Продромальная фаза множественного склероза: данные, полученные в результате отсутствия по болезни до возникновения болезни - соответствующее когортное исследование.

WHAT IS ALREADY KNOWN ON THIS TOPIC

Research has shown that individuals in the prodromal phase of multiple sclerosis (MS) often experience increased healthcare utilisation, such as more frequent physician visits and higher rates of hospitalisations.

WHAT THIS STUDY ADDS

This study provides robust evidence of a significant increase in sickness absence rates in the years leading to an MS onset. The increase in sickness absence may signal early health deterioration before the clinical onset of MS. These findings contribute to the growing body of knowledge on the prodromal phase of MS and the understanding of early disease impact.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

These findings may suggest that monitoring work disability patterns as potential early indicators of MS allows for more timely intervention and support. Sickness absence data can be a valuable tool for gaining a deeper understanding of the early effects of MS on individuals’ lives, offering new avenues for research into the disease’s prodromal phase.

Introduction

Multiple sclerosis (MS) is an immune-mediated disease of the central nervous system characterised by demyelination, inflammation and neurodegeneration. While substantial progress has been made in understanding clinical features, diagnosis and treatment of MS, the preclinical or prodromal phase remains poorly delineated and a challenging area of research. The prodromal phase occurs before the first classical manifestation of the disease, during which subtle pathological processes may occur. Examining this phase is essential for a more complete understanding of MS and its aetiology by bringing us closer to the biological onset of MS. It also offers the potential for early intervention and improved quality of life for individuals affected by MS. 1 2 Recent studies have provided intriguing clues about the earliest signs of MS, indicating that it is characterised by specific neurological, immunological and systemic changes before the classical clinical onset of MS. 1 6 Detecting these subtle indicators is challenging due to their vague nature and lack of disease specificity. Therefore, innovative methodologies and measures are needed to help us understand the features of the prodromal phases of MS.

Employment-related sickness absence data 7 8 present a novel measure to investigate the prodromal phase of MS. Sickness absence data can serve as a sensitive indicator of an individual’s overall health and well-being. 9 While healthcare use, such as the number of physician visits, prescriptions filled or hospitalisations, also indicates health issues, sickness absence can reflect a broader range of health impacts, including preventative care taken by individuals to manage or mitigate their emerging health problems. Additionally, individuals might be granted sickness absence for symptoms that are not severe enough to warrant medical treatment, specialised care or hospitalisation, thus providing important early signals of potential health issues.

This work aimed to examine the prodromal phase of MS by using sickness absence data from Sweden. Such information can help us better understand the early symptoms and potential triggers of MS.

Methods

Data sources and identification of individuals with MS

We used the Swedish National Inpatient and Outpatient Register, maintained by the National Board of Health and Welfare (Socialstyrelsen), 10 and the Swedish MS Register 11 12 to identify MS cases. We accessed the Swedish Total Population Register, maintained by Statistics Sweden (Statistikmyndigheten), 13 and the Swedish National Cause of Death Register 14 for the selection of controls and to gather essential information required to calculate follow-up time, such as date of birth, death, immigration and emigration status. We obtained socioeconomic information from the Swedish Longitudinal Integrated Database for Health Insurance and Labour Market Studies, including annual individual-level income. 15

We constructed two study cohorts: health administrative and clinical. In the health administrative cohort, we identified MS cases via a validated algorithm requiring ≥3 hospitals or physicians International Classification of Diseases version 10 (ICD10) diagnostic codes for MS. 16 In the clinical cohort, MS cases were identified as those registered in the Swedish MS registry and diagnosed by an MS neurologist. 11 The index date (‘MS onset’) in the health administrative cohort was determined as the earliest date of the first MS or demyelinating disease ICD code suggestive of MS ( online supplemental eTable 1 ). In the clinical cohort, the index date was the MS symptom onset date, as determined by an MS neurologist, or the first ICD code for the first MS/demyelinating disease if earlier. The MS diagnosis date was defined as the date of the third MS ICD code in the health administrative cohort and the recorded diagnosis date in the Swedish MS registry for the clinical cohort to be resident in Sweden for at least 5 years before the index date through until MS diagnosis ( online supplemental eFigure 1 ).

Selection of controls

Cases were matched by the residency time, sex, birth year and county of residence at the index year, with up to five randomly selected controls (without replacement) from the Swedish general population with no record of MS or any demyelinating disease codes ever. The controls were assigned the same index and diagnosis dates as their matched case, and they had to be residents until their case’s date of diagnosis. Since sickness absence eligibility is restricted to working-age individuals, all cases and controls had to be 19 years or older, at least in the 5-year preindex date (the minimum follow-up period), 17 such that the age at the index date for those included was ≥24 years.

Follow-up time

The study period was 1 January 2001–31 December 2019, with 2001 marking the initiation of nationwide data collection in the Swedish outpatient register. The follow-up time for assessing sickness absence annual rates was from 1 January 2001 (or the date of immigration to Sweden) in individuals 19 years or older until each participant’s index date.

Sickness absence and disability pension data

The Swedish Social Insurance Agency (Försäkringskassan) administers benefits and compiles individual-level data on periods of sickness absence. 18 In Sweden, sickness absence benefits can be granted to working-age residents with earnings or unemployment benefits, provided they have reduced work capacity due to illness or injury. Individuals are entitled to receive sickness absence benefits from day 15 of the sickness absence spell if employed (the employer is responsible for paying days 2–14) or day 2 if the individual is unemployed. As a result, the data collected by the agency do not cover spells lasting less than 15 days if the individual is employed. Sickness absence covers regular working hours missed as a whole day (100%) or partial day (25%, 50% or 75%). To make the spells comparable, the gross days of sickness absence are converted into net days. For instance, two gross days at 50% are counted as one net day. The net days were rounded to the nearest integer.

Statistical analysis

Yearly rates of net sickness absence days (converted from the part-time spells), up to 18 years preceding the index date, were compared between cases and controls using negative binomial regression models fitted via generalised estimating equations with an exchangeable working correlation matrix. All models included the logarithm of residency time during a given year as an offset. Additional models with interaction terms were constructed to examine heterogeneity across sociodemographic variables. These variables included sex, age at the index categorised into four groups (24–34, 35–44, 45–54 and 55–65), socioeconomic status quantiles derived from individual disposable income at the index date (ranging from the 1st quantile representing the most disadvantaged to the 5th quantile signifying the most advantaged), and county of residence at the index date, classified into ‘urban’ and ‘rural’ areas. Urban areas included counties with a population density ≥ the median Swedish county population density, based on data from Statistics Sweden. Rural areas included counties with a population density below the median. Immigration status was defined as ‘immigrant’ (individuals with a history of immigration to Sweden) and ‘non-immigrant’ (ie, those born in Sweden).

Sensitivity analyses included using gross sickness absence days. Findings were reported as adjusted rate ratios (aRRs) and 95% CIs. Analyses were conducted using R (V.4.3.1, R Foundation for Statistical Computing, Vienna, Austria) and SAS V.9.4 (SAS Institute).

Results

Cohorts

The health administrative cohort comprised 8618 MS cases and 43 072 matched controls, with an average age at the index date of 42.0 (SD: 10.8, table 1 ). The clinical cohort included 6361 MS cases and 31 776 matched controls. The average age at the index date was 39.0 (SD: 9.8), 3 years younger than the health administrative cohort. Females represented 69% of each cohort. The MS cases largely overlapped between the two cohorts, with 66.2% of cases in the health administrative cohort represented in the clinical cohort. Conversely, 89.7% of the cases in the clinical cohort were represented in the health administrative cohort.

Table 1 Characteristics of the multiple sclerosis (MS) cases and matched controls in the health administrative and clinical cohorts, Sweden (2001–2019) Health administrative cohort Clinical cohort MS N=8618 Control N=43 072 MS N=6361 Control N=31 776 Sex  Female 5908 (69%) 29 517 (69%) 4329 (68%) 21 634 (68%) Age at index  Mean (SD) 42.0 (10.8) 42.0 (10.7) 39.0 (9.8) 39.0 (9.8) Age at Index (categories)  24–34 2652 (31%) 13 265 (31%) 2537 (40%) 12 673 (40%)  35–44 2590 (30%) 12 992 (30%) 2054 (32%) 10 316 (32%)  45–54 2129 (25%) 10 575 (25%) 1307 (21%) 6456 (20%)  55–65 1247 (14%) 6240 (14%) 463 (7%) 2331 (7%) Index year  2001–2010 3560 (41%) 17 804 (41%) 2430 (38%) 12 161 (38%)  2011–2019 5058 (59%) 25 268 (59%) 3931 (62%) 19 615 (62%) Socioeconomic status (quantiles)  1 (least affluent) 909 (11%) 3940 (9.1%) 642 (10%) 3192 (10%)  2 1415 (16%) 6006 (14%) 951 (15%) 4462 (14%)  3 2301 (27%) 10 640 (25%) 1727 (27%) 7939 (25%)  4 2196 (25%) 11 797 (27%) 1701 (27%) 8597 (27%)  5 (most affluent) 1797 (21%) 10 689 (25%) 1340 (21%) 7586 (24%) Immigrant status  Immigrant 914 (11%) 4381 (10%) 662 (10%) 3193 (10%) Area of residence  Rural 2090 (24%) 10 457 (24%) 1467 (23%) 7329 (23%) Follow-up  Mean (SD) 10.8 (3.5) 10.9 (3.6) 10.8 (3.6) 10.9 (3.6)

Characteristics of the multiple sclerosis (MS) cases and matched controls in the health administrative and clinical cohorts, Sweden (2001–2019)

Sickness absence

In the health administrative cohort, during the year before the index date (ie, MS onset), the average number of net sickness absence days was 20.1 (SD: 53.7) for MS cases and 7.3 (SD: 32.8) for controls. Additionally, 28.7% of the MS cases had at least one net day of sickness absence, compared with 7.3% of the controls. In the clinical cohort, the average number of net sickness absence days in the year before the index date was 9.2 (SD: 36.1) for MS cases and 6.9 (SD: 31.9) for controls. In this cohort, 15.4% of the MS cases had at least one net day of recorded sickness absence, compared with 11.4% of the controls.

Rates of sickness absence

In the administrative cohort, the aRRs for sickness absence in the most recent 6 years preceding the index date revealed statistically significantly higher rates of sickness absence in cases than controls. In the year before the index date, cases exhibited a substantially higher sickness absence rate, with an aRR of 2.59 (95% CI 2.40 to 2.79). The elevated rate persisted in the second year before the index date, with an aRR of 1.44 (95% CI 1.31 to 1.59). The differences in sickness absence rates became less pronounced further away from the index date, and none were statistically significant after the tenth year before the index. While a similar pattern was observed in the smaller clinical cohort, findings were more modest, reaching statistical significance only 2 years before the index date ( figure 1 and online supplemental eTable 2 ). Results remained largely the same when using gross sickness absence days ( online supplemental eTable 3 ).

Adjusted rate ratios for sickness absence in the prodromal phase of MS. Each point represents the rate ratio for a given year, with vertical bars indicating 95% CIs. MS, multiple sclerosis.

Effect of age and sex on sickness absence rates

We observed age-related differences in sickness absence rates in both cohorts, primarily in the year before MS onset. In the health administrative cohort, while individuals aged 24–34 showed an aRR of 1.94 (95% CI 1.65 to 2.28) in the year before MS onset, the aRR was 2.44 (95% CI 2.13 to 2.80) in the 35–44 age range, 3.07 (95% CI 2.65 to 3.55) in those between 45-year-olds and 54-year-olds and the highest in those aged 55–65 with an aRR of 3.86 (95% CI 3.23 to 4.61). Results were less pronounced, but a similar age gradient in aRRs was observed in the clinical cohort ( online supplemental eTable 4 ). We also observed that males exhibited a higher propensity for absence from work than their female counterparts in the health administrative cohort in the 2 years preceding the index date (aRRs for males: 3.53 (95% CI 3.05 to 4.10) and 1.74 (95% CI 1.43 to 2.12), females: 2.09 (95% CI 1.92 to 2.28) and 1.27 (95% CI 1.14 to 1.42)). Although the aRRs for males were higher than those for females in the 2 years prior to the index date, we did not observe statistically significant sex differences in the clinical cohort during the study follow-up as 95% CIs overlapped ( online supplemental eTable 5 ).

Effect of income, immigration and area of residence on sickness absence rates

A gradient was observed in the health administrative cohort, with the highest-income group exhibiting the greatest aRR of 2.81 (95% CI 2.30 to 3.43) in the year preceding MS onset. Conversely, the lowest-income group had the lowest aRR of 1.88 (95% CI 1.37 to 2.56). Most aRRs did not reach statistical significance in the clinical cohort except for the second and third quantiles, and no apparent gradient was observed concerning income level ( online supplemental eTable 6 ).

In the administrative cohort, immigrants to Sweden (10% of the cases and controls combined) demonstrated similar aRRs (2.60, 95% CI 2.03 to 3.33) in the year preceding MS onset to those born in Sweden (aRR 2.54, 95% CI 2.34 to 2.76). In the clinical cohort, immigrants to Sweden consistently showed lower aRRs in the 5 years preindex date, with values ranging from 0.51 (95% CI 0.32 to 0.81) to 0.86 (95% CI 0.60 to 1.23), although some were not significant. Conversely, non-immigrants exhibited higher aRRs, ranging from 1.07 (95% CI 0.92 to 1.25) to 1.23 (95% CI 1.09 to 1.40) ( online supplemental eTable 7 ).

Finally, we did not observe any noticeable differences based on area of residence in either of the cohorts (urban vs rural, data not shown).

Discussion

Here, we analysed the annual rates of sickness absence among individuals in the prodromal phase of MS compared with a matched cohort from the Swedish general population, using comprehensive nationwide administrative and clinical data from Sweden. Our results indicate significant increases in absence due to short-term illness or injury among MS cases compared with matched controls, particularly in the few years immediately preceding the classical onset of MS. Individuals with MS showed 2.6 times higher rates of sickness absence in the year before the index date. The difference was less pronounced but still significant in the clinical cohort, underscoring the presence of the prodromal phase as measured by the number of sickness absence days in individuals with MS. We also observed age-related and sex-related variations, with older individuals and males exhibiting higher sickness absence rate ratios in the 2 years leading up to MS onset. While we observed a gradient in income estimates in the year before the index in the health administrative cohort with higher aRRs in individuals in the highest income level, we did not observe a notable influence of area of residence (urban or rural) or immigration status on sickness absence.

Our study adds novel insights into the prodromal phase of MS. Previous studies of the period before MS onset have primarily focused on physician visits, hospitalisations and prescriptions filled. 19 Our approach, leveraging sickness absence data, offers an additional and unique perspective on the socioeconomic implications of the early manifestations of disease. Notably, the significant differences in sickness absence rates observed years before the index date align with studies suggesting that the MS prodromal phase can extend 5–15 years before its onset. 3 20 Our results also align with a recent Danish study that found significant differences in occupation levels and average medical costs before and after diagnosis in patients with optic neuritis with and without MS. 21 A previous Swedish study found that compared with controls, individuals with MS experience an increased combined annual net days of sickness absence and disability pension days from 5 years before diagnosis to 5 years after. The difference between individuals with MS relative to controls gradually increased, with a steep increase around the time of diagnosis. 22

Our approach also underscores the sickness absence data as a valuable measure in epidemiological research, providing a broader perspective on the health trajectories of chronic conditions like MS before their clinical manifestation. An advantage of sickness absence data is its accessibility and standardisation. In Sweden, recordings of sickness absence episodes are standardised due to legal and policy requirements. This standardisation can offer more uniform data across different demographics and regions than physician visits or hospitalisation records, which might vary more significantly in how data are recorded based on varying medical protocols or patients’ access to care. 23 24 Nevertheless, because sickness absence depends on approval from a physician, limitations within the healthcare system can affect the quality of sickness absence data.

Few studies have examined the effects of sex and age on the MS prodrome. We found age to be an influential factor, with a notable age-related increase in sickness absence days, particularly pronounced among older individuals approaching the onset of MS in the health administrative cohort. Sex differences were also prominent in the health administrative cohort, with males demonstrating a higher propensity for work absence in the immediate period leading up to the MS onset. Overall, our results indicate possible age-related and sex-related variations in the experience of early MS symptoms and their impact on the ability to maintain employment. A previous study from Canada indicated age-related differences in healthcare utilisation (hospitalisations and physician visits) rates with higher aRRs for those aged ≥50 at MS onset compared with the <30 in the 5 years preceding MS onset. 4 Similar sex-related differences and patterns were reflected in our findings. Several possible reasons exist for the observed increase in sickness absence rates among males. Males often exhibit a more progressive and disabling form of MS, which could manifest in higher sickness absence rates compared with females even before MS onset. Occupational differences between sexes can also contribute to this disparity, as males may be more likely to engage in physically demanding jobs that necessitate sick days in the event of a health issue. Additionally, differences in health-seeking behaviour may also play a role. 25

The impact of socioeconomic status, as reflected through individual disposable income, revealed an income-dependent gradient in sickness absence rates, with those in the highest income quintile showing the highest rates of sickness absence in the year before MS onset. This suggests that the prodromal phase of MS might affect employment ability differently across income groups.

We found a difference in the overall patterns of sickness absence between immigrants and those born in Sweden, particularly in the clinical cohort, with immigrants showing lower aRRs of sickness absence during the 5 years leading up to MS symptom onset. Although only 10% of our cohorts were immigrants, and the results were not significant for all years, the lower aRRs among immigrants may reflect systemic challenges and barriers to accessing sickness absence benefits. Immigrants might struggle with navigating a foreign healthcare system and social support systems due to language barriers, limited understanding of their entitlements or unfamiliarity with the processes involved in claiming sick leave. 26 28 Nevertheless, the possibility of the healthy migrant effect on our findings should not be dismissed. 29 Overall, our findings highlight the multifaceted effects of sociodemographic factors on the prodromal phase of MS.

Our findings are consistent with the notion that MS is a heterogeneous disease, and the risk of MS may not arise from a single uniform pathway. Whether our observations are a consequence of early demyelination, immunological changes, or are caused by a broader chronic ill health trajectory leading to MS, or a combination of all these remains unknown. A clearer understanding of these pathways would require further investigation integrating clinical, imaging and molecular data. Identifying how each subset of patients transitions from prodrome to classical onset and diagnoses of MS may help refine risk stratification and facilitate timely detection and targeted intervention.

While our study offers insights into the timing and magnitude of increased sickness absence before MS onset, linking diagnosis codes, treatment histories, imaging, biological and genetic data may help identify specific patient clusters or pathways and clarify whether particular symptoms or clinical trajectories signal elevated MS risk. This could further refine risk‐prediction models or decision-aid tools and inform the development of prodromal criteria. 2 Strengthening cross‐jurisdictional collaborations, standardising data collection (eg, MRI protocols) and adopting inclusive research practices (eg, including non-white/non-European individuals in genetic studies) will be crucial for translating these emerging tools into effective, ethical and broadly generalisable strategies for early MS detection and intervention.

The strengths of our study include the use of a large, nationally representative sample and the comprehensive, real-time capture of sickness absence data, which allow for robust analysis and comparison of sickness absence trends over time. The addition of a clinical cohort in this study provided significant advantages despite its smaller size, as it generally confirmed the patterns observed in the larger health administrative cohort. This strengthens the validity of the results, demonstrating that the observed trends are robust across different data sources and methods of defining the index date.

Nonetheless, the study faces limitations, including the reliance on administrative data not collected for research purposes. However, such data have been validated and quality checked and have shown value in the research setting. 10 12 16 30 31 Moreover, sickness absence, as a measure of disease impact, has been extensively used in studying MS, primarily focusing on the period after its onset, and has provided valuable insights into its societal impact. 7 8 17

Additionally, our study may not have captured subtle regional, migrational or the full extent of socioeconomic and demographic influences on sickness absence. We lacked information on ‘employment status’ each year and therefore could not know if a person was started on sickness absence after 2 or 15 days (if employed, the employer pays days 2–14, but we do not have access to these data). In other words, it is possible that there was a misclassification bias if there were differences in employment rates between cases and controls. If cases are employed at higher rates, the true aRR for sick days among cases is likely to increase if days 2–14 data are included, whereas if cases are less employed, the aRR for cases is likely to show little change or a slight decrease.

An intriguing avenue for future research involves clarifying whether individuals who already harbour heightened susceptibility to MS—such as those with elevated MS polygenicity or specific environmental exposures—are more likely to transition to clinical MS when also burdened by chronic ill health and comorbidities.

Despite these limitations, the use of sickness absence data represents a novel and promising approach to understanding the prodromal phase of MS. Our results suggest that there is a measurable reduction in the ability to work among those who go on to develop MS up to 6 years before the clinical recognition of the disease. The reasons for these increased spells of temporary work loss should be explored further in future studies.

Supplementary material