Early Detection of COVID-19 in Female Athletes Using Wearable Technology

Abstract

Background:

Heart rate variability (HRV), respiratory rate (RR), and resting heart rate (RHR) are common variables measured by wrist-worn activity trackers to monitor health, fitness, and recovery in athletes. Variations in RR are observed in lower-respiratory infections, and preliminary data suggest changes in HRV and RR are linked to early detection of COVID-19 infection in nonathletes.

Hypothesis:

Wearable technology measuring HRV, RR, RHR, and recovery will be successful for early detection of COVID-19 in NCAA Division I female athletes.

Study Design:

Cohort study.

Level of Evidence:

Level 2.

Methods:

Female athletes wore WHOOP, Inc. bands through the 2020 to 2021 competitive season. Of the athletes who tested positive for COVID (n = 33), 14 had enough data to be assessed (N = 14; 20.0 ± 1.3 years; 69.8 ± 7.2 kg; 172.0 ± 8.3 cm). Roughly 2 weeks of noninfected days were used to set baseline levels of HRV, RR, recovery, and RHR to compare with -3, -2, and -1 days before a positive COVID-19 result.

Results:

Increases in RR (P = 0.02) were detected on day -3. RHR (P < 0.01) and RR increased (P < 0.01), while HRV decreased (P < 0.05) on day -1, compared with baseline. Differences were noted in all variables on the day of the positive COVID-19 result: decreased HRV (P < 0.05) and recovery scores (P < 0.01), and increased RHR (P < 0.01) and RR (P < 0.01).

Conclusion:

In female athletes, wearable technology was successful in predicting COVID-19 infection through changes in RR 3 days before a positive test, and also HRV and RHR the day before a positive test.

Clinical Relevance:

Wearable technology may be used, as part of a multifaceted approach, for the early detection of COVID-19 in elite athletes through monitoring of HRV, RR, and RHR for overall team health.

As of July 2022, the World Health Organization (WHO) has reported globally 564,126,546 confirmed cases of SARS-CoV-2 - the virus responsible for the coronavirus 2019 (COVID-19) pandemic. Of these confirmed infections, 6,371,354 have resulted in death. Hence, intensive efforts are in place to better understand its pathogenesis, viral transmission, and identification strategies. According to the Centers for Disease Control and Prevention (CDC), known symptoms include shortness of breath, tightness of chest, fever, headache, loss of taste or smell, fatigue, tachycardia, and cough, characteristic of both upper and lower respiratory infections. This is of particular importance in the sports community, as athletes have consistently reported residual symptoms weeks to months after their initial infection of COVID-19. ¹⁴ However, a recent review suggests that approximately 94% of athletes have mild-to-no symptoms, making them more likely to unknowingly infect those around them. ⁹ In addition, due to heightened risk of transmission between athletes and coaching staff, it is imperative to minimize infections to maintain health while keeping more athletes in play and off the bench, and it is important to implement an established protocol of COVID-19 detection. ⁴ Currently, there is no standard of practice for early detection of COVID-19 in athletes.

Wearable technology is commonly utilized by athletes to track their activity, sleep, and quantify recovery. However, during the COVID-19 pandemic, wearable technology also gained traction throughout the general population for its potential to identify early onset of COVID-19 infection from alterations in recovery metrics including heart rate variability (HRV), respiratory rate (RR), and resting heart rate (RHR). ⁷ Research has indicated that lower respiratory infections, such as pneumonia or COVID-19, may alter recovery metrics by increasing RR and RHR, and decreasing HRV up to 7 days before appearance of COVID-19 symptoms. ⁷ Therefore, wearable technology may play a key role in early detection of COVID-19 in athletes as well. This is especially important in the context of team sports since the ability to identify and isolate an infected athlete is critical to overall team health and, thus, the ability for the team to compete.

WHOOP bands (WHOOP Inc) are wrist-worn activity monitors that provide users with a recovery score that is generated from sleep, activity, daily strain, HRV, RHR, and RR data. Photoplethysmography is used to measure heart rate metrics while activity and sleep data are obtained through accelerometry. WHOOP bands have been validated against the gold standard of polysomnography, making them accurate wearable technology to quantify and measure sleep. ¹⁰ As a result, team performance staff often implement the use of wearable technology with their elite athletes in an effort to monitor recovery metrics. In addition, these technologies may help athletes gain better control of their training and recovery and provide them with insight on their overall health.

Of the metrics that WHOOP bands record, HRV predominates as it serves as a measure of the autonomic nervous system activity and is therefore a strong metric of physical fitness and overall health. ⁶ Trained athletes typically exhibit a higher HRV compared with nonathletic populations. ² Whereas HRV is highly individualized and can vary from day to day, a drastic decrease in HRV may indicate illness, injury, or under-recovery. Recent research has observed a significant drop in HRV throughout the week preceding a positive COVID-19 test in an adult, nonathletic population, when compared with persons without COVID-19. ⁷ While HRV can be impacted by training load, a rapid decrease in HRV coupled with a reduction in RR may indicate infection and warrant further medical observations from health professionals. ¹²

Unlike HRV or RHR, RR is a stable metric that is not easily perturbed. On average, in healthy adults, RR ranges from 12 to 20 respirations per minute (rpm). ¹² Among incidence of lower respiratory infections such as pneumonia or COVID-19, however, a rise in RR of 3 to 5 rpm is seen as a clinically significant indicator of respiratory deterioration. ¹¹ While RR may be influenced by unrelated factors, such as high altitude or excessive caffeine consumption, an increase by 17% or more has been deemed a significant variation from day-to-day measurements. Because of this, according to their website, WHOOP Inc programmed its fitness bands to flag users if their RR increases by ≥17%.

Research shows that monitoring recovery variables such as HRV and RR may serve as early predictors of COVID-19 in nonathletic populations. In addition, detection of COVID-19 has not yet been studied in athletic populations, more specifically in elite female athletes. Therefore, the purpose of this investigation was to determine the efficacy of wearable technology for early detection of COVID-19 in NCAA Division I female athletes through monitoring of HRV, RR, recovery, and RHR.

Methods

This study was approved by the institutional review board (IRB) at Florida State University to utilize data collected on student-athletes. The corresponding IRB identification number is 00000058.

NCAA Division I elite female athletes (soccer, golf, softball, indoor volleyball, beach volleyball, and tennis) were asked to wear WHOOP bands daily between August 1, 2020 and May 31, 2021 by their strength and conditioning team. Of the 113 athletes being monitored, 33 tested positive for COVID-19 with a polymerase chain reaction (PCR) test. However, only 14 of these athletes wore the WHOOP bands consistently enough to establish baseline levels of HRV, RHR, recovery, and RR. Demographic data of these 14 athletes are represented in Table 1. Athlete data were deidentified and provided to researchers by WHOOP to maintain anonymity. Strength and conditioning staff at the university observed the spike in RR roughly 2 to 3 days before a positive COVID-19 test. Based on this anecdotal data from strength and conditioning staff and based on the CDC guidelines at the time, the research team decided data 3 days before infection should be analyzed.

Table 1.

Demographic data ^a

Metric	Mean (n = 14)
Age, y	20.0 ± 1.3
Height, cm	172.0 ± 8.3
Weight, kg	69.8 ± 7.2
BMI, kg/m2	23.7 ± 3.3

BMI, body mass index.

^aAll demographic data for the female athletes included in the study. Data are represented as mean ± SD.

Throughout the 10 months WHOOP bands were worn, athletes were instructed to quarantine with onset of symptoms and then take a PCR test for COVID-19. The 2 weeks of noninfected, healthy baseline days starting between 1 and 3 months before the positive COVID-19 test were analyzed. All baseline healthy data ended ≥2 weeks before a positive COVID-19 test as previous literature suggests an incubation period of >14 days is unlikely.^1,8 These data were utilized to establish the athlete’s typical measurements for HRV, RR, recovery, and RHR, including the means and standard deviations. The standard deviation from the baseline mean for 3 days before (day -3), 2 days before (day -2), 1 day (day -1), and the positive COVID-19 test day were calculated for HRV, RR, recovery, and RHR values as follows:

$$\begin{array}{l}\left(\mathrm{Daily}\mathrm{metric-baseline}\mathrm{mean}\mathrm{metric}\right)/\\ \mathrm{Baseline}\mathrm{standard}\mathrm{deviation}=\\ \mathrm{Standard}\mathrm{deviation}\mathrm{from}\mathrm{baseline}\end{array}$$

Not all athletes wore their WHOOP bands for the 3-day period leading to the positive COVID-19 test; however, each set of data has the associated n provided.

Statistical Analyses

Analyses were conducted using R Studio (Version 3.0). All data are reported as average number of standard deviation; range of standard deviation observed, unless defined otherwise. The z-score relative to baseline was calculated for each metric on each day with the assumption that the distribution of these z-scores is approximately N(0,1), and the sample mean of n of these resulting values on a given day is distributed as approximately N(0,1/√n), which allowed for the calculation of P values.

Results

A total of 14 Division I NCAA female athletes (n = 14; mean ± SD, 20.0 ± 1.3 years; 69.8 ± 7.2 kg; 172.0 ± 8.3 cm) had data analyzed in this study. Baseline data were calculated over 12.7 ± 2.8 days for each athlete. Mean baseline HRV, RR, recovery, and RHR are represented in Table 2 and are utilized to set the “0” y axis in Figure 1. All HRV, RR, recovery, and RHR data on days -3, -2, -1, and positive COVID-19 day are represented in Figure 1.

Table 2.

Baseline data ^a

Metric	Mean	SD
RHR, bpm	59.8	4.6
HRV, ms	89.3	22.1
Recovery, AU	57.2	22.1
RR, rpm	15.6	0.4

AU, arbitrary units; HRV, heart rate variability; RHR, resting heart rate; RR, respiratory rate.

^aData are representative of the mean and SD of each metric from 12.71 ± 2.76 (mean ± SD) noninfected days. The first day of the baseline measurements was 1 month before the positive COVID-19 test.

Figure 1.

Daily variations from baseline levels days -3 (n = 13), -2 (n = 13), -1 (n = 12), and on a positive COVID-19 test day (n = 14) for (a) HRV, (b) RHR, (c) RR, and (d) recovery as measured by WHOOP fitness bands. Data are presented as median with bars denoting overall range. Dotted gray line represents ±2 SD from the predetermined baseline means of that specific metric. *P < 0.05; **P < 0.01. HRV, heart rate variability; RHR, resting heart rate; RR, respiratory rate.

At -3 days (n = 13), there was a noticeable increase in RR by 0.7 SD (-0.9 to 2.7 SD; P = 0.02). In addition, there were no meaningful changes observed in RHR (-0.03 SD; -1.9 to 4.6 SD; P = 0.90), HRV (-0.1 SD; -6.2 to 1.9 SD; P = 0.6), or recovery (-0.5 SD; -11.4 to 0.6 SD; P = 0.07).

On day -2 (n = 13), only RR was observed to significantly change from baseline values. RR increased by 1.1 SD (-1.5 to 5.7 SD; P < 0.01) while recovery approached significance with a decrease of -0.5 SD (-10.4 to 1.9 SD; P = 0.05). Neither daily average for HRV (-0.3 SD; -8.3 to 2.7 SD; P = 0.30) nor RHR (0.1 SD; -1.3 to 3.8 SD; P = 0.70) deviated from baseline values.

The day before the positive COVID-19 test (day -1; n = 12), the only metric that did not have an observable change on the overall average was recovery (0.02 SD; -1.9 to 5.2 SD; P = 0.95). RHR, HRV, and RR all had marked changes from baseline averages. RHR increased by 1.00 SD (-1.7 to 3.9 SD; P < 0.01), RR increased by 0.8 SD (-1.5 to 9.6 SD; P < 0.01), while HRV decreased by -0.7 SD (-10.2 to 3.6 SD; P < 0.05) from baseline data.

On the day of the positive COVID-19 test (n = 14), there was an apparent difference from baseline in all 4 variables. HRV and recovery decreased by -0.5 SD (-10.2 to 3.6 SD; P < 0.05) and -1.2 SD (-17.8 to 2.6 SD; P < 0.01), respectively. RHR increased by 0.7 SD (-2.7 to 8.9 SD; P < 0.01). In addition, the day of the positive test marked the greatest deviation on average from the baseline metrics with RR increasing by 3.2 SD (-3.0 to 17.5 SD; P < 0.01) from baseline.

Discussion

This study observed variations in HRV, RR, recovery, and RHR in Division I NCAA female athletes to determine whether a metric exists to help inform team sports science staff of a possible COVID-19 infection before a confirmed positive test. Based on our findings using the WHOOP band, RR may be the most reliable indicator for early detection of COVID-19, followed by RHR and HRV, with the least effective metric for being “recovery,” RR demonstrated a steady increase in deviations from baseline, beginning 3 days before a positive COVID-19 test. Conversely, RHR and HRV did not have significant changes until 1 day before. In addition, significant overall changes in recovery were not observed until the day of a positive COVID-19 test, making recovery less effective as an ‘early’ detector of infection.

Previously, HRV - measured by the amplitude of standard deviations of NN pattern (SDNN) - has been demonstrated to decrease (-4.07 ms; P < 0.01) 7 days before a positive COVID-19 test in healthy healthcare workers (NN is the normalized time between R to R intervals). ⁷ However, in the present study, this significant decrease in HRV was not observed until 1 day before a positive COVID-19 test. This is likely due to differences in populations, methods, and strain variances. Also, the study by Hirten et al ⁷ utilized Apple watches, while the current study utilized WHOOP bands.

Companies that produce wearable technology may utilize different methods of calculating metrics, such as HRV. According to their respective company websites, Apple watches utilize SDNN intervals while WHOOP averages the RR intervals between successive heart beats. Other methodological differences include when the athlete decided to have a COVID-19 test conducted. While Hirten et al ⁷ instructed participants to take a COVID-19 test when they deemed fit; athletes in the present study took a COVID-19 test 3 days after an increase in RR. This potentially led to greater consistency and earlier detection of the viral infection in our study compared with earlier work. In addition, as the COVID-19 pandemic has progressed, multiple strains have developed. Different strains of the virus have different incubation times, resulting in symptoms presenting anywhere from 2 to 14 days after initial exposure.^1,8 In addition, HRV is also a metric that can be easily influenced by outside factors from illness, including altitude exposure or alcohol consumption.^3,13 Similarly, RHR and recovery score also exhibit high variation from baseline metrics with stimuli unrelated to COVID-19 infection. Because of this, it would be inappropriate to observe changes in these metrics and assume they are indicative of serious viral infection. However, with a more stable metric like RR, it may be more appropriate to take precautions based on variations from typical measurements.

RR has low day-to-day variation (average SD for baseline, 0.39 rpm); thus, an increase can be a potential sign of serious physiological dysregulation. ⁵ Because of these characteristics, healthcare workers may choose to recognize RR as a vital marker for severe illness. ⁵ Now, in a time where COVID-19 does not always present with noticeable symptomology, it may be important for not only healthcare workers to monitor RR but also coaches and athletic training staff. Our data suggest a meaningful increase in RR from the athletes’ baseline 3 days before a positive COVID-19 test that remains elevated until the day of a positive COVID test as demonstrated in Figure 1. Using RR as the primary indicator of COVID-19 detection may be most appropriate compared with other metrics utilized by wearable devices, as it has the most consistent deviations from baseline.

Because of the lack of consistency in changes with HRV, RHR, and recovery in conjunction to daily variability observed in these metrics, the present research does not suggest using these metrics as a primary indicator for early detection of COVID-19. However, it may be meaningful to utilize elevated RR as an additive metric, since an increase in RR indicated by the technology (≥17%) is likely the best primary “alert” for coaching staff to suggest quarantining for athletes. If RR continues to rise, and if changes in HRV, RHR, and recovery are observed from what is typical of the athlete, illness is likely, and the athlete should continue to quarantine and be tested for COVID-19 roughly 3 days after the initial change in RR.

It must be noted that the sample size analyzed in this retrospective study was small as it was data on in-season athletes, causing difficulty with creating definitive scientific conclusions. However, there is great external validity of this project as it was observed in athletes during a real season without external intervention. The small sample size is also testament to the success of the method in predicting a positive COVID-19 test with the wearable technology to allow for quicker responses. In having greater physiological insight in real time, coaches were able to instruct athletes to quarantine and decreased the spread of the virus, thus keeping more athletes healthy and in play.

This is the first study in NCAA Division I Female Athletes demonstrating the utility of wearable technology for predicting the onset of COVID-19 infection before a positive clinical test in asymptomatic or presymptomatic athletes. With a high prevalence of asymptomatic cases and variable incubation periods, it has been challenging to contain COVID-19 outbreaks. ⁵ Athletes practice in proximity to each other with increased respiration, thus, transmission is highly probable. Accessibility and popularity of wearable technology provides a unique solution to this dilemma. Using these devices to continually track physiological parameters could provide an additional safeguard in conjunction with temperature checks and symptoms monitoring to prevent further spread of COVID-19 in athletic populations.

While this retrospective study does show promise with the potential to detect COVID-19, this study has significant limitations. The student athletes were not tested for other viruses, such as the flu or respiratory syncytial virus, only COVID-19. Also, athletes were not tested on days -3, -2, or -1. However, this protocol was in line with COVID-19 guidelines suggested at the time. It is possible they may have tested positive earlier. In addition, while baseline measures were taken before predicted possible COVID-19 infection, we did not control whether athletes were in or out of season, at a game or tournament, or traveling during baseline measures. Because of this, athletes may have been more fatigued during the time baseline measures were determined. Further, even though baseline data was within participants, there was no control group used to compare against each athlete. Importantly, diet, travel, and sleep could impact many of the metrics measured.

Clinical Recommendations

Performance staff should consider utilizing wearable technology and educate athletes on the importance of monitoring HRV, RR, recovery, and RHR throughout their season. Because training and daily life can easily change an athlete’s HRV, recovery, and RHR measured by wearable technology, RR may be the best metric to use as the first indicator of infection. By considering allowing an athlete to quarantine early with a high deviation from normal RR, coaching staff and athletes may monitor changes in certain variables that may be associated with infection, such as HRV. By having a standard practice of COVID-19 testing roughly 3 days after the initial variation in RR, it is less likely for athletes to go undetected with COVID-19 infection and unknowingly spread the virus. This not only allows athletes to stay healthy and decrease adverse events from the still unknown long-term effects of COVID-19 but could also potentially save the team’s season by keeping athletes in play.