Climate change poses an unprecedented threat to planetary and human health, with 2024 identified as the warmest year on record, surpassing preindustrial temperatures by 1.55°C.1 This substantial increase, marked by a consistent rise of 0.2°C per decade,2 aligns with IPCC projections predicting that global warming will reach the critical 1.5°C threshold by the early 2030s.3 Without substantial mitigation, warming could reach 2.7°C by 2100, severely impacting human health, agriculture, economies and ecosystems.4 5

While climate change is a global issue, its impacts are disproportionately severe in sub-Saharan Africa (SSA) due to the region’s limited adaptive capacity and heightened vulnerability to climate variability.6 Despite contributing minimally to global greenhouse gas emissions, SSA accounts for 34% of climate-related global disability-adjusted life-years.7 According to the IPCC’s Sixth Assessment Report, SSA will experience pronounced temperature increases, particularly in Eastern and Western Africa.8–11 Kenya, within the Greater Horn of Africa, exemplifies this vulnerability despite contributing less than 0.1% of global emissions.12 13

SSA disproportionately carries the global HIV burden, hosting approximately 67% of the 38.4 million people living with HIV (PLWH) worldwide in 2021.14 Despite recent declines in HIV incidence rates, HIV prevalence in Kenya remains high, estimated at 3.7% among adults aged 15–49 in 2022,15 16 with significant geographic variation across counties; Kisumu, Nairobi, Homabay and Siaya have the highest numbers of PLWH.17 Amid other global health priorities such as COVID-19, reduced international funding threatens the sustainability of HIV prevention and management programmes in SSA.

The intersection between HIV and climate change in SSA constitutes a syndemic, where interacting health threats synergistically heighten disease burdens and vulnerabilities among PLWH.18 Despite this critical public health issue, original research explicitly addressing these interactions remains limited, with current literature predominantly offering conceptual frameworks.18–21 Talman et al described this bidirectional interaction through a syndemic model highlighting vulnerabilities driven by poverty, food and water insecurity, gender inequality, migration and conflict.18 Lieber et al proposed pathways linking climate change and HIV through food insecurity, migration, infrastructure erosion and opportunistic infections,19 while Guinto et al expanded this framework to emphasise climate adaptation strategies relevant to the Philippines.20 Furthermore, Orievulu et al systematically reviewed drought-related impacts on HIV treatment adherence.21

Meanwhile, empirical research in SSA primarily focuses on the impact of droughts on various HIV outcomes, such as prevalence and sexual behaviour,22 testing and transmission,23 treatment outcomes and adherence.24 25 Moreover, a study in Kenya qualitatively explored the pathways connecting climate change and mental and emotional well-being.26

Several key gaps remain: geographical coverage is uneven, with some regions underrepresented; the impact of other extreme weather events, such as heat stress, floods and storms on HIV is not well explored; mechanisms through which these events affect HIV, including changes in health infrastructure, are understudied; longitudinal studies are scarce; and there is a need for research on effective policy responses and adaptation strategies. Moreover, biological mechanisms through which HIV increases vulnerability to heat stress are unexplored.

Technological innovations have greatly enhanced medical research, yet the potential of health data from these technologies remains underused, particularly in data-scarce settings like SSA.27–29 Consumer-grade wearable devices, or wearables, provide non-invasive, objective and continuous monitoring of individual-level health parameters, facilitating real-world health research through ecological momentary assessment.29 By correlating continuous wearable-generated health data with environmental data, researchers can effectively examine the impacts of climate change-weather changes on health outcomes.29 Longitudinal studies in the Nouna (Burkina Faso) and Siaya (Kenya) Health and Demographic Surveillance Systems (HDSS) demonstrated feasibility and acceptability of wearables in rural settings.30–33 In addition, wearables have previously identified associations between heat stress and decreased activity levels and sleep quality,34 35 but they have also been used to capture health metrics in HIV populations.36–38

To date, no studies have comprehensively assessed the real-time physiological effects of extreme heat and precipitation among PLWH in SSA using continuous, individual-level data. This study addresses that gap by employing wearable technology in a longitudinal cohort, paired with localised weather data, to explore how weather extremes affect heart rate (HR), physical activity and sleep in PLWH compared with HIV-negative individuals.

This study aims to investigate the relationship between weather impacts which are projected to increase due to climate change in the region, specifically heat and rainfall, and health parameters (ie, HR, step count, sleep) among PLWH in rural Kenya using consumer-grade wearable devices (see table 1, WEATHear study). Furthermore, we adopt a mixed-methods approach to examine the intertwining factors contributing to increased vulnerability in this population (see table 1, PERCePt-HIV study). The longitudinal quantitative data from the WEATHear study and the cross-sectional quantitative and qualitative findings from the PERCePt-HIV study are designed to complement each other methodologically and conceptually, offering a more complete understanding of how weather extremes affect health outcomes in PLWH. Specific objectives are detailed in table 1.

The study is situated within the Kenya Medical Research Institute (KEMRI)/Centers for Disease Control and Prevention (CDC) HDSS in rural western Kenya. The surveillance area spans three largely rural districts—Gem, Karemo and Asembo—lying northeast of Lake Victoria in Nyanza Province. The population is predominantly from the Luo ethnic group, who depend on subsistence farming and local trade for their livelihoods. Residents typically live in mud, brick or cement houses, with roofs of iron sheets or grass (figure 1). Siaya’s climate is warm year-round with a bimodal rainfall pattern: heavier, more reliable ‘long rains’ from January to June and lighter, less uniform ‘short rains’ from July to December.39 The Siaya HDSS gathers comprehensive and regularly updated sociodemographic and health-related data, including disease prevalence, birth and death rates, immunisations and healthcare utilisation. Despite a significant decline in the annual number of new HIV cases, Siaya County still has one of the highest burdens of HIV in Kenya, with a prevalence of 15.3% among adults aged 15–49.40

We actively collaborated with Siaya County stakeholders through multiple community meetings to codevelop our research question, shape the study design and ensure culturally sensitive methods. Survey and in-depth interview (IDI) questions were codesigned with the local Kenyan team and piloted with participants to confirm clarity and alignment with intended purposes. Prior to recruitment, trained community health workers conducted village-level barazas (public forums) in Luo, incorporating live device demonstrations to clearly communicate data collection processes, privacy protections and participant rights. On study completion, findings will be disseminated through co-hosted workshops and tailored, illustrated policy briefs for county health officials, ensuring results are accessible, relevant and actionable for community-driven climate-health adaptations.

Given the interplay between weather exposures and the health and activity of PLWH, this study employs a convergent mixed-methods approach (QUAL+QUAN) to comprehensively explore this relationship. We combine two complementary components: the WEATHear study, which provides longitudinal quantitative data from wearable device logs, and the PERCePt-HIV study, which adds cross-sectional quantitative data from Likert-scaled questionnaires and qualitative insights from semistructured IDIs. Together, these components will allow us to capture both the objective, physiological effects of weather extremes and the subjective perceptions and lived experiences of PLWH.

Participants were recruited if they were 18 years or older, registered within the Siaya HDSS, live within a 10 km radius of the established weather stations, and consent to participate in the study for the 6-month period. Based on their HIV status, participants were assigned to either the HIV-positive group or the HIV-negative control group. The HIV-positive group required (1) a documented positive HIV test and (2) stable HIV antiretroviral treatment for at least 12 weeks before enrolment, while the control group required a confirmed negative HIV status.

Exclusion criteria include (1) inability to be physically active without an assistive device (ie, wheelchair, walker, cane), (2) recent history of a limb fracture, (3) pregnancy or having a baby less than 6 months old (4) hospitalisation for alcoholism or drug use within the past year, or (5) plans to migrate or change residence during the study period and (6) refusal to agree to and sign the informed consent.

For the WEATHear study, we selected average daily step count as our primary endpoint based on preliminary surveillance data from the Siaya HDSS.33 Using G*Power, we modelled a two‐tailed test (α=0.05) with 80% power to detect a small‐to‐moderate effect (Cohen’s d=0.30) and assumed an intraparticipant correlation of r=0.50 between seasons. This calculation yielded a requirement of 92 participants per arm (HIV-positive vs HIV-negative) for a 1:1 matched design. To mitigate potential attrition and loss to follow-up, we increased the sample size to 200 participants, consisting of 100 HIV+participants and 100 HIV− controls.

In the PERCePt-HIV study, all 200 enrolled participants completed the climate–HIV perception questionnaire (see online supplemental material S1). A purposive subsample of HIV-positive participants was then invited for IDIs. We employed the established ‘10+3’ saturation strategy—conducting an initial batch of 10 interviews and then adding three at a time until thematic saturation was reached.41 This approach balances feasibility with the need to capture the full range of participant experiences (see online supplemental material S2).

We employed a proportional stratified sampling technique based on age and gender to ensure our study population accurately represents all age groups in Siaya county. This method enhances the accuracy of group-specific estimates, reduces recruitment bias and facilitates the examination of variations in outcomes across subgroups.

Stratification by age (18–64 years and ≥65 years) and gender (1:1 ratio) was chosen because these variables are consistently identified in the literature as significant modifiers of physiological and behavioural responses to weather exposure and HIV-related health outcomes.42 43 According to the 2019 Kenya Population and Housing Census,44 88% of the adult population in Siaya County is aged 18–64 while 12% are ≥65 years. With a sample size of 200 participants, evenly divided between target and control groups, we recruited 88 participants from each group aged 18–64 and 12 participants aged ≥65. To ensure gender balance, we recruited 44 female and 44 male participants aged 18–64, and 6 females and 6 males aged ≥65 (figure 2).

Additional covariates such as socioeconomic status, occupation or underlying health conditions were not included in the stratification due to the lack of comprehensive data within the census, limiting their utility for generalisation at the population level. However, these covariates will be considered and adjusted for in subsequent analyses as potential confounders. For the IDIs, we employed a purposive sampling technique to recruit a subset of 13 study participants from the initial study population.

Study participants were recruited from the Akala healthcare facility within the Siaya HDSS. This health facility was chosen after a geographic information system-based mapping of all Siaya County clinics and the five study weather stations identified five facilities within a 5 km radius of a station (see figure 5). We further shortlisted sites with a high volume of antiretroviral therapy patients (>800 on treatment) to ensure efficient enrolment. Ultimately, we selected the Akala healthcare facility based on administrative ease, availability of trained staff and past collaborative experiences.

HIV testing is offered free-of-charge to all patients at this facility as part of routine health services, irrespective of the visit’s purpose. The HIV-positive group comprises individuals with documented positive HIV test results, while the HIV-negative control group includes individuals with negative HIV test results. A study nurse and trained field workers at the health facility approached potential participants to screen for eligibility and consent. After screening for eligibility, they provided potential participants with information leaflets and study details. Interested individuals gave written informed consent (figure 6). Recruitment continued until the target number of subjects in both groups is reached, with privacy and confidentiality maintained throughout.

Given potential resource constraints for regular monitoring, participants will be analysed according to their original group assignment using an intention-to-treat analysis approach. This method ensures that participants are analysed based on their initial HIV status at enrolment, even if an HIV-negative participant tests positive during the study. This approach is justified by the low HIV incidence rate in the region (2 new infections per 1000 person-years in 2022), which minimises the impact of changes in HIV status on the overall study findings.

After consenting and signing the informed consent form, field workers visited the participants in their houses (figure 7). Participants received a Garmin Vivosmart 5 wearable device with instructions on application, adjustment and usage. Participants were asked to consistently wear the device on their non-dominant wrist around the clock, throughout the study period. Fieldworkers visited weekly to synchronise data via Bluetooth or cable connection, recharge the devices and perform routine checks.

Primary data collection—6 months of continuous wearable and environmental measurements—was completed in December 2024. As of 12 May 2025, we are merging supplementary datasets, notably longitudinal demographic records from the Siaya HDSS, with the primary database; consequently, formal analyses have not yet begun, and no results have been generated.

We securely store the data in a cloud-based system that is accessible only to study investigators. Wearable-device files are uploaded via encrypted transfer to a password-restricted institutional cloud, where they are immediately pseudonymised; the reidentification key is stored offline solely by the principal investigators, retained for one follow-up contact roughly 5 years postparticipation and destroyed after 10 years. Access to the deidentified dataset is limited—under written data-sharing agreements—to authorised researchers at KEMRI and the Heidelberg Institute of Global Health, while fieldworkers remain blinded to participants’ HIV status throughout recruitment and monitoring to minimise bias and protect confidentiality. Participants are explicitly informed in the consent document of these safeguards and of their rights to inspect, correct, delete or withdraw their data at any time.

We will perform statistical analysis using the open-source statistical software R (The R Foundation for Statistical Computing, Vienna, Austria).48

The expected findings of this study carry important public health and policy implications. By quantifying how climate-related stressors affect PLWH in real time, our study will generate evidence to inform climate-resilient HIV care strategies. Extreme heat and rainfall events are increasingly recognised as threats that could disrupt HIV services and daily self-management, potentially undermining progress toward the UNAIDS 95-95-95 targets if left unmitigated.69 70 For example, weather-driven increased cardiovascular strain, poor sleep or reduced mobility might impair medication adherence or clinic attendance, thereby jeopardising treatment continuity and viral suppression. Recognising this, global HIV programmes have begun calling for the integration of HIV services with climate adaptation efforts.70 Our hypothesised results will bolster this agenda by identifying specific pathways through which climate extremes affect the health and well-being of PLWH. Ultimately, evidence on the links between weather and health in PLWH can guide targeted interventions (eg, heat alert systems, enhanced counselling on hydration and rest during heatwaves, or infrastructure to maintain access to care during heavy rains) that improve resilience of both patients and health systems.

Future work could replicate this mixed-methods design in other climate zones and vulnerable groups to test generalisability and capture a broader spectrum of weather hazards. Follow-up studies should pair consumer devices with research-grade wearables—such as multisensor patches or core-temperature loggers—to obtain finer-grained physiological data and validate our findings. Embedded trials could then evaluate low-cost cooling or shelter interventions in real time, providing actionable evidence for policy.

This study aims to understand the impact of climate change on the health of PLWH in SSA, acknowledging several limitations. First, the recruitment strategy, designed to protect participant privacy, may exclude those not seeking medical care, introducing potential bias. Future studies should explore ethically sound household-level recruitment to address this gap. Second, while wearable devices enable continuous, objective data collection, consumer-grade devices such as the Garmin Vivosmart 5 have inherent limitations compared with medical-grade tools. For instance, optical HR sensors may exhibit reduced accuracy during high-intensity activity or for individuals with darker skin tones due to signal interference,71 72 and sleep staging algorithms may lack the precision of polysomnography.73 These technical constraints could affect the granularity and reliability of physiological measurements. Third, self-reported data from questionnaires and interviews may be subject to recall bias or social desirability bias, as participants might underreport challenges or overstate adaptive behaviours. Fourth, generalisability beyond rural Siaya County may be limited due to the study’s geographic and demographic specificity (eg, high HIV prevalence, Luo ethnic majority, subsistence farming livelihoods). Extrapolating results to urban settings, regions with distinct climatic conditions or populations with lower HIV prevalence requires further validation. Finally, while we address missing wearable data through statistical imputation and weekly monitoring, intermittent device non-compliance or signal loss in low-resource settings may still introduce noise. Despite these limitations, the study provides novel insights into the climate-HIV syndemic, leveraging wearable technology to advance ecological momentary assessment in understudied populations. Since all participants live in the same district, exposure to weather variables will be consistent, reducing bias.

Climate change and the HIV epidemic, two major public health threats, exacerbate vulnerabilities in PLWH, especially in SSA. This study proposes a novel method to assess the impact of weather conditions, such as heat and rainfall, on the health parameters of PLWH using consumer-grade wearable devices, allowing for ecological momentary assessments. Additionally, a mixed-method approach explores PLWH perceptions of climate change on their health. Objective data on how weather impacts health in PLWH is scarce but essential for developing targeted interventions.

Not applicable.