Utilizing wearable sensors for continuous and highly-sensitive monitoring of reactions to the BNT162b2 mRNA COVID-19 vaccine.

Background: Clinical trial guidelines for assessing the safety of vaccines, are primarily based on self-reported questionnaires. Despite the tremendous technological advances in recent years, objective, continuous assessment of physiological measures post-vaccination is rarely performed.
Methods: We conducted a prospective observational study during the mass vaccination campaign in Israel. 160 participants >18 years who were not previously found to be COVID-19 positive and who received the BNT162b2 COVID-19 (Pfizer BioNTech) vaccine were equipped with an FDA-approved chest-patch sensor and a dedicated mobile application. The chest-patch sensor continuously monitored 13 different cardiovascular, and hemodynamic vitals: heart rate, blood oxygen saturation, respiratory rate, systolic and diastolic blood pressure, pulse pressure, mean arterial pressure, heart rate variability, stroke volume, cardiac output, cardiac index, systemic vascular resistance and skin temperature. The mobile application collected daily self-reported questionnaires on local and systemic reactions.
Results: We identify continuous and significant changes following vaccine administration in nearly all vitals. Markedly, these changes are observed even in presumably asymptomatic participants who did not report any local or systemic reaction. Changes in vitals are more apparent at night, in younger participants, and in participants following the second vaccine dose.
Conclusion: the considerably higher sensitivity of wearable sensors can revolutionize clinical trials by enabling earlier identification of abnormal reactions with fewer subjects.

Introduction

Vaccination is widely accepted as the most prominent measure in the fight against COVID-19, posing the greatest hope for ending this major global health pandemic and related economic crisis[1,2]. Consequently, an unprecedented international effort by private and public institutions alike was directed at accelerating the traditionally lengthy vaccine-development process[3-5]. On 2 December 2020, less than a year from the pandemic outbreak, the first vaccine, BNT162b2 mRNA (Pfizer-BioNTech), was granted an Emergency Use Authorization (EUA) by the UK Medicines and Healthcare products Regulatory Agency (MHRA)[6]. This initial authorization was followed by rapid authorizations for emergency use in several countries, with the US Food and Drug Administration (FDA) among the first to do so[7]. The promising BNT162b2 vaccine was demonstrated to have 95% efficacy in preventing symptomatic COVID-19 in clinical trials[8], and 92% efficacy in a nationwide mass vaccination[9].

Safety data from a randomized, controlled trial suggests a favorable safety profile for the BNT162b2 vaccine[8]. Specifically, the local and systemic reactions reported during the first seven days after vaccination were mostly mild to moderate, with a median onset of 0–2 days after vaccine administration and a median duration of 1–2 days. The most frequently reported reactions were fatigue, headache, muscle pain, chills, joint pain, and fever. The incidence of serious adverse events was low and was similar between vaccine- and placebo-treated participants. The safety of the new vaccine over a median of two months post-vaccination was similar to that of other viral vaccines. A considerable fraction of the participants did not report any reaction or adverse event. Likewise, several other vaccine candidates, including ChAdOx1 nCoV-19 (Oxford/AstraZeneca) and mRNA-1273 (Moderna), received EUAs following similar encouraging safety results in randomized, controlled trials[10-12].

Nevertheless, concerns regarding potential adverse effects from vaccines have recently led to the suspension of the ChAdOx1 nCoV-19 vaccination campaigns in several European countries[13] and may have reinforced the public hesitancy towards COVID-19 vaccines. These concerns underscore the importance of extracting as much information as possible from clinical trials. However, to date, clinical trial guidelines for assessing the safety of vaccines, including the FDA criteria[14], are primarily based on subjective, self-reported questionnaires. Despite the tremendous technological advances in recent years, objective, continuous assessment of physiological measures post-vaccination is rarely performed.

Here, we evaluated the short-term effects of the BNT162b2 COVID-19 vaccine on physiological measures. We followed a cohort of 160 participants who received the second dose of the BNT162b2 vaccine for 96 hours, from 24 h prior vaccine administration until 72 h after the inoculation. Participants were fitted with a chest-patch sensor that monitored objectively and continuously 13 different physiological indicators. Additionally, a dedicated mobile application was used to record daily self-reported questionnaires on local and systemic reactions. We identified considerable changes in chest-patch indicators during the first 48 h post-vaccination also in this group of presumably asymptomatic participants. These measures returned to the levels observed during the day prior vaccination in both groups, further supporting the safety of the vaccine. Our findings underscore the importance of accounting for objective technological advances in clinical trials to more accurately understand the invisible impact of the vaccine on our respiratory, cardiovascular, and hemodynamic systems. Extracting as much information as possible from clinical trials, particularly during an emergency, is crucial for a more comprehensive determination of vaccine safety.

Methods

Study design and participants

Our study includes a prospective cohort of 160 participants who were not previously found to be COVID-19 positive and received the second dose of the BNT162b2 mRNA COVID-19 vaccine between 1 January 2021 and 13 March 2021. This sample size of >150 participants was chosen to ensure a sufficient amount of participants will present substantial reactions (such as fever)[8]. Specifically, as reported reactions were considerably more severe after the second dose compared to the first[8], we focused on the second dose of vaccine. Out of the 160 participants in this study, 90 (56.25%) were women and 70 (43.75%) were men. Their age ranged between 21 and 78 years, with a median age of 40. Participants were equipped with a chest-patch sensor and were monitored for a period of four days, starting one day prior to vaccine administration. In addition, participants installed a dedicated mobile application and were requested to fill in a daily questionnaire, starting one day prior to the inoculation, for a period of 15 days. For each participant, the measurements levels were compared to the levels observed on the 24-h period prior to vaccination.

In order to recruit participants and ensure they complete all the study’s requirements, we hired a professional survey company. Potential participants were recruited through advertisements in social media, online banners, and word-of-mouth. The survey company was responsible for guaranteeing the participants met the study’s requirements, in particular, that they agreed to wear the chest-patch sensor and fill in the daily questionnaires.

Participants were met in person, roughly 24 h prior to vaccination, and received a detailed explanation about the study, after which they were requested to sign an informed consent form. Then, participants were asked to complete a one-time enrollment questionnaire and install two applications on their mobile phones: an application that passively collects data from the chest-patch sensor and the PerMed application, allowing participants to fill the daily questionnaires.

To better understand the effects of the second vaccine dose on physiological measures, we also monitored 24 participants, with an identical procedure, when receiving their first vaccination dose.

The chest-patch sensor

The photoplethysmography (PPG)-based chest monitors purchased and used in this study collects the following 13 indicators of vital signs: heart rate (bpm), blood oxygen saturation (%), respiratory rate (br/min), systolic and diastolic blood pressure (mmHg), pulse pressure (mmHg), mean arterial pressure (mmHg), heart rate variability (HRV—the calculation is based on a time-domain method in which the root mean square of successive RR interval differences are measured for a segment of ten beats and are presented in %), stroke volume (mL/beat), cardiac output (L/min), cardiac index (L/min/m2), systemic vascular resistance (SVR) (dynes·sec·cm−5), and skin temperature (c). Together, these indicators of vital signs provide an accurate and comprehensive assessment of the respiratory, cardiovascular, and hemodynamic. The chest patch sensor received FDA clearance for measures of heart rate, blood oxygen saturation, and systolic and diastolic blood pressure (clearance number K190792) and CE Mark approval for all 13 measures (CE 2797) (Biobeat Technologies Ltd), (see also a technological validation study for blood pressure[15]). The sensor tracks vital signs derived from changes in the pulse contour, following calibration using an approved non-invasive, cuff-based device, and is based on Pulse Wave Transit Time (PWTT) technology, combined with pulse wave analysis (PWA) (see, for example, refs. [16,17]). To increase its validity, the device has an inherent component that prevents showing values if the movement has crossed a pre-set threshold. To the best of our knowledge, this is the only cleared wearable wireless medical-grade device to provide all these measurements. As the chest patch sensors’ battery typically lasts for 4–5 days, participants were asked to remove the chest patch sensors 3 days after vaccination. Accordingly, the sensor continuously collected data at 10-min intervals for the entire duration of the 96-h experiment.

PerMed mobile application

Developed originally to support the PerMed study[18], the PerMed mobile application allows participants to fill the daily questionnaires. The daily questionnaire we used included questions about clinical symptoms from a closed list of local and systemic reactions observed in the BNT162b2 mRNA Covid-19 clinical trial[8], with an option to add other symptoms as free text.

In order to improve the quality and reliability of the data and to ensure its continuous collection, we applied the following two measures: (1) Participants who did not complete the daily questionnaire by 7 p.m. received a notification in their mobile app to fill the questionnaire; (2) We developed a dedicated dashboard that helped us identify when participants did not fill in the daily questionnaires. Those participants were contacted by the survey company and were encouraged to cooperate better.

Statistics and reproducibility

Before analyzing the data, we performed several preprocessing steps. With regard to the daily questionnaires, in cases where participants filled in the daily questionnaire more than once on a given day, only the last entry for that day was considered, as it was reasoned that the last one likely best represented the entire day. Self-reported symptoms that were entered as free text were manually categorized. With regard to the chest-patch indicators, data were first aggregated per hour (by taking the mean value). Then, to impute missing values, we performed a linear interpolation. To validate the consistency of our findings, we split the data into two groups in a balanced fashion in terms of their age group and gender. One group contained 80%, and the other one contained the other 20%. The final analyses presented are based on the entire sample. All data and code required to reproduce the results reported in this paper are available in the GitHub repository[19].

To examine the changes in each chest-patch indicator over the 72 h post-vaccination compared with the 24-h period prior to vaccination, we performed the following steps. For each indicator, for each hour h of the 72 h post-vaccination, we calculated for each individual the mean value in the 5-h sliding window: [h−4, h−3, h−2, h−1, h]. Then, we calculated the relative change in the percentage of this value compared to the corresponding 5-h window in the day prior to vaccination. Finally, we calculated the mean value for hour h over all 160 participants, as well as the 90% confidence interval, corresponding to a significance level of 0.05 in a one-sided t-test (Fig. 1). We performed a similar analysis for the 24 participants who received the first vaccine dose (Supplementary Fig. S1).

Fig. 1

Percentage of change in chest-patch indicators compared to their levels observed on the day prior to vaccination.

Percentage of change in respiratory, cardiovascular, and physiological indicators recorded by the chest-patch sensor compared to their levels observed on the day prior vaccination: a skin temperature, b heart rate, c cardiac output, d systemic vascular resistance, e systolic blood pressure, f diastolic blood pressure, g respiratory rate, and h oxygen saturation. Mean values are depicted as solid lines, 90% confidence intervals are presented as shaded regions, and horizontal dashed lines represent no change compared to the levels observed on the day prior to vaccination. The analysis is based on participants.

Next, for each of the 3 days after the vaccination, we calculated the percentage of participants who reported new local or systemic reactions compared to their reports on the day prior vaccination (Fig. 2). For each reaction, a 90% confidence interval was calculated assuming a beta distribution, with parameter corresponding to the number of participants reporting that reaction plus one (i.e., “successes”), and parameter corresponding to the number of participants who did not report that reaction plus one (i.e., “failures”).

Fig. 2

Local and systemic reactions reported by participants through the mobile application.

Error bars represent 90% confidence intervals. The number of participants who filled in the daily questionnaire on the first-, second-, and third-day post-inoculation was n = 118, 116, and 108, respectively.

Finally, we examined the difference between symptomatic and asymptomatic participants with regard to changes in the chest-patch indicators, stratified by the number of days post-vaccination (1–3) and part of the day (day or night) (Fig. 3). For a given day post-vaccination, symptomatic participants were defined as those who reported at least one reaction on that day that they did not report the day prior vaccination. Asymptomatic individuals were defined as those who reported no reactions on that day. We defined nighttime as the time interval between 12 a.m. and 7 a.m. and daytime between 7 a.m. and 12 a.m. This day-night definition is consistent with the observed movement patterns of the participants throughout the study. For each participant, we calculated the mean indicator value for each day and part of the day post-vaccination. Then, we calculated the relative change in percentages of these values compared to their corresponding values in the day prior vaccination. Next, we calculated the mean values of symptomatic participants and asymptomatic participants, as well as their corresponding 90% confidence intervals using a t distribution. Finally, unequal variances t-tests were used to evaluate the differences between symptomatic and asymptomatic participants.

Fig. 3

Percentage of change in chest-patch indicators for participants who reported at least one local or systemic reaction and those who reported no reaction during the daytime and the nighttime.

Percentage of change in chest-patch indicators for participants who reported at least one local or systemic reaction (symptomatic, participants) and those who reported no reaction (asymptomatic, participants) during the daytime and the nighttime: a skin temperature, b heart rate, c cardiac output, and d respiratory rate. Error bars represent 90% confidence intervals. Horizontal dashed lines represent no change compared to the levels observed on the day prior to vaccination. Significant differences compared to the measurements levels observed on the day prior to vaccination at a 0.05 level are marked with *. Significant differences between symptomatic and asymptomatic participants at a 0.05 level are marked with #.

Ethical approval

Before participating in the study, all subjects were advised, both orally and in writing, as to the nature of the study and gave written informed consent to the study protocol, which was approved by the Tel Aviv University Institutional Review Board (0002522-1). All data were de-identified and no personally identifiable information was gathered.

Table 1

Characteristics of the participants.

Characteristic	Number of participants (%) (N = 160)
Gender
Male	70 (43.75%)
Female	90 (56.25%)
Age group
18–39 years	91 (56.87%)
40–59 years	52 (32.50%)
≥60 years	17 (10.63%)
Body-mass index* (kg/m2)
<30.0	128 (80.00%)
≥30.0	22 (13.75%)
Unspecified	10 (6.25%)
High blood pressure measured at the day before vaccination (Systolic >140 mmHg or Diastolic >90 mmHg)**
No	131 (8.88%)
Yes	29 (18.12%)
Household size
≤2	67 (41.87%)
>2	83 (51.88%)
Unspecified	10 (6.25%)
Self-reported illness in the past 30 days
No	104 (65.00%)
Yes	46 (28.75%)
Unspecified	10 (6.25%)
Comorbidities
No	128 (80.00%)
Hypertension	11 (6.87%)
Diabetes	7 (4.37%)
Heart disease	3 (1.87%)
Chronic lung disease	5 (3.12%)
Renal failure	1 (0.62%)
Unspecified	12 (7.50%)

* The body-mass index is the weight in kilograms divided by the square of the height in meters.

** Blood pressure was measured a day before inoculation.