using the scrambled unique identifier provided by CIHI for this project. Inclusion criteria included: (i) a confirmed diagnosis of MS, and (ii) at least one recorded EDSS measurement (n=3532). Subjects were excluded if their date of symptom onset was missing because this was used to calculate disease duration, or if age at symptom onset was <10 years (Fig. e-1). Exclusion criteria for individual EDSS measurements were: values predating the listed date of symptom onset, or within 120 days of the end of the DAD data because hospitalization records were dated by discharge dates, and the 99th percentile of hospital length of stay in this dataset was 120 days. Finally, to minimize confounding due to effects of prior hospitalizations, we required that the first EDSS measurement used was not predated by a hospitalization within five prior years, which also excluded all measurements before April 1, 2010.

      2.4 Variables

      The primary outcome was the EDSS, while the primary exposure of interest was an inpatient, non-obstetric, hospitalization. Obstetric admissions were removed from the DAD dataset by CIHI using obstetric-related diagnosis and procedure codes (Table e-1). Covariates included in this analysis were age at symptom onset (continuous), sex (male as reference group), whether the reason for admission was for MS or not based on whether MS was listed as the most responsible hospital diagnosis (ICD-10-CA diagnosis code G35), clinical course at onset (relapsing [reference group], progressive, unknown/not recorded), whether a disease-modifying therapy was being used at the time of each EDSS assessment, and socioeconomic status (SES, continuous). We linked the first three digits of the postal code to the 2011 Canadian census to derive area-level median household income as a measure of SES.

      2.5 Statistical analysis

      Clinical and demographic characteristics were summarized using frequency (percent), mean (standard deviation) or median (interquartile range [IQR]) as appropriate. We conducted the multivariable analysis using a generalized linear model with generalized estimating equations (GEE) to account for clustering of the repeated EDSS measures within individuals (
      • Hardin J.
      • Hilbe J.
      General Estimating Equations.
      ). The model was parameterized to estimate three key aspects of the trajectory of EDSS over time, starting from MS symptom onset. First was a slope representing the rate of change of EDSS during the whole study interval for those who had no hospitalizations; this same slope also applied to the time from MS onset until right before the first hospitalization for those who had any hospitalizations during the study interval. Second, for those who had any hospitalizations during the study interval, we allowed for a possible step change in EDSS from before to after the hospitalization. Finally, for those who had any hospitalizations during the study interval, we estimated a slope representing the change in EDSS over time after the hospitalization, which was allowed to differ from that before hospitalization. The core of the analysis was to assess whether the pre-hospital to post-hospital EDSS step change and slope change were statistically and clinically significant.
      We utilized an identity link and an exchangeable correlation matrix based on observed pairwise correlations of EDSS values that all ranged from 0.71 to 0.89. GEE models generate effect estimates that are population averages of within-subject and between-subject effects over time (
      • Neuhaus J.M.
      • Kalbfleisch J.D.
      Between- and within-cluster covariate effects in the analysis of clustered data.
      ). Since we were interested in assessing the change in EDSS related to hospitalization within individuals, we created separate “within-patient” and “between-patient” components of: (a) time from symptom onset (TIME), and (b) an indicator variable identifying whether the EDSS was measured pre- or post-hospitalization (PREPOST) (
      • Neuhaus J.M.
      • Kalbfleisch J.D.
      Between- and within-cluster covariate effects in the analysis of clustered data.
      ;
      • Begg M.D.
      • Parides M.K.
      Separation of individual-level and cluster-level covariate effects in regression analysis of correlated data.
      ). We also created interaction terms between these components of TIME and PREPOST. For this study, the coefficients of main interest are the within-patient versions, which have the following meanings: (a) TIMEwithin is the slope of the temporal change in EDSS for patients before the first (index) hospitalization, or for the entire interval if the person had no hospitalizations; (b) PREPOSTwithin is the step-change in EDSS from before the first hospitalization to after it; and (c) the interaction between PREPOSTwithin and TIMEwithin is the difference between the pre-hospitalization and post-hospitalization slopes of temporal change in EDSS.
      The baseline model considered EDSS values as being before or after the index hospitalization, and included the variables TIMEwithin, TIMEbetween, PREPOSTwithin and PREPOSTbetween, the interaction terms, as well as age at symptom onset, sex, course at onset, and use of disease-modifying therapy. Sensitivity analyses included: (i) exclusion of individuals with only one recorded EDSS (n =267); (ii) inclusion of SES in the model [omitted for the baseline model due to some missing values]; (iii) categorization of hospitalizations as elective vs. urgent or emergent; (iv) categorization of hospitalizations as including an ICU admission or not; (v) categorization of hospitalizations as due to MS or not; (vi) categorization of hospitalizations as short stay (≤10 days) or longer stay (>10 days); and (vii) accounted for the presence of a second hospitalization if it occurred after the first hospitalization by creating additional TIMEwithin, TIMEbetween, PREPOSTwithin and PREPOSTbetween terms. As extensions of the baseline model, analyses (iii)–(vi) were effected by creating two sets of within-patient and between-patient TIME, PREPOST and interaction variables.
      We used Stata 14.1 (StataCorp, College Station, Texas) for analysis; P-values <0.05 were considered statistically significant.

      3. Results

      The analysis included 2104 individuals with MS, with 9076 EDSS measurements; each individual had a median of 4 EDSS measurements (Table 1). Most subjects were female, with a relapsing disease course at onset, and a mean (SD) age at symptom onset of 33.0 (10.0) years. About half used disease-modifying agents at some point in the study interval. Approximately one-fifth had at least one hospitalization, of which 29% were elective admissions, 12% were for a primary diagnosis of MS. Reasons for admission are summarized by ICD-10-CA chapter in Table e-2. Five percent of admissions included time in an ICU. One hundred and fifty-six (7%) members of this cohort had at least two hospitalizations. The frequency of second hospitalizations following MS-related admission was higher (44.2%) than following non-MS related admissions (30%, p=0.028). The median length of hospitalization was 3 days, and 80% of those hospitalized had a length of stay of 10 days or less. The median (IQR) number of EDSS measurements was the same before (2 [1–4]) and after (2 [1–3]) hospitalization.
      Table 1Demographic and clinical characteristics of study cohort (n=2104).
      VariableValue
      Age (years) at symptom onset, mean (SD)33.2 (10.0)
      Females, N (%)1538 (73.1)
      Area-level median household income at first EDSS value, mean (SD)
      Unavailable for 114 subjects.
      67,231 (15,207)
      Disease course at onset, N (%)
       progressive224 (10.7)
       relapsing1844 (87.6)
       not documented36 (1.7)
      Using disease-modifying therapies, N (%)
       at time of first EDSS611 (29.0)
       at any point1024 (48.7)
      EDSS measurements, N9076
       EDSS measurements/person, median (IQR)4 (2, 5)
       EDSS measurements/person, range1–21
      Years from symptom onset to first EDSS used in analysis, median (IQR)10.8 (5.3, 17.1)
      Years from first to last EDSS used in analysis, median (IQR)2.9 (1.1, 3.8)
      EDSS values, median (IQR)
       first EDSS measurement2.0 (1.5, 4.0)
       all measurements2.5 (1.5, 5.0)
      Hospitalized, N (%)491 (23.3)
       Elective admissions142
       Most Responsible Hospital Diagnosis of MS61
       Including time in an Intensive Care Unit24
       Years from first EDSS used to index hospitalization, median (IQR) (n=491)1.4 (0.3, 2.6)
      Hospitalization characteristics
       Modified Charlson Comorbidity Score
      Modified by omitting points due to hemiplegia/paraplegia to avoid misclassifying these symptoms of MS as comorbidity (Marrie et al., 2014a, 2014b).
      , median (IQR)
      0 (0, 0)
       Modified Charlson Comorbidity Score
      Modified by omitting points due to hemiplegia/paraplegia to avoid misclassifying these symptoms of MS as comorbidity (Marrie et al., 2014a, 2014b).
      , mean (SD)
      0.26 (0.77)
       Length of stay, days3 (2, 7)
       Died in hospital, N5
      Hospitalized and any EDSS measurements afterwards, N (%)
      81 additional subjects had hospitalization without any EDSS measurements afterwards.
      315 (18.3)
      Years (median, (IQR)) from index hospitalization to:
       immediately prior EDSS (n=404)0.41 (0.19, 0.75)
       immediately subsequent EDSS (n=382)0.47 (0.19, 1.19)
       last EDSS used (n=491)1.1 (0.1, 2.8)
      Individuals with at least 2 hospital admissions, N (%)156 (7.4)
       Years from 1st to 2nd hospital admissions, mean (SD) (n=156)0.65 (0.17, 1.55)
      EDSS=Expanded Disability Status Scale; IQR=interquartile range, MS=multiple sclerosis.
      a Unavailable for 114 subjects.
      b Modified by omitting points due to hemiplegia/paraplegia to avoid misclassifying these symptoms of MS as comorbidity (
      • Marrie R.A.
      • Bernstein C.N.
      • Peschken C.A.
      • Hitchon C.A.
      • Chen H.
      • Fransoo R.
      • et al.
      Intensive care unit admission in multiple sclerosis: increased incidence and Increased mortality.
      ,
      • Marrie R.A.
      • Elliott L.
      • Marriott J.
      • Cossoy M.
      • Blanchard J.
      • Tennakoon A.
      • et al.
      Dramatically changing rates and reasons for hospitalization in multiple sclerosis.
      ).
      c 81 additional subjects had hospitalization without any EDSS measurements afterwards.
      The baseline model showed that the rate of disability progression, as assessed by a rise in the EDSS, averaged almost 1 EDSS point per decade (Table 2). Hospitalization altered disability progression; on average, the EDSS increased by 0.23 points, equivalent to approximately 2.5 years of time-related disease progression. However, the interaction term (TIMEwithin-patient×PREPOSTwithin-patient) was not statistically significant, indicating that other than the step increase in disability, hospitalization did not change the subsequent rate (slope) of disability progression. On average, disability was lower among women, subjects with an earlier age at symptom onset, and those with a relapsing (vs. progressive) course at onset. After we excluded subjects with only one EDSS measurement, the findings were similar (Table 3). The addition of SES to the model did not change the main findings, but we found that for each additional $1000 of income, the EDSS was statistically significantly, but only slightly lower (0.006 points; 95%CI: −0.002, −0.009).
      Table 2Baseline regression model of disability progression.
      VariableCoefficient
      Change in disability as measured by the Expanded Disability Status Scale.
      95% CIP-value
      Sex
       maleReferenceReferenceReference
       female−0.27−0.10, −0.400.002
      Age at symptom onset (per 10 years)0.450.37, 0.53<0.0001
      On disease-modifying drug(s)−0.36−0.20, −0.42<0.0001
      Disease course at onset
       relapsingReferenceReferenceReference
       progressive1.441.18, 1.69<0.0001
       not recorded−0.31−0.91, 0.280.31
      TIMEwithin-patient (per year)0.0910.074, 0.107<0.0001
      PREPOSTwithin-patient0.230.14, 0.33<0.0001
      TIMEwithin-patient×PREPOSTwithin-patient−0.02−0.10, 0.0640.64
      TIMEbetween-patients0.0970.088, 0.105<0.0001
      PREPOSTbetween-patients0.950.48, 1.42<0.0001
      TIMEbetween-patients×PREPOSTbetween-patients−0.01−0.037, 0.170.47
      a Change in disability as measured by the Expanded Disability Status Scale.
      Table 3Key coefficients from sensitivity analyses that are simple base case modifications.
      ModelNo. subjectsTIMEwithin-patientPREPOSTwithin-patient
      (per year)
      Coefficient
      Change in disability as measured by the Expanded Disability Status Scale.
      95% C.I.Coefficient95% C.I.
      Base case21040.0910.074, 0.110.230.14, 0.33
      Exclude subjects with only 1 EDSS value18370.0910.075, 0.110.230.13, 0.33
      Include median household income20220.0870.070, 0.100.250.15, 0.35
      a Change in disability as measured by the Expanded Disability Status Scale.
      The remaining sensitivity analyses suggested heterogeneity in the effects of hospitalization on disability, with this heterogeneity only affecting the step change in EDSS following hospitalization (final four data columns of Table 4). The step change worsening in the EDSS was greater for urgent/emergent vs. elective hospitalizations (0.26 vs. 0.16, p=0.32), for hospitalizations that required ICU care vs. those that did not (0.65 vs. 0.21, p=0.10), when the main admission diagnosis was not MS vs. MS (0.32 vs. 0.22, p=0.56), and for longer rather than shorter lengths of stay (0.35 vs. 0.20, p=0.22) although these differences were not statistically significant. Finally, assessing the step changes in EDSS with up to two successive hospitalizations showed similar average effects for the first and second hospitalization (both 0.19, p=0.98 for the difference between them, Fig. 1). Just as in the baseline model, hospitalization did not significantly change the trajectory (slope) of EDSS over time in any of the sensitivity analyses.
      Table 4Key coefficients from sensitivity analyses adding a variable to the base case (n=2104).
      ModelTIMEwithin-patient (per year)PREPOSTwithin-patient
      Coefficient
      Change in disability as measured by the Expanded Disability Status Scale.
      95% CICoefficient
      Change in disability as measured by the Expanded Disability Status Scale.
      95% CICoefficient
      Change in disability as measured by the Expanded Disability Status Scale.
      95% CI
      Base case0.0910.074, 0.1070.230.14, 0.33
      First hospitalization being elective vs. urgent/emergent0.0910.074, 0.11Elective admissionUrgent/emergent admission
      0.160.001, 0.330.260.15, 0.37
      First hospitalization with vs. without time in an ICU0.0910.074, 0.11Without ICU careWith ICU care
      0.210.12, 0.310.650.13, 1.18
      Main hospital diagnosis was vs. was not MS0.0910.074, 0.11Hospitalized for MSHospitalized for non-MS reason
      0.220.12, 0.320.320.012, 0.62
      First and second hospitalizations0.0890.073, 0.11First hospitalizationSecond hospitalization
      0.190.088, 0.290.19−0.006, 0.38
      Short stay vs. long stay hospitalizations0.0910.074, 0.11Length of stay≤10 daysLength of stay>10 days
      0.200.099, 0.310.350.14, 0.57
      a Change in disability as measured by the Expanded Disability Status Scale.
      Fig. 1
      Fig. 1Average trajectory of Expanded Disability Status Scale Score according to number of hospitalization.

      4. Discussion

      We used population-based clinical and administrative datasets to show that following acute hospitalization for any cause, people with MS experienced a step worsening of disability, as measured by the EDSS, equivalent to 2.5 years of time-related disease progression. Hospitalization did not alter the subsequent temporal rate of disability progression. The effect of a second hospitalization was similar and additive. While none of our sensitivity analyses showed statistically significant heterogeneity in the change in disability with hospitalization, the point estimates showed a greater than 3-fold higher rise for hospitalizations that included time in an ICU; since only 24 subjects had ICU-containing hospitalizations, it seems plausible that this effect is real but lacks statistical significance due to the small size of that subgroup. Previously we showed that among hospitalized persons with MS, in the year following admission those who were admitted to the ICU had statistically significantly greater mean numbers of hospital days (15.6 vs. 7.8), and physician visits (27.0 vs. 19.9) (
      • Marrie R.A.
      • Bernstein C.N.
      • Peschken C.A.
      • Hitchon C.A.
      • Chen H.
      • Fransoo R.
      • et al.
      Health care utilization before and after intensive care unit admission in multiple sclerosis.
      ,
      • Marrie R.A.
      • Elliott L.
      • Marriott J.
      • Cossoy M.
      • Tennakoon A.
      • Yu N.
      Comorbidity increases the risk of hospitalizations in multiple sclerosis.
      ). These findings would be consistent with increased health care needs due to increased disability post-hospitalization. Collectively these findings and our prior work demonstrate that there are bidirectional effects between acute illness and MS.
      Little is known about the effect of acute intercurrent illness on the progression of chronic disease including MS. In community-dwelling persons aged 70 years or older without disability, illness or injuries leading to hospitalization are associated with an increased risk of developing disability (
      • Gill T.M.
      • Allore H.G.
      • Gahbauer E.A.
      • Murphy T.E.
      Change in disability after hospitalization or restricted activity in older persons.
      ). In chronic obstructive lung disease (COPD), lung function tests provide a measure of disease progression, and two studies have explored how acute COPD exacerbations influence the trajectory of forced expiratory volume in one second (FEV1), which normally falls with age but declines faster in COPD (
      • Kanner R.E.
      • Anthonisen N.R.
      • Connett J.E.
      Lower respiratory illnesses promote FEV(1) decline in current smokers but not ex-smokers with mild chronic obstructive pulmonary disease: results from the lung health study.
      ;
      • Donaldson G.C.
      • Seemungal T.A.
      • Bhowmik A.
      • Wedzicha J.A.
      Relationship between exacerbation frequency and lung function decline in chronic obstructive pulmonary disease.
      ). Both studies showed that the rate of FEV1 decline became more rapid after hospitalization for a COPD exacerbation, but they did not evaluate a possible step-change effect.
      Factors that influence disease progression in MS remain poorly understood and therapies to slow progression are only modestly effective therefore it is important to understand potentially modifiable events. Our finding that acute illness precedes a worsening of the chronic disease is consistent with evidence showing that hospitalization with or without critical illness results in long-lasting functional decline (
      • Cuthbertson B.H.
      • Roughton S.
      • Jenkinson D.
      • Maclennan G.
      • Vale L.
      Quality of life in the five years after intensive care: a cohort study.
      ;
      • Feemster L.C.
      • Cooke C.R.
      • Rubenfeld G.D.
      • Hough C.L.
      • Ehlenbach W.J.
      • Au D.H.
      • et al.
      The influence of hospitalization or intensive care unit admission on declines in health-related quality of life.
      ). Given the high frequency of hospitalizations in MS (
      • Evans C.
      • Kingwell E.
      • Zhu F.
      • Oger J.
      • Zhao Y.
      • Tremlett H.
      Hospital admissions and MS: temporal trends and patient characteristics.
      ), our finding that acute illness requiring hospitalization affects disability in MS progression offers a potential avenue for mitigating some disease progression. Individuals with MS are at increased risk of hospitalizations due to influenza, urinary tract infections and pressure ulcers (
      • Marrie R.A.
      • Bernstein C.N.
      • Peschken C.A.
      • Hitchon C.A.
      • Chen H.
      • Fransoo R.
      • et al.
      Intensive care unit admission in multiple sclerosis: increased incidence and Increased mortality.
      ,
      • Marrie R.A.
      • Elliott L.
      • Marriott J.
      • Cossoy M.
      • Blanchard J.
      • Tennakoon A.
      • et al.
      Dramatically changing rates and reasons for hospitalization in multiple sclerosis.
      ), all of which can potentially be prevented. Comorbid illnesses are associated with increased risks of hospitalization, (
      • Marrie R.A.
      • Bernstein C.N.
      • Peschken C.A.
      • Hitchon C.A.
      • Chen H.
      • Fransoo R.
      • et al.
      Health care utilization before and after intensive care unit admission in multiple sclerosis.
      ,
      • Marrie R.A.
      • Elliott L.
      • Marriott J.
      • Cossoy M.
      • Tennakoon A.
      • Yu N.
      Comorbidity increases the risk of hospitalizations in multiple sclerosis.
      ) therefore better management of comorbidities is also important. The increased frequency of second hospitalizations following MS-related admissions suggests that individuals hospitalized for MS may warrant closer follow-up after discharge. Our findings cannot be solely attributed to an intrinsic effect of MS (e.g. relapses), as we found that the effects of hospitalization were similar for MS-related and non-MS related admissions.
      This study had several strengths, including the use of a linked population-based clinical and administrative dataset, and an analytic approach aimed at understanding the effects of exposure to hospitalization on disability progression at the individual level. We considered several potential confounders, and conducted several sensitivity analyses. The effects of sex and SES were consistent with findings in other cohorts (
      • Shirani A.
      • Zhao Y.
      • Karim M.
      • Evans C.
      • Kingwell E.
      • van der Kop M.
      • et al.
      Association between use of interferon beta and progression of disability in patients with relapsing-remitting multiple sclerosis.
      ), and the average increase in the EDSS of 0.9 over 10 years is similar to the average increase of 1 point over 10 years reported elsewhere (
      • Pittock S.J.
      • Mayr W.T.
      • McClelland R.L.
      • Jorgensen N.W.
      • Weigand S.D.
      • Noseworthy J.H.
      • et al.
      Change in MS-related disability in a population-based cohort: a 10-year follow-up study.
      ), suggesting that findings from our cohort are generalizable.
      Some study limitations should also be noted. First, the number of people who were hospitalized was relatively small. Second, due to the relatively small number of EDSS measurements per individual, we were constrained to assess linear rates of EDSS change with time, rather than allowing for a nonlinear relationship. Although it is possible that if our dataset encompassed more years and allowed for assessing a nonlinear relationship, we might have observed a return of hospital-associated worsening in disability towards the pre-hospital value, this seems unlikely given that MS relapses generally have completed their recovery within weeks to a few months (
      • Vollmer T.
      The natural history of relapses in multiple sclerosis.
      ), while our median interval from hospitalization to the final EDSS measurement was 1.1 years. Third, the EDSS has known limitations and would be relatively insensitive to cognitive changes, potentially underestimating the impact of hospitalization on MS. Fourth, we did not have Functional System Scores, therefore we could not determine what specific changes were driving the observed changes in the EDSS. Fifth, we used an area-level measure of SES as a proxy for individual-level SES, but a study comparing the associations between area-level and individual-level measures of SES and health outcomes produced similar findings (
      • Mustard C.A.
      • Derksen S.
      • Berthelot J.-M.
      • Wolfson M.
      Assessing ecologic proxies for household income: a comparison of household and neighbourhood level income measures in the study of population health status.
      ). Finally, we did not account for the effects of comorbid disease as this information was not available in the clinical database, and could only be derived from the DAD for those hospitalized. As comorbidity is associated with the risk of hospitalization in MS and with disability progression (
      • Marrie R.A.
      • Rudick R.
      • Horwitz R.
      • Cutter G.
      • Tyry T.
      • Campagnolo D.
      • et al.
      Vascular comorbidity is associated with more rapid disability progression in multiple sclerosis.
      ,
      • Marrie R.A.
      • Bernstein C.N.
      • Peschken C.A.
      • Hitchon C.A.
      • Chen H.
      • Fransoo R.
      • et al.
      Health care utilization before and after intensive care unit admission in multiple sclerosis.
      ,
      • Marrie R.A.
      • Elliott L.
      • Marriott J.
      • Cossoy M.
      • Tennakoon A.
      • Yu N.
      Comorbidity increases the risk of hospitalizations in multiple sclerosis.
      ), future studies should consider this factor.
      Acute illness requiring hospitalization is associated with greater disability in MS, regardless of the reason for hospitalization. This suggests that acute illness may be a modifiable risk factor for disability in MS.

      Disclosures

      Allan Garland receives research funding from the Canadian Institutes of Health Research, the Multiple Sclerosis Society of Canada, the Heart & Stroke Foundation of Canada, and the Canadian Frailty Network.
      Luanne Metz has a grant from Hoffmann – La Roche to undertake a study of the Economic Burden of MS.
      Charles Bernstein receives research funding from the Canadian Institutes of Health Research, Multiple Sclerosis Scientific Research Foundation, Crohn's and Colitis Canada. He has consulted to Abbvie Canada, Ferring Canada, Janssen Canada, Mylan Pharmaceuticals, Pfizer Canada, Shire Canada, Takeda Canada, and has received unrestricted educational grants from Abbvie Canada, Janssen Canada, Shire Canada, and Takeda Canada. He has been on speaker's bureau of Abbvie Canada Ferring Canada, and Shire Canada.
      Christine Peschken receives research funding from: Canadian Institutes of Health Research.
      Carol Hitchon receives research funding from: Canadian Institutes of Health Research.
      Ruth Ann Marrie receives research funding from: Canadian Institutes of Health Research, Research Manitoba, Multiple Sclerosis Society of Canada, Multiple Sclerosis Scientific Foundation, National Multiple Sclerosis Society, Crohn's and Colitis Canada, the Waugh Family Chair in Multiple Sclerosis and has conducted clinical trials funded by Sanofi-Aventis.

      Acknowledgement

      This study was funded by the Canadian Institutes of Health Research (MOP 119318) and the Waugh Family Chair in Multiple Sclerosis. The sponsors had no role in the study design, collection, analysis or interpretation of data, writing of the article, or the decision to submit it for publication.

      Appendix A. Supplementary material

      References

        • Begg M.D.
        • Parides M.K.
        Separation of individual-level and cluster-level covariate effects in regression analysis of correlated data.
        Stat. Med. 2003; 22: 2591-2602
        • Broemeling A.
        • Watson D.E.
        • Prebtani F.
        • on behalf of the Councillors of the Health Outcomes Steering Committee of the Health Council of Canada
        Population patterns of chronic health conditions, co-morbidity and health care use in Canada: implications for policy and practice.
        Healthc. Q. 2008; 11: 70-76
      1. Canadian Institute for Health Information, 2014a. Data Quality Documentation: Discharge Abstract Database 2013–2014.

      2. Canadian Institute for Health Information, 2014b. Discharge Abstract Database (DAD) Metadata.

        • Cuthbertson B.H.
        • Roughton S.
        • Jenkinson D.
        • Maclennan G.
        • Vale L.
        Quality of life in the five years after intensive care: a cohort study.
        Crit. Care. 2010; 14: R6
        • Donaldson G.C.
        • Seemungal T.A.
        • Bhowmik A.
        • Wedzicha J.A.
        Relationship between exacerbation frequency and lung function decline in chronic obstructive pulmonary disease.
        Thorax. 2002; 57: 847-852
        • Evans C.
        • Kingwell E.
        • Zhu F.
        • Oger J.
        • Zhao Y.
        • Tremlett H.
        Hospital admissions and MS: temporal trends and patient characteristics.
        Am. J. Manag. Care. 2012; 18: 735-742
        • Feemster L.C.
        • Cooke C.R.
        • Rubenfeld G.D.
        • Hough C.L.
        • Ehlenbach W.J.
        • Au D.H.
        • et al.
        The influence of hospitalization or intensive care unit admission on declines in health-related quality of life.
        Ann. Am. Thorac. Soc. 2015; 12: 35-45
        • Fransoo R.
        • Yogendran M.
        • Olafson K.
        • Ramsey C.
        • McGowan K.L.
        • Garland A.
        Constructing episodes of inpatient care: data infrastructure for population-based research.
        BMC Med. Res. Methodol. 2012; 12: 133
        • Gill T.M.
        • Allore H.G.
        • Gahbauer E.A.
        • Murphy T.E.
        Change in disability after hospitalization or restricted activity in older persons.
        JAMA. 2010; 304: 1919-1928
        • Hardin J.
        • Hilbe J.
        General Estimating Equations.
        Chapman & Hall/CRC, Boca Raton2003
        • Kanner R.E.
        • Anthonisen N.R.
        • Connett J.E.
        Lower respiratory illnesses promote FEV(1) decline in current smokers but not ex-smokers with mild chronic obstructive pulmonary disease: results from the lung health study.
        Am. J. Respir. Crit. Care Med. 2001; 164: 358-364
        • Karamyan A.
        • Dünser M.W.
        • Wiebe D.J.
        • Pilz G.
        • Wipfler P.
        • Chroust V.
        • et al.
        Critical illness in patients with multiple sclerosis: a matched case-control study.
        PLoS One. 2016; 11: e0155795
        • Kurtzke J.F.
        Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS).
        Neurology. 1983; 33: 1444-1452
        • Marrie R.A.
        • Rudick R.
        • Horwitz R.
        • Cutter G.
        • Tyry T.
        • Campagnolo D.
        • et al.
        Vascular comorbidity is associated with more rapid disability progression in multiple sclerosis.
        Neurology. 2010; 74: 1041-1047
        • Marrie R.A.
        • Bernstein C.N.
        • Peschken C.A.
        • Hitchon C.A.
        • Chen H.
        • Fransoo R.
        • et al.
        Intensive care unit admission in multiple sclerosis: increased incidence and Increased mortality.
        Neurology. 2014; 82: 2112-2119
        • Marrie R.A.
        • Elliott L.
        • Marriott J.
        • Cossoy M.
        • Blanchard J.
        • Tennakoon A.
        • et al.
        Dramatically changing rates and reasons for hospitalization in multiple sclerosis.
        Neurology. 2014; 83: 929-937
        • Marrie R.A.
        • Bernstein C.N.
        • Peschken C.A.
        • Hitchon C.A.
        • Chen H.
        • Fransoo R.
        • et al.
        Health care utilization before and after intensive care unit admission in multiple sclerosis.
        Mult. Scler. Relat. Disord. 2015; 4: 296-303
        • Marrie R.A.
        • Elliott L.
        • Marriott J.
        • Cossoy M.
        • Tennakoon A.
        • Yu N.
        Comorbidity increases the risk of hospitalizations in multiple sclerosis.
        Neurology. 2015; 84: 350-358
        • Marrie R.A.
        • Miller A.
        • Sormani M.P.
        • Thompson A.
        • Waubant E.
        • Trojano M.
        • et al.
        Recommendations for observational studies of comorbidity in multiple sclerosis.
        Neurology. 2016; 86: 1446-1453
        • McDonald W.I.
        • Compston A.
        • Edan G.
        • Goodkin D.
        • Hartung H.-P.
        • Lublin F.D.
        • et al.
        Recommended diagnostic criteria for multiple sclerosis: guidelines from the international panel on the diagnosis of multiple sclerosis.
        Ann. Neurol. 2001; 50: 121-127
        • Mustard C.A.
        • Derksen S.
        • Berthelot J.-M.
        • Wolfson M.
        Assessing ecologic proxies for household income: a comparison of household and neighbourhood level income measures in the study of population health status.
        Health Place. 1999; 5: 157-171
        • Neuhaus J.M.
        • Kalbfleisch J.D.
        Between- and within-cluster covariate effects in the analysis of clustered data.
        Biometrics. 1998; 54: 638-645
        • Pittock S.J.
        • Mayr W.T.
        • McClelland R.L.
        • Jorgensen N.W.
        • Weigand S.D.
        • Noseworthy J.H.
        • et al.
        Change in MS-related disability in a population-based cohort: a 10-year follow-up study.
        Neurology. 2004; 62: 51-59
        • Polman C.H.
        • Reingold S.C.
        • Edan G.
        • Filippi M.
        • Hartung H.P.
        • Kappos L.
        • et al.
        Diagnostic criteria for multiple sclerosis: 2005 revisions to the McDonald criteria.
        Ann. Neurol. 2005; 58: 840-846
        • Polman C.H.
        • Reingold S.C.
        • Banwell B.
        • Clanet M.
        • Cohen J.A.
        • Filippi M.
        • et al.
        Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria.
        Ann. Neurol. 2011; 69: 292-302
        • Poser C.M.
        • Paty D.W.
        • Scheinberg L.
        • McDonald W.I.
        • Davis F.A.
        • Ebers G.C.
        • et al.
        New diagnostic criteria for multiple sclerosis: guidelines for research protocols.
        Ann. Neurol. 1983; 13: 227-231
        • Shah F.A.
        • Pike F.
        • Alvarez K.
        • Angus D.
        • Newman A.B.
        • Lopez O.
        • et al.
        Bidirectional relationship between cognitive function and pneumonia.
        Am. J. Respir. Crit. Care Med. 2013; 188: 586-592
        • Shirani A.
        • Zhao Y.
        • Karim M.
        • Evans C.
        • Kingwell E.
        • van der Kop M.
        • et al.
        Association between use of interferon beta and progression of disability in patients with relapsing-remitting multiple sclerosis.
        JAMA. 2012; 308: 247-256
        • van den Akker M.
        • Buntinx F.
        • Metsemakers J.F.
        • Roos S.
        • Knottnerus J.A.
        Multimorbidity in general practice: prevalence, incidence, and determinants of co-occurring chronic and recurrent diseases.
        J. Clin. Epidemiol. 1998; 51: 367-375
        • Vollmer T.
        The natural history of relapses in multiple sclerosis.
        J. Neurol. Sci. 2007; 256: S5-13
        • Zhang T.
        • Tremlett H.
        • Leung S.
        • Zhu F.
        • Kingwell E.
        • Fisk J.D.
        • et al.
        Examining the effects of comorbidities on disease-modifying therapy use in multiple sclerosis.
        Neurology. 2016; 86: 1287-1295