key: cord-0756554-4hxzf98w
authors: Schwellnus, Martin; Sewry, Nicola; Snyders, Carolette; Kaulback, Kelly; Wood, Paola Silvia; Seocharan, Ishen; Derman, Wayne; Hull, James H; Valtonen, Maarit; Jordaan, Esme
title: Symptom cluster is associated with prolonged return-to-play in symptomatic athletes with acute respiratory illness (including COVID-19): a cross-sectional study—AWARE study I
date: 2021-03-22
journal: Br J Sports Med
DOI: 10.1136/bjsports-2020-103782
sha: d2f8b9afd1425e1c53365c147dbfadf02bfdd5a3
doc_id: 756554
cord_uid: 4hxzf98w

BACKGROUND: There are no data relating symptoms of an acute respiratory illness (ARI) in general, and COVID-19 specifically, to return to play (RTP). OBJECTIVE: To determine if ARI symptoms are associated with more prolonged RTP, and if days to RTP and symptoms (number, type, duration and severity) differ in athletes with COVID-19 versus athletes with other ARI. DESIGN: Cross-sectional descriptive study. SETTING: Online survey. PARTICIPANTS: Athletes with confirmed/suspected COVID-19 (ARI(COV)) (n=45) and athletes with other ARI (ARI(OTH)) (n=39). METHODS: Participants recorded days to RTP and completed an online survey detailing ARI symptoms (number, type, severity and duration) in three categories: ‘nose and throat’, ‘chest and neck’ and ‘whole body’. We report the association between symptoms and RTP (% chance over 40 days) and compare the days to RTP and symptoms (number, type, duration and severity) in ARI(COV) versus ARI(OTH) subgroups. RESULTS: The symptom cluster associated with more prolonged RTP (lower chance over 40 days; %) (univariate analysis) was ‘excessive fatigue’ (75%; p<0.0001), ‘chills’ (65%; p=0.004), ‘fever’ (64%; p=0.004), ‘headache’ (56%; p=0.006), ‘altered/loss sense of smell’ (51%; p=0.009), ‘Chest pain/pressure’ (48%; p=0.033), ‘difficulty in breathing’ (48%; p=0.022) and ‘loss of appetite’ (47%; p=0.022). ‘Excessive fatigue’ remained associated with prolonged RTP (p=0.0002) in a multiple model. Compared with ARI(OTH), the ARI(COV) subgroup had more severe disease (greater number, more severe symptoms) and more days to RTP (p=0.0043). CONCLUSION: Symptom clusters may be used by sport and exercise physicians to assist decision making for RTP in athletes with ARI (including COVID-19).

Protecting the athlete against acute illness is a key consideration for health professionals involved in sport. About 50% of all acute illness in athletes during competitions and tournaments affect the respiratory system, 1 2 and the most frequent illness is an acute infection caused by a wide range of pathogens, mostly viruses. 3 At times, athletes can be more prone to acute respiratory illness (ARI), especially during times of increase training load, competitions or travel. [4] [5] [6] [7] In 2019, the SARS-CoV-2 was identified as a novel respiratory pathogen responsible for COVID-19. COVID-19 is associated with a broad range of symptoms, including symptom clusters localised to the upper respiratory tract (eg, anosmia, dysgeusia, coryzal symptoms and throat discomfort), general thoracic symptoms (eg, cough, shortness of breath, headache or chest pain) and systemic symptoms (eg, myalgia, fever or excessive fatigue). 8 In addition to causing acute respiratory disease (ie, a viral pneumonitis), data indicate that COVID-19 can impact multiple organ systems, including the cardiovascular, neurological, gastrointestinal and renal systems and can impede skeletal muscle function. 9 10 Even mild / moderate COVID-19 infection (ie, non-hospitalised individuals), can be associated with protracted or 'Long-COVID' symptoms in previously healthy individuals. 11 Serious potential complications, resulting from COVID-19, include myocarditis, which is also reported in athletic individuals 12 placing an athlete at significant health risk, when returning to vigorous exercise.

A key role within sport and exercise medicine is to provide guidelines for safe return to play (RTP) after illness. A number of expert position statements have been published to guide physicians in decision making on RTP after COVID-19 infection, but these focused on cardiovascular assessment and are based on expert opinion and experience. 13 14 There are limited published data that describe or characterise symptoms or groups of symptoms (ie, symptom clusters) in athletes, during the acute phase of an ARI in general, and COVID-19 specifically. In the general (ie, non-athletic) population, recently published data indicates that symptom clusters of COVID-19 are valuable and can predict probable infection 15 and other clinical outcomes such as need for respiratory support, 16 hospitalisation 17 and 'Long-COVID'. 18 The primary aim of this study was to determine if symptoms experienced during the acute phase of an ARI in athletes was associated with days to RTP. A secondary aim was to determine if the number, type, duration and severity of ARI symptoms differ in athletes with confirmed or suspected COVID-19, compared with athletes with other causes of ARI Original research . These data could be used to develop a future prediction model (based on symptoms and other variables) to guide the RTP clinical decision making for athletes with recent ARI. This information is required urgently so that health professionals can advise competitive and recreational athletes and their coaching support systems as they return to play following COVID-19 or another ARI.

The Athletes With Acute Respiratory InfEctions (AWARE studies) is a multicentre study, led by the Sport, Exercise Medicine and Lifestyle Institute at the University of Pretoria in South Africa, together with researchers from a number of academic institutions, sports federations and some members of 'subgroup 7' of the IOC Consensus group on 'Acute Respiratory Illness in the Athlete'. This is a descriptive cross-sectional study of data collected between 20 July 2020 and 13 October 2020.

Potential participants for this study were athletes (age 18-60 years), defined as 'competing at varying levels in any sport, training for a minimum of 3 hours per week'. Following informed electronic consent, participants completed an online survey using the Research Electronic Data Capture platform. 19 20 A link to the online survey was distributed to potential participants using existing databases of athletes and via several social media platforms. For participants to be included in this study, they had to report at least one symptom lasting ≥1 day. Details regarding participant recruitment is shown in online supplemental figure 1.

Participants completed their history of respiratory health, including symptoms of a recent (<6 month period) ARI, including COVID-19. The survey evaluated athlete demographics, comorbidities, training history, RTP and residual symptoms. Participants reported the following details about ARI: number, type, severity and duration of symptoms (days). Symptom type consisted of 26 symptoms of ARI in three categories: (1) 'nose and throat' (eight symptoms), (2) 'chest and neck' (eight symptoms) and (3) 'whole body' (10 symptoms, including gastrointestinal symptoms). Symptom severity was reported on a scale from 1 to 7, which was adapted from the validated Wisconsin Upper Respiratory Symptom Survey (WURSS). 21 The duration (days) that each symptom lasted was recorded. The days to RTP were reported by participants in response to the question 'How many days were there between the start of your symptoms and the return to your first training session?'.

The symptom list and severity score were adapted from the validated WURSS, 21 to which we added additional COVID-19 symptoms. 8 A pilot study of 16 participants, consisting of athletes and medical practitioners directed early development and assessed accessibility of the survey (the symptoms, the length of the questionnaire and terminology used).

PPI was included in this study. Athletes and medical practitioners treating athletes who had experienced an ARI (including COVID-19) were asked to provide feedback on the questionnaire in the development stages.

Participants (ARI ALL ) (n=84) in this study were divided into subgroups, based on their responses to questions in the survey that related to the diagnosis of the ARI (online supplemental figure 1 . For the purpose of our cross-sectional analysis of subgroups, we combined the ARI COV+ve and the ARI COV-S subgroups into one group: ARI with COVID-19 (ARI COV ) (n=45).

The primary outcome measure was the self-reported number of days to RTP. Secondary analyses included evaluation of the association between symptom type and the days to RTP. We also compare the number of symptoms/athlete (maximum of 26), type of symptoms (by region and individual symptoms), duration of each symptom (number of days) and the severity of each symptom in two categories (mild: score 1-3; moderate/severe: score 4-7) between the ARI COV and ARI OTH subgroups.

Demographic data were described for the ARI COV (including ARI COV+ve and ARI COV-S ) and ARI OTH subgroups. The responses to the type of ARI symptoms were modelled using the log-binomial model, and the prevalence, 95% CI and prevalence ratios were reported. χ 2 (p values) were used for significance testing (type 3). The negative binomial model was used for the comparison of the mean number of symptoms between the ARI COV and the ARI OTH groups. The median number of days and IQRs were reported for describing the duration of symptoms. The comparison of duration of symptoms (median duration in days) between subgroups was done using non-parametric survival analysis. For participants that did not report a specific symptom, the days of a symptom was considered censored; that is, the comparison of days with a symptom between groups was for those participants reporting the symptom. Nelson-Aalen estimates were reported, and the log-rank χ 2 (p values) were reported for testing of the homogeneity of survival curves over the groups.

The Cochran-Armitage trend test was used to test for a difference in severity (no symptoms, mild and moderate/severe) of ARI symptoms between subgroups. A one-sided z test (p value) was reported.

For the days to RTP analysis, six participants did not report RTP days, resulting in 78 cases for this analysis. For the 49 participants reporting response 1, the actual days RTP were recorded. For the 23 participants reporting response (2) (above), censoring was indicated for their RTP days. For the six participants reporting response (3) above, 0 days RTP were recorded. Non-parametric survival analysis was done to compare the median duration (in days) between the two subgroups. The log-rank χ 2 (p values) were reported. A figure with separate Product-Limit Survival Estimates for the two subgroups was included.

For the analysis to predict the RTP from the 'type of symptom', a Cox regression was done including the individual ARI symptoms in the model. The interaction for ARI group × symptom was tested, and separate HRs for the groups were reported when the interaction was significant. Symptoms with less than 15 events were not analysed. The Firth's corrected estimates were computed but not reported since they were consistent with the uncorrected estimates. 22 Three multiple models were presented, one for each subgroup of symptoms: nose and throat, chest and neck and whole body. For each model, significant variables from the univariate models were entered (p<0.05). However, the impact of multicollinearity †Symptoms where the numbers were too small to analyse between groups or produce CIs. ARI, acute respiratory illness; ARI COV , acute respiratory illness COVID-19 subgroup that is the combined ARI COV+ve and ARI COV-S subgroups.; ARI OTH , acute respiratory illness Other subgroup; PR, prevalence ratio.

was considered in the modelling. 23 Tetrachoric correlations were calculated for variables within each subgroup of symptoms and variables with a correlation ≥0.8 were not entered together. The three final models were presented without doing any stepwise selection.

Data from 181 participants were available, and 97 were excluded because they had no ARI symptoms, including 11 participants who had a positive COVID-19 test but were asymptomatic (online supplemental figure 1 ). The remaining 84 participants were included in this analysis and their demographics are shown in table 1.

There was no significant difference in the demographic variables (age, sex, height and weight), professional sporting level and years of sporting experience between ARI OTH and the ARI COV subgroups.

The mean number of ARI symptoms (out of 26) in the ARI ALL group was 9.2 (95% CI 8.1 to 10.3).

The mean number of symptoms was significantly higher in the ARI COV subgroup (10.4 (95% CI 8.9 to 12.1) compared with the ARI OTH subgroup (7.8; 95% CI 6.5 to 9.2) (p=0.016). The mean number of 'nose and throat' symptoms for ARI ALL was 3.4 (95% CI 3.0 to 3.8), and this was not different between the ARI COV (3.6; 95% CI 3.1 to 4.2) and ARI OTH (3.1; 95% CI 2.5 to 3.7) (p=0.139). The mean number of 'chest and neck' symptoms for ARI ALL was 2.9 (95% CI 2.5 to 3.4), and this was not different between subgroups (ARI COV : 3.2; 95% CI 2.6 to 3.9); ARI OTH : (2.5; 95% CI 2.0 to 3.2) (p=0.14). The mean number of 'whole body' symptoms for ARI ALL 

The symptoms (type, number and percentage with 95% CI) by region and specific symptoms of ARI in the ARI ALL and the subgroups (ARI COV and ARI OTH ) are shown in table 2.

In the ARI ALL group, the four most common symptoms were 'excessive fatigue' (73.8%), 'headache' (64.3%), 'blocked/plugged nose' (61.9%) and 'sore/scratchy throat' (52.4%). The following Table 3 The duration (median number of days; 95% CI) of symptoms by region and specific symptoms in the ARI ALL , and the ARI COV and the ARI OTH subgroups †Symptoms where the numbers were too small to analyse between groups or produce CIs. ARI, acute respiratory illness; ARI COV , acute respiratory illness COVID-19 subgroup that is the combined ARICOV+ve and ARICOV-S subgroups; ARI OTH , acute respiratory illness Other subgroup. symptoms were significantly more common in the ARI COV compared with the ARI OTH subgroup: 'altered/loss of sense of smell' (prevalence ratio (PR)=4.48; p=0.0001), 'altered/loss of sense of taste' (PR=3.59; p=0.0001), 'fever' (PR=2.95; p=0.008), 'loss of appetite' (PR=2.17; p=0.005), 'chills' (PR=2.1; p=0.042) and 'headaches' (PR=1.6; p=0.005). Other symptoms included: dry eyes, hives, hunger, racing heart beats, vertigo and insomnia (all n=1).

The duration of symptoms (days) by region and specific symptoms in ARI ALL and the ARI COV and the ARI OTH subgroups is shown in table 3 (for ARI ALL the median and IQR is presented in online supplemental table S1).

For 'nose and throat' symptoms, 'altered/loss sense of smell' and 'altered/loss sense of taste' took the longest to recover (median 10 and 9.5 days, respectively). For 'chest and neck' symptoms, 'difficulty in breathing' and 'fast breathing or shortness of breath' took the longest to recover (median 14 days), and for 'whole body' symptoms, 'excessive fatigue' took the longest to recover (median 11 days). The only symptom with a significantly longer duration in the ARI COV versus the ARI OTH subgroup was 'blocked/ plugged nose' (median of 10 days vs 5 days, p=0.006).

The number and percentage (%) of athletes with mild (score 1-3) versus moderate/severe (score 4-7) symptoms by region and specific symptoms in the ARI COV and the ARI OTH subgroups is shown in table 4 (the 'no symptom' group is not noted in table 4 but is included in the analyses).

The % moderate/severe symptoms was higher in the ARI COV subgroup compared with the ARI OTH subgroup for the following symptoms: 'altered/loss of smell' (58% vs 10%; p=0.0001), 'altered/loss of taste' (47% vs 13%; p=0.0001), 'headache' (64% vs 31%; p=0.0008), 'fever' (22% vs 10%; p=0.013), 'chills' (24% vs 13%; p=0.035) and 'loss of appetite' (42% vs 13%; p=0.001).

Of the 84 ARI ALL athletes, six did not report number of days to RTP, resulting in 78 data points for the analysis of days to RTP Table 4 The number and percentage (N, %) of athletes with mild (score 1-3) and moderate/severe (score 4-7) symptoms by region and specific symptoms in the ARI COV and the ARI OTH subgroups (athletes who did not report the symptom, had a severity of 0 and make up the rest to 100%) 

(ARI COV : n=43; ARI OTH : n=35). In the ARI ALL group, the estimated median days to RTP was 20 (95% CI 14 to 30), indicating that 50% of athletes would have returned to play by day 20, and 75% of athletes would have returned to play by day 30. The survival probability (%) of athletes in the ARI COV and the ARI OTH group after the onset of symptoms and RTP is shown in figure 1 .

The median days to RTP for the ARI COV subgroup (30 days; 95% CI 16 to 40) was significantly longer than the ARI OTH subgroup (10 days; 95% CI 7 to 22) (log-rank test: p=0.0043). The survival curves for the two groups both drop substantially at around day 7 (13 athletes returning to play by day 7). After 7 days, the rate of RTP was lower in the ARI COV group than the ARI OTH group.

The HR (HR and 95% CI) for symptoms (by region and specific symptoms) in the ARI ALL group is shown in table 5 (Univariate model). The HR was derived as the ratio of the hazard of RTP for an individual with the symptom, compared with the hazard of RTP for an individual without the symptom. An HR <1 indicates a lower % chance of RTP in the 40-day period after the onset of symptoms for an individual with the symptom compared with an individual without the symptom, indicating a more prolonged RTP. The interaction for group × symptom was tested when analysing the individual symptoms, and separate HRs were reported for the two groups where the interaction was significant.

Over the 40-day period, athletes with the following specific symptoms had a lower chance (% lower chance) and thus a more prolonged RTP compared with athletes without the symptom: 'excessive fatigue' (75%; p<0.0001), 'chills' (65%; p=0.004), 'fever' (64%; p=0.004), 'headache' (56%; p=0.006), 'altered/ loss sense of smell' (51%; p=0.009), 'chest pain/pressure' (48%; p=0.033), 'difficulty in breathing' (48%; p=0.022) and 'loss of appetite' (47%; p=0.022). Athletes with altered loss of taste or chest pain experienced a marginal but potentially clinically significantly longer RTP compared with athletes without the symptoms. 'Difficulty in breathing' was the only symptom that had a significant interaction effect (p=0.029), with the separate HR estimates for ARI OTH (0.18; 95% CI 0.08 to 0.45, p=0.0002) and ARI COV (0.67; 95% CI 0.30 to 1.49, p=0.328). In the ARI OTH RTP was significantly longer in those reporting 'difficulty in breathing' (median 30 days; 95% CI 10 to 40) versus those not reporting the symptom (median 7 days; 95% CI 1 to 10).

The HR (95% CI) of days to RTP for symptoms in the ARI ALL group is shown in online supplemental table S2 (multiple model) (further details on the collinearity are shown in online supplemental table S3). The multiple model included six significant symptoms from the univariate analysis (without correlation >0.8). Of these included symptoms, 'altered/loss sense of smell', 'headache' and 'excessive fatigue' remained significant. 'Excessive fatigue' was associated with a 70% lower chance of RTP over the 40-day period, compared with not reporting 'excessive fatigue'. 'Altered/loss sense of smell' was significant (p=0.009), with a 51% lower chance of RTP, as well as 'headache' having a 49% lower chance of RTP.

The novel and clinically most important finding of this study is that the following cluster of symptoms of ARI are significantly associated (univariate model) with a longer RTP (listed in order by the % chance of more prolonged RTP in the 40-day period Figure 1 The survival probability of days to return to play in the ARI COV (n=43) and the ARI OTH group (n=35) (p=0.004). Shaded areas depict the 95% CIs. Numbers above the x-axis represent the number of athletes who have not yet returned to play in their respective groups at that time point. ARI, acute respiratory illness.

after the onset of symptoms): 'excessive fatigue' (75%), 'chills' (65%), 'fever' (64%), 'headache' (56%), 'altered/loss sense of smell' (51%), 'chest pain/pressure' (48%), 'difficulty in breathing' (48%) and 'loss of appetite' (47%). All these associations were valid in the ARI OTH and the ARI COV subgroups, with the exception of the symptom 'difficulty in breathing' that was significantly associated with a longer RTP only for the ARI OTH subgroup. From our multiple model, we show that the most important ARI symptom associated with a longer RTP was 'excessive fatigue'-a 70% lower chance of RTP over the 40-day period. The symptoms 'altered/loss sense of smell' and 'headache' were also significantly associated with a longer RTP, a 51% and 49% less chance of RTP over the 40-day period, respectively. We are not aware of data from any other study that relate individual symptoms or clusters of ARI symptoms to RTP in athletes.

To date, the so-called neck check has been used in clinical decision making on RTP in athletes with ARI. 24 This guideline is not based on any scientific data, and its validity has recently been challenged also in the context of athletes with confirmed COVID-19. 25 We do note that the symptom cluster significantly associated with prolonged RTP in our study includes 'above the neck' and 'below the neck' symptoms. The simplified 'neck check' as a tool to guide RTP is not supported by our preliminary data, and we will investigate this in future AWARE studies, as we accumulate more data.

The symptom cluster is associated with prolonged RTP in the ARI ALL group and holds for the ARI COV subgroup and the ARI OTH subgroup, where we do not have data on the specific pathogen causing the ARI. We therefore tentatively suggest that this symptom cluster can be used by sport and exercise physicians to assist their decision making for RTP in athletes with COVID-19 and athletes with any ARI. Further research is needed to determine if this or other symptom clusters are associated with prolonged RTP in ARI caused by other pathogens, for example, adenovirus or influenza virus. We also suggest testing these preliminary findings in a predictive model of RTP for all ARI and pathogen-specific ARI and to refine such a model by including other variables, for example, age, sex, sport type, treatment modalities used and existing comorbidities.

The second novel finding of this study is that the clinical presentation of symptoms and the RTP was significantly different in athletes with confirmed or suspected COVID-19 (ARI COV ), compared with the ARI OTH subgroup. Compared with ARI OTH , athletes in the ARI COV subgroup: (A) had significantly greater number of symptoms, (B) were more likely to have 'altered/loss of sense of smell' (PR=4.48) and 'altered/loss of sense of taste' (PR=3.59), 'fever' (PR=2.95), 'chills' (PR=2.1), 'loss of appetite' (PR=2.17) and 'headaches' (PR=1.6), (C) reported higher % of more moderate/severe symptoms, specifically for 'altered/ loss of smell', 'altered/loss of taste', 'headache', 'fever', 'chills' and 'loss of appetite' and (D) reported a longer RTP (median days=30 vs 10 in the ARI OTH subgroup). The number, type and severity of symptoms differed between athletes with COVID-19 and the ARI OTH subgroup, which is not surprising as there are data to show that COVID-19 is a more severe disease that is associated with specific symptoms (eg, 'altered/loss of smell and taste'), 26 greater number of symptoms and more severe symptoms. 8 The RTP in our COVID-19 subgroup is longer because many of the symptoms in the cluster associated with prolonged RTP were more common in the ARI COV subgroup.

Finally, we show that in our population of symptomatic athletes (n=84), the mean number of ARI symptoms per athlete was 9.2; the four most common symptoms were 'excessive fatigue' (73.8%), 'headache' (64.3%), 'blocked/plugged nose' (61.9%) and 'sore/scratchy throat' (52.4%) and that 20 days after the onset of symptoms, 50% of the athletes returned to their first training session. These data are of interest, but we recognise that this is a self-selected sample of athletes with ARI that we recruited during the COVID-19 pandemic and that the sample is not necessarily representative of a general population of athletes with ARI.

The main strength of our study is the novel data relating symptoms of ARI to RTP in a population of athletes with confirmed / suspected COVID-19 (only 5/45 were suspected but not confirmed). Although we compare the number, type, duration and severity of symptoms in athletes with COVID-19 to a subgroup of athletes with other ARI, we recognise that this ARI OTH subgroup could have included athletes with noninfective illness or COVID-19, despite not presenting with more 

typical symptoms of COVID-19. This is a limitation, but it was not possible to obtain data to confirm an infective illness or the specific pathogen in this subgroup. Other study limitations are that this is an observational study showing an association and does not demonstrate cause-effect, that factors other than symptom type could have influenced days to RTP, that data are self-reported and was reliant on recall and that our sample was heterogeneous with potential selection bias, that is, athletes with more severe illness completing the questionnaire. We also recognise that about a third of our study participants (all and in the two subgroups) were professional athletes, and in future studies, with a larger sample size, we can analyse this group separately. These are preliminary findings from a small sample of data collected over a short period, and this study is ongoing. In future, with a larger sample, we will be able to address many of the limitations in this study, including adjusting for possible confounders in the models. We do feel that the main practical clinical finding of the relationship between a symptom cluster and days to RTP holds for our sample of athletes with ARI and that it is important to communicate this finding at this stage of the COVID-19 pandemic. Further research is needed to determine possible differences in symptom clusters and days to RTP for other specific ARI pathogens.

In summary, we show that in our population of symptomatic athletes with ARI symptoms, a cluster consisting of 'excessive fatigue', 'chills', 'fever', 'headache', 'altered/loss sense of smell', 'chest pain/pressure', 'difficulty in breathing' and 'loss of appetite' was associated with a more prolonged RTP. The most important symptom of ARI associated with a longer RTP was 'excessive fatigue', with a 70% lower chance of RTP over the 40-day period. We show that athletes with COVID-19 presented with a greater number and more severe symptoms, compared with a subgroup of athletes with other ARI. These findings may be used by sport and exercise physicians to assist their decision making for RTP in athletes with ARI. For example, in athletes presenting with ARI, clinicians can document the type, duration and severity of symptoms at the time of acute illness and then use the symptom cluster to identify athletes that may be at higher risk for a more prolonged RTP. The data on the % chance of RTP in the 40-day period can assist clinicians to assign a relative 'weighting' to specific symptoms in the cluster; for example, the most important symptom is 'excessive fatigue'.

We will now extend our work in a larger sample size and test these preliminary findings in a predictive model of RTP. We will also refine the model by including other variables, for example age, sex, sport type, treatment modalities used and existing comorbidities.

Contributors MS: responsible for the overall content as guarantor, study concept, study planning, data collection, data interpretation, manuscript (first draft), manuscript editing and facilitating funding. NS: study planning, data collection, data cleaning, data interpretation, manuscript (first draft) and manuscript editing. CS: study planning, data collection, data interpretation and manuscript editing. KK and PSW: study planning, data collection, data interpretation, manuscript (first draft) and manuscript editing. IS: study planning, development of the data management system, data collection, data cleaning and manuscript editing. WD, JHH and MV: data interpretation and manuscript editing. EJ: study planning, data cleaning, data management, data analysis including statistical analysis, data interpretation and manuscript editing.

Funding IOC Research Centre (South Africa) (partial funding). South African Medical Research Council (partial funding, statistical analysis).

Patient consent for publication Not required.

Committee of the Faculty of Health Sciences at the University of Pretoria (REC 409/2020).

Provenance and peer review Not commissioned; externally peer reviewed.

► The most important symptom of acute respiratory illness (ARI) associated with a longer return to play (RTP) was 'excessive fatigue', with a 70% lower chance of RTP in the 40-day period after the onset of symptoms. ► We also recommend that clinicians consider the following symptom cluster as indicators of a more prolonged RTP in their athletes presenting with symptomatic ARI: 'excessive fatigue', 'chills', 'fever', 'headache', 'altered/loss sense of smell', 'Chest pain/pressure', 'difficulty in breathing' and 'loss of appetite'. ► The association between these symptoms of ARI hold for athletes with confirmed/suspected COVID-19 and athletes with other (non-COVID-19) ARI. ► Athletes with COVID-19 present with more severe disease (greater number of symptoms and higher % of moderate/ severe symptoms) compared with a subgroup of athletes with other ARI. ► Athletes with COVID-19 have a significantly longer RTP compared with a subgroup of athletes with other ARI.

How might it impact on clinical practice in the future? ► We suggest that a specific symptom cluster can be used by sport and exercise physicians to assist their decision making for RTP in athletes with ARI (including COVID-19).

Data availability statement No data are available. No additional data are available.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Sports injuries and illnesses during the London summer Olympic Games 2012

Sports injuries and illnesses during the winter Olympic Games

Common cold in team Finland during 2018 winter Olympic Games (PyeongChang): epidemiology, diagnosis including molecular point-of-care testing (POCT) and treatment

How much is too much? (Part 2) International Olympic Committee consensus statement on load in sport and risk of illness

Influence of training load on upper respiratory tract infection incidence and antigen-stimulated cytokine production

Training-related and competition-related risk factors for respiratory tract and gastrointestinal infections in elite cross-country skiers

Elite athletes travelling to international destinations >5 time zone differences from their home country have a 2-3-fold increased risk of illness

Symptom Profiles of a Convenience Sample of Patients with COVID-19 -United States

COVID-19 and multiorgan response

COVID-19 and multiorgan failure: a narrative review on potential mechanisms

Multi-Organ impairment in low-1 risk individuals with long COVID

Cardiovascular magnetic resonance findings in competitive athletes recovering from COVID-19 infection

Coronavirus disease 2019 and the athletic heart: emerging perspectives on pathology, risks, and return to play

Cardiorespiratory considerations for return-to-play in elite athletes after COVID-19 infection: a practical guide for sport and exercise medicine physicians

Real-Time tracking of self-reported symptoms to predict potential COVID-19

Symptom clusters in Covid19: a potential clinical prediction tool from the COVID symptom study APP

Key predictors of attending hospital with COVID19: an association study from the COVID symptom Tracker APP in 2,618,948 individuals

Attributes and predictors of Long-COVID: analysis of COVID cases and their symptoms collected by the Covid symptoms study APP

Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support

The REDCap Consortium: building an international community of software platform partners

The Wisconsin upper respiratory symptom survey is responsive, reliable, and valid

Sparse data bias: a problem hiding in plain sight

Multicollinearity in regression analyses conducted in epidemiologic studies

Return to play after acute infectious disease in football players

Respiratory health in athletes: facing the COVID-19 challenge

Recent smell loss is the best predictor of COVID-19: a preregistered, cross-sectional study

Li) colleagues for their willingness to assist this study group with the ongoing distribution of the link containing the survey. In some cases, colleagues have now formally joined as collaborators, following approvals by their respective institutions. We would also like to sincerely thank all the athletes for their participation in this study.This article is made freely available for use in accordance with BMJ's website terms and conditions for the duration of the covid-19 pandemic or until otherwise determined by BMJ. You may use, download and print the article for any lawful, non-commercial purpose (including text and data mining) provided that all copyright notices and trade marks are retained.