Evaluating the feasibility and acceptability of using a smart watch for atrial fibrillation detection among inpatients in Uganda

Abstract Background/Introduction Atrial fibrillation (AF) affects an estimated 37.6 million people, and its prevalence is projected to increase by more than 60% by 2050. Low- and middle-income countries (LMICs) are showing rapid increases in AF prevalence. Addressing AF is particularly important in Uganda, where there is a paucity of data despite high rates of both AF’s risk factors and ischemic stroke. The tools for detecting AF are limited in Uganda, and new, wearable technologies are an exciting opportunity to improve diagnosis. Purpose Few studies have looked at the feasibility and acceptability of smart watches in LMICs. We conducted a study evaluating the feasibility and acceptability of FitBit watches as compared with Holter monitor in detection of AF in Uganda in high-risk inpatients. Methods Between Nov 4, 2024, and Jan 28, 2025, we recruited adult patients admitted with a clinical diagnosis of heart failure or ischemic stroke detected on CT to our Hospital in Uganda. Participants were monitored concurrently with Holter Monitor and FitBit Charge5 watch for 24 hours. Both presence and duration of AF were recorded. Baseline clinical information was collected on and a transthoracic echocardiogram was performed on all patients. We evaluated the feasibility and acceptability of implementing AF detection devices into clinical care in Uganda using the Proctor Model to assess key implementation, service, and client outcomes, as well as through in-depth qualitative interviews with patients and healthcare providers. Results We recruited 50 patients, of whom 30 (78%) were women, 28 (56%) had heart failure, and 22 (44%) had ischemic stroke. AF was detected in 23 (46%), with a mean duration of 650 mins (SD 123.8). Concordance between FitBit watch and Holter monitor was 52%, driven by the need for regular rhythm spot checks with FitBit since continuous rhythm monitoring is not yet available in Uganda. The System Usability Score was 63.6 (Standard Deviation SD 1.7), below the target of 70. The Acceptability of Intervention Measure, Intervention Appropriateness Measure, and Feasibility of Intervention Measure were all above the target score of 3 and were 4.6 (SD 0.1), 4.3 (SD 0.1), and 4.1 (SD 0.1), respectively. In qualitative interviews, patients and providers found watches easy to use and associated with more attentive medical care. Both groups were excited about the possibility of smart watches improving health both inside and outside the hospital, though noting logistical issues (cost, access to charging and repair) limiting its broader use. Conclusions We found the FitBit watch had high feasibility and acceptability on 3 of 4 metrics but had suboptimal AF detection when compared to a Holter monitor, likely due to logistical and country-specific limitations. Further work should focus on ensuring continuous monitoring in Uganda and conducting population-level studies to understand AF prevalence.