Использование услуг здравоохранения 2-3 За десять лет до первого события, связанного с множественным склерозом.

Introduction

Chronic conditions may start years before their classical manifestation, often preceded by nonspecific signs and symptoms, forming a prodromal phase [ 1 , 2 ]. The neurodegenerative condition, Parkinson's disease, has a well‐recognized prodromal phase lasting decades before the onset of classical motor symptoms [ 3 ]. In contrast, multiple sclerosis (MS), a neurodegenerative, immune‐mediated disease, was initially assumed not to have a prodrome [ 4 ]. However, recent studies have demonstrated higher rates of healthcare use, including for psychiatric conditions, gastrointestinal diseases, fatigue, anemia, and pain, in the 5–10 years before MS onset compared to a matched population [ 5 , 6 , 7 , 8 , 9 ]. Supporting biological evidence of preclinical disease includes rising serum neurofilament levels 6–10 years before MS onset versus matched controls [ 10 , 11 , 12 , 13 ].

Yet significant uncertainties remain, including how long the increased healthcare usage occurs before MS onset and what its defining features are as follows: Most studies have examined only a limited period before MS onset [ 5 , 6 , 8 , 14 ], or were unable to delineate its key characteristics [ 15 ]. Studies have also been limited in their sources of healthcare data, often only accessing primary care information [ 9 , 16 ]. This is despite the importance of secondary care, including the emergency department (ED), which can provide a critical resource for emerging health issues, and may help identify early warning signs of MS [ 17 ].

We accessed and characterized population‐based primary and secondary healthcare‐related administrative data, including hospital, physician, and ED visit‐related diagnoses up to 29 years before MS onset (the first demyelinating event) compared to a matched population without MS.

Methods

Study Design, Setting, and Data Sources

We conducted a matched retrospective cohort study using population‐based health administrative data from 1991 to 2020 from Canada's most populous province, Ontario (14.6 million residents). Comprising approximately 39% of Canada's population, Ontario's publicly funded universal health insurance provides residents with coverage for all medically necessary hospital and physician care.

Study databases (and measures) were accessed through ICES and included: the Ontario Health Insurance Plan (dates of physician claims and diagnoses recorded using International Classification of Disease, ICD‐9 codes); the Discharge Abstract/Same Day Surgery and the National Ambulatory Care Reporting System (admission/discharge dates for hospital and ED visits and diagnoses recorded as ICD‐9 codes, then ICD‐10 with Canadian modifications [CA] from April 1, 2002); the Physician Database (physician specialty); and the Registered Persons Database (sex, postal code and dates of birth, death and health insurance coverage), Table  S1 . Postal codes were linked to Statistics Canada's census geographic data and Postal Code OM Conversion File Plus to provide each individual's income quintile (area‐level) and urban/rural residency [ 18 ]. Databases were linked using unique encoded identifiers and analyzed at ICES. Data were available/accessed from April 1, 1991 to March 1, 2020, except for ED visits (available from July 1, 2000). The study end date was chosen, in part, to avoid SARS‐CoV‐2/COVID‐19‐related changes in healthcare use.

Study Cohorts

Persons with MS (PwMS) were identified using a validated algorithm, requiring ≥ 3 hospital or physician claims (any combination) for MS (ICD‐9/10‐CA:340/G35) [ 19 ]. For those meeting this definition, the earliest ICD‐9/10‐CA code for demyelinating disease (Table  S2 ) formed the index date (“MS onset”). At least 5 years of residency in the province (i.e., eligible for healthcare coverage for ≥ 90% of each year, or from birth if < 5‐years old) with no MS/demyelinating disease‐related codes pre‐index date were required.

Each PwMS was matched with up to five persons without any MS/demyelinating codes, without replacement, from the general population by sex, exact birth year, first three digits of the postal code at the index year, and residency time in province pre‐index date and through until their match fulfilled the MS algorithm (i.e., the 3rd MS diagnostic code date).

Outcomes

We characterized healthcare use preceding the index date for the MS versus matched cohort by estimating rates of use by ICD chapters for each of the hospital, physician, and ED visits and by medical specialty for the physician visits (Tables  S3 and S4 ).

Analysis

Cohort characteristics were summarized with descriptive statistics at the index date, including subjects' age, sex, area‐level neighborhood income quintile, urban/rural region of residence, and index year. Rates of each healthcare use outcome (ICD chapters for hospital, physician and ED visits, and physician specialty) were compared between PwMS and the matched cohorts pre‐index date, initially as all years combined, then yearly, up to 29 years pre‐index date for hospital and physician visits and 20 years for ED visits. We fit overdispersed‐Poisson regression models with person time (natural logarithm) as an offset. For the initial analyses (all years combined pre‐index date) we used generalized estimating equations with an exchangeable working correlation matrix and robust standard error estimate, and, for the yearly, standard errors were adjusted by scaled Pearson's chi‐square. Rate ratios (RR) and 95% confidence intervals (CI) were estimated. Secondary analyses included stratification by sex and age groups (18 < 30, 30– < 50, ≥ 50 years) for all the years combined pre‐index date. Post hoc examination of physician visits included bubble charts depicting individual ICD codes that most commonly contributed to four ICD chapters: three chapters with the highest RRs and infection‐related visits because of its special interest in MS (and relevance as both a risk factor/trigger of MS and potential prodromal feature) [ 6 , 20 ]. SAS_V9.4 (SAS Institute, Cary, USA) and R_V4.0.5 (R Foundation, Vienna, Austria) were used to conduct analyses.

Results

The hospital and physician visit‐related analyses included 35,018 PwMS and 136,007 people in the matched cohort (Table  1 ). The average age at index date (MS onset) was 43 (SD = 14) years, more than two‐thirds of the cohorts were female, and most resided in urban areas. The cohorts showed a comparable distribution for neighborhood income quintile and average follow‐up time pre‐index date (approximately 15 [SD = 7] years). For the ED visit‐related analyses, the characteristics of the 19,598 PwMS and 73,166 matched subjects included were similarly comparable (Table  S5 ). Further details on these cohorts are available elsewhere [ 15 ].

TABLE 1 Characteristics of the multiple sclerosis (MS) and matched cohort drawn from the general population at the index date (first demyelinating event) during the study period (1991–2020), Ontario, Canada. Characteristic, n (%) unless otherwise stated MS cohort N  = 35,018 Matched cohort N  = 136,007 Sex Female 24,239 (69.2%) 94,671 (69.6%) Male 10,779 (30.8%) 41,336 (30.4%) Age, years: Mean (SD) 42.7 (14.1) 42.9 (13.6) < 30 6368 (18.2%) 22,818 (16.8%) 30–< 50 18,583 (53.1%) 73,608 (54.1%) ≥ 50 10,067 (28.7%) 39,585 (29.1%) Calendar year 1991–2000 8984 (25.7%) 37,190 (27.3%) 2001–2010 14,727 (42.1%) 57,200 (42.1%) 2011–2020 11,307 (32.3%) 41,617 (30.6%) Neighborhood income quintile
a

1 (least affluent) 6361 (18.2%) 23,483 (17.3%) 2 6658 (19.0%) 26,099 (19.2%) 3 7287 (20.8%) 28,204 (20.7%) 4 7387 (21.1%) 29,348 (21.6%) 5 (most affluent) 7325 (20.9%) 28,873 (21.2%) Rurality of residence
b

Rural 4189 (12.0%) 17,434 (12.8%) Urban 30,829 (88.0%) 118,573 (87.2%)
Note: 3127 MS participants had 1 general population match; 4162 had 2 matches; 4651 had 3 matches; 4787 had 4 matches; 18,291 had 5 matches. Abbreviation: SD, standard deviation.
a
Those with missing neighborhood income status (0.3%; 103 MS; 423 matched cohort) were assigned to category 3.
b
Those with missing residential postal code (0.1%; 45 MS and 100 matched cohort) were assigned to urban status.

Characteristics of the multiple sclerosis (MS) and matched cohort drawn from the general population at the index date (first demyelinating event) during the study period (1991–2020), Ontario, Canada.

Note: 3127 MS participants had 1 general population match; 4162 had 2 matches; 4651 had 3 matches; 4787 had 4 matches; 18,291 had 5 matches.

Abbreviation: SD, standard deviation.

Those with missing neighborhood income status (0.3%; 103 MS; 423 matched cohort) were assigned to category 3.

Those with missing residential postal code (0.1%; 45 MS and 100 matched cohort) were assigned to urban status.

Healthcare Visits by ICD Chapter Before the Index Date

Comparing the MS and matched cohorts, the nervous system ‐related RRs (range = 3.21–5.48) were the highest across all visit types (physician, hospital, and ED) in the years pre‐index date (Figure  1 ), followed by mental disorders and ill‐defined signs / symptoms which were > 1.70 across all healthcare visit types (RRs = 1.71–2.11). Musculoskeletal , sensory , digestive , circulatory, infection‐related, genitourinary, injury‐related, endocrine / metabolic issues, blood , s kin , and respiratory‐related visits all reached 1.17 or more (all RRs = 1.18–2.08).

(Panels A–C) The associations between multiple sclerosis (MS) and diagnoses grouped by ICD chapters in the years before a first demyelinating event (the index date) resulting from physician, hospital, and emergency department visits, Ontario, Canada (1991–2020). Panel A (left): Physician visits. Panel B (middle): Hospital visits. Panel C (right): ED visits. Healthcare use is shown as all years combined pre‐index date. Pregnancy, childbirth (depicted last in each panel) included females only. “Other” health system contact available for the hospital and ED visits only. The y axis summarizes the diagnoses by ICD chapter linked to each visit (Table  S3 provides further details). The x axis shows the rate ratios (RR) and corresponding 95% confidence intervals (CI); a RR > 1 (right of the red vertical line) indicates higher healthcare use in the MS cohort relative to the matched cohort. An arrow indicates that the 95% CI extends beyond the edge of the scale.

When examined yearly, up to 29 years pre‐index date were available for hospital and physician visits and 20 years for ED visits (Figure  S1 ). The highest RRs reached for PwMS versus the matched cohort were for the nervous system , peaking in the year pre‐index for all three healthcare types (RR range = 12.06–17.13). Nervous system‐ related RRs were also consistently elevated from year 24 through to the year pre‐index for physician (exceeding 1.47 in each of those years), and from years 13 and 8 for hospital and the ED, respectively. These elevations were preceded by some other ICD chapters. For example, mental health ‐related RRs were consistently elevated for physician visits from year 28 pre‐index (where they exceeded 1.29 in each of these years), and were elevated by > 1.42 for almost each of the 20 years pre‐index date for hospital and 9 years for ED visits. The magnitude of the association rose as the index date approached across all healthcare types, with the peak RR ranging between 2.14 and 2.70 in the year pre‐index. Ill‐defined signs / symptoms ‐related RRs were consistently elevated for almost three decades pre‐index (from year 28) for physician visits and from years 14 to 16 for hospital and ED visits. All RRs for the three visit types were ≥ 1.33 from years 12 to 15 pre‐index before rising steadily and peaking the year pre‐index (RRs range = 3.51–5.45). Other ICD chapter‐related RRs exhibiting sustained elevations which rose as the index date approached included, for example, digestive system ‐related which were persistently high, from year 22 pre‐index for physician, year 20 for hospital, and year 16 pre‐index for ED visits, with all three exceeding 1.24 from years 13 to 19 pre‐index. Respiratory systems ‐related visit RRs were consistently elevated from year 24 pre‐index for physician (RR > 1.07), and years 15–16 for hospital (RR > 1.31) and the ED (RR > 1.18). Musculoskeletal and injury‐related physician visit RRs were also consistently elevated from year 24, peaking between 2.13 and 2.32 in the year pre‐index. Infection‐ related physician visits, RRs were consistently higher for PwMS versus the matched cohort by > 1.09 in each of the 22 years pre‐index date, peaking at 1.73 (95% CI: 1.64–1.83) in the year pre‐index. ED and hospital visit RRs consistently exceeded 1.25 in the 13 years (ED) and 4 years (hospital) pre‐index, with both peaking in the year pre‐index (range = 1.83–2.59). Finally, circulatory system RRs were consistently, significantly elevated from year 21 pre‐index date for physician visits, reaching 1.22 or more in each of the 12 years pre‐index, peaking at 2.23 (95% CI: 2.13–2.34) in the year pre‐index date. The Supplemental Results summarize the remaining ICD chapters, with salient examples provided here only. Briefly, ED visit RRs for endocrine / metabolic‐ related issues were particularly high and sustained (versus hospital or physician visits), exceeding 1.56 from year 10 pre‐index date, peaking at 3.14 (95% CI: 2.32–4.26) in the year pre‐index. Sense organ‐ related ED RRs were particularly high in the year pre‐index date, reaching 7.57 (95% CI: 6.96–8.23).

Visits by Physician Specialty Before the Index Date

RRs were significantly higher for PwMS versus the matched cohort across all 13 medical specialties examined pre‐index date (Figure  2 ). Visit rates were highest for neurology (RR = 5.38), neurosurgery (RR = 3.27) and physiatry (RR = 2.80), followed by psychiatry, urology, internal medicine , and emergency‐ related specialties (RRs = 1.58–1.72).

The associations between multiple sclerosis (MS) and physician visits in the years before a first demyelinating event (the index date) by medical specialty, Ontario, Canada (1991–2020). Healthcare use is shown as all years combined pre‐index date. Obstetrics and gynecology (depicted last) included females only. The y axis summarizes the medical specialty visited (Supplemental Table  S4 provides further details). The x axis shows the rate ratios (RR) and corresponding 95% confidence intervals (CI); a RR > 1 indicates higher healthcare use in the MS cohort relative to the matched cohort. An arrow indicates that the 95% CI extends beyond the edge of the scale.

Yearly visit RRs were particularly high and sustained for PwMS versus the matched cohort (Figure  S2 ) for primary care which were elevated in each of the 28 (of 29) years pre‐index, exceeding 1.22 by year 11, peaking in the year pre‐index at 1.93 (95% CI: 1.90–1.95). Internal medicine RRs were 1.20 or more in each of the 25 years pre‐index date, gradually rising, peaking at 2.60 (95% CI: 2.51–2.70) in the year pre‐index. Psychiatrists and neurologist visit RRs were consistently > 1.52 in each of the 24 years pre‐index date, peaking in the year pre‐index ( psychiatrists  = 1.90, 95% CI: 1.74–2.08); neurologist  = 21.91, 95% CI: 20.69–23.20). Ophthalmologist / otolaryngologist RRs were consistently elevated from year 20 pre‐index, exceeding 1.31 by year 13, peaking at 2.68 (95% CI: 2.58–2.78) in the year pre‐index. Physiatrist visit RRs pre‐index were rather variable, but were consistently > 1.75 in each of 12 years pre‐index date, peaking at 7.61 (95% CI: 6.55–8.83) in the year pre‐index.

Secondary Analyses: Sex and Age Stratification

Males exhibited higher RRs than females across several ICD chapters and medical specialties in the years pre‐index date (Figures  S3 and S4 ). By ICD chapter, nervous system‐ related visits were particularly high for all visit types (e.g., physician: males = 9.31, females = 4.37 and ED: males = 5.02, females = 2.82). Males also exhibited higher RRs (2.19) for mental health‐ related ED visits than females (RR = 1.64), and genitourinary ‐related physician and hospital visits (e.g., hospital: males = 1.67, females = 1.40). Physician and ED visits for ill‐defined signs / symptoms were also higher among males (e.g., ED = 2.28 vs. females = 2.05). By medical specialty, these included (ordered by RR): neurology (males = 6.57, females = 4.95); neurosurgery (males = 3.92, females = 2.91); physiatry (males = 3.61, females = 2.44); and primary care (males = 1.45, females = 1.32). Only urology visits were higher for females (RR = 1.85) than for males (RR = 1.51).

An age‐gradient was observed for some ICD chapters and medical specialties (Figures  S5 and S6 ), with those older at the index date exhibiting higher visit RRs, for example, for blood‐ related physician, hospital, and ED visits (e.g., physician: 18–< 30, 30–< 50, ≥ 50 years = 1.31, 1.45, 1.70). Nervous system ‐related physician visits were particularly high for those oldest at the index date (RRs: < 50‐age groups = 4.42–5.01 vs. ≥ 50 years = 7.45), with no remarkable age‐gradient for hospital or ED visits. Although injury‐ related physician visits differed little by age, hospital visits increased (18–< 30, 30–< 50, ≥ 50 years = 1.19, 1.39, 1.56). Examples where ICD chapters were higher for younger PwMS included circulatory‐ related hospital visit RRs (18–< 30, 30–< 50, ≥ 50 years = 2.13, 1.42, 1.29). Other ICD chapters, including mental health , exhibited minimal differences across age groups. By medical specialty, primary care visit RRs increased modestly across age‐groups (< 50‐age groups = 1.28–1.34 vs. ≥ 50 years = 1.40).

Post hoc descriptive analyses of the individual ICD codes contributing to the four select chapters are depicted as bubble charts (Figure  S7 ). Briefly, for the nervous system , there was an increase in the rate difference [RD] (excess number of physician visits per 1000 person‐years for PwMS versus the matched cohort) for migrain e (ICD‐9 code:346) starting at 20 years pre‐index and for other CNS‐ related diseases (ICD‐9:349) at 15 years pre‐index. The RD increased as the index date approached, peaking in the 5 years pre‐index, particularly for the latter (ICD‐9:349). For mental disorders , the largest RD was for anxiety (ICD‐9:300), evident 30 years pre‐index, peaking in the 5 years pre‐index date. The yearly RD for each of these ICD codes also steadily climbed in the 5 years pre‐index date.

For both ill‐defined signs / symptoms and infection‐ related physician visits, gastrointestinal‐related issues predominated, being evident up to 25 years pre‐index date. For example, nausea / vomiting / heartburn / dysphagia / abdominal pain (ICD‐9:787) and diarrhea / gastroenteritis (ICD‐9:009) RDs rose as the index date approached (in both the five‐yearly and yearly results). In contrast, the RD for infectious mononucleosis (ICD‐9:075) remained relatively stable pre‐index.

Discussion

In this large population‐based matched cohort study, we observed an extended period of higher healthcare use for PwMS, measured using physician, hospital, and ED visits before the first MS‐related demyelinating event (the index date) versus the matched cohort without MS. There was a particularly high and sustained mental health‐related burden; RRs consistently exceeded 29% for mental disorders‐related physician visits from 28 years pre‐index, with visits to the psychiatrist exceeding 52% from 24 years pre‐index date. Other condition‐related visits also rose early and remained high for an extended period, including ill‐defined signs/symptoms and primary care visits, which were elevated for 28 years pre‐index, followed by internal medicine consults (elevated for 25 years), nervous system, musculoskeletal, injury, and respiratory‐related physician visits (24 years) and digestive system (elevated for 22 years pre‐index). Combined, the results show a sustained and durable excess of healthcare use evident across different conditions and physician specialties for over two decades before the first MS‐related demyelinating event. Findings suggest health‐related phenotypical differences between those who develop MS and those who do not early in the disease process.

Although we were unable to find another study examining healthcare use by reason (diagnosis) and physician specialty for such an extended period before MS onset, our findings are consistent with the lower cognitive performance measurable for an extended period—up to 20 years—before MS onset in a male Norwegian military cohort study [ 21 ]. By measuring different types of healthcare use each year, we could also examine the change over time, demonstrating steady rises over time for physician, hospital, and ED visits, escalating as the index date approached. Use of the ED was particularly high for an extended period (relative to physician or hospital visits) pre‐index date for issues, such as endocrine/metabolic‐related diseases, with visit rates consistently exceeding 56% or higher from year 10, and musculoskeletal and genitourinary system‐related issues being 30% higher from years 12 to 18 pre‐index. Although some studies have examined ED use after MS onset or diagnosis [ 22 , 23 ], we found few did so for the period before onset or diagnosis. Of those that did, three small US‐based single center studies (ranging in size from 21 to 108 PwMS) examined the 2–4 years either before MS diagnosis [ 24 , 25 ] or the first demyelinating event [ 26 ]. Of the two examining healthcare use over time (which included ED visits), both found use to increase as these dates approached [ 24 , 26 ]. However, neither reported the reason nor diagnosis linked to the ED visit [ 24 , 26 ]. Moreover, none had a comparator cohort without MS, limiting interpretation of findings [ 24 , 25 , 26 ]. The ED provides a critical source of healthcare [ 27 ]; our findings suggest that understanding patterns across different healthcare types, including ED visits, before MS onset could enhance the ability to identify high‐risk patients earlier.

Our findings greatly extend prior work which was largely limited to a 5 year period pre‐MS onset or did not report the “reasons” for elevations in healthcare use [ 5 , 6 , 7 , 8 , 15 ]. Findings also have research, practice and health policy‐related implications. First, we demonstrate that elevated healthcare use for issues commonly recognized after MS diagnosis are emerging decades earlier than previously thought. Moreover, nonnervous system ICD chapters, such as mental‐health, were among the first (earliest) to rise. Our observations align with, and extend, the concept of the MS continuum [ 28 ]. Findings also position MS within the life‐course model of disease, providing a useful framework and approach to disease management and prevention of MS [ 15 , 29 ]. Drawing on life‐course principles combined with recognized risk factors for MS [ 29 ], an early sensitive period for exposure could start at conception, with factors, such as maternal diabetes [ 30 ], suboptimal sunlight or low serum vitamin D [ 31 ], smoking [ 32 ], higher body size/obesity, and adversity (e.g., physical or psychological trauma and social hardships) [ 33 , 34 , 35 ], increasing susceptibility through into early adulthood. In turn, several of these exposures increase risk for morbidities, such as psychiatric illness [ 36 ], which could contribute to some of the early rises in mental health‐related healthcare observed in our study. Whether there is a specific time‐limited sensitive period is unclear, or if cumulatively, these exposures, along with the critical event—EBV exposure—could trigger MS in a genetically susceptible individual [ 20 ]. Evidence of underlying disease processes lasting some years before classical presentation of MS includes smaller head size and brain volume measurable around MS onset in children and youth [ 37 ]. Second, our findings demonstrate that accessing a breadth of healthcare sectors, including primary, secondary and ED visits may be instrumental in furthering understanding of the period before MS onset. Third, findings suggest that targeted training for specific physicians, such as psychiatrists, internal medicine, primary care physicians or those working in the ED to recognize early signs of MS in high‐risk individuals warrants consideration and further work. As does development of screening tools, prioritizing tools that seamlessly integrate into the existing workflow, minimizing burden [ 38 ]. Use of electronic health records to identify early MS disease activity and predict an MS diagnosis also hold promise [ 39 , 40 ]. These approaches, if rigorously tested in high‐risk individuals and appropriate settings, could enable prompt management of MS and early intervention aimed at minimizing disease progression [ 2 ]. Prospective studies of individuals at high risk of developing MS, such as those with a close relative with MS provide valuable insights [ 41 , 42 , 43 , 44 ]. For example, among monozygotic twins discordant for MS, a distinct peripheral immunological signature was observed which differentiated the twin pairs [ 42 ]. Also, in asymptomatic females with a high ( n  = 27) versus low ( n  = 20) gene–environment risk score for developing MS were found to have lower vibration perception in their great toe (mean [SD] = 2.48 [0.60] vs. 1.83 [0.54], p  = 0.008, adjusted for age, height and test date) [ 44 ]. Which of these individuals will go onto develop MS remains unknown, but findings are intriguing nonetheless. In addition, individuals with radiologically isolated syndrome (RIS, n  = 197) were more likely to visit a psychiatrist (odds ratio, OR: 4.43, 95% CI: 2.79–6.92) and fill a prescription for an anxiolytic, antidepressant or antimigraine drug (ORs: 2.22–2.73) in the 3 years before their diagnosis compared to matched general population controls [ 45 ]. Moreover, the RIS participants who subsequently developed clinically definite MS (versus those who did not) had a higher hazard (risk) of accessing the ED (hazard ratio, HR: 1.87, 1.17–2.96) [ 45 ].

We were also able to examine effects of age and sex on findings in the years pre‐index date. Although healthcare use linked to many chronic diseases increases with age, including cardiovascular [ 46 ], we found a disproportionately higher circulatory system‐related hospitalization rate among younger PwMS. This observation is broadly consistent with a smaller study from British Columbia ( n  = 6863) [ 47 ], albeit reported as higher rates of cardiovascular drug fills before the first demyelinating event [ 47 ]. After an MS diagnosis, a higher prevalence of ischemic heart disease in younger (20–44 year‐olds) versus older (45–59 year‐olds) has also been found [ 48 ]. Whether this reflects, for example, early autonomic symptoms or more MS‐related inflammatory disease processes in younger persons, even before MS onset, or underdiagnosis of circulatory diseases in older PwMS is unclear [ 48 ]. Males had higher visit rates across several conditions, including for mental health‐related ED visits. This concurs with our prior work in British Columbia [ 47 ], and also a large population based study from Canada showing a disproportionately higher prevalence of depression and anxiety in males with MS at the time of an MS diagnosis [ 49 ]. However, neither study included ED visits. If males were more likely than females to have their MS “missed,” this could in part explain findings, including the increase in mental health issues. However, a 2024 systematic review was unable to find a sex difference with respect to diagnostic delay or underdiagnosis [ 50 ]. Further work is warranted to delineate age and sex‐specific effects on the years leading up to MS onset.

Strengths and Limitations

Strengths included the population‐based data examined, with healthcare visits available for everyone regardless of location (i.e., urban or rural) or ability to pay for health services, which are free at the point of access. Data are captured “real time” as a person moves through the health system, limiting recall bias. Further, findings are not subject to volunteer bias and included primary and secondary care, including the ED. However, we are unable to capture health issues that are self‐managed or do not prompt a person to access the health system. This may suggest that our estimates are conservative (underestimate the true rates of healthcare use). Studies in MS have shown that some groups (e.g., racialized groups, rural dwellers, those of lower socioeconomic status) maybe less likely to use the health system due to specific barriers such as access (due to travel distances, lack of public transport), health literacy or personal beliefs [ 51 , 52 , 53 ], which could result in a lower likelihood of receiving a timely diagnosis for their MS or other health conditions. However, for others, low healthcare use could be a result of having fewer health issues. Teasing apart these competing issues is challenging, and the available literature is sparse. It is also possible that persons more likely to be frequent healthcare users are more likely to be diagnosed or even misdiagnosed with MS. Nevertheless, our MS case finding algorithm has a high positive predictive value (98.8%), such that the likelihood of false positives is very low [ 19 ]. Other limitations included lack of access to clinical data from tertiary care MS clinics, such as MRI information or the date of MS symptom onset. Although this latter date, as determined retrospectively by the neurologist, often at the time of diagnosis, may represent the “true” end of a potential prodromal period, it can also be subject to recall bias. Moreover, we previously showed that the time between the neurologist determined symptom onset date and the first demyelinating diagnostic claim in administrative data are relatively short, averaging 1.1 years (SD: 1.74) in Sweden (based on n  = 5835 with both dates recorded) [ 15 , 54 ]. Also in that study, the average age at MS symptom onset was 39 years (SD: 9.8) [ 54 ], just a few years younger than our index date (42 years). Given that our observation period spans 20–30 years before the first demyelinating diagnostic claim, most of the studied time will fall within a clinically relevant putative prodromal phase. Some findings will naturally represent a missed opportunity for “correct” identification of a demyelinating/MS event and an earlier diagnosis, whereas others represent a true feature of a prodromal phase. Either way, findings are relevant to future patient care as they suggest that there are opportunities for earlier recognition of MS. Our study was not designed to examine risk factors for MS onset. Disentangling any capture of such markers would be challenging, although no remarkable patterns were found, for example, for physician visits for infectious mononucleosis, a major risk factor for MS onset [ 55 ]. Although we were able to match each PwMS with individuals in the general population by key characteristics, such as sex, birth year and place of residency, we did not have access to other factors, such as ancestry, which may be of value for future studies to consider. Work from the US suggests, for example that racialized individuals, such as Black or Hispanics can experience diagnostic delay, especially when MS is not suspected [ 51 , 56 , 57 ].

Conclusion

We found a sustained excess of healthcare use evident across a diversity of conditions and physician specialties. These included higher hospital, physician, and ED visits related to mental health, nervous‐system, ill‐defined signs/symptoms, and digestive issues, present for over two decades before the first demyelinating event. Findings suggest differences between those who develop MS and those who do not early in the disease process.

Author Contributions

H.T. conceptualization, funding acquisition, investigation, methodology, project administration, resources, supervision, writing – original draft. F.Z. formal analysis, funding acquisition, investigation, methodology, visualization, writing – review and editing. K.E. data curation, formal analysis, writing – review and editing. A.A. data curation, formal analysis, writing – review and editing. P.L. data curation, formal analysis, writing – review and editing. K.A.M. conceptualization, data curation, funding acquisition, investigation, methodology, writing – review and editing. Y.Z. conceptualization, formal analysis, funding acquisition, investigation, methodology, supervision, writing – review and editing. C.J.M. conceptualisation, funding acquisition, investigation, methodology, project administration, supervision, validation, writing – review and editing. R.A.M. conceptualization, funding acquisition, investigation, methodology, visualization, writing – review and editing.

Disclosure

Feng Zhu, Karl Everett, Ayesha Asaf, Ping Li, and Yinshan Zhao: nothing to report. Kyla A. McKay is funded by the Swedish Research Council, StratNeuro, Bjarne Ahlström's Foundation, and a salary award from Karolinska Institutet. She has received speaker honoraria from Biogen (2022) and Sanofi‐Aventis (2023). Colleen Maxwell receives research funding from a University of Waterloo Research Chair, CIHR, Multiple Sclerosis Society of Canada, National Multiple Sclerosis Society, CMSC, and the Public Health Agency of Canada. Ruth Ann Marrie receives research funding from: CIHR, MS Canada, Crohn's and Colitis Canada, National Multiple Sclerosis Society, CMSC, The Arthritis Society, US Department of Defense, Pfizer Foundation. She is a coinvestigator on a study funded in part by Biogen Idec and Roche (no funds to her). Helen Tremlett has, in the last 5 years, received research support from the Canada Research Chair Program, the National Multiple Sclerosis Society, the Canadian Institutes of Health Research, Multiple Sclerosis Canada, the Multiple Sclerosis Scientific Research Foundation and the EDMUS Foundation (“Fondation EDMUS contre la sclérose en plaques”). In addition, in the last 5 years, has had travel expenses or registration fees prepaid or reimbursed to present at CME conferences or attend meetings (as a member of the International Advisory Committee on Clinical Trials in Multiple Sclerosis) from the Consortium of MS Centers (2023), the Canadian Neurological Sciences Federation (2023), National MS Society (2022, 2025), ECTRIMS/ACTRIMS (2017–2025), American Academy of Neurology (2019). Speaker honoraria are either declined or donated to an MS charity or to an unrestricted grant for use by HT's research group.

Ethics Statement

Institutional ethics approval was provided by the University of British Columbia's Clinical Research Ethics Board. Linkage and analyses were performed at ICES.

Conflicts of Interest

The authors declare no conflicts of interest directly related to this work. Please see 'Disclosures' for other relevant information.

Supporting information

Appendix S1.

Appendix S2.

Acknowledgements

This study was in part supported by ICES (formerly the “Institute for Clinical Evaluative Sciences”), which is funded by an annual grant from the Ontario Ministry of Health (MOH) and the Ministry of Long‐Term Care (MLTC). ICES is an independent, nonprofit research institute whose legal status under Ontario's health information privacy law allows it to collect and analyze healthcare and demographic data, without consent, for health system evaluation and improvement. The use of the data in this project is authorized under section 45 and approved by ICES' Privacy and Legal Office. This document used data adapted from the Statistics Canada Postal Code OM Conversion File, which is based on data licensed from Canada Post Corporation, and/or data adapted from the Ontario Ministry of Health Postal Code Conversion File, which contains data copied under license from Canada Post Corporation and Statistics Canada. Parts of this material are based on data and/or information compiled and provided by the Ontario Ministry of Health and the Canadian Institute for Health Information (CIHI). The analyses, conclusions, opinions and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred. We would like to acknowledge the members of the team who facilitated gaining funding for the wider program of work related to the MS prodrome (Dr M. Ehsan Karim, Dr. José Wijnands and Fardowsa L.A. Yusuf, MSc). JW also contributed to designing the technical plan while a research associate at the University of British Columbia, Canada.

Funding: The study was supported by the National MS Society and MS Canada (RG5063A4/1/RFA‐2103‐37392; EGID: P002/903124; PI: Tremlett). The study funders had no role in the methods, analyses, writing of the manuscript, or the decision to submit it for publication.

Funding Statement

This work was funded by
Multiple Sclerosis Society of Canada
grant EGID: P002/903124;
National Multiple Sclerosis Society
grant RG5063A4/1/RFA‐2103‐37392.

Data Availability Statement

Data sharing agreements prohibit public availability. Access maybe granted via ICES. The dataset from this study is held securely in coded form at ICES. Although legal data sharing agreements between ICES and data providers (e.g., healthcare organizations and government) prohibit ICES from making the dataset publicly available, access may be granted to those who meet prespecified criteria for confidential access, available at www.ices.on.ca/DAS (email: das@ices.on.ca ). The full dataset creation plan and underlying analytic code are available from the authors upon request, understanding that the computer programs may rely upon coding templates or macros that are unique to ICES and are therefore either inaccessible or may require modification.

References

Associated Data

Supplementary Materials

Appendix S1.

Appendix S2.

Data Availability Statement

Data sharing agreements prohibit public availability. Access maybe granted via ICES. The dataset from this study is held securely in coded form at ICES. Although legal data sharing agreements between ICES and data providers (e.g., healthcare organizations and government) prohibit ICES from making the dataset publicly available, access may be granted to those who meet prespecified criteria for confidential access, available at www.ices.on.ca/DAS (email: das@ices.on.ca ). The full dataset creation plan and underlying analytic code are available from the authors upon request, understanding that the computer programs may rely upon coding templates or macros that are unique to ICES and are therefore either inaccessible or may require modification.