Introduction
After a COVID-19 infection, individuals may continue to have long-term symptoms for weeks or months. The World Health Organization (WHO) defines post COVID-19 condition (PCC) as persistent symptoms occurring 12 or more weeks after acute COVID-19, which have persisted or re-occurred for a minimum of 8 weeks and cannot be explained by alternative diagnoses [1]. Other institutions have adopted similar definitions, including the United States Centers for Disease Control and Prevention [Reference Fineberg, Brown, Worku and Goldowitz2, 3]. The most common PCC symptoms include fatigue, insomnia, general pain and discomfort, shortness of breath, cognitive issues, and anxiety or depression [1, 4, Reference Domingo, Waddell, Cheung, Cooper, Belcourt and Zuckermann5].
The estimated prevalence of PCC after COVID-19 infection has varied widely from <10% to >50% of people affected by PCC depending on the sample population, definition of PCC used to define the outcome, how the outcome was collected, and time from infection to follow-up [6–Reference Taher, Salzman, Banal, Morissette, Domingo and Cheung9]. The most recent self-report survey data estimated that among adults who had COVID-19, 19% had experienced PCC in Canada (6.8% point prevalence in June 2023) [Reference Kuang, Earl, Clarke, Zakaria, Demers and Aziz10] and 29.8% (95% CI 28.7–30.8) in the United States (8.7% point prevalence in September 2024) [11]. By the end of 2023, 56% of the world population had received a complete primary series, and 28% had received at least one booster dose of COVID-19 vaccines [12]. Given that COVID-19 and the burden of PCC continue to persist, it is important to evaluate the impact of COVID-19 vaccination on PCC, including potential benefits and/or safety concerns.
The systematic reviews (SRs) that have previously been completed on the impact of COVID-19 vaccination on PCC have all included post-acute sequalae (PAS) occurring 4 to 12 weeks post-infection [Reference Byambasuren, Stehlik, Clark, Alcorn and Glasziou13–Reference Ceban, Kulzhabayeva, Rodrigues, Di Vincenzo, Gill and Subramaniapillai19], except for one SR that reported on the association between two doses and PCC development [Reference Tsampasian, Elghazaly, Chattopadhyay, Debski, Naing and Garg20]. This living SR addresses the impact of vaccination on only PCC, which may reduce heterogeneity in the results, and includes all options for timing of vaccination relative to infection and/or PCC (pre-infection, post-infection, and post-PCC). The first version of this SR was published with data up to 13 December 2022 [Reference Jennings, Corrin and Waddell21]. Therefore, the objective of this updated living SR and meta-analysis was to assess the global evidence on the associations and safety of COVID-19 vaccination and PCC (symptoms ≥12 weeks from infection), with data up to 31 January 2024.
Methods
This living SR was conducted using standard SR methodology outlined by the Cochrane Collaboration and reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [Reference Liberati, Altman, Tetzlaff, Mulrow, Gøtzsche and Ioannidis22, Reference Higgins, Thomas, Chandler, Cumpston, Li, Page and Welch23]. A protocol was determined a priori and registered in PROSPERO (CRD42022365386); all deviations have been noted in the updated protocol document on PROSPERO.
Research question and eligibility criteria
The research questions of this SR were: (1) Does COVID-19 vaccination before COVID-19 decrease the risk of developing PCC or the risk of developing specific PCC symptoms? (2) Does COVID-19 vaccination after COVID-19 decrease the risk of developing PCC or the risk of developing specific PCC symptoms? (3) Among those that already have PCC, does COVID-19 vaccination lead to symptom changes? (4) Is it safe to get a COVID-19 vaccine for individuals who have PCC?
PCC was defined as persistent symptoms occurring 12 or more weeks after acute COVID-19, in accordance with WHO [1]. The population of interest was anyone who had COVID-19, and the intervention was vaccination with any authorized COVID-19 vaccine. The comparison group was individuals who had COVID-19 and were unvaccinated or received a different number of doses. The primary outcomes of interest were the risk of developing PCC or resolution of PCC. Published and preprint studies with an observational or experimental study design were considered for inclusion. The inclusion and exclusion criteria are detailed in the previously published review [Reference Jennings, Corrin and Waddell21] and protocol. A list of excluded studies is provided in Supplementary Table S1.
Search strategy
The Public Health Agency of Canada curated a database of COVID-19 literature from February 2020 to August 2024 [Reference Corrin, Ayache, Baumeister, Young, Pussegoda, Ahmad and Waddell24], with results maintained in the bibliographic management software EndNote20 (Clarivate, Philadelphia, PA). The search algorithm for this SR was run within the EndNote20 database with no restrictions on language and included a combination of PCC OR non-specific symptom terms AND vaccination terms (see protocol for details). The search was conducted on 21 September 2022 and has been updated four times, most recently on 31 January 2024.
Search verification
In this update, the reference lists of six relevant review articles were searched as part of search verification [Reference Marra, Kobayashi, Callado, Pardo, Gutfreund and Hsieh17–Reference Tsampasian, Elghazaly, Chattopadhyay, Debski, Naing and Garg20, Reference Fernández-de-las-Peñas, Notarte, Peligro, Velasco, Ocampo and Henry25, Reference Calabrò, Pappalardo, D’Ambrosio, Vece, Lupi and Lontano26], which yielded four studies that were added to the screening process [Reference Antonelli, Pujol, Spector, Ourselin and Steves27–Reference van der Maaden, Mutubuki, de Bruijn, Leung, Knoop and Slootweg30].
Study selection and data extraction
Search results were imported into EndNote20 (Clarivate, Philadelphia, PA) and de-duplicated. Unique references were imported into DistillerSR software (DistillerSR, Inc.) for SR management. Title/abstract and full-text relevance screening forms and a data extraction form were developed a priori and piloted by all reviewers to determine functionality. Title/abstract screening, full-text screening, study characterization, and data extraction were performed in duplicate by two independent reviewers. Study characterization included publication details (e.g. language, year), funding, conflict of interest, and study design. Data extraction included country and sampling frame, study period, population characteristics (e.g. demographics, COVID-19 severity), vaccination information (e.g. number of doses, vaccine product), and outcome-related data. Conflicts at each stage of screening and data extraction were resolved by consensus or by a third reviewer where necessary. Upon publication of a previously captured preprint, the reference was updated and re-evaluated to ensure all extracted data and risk of bias assessment reflected the published version of the article.
Risk of bias assessment
Included articles were evaluated for their risk of bias (ROB) using the Newcastle-Ottawa Scale (NOS) [Reference Wells, Shea, O’Connell, Peterson, Welch, Losos and Tugwell31]. The NOS was selected over the Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I) because the NOS is more efficient and easier to implement on a range of observational studies, and the relationship between COVID-19 vaccination and development or remission of PCC may not be a direct relationship [Reference Sterne, Hernan, Reeves, Savovic, Berkman and Viswanathan32]. ROB assessments were performed in duplicate by two independent reviewers using two pre-existing NOS forms for case-control and cohort studies, as well as a modified tool for cross-sectional studies [Reference Ribeiro, Beserra, Silva, Lima, Rocha and Coelho33]. The forms used are available in the protocol; the questions assessed selection, information, confounding, and/or reporting biases. Each tool was pretested on one article by all reviewers, and then articles were independently assessed by two reviewers. Conflicts were resolved by consensus.
Data synthesis
The complete dataset was exported into Microsoft Excel (2016), where results were grouped according to the review question addressed and tabulated to summarize the primary and secondary outcomes. Narrative synthesis of results was performed for each review question. When there were two or more studies measuring the same association for a primary outcome, random-effects meta-analyses using the restricted maximum likelihood estimator for between-study variance were performed on STATA18 (StataCorp). Meta-analyses were sub-grouped by number of doses received, the reported outcome measures, and population sub-groups including children. Those who received one dose of Janssen were considered to have a complete primary vaccine series and were placed into the two doses subgroup. For meta-analysis, risk ratios (RR) and prevalence ratios (PR) were converted to odds ratios (OR) to calculate a pooled effect (pOR) [Reference Zhang and Yu34, Reference Carazo, Skowronski, Laforce, Talbot, Falcone and Laliberté35]. Hazard ratios (HR) and incidence rate ratios (IRR) were pooled together but kept separate from ORs because HRs and IRRs measure rate of change over a defined period, whereas OR and RR report the associations across the entire study period, thus their meaning and value are different [Reference George, Stead and Ganti36].
The impact of ROB (low, moderate, and high) was examined for outcomes considered in meta-analysis and reported in Supplementary Table S2a–b. Testing for small study effects was only considered where meta-analyses included more than ten observations/lines of data; only the meta-analysis on two doses before infection (OR) met this criterion. In the sensitivity analysis for meta-analysis subgroups with more than three studies, the Hartung-Knapp-Sidik-Jonkman method for estimating more conservative confidence intervals was examined and reported in Supplementary Table S2a–e [Reference IntHout, Ioannidis and Borm37]. For meta-analysis subgroups with at least three observations, prediction intervals were calculated to provide a plausible range of effect size in a future new study and reported in Supplementary Table S2a–e [Reference IntHout, Ioannidis, Rovers and Goeman38].
Certainty of evidence
Grading of Recommendations, Assessment, Development and Evaluation (GRADE) criteria were used to indicate the level of confidence in the body of evidence for the primary outcomes of PCC development or resolution [Reference Schünemann, Hill, Guyatt, Akl and Ahmed39]. The GRADE domains risk of bias, inconsistency, imprecision, indirectness, and dose response were evaluated independently by two reviewers to determine a one-to-four-star grade. The evaluation scheme is provided in Supplementary Table S2f. Conflicts were resolved by consensus.
Results
Study selection
In this update, 971 new citations underwent title/abstract screening, of which 210 potentially relevant citations underwent full-text screening, and 47 new studies were included. This SR summarizes 78 studies: 74 peer-reviewed research articles, two preprints, one letter to the editor, and one short communication (Figure 1 and Supplementary Table S3–6). Articles that only assessed PAS (n = 29), did not differentiate between study participants with PAS and PCC (n = 73), or did not report the timing of vaccination (n = 41) were excluded (Supplementary Table S1).
PRISMA flow diagram of articles through the systematic review process, including studies from the previous version and new studies in this update.
Characteristics of the included studies
The included studies addressed the following subtopics: the effect of vaccination administered (1) before (n = 50) or (2) after (n = 2) COVID-19; (3) among previously unvaccinated individuals already experiencing PCC (n = 29); and (4) adverse events post-vaccination among those with PCC (n = 4). All studies were observational (prospective cohort, n = 39; retrospective cohort, n = 14; cross-sectional, n = 20; case-control, n = 5), and had high (n = 55), moderate (n = 21), or low (n = 2) risk of bias (Table 1). Three studies were funded by the pharmaceutical industry [Reference Di Fusco, Sun, Moran, Coetzer, Zamparo and Alvarez40–Reference Morello, Mariani, Mastrantoni, De Rose, Zampino and Munblit42]; all were funded by Pfizer Inc. and examined the impact of the Pfizer-BioNTech COVID-19 vaccine or mRNA vaccines on PCC. For one of these studies, six authors were Pfizer employees [Reference Di Fusco, Sun, Moran, Coetzer, Zamparo and Alvarez40], and in another study, an author had multiple conflicts related to PCC work [Reference Morello, Mariani, Mastrantoni, De Rose, Zampino and Munblit42] (Supplementary Table S3). Most studies were conducted in Europe (n = 40), Asia (n = 15), or North America (n = 14), with a few in South America (n = 4; all Brazil) and Africa (n = 2), and three had a multi-national sampling frame. More than half (n = 55) assessed individuals with mixed severities of COVID-19. Two studies reported on elderly populations, and five studies reported on children. Vaccine products received were mostly BNT162b2 (Pfizer-BioNTech, Comirnaty; n = 50) and mRNA-1273 (Moderna, Spikevax; n = 37). Most studies (n = 63) included individuals with a completed primary series (two vaccine doses for most individuals), while booster doses were examined in 28 studies that included individuals with three doses and four studies that included individuals with four doses.
General characteristics of the 78 included primary research publications on post-COVID-19 condition and vaccination, grouped by research question a
a Each group may sum to >78 because studies can be included in more than one category and more than one question.
(Q1) Risk of developing PCC in those vaccinated before COVID-19
The association between PCC and vaccination before COVID-19 was assessed in 50 studies, including 23 prospective cohorts, 14 retrospective cohorts, four case-control studies, and nine cross-sectional studies (Supplementary Table S3). Studies examined individuals with two doses (n = 43), three doses (n = 21), and four doses (n = 2). The following studies contributed to respective meta-analyses: 24 studies on vaccinated versus unvaccinated general population [Reference Carazo, Skowronski, Laforce, Talbot, Falcone and Laliberté35, Reference Hastie, Lowe, McAuley, Winter, Mills and Black43–Reference Fatima, Ismail, Ejaz, Shah, Fatima, Shahzaib and Jafri65]; three studies on a booster dose versus primary series [Reference Diexer, Klee, Gottschick, Xu, Broda and Purschke64, Reference de Bruijn, van Hoek, Mutubuki, Knoop, Slootweg and Tulen66, Reference Antonelli, Penfold, Canas, Sudre, Rjoob and Murray67]; and four studies on children [Reference Morello, Mariani, Mastrantoni, De Rose, Zampino and Munblit42, Reference Atchison, Whitaker, Donnelly, Chadeau-Hyam, Riley and Darzi68–Reference Li, Nadua, Chong and Yung70] (Figures 2–5). The certainty of evidence was evaluated for each subgroup in the meta-analyses and the GRADE Summary of Findings tables are provided in Tables 2–6. In many studies, individuals classified as unvaccinated before infection may have become vaccinated during the follow-up period and are therefore referred to as those ‘unvaccinated before infection’.
Summary of findings table for the main outcome of PCC development in individuals vaccinated before COVID-19, compared to unvaccinated. Separated by odds ratios/hazard ratios, number of vaccine doses, and type of vaccine. The illustrative example is based on a PCC prevalence of 25% in the unvaccinated population
Note: The illustrative example is based on a PCC prevalence of 25% in the unvaccinated population. For explanations see the GRADE data in Supplementary Table S2a.
Abbreviations: aHR: adjusted HR; aIRR: adjusted incidence rate ratio; aOR: adjusted OR; CI: confidence interval; GRADE: grade of evidence; HR: hazard ratio; OR: odds ratio; pHR: pooled HR; pOR: pooled odds ratio.
*The basis for the assumed risk was a base rate of 25.0% (95%CI 21.5–28.8) reported by unvaccinated Canadians, 13.2% (11.3–15.3) for those with two doses of COVID-19 vaccine before infection and 12.2% (9.2–15.7) for those with three doses before infection up to 31 August 2022 in the Canadian COVID-19 Antibody and Health Survey [7]. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the intervention group and the relative effect of the intervention (and its 95% CI). GRADE: grade of evidence based on a four-star scale of **** (high confidence that the effect estimate is close to the true effect) to * (very low confidence in the effect estimate, the true effect is likely to be substantially different).
Summary of findings table for the main outcomes of PCC development or persistence in individuals vaccinated after COVID-19 or after PCC, compared to unvaccinated. The illustrative example is based on a PCC prevalence of 25% in the unvaccinated population
Note: The illustrative example is based on a PCC prevalence of 25% in the unvaccinated population. For explanations see the GRADE data in Supplementary Table S2b.
Abbreviations: aHR: adjusted HR; aIRR: adjusted incidence rate ratio; aOR: adjusted OR; CI: confidence interval; GRADE: grade of evidence; HR: hazard ratio; OR: odds ratio; pHR: pooled HR; pOR: pooled odds ratio.
*The basis for the assumed risk was a base rate of 25.0% (95%CI 21.5–28.8) reported by unvaccinated Canadians, 13.2% (11.3–15.3) for those with two doses of COVID-19 vaccine before infection and 12.2% (9.2–15.7) for those with three doses before infection up to 31 August 2022 in the Canadian COVID-19 Antibody and Health Survey [7]. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the intervention group and the relative effect of the intervention (and its 95% CI). GRADE: grade of evidence based on a four-star scale of **** (high confidence that the effect estimate is close to the true effect) to * (very low confidence in the effect estimate, the true effect is likely to be substantially different).
Summary of findings table for the main outcome of PCC development in individuals vaccinated after COVID-19 versus vaccinated before COVID-19. The illustrative examples are based on a PCC prevalence of 13.2% for those with two doses before infection and 12.2% for those with three doses before infection
Note: The illustrative example is based on a PCC prevalence of 13.2% for those with two doses before infection and 12.2% for those with three doses before infection. For explanations see the GRADE data in Supplementary Table S2c.
Abbreviations: aHR: adjusted HR; aIRR: adjusted incidence rate ratio; aOR: adjusted OR; CI: confidence interval; GRADE: grade of evidence; HR: hazard ratio; OR: odds ratio; pHR: pooled HR; pOR: pooled odds ratio.
*The basis for the assumed risk was a base rate of 25.0% (95%CI 21.5–28.8) reported by unvaccinated Canadians, 13.2% (11.3–15.3) for those with two doses of COVID-19 vaccine before infection and 12.2% (9.2–15.7) for those with three doses before infection up to 31 August 2022 in the Canadian COVID-19 Antibody and Health Survey [7]. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the intervention group and the relative effect of the intervention (and its 95% CI). GRADE: grade of evidence based on a four-star scale of **** (high confidence that the effect estimate is close to the true effect) to * (very low confidence in the effect estimate, the true effect is likely to be substantially different).
Summary of findings table for the main outcome of PCC development, in individuals who received a booster dose versus only primary series. The illustrative example is based on a PCC prevalence of 13.2% in the primary series population
Note: The illustrative example is based on a PCC prevalence of 13.2% for those with two doses before infection. For explanations see the GRADE data in Supplementary Table S2d.
Abbreviations: CI: confidence interval; GRADE: grade of evidence; pOR: pooled odds ratio.
*The basis for the assumed risk was a base rate of 13.2% (95%CI 11.3–15.3%) for those with two doses of COVID-19 vaccine up to 31 August 2022 in the Canadian COVID-19 Antibody and Health Survey [7]. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the intervention group and the relative effect of the intervention (and its 95% CI). GRADE: grade of evidence based on a four-star scale of **** high confidence to * very low confidence in the evidence.
Summary of findings table for the main outcome of PCC development in children up to 18 years old. Separated by number of vaccine doses. The illustrative example is based on a PCC prevalence of 5.8% in the unvaccinated children population
Note: The illustrative example is based on a PCC prevalence of 5.8% in the unvaccinated children population. For explanations see the GRADE data in Supplementary Table S2e.
Abbreviations: CI: confidence interval; GRADE: grade of evidence; pOR: pooled odds ratio.
*The basis for the assumed risk was a base rate of 5.8% reported by unvaccinated children under 18 years old in eight countries (Argentina, Canada, Costa Rica, Italy, Paraguay, Singapore, Spain, and the United States) [Reference Funk, Kuppermann, Florin, Tancredi, Xie and Kim103]. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the intervention group and the relative effect of the intervention (and its 95% CI). GRADE: grade of evidence based on a four-star scale of **** high confidence to * very low confidence in the evidence.
One dose prior to COVID-19 may reduce the odds of developing PCC compared to those unvaccinated before infection, across eight studies (pOR 0.62, 95% CI 0.41–0.92) with high heterogeneity (I2 = 96.9%) (Figure 2). When stratified by ROB, moderate ROB studies (n = 5) showed a pooled protective effect, while low (n = 1) and high ROB studies (n = 2) showed no association.
Meta-analysis of the effect of vaccination prior to COVID-19 compared to unvaccinated on the odds of developing PCC, stratified by number of doses.
Two doses prior to COVID-19 likely reduced the odds of developing PCC compared to those unvaccinated before infection (pOR 0.69, 95% CI 0.64–0.74, I2 = 35.16%, 13 studies) with a 95% prediction interval of 0.57–0.83 (Figure 2). There was no indication of small study effects in the two-dose subgroup (Egger test p = 0.36, Begg test p = 0.39; funnel plot was symmetrical). When studies were stratified by low ROB (n = 1), moderate (n = 7), or high (n = 5), protective associations were still found and there was no difference across subgroups. Two studies that examined those with one or two doses aligned with the two-dose analysis (pOR 0.32, 95% CI 0.14–0.73, I2 = 81.87%). One study provided effect estimates of two or more doses separated by vaccine product (CoronaVac and Pfizer-BioNTech), and when these estimates were pooled together in meta-analysis, there was a protective association (pOR 0.28, 95% CI 0.14–0.57) [Reference Wong, Huang, Wong, Wong, Yip and Chan57].
Across two studies reporting hazard ratios, one dose prior to COVID-19 may have little to no effect on the average hazard of developing PCC from 6 months to 1 year post-infection (pHR 0.72, 95% CI 0.41–1.28) with high heterogeneity (I2 = 96.8%) (Figure 3). This heterogeneity may be explained by Taquet et al. having high ROB [Reference Taquet, Dercon and Harrison45] and only including those infected during Alpha or Delta waves, in contrast to Catala et al., which had low ROB and included those infected during Alpha, Delta, and Omicron [Reference Català, Mercadé-Besora, Kolde, Trinh, Roel and Burn60].
Meta-analysis of the hazard ratios for developing PCC in those vaccinated prior to COVID-19 compared to unvaccinated, stratified by number of doses.
Across four studies reporting hazard ratios, two doses prior to COVID-19 may have little to no effect on the average hazard of developing PCC from 6 months to 300 days post-infection (pHR 0.82, 95% CI 0.67–1.00) with high heterogeneity (I2 = 96.7%) and a wide 95% prediction interval (0.39–1.73), suggesting the results are imprecise (Figure 3). The two studies reporting a reduction in the hazard of PCC were at moderate ROB and the two reporting no association were at high ROB.
Three or more doses prior to COVID-19 may have little to no effect on the odds of developing PCC compared to those unvaccinated before infection (pOR 0.82, 95% CI 0.62–1.08, I2 = 60.5%, five studies) (Figure 2). The two moderate ROB studies (pOR 0.77, 95% CI 0.59–1.00) and three high ROB studies (pOR 0.80, 95% CI 0.44–1.44) showed no association. The finding of no association may be explained by individuals becoming vaccinated post-infection, re-infections, and different variants, which is detailed in the discussion. The multivariate analysis in Marra et al. demonstrated a strong protective effect with four doses before COVID-19 on the odds of developing PCC compared to unvaccinated before infection (aOR 0.05, 95% CI 0.01–0.19) [Reference Marra, Sampaio, Ozahata, Lopes, Brito and Bragatte54].
Among children (≤18) with one or more doses prior to COVID-19, the overall meta-analytic estimate indicated no association (pOR 0.79, 95% CI 0.56–1.11, I2 = 37.2%, four studies) with the odds of developing PCC compared to unvaccinated before infection, and no difference between one versus two or three doses (p = 0.80) (Figure 4). In the two or three doses subgroup, Morello et al. reported an OR > 1 for 12–18 year olds, which was the main source of heterogeneity and pulled the pooled estimate towards the null [Reference Morello, Mariani, Mastrantoni, De Rose, Zampino and Munblit42]. For those who received one to three doses in Morello et al., the estimates were similar to those who received two or three doses for both 5–11 year olds and 12–18 year olds [Reference Morello, Mariani, Mastrantoni, De Rose, Zampino and Munblit42]. All studies were evaluated as high ROB.
Meta-analysis of the effect of vaccination prior to COVID-19 compared to unvaccinated on the odds of developing PCC in children up to 18 years old, stratified by number of doses.
A booster dose before Delta or Omicron infection may reduce the odds of developing PCC compared to those with only a primary series (pOR 0.85, 95% CI 0.74–0.98, I2 = 16.85%, three studies) (Figure 5). All three studies were high ROB and did not specify the length of time between vaccination and infection.
Meta-analysis of the effect of booster vaccination prior to COVID-19 compared to only a primary series on the odds of developing PCC.
Impact of vaccination before infection on individual PCC symptoms
Twenty-three studies reported on differences in individual PCC symptoms in those vaccinated before COVID-19. Among studies reporting an effect estimate for vaccinated versus unvaccinated, vaccination was associated with a protective effect for the most common PCC symptoms, including fatigue in 5/7 studies, anxiety/depression in 3/5 studies, dyspnea in 4/7 studies, pain in 3/6 studies, insomnia in 1/2 studies, and cognitive impairment in 3/8 studies (Supplementary Table S3). Only one study reported that vaccination was associated with a higher risk of individual symptoms (concentration and memory impairment, voice disorder) compared to unvaccinated; however, when excluding those vaccinated >3 months before COVID-19, there was no association with concentration/memory impairment [Reference Mizrahi, Sudry, Flaks-Manov, Yehezkelli, Kalkstein and Akiva71]. Seven studies reported on the prevalence of individual PCC symptoms in vaccinated versus unvaccinated without an effect estimate [Reference van der Maaden, Mutubuki, de Bruijn, Leung, Knoop and Slootweg30, Reference Di Fusco, Sun, Moran, Coetzer, Zamparo and Alvarez40, Reference Mariani, Morello, Traini, La Rocca, De Rose, Valentini and Buonsenso41, Reference Nascimento, Costa, Ruiz, Ledo, Fernandes and Cardoso46, Reference Kuodi, Gorelik, Zayyad, Wertheim, Beiruti Wiegler and Abu Jabal61, Reference Babicki, Kapusta, Pieniawska-Śmiech, Kałuzińska-Kołat, Kołat and Mastalerz-Migas63, Reference Fernández-de-las-Peñas, Ortega-Santiago, Fuensalida-Novo, Martín-Guerrero, Pellicer-Valero and Torres-Macho72, Reference Mitrović-Ajtić, Stanisavljević, Miljatović, Dragojević, Živković, Šabanović and Čokić73]; one of these examined children (≤18) and found no significant differences in anosmia or dysgeusia [Reference Mariani, Morello, Traini, La Rocca, De Rose, Valentini and Buonsenso41]. Two studies reported on individuals with a booster versus primary series before Omicron infection. One found that those with a booster had reduced incidence rates of specific PCC symptoms (physical symptoms, depression, anxiety, fatigue, and cognitive complaints) at 4 months post-infection [Reference Spiliopoulos, Sørensen, Bager, Nielsen, Vinsløv Hansen and Koch74], while the other found no significant difference in the prevalence of individual symptoms (fatigue, dyspnoea, difficulties with a busy environment, memory problems, or brain fog) at 3 months post-infection [Reference de Bruijn, van Hoek, Mutubuki, Knoop, Slootweg and Tulen66].
There was no association with one to three doses before infection and the number of PCC symptoms compared to unvaccinated, with a rate ratio of 1.27 (95% CI 0.82–1.94) adjusted for variant among other variables; this represents the multiplicative effect of vaccination on the number of symptoms [Reference Kahlert, Strahm, Güsewell, Cusini, Brucher and Goppel75]. Among those infected during Omicron dominance, 3+ doses before infection were associated with lower odds of 3+ PCC symptoms at 6 months post-infection compared to unvaccinated (aOR 0.36, 95% CI 0.15–0.87, p = 0.019), while no association was found with two doses [Reference Di Fusco, Sun, Moran, Coetzer, Zamparo and Alvarez40].
Differences between vaccine products
Three studies addressed differences between vaccine products and showed that all vaccine products reduced the risk of developing PCC. One showed that mRNA vaccines resulted in a decreased risk of PCC compared to Ad26.COV2.S (Johnson & Johnson) (aHR 0.89, 95% CI 0.81–0.97) [Reference Al-Aly, Bowe and Xie50]. Similarly, another study found that BNT162b2 resulted in a reduced hazard of PCC compared to ChAdOx1 (AstraZeneca) (aHR 0.84, 95% CI 0.75–0.94) and some individual symptoms [Reference Català, Mercadé-Besora, Kolde, Trinh, Roel and Burn60]. A third study found no significant difference between mRNA, viral vector, or inactivated vaccines for protection against developing ‘neuropsychiatric PCC’ (various neurological and mental health symptoms) [Reference Elmazny, Magdy, Hussein, Elsebaie, Ali and Abdel Fattah56].
Timing of vaccination before infection
One study found that vaccination (one to three doses) within 6 months before Omicron infection was associated with a lower odds of PCC compared to vaccination more than 6 months before infection [Reference Ballouz, Menges, Kaufmann, Amati, Frei and von Wyl52]. Another study found that vaccination (three doses) within 3 months before Omicron infection was associated with higher odds of PCC compared to vaccination 4–6 months before, which may be explained by a limited number of PCC cases [Reference Antonelli, Penfold, Canas, Sudre, Rjoob and Murray67]. A third study found that vaccination (one to three doses) before infection with mostly Omicron was not associated with PCC compared to unvaccinated, regardless of the timing of last dose (>6 months, 3–6 months, or < 3 months before infection) [Reference Hernández-Aceituno, García-Hernández and Larumbe-Zabala76]. There was no association between timing of vaccination (up to three doses) before Delta infection and the odds of developing PCC in two studies [Reference Ballouz, Menges, Kaufmann, Amati, Frei and von Wyl52, Reference Antonelli, Penfold, Canas, Sudre, Rjoob and Murray67].
Differences by age and sex
One study suggested older adults (≥60 years), who received a third dose at 4–6 months prior to Omicron infection had significantly lower odds of PCC compared to those who received a third dose within 3 months, and this association was not found in the 18–59 age group [Reference Antonelli, Penfold, Canas, Sudre, Rjoob and Murray67]. Another study found no association between vaccination with one or two doses and the hazard of developing PCC in either the <60 age group or ≥ 60 age group [Reference Taquet, Dercon and Harrison45]. A third study found no association between vaccination with one to three doses and the odds of PCC in either younger (5–11 years old) or older (12–18 years old) children [Reference Morello, Mariani, Mastrantoni, De Rose, Zampino and Munblit42]. None of the studies examined differences by sex regarding the association between vaccination prior to COVID-19 and risk of developing PCC.
(Q2) Risk of developing PCC in those vaccinated after COVID-19
Two studies assessed the association between PCC and vaccination post-infection (up to 12 weeks), including one retrospective cohort study with moderate ROB [Reference Simon, Luginbuhl and Parker44] and one cross-sectional study with high ROB [Reference Diez-Cirarda, Yus, Gomez-Ruiz, Polidura, Gil-Martinez and Delgado-Alonso77] (Supplementary Table S4). The GRADE Summary of Findings is provided in Table 3.
In the retrospective cohort study, the protective effect against PCC development was stronger when one dose was given earlier post-infection (aOR 0–4 weeks post-infection 0.38, 95% CI 0.35–0.41; aOR 4–8 weeks post-infection 0.54, 95% CI 0.51–0.57; aOR 8–12 weeks post-infection 0.75, 95% CI 0.71–0.78) compared to unvaccinated (Table 3) [Reference Simon, Luginbuhl and Parker44].
The cross-sectional study found no significant difference in cognition and neuroimaging results (grey matter volume, white matter hyperintensities, functional connectivity) between those with one or two doses versus unvaccinated; however, vaccinated individuals performed better on Visual Object and Space Perception Battery discrimination [Reference Diez-Cirarda, Yus, Gomez-Ruiz, Polidura, Gil-Martinez and Delgado-Alonso77].
(Q3) Changes in PCC following vaccination among individuals with established PCC
Nineteen studies examined the effect of COVID-19 vaccination on individuals with established PCC, and ten studies examined individuals where it was unclear if vaccination occurred after developing PCC or within 12 weeks of infection. These 29 studies included 18 prospective cohorts, eight cross-sectional, two retrospective cohorts, and one case-control (Supplementary Table S5).
Seven studies contributed to the meta-analysis on the odds of PCC persistence among vaccinated individuals compared to unvaccinated, stratified by two subgroups: vaccinated after PCC and vaccinated anytime after COVID-19 infection (Figure 6). The certainty of evidence was evaluated for each subgroup and the GRADE Summary of Findings is provided in Table 3. Among those vaccinated after PCC, vaccination may reduce the odds of PCC persistence compared to unvaccinated (pOR 0.73, 95% CI 0.57–0.92, I2 = 15.5%, three studies). When stratified by ROB, a pooled protective effect was found for the high ROB studies (n = 2), while there was no association for the moderate ROB study. Among those vaccinated anytime after infection, there may be little to no effect on the odds of PCC development or persistence compared to unvaccinated (pOR 0.65, 95% CI 0.32–1.31, I2 = 67.8%, four studies). These four studies were high ROB.
Meta-analysis of the effect of vaccination after COVID-19 compared to unvaccinated on the odds of developing PCC or persistent PCC, stratified by vaccination after established PCC and vaccination anytime after COVID-19 infection.
Findings were inconsistent regarding symptom improvement, worsening, or no change for vaccinated versus unvaccinated individuals with established PCC across three studies. One study found a significantly higher proportion of vaccinated had improved symptoms (23.3% vs. 15.4%, p = 0.035) [Reference Arnold, Milne, Samms, Stadon, Maskell and Hamilton78], while two studies found no significant difference by vaccination in PCC symptom improvement [Reference Peghin, De Martino, Palese, Gerussi, Bontempo and Graziano79, Reference Salmon, Slama, Linard, Dumesges, Lebaut and Hakim80].
Across nine studies that compared PCC symptoms pre- and post-vaccination in the same individuals, symptoms tended to improve or remain the same following vaccination, rather than worsen. Across four studies, one dose resulted in symptom improvement [Reference Krishna, Lim, Metaxaki, Jackson, Mactavous and BioResource81, Reference Strain, Sherwood, Banerjee, Van der Togt, Hishmeh and Rossman82], a reduction in the proportion of those with more than one PCC symptom [Reference Fong, Goh, Torres-Ruesta, Chang, Chan and Neo83], and a slightly reduced odds of on-going PCC following both the first (OR 0.87, 95% CI 0.81–0.93) and second doses (OR 0.91, 95% CI 0.86–0.97) [Reference Ayoubkhani, Bermingham, Pouwels, Glickman, Nafilyan and Zaccardi84]. Three studies examined one to three doses post-infection; two studies reported a greater proportion of PCC cases had symptom improvement post-vaccination compared to worsening [Reference Nakagawara, Morita, Namkoong, Terai, Chubachi and Asakura85, Reference Malesevic, Sievi, Schmidt, Vallelian, Jelcic, Kohler and Clarenbach86], while a third reported the opposite (12.7% improved vs. 20% worsened) [Reference Romero-Rodríguez, Perula-de-Torres, González-Lama, Castro-Jiménez, Jiménez-García and Priego-Pérez87]. The other studies on one to two doses post-infection [Reference Gerhards, Kittel, Ast, Bugert, Froelich and Hetjens88] and two or more doses [Reference da Silva, de Araújo, Santos, de Souza, de Araújo and Cruz89] found no significant change in PCC symptoms post-vaccination.
Nine studies reported on individual PCC symptoms: one pre-post study [Reference Ayoubkhani, Bermingham, Pouwels, Glickman, Nafilyan and Zaccardi84], five comparing those vaccinated after PCC versus unvaccinated [Reference Arnold, Milne, Samms, Stadon, Maskell and Hamilton78, Reference Peghin, De Martino, Palese, Gerussi, Bontempo and Graziano79, Reference Wisnivesky, Govindarajulu, Bagiella, Goswami, Kale and Campbell90–Reference Nehme, Braillard, Salamun, Jacquerioz, Courvoisier, Spechbach and Guessous92], and three comparing those vaccinated anytime after infection versus unvaccinated [Reference Babicki, Kapusta, Pieniawska-Śmiech, Kałuzińska-Kołat, Kołat and Mastalerz-Migas63, Reference Yin, Guo, Xiao, Chen, Long and Wang93, Reference Kim, Bae, Chang and Kim94], with follow-up ranging from 6 to 30 months post-infection. One study did not have extractable data [Reference Arnold, Milne, Samms, Stadon, Maskell and Hamilton78] and three studies did not find a significant difference for any symptoms [Reference Wisnivesky, Govindarajulu, Bagiella, Goswami, Kale and Campbell90, Reference Augustin, Stecher, Wüstenberg, Di Cristanziano, de Silva and Picard91, Reference Kim, Bae, Chang and Kim94]. Four studies found significant differences in PCC symptoms between vaccinated versus unvaccinated. Fatigue was less prevalent for those with three doses [Reference Yin, Guo, Xiao, Chen, Long and Wang93], as well as headache and arthralgia for those with two doses [Reference Babicki, Kapusta, Pieniawska-Śmiech, Kałuzińska-Kołat, Kołat and Mastalerz-Migas63]. Worsening ocular symptoms were less prevalent [Reference Peghin, De Martino, Palese, Gerussi, Bontempo and Graziano79] and dyspnoea and change in taste were lower [Reference Nehme, Braillard, Salamun, Jacquerioz, Courvoisier, Spechbach and Guessous92] in those with one or two doses. More vaccinated individuals (one to two doses) reported persistent hair loss in one study [Reference Peghin, De Martino, Palese, Gerussi, Bontempo and Graziano79]. The pre–post-study found individuals with PCC had significantly lower odds of fatigue after two doses and loss of smell after one dose but not two [Reference Ayoubkhani, Bermingham, Pouwels, Glickman, Nafilyan and Zaccardi84].
Vaccination anytime after COVID-19 versus vaccination before COVID-19
Three studies compared individuals vaccinated anytime after infection versus individuals vaccinated before infection. The GRADE Summary of Findings is provided in Table 4. Two studies compared those who received a primary series after infection versus before infection: one found no significant difference in the odds of developing PCC [Reference Bostanci, Gazi, Tosun, Suer, Unal Evren, Evren and Sanlidag95], and the other found no difference in the rate of PCC at 6 months follow-up (aIRR 0.91, 95% CI 0.75–1.10) [Reference Jassat, Mudara, Vika, Welch, Arendse and Dryden51]. A third study found that a significantly higher proportion of individuals vaccinated (two to three doses) after infection reported PCC symptoms at 3 to 6 months compared to those vaccinated (three doses) before infection (74.5% vs. 51.9%, p < 0.001), but significantly fewer reported symptoms at more than 6 months (13.7% vs. 38%, p < 0.001) [Reference Adly, Saleh, Garout, Abdulkhaliq, Khafagy and Saati96].
Differences between vaccine products
Three studies found no significant differences between mRNA vaccines (BNT162b2 or mRNA-1273) and adenoviral vector vaccines (ChAdOx1 or Ad26.COV2.S) [Reference Arnold, Milne, Samms, Stadon, Maskell and Hamilton78, Reference Peghin, De Martino, Palese, Gerussi, Bontempo and Graziano79, Reference Ayoubkhani, Bermingham, Pouwels, Glickman, Nafilyan and Zaccardi84]. However, a fourth study suggested those who received mRNA-1273 after PCC experienced improvement in certain symptoms significantly more than those who received ChAdOx1, including fatigue, brain fog, myalgia, gastro-intestinal symptoms, and autonomic dysfunction [Reference Strain, Sherwood, Banerjee, Van der Togt, Hishmeh and Rossman82].
Differences by age and sex
One study found that only individuals ≥60 years who received two doses post-infection had significantly lower odds of persistent PCC compared to unvaccinated, and there was no association with sex [Reference Nehme, Braillard, Salamun, Jacquerioz, Courvoisier, Spechbach and Guessous92]. Worsening PCC symptoms after one to three doses was significantly higher among individuals aged 14–40 compared to older individuals (aged 41–76) and among males compared to females [Reference Romero-Rodríguez, Perula-de-Torres, González-Lama, Castro-Jiménez, Jiménez-García and Priego-Pérez87].
(Q4) Safety and risk of adverse events following COVID-19 vaccination among individuals with PCC
Four studies reported on the safety or adverse events among those with PCC following COVID-19 vaccination, including three cross-sectional and one prospective cohort, all of which were high ROB (Table S6). Only one study included a vaccinated comparator group with no previous COVID-19 and found no significant difference in the number or type of side effects following one dose (BNT162b2) among those with PCC (n = 30) compared to controls [Reference Raw, Kelly, Rees, Wroe and Chadwick97]. Previous COVID-19 infection, but not PCC, was associated with an increased risk of adverse events post-vaccination. Another study found that only 5.7% (n = 26/455) of participants with PCC reported adverse events after one dose (various brands) [Reference Tran, Perrodeau, Saldanha, Pane and Ravaud98]. However, the control group was unvaccinated individuals with PCC; therefore, this study does not show if the effects of vaccination were like those without PCC. In a survey of 67 healthcare workers with PCC, 72% reported immediate but self-limiting side effects at 2 weeks after one dose (BNT162b2) [Reference Gaber, Ashish, Unsworth and Martindale99]. A fourth study found that the most common adverse effects after one to three doses (various brands) in those with PCC were pain at the injection site (90.8%), tiredness or fatigue (76.7%), and muscle pain (68.3%) [Reference Romero-Rodríguez, Perula-de-Torres, González-Lama, Castro-Jiménez, Jiménez-García and Priego-Pérez87]. A significantly higher proportion of those aged 14–40 reported dizziness post-vaccination (p = 0.017); otherwise, there were no significant differences in adverse effects by age or gender [Reference Romero-Rodríguez, Perula-de-Torres, González-Lama, Castro-Jiménez, Jiménez-García and Priego-Pérez87].
Discussion
The results of this updated living SR are aligned with the previous version and other evidence syntheses, which suggest that vaccination before COVID-19 provides protection against the risk of developing PCC [Reference Byambasuren, Stehlik, Clark, Alcorn and Glasziou13–Reference Marra, Kobayashi, Callado, Pardo, Gutfreund and Hsieh17]. There was moderate confidence that two vaccine doses before COVID-19 decreased the odds of developing PCC by 31%, compared to unvaccinated. Vaccination within 12 weeks after COVID-19 may offer additional protection against developing PCC compared to unvaccinated, but the evidence was very uncertain from only one study. There was low confidence that vaccination after PCC may reduce the odds of PCC persistence. Preliminary evidence suggested that a booster dose before infection may offer additional protection against PCC compared to only primary series [Reference Diexer, Klee, Gottschick, Xu, Broda and Purschke64, Reference de Bruijn, van Hoek, Mutubuki, Knoop, Slootweg and Tulen66, Reference Antonelli, Penfold, Canas, Sudre, Rjoob and Murray67]. Among children up to 18 years old, vaccination may have little to no effect on the odds of developing PCC [Reference Morello, Mariani, Mastrantoni, De Rose, Zampino and Munblit42, Reference Atchison, Whitaker, Donnelly, Chadeau-Hyam, Riley and Darzi68–Reference Li, Nadua, Chong and Yung70].
More recent studies examining the effect of three or more vaccine doses before infection on PCC frequently reported no association compared to unvaccinated [Reference Ballouz, Menges, Kaufmann, Amati, Frei and von Wyl52, Reference Marra, Sampaio, Ozahata, Lopes, Brito and Bragatte54, Reference Angarita-Fonseca, Torres-Castro, Benavides-Cordoba, Chero, Morales-Satán and Hernández-López59, Reference Kuodi, Gorelik, Zayyad, Wertheim, Beiruti Wiegler and Abu Jabal61, Reference Diexer, Klee, Gottschick, Xu, Broda and Purschke64]. There are several explanations for the association with vaccination becoming less clear than earlier in the pandemic. Population immunity has become more complex, with most people having hybrid immunity. Furthermore, the risk of PCC has changed over time as different variants have become dominant [Reference Thaweethai, Jolley, Karlson, Levitan, Levy and McComsey62, Reference Diexer, Klee, Gottschick, Xu, Broda and Purschke64, Reference Reme, Gjesvik and Magnusson100], and there are likely some differences in virulence between variants. The potential impact of variants was seen in a couple of studies where a significant association between vaccination and PCC was found in univariate analysis; however, after controlling for variant in multivariate analysis, the association became non-significant [Reference Marra, Sampaio, Ozahata, Lopes, Brito and Bragatte54, Reference Reme, Gjesvik and Magnusson100]. Finally, individuals becoming vaccinated during the follow-up period may also impact the development or persistence of PCC but this was not accounted for in analyses; this would bias the estimated association between vaccination before infection and PCC towards the null. Overall, it has become increasingly more complex to measure the impact of vaccination on PCC.
Vaccination for those with PCC was safe across four studies, and there is low confidence that vaccination may reduce odds of PCC persistence. Although most studies suggested there was an improvement or resolution of PCC following vaccination, some suggested that PCC worsened or remained unchanged. Some of this heterogeneity in results may be due to recall bias in the self-reported PCC assessments. Improvement in PCC symptoms post-vaccination may also be conflated with natural recovery over time. Some studies did not specify whether individuals were vaccinated before PCC (<12 weeks post-infection) or after PCC (>12 weeks), so it was unclear whether the outcome was PCC development or persistence. Only one study reported results on the association between vaccination and PCC stratified by re-infection status [Reference Reme, Gjesvik and Magnusson100]. Clear reporting on the timing of vaccination and re-infections after the infection that resulted in PCC would be useful in future studies.
Only a few studies examined differences in the association between vaccination and PCC by sociodemographic variables. However, many studies controlled for potential confounding variables, such as sex, age, and severity of initial COVID-19, which have been reported as risk factors for PCC [Reference Hastie, Lowe, McAuley, Winter, Mills and Black43, Reference Nascimento, Costa, Ruiz, Ledo, Fernandes and Cardoso46, Reference Emecen, Keskin, Turunc, Suner, Siyve and Basoglu Sensoy48]. Any differences by sociodemographic variables would be important to consider when developing recommendations for treatment and equitable resource allocation.
Across studies, there were various methodological differences in how PCC was defined and measured. Prospective studies often assessed PCC using self-reported surveys, while retrospective studies examined ICD-10 codes in health records; both of which could have resulted in the misclassification of PCC due to sequelae that are related to other conditions. Using a consistent PCC definition and developing validated PCC diagnostic tools in future research will help improve our understanding of this condition.
Limitations to this SR process include using the NOS tool for risk of bias assessments, which has not been validated, and a modified version of the tool to assess cross-sectional studies [Reference Ribeiro, Beserra, Silva, Lima, Rocha and Coelho33]. Furthermore, even though updated searches were conducted, the findings of this SR may change with emerging evidence on this evolving topic.
Conclusion
This updated SR indicates that there is moderate confidence that two vaccine doses before COVID-19 reduces the odds of developing PCC. For those with PCC, getting a COVID-19 vaccine appears to be safe, and there is low confidence that vaccination may reduce the odds of PCC persistence. Understanding the impact of vaccination on PCC, in the context of booster doses and re-infections, is important for informing public health recommendations.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0950268825000378.
Data availability statement
The data that support the findings of this study are openly available in the Supplementary Materials (Tables S1–S6).
Acknowledgements
The authors would like to acknowledge Sydney Jennings, the first author on the previous version of this living systematic review [Reference Jennings, Corrin and Waddell21].
Author contribution
Data curation: T.C., L.A.W., T.N., M.S.; Investigation: T.C., L.A.W., T.N., M.S.; Methodology: T.C., L.A.W., M.S.; Validation: T.C., L.A.W., T.N., M.S.; Writing – review & editing: T.C., L.A.W., T.N., M.S.; Conceptualization: L.A.W.; Project administration: L.A.W., M.S.; Supervision: L.A.W.; Writing – original draft: L.A.W., M.S.; Formal analysis: M.S.
Competing interests
The authors declare none.
Open access
