Seasonal migration and healthy ageing in China: protocol for the Frigid-Tropical Migratory Population Health (ftMPH) cohort Authors: Junli Guo, Jinwei Tian, Yi Han, Shijiao Yan, Cheng Jin, Xinyi Chang, Xiaohong Wu, Wuqi Song, Fan Wang, Ben Zhang, Yong Ji Published: 2026 ABSTRACT AbstractIntroductionAs populations age and climate variability intensifies, seasonal migration has emerged as a strategy among older adults to reduce exposure to climatic extremes and optimise living conditions. In China, millions of older adults migrate annually between cold northern and tropical southern regions, providing a natural setting to examine the health effects of seasonal migration. The Frigid-Tropical Migratory Population Health (ftMPH) cohort was established to assess the short- and long-term health effects of this migration and the underlying biological, behavioural and social pathways.Methods and analysisThe ftMPH cohort is a prospective, dynamic cohort jointly conducted in Heilongjiang (a cold region) and Hainan (a tropical region) in China, with a planned sample size of approximately 26 000 participants. Adults aged ≥60 years are recruited and classified into four subcohorts: cold-origin seasonal migrants, cold-region residents, tropical-origin seasonal migrants and tropical-region residents. Baseline assessments include questionnaires, clinical examinations, biospecimen collection and environmental exposure measurements. Each migration cycle is assessed at predefined time points spanning the departure and return phases. Migration exposure will be updated longitudinally at each follow-up interval.The primary outcome is the incidence of major cardiovascular disease events, including non-fatal myocardial infarction, coronary revascularisation, stroke and cardiovascular mortality. Secondary outcomes include non-cardiovascular mortality, chronic disease events, healthcare utilisation and ageing-related measures. Time-to-event analyses will be performed using Cox proportional hazards models. Inverse probability of treatment weighting (IPTW) will be used to control for baseline confounding, with stabilised weights and covariate balance assessed using standardised mean differences. Longitudinal changes in repeated measures will be analysed using mixed-effects models.Ethics and disseminationEthical approval was obtained from the Ethics Review Committee of Harbin Medical University (approval No HMUIRB2024029PRE) and the Ethics Review Committee of Hainan Medical University (approval No HYMLL-2025-102). Findings will be disseminated through peer-reviewed publications, conference presentations and policy or technical reports.Trial registration numberThe ftMPH cohort is registered with the Chinese Clinical Trial Registry (ChiCTR2500102337; registered on 13 May 2025).Keywords: Prospective studies, Cardiovascular Disease, Climate Change, Aging, GERIATRIC MEDICINE STRENGTHS AND LIMITATIONS OF THIS STUDY This study uses a prospective cohort design with harmonised protocols across climatically distinct regions. This study uses a prospective cohort design with harmonised protocols across climatically distinct regions. A repeated-measures design with four time points per migration cycle, combined with parallel migrant and resident subcohorts, enables within-individual longitudinal analysis and between-group comparisons. A repeated-measures design with four time points per migration cycle, combined with parallel migrant and resident subcohorts, enables within-individual longitudinal analysis and between-group comparisons. Migration exposure is modelled as a time-varying variable, allowing dynamic assessment of exposure–response relationships. Migration exposure is modelled as a time-varying variable, allowing dynamic assessment of exposure–response relationships. Standardised data collection, including questionnaires, clinical assessments and biospecimen sampling, ensures consistency across study sites. Standardised data collection, including questionnaires, clinical assessments and biospecimen sampling, ensures consistency across study sites. Limitations include potential loss to follow-up in a mobile population, residual confounding and healthy migrant bias due to baseline differences between groups. Limitations include potential loss to follow-up in a mobile population, residual confounding and healthy migrant bias due to baseline differences between groups. Background and rationale Seasonal migration has become an increasingly common strategy among older adults who seek relief from climatic extremes and improvements in daily living conditions.1,4 Age-related declines in thermoregulation, combined with multimorbidity and reduced physiological reserve, heighten older adults’ vulnerability to temperature variability.5,7 As the frequency of extreme weather events continues to rise, it is increasingly important to understand how seasonal migration, as a form of climate adaptation, shapes both acute and long-term health outcomes.8 9 In China, millions of older adults migrate each year between cold northern provinces and warm or tropical southern regions for several months at a time.10 This large-scale, predictable and sustained pattern of seasonal movement differs fundamentally from short-term tourism. It involves coordinated transitions across multiple domains, including ambient temperature, humidity, daylight duration, physical activity, dietary patterns, sleep rhythms, social engagement and continuity of medical care. These multidimensional changes may produce both benefits and risks. For example, warmer climates may support more outdoor activities and a better mood, whereas disruptions in medication regimens or healthcare access may worsen the control of chronic diseases.11 12 Despite the relevance and scale of this phenomenon, substantial gaps remain in the evidence base. Most existing studies rely on cross-sectional data or small samples, lack premigration and postmigration assessments, and do not capture dynamic exposure–response processes.13,19 Research conducted among North American ‘snowbird’ populations offers important preliminary observations but is limited by a short follow-up period, single-region recruitment, limited clinical detail and the absence of mechanistic or system-level data.20 Crucially, it remains unclear how repeated transitions across contrasting climate zones cumulatively reconfigure physiological regulation, behavioural adaptation and healthcare utilisation in ageing populations. China offers a unique real-world setting to address this question. Seasonal migration follows predictable timing and directional patterns, with sufficient duration to enable physiological and behavioural adaptation. The pronounced climatic contrast between Heilongjiang and Hainan makes this setting particularly well suited to quasi-experimental analytical frameworks. However, no large-scale prospective cohort has systematically characterised the short-term changes, long-term trajectories and underlying mechanisms associated with this migration pattern. To address these gaps, we established the Frigid-Tropical Migratory Population Health (ftMPH) cohort, a large-scale, multiregional prospective study of seasonal migration in older adults. The cohort combines repeated assessments across migration phases with multimodal data collection, including clinical phenotyping, imaging, biospecimens, multiomics profiling, wearable monitoring and high-resolution environmental exposure estimation. By enabling within-individual comparisons across contrasting environments and incorporating resident comparison groups, the ftMPH cohort enables the investigation of dynamic environmental, behavioural and biological processes underlying changes in health over time. The findings are expected to provide empirical evidence to inform climate-adaptive and migration-inclusive strategies for healthy ageing, contributing to the WHO Healthy Ageing framework and China’s National Climate Change Adaptation Strategy 2035.21 Methods and analysis Study design and settings The ftMPH cohort is a prospective, dynamic cohort jointly coordinated by Harbin Medical University and Hainan Medical University, with support from community health centres and local public health authorities. The cohort adopts repeated assessments across migration cycles (figure 1), enabling within-individual analyses over time with standardised measurement protocols across sites. In addition, parallel migrant and resident subcohorts are maintained to facilitate complementary between-group comparisons. This design allows evaluation of dynamic changes associated with migration while reducing between-individual confounding. Study design of the Frigid-Tropical Migratory Population Health cohort. Study sites are located in two regions with pronounced climatic contrast. Heilongjiang, in northeastern China (approximately 45–53°N), experiences long, cold winters and substantial seasonal temperature variation.22 In contrast, Hainan, in southern China (approximately 18–20°N), has a stable tropical monsoon climate.23 The marked latitudinal gradient results in substantial differences in temperature, humidity and related environmental exposures between regions. The study will be conducted in stages, beginning with a pilot phase at selected sites to establish procedures and assess feasibility, followed by expansion as data quality and study capacity are confirmed. Predefined feasibility criteria, including recruitment performance, data completeness and protocol adherence, will guide progression. Participant recruitment and baseline data collection began in May 2025, with primary follow-up expected to be completed by May 2030. Extension of follow-up may be considered, subject to additional approvals and funding. Follow-up duration will vary by enrolment date, reflecting the dynamic nature of cohort recruitment. Study population The study will enrol adults aged 60 years and above, divided into four subcohorts based on seasonal migration patterns and long-term residence: cold-origin seasonal migrants, cold-region residents, tropical-origin seasonal migrants and tropical-region residents. Detailed eligibility criteria for migrants and residents are described below. Migrant subcohorts are defined as individuals who seasonally relocate between cold northern regions (eg, Heilongjiang) and warm or tropical southern regions (eg, Hainan) for at least two consecutive months per year. To ensure sustained environmental exposure and distinguish seasonal migration from short-term travel, a minimum duration of 2 months was required.24,26 Migrants are additionally required to have maintained this pattern and to plan to continue this behaviour. Residents are defined as individuals who have lived continuously in Heilongjiang or Hainan for at least 5 years and intend to remain in the area. Participants are excluded if they have had a life-threatening cardiovascular or cerebrovascular event within the past 30 days; have advanced malignancy or end-stage organ failure; are receiving antiviral therapy or long-term antipsychotic treatment; have moderate to severe cognitive impairment that precludes participation; have an acute infectious illness; are enrolled in another interventional study; decline informed consent; or are deemed unsuitable based on clinical judgement. These criteria aim to ensure participant safety and the feasibility of longitudinal follow-up. Recruitment procedures Recruitment is conducted on a rolling basis over a multiyear period until the planned cohort size is achieved. Participants are recruited across the two study regions through healthcare institutions (including community health centres) and public outreach activities. To facilitate initial contact and improve accessibility, an online registration system collects basic demographic and residential information and enables communication with potential participants. A two-stage screening procedure is implemented. In the first stage, preliminary eligibility is assessed through telephone interviews, online questionnaires or on-site screening, focusing on age, residential history and seasonal migration patterns. In the second stage, trained staff at study centres conduct in-person confirmation of identity and migration history and provide detailed information on study objectives, procedures, potential risks and benefits, and data privacy safeguards. Sample size considerations The planned cohort will include approximately 26 000 participants, distributed across four subcohorts: approximately 5000 cold-origin seasonal migrants, 1000 tropical-origin seasonal migrants, 10 000 cold-region residents and 10 000 tropical-region residents. The smaller number of tropical-origin seasonal migrants reflects the lower prevalence of this migration pattern and the feasibility of recruitment. Based on published estimates, annual rates of cardiovascular events among older Chinese adults range from 0.8% to 1.5%, corresponding to a cumulative 5-year incidence of approximately 4%–8%.27 This incidence range is expected to yield 1040–2080 primary outcome events in the planned cohort. Sample size considerations are based on Schoenfeld’s method for Cox proportional hazards models. Assuming a two-sided α of 0.05 and 80% power, approximately 1000–1400 primary outcome events are required to detect a HR of 1.20, depending on the comparison and allocation ratio.28 The projected number of events is expected to provide adequate power for the main comparisons involving the overall migrant and resident groups; however, comparisons involving the tropical-origin migrant subgroup may be underpowered and will be considered exploratory. Baseline assessment procedures All participants will undergo a standardised baseline assessment comprising core components, with enhanced components conducted in selected subcohorts according to predefined strategies and operational feasibility. Core assessments include structured questionnaires, clinical examinations, routine laboratory testing and physiological measurements. Questionnaires capture demographic characteristics, lifestyle behaviours, medical history, medication use, psychological status, sleep patterns and self-reported environmental exposures. Clinical examinations include anthropometric measurements, blood pressure and pulse rate. Routine laboratory testing includes complete blood count, liver–renal function panels, lipid profiles, fasting glucose, hemoglobin A1c (HbA1c), and urinalysis. Physiological measurements include assessments of body composition and arterial stiffness. Enhanced assessments include imaging and ambulatory monitoring, implemented in selected subcohorts. Imaging (eg, echocardiography, vascular ultrasound and coronary artery calcium scoring using CT) will be conducted using standardised protocols, with specific components deployed according to study phase and site capacity. Ambulatory monitoring (eg, blood pressure and electrocardiographic monitoring) will be performed in predefined subcohorts where feasible. Seasonal migration follow-up protocol A central feature of the cohort is the structured monitoring of seasonal migration across repeated cycles. Each migration cycle will be assessed at four predefined time points: shortly before departure, shortly after arrival at the destination region, shortly before return and shortly after return. A set of core measurements will be collected at all migration-related visits using standardised operating protocols across study sites. At each assessment, participants will complete structured questionnaires, undergo focused physical examinations and provide blood and urine samples. Selected imaging and wearable monitoring will be conducted in predefined subcohorts, where feasible. These repeated assessments are designed to capture short-term physiological responses, behavioural changes, environmental transitions and early variations in chronic disease control associated with migration. Information on migration patterns, including timing, duration of stay and residence locations, will be collected at each follow-up using structured questionnaires and, where available, supplemented by supporting records to enhance data completeness. An annual attrition rate of approximately 3%–5% is anticipated. To minimise loss to follow-up, multiple retention strategies will be implemented, including multimodal contact, coordination with community health centres and flexible scheduling of follow-up assessments aligned with migration patterns. Migration exposure assessment These migration data will be cross-validated, where available, using supporting sources such as travel records or residence-related documentation. Predefined reconciliation rules will be applied to resolve discrepancies, prioritising objective records and ensuring consistency is maintained across repeated measurements. Based on longitudinal follow-up data across migration cycles, migration exposure will be defined as a time-varying variable updated at each follow-up interval. For primary analyses, migration exposure will be operationalised as a time-varying indicator of the current migration phase or residence location. Meanwhile, migrant status will be treated as a baseline group-defining characteristic. Secondary parameterisations will include cumulative months of migration, number of migration cycles and time since last migration, capturing variation in exposure intensity and duration across individuals and over time. This approach enables assessment of both acute and cumulative effects of migration exposure. Partial adherence to expected migration patterns (eg, irregular or shortened cycles) will be captured using continuous measures of migration duration and cycle characteristics. Environmental exposure assessment Environmental exposure data include temperature, relative humidity, air pollutants, environmental noise and surrounding green space. Meteorological data are obtained from the national meteorological monitoring network, and air pollution data from the national environmental monitoring network. Green space exposure is derived from satellite-based indices, while noise exposure is estimated from publicly available monitoring data or model-based estimates. All environmental data are spatially assigned to individuals based on geocoded residential addresses using nearest-neighbour or inverse-distance weighting. Exposures are characterised at a daily resolution (with hourly data aggregated when necessary) and aligned with individual follow-up time windows, including baseline and migration-related periods. In addition to absolute exposure levels, environmental contrast between origin and destination regions (eg, differences in temperature and humidity) will be characterised as an alternative exposure metric in secondary analyses. These contrast-based measures capture the magnitude of environmental change associated with migration and may provide additional insight into exposure–response relationships beyond static exposure levels. Despite these approaches, some degree of exposure misclassification is inevitable due to spatial variability, individual mobility and limitations in exposure modelling. Outcomes and event ascertainment The primary outcome is the incidence of major cardiovascular disease (CVD) events, including non-fatal myocardial infarction, coronary revascularisation, stroke and cardiovascular mortality, defined according to standard International Classification of Diseases (10th revision) codes and validated using clinical, imaging or procedural records where available. The primary outcome captures clinically meaningful vascular events that may be influenced by transitions across contrasting climatic environments during seasonal migration.29 Secondary outcomes encompass a broad range of health domains beyond major CVD events. These include non-cardiovascular mortality and chronic disease events, enabling assessment of morbidity and mortality unrelated to cardiovascular causes. Healthcare utilisation will be assessed through outpatient visits, hospital admissions, medication changes and associated medical expenditures. Health-related quality of life will be evaluated using validated instruments to capture changes in subjective well-being and functional capacity. Indicators of functional and biological ageing, including frailty, sarcopenia, cognitive performance and molecular ageing markers, will be collected to characterise ageing trajectories over time. Additionally, subclinical indicators derived from imaging, physiological measurements and laboratory tests will be used to detect early pathophysiological changes that may precede the onset of clinical disease. Outcome ascertainment will follow a predefined hierarchy of data sources, including electronic health records, health insurance claims, hospital discharge summaries and death registries, with linkage established within study regions subject to data availability and regulatory approvals. Where cross-provincial linkage is not feasible, outcome ascertainment will be supplemented by participant follow-up and local health system records to minimise misclassification and incomplete event capture. When multiple sources report the same event, hospital records with imaging or laboratory confirmation will serve as the reference standard. Event dates will be defined as the earliest clinically verified occurrence. Recurrent events of the same type will be recorded but excluded from primary analyses and examined in sensitivity analyses. An independent adjudication committee, blinded to migration status, will review all fatal events and cases with incomplete or conflicting information to ensure accuracy and comparability across study sites. All time-varying exposure variables will be aligned with corresponding follow-up intervals and linked to subsequent outcome assessments to ensure correct temporal ordering and minimise reverse causation. Biospecimen and multiomics analyses Biospecimens, including serum, plasma, buffy coat, urine, stool, hair and nail samples, as well as peripheral blood mononuclear cells from a subset of participants, are processed using harmonised protocols and stored in controlled biobank facilities for future biomarker and multiomics research. Multiomics analyses, including genomics, epigenomics, metabolomics and microbiome profiling, will be conducted in nested subcohorts. Sample selection will be guided by predefined case–control or case–cohort strategies, depending on the study objectives and the availability of biospecimens. Omics datasets will undergo appropriate preprocessing, dimension reduction and pathway-level integrative analyses to identify molecular signatures associated with migration-related exposures and disease risk. Statistical analysis Analyses will include time-to-event, longitudinal and causal inference approaches and will be conducted using R (V.4.2 or later) or SAS (V.9.4 or later). All statistical tests will be two-sided, with a significance level of 0.05. Time-varying migration exposure will be incorporated into regression models as updated covariates to capture temporal changes. Secondary exposure metrics, including cumulative duration of migration, number of migration cycles and time since last migration, will be used to examine dose–response relationships. The primary outcome will be analysed using Cox proportional hazards models, adjusting for baseline demographic, socioeconomic (eg, education, income and health insurance status), behavioural and clinical characteristics, with additional time-varying covariates included as appropriate. Time-to-event analyses will account for staggered entry using delayed-entry (left-truncation), with time since study entry as the primary time scale. Competing risks (eg, non-cardiovascular death) will be addressed using Fine and Gray subdistribution hazard models. To address confounding and selection bias, propensity score-based inverse probability of treatment weighting will be applied in weighted regression models to improve covariate balance between migrant and resident groups. In secondary analyses, marginal structural models with stabilised inverse probability weights, incorporating both treatment and censoring processes, will be used to account for time-varying confounding and migration-related selection bias. Attrition bias will be assessed by comparing baseline characteristics and key covariates between participants retained in follow-up and those lost to follow-up. Repeated measures (eg, blood pressure, metabolic biomarkers and psychological indicators) will be analysed using linear and generalised additive mixed-effects models to characterise within-individual longitudinal changes over time. In complementary analyses, comparisons between migrant and resident subcohorts will be conducted using Cox proportional hazards models and mixed-effects models, with propensity score-based methods applied to control for confounding. Sensitivity analyses will assess robustness to alternative exposure definitions, lagged exposure specifications and different time scales (eg, using age as the underlying time scale). Missing data will be handled using multiple imputation under the missing-at-random assumption, with additional analyses (eg, pattern-mixture and scenario-based approaches) to assess potential departures from this assumption. Robust standard errors will be used to account for clustering at the study site or community level. Patient and public involvement Older adults from participating communities provided feedback on recruitment procedures, questionnaire modules and communication strategies, informing the feasibility of repeated assessments and preferences for follow-up contact. Participants were not involved in selecting outcomes or in the analytical design. Study findings will be disseminated through newsletters, informational sessions and online updates. Ethics and dissemination Ethics approval and consent The study has been approved by the Ethics Review Committee of Harbin Medical University (approval No HMUIRB2024029PRE) and the Ethics Review Committee of Hainan Medical University (approval No HYMLL-2025-102). Written informed consent is obtained from all participants, or their legal representatives where appropriate, before participation. The study is conducted in accordance with institutional, national and international guidelines for research involving human participants. Data management and confidentiality Participant information and biological samples are handled in accordance with strict confidentiality protocols to protect personal data. Identifiers are stored separately from research datasets, and analytical datasets are deidentified before use. Data are stored on secure, access-controlled servers, and biospecimens are maintained in accordance with approved biobank governance procedures. Access to data is restricted to authorised personnel and monitored to ensure compliance with regulatory requirements. Data sharing and dissemination strategy Deidentified data will be curated in a secure, access-controlled environment. Researchers may request access through a formal review process overseen by the ftMPH Cohort Data Access Committee. Requests will be evaluated based on scientific merit, ethical compliance and consistency with participant consent. Study findings will be reported through peer-reviewed publications, conference presentations, policy briefs and public-facing communication platforms. Analytical code and aggregated results will be made available on reasonable request after completion of primary analyses. Discussion The ftMPH cohort enables examination of how older adults respond to seasonal migration across biological, behavioural and health system dimensions. By treating migration as a recurrent cycle rather than a single relocation, the study captures both short-term physiological responses and longer-term trajectories of ageing. Transitions between contrasting climatic regions allow assessment of responses to changes in environmental factors, which may be particularly relevant for cardiovascular risk and disease progression. Modelling migration as a time-varying exposure enables evaluation of cumulative effects across cycles. This design also allows examination of the temporal structure of exposure–response relationships across migration phases and may improve causal interpretation by enabling within-individual comparisons over time. The four-stage assessment design allows characterisation of the timing and reversibility of physiological responses to climatic transitions. Acute changes may occur in vascular tone, autonomic activity, sodium balance and inflammatory processes, whereas metabolic and molecular adaptations may develop over longer periods or accumulate across repeated cycles.30,33 Integration of imaging, biospecimens and wearable-derived measures enables examination of pathways linking environmental exposures to cardiometabolic and ageing-related outcomes.34 35 Seasonal migration also alters behaviours that may modify environmental effects.36 37 Changes in physical activity, diet, medication routines and sleep patterns vary across individuals and may influence physiological responses. Repeated measurements within migration cycles allow assessment of these behavioural adaptations and their interaction with biological processes over time, including their temporal alignment with environmental exposures. Migration introduces variation in healthcare access and continuity of care across regions. Cross-provincial movement exposes older adults to differences in medical record systems, medication availability, reimbursement policies and access to specialist services.38 By linking clinical assessments with electronic health records, insurance claims and healthcare utilisation data, the ftMPH cohort allows evaluation of how system fragmentation affects chronic disease management and whether disruptions in care continuity are associated with increased physiological vulnerability or acute-care use. These findings may inform approaches to improve mobility-inclusive models of chronic disease management, including interoperable electronic records, portable medication plans and telehealth-enabled follow-up strategies. As climate variability increases, older adults who are unable to migrate or lack the socioeconomic or functional capacity to do so may be disproportionately exposed to climatic stressors.39 40 Inclusion of both migrant and resident subcohorts allows comparison of adaptive and non-adaptive patterns and identification of vulnerable groups that may require targeted prevention or community support. The study’s digital follow-up infrastructure and integration of high-resolution environmental data may also inform strategies to strengthen system resilience in ageing populations facing climate-sensitive health risks. Findings from the ftMPH cohort may be relevant beyond China, particularly in settings undergoing rapid demographic and environmental change and other contexts with climate-related population mobility. Several limitations should be considered. Loss to follow-up in this mobile population may introduce attrition-related selection bias, which is mitigated through retention strategies and appropriate analytical approaches. Measurement error may affect behavioural and environmental exposures despite high spatiotemporal resolution. Migrants may be healthier and more socioeconomically advantaged than residents (the ‘healthy migrant’ effect), potentially introducing baseline selection bias; this is addressed through adjustment for baseline characteristics and causal inference approaches, although residual confounding cannot be excluded. In summary, the ftMPH cohort conceptualises seasonal migration as a recurrent, real-world environmental transition shaping health in older adults. By integrating longitudinal clinical and environmental data within a multidomain framework, the study is positioned to elucidate the dynamic links between climate variability, mobility and healthy ageing. The findings may inform the development of climate-adaptive and mobility-inclusive strategies for ageing populations in China and beyond. Acknowledgements We gratefully acknowledge all study participants, collaborating communities and healthcare institutions for their invaluable contributions to this research. Footnotes Funding: This work was supported by the Central Government Funds for Guiding Local Science and Technology Development in Heilongjiang Province (grant number ZY04JD01). Prepublication history for this paper is available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2025-115657). Patient consent for publication: Not applicable. Provenance and peer review: Not commissioned; externally peer reviewed. Patient and public involvement: Patients and/or the public were involved in the design, conduct, reporting or dissemination plans of this research. Refer to the Methods section for further details. References