This study used a sequential explanatory mixed methods approach where we conducted quantitative survey and wearable data collection for 30 days, after which qualitative interviews were completed to provide context to the quantitative outcomes. Our results are reported in line with the CONSORT (Consolidated Standards of Reporting Trials) extension for randomized pilot and feasibility trials (Multimedia Appendix 1) [22].

HCWs, including nurses, doctors, psychologists, pharmacists, radiographers, physiotherapists, and nutritionists, were recruited from 4 urban health care facilities in Nairobi, Kenya from July to August 2022 (Table 1). Study inclusion criteria were aged 18 years or older, employment as an HCW, and smartphone ownership.

Recruitment strategies have been previously described [23]. Briefly, in-person sensitization meetings were conducted in study sites. During these sessions, HCWs were informed about the study objectives, duration, and study procedures. Additional recruitment strategies included email blasts, posters, and pamphlets placed around the health care facilities, and snowballing through participants who had already registered. Those interested in participating were invited to complete an online registration form that collected names, work cadre, email addresses, and phone numbers for enrollment. To increase reach, the registration form, along with brochures containing frequently asked questions sheets and QR codes for registering, was circulated in departmental WhatsApp (Meta Platforms) groups, Continuous Medical Education meetings, and virtual sensitization meetings. In total, 356 HCWs expressed interest in the study.

Of these, 60 individuals were randomly selected and stratified by facility using a random number generator. The target sample size for the pilot was n=50; thus, 60 individuals were contacted with an anticipated participation rate of 80%. The randomly selected individuals were contacted by email and invited to enroll in the pilot study within 7 days. Participants enrolled using the MyDataHelps (CareEvolution) mobile app. Within a week, 51 individuals provided informed consent and were enrolled in the study. Qualitative exit interviews were performed at the end of the study, with a subset of participants (n=22) selected through convenience sampling. Participants were recruited from individuals who were readily available and had participated in the broader study, rather than being selected through random sampling. Potential participants were identified based on their availability and willingness to participate in follow-up qualitative interviews. Representation across professional cadres and institutions was desired; however, the availability of participants was a key factor for inclusion due to demanding schedules. Participants were contacted by phone to schedule an appointment to meet face-to-face with the researchers to review the consent form. Of the 22 who were contacted, 1 participant declined to participate in the interviews.

Following study enrollment, participants completed a baseline questionnaire that included information on sociodemographics, work environment, and the following assessments: (1) the Neuroticism, Extraversion, Openness to Experience Five-Factor Inventory to assess domains of personality in 5 areas [24]; (2) Risky Families Questionnaire to assess the degree of physical, mental, and emotional distress experienced in their homes during childhood and adolescence [25]; (3) Generalized Anxiety Disorder-7 to measure anxiety scores [26]; (4) Patient Health Questionnaire-9 to assess symptoms of depression [27]; (5) posttraumatic stress disorder (PTSD) checklist to assess presence and severity of PTSD symptoms [28]; and (6) World Health Organization Alcohol, Smoking and Substance Involvement Screening Test to assess substance use–related health risks and substance use disorders [29].

Following submission of the baseline questionnaire, participants received a Fitbit Inspire 2 (Fitbit Inc) and were instructed to wear the device continuously for the duration of the 30-day study. Participants were instructed on how to use wrist mode, verify that the sleep feature was on, and ensure syncing of the data with the mobile apps. Additionally, participants were advised to charge the device as soon as a low battery notification appeared. The battery life on Fitbit is up to 10 days on a single charge and charges to full in about 1 to 2 hours. Sleep, step count, heart rate, and mood rating were collected daily for 30 days.

The mobile app, MyDataHelps, was used to collect and integrate active (baseline questionnaire and daily mood) and passive (wearable) data. The app was available on Android and iOS operating systems, and on the web. Step-by-step instructions were provided for the web-based baseline questionnaire. The app provided a daily prompt to complete the mood rating (scale of 1-10). Wearable data were uploaded daily to the mobile app and integrated into the data collection platform through automatic synchronization from the Fitbit device to the Fitbit app on the participant’s mobile phone. To ensure the data synchronization from the Fitbit app to the MyDataHelps app, participants were required to have their mobile phone connected to the internet and Bluetooth enabled. The data synchronization process retrieved data stored on the device for up to 7 days. When gaps in data upload exceeded 3 days, the app sent a prompt to remind participants to synchronize their data.

At the end of the 30-day pilot study period, a subset of participants was interviewed by members of the research team (LK, WN, and A Aballa) at the participants’ respective health facilities. The interview guide consisted of questions on user experience, challenges, and recommendations regarding the content and technology.

The interviews were conducted by trained social scientists. Prior to the interview, the interviewers completed a 3-day training on research ethics and the interview guide. A pretest was conducted with HCWs who were not part of the study, and the interview guide was later revised to ensure the flow and clarity of the questions. The interviews took approximately 40 to 70 minutes.

Feasibility of the pilot study was evaluated using baseline questionnaire response time, data completeness, recruitment results, and study attrition. For the baseline questionnaire, data completeness was calculated based on the proportion of visible questions answered, which excludes branching questions. Data completeness was also analyzed based on the number of days participants completed the daily mood rating and uploaded any steps, heart rate (heart rate intraday count), and sleep (duration) data. The average daily step count was also reported, excluding days with zero step count. The number of days participants wore the wearable for at least 10 hours was calculated using heart rate intraday minute count, as studies have suggested this is the minimal wear time to indicate a full day of use [30]. Descriptive statistics were calculated using R (version 4.1.1; R Core Team), and summary statistics were presented as mean and SD or median and IQR for continuous data, and frequencies and percentages for categorical data.

Acceptability of the baseline questionnaire and wearables was based on user experiences provided in qualitative exit interviews.

The interviews were recorded and transcribed verbatim, and a deductive thematic analysis approach was used to analyze the data with the support of NVivo Software (version 12; QSR International). Interviewers created field notes to capture participants’ impressions and observations during the interviews. The analysis followed the key stages of thematic analysis: familiarization with the data, generating initial codes, searching for themes, reviewing and refining themes, and defining and naming themes [31]. In the initial phase, transcripts were read thoroughly, and data were coded line by line, with concepts and ideas identified. As the analysis progressed, codes were refined and grouped into potential themes, which were then reviewed and further developed to reflect deeper meanings. In the final phase, the themes were clearly defined and organized to capture the core ideas of the study based on the objectives of the study. The coding matrices and final themes were reviewed and refined by LK, A Aballa, WN, and RM to ensure the accuracy and depth of the analysis.

The study obtained approvals from Aga Khan University Institutional Scientific Ethics and Review Committee (reference number 2021/IERC-166(V2)), the University of Nairobi and the Kenyatta National Hospital Ethics Review Committee (reference number KNH-ERC/A/473), regulatory bodies including the National Commission for Science Technology and Innovation (778557) and the Nairobi County and Nairobi Metropolitan Services (SMS; reference number EOP/NMS/HS/167), and hospital administrators in each study site. The study was conducted in accordance with the Declaration of Helsinki and written and informed consent was obtained from all participants. The consent form provided information on the study objective, willingness to participate, data safety, and confidentiality. For qualitative interviews, the consent included exit interview procedures and consent to be audio recorded. All participant data were deidentified using study identifier and not individual names, and data were securely transferred to the institution server. Due to budget constraints in maintaining prepaid data packages in LMICs, participants were compensated (KSh 500 or US $3.58 equivalent) for the purchase of prepaid data packages that were needed to support timely synchronization. Qualitative participants were compensated (KSh 1000 or US $7.06 equivalent) for their time. All participants were able to keep the wearable device at the end of the study.

The participants were primarily female (70%), with 69% reporting attainment of a bachelor’s or master’s degree (Table 2). The median age of the participants was 32 (IQR 29‐36) years, and nurses (47%) were the largest health care cadre. The median duration of employment in health care was 8 (IQR 5‐12) years. The retention rate for the 30-day study duration was 98% (n=50). Characteristics of those who participated in the qualitative interviews are in Table 2 and Figure 1.

The median time to complete the baseline questionnaire was 1 (IQR 1‐2, range 1-11) day, from the date the link was received to submission. All participants (N=51) completed 100% of questions in the Risky Families Questionnaire, Generalized Anxiety Disorder-7, Patient Health Questionnaire-9, and PTSD assessments, while 92% (n=47) and 84.4% (n=42) completed all questions in the Neuroticism, Extraversion, Openness to experience Five-Factor Inventory and World Health Organization Alcohol, Smoking and Substance Involvement Screening Test assessment, respectively.

The wearable was worn for >10 hours on 63% of study days (n=50). Nearly all participants uploaded step (n=50), heart rate (n=46), and sleep (n=45) data within 5 days of receiving the wearable. On average, participants completed 58.4% of the daily mood questions over the 30 days. Data on steps, heart rate, and sleep were uploaded on 92.6%, 76.6%, and 49.8% of the days, respectively. The average number of daily steps was 8180 (SD 4872).

To further characterize variability in adherence, we examined the median (IQR) and distribution of the number of days each participant submitted data (Table 3 and Multimedia Appendix 1). Approximately half of the participants submitted heart rate (n=26) and step count (n=31) data on all 30 study days, indicating higher adherence for these 2 passively collected metrics (Multimedia Appendix 1). In contrast, submissions for mood and sleep and the number of days with >10 hours of data (ie, wear time) were more variable across participants and across days. Histograms of the number of submission days show a more dispersed pattern for these measures; however, the distributions were not strongly skewed toward a small subset of highly adherent individuals.

This pilot study supports the feasibility of deploying mental health surveys, collecting data through wearables, and integrating wearable and survey data within a single mobile platform under real-world infrastructure constraints among Kenyan HCWs. Participant retention and completion of the initial questionnaire were high, indicating strong initial engagement and comfort reporting on mental health symptoms. Daily adherence was moderate; approximately half of the participants submitted heart rate and step count data on all 30 study days, while mood, steps, and wear time (≥10 h/d) were more variable across participants and over time. Sleep data were only captured on about half of all study days. Together, these metrics suggest that while continuous daily engagement was not fully achieved, the high retention and collection of multidimensional data support that these technologies are logistically viable in similar LMIC settings. Moreover, participants reported largely positive experiences with the platform, especially appreciating the self-monitoring aspect and feedback features, further supporting the feasibility and acceptability of this approach.

Mental health research in HCWs is critical to identifying and reducing stress, burnout, and psychological harm in those who care for others, which in turn improves worker well-being and patient care. This pilot study revealed that HCWs were willing to provide sensitive mental health information, such as depression, anxiety, posttraumatic stress disorder, and substance use, within a day of enrollment. Studies in Africa have highlighted that the absence of native terminology for mental health and the existence of unwritten nondisclosure norms make it difficult to openly initiate conversations about mental health [32]. Willingness to provide sensitive information on mental health is a critical step in identifying people at risk for mental health disorders. In this pilot, the web-based mode of administration and the preenrollment sensitization process, which helped build rapport between the study team and participants while addressing their concerns and questions about data confidentiality, may have mitigated some stigma-related barriers to disclosure as other studies have suggested [33,34].

The use of wearables in mental health research to infer, monitor, and predict changes in health states over time is an emerging area. On average, studies that have used wearables to collect data among HCWs with small sample sizes or shorter duration of data collection have reported a retention rate of 83%‐98% [35,36], which aligns with the retention rate in this pilot. In our study, participants reported that motivation to know their health information was a contributor to high retention, and other studies have shown motivational support to be a contributor in reducing attrition [37]. In addition, all 51 participants completed the baseline questionnaire, a rate that was 2 times higher than a similar study conducted in the United States where the completion rate was 48% [38]. The differences could be due to the varied recruitment approaches. Unlike the US-based study that used a mainly monetary incentive-driven approach to recruit participants, our study used a multipronged recruitment approach, including increased physical contact through sensitization, where we had one-on-one discussions with participants as opposed to sharing information via email or website, which could be more impersonal reimbursement for internet connectivity, email, and phone notifications. Similar studies have reported that adopting such a comprehensive approach during recruitment of participants for web-based surveys leads to better engagement [39].

In terms of adherence to daily data collection, our findings are consistent with results from similar digital health studies in both low- and high-income settings. For example, a 5-day pilot in Uganda reported high compliance with wearable use and daily diary completion, despite challenges with comfort during sleep [40]. Similarly, a 3-week study in rural Kenya demonstrated high step and sleep data completeness, but limited heart rate data availability [41]. The wear time observed in our study (63% of days with over 10 h of usage) is within the typical range of 50%‐70% reported in free-living environments [30]. Additionally, mood surveys were completed on two-thirds of the days, which is encouraging given the known challenges of sustaining engagement in digital health monitoring. For example, the average 30-day retention rates for mobile apps in general are less than 6% [42,43]. Collectively, these findings highlight the feasibility of digital monitoring among HCWs in LMICs.

Several barriers likely contributed to moderate adherence to the daily data collection, particularly for sleep and daily mood entries. Participants cited discomfort wearing the device overnight, contributing to reduced sleep data availability. Similar issues have been documented in other wearable studies [40,41] and highlight the importance of comfort for extended wear. Technical challenges with the wearable included instances where the device was accidentally switched from wrist mode to clip mode, which turns off sleep monitoring. Participants were guided on how to check if the wearable was in wrist mode during the initial onboarding to minimize potential data loss. However, this is an important consideration for future studies that leverage wearable devices with the functionality to switch modes. Regarding daily mood, reporting may have been impacted by notification delivery failures. Push notifications relied on continuous internet access, which was not always available. Interestingly, studies reporting higher adherence to daily mood diaries have reported using SMS as opposed to push notifications [44], which may circumvent the need for Wi-Fi. Participants also reported forgetfulness, competing work demands, and survey fatigue, which are consistent with prior research indicating that survey delivery mode, timing, and user workload influence daily self-report adherence [40,45]

Despite these challenges, our findings suggest strong participant engagement. Qualitative feedback revealed largely positive perceptions about using the mobile app and wearable device. Many participants appreciated tracking their health metrics and receiving motivational messages (eg, congratulatory prompts for meeting step count goals), indicating that these features likely enhanced engagement. Future studies that use wearables in mental health research may need to consider that beyond serving as an assessment tool, the wearable itself may also influence participant behavior. Increased awareness of sleep patterns and physical activity may motivate behavior change, which may function as an unintended intervention and potentially confer mental health benefits.

The study was conducted on a small sample size, which is appropriate for a pilot study, but limits our ability to detect subgroup or temporal comparisons. In addition, the population was limited to 4 urban hospitals with an overrepresentation of nurses, which may limit the generalizability of the findings to rural areas and other cadres. In addition, the trial duration was only 1 month, which captures feasibility in the short term, but adherence patterns might change over longer periods. It remains unclear whether participants would continue using the app and wearable for an extended period of time (eg, months). Qualitative feedback was obtained from a subset of participants who completed the study; those who had more difficulties might have been less likely to participate in the interviews, potentially leading to a positive bias in qualitative results. Collectively, this pilot focused on the practical ability to deploy surveys, sustain device wear, and synchronize wearable data under real-world infrastructure constraints, rather than on deriving clinically interpretable mental health metrics such as heart rate variability or complete daily physical activity profiles, which were beyond the scope of this pilot. Future studies are warranted to better link wearable-derived measures with mental health outcomes and to evaluate clinically interpretable metrics in larger, more comprehensive studies. Despite these limitations, the study provides valuable preliminary evidence to inform the scaling of similar digital mental health studies in HCWs in LMICs.

This pilot study provides justification for the use of a mobile app and wearable in mental health research in LMICs. The high survey completion rate and study retention, as well as qualitative acceptability, reflect strong engagement and suggest that such technologies are logistically viable in similar contexts. Future studies looking to scale similar approaches should consider enhancing prompts, improving device comfort, and ensuring reliable internet connectivity or finding an alternative solution.