JMIR Publications

ABSTRACT

Background:

Long COVID encompasses a range of health problems that can be highly debilitating. While some research has relied on self-reported measures of symptoms and functioning, few studies have characterized symptoms in relation to behaviors and physiology measured objectively through wearable devices. This high-resolution approach may provide unique insights into presentation of LC and inform monitoring and treatment strategies.

Objective:

The primary aim of the study was to identify longitudinal patterns in physical activity, physiology, and patient reported outcomes (PROs) among Long COVID patients at a Federally Qualified Health Center in the US. The secondary aim was to identify meaningful subgroups or phenotypes within the study population and examine the overlay of PROs and symptoms in relation to physical activity characteristics of Long COVID patients.

Methods:

From March 2022 to May 2023, 172 patients with Long COVID and/or myalgic encephalomyelitis/chronic fatigue syndrome were given Fitbit Charge 5 devices and instructed to wear them continuously for up to a year. Patients also completed PRO questionnaires (PROMIS-29 and symptom questionnaires, etc.) at baseline, 3, and 6 months. Inclusion in longitudinal analysis required at least 20 hours of data per day for a minimum of 7 days. Statistical differences between patient subgroups at different time points were examined using corresponding statistical tests. Growth mixture modeling and linear mixed effects regression were used to identify patient clusters using day-level Fitbit data and assess longitudinal associations between activity levels and PROs, respectively.

Results:

Among 172 patients, 80% were female, 88% were white, 46% identified as Hispanic, 51% were unemployed, and 97% had estimated annual income below $50,000. Of these patients, 82 (48%) met valid wear criteria. Each valid wear patient provided 50.5 days of Fitbit data on average (median=42, Q1=20, Q3=70.75). The growth mixture model of daily minutes of moderate to vigorous physical activity (MVPA) identified two clusters of patients. The WHO guideline of 150 MVPA minutes per week was then used to differentiate MPVA-active and -inactive patients. At baseline, the MVPA-inactive group (n=41) reported more severe problems on physical function (P=.031), fatigue (P=.006) and dyspnea (P=.007) scales compared to the MVPA-active group (n=41). Longitudinal analysis found the inactive group showed an improved ability to participate in social roles (P=.004) and reduced intensity of sleep symptoms (P=.015) over time compared with the active group.

Conclusions:

This study underscores the complex nature and varying impact of Long COVID on individuals. The level of physical activity among our patients was higher than anticipated, perhaps reflecting inflexible demands on our low-income clinic population. Our findings offer valuable insight for healthcare providers in tailoring rehabilitation efforts with personal activity profiles and for researchers seeking to phenotype Long COVID.