Kenny et al. BMC Infectious Diseases (2023) 23:804 BMC Infectious Diseases
RESEARCH Open Access
Impact of vaccination and variants
of concern on long COVID clinical phenotypes
Grace Kenny1,2, Kathleen McCann2, Conor O’Brien3, Cathal O’Broin1,2, Willard Tinago1, Obada Yousif4,
Tessa O’Gorman1,5, Aoife G. Cotter1,3,5, John S. Lambert1,3,5, Eoin R. Feeney1,2, Eoghan de Barra6,7,
Corinna Sadlier8, Alan Landay9, Peter Doran10, Stefano Savinelli1,2, Patrick W. G. Mallon1,2 and The All Ireland
Infectious Diseases Cohort Study
Background Defining patterns of symptoms in long COVID is necessary to advance therapies for this heterogeneous
condition. Here we aimed to describe clusters of symptoms in individuals with long COVID and explore the impact
of the emergence of variants of concern (VOCs) and vaccination on these clusters.
Methods In a prospective, multi centre cohort study, individuals with symptoms persisting > 4 weeks from acute
COVID-19 were divided into two groups based on timing of acute infection; pre-Alpha VOC, denoted wild type (WT)
group and post-Alpha VOC (incorporating alpha and delta dominant periods) denoted VOC group. We used multiple
correspondence analysis (MCA) and hierarchical clustering in the WT and VOC groups to identify symptom clusters.
We then used logistic regression to explore factors associated with individual symptoms.
Results A total of 417 individuals were included in the analysis, 268 in WT and 149 in VOC groups respectively.
In both groups MCA identified three similar clusters; a musculoskeletal (MSK) cluster characterised by joint pain
and myalgia, a cardiorespiratory cluster and a less symptomatic cluster. Differences in characteristic symptoms were
only seen in the cardiorespiratory cluster where a decrease in the frequency of palpitations (10% vs 34% p = 0.008)
and an increase in cough (63% vs 17% p < 0.001) in the VOC compared to WT groups was observed. Analysis of the fre-
quency of individual symptoms showed significantly lower frequency of both chest pain (25% vs 39% p = 0.004)
and palpitations (12% vs 32% p < 0.001) in the VOC group compared to the WT group. In adjusted analysis being
in the VOC group was significantly associated with a lower odds of both chest pain and palpitations, but vaccination
was not associated with these symptoms.
Conclusion This study suggests changes in long COVID phenotype in individuals infected later in the pan-
demic, with less palpitations and chest pain reported. Adjusted analyses suggest that these effects are mediated
through introduction of variants rather than an effect from vaccination.
Keywords SARS-CoV-2 variants, Long COVID, Post-Acute Sequelae of SARS-CoV-2 infection
Full list of author information is available at the end of the article
© The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecom-
mons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Kenny et al. BMC Infectious Diseases (2023) 23:804 Page 2 of 8
Introduction World Health Organisation (WHO) scale , Medical
The management of long COVID remains one of the most Research Council (MRC) dyspnea scale  and Short
challenging aspects of the COVID-19 pandemic . Long Form-36 (SF-36)  score were assessed at each clinic
COVID is used to describe a number of symptoms that visit using a standardised proforma. Participants self-
persist beyond the acute viral illness, but criteria to dif- completed SF-36 questionnaires, while other compo-
ferentiate these into distinct clinical categories are lack- nents were assessed by structured interview and review
ing. While it is increasingly acknowledged long COVID of medical records. The MRC dyspnea scale is a validated
is not one distinct entity, frequently studies exploring the 5 point scale that assesses functional disability due to
pathogenesis or long COVID attempt to overcome this dyspnea, ranging from no disability (point 1) to dyspnea
heterogeneity by correlating individual symptoms with limiting basic activities of daily living (point 5). The SF-36
specific immune abnormalities , running the risk of survey is a generic measure of health status and quality
multiple comparisons and potentially erroneous conclu- of life, evaluating an individual’s perception of their per-
sions. Identification of consistent patterns of symptoms formance in 8 domains, and has shown to be a reliable,
in long COVID that could be used to guide therapeutic valid and sensitive measure of health status in a variety
and translational studies is urgently needed. We and oth- of settings. Additional investigations were performed if
ers have used cluster analysis to identify clinical phe- determined to be indicated by the assessing clinician. For
notypes of long COVID [3–5], but a variety of factors analysis we included only symptoms present in at least
have emerged that may impact these patterns. Specifi- 10% of individuals, to maintain an appropriate number of
cally, vaccination against SARS-CoV-2 has been associ- variables for the sample size .
ated with both a lower risk of developing long COVID As sequencing data was not available, individuals were
 and an improvement in long COVID symptoms in allocated to two groups based on timing of acute COVID-
observational studies . In addition, variants of con- 19. The first period (“wild-type” (WT) period) included
cern (VOCs) have emerged which differ in virulence  those with onset of symptoms prior to 26th December
and transmission dynamics  to wild-type SARS-CoV-2 2020 when the alpha variant became dominant in Ire-
but the impact of VOCs on post-acute symptoms has not land, and the VOC period individuals with symptoms
been determined. Here, we aimed to identify phenotypes after this date. The delta variant became dominant in
of long COVID in individuals infected prior to and sub- April 2021 and the Omicron variant in December 2021 in
sequent to the emergence of VOCs, explore changes in Ireland . For individuals who reported more than one
long COVID phenotype across these time periods, and infection prior to their review, we allocated group based
the association of factors including VOCs and vaccina- on the infection that the participant attributed their long
tion with these changes. COVID symptoms to.
Methods Statistical analysis
The All Ireland Infectious Diseases (AIID) cohort is a Categorical variables were summarised using number
prospective, multicentre, observational cohort study and percentage and continuous variables with median
recruiting individuals presenting with issues pertaining and interquartile range (IQR). Cluster analysis was per-
to infectious diseases in participating hospitals in Ire- formed as described in detail previously . Briefly, we
land, described in detail elsewhere . The AIID cohort used multiple correspondence analysis (MCA) to remove
study was approved in line with national and European dimensionality from the dataset. MCA is a principal
regulations on health research by the St Vincent’s Hos- component analysis method that transforms categorical
pital group Research Ethics committee and the National data into coordinates in multidimensional space (using
Research Ethics Committee for COVID-19 in Ireland. χ2 distance between coordinates so similar individuals
All study procedures adhered to required guidelines. lie closer together). The smallest number of dimensions
Participants provide written informed consent for col- that account for the largest total explained variance are
lection of data on demographics, clinical characteristics retained for further analysis, resulting in a reduction in
and investigations undertaken as part of routine care. For the number of variables needed to summarize the data.
this study, individuals were included if they were attend- Then, agglomerative hierarchical clustering was per-
ing for assessment of long COVID, had PCR-confirmed formed on the results of the MCA, using squared Euclid-
COVID-19, and were still symptomatic at least 4 weeks ean distance and Ward’s minimum variance linkage. The
post onset of acute symptoms. number of clusters was selected at the partition where
The details of the long COVID assessments have there was the greatest within cluster loss of inertia. We
been described in detail elsewhere . Briefly, 19 symp- used heatmaps to visualise the prevalence of individual
toms, maximum acute disease severity as graded by the symptoms within each cluster and compared continuous
K enny et al. BMC Infectious Diseases (2023) 23:804 Page 3 of 8
variables and categorical variables with Kruskal–Wallis periods. Two individuals infected within the WT period
and chi square test respectively. As this was an explora- had a second infection in the VOC period prior to review,
tory analysis, we did not correct for multiple compari- and one individual infected in the VOC period reported a
sons . prior infection in the WT period that did not lead to long
Univariate and multivariable logistic regression models COVID symptoms.
were constructed to explore the association of infection Participant demographics are shown in Table 1. Basic
period and vaccination status on individual symptoms. demographics were similar across both periods, median
In multivariable models, we adjusted for age, sex, ethnic- (IQR) age was 45 (35–55) years in the WT period and 47
ity, WHO acute disease severity and time from symptom (34–57) years in the VOC period (p = 0.2). 73% of indi-
onset. All analysis was carried out using R version 4.2.1. viduals were female in both periods (n = 196 and n = 109
in WT and VOC periods respectively, p = 0.9), and the
Results majority experienced a mild acute illness (83% in WT
Participant characteristics and 75% in VOC, p = 0.07). There were no significant dif-
From 1st March 2020 to 10th March 2022, 2,392 individu- ferences in ethnicity, BMI or comorbidities across time
als were recruited to the AIID cohort study, of these 1,644 periods, but significantly more individuals were vac-
were recruited for conditions other than post COVID cinated both prior to infection and after acute infection
review. Of the 748 individuals recruited at post COVID but prior to review in the VOC compared to the WT
review, 519 were still experiencing symptoms. Of these, period (0 (0%) and 15 (11%) vaccinated at time of infec-
52 did not have COVID-19 confirmed by PCR and 23 tion (p < 0.001) and 60 (25%) and 97 (75%) vaccinated by
were seen < 4 weeks from symptom onset. Of the remain- the time of clinic review (p < 0.001) in the WT and VOC
ing 444, 417 had complete data and were included in this periods respectively.
analysis. Of these, 268 were infected prior to the intro-
duction of the alpha variant in Ireland (wild-type (WT)
period) and 149 after the alpha variant became domi- Symptom clusters
nant (VOC period). Within the VOC period, 124 were In both WT and VOC periods, multiple correspond-
infected in alpha dominant and 25 in delta dominant ence analysis and hierarchical clustering revealed three
Table 1 Participant demographics
Whole cohort (n = 417) WT (n = 268) VOC (n = 149) P value
Age (years) 45 (35–55) 43 (36–54) 47 (34–57) 0.2
Sex (F) 307 (73) 196 (73) 109 (73) 0.9
Ethnicity (Caucasian) 352 (84) 222 (84) 129 (87) 0.31
BMI (kg/m2) 28 (24–32) 28 (24 – 32) 28 (24 – 33) 0.6
HCW 215 (51) 163 (61) 51 (35) < 0.001
Smoking 19 (4) 12 (4.5) 7 (4.7) 0.9
Alcohol 125 (30) 80 (30) 45 (30) 0.9
Comorbidity (Any) 243 (56) 152 (59) 89 (60) 0.9
Hypertension 74 (18) 45 (18) 28 (19) 0.8
Diabetes 23 (5) 14 (5.5) 9 (6.1) 0.9
Respiratory 78 (19) 45 (18) 32 (22) 0.4
-Asthma 62 (15) 36 (13) 25 (17) 0.43
Cardiac 21 (5) 13 (5.1) 8 (5.4) 0.9
Immunosuppression 5 (1) 2 (0.8) 3 (2) 0.5
Psychiatric 10 (2) 8 (3) 2 (1) 0.47
Vaccinated at time of infection 15 (3) 0 (0) 15 (11) < 0.001
Vaccinated at time of review 157 (37) 60 (25) 97 (75) < 0.001
Time from symptom onset (weeks) 22 (13–35) 24 (16–38) 18 (10–31) < 0.001
Mild initial disease severity 330 (79) 217 (83) 112 (75) 0.07
Hospitalised at acute infection 129 (31) 75 (28) 54 (36) 0.11
Legend: BMI Body mass index. HCW Healthcare worker. Continuous variables compared with Kruskal–Wallis test and categorical variables by chi square test.
Continuous data are median (IQR) and categorical data number (%)
Kenny et al. BMC Infectious Diseases (2023) 23:804 Page 4 of 8
distinct symptom clusters (Fig. 1A). Table S1 shows the the WT and VOC periods), poor concentration (pre-
cluster symptom profiles across groups. sent in 68% and 76% in the WT and VOC periods) and
The first cluster (n = 69 (26%) in WT and n = 25 (17%) GI symptoms (present in 19% and 32% in the WT and
in VOC) we labelled a musculoskeletal (MSK)/pain VOC periods) were also overrepresented in this clus-
cluster. This cluster was characterised by the highest ter. This cluster reported the greatest functional impact
number of symptoms (median (IQR) 6 (5–7) per indi- compared to other clusters across both time periods
vidual in both time periods), with joint pain and myal- (Table S2), with the longest time missed from work
gia in particular overrepresented compared to the other (median (IQR) 11.5 (4–18.75) weeks in the WT and
two clusters (joint pain present in 74% and 84% and 12 (4–32) weeks in the VOC period), and worst SF-36
myalgia in 64% and 92% in the WT and VOC period scores in the domains of physical functioning, pain,
respectively). Headache (present in 54% and 68% in general health and social functioning.
Fig. 1 Heatmaps of self-reported symptom clusters and forest plots showing odds of chest pain and palpitations associated with individual
symptoms. Legend: A Heatmaps demonstrating hierarchical clustering on self-reported symptoms. WT group ( 1st March 2020-25th December 2020)
is shown on the left and VOC group (26th December 2020- 1 st March 2022) on the right. VOC group included 124 individuals with acute infection
in the alpha variant dominant period and 25 infected in the delta dominant period. In both groups the top cluster is a musculoskeletal cluster
(n = 69 (26%) in WT and n = 23 (15%) in VOC), the middle is a cardiorespiratory cluster (n = 134 (50%) in WT and n = 30 (20%) in VOC) and the bottom
is a less symptomatic cluster (n = 65 (24%) in WT and n = 94 (63%) in VOC). B Forest plot showing the unadjusted and adjusted odds ratios
for reporting chest pain and palpitations
K enny et al. BMC Infectious Diseases (2023) 23:804 Page 5 of 8
Individuals in Cluster 2 self-reported a median of Looking next at cardiac imaging, results of 64 echocar-
3 (IQR 4–5) symptoms in the WT period and 4 (IQR diograms were available, 56 in the WT period (21%
3–5) in the VOC period. Characteristic symptoms were of WT period) and 8 in the VOC period (5% of VOC
cardiorespiratory in both periods. Dyspnea (present in period). 9 individuals had findings consistent with peri-
91% and 97% in the WT and VOC periods) and chest carditis or a pericardial effusion, 8 within the WT period
pain (present in 57% and 60% in the WT and VOC peri- and 1 in the VOC period 7 (78%) of these were in indi-
ods) were common in this cluster, however the propor- viduals in the cardiorespiratory cluster.
tion of individuals in this cluster reporting palpitations Results of 40 cardiac MRIs were available, 34 in the
was significantly higher in the WT period (34% vs 10% WT (13% of WT period) and 6 in the VOC period (4%
p = 0.008), and cough significantly lower (17% vs 63% of VOC period). There were 4 cardiac MRI confirmed
p < 0.001) in the WT than the VOC period. Overall this diagnoses of myocarditis, all of these were in individuals
cluster incorporated fewer individuals within the VOC within the cardiorespiratory cluster in the WT period.
period compared to the WT period (50% of WT and 20%
of VOC period, p < 0.001), and had less functional impact
in terms of illness associated work absence in the VOC Change in symptoms across periods of infection
compared to the WT period (median (IQR) 8 (3–17) Given the change in characteristic symptoms in the car-
weeks missed in the WT period and 4 (0–7) in the VOC diorespiratory cluster, we next compared the frequency
period, p = 0.007) (Table S4). of all symptoms across the full cohort during WT and
The third cluster reported less overall symptoms VOC periods to determine if an overall change in symp-
(median 2 (IQR 1–3) per patient across both time peri- tom frequency was mediating this difference. There were
ods. In the WT period, anosmia was more frequent in significant differences only in the frequency of palpita-
this cluster than the other two clusters, (present in 25%), tions (32% vs 12% in the WT vs VOC groups p < 0.0001)
whereas in the VOC period no single symptom predomi- and chest pain (39% vs 25% p = 0.004) but not cough (16%
nated. In line with the lower frequency of symptoms, this vs 15% p = 0.9), or any other symptom.
cluster had the least functional impact across both peri- Looking first at chest pain, in univariate analysis, along
ods, with the best breathlessness scores as measured by with being in the VOC group (OR 0.51 (95% CI 0.33–0.8),
the MRC dyspnea scale, and the highest health related p = 0.003), increasing age (OR 0.97 (95% CI 0.96–0.99),
quality of life in the SF36 domains of physical function- p < 0.001), and moderate severity acute COVID-19 (OR
ing, pain, general health and social functioning in both 0.43 (95% CI 0.19–0.9), p = 0.033) were significantly asso-
periods (Table S4). ciated with a reduced odds of chest pain (Fig. 1B, Table
S3). There was no association between vaccination, either
Vital signs and investigations at the time of initial infection or at the time of review
Three hundred thirteen (75%) of individuals had rest- and odds of reporting chest pain. In adjusted analysis
ing vital signs available. Median (IQR) heart rate was only older age (OR 0.98 (95% CI 0.96–0.99), p = 0.02)
74 (66–85) for the overall cohort and there was no dif- and being in the VOC group (OR 0.5 (95% CI 0.29–0.87),
ference between WT and VOC periods (median (IQR) p = 0.02) remained significantly associated with a reduced
heart rate 74 (65–85) in the WT period, 74 (66–84) in odds of reporting chest pain.
the VOC period, p = 0.72). Given the change in subjective Similarly, regarding palpitations, in univariate analy-
palpitations within the cardiorespiratory cluster across sis being in the VOC group (OR 0.29 (95% CI 0.16–0.5),
WT and VOC periods we next looked specifically at p < 0.0001), older age (OR 0.97 (95% CI 0.96–0.99),
vital signs within this cluster. 102 (76% of the WT and 21 p = 0.02), and more severe acute disease (moderate dis-
(70%) of the VOC cardiorespiratory cluster had resting ease OR 0.33 (95% CI 0.11–0.78), p = 0.022, severe disease
vital signs available and 72 (54%) of the WT and 20 (66%) OR 0.21 (95% CI 0.05–0.61), p = 0.011) were associated
of the VOC cardiorespiratory cluster had orthostatic vital with a reduced odds of reporting palpitations. Interest-
signs available. There was no difference in heart rate in ingly, vaccination at the time of review (OR 0.46 (95% CI
the WT cardiorespiratory cluster at rest (median (IQR) 0.28–0.76), p = 0.003) (Fig. 1B, Table S4), but not at the
resting HR 75 (67–85) in WT and 69 (65–80) in VOC time of infection (OR 0.21 (95% CI 0.01–1.04), p = 0.13)
period p = 0.35), or at 1, 3 and 5 min standing (median was also associated with a reduced odds of reporting
(IQR) HR at 1 min standing; 85 (78–96) in WT and 78 palpitations. However in fully adjusted analysis, only
(72–89) in VOC, at 3 min standing; 86 (78–97) in WT being in the VOC group (OR 0.37 (95% CI 0.18–0.71),
and 79 (72–90) in VOC, at 5 min standing; 88 (79–97) p = 0.004), but not vaccination (OR 0.59 (95% CI 0.33–
in WT and 79 (75–91) in VOC p = 0.19) compared to the 1.07), p = 0.08), remained significantly associated with a
VOC cardiorespiratory cluster. reduced odds of palpitations.
Kenny et al. BMC Infectious Diseases (2023) 23:804 Page 6 of 8
Discussion D614G found in both alpha and delta variants, increased
In this analysis we demonstrate three similar symp- infectivity in the upper but not lower respiratory tract
tom clusters in two, independent groups of participants , and may have altered viral tropism to cardiac tis-
with confirmed SARS-CoV-2 at different periods of the sues. Atypical cardiac inflammation has been described
pandemic. Validation of these clusters across independ- in individuals with these long COVID symptoms .
ent time points suggests distinct clinical phenotypes While we observed fewer diagnoses of pericarditis and
that may share underlying mechanisms, while variation myocarditis within the VOC compared to the WT period
in reported symptoms suggests modification of long within this analysis, relatively few individuals had cardiac
COVID over time. imaging results available, and the relationship between
Similarly, we observed a decrease in cardiac symptoms VOC infection and cardiac MRI findings requires further
of chest pain and palpitations between the pre and post study.
VOC periods. This association between infection period There has been considerable interest in the effect of
and lower odds of palpitations or chest pain remained vaccination on long COVID symptoms. While most
robust in fully adjusted models, suggesting this change studies have focused on the effect of vaccination prior
may be mediated by change in the infecting variant. to infection, a number of observational studies have sug-
Interestingly, while the symptoms contributing to the gested an improvement in symptoms with vaccination
clusters were broadly similar across the two time peri- [7, 24]. While we observed an association between vac-
ods, we observed a shift from cardiac to respiratory pre- cination and a reduced odds of reporting palpitations
dominant symptoms within the cardiorespiratory cluster in univariate analysis, this was significantly attenuated
between WT and VOC periods, with a decrease in pal- after adjustment, while the association between reduced
pitations, an increase in cough but persistent dyspnea. reporting of palpitations and infection in the VOC period
There have been a number of studies focused on pheno- persisted. This would suggest that the change in long
typing long COVID. While these differ in data source, COVID symptoms was mediated more by time period of
analytic approach and symptoms included, an MSK/pain infection rather than by vaccination status. This is con-
dominant, cardiorespiratory and less symptomatic clus- sistent with observations elsewhere that while vaccina-
ter have been found in other studies [5, 17–20]. However, tion may reduce long COVID incidence , in those
few studies have examined the change in long COVID who develop post-acute symptoms patterns are similar
phenotype with infection period. One study  using , and demonstrates the importance of considering
smartphone application data examined long COVID the introduction of VOCs when determining the effect of
symptom profiles across WT, alpha and delta waves. vaccination on long COVID.
While this study demonstrated two clusters similar to This analysis has limitations. Although there was a
the cardiorespiratory cluster and musculoskeletal clus- distinct time period in Ireland when the alpha variant
ter observed here across all periods, it found different became dominant, most subjects attended for clinic visits
numbers of clusters across variants due to the inclusion long after the initial infection, therefore sequencing data
of less frequently reported symptoms, limiting the abil- to confirm the original infecting variant was not avail-
ity to examine changes in the consistent phenotypes. able. In addition, an insufficient number of individuals
This study also observed a decrease in size of the cardi- reporting SARS-CoV-2 infection during the Delta domi-
orespiratory cluster with the introduction of VOCs. The nant period, meant we were unable to analyse these VOC
authors postulate that this reflects a reduction in lung independently, and this analysis included no individuals
damage with the introduction of vaccines, however phy- infected during Omicron dominant periods prevent-
sician assessed acute disease severity available in this ing a more detailed examination of the different com-
study allowed us to test this hypothesis, finding timing of mon VOC. We included only the twelve most commonly
infection, rather than disease severity or vaccination may reported symptoms, and may therefore have missed the
mediate this reduction in cardiac symptoms. impact of VOC on less frequently reported symptoms,
We report a decrease in cardiac symptoms with the however we believe including only the most common
introduction of VOCs that has not been demonstrated symptoms improved cluster interpretability compared
elsewhere. Magnusson et al. looked at the difference in to some larger studies derived from electronic health
long COVID symptoms between only delta and omicron records . We do not explore classification methods
infections, and found no difference in any post COVID in this analysis, and further research is needed to iden-
outcome between these two variants , suggesting the tify how to best assign individuals to the most appropri-
change in cardiac symptoms may be due to the initial ate cluster. Although we report data from a multicentre
adaptations of the alpha variant that have been retained study, participating hospitals are all within Ireland and
in subsequent variants. For example the spike mutation further research is needed to replicate these findings in
K enny et al. BMC Infectious Diseases (2023) 23:804 Page 7 of 8
independent cohorts. Cardiac imaging and vital sign Funding
data was incomplete, limiting power to detect differences GK was funded through a fellowship from the Embassy of the United States
in Ireland during the study. This analysis was funded through an unrestricted
across symptom clusters and variants, and precluding grant from Smurfit Kappa.
conclusions of the effect of VOCs on these findings.
In summary, this analysis demonstrates three clinical Availability of data and materials
Researchers can apply for access to pseudonymised data by submitting a data
phenotypes of long COVID, observed across two distinct access request to the All Ireland Infectious Diseases Cohort steering group.
time periods. We observed a change in pattern of symp- Requests can be sent to the corresponding author, and access will be granted
toms between the two time periods, which were particu- depending on a data protection impact assessment and assessment of the
larly associated with the cardiorespiratory phenotype,
driven by infection during the VOC period rather than Declarations
vaccination. Further research is needed to determine the
pathophysiologic basis of these differences. Ethics approval and consent to participate
All study procedures adhered to required guidelines. Participants provide
written informed consent for collection of data on demographics, clinical
Supplementary Information characteristics and investigations undertaken as part of routine care. The AIID
The online version contains supplementary material available at https://d oi. cohort study was approved in line with national and European regulations on
org/1 0. 1186/ s12879-0 23- 08783-y. health research by the St Vincent’s Hospital group Research Ethics committee
and the National Research Ethics Committee for COVID-19 in Ireland.
Additional file 1: Supplemental Table 1. Proportion of individuals
Consent for publication
experiencing symptoms in each cluster in wild type and variant of con-
cern groups. Supplemental Table 2. Not applicable. Functional impact across clusters.
Supplemental Table 3. Univariate and multivariate models demonstra-
tion association of factors with self-reported chest pain. Supplemental
table 4. E. F. has received consulting fees from Gilead, ViiV, and Vidacare Ireland, and Univariate and multivariate models demonstration association of
has been awarded a grant from Science Foundation Ireland, outside the
factors with self-reported palpitations
submitted work. E. D. B. has received consulting fees from Sanofi Pasteur and
honoraria/travel grant from Pfizer. P. W. G. M. has received honoraria and/or
Acknowledgements travel grants from Gilead Sciences, MSD, Bristol-Myers Squibb, and ViiV Health-
The authors wish to thank all study participants and their families for their care, and has been awarded grants by Science Foundation Ireland, outside the
participation and support in the conduct of the All Ireland Infectious Diseases submitted work. All other authors report no potential conflicts of interest.
The All Ireland Infectious Diseases Cohort Study Investigators: Author details1
Grace K enny1,2, Kathleen M cCann2, Rachel MacCann1,2, Cathal O’Broin1,2, Ste- Centre for Experimental Pathogen Host Research, University College Dublin, 2
fano Savinelli 1,2, Eoin R Feeney1,2, Patrick WG M allon1,2. Alejandro Garcia Leon1, Belfield Dublin 4, Ireland. St Vincent’s University Hospital, Dublin, Ireland. 3
Sarah M iles1, Dana Alalwan1, Riya N egi1 , Obada Y ousif4, Aoife G Cotter5, Eavan School of Medicine, University College Dublin, Dublin, Ireland.
Muldoon5, Gerard S heehan5, Tara McGinty5, John S. Lambert5, Sandra Green5, General Hospital, Wexford, Ireland. Mater Misericordiae University Hospital, 6 7
Kelly Leamy5, Christine Kelly5, , Eoin de B arra6,7, Samuel McConkey6,7, Killain Dublin, Ireland. Beaumont Hospital, Beaumont, Dublin 9, Ireland. Depart-
Hurley6, Imran Sulaiman6, Mary Horgan8, Corinna S adlier8, Joseph Eustace8, ment of International Health and Tropical Medicine, Royal College of Surgeons
Tommy Bracken11 Bryan Whelan12, Justin Low13, Bairbre McNicholas14, Garry in Ireland, Dublin, Ireland.
8 Department of Infectious Diseases, Cork University
Hospital, Wilton, Cork, Ireland. 9 Courtney15, Patrick Gavin16. Department of Internal Medicine, Rush Uni-10
1Centre for Experimental Pathogen Host Research, University College Dublin, versity, Chicago, IL, USA. Clinical Trials Institute, University of Galway, Galway,
Dublin, Ireland Ireland.
2St Vincent’s University Hospital, Dublin, Ireland
3School of Medicine, University College Dublin, Dublin, Ireland Received: 8 June 2023Accepted: 2 November 2023
4Wexford General Hospital, Wexford, Ireland
5Mater Misericordiae University Hospital, Dublin, Ireland
6 Beaumont Hospital, Beaumont, Dublin 9, Ireland
7Department of International Health and Tropical Medicine, Royal College of
Surgeons in Ireland, Dublin, Ireland References
8Department of Infectious Diseases, Cork University Hospital, Wilton, Cork, 1. Kenny G, Mallon PW. COVID19- clinical presentation and therapeutic
Ireland. 9Department of Internal Medicine, Rush University, Chicago, IL, USA considerations. Biochem Biophys Res Commun. 2021;538:125–31.
10Clinical Trials Institute, University of Galway, Galway, Ireland 2. Su Y, Yuan D, Chen DG, et al. Multiple early factors anticipate post-acute
11University College Dublin, Dublin, Ireland COVID-19 sequelae. Cell. 2022;185:881-895e20.
12Sligo University Hospital, Sligo, Ireland 3. Kenny G, McCann K, O’Brien C, et al. Identification of distinct long COVID
13Our Lady of Lourdes Hospital, Drogheda, Ireland clinical phenotypes through cluster analysis of self-reported symptoms.
14University Hospital Galway, Galway, Ireland Open Forum Infect Dis. 2022;9: ofac060.
15St Luke’s Hospital Kilkenny, Kilkenny, Ireland 4. Canas LS, Molteni E, Deng J, et al. Profiling post-COVID syndrome across
16Children’s Health Ireland, Dublin, Ireland different variants of SARS-CoV-2. MedRxiv. 2022. https:// doi. org/ 10. 1101/
2022.0 7.2 8.2 22781 59.
Authors’ contributions 5. Zhang H, Zang C, Xu Z, et al. Machine learning for identifying Data-Driven
GK recruited participants, conceived the study, performed the analyses and subphenotypes of Incident Post-acute SARS-CoV-2 Infection conditions
wrote the manuscript. KMcC recruited participants and contributed to data with Large Scale Electronic Health Records: findings from the RECOVER
entry. C’OB and T’OG contributed to data entry. WT supervised the statistical Initiative. Public Global Health. 2022. https://d oi. org/1 0.1 101/2 022. 05.2 1.
analysis. C’OB, OY, AC, JL, EF, EdB, CS, PD, SS and PM all oversaw participant 222754 12.
recruitment and data acquisition. AL contributed to the design and analysis of 6. Al-Aly Z, Bowe B, Xie Y. Long COVID after breakthrough SARS-CoV-2 Infec-
the study. PM contributed to the study design, analysis, manuscript prepara- tion. Nat Med. 2022;28:1461–7.
tion. All authors reviewed the final manuscript.
Kenny et al. BMC Infectious Diseases (2023) 23:804 Page 8 of 8
7. Ayoubkhani D, Bermingham C, Pouwels KB, et al. Trajectory of long covid
symptoms after covid-19 vaccination: community based cohort study.
8. Twohig KA, Nyberg T, Zaidi A, et al. Hospital admission and emergency
care attendance risk for SARS-CoV-2 delta (B.1.617.2) compared with
alpha (B.1.1.7) variants of concern: a cohort study. Lancet Infect Dis.
9. Davies NG, Abbott S, Barnard RC, et al. Estimated transmissibility and
impact of SARS-CoV-2 lineage B.1.1.7 in England. Science. 2021;372:
10. Kenny G, Negi R, O’Reilly S, et al. Performance and validation of an adapt-
able multiplex assay for detection of serologic response to SARS-CoV-2
infection or vaccination. J Immunol Methods. 2022;510:113345.
11. World Health Organisation. COVID-19 Clinical management Living guid-
ance. 2021; published online Jan 25. https:// www. who. int/ public atio ns/i/
item/ WHO- 2019- nCoV- clinic al-2 021-1Accessed 23 Sept 2021.
12. Stenton C. The MRC breathlessness scale. Occup Med. 2008;58:226–7.
13. Blake C, Codd MB, O’Meara YM. The short form 36 (SF-36) Health Survey:
normative data for the Irish population. Ir J Med Sci. 2000;169:195–200.
14. Di Franco G. Multiple correspondence analysis: one only or several tech-
niques? Qual Quant. 2016;50:1299–315.
15. Health Protection Surveillance Centre. Summary of COVID-19 virus vari-
ants in Ireland Report prepared by HPSC and NVRL on 23/03/2022. http://
hdl. handle. net/1 0147/6 31713Accessed 9 Aug 2023.
16. Althouse AD. Adjust for multiple comparisons? It’s not that simple. Ann
Thorac Surg. 2016;101:1644–5.
17. Sahanic S, Tymoszuk P, Ausserhofer D, et al. Phenotyping of acute and
persistent COVID-19 features in the outpatient setting: exploratory
analysis of an international cross-sectional online survey. Clin Infect Dis.
2022;75(1):e418–31. https:// doi. org/ 10. 1093/ cid/ ciab9 78.
18. Frontera JA, Thorpe LE, Simon NM, et al. Post-acute sequelae of COVID-19
symptom phenotypes and therapeutic strategies: a prospective, observa-
tional study. PLoS One. 2022;17:e0275274.
19. Whitaker M, Elliott J, Chadeau-Hyam M, et al. Persistent COVID-19 symp-
toms in a community study of 606,434 people in England. Nat Commun.
20. Fernández-de-Las-Peñas C, Martín-Guerrero JD, Florencio LL, et al. Clus-
tering analysis reveals different profiles associating long-term post-COVID
symptoms, COVID-19 symptoms at hospital admission and previous
medical co-morbidities in previously hospitalized COVID-19 survivors.
Infection. 2022. https:// doi.o rg/ 10. 1007/ s15010- 022- 01822-x. published
online April 22.
21. Magnusson K, Kristoffersen DT, Dell’Isola A, et al. Post-covid medical com-
plaints following Infection with SARS-CoV-2 omicron vs Delta variants.
Nat Commun. 2022;13:7363.
22. Plante JA, Liu Y, Liu J, et al. Spike mutation D614G alters SARS-CoV-2 fit-
ness. Nature. 2021;592:116–21.
23. Puntmann VO, Martin S, Shchendrygina A, et al. Long-term cardiac
pathology in individuals with mild initial COVID-19 Illness. Nat Med.
24. Nehme M, Braillard O, Salamun J, et al. Symptoms after COVID-19 vac-
cination in patients with Post-acute Sequelae of SARS-CoV-2. J Gen Intern
25. Antonelli M, Penfold RS, Merino J, et al. Risk factors and Disease profile of
post-vaccination SARS-CoV-2 Infection in UK users of the COVID Symp-
tom Study app: a prospective, community-based, nested, case-control
study. Lancet Infect Dis. 2022;22:43–55.
26. Reese JT, Blau H, Bergquist T, et al. Generalizable long COVID subtypes: Ready to submit your research ? Choose BMC and benefit from:
findings from the NIH N3C and RECOVER programs. Infect Dis (except • fast, convenient online submission
HIV/AIDS). 2022. https:// doi.o rg/ 10. 1101/2 022. 05.2 4.2 2275 398.
• thorough peer review by experienced rese archers in your field
Publisher’s Note • rapid publication on acceptance
Springer Nature remains neutral with regard to jurisdictional claims in pub- • support for research data, including large and complex data types
lished maps and institutional affiliations. • gold Open Access which fosters wider collaboration and increased citations
• maximum visibility for your research: over 100M website views per year
AtBMC, research is always in progress.
Learn more biomedcentral.com/submissions