Flare Association

Multiple publications quantify rates of postvaccination flare by identifying flares in a single group of subjects exposed to vaccination. Flare is defined variably, most frequently via cross-sectional or retrospective participant self-report, proxy measures such as medication adjustments, as well as determination by researchers via retrospective chart review. Only a minority of studies ascertain flare through direct clinical assessment by investigators. Across studies, rates of flare following vaccination range from as low as 2.1% to as high as 26.6%, within postvaccination windows varying from a few days to 6 months.

Relatively few studies utilize comparator groups to perform hypothesis testing for a statistical association between vaccination and flare. Early investigations undertaking such analyses have suggested COVID-19 vaccines are well tolerated among individuals with SARD. A multicenter European prospective cohort study of 623 patients with SARD and non-rheumatic autoimmune disease enrolled consecutive patients to compare incidence of flare in those who received vaccination and those who did not. The investigators administered questionnaires electronically and, during in-person medical consultations, monitored patients telephonically every 2 months to assess for increased disease activity. Data were collected from February to May 2021. Flare was defined as an increase in activity of typical disease symptoms, as evidenced by hospital admission, treatment escalation, or serologic evidence of inflammation. Over a median follow-up time of 180 days, the investigators found no statistically significant difference in rates of flare between vaccinated versus unvaccinated participants (6% vs 8% of subjects, P = .302). Three out of 25 flares in the vaccinated group occurred after the first dose of vaccination, with the remainder occurring after the second dose. Notably, more than half of the participants had already received the first dose of the vaccine prior to enrolling in the study, introducing potential selection bias, and confounding by indication. Additionally, there was imbalanced recruitment in the cohort, with overrepresentation of vaccinated participants recruited online and unvaccinated patients recruited in the hospital setting. This differential method of enrollment may have introduced ascertainment bias if the unvaccinated patient group had been systematically more likely than the vaccinated group to be identified as flaring.

A prospective, multicenter cohort study from December 2020 to March 2021 examined the immunogenicity, efficacy, and safety of the BNT162b2 vaccine in adult patients with rheumatoid arthritis (RA), psoriatic arthritis (PsA), axial spondyloarthritis (AxSpA), systemic lupus erythematosus (SLE), idiopathy inflammatory myositis (IIM), and a variety of vasculitides as compared to unmatched healthy controls. Participants received 2 doses of vaccine without adjustment of immunosuppression (with the exception of rituximab) and were contacted within 2 weeks of the first dose and 2 to 6 weeks of the second dose for a telephone-based survey assessing self-reported adverse events. The primary outcomes were levels of immunoglobulin G (IgG) neutralizing antibody levels against the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spike glycoproteins, self-report, or medical record documentation of laboratory-confirmed COVID-19 infection and participant-reported adverse events following vaccination. As an exploratory, descriptive secondary analysis, the authors evaluated pre-vaccination and postvaccination disease activity for subjects with RA, PsA, AxSpA, and SLE within variable pre-vaccination and postvaccination time periods. Baseline disease activity in the 3 months preceding vaccination was determined from chart review, and postvaccination disease activity was assessed via in-person clinical evaluation using validated disease activity indices within 2 to 6 weeks of the second dose. Data from 686 individuals with SARD and 121 controls were analyzed, with similar self-reported rates of mild adverse events following vaccination. Two patients in the SARD group died after vaccination—one from a myocardial infarction and the other from fulminant hemorrhagic cutaneous vasculitis—although the authors report there was no obvious causal link to the vaccine identified in either case. Disease activity scores pre/postvaccine were stable for most subjects with RA, PsA, SLE, and ankylosing spondylitis across all clinical disease activity indices, and there were roughly as many participants with increased disease activity scores as decreased disease activity scores. Notably, no statistical testing was reported due to the exploratory, descriptive nature of the analysis.

Another observational study prospectively followed participants with SARD who received an mRNA-based vaccine between December 2020 and April 2021. Subjects completed a baseline survey and received surveys 7 days after each vaccine dose, which included questions on local and systemic side effects. One month after the second dose, participants were sent an additional survey on incidence of flare, as well as recent prior history of flares. In total, 1377 individuals participated in the study, with 47% having a diagnosis of inflammatory arthritis, 20% SLE and 20% overlap connective tissue disease. Prevalence of at least one flare in the 6 months prior to the first dose of vaccine was reported by 56% of subjects; following vaccination, 11% of subjects reported a flare requiring treatment, 60% of which occurred following the second dose. Flares that required treatment lasted a median of 10 days (interquartile range [IQR] 6–22 days), with the most common treatment reported as oral corticosteroids (75%).

A retrospective cohort study of 5493 patients in Hong Kong with RA compared the risk of arthritis flare in age and sex-matched unvaccinated versus vaccinated who had received 2 doses of immunization. Cohort creation was performed via a territory-wide search of the electronic health record (EHR) using the International Classification of Diseases Ninth version, Clinical Modification diagnosis codes, with linked immunization records from the Hong Kong Department of Health. Vaccine recipients received either BNT162b2 (Pfizer-BioNTech) or CoronaVac, and inactivated virus vaccine. Flare was defined by any outpatient clinic visit or hospitalization related to RA. The index date was set for both vaccinated subjects and their matched controls as the date of vaccination for vaccinated participants, and the investigators evaluated for the occurrence of the outcomes or death through the end of July 2021. After propensity score weighting, there was no statistically significant association between vaccination with either vaccine and proxy measures of flare over a median follow-up period of 32 days (IQR 14–72).

Several other studies focusing on a range of SARD diagnoses have found no differences between within-subject rates of flare before and after vaccination, including for individuals with RA, PsA, SLE, and mixed disease cohorts. A cohort study of patients with RA and PsA enrolled in a multicenter registry in Spain examined disease activity in participants who received at least 2 doses of an mRNA vaccine or one dose of either the Ad26.COV2.S (manufactured by Johnson & Johnson-Janssen) or ChAdOx1-S/nCoV-19 (produced by Oxford-AstraZeneca) vaccines between April and December 2021. Investigators collected data on immunosuppression use and vaccinations, as well as disease activity (DAS28) at every visit. The authors evaluated change in DAS28 as a function of vaccination, comparing the cohort-wide distribution of pre/postvaccination DAS28 scores by strata of disease activity (remission: <2.6, low: 2.6–3.2, moderate: 3.2–5.1, high:≥5.1). The researchers used the DAS28 score that most closely preceded date of vaccination for pre-vaccination disease activity and then compared this to the most recent DAS28 score within 3 months following vaccination; flare was defined as an increase in DAS28 greater than 1.2. 731 patients with RA and 310 patients with PsA had paired pre/postvaccination DAS28 scores available for comparison. Using McNemar’s test for paired samples, the authors found no significant difference in the distribution of disease activity before and after vaccination across both RA and PsA populations. The rates of flare for patients in the 6 months before and after vaccination were similar (RA: 12.6% and 14.2%; PsA: 10.2% and 9.9%), with no significant difference found on multivariate analysis (RA: odds ratio [OR] 1.14, 95% confidence interval [CI] 0.8–1.56; PsA: OR: 0.93, 95% CI 0.54–1.60) or on sensitivity analysis controlling for seasonality (RA: OR 1.42, 95% CI 0.94–2.13; PsA: OR 1.25, 95% CI 0.83–2.02).

A self-controlled case series analysis studied the incidence of flare in the 21 days following COVID-19 vaccination in a large administrative dataset in the United Kingdom. Adults with one or more outpatient visits for SARD and one or more prescriptions for an immunosuppressive medication before December 2020 formed a cohort of 3554 individuals. Vaccine exposure was determined through linkage to the National Health Service database. Flare was defined by an outpatient consultation with a diagnostic code for SARD in combination with a steroid prescription on the same or following day; flares within a 14 day period were considered part of the same event. The authors found a negative association between first dose vaccination and subsequent flare in a 21 day risk period, with an adjusted incidence rate ratio for all 3 doses of 0.89 (95% CI, 0.80–0.98). Analyses of first dose stratified by prior COVID-19 infection and vaccine type found no significant differences between strata and continued to show a negative association with flare in the 21 day postvaccination period relative to baseline. There were no other significant associations between vaccination and flare for subsequent vaccine dose number or type (BNT162b2 vs vectored-DNA). On sub-analysis stratified by disease (RA, spondyloarthritis and polymyalgia rheumatica/giant cell arteritis), this protective association remained significant only for RA, likely due to the relatively small sample sizes for other diseases. A strength of this study is its methodologic design, which minimizes recall bias and leverages large datasets, though a notable weakness is its use of proxy measures for flare.

A survey-based case-crossover study by our group conducted through a COVID-19 registry at a large rheumatology referral center in New York City examined risks of flare associated with vaccination in a subset of 434 subjects from March 2021 to September 2022. This study used a case-crossover design, in which data are longitudinally collected from subjects during periods when their SARD is flaring as well as during periods when it is not. Participants serve as their own within-person controls, which mitigates confounding by time-invariant covariates. A variety of exposures are ascertained during the risk windows, including COVID-19 vaccination; if vaccination is associated with increased risk of flare, then it will be encountered more than expected by random chance during periods preceding report of flare. In this study, vaccinations were detected by EHR review and participant self-report, as well as via direct review of immunization cards. Flare was ascertained by participant report and included validated disease-specific flare instruments for RA, SLE, and scleroderma. This study found that there was no association between reporting SARD flare and receiving COVID-19 vaccination in the preceding 2, 7, or 14 days. Further, there remained no association in analyses stratified by disease type (inflammatory arthritis vs connective tissue disease), age, sex, or vaccine manufacturer. When flares occurred, they tended to be rated as more severe by subjects with inflammatory arthritis relative to connective tissue disease, male sex, and in those who had received a vaccine in the preceding 3 to 7 days as compared to no recent vaccine exposure. An earlier analysis among participants in the parent COVID-19 registry who had received 2 doses detected a 17.0% rate of self-reported flares following vaccinations but did not compare this to rates of flare during vaccine unexposed periods. These results illustrate the importance of a comparator in contextualizing point estimates of flare incidence.

While the vast majority of studies have found no association between COVID-19 immunization and increased disease activity, a recent study of patients with IIM is notable for its finding of a statistically significant association between flare and second-dose vaccination. The analysis evaluated flare in subjects with IIM who completed surveys as part of the COVID-19 vaccination in autoimmune disease (COVAD) study. COVAD is an international survey collaboration involving over 100 investigators from over 20 countries, with the intention of better understanding a variety of vaccine experiences and outcomes of patients with rheumatic diseases. Cross-sectional online questionnaires were distributed from March 2021 to February 2022 and February to June 2022 to a convenience sample of patients seen in participating clinics. Survey items queried demographics, SARD diagnosis, treatment and symptom status, COVID-19 infection and vaccination history, vaccine adverse events, SARD flares, and patient-reported outcome measures. Vaccinated subjects were also asked about the status of their SARD in the period prior to vaccination (inactive and in remission, active but stable, active and improving, active and worsening). A flare of IIM was ascertained by any 1 of 4 criteria: participant self-report of flare, an escalation or addition of immunosuppression, an increase of greater than 7.9 in 10 item Patient-Reported Outcome Measurement Information System physical function form scores (defined as the minimally clinically significant improvement difference), or self-report of new clinical features deemed consistent with flare by the investigators (rash, myositis, arthritis, or rise in muscle enzymes). The authors found that respondents were 1.2 times more likely to report active or worsening disease following the second dose as compared to the before the first dose (95% CI 1.03–1.6, P = .025). However, the strength of this conclusion may be limited by recall, response, and selection biases.

Multiple systematic reviews and meta-analyses have examined observational data—inclusive of these studies reviewed earlier, and others—in aggregate. An early systematic review and meta-analysis in 2022 collated observational studies published through August 31, 2021, finding a 6.9% rate of disease flare. History of flare in the prior year and greater disease activity were significantly associated with increased risk of flare. An updated systematic literature search up to April 2022 of studies reported efficacy and safety of both mRNA vaccines and the AstraZeneca vaccine. Across 36 observational studies, the authors determined a range of postvaccination flare rates from 0% to 15.9%. Predictors for postvaccination flare included prior COVID-19 infection, use of combination immunosuppression, disease flare within the past 6 months, and peri-vaccination interruption of SARD treatment. A systematic review and meta-analysis of studies published between June 2020 and September 2022 pooled results across 74 studies including SARD and non-SARD immune mediated diseases, with an estimated prevalence of relapse, flare, or disease exacerbation of 6.24% after removing outliers. A methodologic limitation is the heterogeneity in study populations, vaccination schemes, and outcome ascertainment, with flare defined variably in heterogenous participant populations.

Several randomized control trials examining methotrexate continuation versus temporary discontinuation following COVID-19 vaccination have reported an incidence of flare as an outcome. While these studies do not directly compare flare rates with and without vaccination exposures, they offer additional data from interventional investigations that modulated immunosuppression in the peri-vaccination period. In a UK-based trial, 254 patients with immune-mediated inflammatory diseases (including various SARD as well as psoriasis and atopic dermatitis) were randomized to continue methotrexate as usual or hold therapy for 2 weeks following COVID-19 booster vaccination. Self-reported flares up to 4 weeks after vaccination were experienced by 56% of participants with available data who held methotrexate as opposed to 31% of those who did not (OR 3.1, 95% CI 1.8–5.4). Flares were generally mild and did not require consultation with a health care professional (14% vs 11%). ^,

Flare Predictors

The literature is mixed on whether particular disease-related covariates are associated with postvaccination flare. The most consistently reported risk factors across studies are higher baseline disease activity or prior flares within a 6 to 12 month window preceding vaccination. ^, In certain analyses, participants with inflammatory arthritis have been found to have significantly increased risk of flare relative to those with connective tissue disease ^, or IIM, whereas other publications report no differences among SARD subgroups ^, older age, ^{, , ,} use of prednisone, ^{, , ,} and female sex ^{, ,} have all been identified variably as both risk factors and protective factors.

Disease-specific risk factors for flare have been described, though these similarly have not been reproducible across different study populations. Such variables include dsDNA positivity and belimumab use in connective tissue disease, receipt of mRNA-1273 vaccine and use of azathioprine (vs hydroxychloroquine) in SLE, ^, anti-IL-1 and high-dose colchicine therapy in Familial Mediterranean Fever, ^, IL-6i and JAKi use (vs TNFi) for RA, IL-12/23i therapy (vs TNFi) for PsA, and receipt of Moderna compared to Pfizer in PsA.