Myalgic Encephalomyelitis/Chronic Fatigue Syndrome After SARS-CoV-2 Infection

This cohort study investigates the incidence of myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS)–like symptoms in individuals tested for SARS-CoV-2 infection and whether ME/CFS symptom prevalence is associated with positive or negative COVID-19 test results.


Introduction
Chronic medical syndromes can occur after a variety of acute infections (eg, postpolio syndrome, post-Epstein Barr virus syndrome). 1,2These postacute infection syndromes (PAISs) share similar symptoms, including functional impairment associated with fatigue, exertion intolerance, and cognitive problems.Myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS), a complex, chronic, debilitating condition with systemic manifestations often linked to a prior acute influenzalike illness, 3,4 is emblematic of the largely enigmatic group of PAISs.Fatigue is the most common symptom among patients with post-COVID condition or long COVID, [5][6][7] and the symptom profile of postacute sequelae of COVID-19 and of other PAISs overlaps with characteristic symptoms of ME/CFS.The prevalence of long COVID symptoms in the US, 8 including those overlapping with ME/CFS symptoms, suggests that millions of individuals will be impacted, with medical costs in the billions, 9 emphasizing the need to understand PAISs.
The COVID-19 pandemic raised awareness of PAISs and provides a unique opportunity to examine the occurrence of ME/CFS following a specific infection.Through the Innovative Support for Patients with SARS-CoV-2 Infections Registry (INSPIRE) study, we collected thousands of selfreported outcome data points, allowing the identification and evaluation of patients with ME/CFSlike illness.The objective of this analysis was to evaluate the occurrence of ME/CFS among INSPIRE participants following symptomatic acute illness that prompted their COVID-19 test and to compare the odds of ME/CFS in the COVID-19-positive cohort and the COVID-19-negative cohort.

Study Design and Data Source
INSPIRE was a multicenter, prospective, longitudinal registry cohort study that enrolled individuals who experienced an acute index illness suggestive of COVID-19 between December 11, 2020, and August 29, 2022.Eight geographically diverse study sites across the US recruited and enrolled participants with a protocol and methods previously described. 10Recruitment occurred in person, by email or telephone, and through electronic advertisement.A secure online platform (Hugo; Hugo Health LLC) facilitated collection of consent and survey distribution.Participants self-enrolled by first completing an online eligibility screener, during which they were asked to self-report COVID-19-like symptoms that they experienced within the past 42 days and provide documentation of their valid COVID-19 test and test result.If respondents were deemed eligible for participation, they were provided a consent form on the online platform.Follow-up surveys were collected through February 28, 2023.The institutional review board at each of the study sites reviewed and approved this study.
The Centers for Disease Control and Prevention (CDC) reviewed the project and determined the study was nonengaged human participants research.The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. 11

Cohort Definition
The INSPIRE study included adult participants (aged Ն18 years) who were fluent in English or Spanish, had access to an internet-enabled device to allow for participation, and had self-reported symptoms suggestive of acute SARS-CoV-2 infection at the time of their SARS-CoV-2 test.
Participants had to have been tested for SARS-CoV-2 with a US Food and Drug Administrationapproved or authorized molecular or antigen-based assay within 42 days before their study enrollment.3][14] We excluded participants who did not link their electronic health portal connection with the online platform, did not complete the baseline survey, died or withdrew from the study before 3 months, or did not have valid COVID-19 test results.The flowchart (eFigure 1 in Supplement 1) has detailed information on recruitment and follow-up completion.
Participants were grouped as either COVID-19 positive or COVID-19 negative based on their index SARS-CoV-2 test result.If they had more than 1 test within 7 days of enrollment and results were discordant, we considered the positive test result to be the true result.If a participant's test result changed more than 7 days after enrollment, we kept the participant in their initial group and adjusted for subsequent positive test results as a covariate.

Variables
Participants completed surveys at baseline and at 4 quarterly follow-up times.Participants selfreported sociodemographic data, including age, gender (female, male, or transgender, nonbinary, or other gender), race (Asian, Native Hawaiian, or Other Pacific Islander; Black or African American; White; or other [groups are listed in eAppendix 3 in Supplement 1] or multiracial), ethnicity (Hispanic, Latino, or of Spanish origin or not Hispanic, Latino, or of Spanish origin), educational level, income, employment status, health insurance status, and marital status.Self-reported race and ethnicity items on the baseline survey were included because patient outcomes have been reported to vary across racial and ethnic groups.Standardized questions assessing physical and mental health, symptoms, access to care, and work-related outcomes were included in the baseline and follow-up surveys. 15Participants listed other conditions that could contribute to ME/CFS symptoms in free-text responses (a list of the free-text entries is included in eAppendix 2 in Supplement 1).The Patient-Reported Outcomes Measurement Information System-29 (PROMIS-29) profile, version 2.1 was administered at baseline and at all follow-up times. 16,17

ME/CFS Outcomes of Interest
We used the 2015 Institute of Medicine (IOM) criteria 15 for binary classification of the primary ME/CFS outcome, operationalized using participants' responses to symptoms from the CDC ME/CFS Symptom Screener-Short Form, version 1.2 (eAppendix 1 in Supplement 1) and the Physical Function subscale of the PROMIS-29, version 2.1 18 profile (algorithm in eTable 1 in Supplement 1).In brief, the criteria include activity limitations associated with fatigue, postexertional malaise, and sleep problems as well as either cognitive impairment or orthostatic intolerance.Because ME/CFS diagnosis requires a full clinical evaluation to identify treatable conditions contributing to symptoms used in diagnosis, the self-reported information in this study only allowed determination of ME/CFS-like illness, hereafter referred to as ME/CFS.Additionally, the survey questions did not allow clear detection of the presence of chronic symptoms before participants' acute index illness.The count of ME/CFS diagnostic criteria met (0-5) was also used as an outcome for this study.
For simplicity of monitoring ME/CFS-related symptoms, we grouped participants as ever having ME/CFS symptoms if they met criteria at any time point (acute index illness through 12 months) or never having ME/CFS symptoms if they did not meet criteria at any time point.We focused regression analysis results comparing ME/CFS classification at 3 months and beyond but, for parsimony, included those who ever met the ME/CFS definition at any time point to summarize participant characteristics descriptively.

Statistical Analysis
Statistical analyses were conducted using SAS, version 9.4 (SAS Institute Inc).All tests were 2-sided with a significance threshold of P = .05.Bivariate analyses were performed to examine the

JAMA Network Open | Public Health
Myalgic Encephalomyelitis/Chronic Fatigue Syndrome After SARS-CoV-2 Infection association between participant characteristics and ME/CFS status at any time point and whether characteristics were balanced between COVID-19-positive and COVID-19-negative participants.This identified significant differences between the COVID-19-positive and COVID-19-negative groups in baseline characteristics that were associated with ME/CFS outcomes.To address the imbalanced distribution of these confounders, we used the inverse propensity score weighting (IPW) technique.
We assessed (Table 1 and eTable 2 and eFigure 2 in Supplement 1) how well the distribution of confounders and covariates were balanced between the COVID-19-positive and COVID-19-negative groups through IPW.Because of the need for balancing groups by COVID-19 status using IPW, all results in the main text are weighted, and observed (unweighted) results are included in eTable2 in Supplement 1.
Incorporating IPW, we used generalized estimating equation (GEE) models to examine the association between initial SARS-CoV-2 infection and ME/CFS outcomes across time and how the risk of ME/CFS outcomes changed over time in each COVID-19 group after the index illness.Specifically, we used a GEE with a binomial distribution and logit link function for the binary outcome of ME/CFS and a GEE negative binomial model with the log link function for the counts of ME/CFS diagnostic criteria met (0-5).Both models included the initial SARS-CoV-2 infection status at enrollment, time points (baseline and 4 quarterly follow-up times), any postbaseline SARS-CoV-2 infection as a timevarying covariate, and hospitalization status for the index illness, which was not fully balanced by IPW (eFigure 2 in Supplement 1).Additionally, we included adjustment for the SARS-CoV-2 variant period at enrollment using previously published methods. 19We also allowed initial COVID-19 status to interact with time points, subsequent SARS-CoV-2 infection, and the variant period.Marginal differences between the COVID-19-positive and COVID-19-negative groups were estimated at each time point.Simultaneously, changes over time from 3 through 12 months in each COVID-19 group were estimated for all possible paired time comparisons.To examine symptom and function measures between participants with and without ME/CFS, effect sizes (ie, Cohen d) were calculated for group mean differences.
After IPW, the frequency of symptoms of ME/CFS reported by participants ranged from 5.1% for dizziness or fainting and 5.2% for forgetfulness or memory problems to 23.5% for fatigue (Table 1).Among all 4738 participants, the survey completion rates ranged from 38.7% (3613) to 76.3% (1835) and decreased over time.Less than one-third (1357 [28.6%]) provided data at all follow-up time points The weighted percentage of participants meeting the ME/CFS criteria at 3 months was 3.4% in the COVID-19-positive group and 3.7% in the COVID-19-negative group, and there was no statistically significant difference between the COVID-19-positive and COVID-19-negative groups in the prevalence of ME/CFS at any time point through 12 months of follow-up (range, 2.8%-3.7% in the COVID-19-positive group and 3.1%-4.5% in the COVID-19-negative group) (  In weighted and adjusted results, there were no statistically significant differences in the odds of ME/CFS between COVID-19-positive and COVID-19-negative participants at any time point a Myalgic encephalomyelitis/chronic fatigue syndrome would not be identified during acute illness, but the index time was included along with all other time points for the dichotomous classification of ever vs never having ME/CFS.
b Other gender included gender nonconforming, not listed, or preferred not to answer.
c Other races listed in free-text responses entered by participants are included in eAppendix 3 in Supplement 1.
d Hospitalization for the index illness was a new question added to the 3-month survey after April 14, 2021.There were 1134 participants missing a response to this question that were included as a separate category in the analysis.
e Viral prevalence greater than 50% was used to determine the dominant variant.
f Any type of tobacco, including smokeless tobacco.

Discussion
Our findings suggest that COVID-19 is no more likely than other acute infections to be associated with ME/CFS and that acute illnesses more broadly may be associated with chronic symptom burden from ME/CFS.The odds of meeting ME/CFS criteria at quarterly follow-up time points through 12 months did not differ between the COVID-19-positive and COVID-19-negative groups, and prevalence was relatively stable over this time in both groups (range, 2.8%-4.5%).Additionally, symptom severity It contains 7 domains, including symptomoriented (sleep, pain, anxiety, depression, and energy or fatigue) and function-oriented (physical function and social role or activity limitation) measures.We assessed cognitive function via the PROMIS Short Form-Cognitive Function 8a measure.The PROMIS measures use T-score metrics with a mean of 50 and SD of 10 in a reference population (ie, the US general population).Higher scores indicate a greater degree of the concept being measured (eg, better functioning for the physical function subscale, more depressed for the depression subscale).

Table 3 )
. At each follow-up survey, approximately one-third of the COVID-19-positive group and one-third of the COVID-19-negative group (range, 31.0%-37.6%)reported 1 or more of the 5 ME/CFS symptoms assessed.In both the COVID-19-positive group and the COVID-19-negative group, unrefreshing sleep

Table 1 .
Weighted Distribution of Confounders and Covariates Between COVID-19 Groups a

Table 4
summarizes the time course of symptoms in the groups that ever or never had ME/CFS symptoms stratified by COVID-19 status.The group that ever had ME/CFS symptoms consistently rated their symptoms as more severe (ie, higher scores) compared with the group that never had ME/CFS symptoms regardless of COVID-19 status.The effect size for the difference in scores between participants who ever and never had ME/CFS symptoms for both the COVID-19-positive

Table 1 .
Weighted Distribution of Confounders and Covariates Between COVID-19 Groups a (continued) a Weighted by inverse propensity scores calculated by accounting for the confounders and covariates in the table.Variables correlated with COVID-19 status but not with the myalgic encephalomyelitis/chronic fatigue syndrome outcomes were excluded from propensity score calculation to avoid introducing noise into estimating the impact of COVID-19 infection and therefore are not included in this table.Hospitalization for the index illness was not well balanced by inverse propensity score weighting and was therefore included in the generalized estimating equation modeling.bOthergenderincludedgendernonconforming, not listed, or preferred not to answer.cOtherraceslisted in free-text responses entered by participants are included in eAppendix 3 in Supplement 1.d Hospitalization for the index illness was a new question added to the 3-month survey after April 14, 2021.There were 1134 participants missing a response to this question that were included as a separate category in the analysis.eViralprevalencegreater than 50% was used to determine the dominant variant.fAnytype of tobacco, including smokeless tobacco.gIncluded coronary artery disease, heart failure, and cardiomyopathy.h

Table 2 .
Observed Characteristics by Ever or Never Meeting Criteria for ME/CFS (continued) marginal odds ratio range, 0.84 [95% CI, 0.42-1.67] to 1.18 [95% CI, 0.55-2.51])(Figure).The odds of ME/CFS also did not significantly differ across time points in either the COVID-19-positive group or the COVID-19-negative group (Figure).The marginal incidence rate ratio (MIRR) of ME/CFS symptom count was not significantly different between the COVID-19-positive and COVID-19-negative groups at any time point (eg, 12-month MIRR: 0.96; 95% CI. 0.74-1.24)(Figure) and was not significantly different over time in the COVID-19-positive or COVID-19-negative group.Results were consistent when using a propensity score-matched sample (eFigure 3 in Supplement 1) compared with the full sample with IPW.
g Included coronary artery disease, heart failure, and cardiomyopathy.h(

Table 3 .
Observed and Weighted ME/CFS Outcomes Across Time by COVID-19 Groups a a Weighted by inverse propensity scores.bDefined as meeting criteria for postexertional malaise, unrefreshing sleep, and fatigue and for either orthostatic intolerance or cognitive impairment.

Table 4 .
Post-Acute Illness Health Status by COVID-19 Group and Ever Having ME/CFS-Like Illness, Weighted by Inverse Propensity Scores

Table 4 .
Post-Acute Illness Health Status by COVID-19 Group and Ever Having ME/CFS-Like Illness, Weighted by Inverse Propensity Scores (continued) Abbreviations: ME/CFS, myalgic encephalomyelitis/chronic fatigue syndrome-like illness; PROMIS-29, Patient-Reported Outcomes Measurement Information System-29.Nonmissing weighted total.The number of missing cases was 3 or fewer across all groups and time points.Percentages were calculated among nonmissing cases.The number of missing cases was less than 10 across all groups and time points.Flowchart of INSPIRE Study Sample eAppendix 1. CDC ME/CFS Symptom Screener-Short Form, Version 1.2 eTable 1. Operationalized Algorithm for the 2015 IOM ME/CFS Case Definition eAppendix 2. List of Conditions Potentially Contributing to ME/CFS Symptoms Entered by Participants eAppendix 3. Other Races Entered by Participants eTable 2. Observed Distribution of Confounders and Covariates Between COVID-19 Groups eFigure 2. Absolute Standardized Differences Between COVID-19 Groups eFigure 3. Marginal Effects of Index COVID-19 Status on the ME/CFS Outcomes Based on Matched Sample a b Symptoms were assessed using the Centers for Disease Control and Prevention (CDC) ME/CFS Symptom Screener.c