ABSTRACT
Background
Managing long COVID in people with multiple sclerosis and related disorders (pwMSRD) is complex due to overlapping symptoms. To address evidence gaps, we evaluated long COVID susceptibility in pwMSRD versus controls and its associations with multi‐domain function and disability.
Methods
In this multicenter cross‐sectional study, participants completed a survey covering 71 post‐infection symptoms, distinguishing new‐onset from worsening symptoms. We defined long COVID using the 2024 NASEM criteria. Logistic regression assessed long COVID odds. Linear and Poisson regression evaluated associations with function and disability.
Results
969 pwMSRD (82.5% female, mean age 51.8 years, 63.5% infected) and 1003 controls (79.4% female, mean age 45.2 years, 61.2% infected) were included. PwMSRD had higher odds of long COVID (aOR = 1.6 [1.2–2.1]), with a stronger association when restricting to worsening symptoms (aOR = 2.3 [1.7–3.1]). Having long COVID was associated with worse physical function, cognition, and depression in both groups. PwMSRD with long COVID experienced greater physical function declines and more depression severity exacerbation than controls, and had faster disability progression compared to those without long COVID.
Conclusion
PwMSRD show increased susceptibility to long COVID, primarily driven by worsening symptoms. Long COVID contributes to more functional decline and disability worsening. Recognizing and managing long COVID is essential for pwMSRD.
Keywords: disability, long COVID, multiple sclerosis and related disorders, neurology, patient‐reported outcomes
1. Introduction
The COVID‐19 pandemic has disproportionately affected individuals with multiple sclerosis and related disorders (pwMSRD) [1]. However, the pathophysiology and clinical manifestation of long COVID, also known as post‐acute sequelae of COVID‐19 (PASC), are poorly understood in this vulnerable population [2]. Long COVID has been widely recognized as a major public health concern in the general population, while few studies have examined its epidemiology or its impact on disability among pwMSRD. As a result, it remains unclear whether pwMSRD are more susceptible to long COVID and whether long COVID exacerbates the clinical outcome of neuroinflammatory diseases [3, 4, 5, 6, 7, 8].
Multiple terms and definitions have been proposed for long COVID, each with distinct strengths and limitations. In 2023, the National Institutes of Health's Researching COVID to Enhance Recovery (NIH RECOVER) initiative introduced a stringent definition, using a weighted scoring system based on 12 symptoms to identify PASC [6]. Although this approach enhances specificity, it is limited by selection bias, low sensitivity, and a lack of external validation among other populations. In 2024, the National Academies of Sciences, Engineering, and Medicine (NASEM) released a consensus definition of long COVID as “an infection‐associated chronic condition occurring after SARS‐CoV‐2 infection, persisting for at least three months in a continuous, relapsing‐remitting, or progressive manner, and affecting one or more organ systems.” [9]. While intentionally inclusive, this definition lacks specificity due to the absence of symptoms or organ system specification.
In pwMSRD, long COVID is particularly challenging to characterize due to overlapping symptoms such as cognitive impairment, fatigue, and psychiatric manifestations [10]. The symptomatic overlap complicates the distinction between the post‐viral sequelae and the underlying neuroinflammatory disease. Recent multimodal proteomic studies using machine learning approaches have suggested that active long COVID involves dysregulated complement activation, elevated antibody responses to herpesviruses, and thrombo‐inflammatory markers, some of which are implicated in MS pathogenesis [11]. Estimates of post‐infectious symptom prevalence in pwMSRD range from 12.4% to 36.9%, but prior studies have not differentiated new symptoms from pre‐existing ones [10, 12, 13, 14]. This distinction is crucial for diagnosing and managing long COVID in pwMSRD. Finally, while long COVID has been shown to substantially worsen patient‐reported outcomes (PROs) in the general population, its specific burden in pwMSRD remains unquantified [2, 15].
In this study, we aimed to evaluate long COVID susceptibility and association with PROs in a large, multi‐center cohort of pwMSRD and controls. We distinguished new symptoms from worsening pre‐existing symptoms to compare how post‐infectious symptoms manifest in pwMSRD versus controls. We hypothesized that pwMSRD had increased odds of long COVID compared to controls and that both MSRD and long COVID independently contributed to worse PROs, with their combined presence leading to even greater impairment.
2. Methods
2.1. Study Design and Participants
This cross‐sectional study utilized data from the Multiple Sclerosis Resilience to COVID‐19 (MSReCOV) Collaborative, a cohort consisting of individuals with MSRD and controls, recruited from five U.S.‐based clinical neuroimmunology centers (Figure 1) [16, 17, 18]. Participants enrolled in MSReCOV between April 2020 and July 2021. The MSRD group included individuals diagnosed with multiple sclerosis (MS), neuromyelitis optica spectrum disorder (NMOSD), myelin oligodendrocyte glycoprotein antibody‐associated disease (MOGAD), and other rare neuroimmunological disorders (NID). The control group had no diagnosis of neuroinflammatory disorders, comprising relatives of pwMSRD, controls from local registries, and individuals recruited through local advertising at each participating center. Inclusion criteria for both groups were: (1) age ≥ 18 years, (2) able to provide informed consent, and (3) proficiency in English [19, 20]. Between August and December 2022, we invited all MSReCOV participants to complete a one‐time survey via a centralized, secure, web‐based Research Electronic Data Capture (REDCap) platform. Each participant completed the informed consent before starting the survey. The University of Pittsburgh Institutional Review Board approved the survey study (STUDY22080007).
FIGURE 1.
Study design. MSRD, multiple sclerosis and related disorders; MSReCOV, Multiple Sclerosis Resilience to COVID‐19; PDDS, Patient Determined Disease Steps; PROMIS, Patient‐Reported Outcomes Measurement Information System.
2.2. Measurements
2.2.1. Demographic and Clinical Profile
The survey collected information on age, sex, race, ethnicity, comorbidities adapted from the Charlson Comorbidity Index (CCI), height, weight, neuroimmunological diagnosis, and pre‐pandemic employment status. Body mass index (BMI) was calculated using height and weight. Comorbidity burden was categorized as none (CCI = 0), mild (1 ≤ CCI ≤ 2), moderate (3 ≤ CCI ≤ 5), or severe (CCI ≥ 6). Employment status was classified as employed (including full‐time, part‐time, self‐employed, students, and military personnel) or unemployed (including those out of work, retired, or unable to work). For pwMS, additional data were collected on symptom onset year, MS subtype, classified as relapsing–remitting MS (RRMS) or progressive MS (including primary and secondary progressive type), and disease‐modifying therapy (DMT) at survey, categorized into B‐cell depletion therapies, sphingosine‐1‐phosphate (S1P) receptor modulators, other DMTs, or none/not answered, based on their prior associations with COVID‐19 severity [21].
2.2.2. COVID‐19 Infection and Vaccine
Participants reported details regarding their COVID‐19 infection history, including method of detection, dates of initial and subsequent infections, and severity of the initial infection. Acute COVID‐19 infection was defined as having a confirmed or suspected infection based on a polymerase chain reaction (PCR) test, a rapid antigen test (in a laboratory, at home, or in an uncertain setting), or self‐reported symptoms consistent with the Centers for Disease Control and Prevention (CDC) list of COVID‐19 symptoms. COVID‐19 severity was categorized as asymptomatic, mild (symptomatic but able to function), moderate (symptomatic and very ill without hospitalization), severe (requiring hospital admission), or critical (requiring intensive care unit management). To account for variant‐specific differences, the time of initial infection was classified as pre‐Omicron (before January 1, 2022) or Omicron era (on or after January 1, 2022). Vaccination status at survey was categorized as fully vaccinated (receiving ≥ 3 doses) or not fully vaccinated (receiving < 3 doses).
2.2.3. Long COVID Symptoms
The primary study objective was to compare the odds of long COVID between pwMSRD and controls. Since no gold‐standard symptom identification for long COVID exists (during the study design period), an expert panel of neurologists (JS, LM, PLD, EL, ZX) specializing in neuroimmunology developed a comprehensive symptom list, classifying 71 symptoms across 12 organ systems. To enhance validity, the questionnaire was adapted from the latest available knowledge of post‐infection syndrome by organ systems at the time of survey deployment [8]. Participants were asked whether they had experienced any of these symptoms occurring one month or later after their first acute COVID‐19 episode, whether these symptoms were new or worsened compared to their pre‐COVID baseline, and the duration (months) of symptom persistence.
To define long COVID, we adopted the 2024 NASEM Long COVID definition for primary analyses [9]. Participants who reported any symptoms that persisted for at least 3 months were classified as the long COVID group. To distinguish between long COVID driven by new symptoms and resulting from the worsening of pre‐existing symptoms, we further restricted the long COVID status to new symptoms (i.e., absent during pre‐COVID baseline) and worsening symptoms (i.e., worsened from the pre‐COVID baseline). Based on these definitions, the three primary outcomes were overall long COVID (Yes vs. No), long COVID based on new symptoms (Yes vs. No), and long COVID based on worsening symptoms (Yes vs. No).
2.2.4. Patient‐Reported Outcomes
To assess functional outcomes, we utilized the Patient‐Reported Outcomes Measurement Information System (PROMIS), including assessments of physical function (version 1.2), cognitive function (version 2.0), and depression (version 1.0). PROMIS instruments were standardized measures of patient‐reported outcomes validated across various diseases. PROMIS T‐scores were standardized to the U.S. general population, with a mean of 50 and a standard deviation of 10, where higher scores indicate greater levels of the measured attribute (e.g., higher depression scores indicate greater depressive severity, while higher physical and cognitive function scores indicate better functions). For pwMS, we additionally evaluated neurological disability both pre‐COVID and at the time of survey using the well‐validated Patient Determined Disease Steps (PDDS) scale, an ordinal scale ranging from 0 (no impairment) to 8 (bed‐bound status).
2.3. Statistical Analysis
The dataset included MSReCOV participants who completed the survey and had complete demographic, vaccination, and COVID‐19 data. Participant characteristics were summarized as mean (SD), median (IQR), or frequency (proportion). Differences between pwMSRD and controls were assessed using the Wilcoxon rank‐sum test for continuous variables and Fisher's exact test for categorical variables.
Among participants with a history of acute COVID‐19, we first compared the prevalence of each post‐infection symptom between pwMSRD and controls using multivariable logistic regression, adjusting for age, sex, race/ethnicity, BMI, comorbidity burden (CCI), pre‐pandemic employment, related factors of initial infection (i.e., detection method, severity, wave, and time to survey completion), reinfection, and vaccination status. We applied separate models for new‐onset and worsening symptoms. A Bonferroni correction for 71 symptoms set the significance threshold at p < 0.0007. We then integrated individual symptoms into a composite outcome to define long COVID status using the NASEM definition and assessed three outcomes: overall long COVID, long COVID based on new symptom, and long COVID based on worsening symptom. We used logistic regression models to compare the odds of long COVID between pwMSRD and controls, adjusting for the same covariates.
To assess associations between long COVID status and PROs, we used multivariable linear regression models, further adjusting for MSRD status and its interaction with long COVID status to assess potential synergistic effects on health outcomes. Separate models examined physical function, cognition, and depression, using overall long COVID, long COVID based on new symptoms, and long COVID based on worsening symptoms as exposures. Within each model, we reported adjusted beta estimates and 95% CIs for long COVID status, MSRD status, and their interaction.
In the subgroup of pwMS, we used multivariable logistic regression to identify factors associated with long COVID, including demographic and clinical profile (age, sex, race/ethnicity, BMI, comorbidities, employment), acute COVID‐19 factors (detection method, wave, severity, reinfection, vaccination, time from infection to survey), and MS‐specific characteristics (onset age, pre‐COVID disability, MS subtype, DMT use). We chose to adjust for onset age rather than disease duration to minimize potential collinearity with age at survey, as disease duration (time from symptom onset to survey completion) is mathematically related to age (time from birth to survey completion) and may obscure independent effects in regression models. Significant factors were selected based on likelihood ratio tests (LRT) comparing full and reduced models. Finally, we employed Poisson regression to assess the relationship between long COVID and patient‐reported disability based on PDDS, adjusting for the same set of demographic and clinical profile, acute COVID‐19 features, and MS‐specific characteristics. We reported adjusted rate ratios (aRRs) and 95% CIs of long COVID‐associated disability worsening.
2.4. Sensitivity Analysis
The NASEM long COVID definition is limited in specificity. To test whether findings were robust to long COVID definitions, we alternatively applied the RECOVER scoring system, which classified the condition by assigning weighted points to each symptom [6]. Symptoms and their respective scores included loss of smell or taste (8 points), post‐exertional malaise (7 points), chronic cough (4 points), brain fog (3 points), thirst (3 points), palpitations (2 points), chest pain (2 points), fatigue (1 point), sexual dysfunction (1 point), dizziness (1 point), gastrointestinal symptoms (1 point), abnormal movements (1 point), and hair loss (1 point). Participants with a total score of 12 or higher were classified as long COVID according to RECOVER [6, 10]. A list of the original symptoms surveyed and their classification within the RECOVER scoring system was provided in Table S1. We separately computed scores for new symptoms and worsening symptoms, applying the same scoring thresholds. We reported the prevalence of RECOVER‐defined long COVID and its association with MSRD groups and PROs, using the same models employed in analyses based on NASEM‐defined long COVID.
3. Results
3.1. Participants
Among the 3527 MSReCOV participants invited to participate in the one‐time survey, 2157 consented. 1972 participants (969 pwMSRD and 1003 controls) completed the survey ≥ 3 months after the initial acute infection (Figure 2). The consent rate was 61.2% (2157/3527), and the response rate among consented participants was 92% (1972/2157).
FIGURE 2.
Participant disposition. MSRD, multiple sclerosis and related disorders; MSReCOV, Multiple Sclerosis Resilience to COVID‐19; PROMIS, Patient‐Reported Outcomes Measurement Information System; pwMSRD, people with multiple sclerosis and related disorders.
Compared to survey responders, non‐responders were younger on average, more likely to be pwMSRD, and less likely to be female or non‐Hispanic White (Table S2). When comparing pwMSRD with controls among participants who completed the survey (Table 1), pwMSRD were older (mean age 51.8 years [SD 12.1] vs. 45.2 years [SD 10.3], p < 0.001), had a lower proportion of non‐Hispanic White participants (84.7% vs. 92.4%, p < 0.001), were less likely to be employed before the pandemic (57.5% vs. 86.0%, p < 0.001), and had a greater comorbidity burden (45.8% vs. 64.5% without comorbidity). 613 pwMSRD and 614 controls (63.3% vs. 61.2%; p = 0.36; Table 1) reported a history of acute COVID‐19. Compared to controls, pwMSRD were more likely to have had a pre‐Omicron infection (44.5% vs. 37.0%; p = 0.008), had a higher rate of severe or critical acute COVID‐19 (4.7% vs. 0.7%; p < 0.001), and were more likely to report multiple acute COVID‐19 infections (27.1% vs. 17.1%; p < 0.001). Vaccination status was similar between groups (26.1% vs. 26.2% not fully vaccinated; p = 0.25), as was the time from initial infection to survey completion (10.0 months [SD 8.0] vs. 9.6 months [SD 8.3], p = 0.45).
TABLE 1.
Participant characteristics.
| Variables | Variable category | Overall | MSRD | Control | p |
|---|---|---|---|---|---|
| N | 1972 | 969 | 1003 | ||
| Age, years; mean (SD) | 48.4 (11.7) | 51.8 (12.1) | 45.2 (10.3) | < 0.001 | |
| Sex; n (%) | Female | 1599 (81.1%) | 803 (82.9%) | 796 (79.4%) | 0.183 |
| Male | 370 (18.8%) | 166 (17.1%) | 204 (20.3%) | ||
| Race and ethnicity; n (%) | Non‐Hispanic White | 1748 (88.6%) | 821 (84.7%) | 927 (92.4%) | < 0.001 |
| Other | 224 (11.4%) | 148 (15.3%) | 76 (7.6%) | ||
| BMI, kg/m2; mean (SD) | 33.3 (8.7) | 33.3 (8.7) | 33.2 (8.7) | 0.753 | |
| CCI category; n (%) | None (CCI = 0) | 1091 (55.3%) | 444 (45.8%) | 647 (64.5%) | < 0.001 |
| Mild (1 ≤ CCI ≤ 2) | 605 (30.7%) | 305 (31.5%) | 300 (29.9%) | ||
| Moderate (3 ≤ CCI ≤ 5) | 235 (11.9%) | 188 (19.4%) | 47 (4.7%) | ||
| Severe (CCI ≥ 6) | 41 (2.1%) | 32 (3.3%) | 9 (0.9%) | ||
| Pre‐pandemic employment status; n (%) | Employed | 1423 (72.1%) | 560 (57.8%) | 863 (86.0%) | < 0.001 |
| Unemployed | 549 (27.9%) | 409 (42.2%) | 140 (14.0%) | ||
| Neurologic disorder; n (%) | MOGAD | N/A | 2 (0.2%) | N/A | N/A |
| MS | N/A | 940 (97.0%) | N/A | ||
| NID | N/A | 18 (1.9%) | N/A | ||
| NMOSD | N/A | 9 (0.9%) | N/A | ||
| MSRD onset age; mean (SD) | N/A | 34.8 (10.4) | N/A | N/A | |
| MS subtype | RRMS | N/A | 714 (76.0%) | N/A | |
| SPMS | N/A | 105 (11.1%) | N/A | ||
| PPMS | N/A | 65 (6.9%) | N/A | ||
| DMT; n (%) | None | N/A | 177 (18.8%) | N/A | N/A |
| Anti‐CD20s | N/A | 339 (36.1%) | N/A | ||
| S1P modulators | N/A | 67 (7.1%) | N/A | ||
| Other | N/A | 357 (38.0%) | N/A | ||
| Acute COVID‐19; n (%) | Present | Adjust | 613 (63.3%) | 614 (61.2%) | 0.353 |
| Absent | 745 (37.8%) | 356 (36.7%) | 389 (38.8%) | ||
| Multiple infection episodes; n (%) | 271 (22.1%) | 166 (27.1%) | 105 (17.1%) | < 0.001 | |
| Initial acute COVID‐19 detection source; n (%) | PCR | 459 (37.4%) | 252 (41.1%) | 207 (33.7%) | 0.030 |
| Rapid antigen test | 583 (47.5%) | 275 (44.9%) | 308 (50.2%) | ||
| Other | 185 (15.1%) | 86 (14.0%) | 99 (16.1%) | ||
| Initial acute COVID‐19 severity; n (%) | No symptoms | 45 (3.8%) | 23 (3.6%) | 22 (4.1%) | < 0.001 |
| Mild | 653 (53.2%) | 290 (47.6%) | 363 (58.8%) | ||
| Moderate | 495 (40.3%) | 271 (44.2%) | 224 (36.5%) | ||
| Severe and critical | 34 (2.7%) | 29 (4.7%) | 5 (0.7%) | ||
| Initial infection wave; n (%) | Before Jan 1, 2022 | 500 (40.1%) | 273 (44.5%) | 227 (37.0%) | 0.008 |
| After Jan 1, 2022 | 727 (59.9%) | 340 (55.5%) | 387 (63.0%) | ||
| Vaccine dose; n (%) | < 3 | 513 (26.1%) | 253 (26.1%) | 263 (26.2%) | 0.245 |
| ≥ 3 | 1456 (73.9%) | 716 (73.9%) | 740 (73.8%) | ||
| Months from the initial infection to survey completion; mean (SD) | 9.8 (8.1) | 10.0 (8.0) | 9.6 (8.3) | 0.450 |
Abbreviations: CCI, Charlson comorbidity index; DMT, disease‐modifying therapy; MOGAD, myelin oligodendrocyte glycoprotein antibody‐associated disease; MSRD, multiple sclerosis and related disorder; NID, neuroimmunological disorders; NMODS, neuromyelitis optica spectrum disorder; PPMS, primary progressive multiple sclerosis; RRMS, relapsing remitting multiple sclerosis; SPMS, secondary progressive multiple sclerosis.
3.2. Post‐Infection Symptoms
The five most frequent new‐onset symptoms persisting over 3 months after acute COVID‐19 differed between pwMSRD (change or loss of smell, change or loss of taste, shortness of breath, cough with mucus, and breathing difficulty; 6.7%–10.4%; Figure S1) and controls (brain fog, fatigue, exercise intolerance, speech and language issues, and memory issues; 5.7%–9.9%; Figure S2). After adjusting for covariates and applying Bonferroni correction, pwMSRD had significantly lower adjusted odds of new‐onset brain fog (0.3 [0.2, 0.6]), fatigue (0.2 [0.1, 0.3]), and speech and language issues (0.2 [0.1, 0.5]) (Figure S3).
The most frequent worsening symptoms (> 10% prevalence) persisting over 3 months after acute COVID‐19 in pwMSRD were fatigue, brain fog, weakness, difficulty concentrating, memory issues, dizziness, speech and language issues, joint pain, muscle aches, muscle spasms, change in sensation, exercise intolerance, and insomnia (Figure S4), none of which exceeded 10% prevalence in controls (Figure S5). PwMSRD had significantly higher adjusted odds of worsening symptoms (Figure S6), including dizziness (aOR = 15.7), weakness (aOR = 14.9), muscle spasms (aOR = 9.3), change in sensation (aOR = 8.0), speech and language issues (aOR = 7.0), difficulty concentrating (aOR = 5.3), brain fog (aOR = 6.0), fatigue (aOR = 4.9), bladder control issues (aOR = 4.2), muscle aches (aOR = 3.4), and joint pain (aOR = 2.8).
3.3. Associations Between MSRD Status and Long COVID
Using the NASEM definition of long COVID overall, 310 (50.6%) pwMSRD met the criteria compared to 216 (35.2%) controls (p < 0.001; Figure 3A). Based on new symptoms, the long COVID prevalence was 39.6% in pwMSRD vs. 29.8% in controls (p < 0.001; Figure 3B). Based on worsening symptoms, the long COVID prevalence was 42.4% in pwMSRD vs. 20.7% in controls (p < 0.001; Figure 3C). When comparing the baseline characteristics stratified by NASEM‐defined overall long COVID status within both the pwMSRD and control groups, individuals with long COVID were more likely to be female, have higher BMI, experience multiple infections, report more severe acute COVID symptoms, and have been infected prior to the Omicron wave (Table 2). Among pwMSRD, those with long COVID were significantly more likely to be not fully vaccinated (35.2% vs. 23.1%; p = 0.001), whereas vaccination status did not differ between groups in the control population (29.2% vs. 28.9%; p = 1.0) (Table 2). PwMSRD with long COVID also had significantly higher baseline PDDS scores compared to those without long COVID, whereas disease onset age and DMT types were similar between groups. After adjusting for confounders, pwMSRD had higher odds of developing long COVID overall (aOR = 1.6 [1.2, 2.1]; Figure 4A) and long COVID based on worsening symptoms (aOR = 2.3 [1.7, 3.1]; Figure 4C), whereas the association with long COVID based on new symptoms was attenuated and no longer statistically significant (aOR = 1.3 [0.9, 1.7]; Figure 4B). Female and moderate‐to‐critical acute COVID‐19 severity (vs. asymptomatic infection) were consistently associated with higher odds of long COVID across all three definitions (Figure 4A–C).
FIGURE 3.
Prevalence of NASEM‐defined long COVID.
TABLE 2.
Baseline characteristics stratified by NASEM‐defined long COVID status within pwMSRD and controls.
| pwMSRD | Controls | |||||
|---|---|---|---|---|---|---|
| NASEM‐defined long COVID (Overall) = Y | NASEM‐defined long COVID (Overall) = N | p | NASEM‐defined long COVID (Overall) = Y | NASEM‐defined long COVID (Overall) = N | p | |
| N | 310 | 303 | 216 | 398 | ||
| Age, years; mean (SD) | 49.0 (11.6) | 50.2 (11.8) | 0.22 | 45.3 (10.8) | 43.8 (9.7) | 0.07 |
| Female; n (%) | 272 (87.7) | 242 (79.9) | 0.01 | 186 (86.1) | 310 (77.9) | 0.02 |
| Non‐Hispanic White; n (%) | 248 (80.0) | 260 (85.8) | 0.07 | 195 (90.3) | 373 (93.7) | 0.17 |
| BMI, kg/m2; mean (SD) | 34.2 (9.1) | 32.7 (7.7) | 0.02 | 35.27 (9.9) | 32.0 (7.8) | 0.001 |
| CCI category; n (%) | ||||||
| None (CCI =0) | 163 (52.6) | 147 (48.5) | 0.49 | 141 (65.3) | 280 (70.4) | 0.21 |
| Mild (1 ≤ CCI ≤ 2) | 90 (29.0) | 101 (33.3) | 60 (27.8) | 102 (25.6) | ||
| Moderate to severe (CCI ≥ 3) | 57 (18.4) | 55 (18.2) | 15 (6.9) | 16 (4.0) | ||
| Pre‐pandemic unemployed; n (%) | 106 (35.0) | 114 (36.8) | 0.706 | 57 (14.3) | 34 (15.7) | 0.72 |
| MS; n (%) | 282 (91.0%) | 276 (91.0%) | 1 | N/A | N/A | N/A |
| MSRD onset age; mean (SD) | 37.2 (11.3) | 37.5 (11.4) | 0.88 | N/A | N/A | N/A |
| DMT; n (%) | 0.36 | N/A | N/A | N/A | ||
| None | 49 (15.8) | 51 (16.8) | N/A | N/A | ||
| Anti‐CD20s | 137 (44.2) | 113 (37.3) | N/A | N/A | ||
| S1P modulators | 24 (7.7) | 25 (8.3) | N/A | N/A | ||
| Other | 100 (32.3) | 114 (37.6) | N/A | N/A | ||
| Multiple infection episodes; n (%) | 114 (36.8) | 52 (17.2) | < 0.001 | 63 (29.2) | 42 (10.6) | < 0.001 |
| Initial acute COVID‐19 severity; n (%) | ||||||
| No symptoms | 10 (3.2) | 13 (4.3) | < 0.001 | 7 (3.2) | 15 (3.8) | < 0.001 |
| Mild | 87 (28.1) | 162 (53.5) | 80 (37.0) | 252 (63.3) | ||
| Moderate | 178 (57.4) | 125 (41.3) | 124 (57.4) | 130 (32.7) | ||
| Severe and critical | 35 (11.3) | 3 (1.0) | 5 (2.3) | 1 (0.3) | ||
| Initial infection wave pre‐Omicron; n (%) | 188 (60.6) | 85 (28.1) | < 0.001 | 129 (59.7) | 98 (24.6) | < 0.001 |
| Vaccine dose < 3; n (%) | 109 (35.2) | 70 (23.1) | 0.001 | 63 (29.2) | 115 (28.9) | 1.0 |
| Pre‐COVID PDDS; median [IQR] | 1.00 [0.00, 3.00] | 1.00 [0.00, 3.00] | 0.003 | N/A | N/A | N/A |
Abbreviations: CCI, Charlson comorbidity index; DMT, disease‐modifying therapy; MSRD, multiple sclerosis and related disorders; NASEM, National Academies of Sciences, Engineering, and Medicine; PDDS, patient‐determined disease steps.
FIGURE 4.
Associations between factors and NASEM‐defined long COVID. Adjusted odds ratios (aORs) with 95% confidence intervals (CIs) from multivariable logistic regression models evaluating factors associated with (A) overall NASEM‐defined long COVID, (B) long COVID based on new symptoms, and (C) long COVID based on worsening symptoms.
3.4. Associations Between Long COVID and PROMIS T‐Scores
One thousand one hundred fifteen (90.9%) participants who completed all three PROMIS assessments were included in the PRO analysis (Figure 2). Compared to the excluded participants, the participants included were more likely to be employed and to develop long COVID, while other demographics were similar (Table S3). Multivariable linear regression showed that long COVID (overall, new, or worsening) was associated with significantly decreased physical function (3.0–5.4 points) and cognitive function (6.2–8.4 points) and with significantly increased depression severity (1.9–5.3 points) (Figure 5; Table S4). Notably, MSRD status modified the association between long COVID and PROs: pwMSRD with long COVID experienced greater physical function declines and depression severity than controls (interaction p < 0.05; Table S4).
FIGURE 5.
Multivariable linear regression for PROMIS physical function (left), cognitive function (middle), and depression (right). Model 1, NASEM‐defined long COVID overall as the main exposure; Model 2, NASEM‐defined long COVID on new symptoms as the main exposure; Model 3, NASEM‐defined long COVID on worsening symptoms as the main exposure.
3.5. Subgroup Results: Long COVID and Disability in pwMS
Among pwMS, 282 (50.3%) had NASEM‐defined long COVID overall, 220 (39.2%) based on new symptoms, and 236 (42.1%) based on worsening symptoms (Table 3). Female, worse pre‐COVID disability, and worse acute infection severity were significant contributors to long COVID development (all LRT p‐values < 0.05), while MS onset age, MS subtype, and DMT use were not. Analyses restricted to new or worsening symptoms yielded similar findings regarding key contributing factors (Table 3). Patients with long COVID exhibited greater patient‐reported disability accumulation based on PDDS at the time of the survey: aRR = 1.2 [1.2, 1.5] for long COVID overall; aRR = 1.2 [1.1, 1.4] for long COVID based on new symptoms; and aRR = 1.2 [1.1, 1.4] for long COVID based on worsening symptoms.
TABLE 3.
Contributing factors to long COVID in multiple sclerosis (MS) patients.
| Variables | Variable categories | NASEM‐defined long COVID (overall) n = 282 (50.3%) | NASEM‐defined long COVID (new symptoms) n = 220 (39.2%) | NASEM‐defined long COVID (worsening symptoms) n = 236 (42.1%) | |||
|---|---|---|---|---|---|---|---|
| OR (95% CI) | p a | OR (95% CI) | p a | OR (95% CI) | p a | ||
| Age at survey completion (years) | — | 1 (0.97–1.02) | 0.96 | 1.01 (0.98–1.04) | 0.549 | 0.99 (0.95–1.02) | 0.524 |
| Sex [ref = male] | Female | 2.96 (1.58–5.82) | < 0.001 | 2.01 (1.01–4.49) | 0.044 | 2.54 (1.02–7.45) | 0.044 |
| Race/ethnicity [ref = Non‐Hispanic White] | Other | 1.72 (1.03–2.9) | 0.038 | 1.77 (1.02–3.03) | 0.041 | 1.41 (0.72–2.67) | 0.305 |
| Body mass index (kg/m2) | — | 1.01 (0.99–1.04) | 0.313 | 1 (0.97–1.03) | 0.974 | 1.02 (0.98–1.05) | 0.327 |
| Comorbidity [ref = no comorbidity] | Mild | 1.49 (0.9–2.49) | 0.295 | 1.19 (0.69–2.06) | 0.067 | 1.72 (0.86–3.48) | 0.274 |
| Moderate to severe | 1.24 (0.59–2.64) | 0.5 (0.21–1.17) | 1.91 (0.68–5.62) | ||||
| Pre‐pandemic employment [ref = employed] | Unemployed | 0.9 (0.57–1.41) | 0.652 | 1.14 (0.69–1.89) | 0.601 | 0.93 (0.5–1.68) | 0.801 |
| MS onset age (years) | — | 1.03 (0.97–1.09) | 0.32 | 1.05 (0.98–1.12) | 0.16 | 1.03 (0.97–1.10) | 0.37 |
| MS subtype [ref = RRMS] | PMS | 0.8 (0.4–1.59) | 0.52 | 0.72 (0.32–1.58) | 0.419 | 0.87 (0.35–2.09) | 0.75 |
| DMT use [ref = none] | Anti‐CD20s | 0.62 (0.34–1.12) | 0.3 | 0.81 (0.42–1.61) | 0.807 | 0.57 (0.27–1.2) | 0.344 |
| S1P modulators | 0.61 (0.25–1.42) | 0.72 (0.25–1.95) | 0.55 (0.14–1.81) | ||||
| Other | 0.57 (0.32–1.03) | 1 (0.52–1.93) | 0.5 (0.24–1.07) | ||||
| Baseline PDDS scale | — | 1.17 (1.04–1.32) | 0.008 | 1.3 (1.11–1.54) | 0.039 | 1.19 (1.02–1.38) | 0.026 |
| Time from survey completion to the first infection (months) | — | 0.99 (0.95–1.03) | 0.599 | 1.02 (0.98–1.07) | 0.391 | 1.04 (0.99–1.1) | 0.121 |
| COVID‐19 detection source [ref = Antigen] | PCR | 1.47 (0.93–2.32) | 0.251 | 1.17 (0.7–1.95) | 0.783 | 1.13 (0.62–2.08) | 0.209 |
| Other | 1.29 (0.65–2.54) | 0.97 (0.44–2.04) | 0.51 (0.18–1.31) | ||||
| Time of initial infection [ref = Omicron era] | Pre‐Omicron | 1.4 (0.74–2.62) | 0.299 | 1.25 (0.61–2.53) | 0.543 | 0.53 (0.21–1.27) | 0.158 |
| Acute infection severity [ref = asymptomatic] | Mild | 1.52 (0.82–2.82) | < 0.001 | 1.55 (0.58–4.14) | < 0.001 | 1.14 (0.11–11.82) | < 0.001 |
| Moderate to critical | 5.09 (3.02–8.58) | 4.1 (2.04–8.24) | 7.32 (3.16–16.96) | ||||
| Vaccination [ref = fully vaccinated] | Not fully vaccinated | 1.24 (0.8–1.93) | 0.334 | 1.19 (0.73–1.93) | 0.481 | 1.65 (0.93–2.91) | 0.088 |
| Multiple infections [ref = no] | Yes | 1.28 (0.78–2.1) | 0.319 | 0.86 (0.49–1.49) | 0.596 | 1.5 (0.79–2.84) | 0.212 |
Abbreviations: PDDS, patient determined disease steps; PMS, progressive multiple sclerosis (primary and secondary); RRMS, relapsing–remitting multiple sclerosis; S1P modulators, sphingosine‐1‐phosphate receptor modulator.
p‐values of likelihood ratio tests (LRT) comparing full and reduced models. Bold text indicates variables reaching pre‐defined statistical significance threshold for association with long COVID in MS patients.
3.6. Sensitivity Results: RECOVER‐Defined Long COVID
When using the RECOVER criteria, the long COVID prevalence declined in both pwMSRD (7.7%–17.5%) and controls (1.8%–8.0%) (Figure S7). The increased odds of long COVID in pwMSRD vs. controls as driven by worsening symptoms remained consistent (aOR = 3.5 [1.7, 7.5]; Figure S8). Similar to NASEM‐defined long COVID, RECOVER‐defined long COVID was also associated with worse PROs (Figure S9; Table S5) and a 15.1%–17.2% greater disability accumulation in pwMS.
4. Discussion
In this multicenter study of pwMSRD and controls without neuroinflammatory disorders, we systematically surveyed long COVID‐19 symptoms. We directly compared each symptom and long COVID development (as defined by the 2024 NASEM and the 2023 RECOVER criteria) between pwMSRD and controls. We refined the NASEM and RECOVER criteria by separately analyzing and distinguishing long COVID based on new‐onset and worsening symptoms, which were applicable for chronic neurological disorders such as MSRD. PwMSRD exhibited a higher prevalence of worsening systemic, musculoskeletal, and neurological symptoms from their pre‐COVID baseline compared to controls. After adjusting for potential confounders, pwMSRD had 56% higher odds of long COVID overall than controls, with a stronger association (OR = 2.3) when long COVID was based on worsening symptoms and a weaker association (OR = 1.3) when based on new symptoms.
Research on long COVID susceptibility and its clinical manifestations in MSRD has been limited, and few distinguished newly onset symptoms and worsening symptoms attributable to long COVID. The main finding of increased odds of long COVID in pwMSRD is primarily driven by worsening symptoms, aligned with a previous study reporting that pwMS were more likely to experience post‐COVID weakness, mobility difficulties, and cognitive dysfunction than controls, after accounting for pre‐existing symptoms [22]. Several possible mechanisms may explain the increased long COVID susceptibility in pwMS. First, dysregulations of the innate and adaptive immune system are common in both MS and long COVID. The chronic proinflammatory state in MS may amplify or prolong the inflammatory responses triggered by SARS‐CoV‐2 infection [5, 23]. Second, pre‐existing neurological impairment as well as mental health and other physical vulnerabilities in pwMS may increase the risk of persistent symptoms after acute COVID [24]. Third, certain DMTs (such as anti‐CD20 B‐cell depletion therapy) could impair immune response to vaccination, increasing the likelihood of long COVID following acute infection [25, 26, 27]. Fourth, long COVID may also cause neurodegenerative changes. People with long COVID who experience neurological symptoms showed elevated levels of neurofilament light chain and glial fibrillary acidic protein, biomarkers of neuroaxonal injury and glial activation which are also central to MS pathology and progression [28, 29]. Finally, SARS‐CoV‐2 has been linked to reactivation of latent herpesviruses, including Epstein–Barr Virus (EBV), which likely contributes to MS etiology. EBV viremia during acute COVID‐19 may predict long COVID and could be compounded by immune dysregulation in pwMS [30, 31].
Assessing long COVID in pwMSRD was challenging due to multiple factors, which we addressed to enhance the study rigor. First, a standardized long COVID definition applicable to pwMSRD is lacking, given the potential overlap of symptoms between MSRD and long COVID [2, 32]. To mitigate misclassification bias, we performed primary analyses using the latest NASEM criteria and performed sensitivity analyses using the widely used RECOVER criteria. The study findings were largely comparable across the two criteria. By distinguishing newly onset symptoms from worsening pre‐existing symptoms, we aimed to clarify the clinical presentation of long COVID in pwMSRD versus controls. Second, the NASEM and RECOVER criteria differed. As expected, NASEM‐defined long COVID prevalence was higher than RECOVER‐defined long COVID prevalence due to fewer restrictions on the type and number of symptoms required. In this study, 17.5% of pwMSRD developed RECOVER‐defined long COVID, consistent with the 15.9% reported by Salter et al. [10] that also used the RECOVER scoring system. Future research is needed to rigorously standardize and optimize long COVID definitions specific to the MSRD populations [13]. Third, pwMSRD differed from controls in demographic and clinical profiles, acute COVID‐19 severity, and vaccination status, all of which were potential confounders [33, 34]. We systematically compared these factors between groups and adjusted for these factors in analyses to reduce confounding bias. However, residual confounding could remain, and we did not conclude a causal relationship between MSRD and long COVID, which warrants future investigation.
The PRO analysis in this study was consistent with the inverse association between long COVID and functional capacity in the general population. More importantly, we found a synergism between MSRD and long COVID on physical function and depression, underscoring the need for better management of long COVID in pwMSRD. While prior research suggested that acute COVID‐19 infection does not immediately affect disability, our findings indicated that long COVID contributed to accelerated long‐term disability worsening or accumulation [35]. Future longitudinal studies with extended follow‐up are necessary to disentangle the relationships between acute infection, immediate disability changes, long COVID, and long‐term disability progression. Interestingly, progressive MS and BCD treatment, which were associated with severe acute COVID‐19, did not significantly contribute to long COVID development in this study. This could be due to the stronger impact of acute infection severity, which was accounted for by our analyses [2].
This study has several strengths. First, the multicenter design provided a large, representative sample of pwMSRD and controls from the Northeastern and Mid‐Atlantic regions of the United States. Second, the inclusion of a control group enabled direct comparisons, aiding future research and clinical guidance. Third, the post‐infection symptom survey systematically assessed a broad range of symptoms while distinguishing between new and worsening symptoms, facilitating a nuanced comparison of symptom manifestations. Finally, we made efforts to reduce misclassification bias and enhance the robustness of findings by assessing different long COVID definitions.
There are also limitations. First, the retrospective survey design might introduce selection and recall biases. Second, some control participants might have been caregivers of pwMSRD. Because we lacked linkage to confirm caregiver–patient pairs, we could not address clustering or dependency, which could bias estimates of the association between MSRD status and long COVID. Third, the cross‐sectional study required cautious interpretation as we did not establish causality. Residual confounding remains possible, particularly due to unmeasured factors not captured in the survey (e.g., treatment duration and socioeconomic factors). Future longitudinal studies with more comprehensive participant profiles and robust causal inference methods are needed to disentangle the complex relationships among the underlying neuroinflammatory disease, long COVID, and patient outcomes. Fourth, we used the PDDS scale, a well‐validated PRO in MS, as the primary disability outcome given the nature of the survey‐based study design. While PDDS complements clinician‐assessed disability measures such as EDSS [36], future studies should examine whether long COVID in MS is also associated with higher EDSS scores and other objective neuroimaging and fluid biomarkers to strengthen the evidence and better guide clinical management. Finally, the study findings may not generalize to people who did not volunteer to answer this survey and to racial and ethnic minorities, highlighting the need for future validation studies leveraging real world clinical data [37, 38].
In summary, pwMSRD exhibited heightened odds for long COVID, particularly due to worsening pre‐existing symptoms, which exacerbated functional decline and accelerated disability progression. Our study provided real‐world evidence of long COVID development and consequences in the vulnerable population of MSRD and posed clinical implications regarding recognition, evaluation, and management of long COVID in the post‐pandemic era. Specifically, symptoms such as worsening dizziness, weakness, and muscle spasms post‐infection were strongly associated with long COVID and should raise greater clinician awareness. For pwMSRD with a history of probable or confirmed SARS‐CoV‐2 infection, long COVID should be considered as a differential diagnosis in the setting of clinical relapse and progression or when managing disabling multi‐system symptoms. In routine clinical practice, it may be beneficial to implement structured disability and symptom monitoring as well as early initiation of rehabilitation programs and symptomatic management for pwMSRD with suspected long COVID to complement the current standard of MS care. Our study highlights the urgent need for validated objective biomarkers to facilitate faster and accurate diagnosis of long COVID, particularly in populations such as those with MSRD, where symptom overlap with long COVID complicates clinical assessment. Refining long COVID definitions and diagnostic criteria to account for these complexities is essential for improving care in these vulnerable patient populations.
Author Contributions
C.H., J.S., L.M., and Z.X. contributed to the conception and design of the study; A.B.‐O., C.P., C.S.R., B.W.G., P.L.D.J., and E.E.L. contributed to the acquisition and analysis of data; M.D., S.F.B., K.K., K.O., E.S., and E.L.S.W. contributed to drafting the text or preparing the figures; C.H., J.S., L.M., and Z.X. contributed to the drafting and editing of the manuscript.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Appendix S1: Supporting information.
Acknowledgments
We thank all research participants for their efforts.
Funding: This study was supported by the National Institute of Neurological Disorders and Stroke (NINDS) (R01NS098023 and R01NS124882 to Z.X.).
Chen Hu, Jiyeon Son, and Lindsay McAlpine share co‐first authorship.
Funding Statement
This work was funded by National Institute of Neurological Disorders and Stroke (NINDS) grants R01NS098023 and R01NS124882.
Data Availability Statement
Individual de‐identified participant data and data dictionaries will be shared upon request and with permission from the participating institutions. Requests for data access should be directed to the corresponding author, and agreements will be facilitated to ensure the ethical use of the shared data. Access to the data will be granted to qualified researchers whose proposals have been reviewed and approved by the corresponding author and participating institutions. Approved data will be provided via a secure transfer mechanism, ensuring compliance with all applicable ethical and institutional guidelines. Code for analyses and figures is publicly available at: https://github.com/xialab2016/Post‐COVID‐Sequelae.
References
- 1. Louapre C., Collongues N., Stankoff B., et al., “Clinical Characteristics and Outcomes in Patients With Coronavirus Disease 2019 and Multiple Sclerosis,” JAMA Neurology 77, no. 9 (2020): 1079–1088, 10.1001/jamaneurol.2020.2581. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Prosperini L., Arrambide G., Celius E. G., et al., “COVID‐19 and Multiple Sclerosis: Challenges and Lessons for Patient Care,” Lancet Regional Health – Europe 44 (2024): 100979, 10.1016/j.lanepe.2024.100979. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Han Q., Zheng B., Daines L., and Sheikh A., “Long‐Term Sequelae of COVID‐19: A Systematic Review and Meta‐Analysis of One‐Year Follow‐Up Studies on Post‐COVID Symptoms,” Pathogens 11, no. 2 (2022): 269, 10.3390/pathogens11020269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Thompson E. J., Williams D. M., Walker A. J., et al., “Long COVID Burden and Risk Factors in 10 UK Longitudinal Studies and Electronic Health Records,” Nature Communications 13, no. 1 (2022): 3528, 10.1038/s41467-022-30836-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Davis H. E., McCorkell L., Vogel J. M., and Topol E. J., “Long COVID: Major Findings, Mechanisms and Recommendations,” Nature Reviews. Microbiology 21, no. 3 (2023): 133–146, 10.1038/s41579-022-00846-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Thaweethai T., Jolley S. E., Karlson E. W., et al., “Development of a Definition of Postacute Sequelae of SARS‐CoV‐2 Infection,” Journal of the American Medical Association 329, no. 22 (2023): 1934–1946, 10.1001/jama.2023.8823. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Griffin D. O., “Postacute Sequelae of COVID (PASC or Long COVID): An Evidenced‐Based Approach,” Open Forum Infectious Diseases 11, no. 9 (2024): ofae462, 10.1093/ofid/ofae462. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Nalbandian A., Sehgal K., Gupta A., et al., “Post‐Acute COVID‐19 Syndrome,” Nature Medicine 27, no. 4 (2021): 601–615, 10.1038/s41591-021-01283-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Ely E. W., Brown L. M., and Fineberg H. V., “Long Covid Defined,” New England Journal of Medicine 391, no. 18 (2024): 1746–1753, 10.1056/NEJMsb2408466. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Salter A., Lancia S., Cutter G. R., Fox R. J., and Marrie R. A., “Post‐Acute Sequela of COVID‐19 Infection in Individuals With Multiple Sclerosis,” Multiple Sclerosis 31, no. 3 (2025): 314–323, 10.1177/13524585241310104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. “Persistent Complement Dysregulation With Signs of Thromboinflammation in Active Long Covid | Science,” accessed April 1, 2025, https://www.science.org/doi/full/10.1126/science.adg7942. [DOI] [PubMed]
- 12. Conway S. E., Healy B. C., Zurawski J., et al., “COVID‐19 Severity Is Associated With Worsened Neurological Outcomes in Multiple Sclerosis and Related Disorders,” Multiple Sclerosis and Related Disorders 63 (2022): 103946, 10.1016/j.msard.2022.103946. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Garjani A., Middleton R. M., Nicholas R., and Evangelou N., “Recovery From COVID‐19 in Multiple Sclerosis,” Neurology Neuroimmunology & Neuroinflammation 9, no. 1 (2022): e1118, 10.1212/NXI.0000000000001118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Czarnowska A., Kapica‐Topczewska K., Zajkowska O., et al., “Symptoms After COVID‐19 Infection in Individuals With Multiple Sclerosis in Poland,” Journal of Clinical Medicine 10, no. 22 (2021): 5225, 10.3390/jcm10225225. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Carlile O., Briggs A., Henderson A. D., et al., “Impact of Long COVID on Health‐Related Quality‐Of‐Life: An OpenSAFELY Population Cohort Study Using Patient‐Reported Outcome Measures (OpenPROMPT),” Lancet Regional Health – Europe 40 (2024): 100908, 10.1016/j.lanepe.2024.100908. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Levin S. N., Venkatesh S., Nelson K. E., et al., “Manifestations and Impact of the COVID‐19 Pandemic in Neuroinflammatory Diseases,” Annals of Clinical Translational Neurology 8, no. 4 (2021): 918–928, 10.1002/acn3.51314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Levit E., Cohen I., Dahl M., et al., “Worsening Physical Functioning in Patients With Neuroinflammatory Disease During the COVID‐19 Pandemic,” Multiple Sclerosis and Related Disorders 58 (2022): 103482, 10.1016/j.msard.2021.103482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Epstein S., Xia Z., Lee A. J., et al., “Vaccination Against SARS‐CoV‐2 in Neuroinflammatory Disease: Early Safety/Tolerability Data,” Multiple Sclerosis and Related Disorders 57 (2022): 103433, 10.1016/j.msard.2021.103433. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Riley C., Venkatesh S., Dhand A., et al., “Impact of the COVID‐19 Pandemic on the Personal Networks and Neurological Outcomes of People With Multiple Sclerosis: Cross‐Sectional and Longitudinal Case‐Control Study,” JMIR Public Health and Surveillance 10, no. 1 (2024): e45429, 10.2196/45429. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Jakimovski D., Kavak K. S., Longbrake E. E., et al., “Impact of Resilience, Social Support, and Personality Traits in Patients With Neuroinflammatory Diseases During the COVID‐19 Pandemic,” Multiple Sclerosis and Related Disorders 68 (2022): 104235, 10.1016/j.msard.2022.104235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Simpson‐Yap S., Pirmani A., Kalincik T., et al., “Updated Results of the COVID‐19 in MS Global Data Sharing Initiative,” Neurology Neuroimmunology & Neuroinflammation 9, no. 6 (2022): e200021, 10.1212/NXI.0000000000200021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Abramoff B. A., Hentschel C., Dillingham I. A., et al., “The Association of Multiple Sclerosis, Traumatic Brain Injury, and Spinal Cord Injury to Acute and Long COVID‐19 Outcomes,” PM & R: The Journal of Injury, Function, and Rehabilitation 16, no. 6 (2024): 553–562, 10.1002/pmrj.13121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Marrodan M., Alessandro L., Farez M. F., and Correale J., “The Role of Infections in Multiple Sclerosis,” Multiple Sclerosis 25, no. 7 (2019): 891–901, 10.1177/1352458518823940. [DOI] [PubMed] [Google Scholar]
- 24. Muñoz‐Jurado A., Escribano B. M., Agüera E., Caballero‐Villarraso J., Galván A., and Túnez I., “SARS‐CoV‐2 Infection in Multiple Sclerosis Patients: Interaction With Treatments, Adjuvant Therapies, and Vaccines Against COVID‐19,” Journal of Neurology 269, no. 9 (2022): 4581–4603, 10.1007/s00415-022-11237-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Sormani M. P., De Rossi N., Schiavetti I., et al., “Disease‐Modifying Therapies and Coronavirus Disease 2019 Severity in Multiple Sclerosis,” Annals of Neurology 89, no. 4 (2021): 780–789, 10.1002/ana.26028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Orrù V., Serra V., Marongiu M., et al., “Implications of Disease‐Modifying Therapies for Multiple Sclerosis on Immune Cells and Response to COVID‐19 Vaccination,” Frontiers in Immunology 15 (2024): 1416464, 10.3389/fimmu.2024.1416464. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Davis‐Porada J., Tozlu C., Aiello C., et al., “Durable T Cell Immunity to COVID‐19 Vaccines in MS Patients on B Cell Depletion Therapy,” NPJ Vaccines 10, no. 1 (2025): 98, 10.1038/s41541-025-01151-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Lai Y. J., Liu S. H., Manachevakul S., Lee T. A., Kuo C. T., and Bello D., “Biomarkers in Long COVID‐19: A Systematic Review,” Frontiers in Medicine 10 (2023): 1085988, 10.3389/fmed.2023.1085988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Zhu W., Chen C., Zhang L., et al., “Association Between Serum Multi‐Protein Biomarker Profile and Real‐World Disability in Multiple Sclerosis,” Brain Communications 6, no. 1 (2023): fcad300, 10.1093/braincomms/fcad300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Su Y., Yuan D., Chen D. G., et al., “Multiple Early Factors Anticipate Post‐Acute COVID‐19 Sequelae,” Cell 185, no. 5 (2022): 881–895.e20, 10.1016/j.cell.2022.01.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Cervia‐Hasler C., Brüningk S. C., Hoch T., et al., “Persistent Complement Dysregulation With Signs of Thromboinflammation in Active Long Covid,” Science 383 (2024): eadg7942, 10.1126/science.adg7942. [DOI] [PubMed] [Google Scholar]
- 32. Delgado‐Alonso C., Delgado‐Alvarez A., Díez‐Cirarda M., et al., “Cognitive Profile in Multiple Sclerosis and Post‐COVID Condition: A Comparative Study Using a Unified Taxonomy,” Scientific Reports 14, no. 1 (2024): 9806, 10.1038/s41598-024-60368-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Montgomery S., Hillert J., and Bahmanyar S., “Hospital Admission due to Infections in Multiple Sclerosis Patients,” European Journal of Neurology 20, no. 8 (2013): 1153–1160, 10.1111/ene.12130. [DOI] [PubMed] [Google Scholar]
- 34. Pérez C. A., Zhang G. Q., Li X., et al., “COVID‐19 Severity and Outcome in Multiple Sclerosis: Results of a National, Registry‐Based, Matched Cohort Study,” Multiple Sclerosis and Related Disorders 55 (2021): 103217, 10.1016/j.msard.2021.103217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Salter A., Lancia S., Cutter G. R., Fox R. J., and Marrie R. A., “Effects of COVID‐19 Infection on Symptom Severity and Disability in Multiple Sclerosis,” Neurology 104, no. 2 (2025): e210149, 10.1212/WNL.0000000000210149. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Learmonth Y. C., Motl R. W., Sandroff B. M., Pula J. H., and Cadavid D., “Validation of Patient Determined Disease Steps (PDDS) Scale Scores in Persons With Multiple Sclerosis,” BMC Neurology 13, no. 1 (2013): 37, 10.1186/1471-2377-13-37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Hong C., Wen J., Zhang H. G., et al., “Label Efficient Phenotyping for Long COVID Using Electronic Health Records,” NPJ Digital Medicine 8, no. 1 (2025): 405, 10.1038/s41746-025-01617-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Huang F., Hou J., Zhou N., et al., “Advancing the Use of Longitudinal Electronic Health Records: Tutorial for Uncovering Real‐World Evidence in Chronic Disease Outcomes,” Journal of Medical Internet Research 27 (2025): e71873, 10.2196/71873. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Appendix S1: Supporting information.
Data Availability Statement
Individual de‐identified participant data and data dictionaries will be shared upon request and with permission from the participating institutions. Requests for data access should be directed to the corresponding author, and agreements will be facilitated to ensure the ethical use of the shared data. Access to the data will be granted to qualified researchers whose proposals have been reviewed and approved by the corresponding author and participating institutions. Approved data will be provided via a secure transfer mechanism, ensuring compliance with all applicable ethical and institutional guidelines. Code for analyses and figures is publicly available at: https://github.com/xialab2016/Post‐COVID‐Sequelae.