Inter-individual variation in objective measure of reactogenicity following COVID-19 vaccination via smartwatches and fitness bands

The COVID-19 pandemic spurred the rapid development and deployment of several vaccines, notably mRNA vaccines (Pfizer-BioNTech and Moderna) and an adenovirus-based vaccine (Janssen/Johnson & Johnson). While the overall efficacy of these vaccines is well-established, individual responses vary significantly, with breakthrough infections occurring even after complete vaccination. Currently, there's no readily available, non-invasive method to objectively assess an individual's response beyond self-reported side effects. The CDC's V-safe program, while revealing high rates of systemic side effects (69% after the second mRNA dose), relies on self-reporting, which is subjective and potentially inaccurate. The relationship between reactogenicity symptoms and immune response remains controversial, although some studies suggest correlations. This study aimed to explore the potential of using wearable sensor data to objectively measure vaccine reactogenicity, hypothesizing that subtle deviations from an individual's normal resting heart rate (RHR), sleep patterns, and activity levels could serve as digital biomarkers. The study leveraged data from the DETECT study, which collects daily wearable sensor data along with self-reported symptoms and vaccination information, to investigate the association between objective physiological changes and individual and vaccine characteristics known to influence reactogenicity, such as prior COVID-19 infection and vaccine type.

Existing literature highlights the substantial variability in individual immune responses to vaccines and the relatively high incidence of breakthrough infections, even with COVID-19 vaccines. The CDC's V-safe program, while providing valuable data on self-reported side effects, lacks objective measures of the immune response. The relationship between reported side effects (reactogenicity) and actual immune response is debated, with some studies showing correlations between reactogenicity, vaccination timing, and humoral immune response. A recent study demonstrated a connection between physiological parameters measured by a smart ring and antibody levels. These studies underscore the need for objective, scalable methods for assessing individual responses to vaccines.

This analysis used data from the Digital Engagement and Tracking for Early Control and Treatment (DETECT) study, an observational study enrolling participants who share their smartwatch data, self-reported symptoms, COVID-19 test results, and vaccination information through a smartphone app. The study included 7298 individuals who received at least one dose of an mRNA COVID-19 vaccine (Pfizer-BioNTech or Moderna). After applying exclusion criteria (inadequate data, vaccination before the official US vaccine rollout date, missing demographic data), 5674 participants were included in the RHR analysis, and a subset for activity and sleep analysis (4628 and 5691 respectively). Data from the two weeks before and after each vaccination dose were collected. Resting heart rate (RHR) was calculated using data from wearable devices (Fitbit and Apple Watch). The individual baseline RHR was determined using a decreasing exponential weighting of data from 60 days to 7 days before vaccination. The daily RHR metric was calculated as the difference between the daily RHR and the baseline RHR. Similar methods were employed for sleep (total sleep time in minutes) and activity (number of steps) metrics. Participants were categorized by gender, age (<40, 40-65, >65), vaccine type (Moderna or Pfizer-BioNTech), and prior COVID-19 infection status. Statistical analyses included two-sided t-tests for comparisons between groups and a chi-squared test to evaluate changes in frequency of observations. The Holm-Bonferroni method was used to correct for multiple hypothesis testing. A multiple linear regression model was used to adjust for confounding factors (age, gender, device, vaccine type, prior COVID-19 infection).

The average RHR significantly increased after vaccination, peaking on day 2 post-vaccination (+0.56 BPM after the first dose and +1.52 BPM after the second). The RHR returned to baseline by day 4 after the first dose and day 6 after the second. A substantial portion of participants (71% after the first dose and 76% after the second) experienced an increase in RHR. Importantly, 37% of participants showed an increase in RHR exceeding one standard deviation above their baseline after the first dose, rising to 47% after the second dose. Analysis revealed several significant associations. Women exhibited a greater RHR increase after the first dose. Individuals under 40 years of age showed a significantly higher RHR increase after the second dose compared to older individuals. Prior COVID-19 infection was associated with a significantly higher RHR increase after the first dose, but not the second. The Moderna vaccine was associated with a significantly greater RHR increase than the Pfizer-BioNTech vaccine after both doses. Regarding sleep and activity, minimal changes were observed after the first dose. After the second dose, there was a significant decrease in activity and an increase in sleep, both returning to baseline by day 2. However, correlations between changes in these behavioral metrics and RHR were weak (-0.05 and -0.01 for sleep, 0.02 and 0.01 for activity after the first and second doses respectively). Multiple regression analysis, adjusting for potential confounders, confirmed independent associations between higher RHR increases and prior COVID-19 infection (first dose only), Moderna vaccine (both doses), and female sex (first dose only), and younger age (<40) after the second dose.

This study demonstrates the feasibility of using wearable sensor data to detect objective physiological changes indicative of reactogenicity following COVID-19 vaccination. While the absolute changes in RHR are subtle, the ability to detect deviations from an individual's baseline using readily available technology is significant. The findings mirror previously reported subjective reactogenicity data, with a stronger response after the second dose (except for those with prior infection), a more pronounced effect with the Moderna vaccine, and variations based on age and sex. The observed differences in RHR responses align with known immunological characteristics of the vaccines and prior infection status. The study highlights the potential of wearable technology to identify individuals with suboptimal or exaggerated immune responses, facilitating targeted interventions such as early booster shots. Although changes in sleep and activity were observed, especially after the second dose, their correlation with RHR was weak, suggesting further investigation is needed to understand their role as reactogenicity markers.

This study demonstrates the potential of leveraging wearable sensor data to objectively measure individual responses to COVID-19 vaccination. The observed changes in resting heart rate, while subtle, provide objective evidence of reactogenicity and correlate with known factors influencing immune response. This approach offers a scalable, non-invasive method for assessing vaccination response, potentially identifying individuals needing further assessment or additional vaccination. Further research should focus on validating these findings by correlating wearable data with immunological assays to establish the true predictive power of this digital biomarker and explore the use of sleep and activity data in identifying vaccine response.

The study relies on self-reported vaccination dates and types, and the DETECT study population may not fully represent the US population. Participants needed to own and use wearable devices, potentially introducing selection bias. The analysis used daily data, excluding finer-grained intra-day information. The effect of self-treatment with anti-inflammatory or anti-pyretic medications was not accounted for. Finally, some individuals might have had undiagnosed prior COVID-19 infection, potentially affecting immune responses.