Protocol for a randomised, double-blind trial of a chronotherapeutic mobile health (mHealth) behaviour change intervention to optimise light exposure among older adults aged ≥ 60 years in Singapore (LightSPAN)

TL;DR

This study tests a mobile app to help older adults in Singapore improve their light exposure, aiming to enhance sleep, mood, and overall health.

Contribution

The study introduces a novel mHealth intervention to optimize light exposure for healthy aging in older adults.

Findings

The LightUP app will be evaluated for its effectiveness in increasing light exposure among older adults.

The study will assess the impact of optimized light exposure on sleep, mood, and cognitive function.

Data and code will be openly shared to support transparency and reproducibility.

Abstract

Suboptimal light exposure among older adults can exacerbate circadian disruption, sleep disturbances, mood disorders, cognitive decline, and frailty. In urbanised environments like Singapore, older adults are particularly vulnerable due to lifestyle and built environment constraints. The LightSPAN study evaluates a chronotherapeutic mobile health (mHealth) behaviour change intervention, delivered via the LightUP app, designed to optimise light exposure patterns and support healthy ageing.

The trial is a community-based, double-blind, parallel-group randomised controlled trial involving approximately 90 community-dwelling older adults (≥ 60 years) recruited through Active Ageing Centres in Singapore. Participants will be randomised using a secure web-based system with stratification by age, sex and recruitment site to receive either the active LightUP app with behaviour change features (goal setting, self-monitoring, personalised feedback) or a placebo version without these features. The primary outcome is daily time above 250 lx melanopic equivalent daylight illuminance (melanopic EDI) during daytime, measured with the ActLumus light logger across baseline, intervention and follow-up periods. Secondary outcomes include sleep parameters and circadian rest-activity rhythms, sleep quality, mood, cognition, frailty status, physical activity, body composition and vitamin D levels. The usability and acceptability of the LightUP app and wearable devices will also be assessed. Analyses will follow the intention-to-treat principle, supplemented by per-protocol analyses. Generalised linear mixed models will be used for repeated measures, with multiple imputation for missing data and exploratory Bayesian analyses. All analyses will be pre-registered on the Open Science Framework and de-identified datasets and code will be openly shared under a CC-BY license.

Despite its central role in circadian regulation, light exposure has rarely been targeted in behavioural preventive interventions for older adults. This study evaluates a novel mHealth-enabled approach that empowers older adults to optimise their light exposure, extending the scope of non-pharmacological strategies for preventive medicine and health promotion.

This study was retrospectively registered at the ISRCTN – The UK’s Clinical Study Registry on 05 September 2025 (ID: ISRCTN12391932, 10.1186/ISRCTN12391932).

The online version contains supplementary material available at 10.1186/s12877-026-07105-6.

Full text

Background

Light is the primary zeitgeber for the human circadian clocks, regulating sleep-wake timing, physiology and health [1]. Inadequate or poorly timed light exposure is associated with circadian disruption, impaired sleep, mood disturbances, cognitive decline and metabolic dysregulation [2–4]. Older adults are particularly vulnerable, as ageing is associated with reduced ocular light transmission, decreased outdoor activity and comorbidities that reduce daily light exposure [5–11].

Evidence suggests that improving light exposure can stabilise circadian rhythms and support sleep and wellbeing in ageing populations, including patients with dementia [12–14]. However, conventional interventions such as bright light therapy boxes, while effective in controlled trials, often face barriers to real-world use: they require active engagement, can be costly and show low adherence [15]. Recent work has highlighted the need for scalable, accessible and user-friendly strategies that deliver sustained behavioural change in light exposure [16].

Mobile health (mHealth) interventions offer a promising alternative. Smartphones can provide personalised feedback, encourage self-monitoring and deliver behaviour change techniques. While mHealth approaches have been widely studied for physical activity, diet and chronic disease management, little evidence exists on their role in modifying light exposure behaviour [16, 17]. Integrating wearable light sensors with mHealth interventions could provide a scalable, data-driven solution to optimise daily light exposure in naturalistic settings.

The LightSPAN study evaluates a chronotherapeutic, mHealth-enabled behaviour change intervention aimed at optimising daily light exposure patterns and supporting circadian health, to facilitate and promote healthy ageing among community-dwelling older adults in Singapore.

Methods

Objectives

The general objective of this protocol is to evaluate the effectiveness of the LightSPAN mHealth intervention in optimising light exposure behaviour and improving physiological outcomes among community-dwelling older adults in Singapore.

The specific objectives are:

1. To assess the effectiveness of the LightUP app in optimising daily light exposure patterns
2. To determine whether the intervention improves circadian rest-activity rhythms, sleep quantity and quality, mood and cognitive performance
3. To evaluate effects on frailty status, physical activity, body composition and vitamin D status
4. To examine feasibility, acceptability, usability and user satisfaction of the intervention
5. To explore the acceptability and user experience of the ActLumus light logger device

Study design

The LightSPAN trial is a randomised, double-blind, parallel-group, placebo-controlled trial conducted among community-dwelling older adults in Singapore. The trial is conducted at Lions Befrienders Active Ageing Centres across Singapore, which serve as community hubs for older adults.

Following eligibility screening and informed consent, participants undergo a 4-week baseline observation period during which habitual light exposure, sleep, and activity patterns are monitored using wearable devices. Participants are then randomised in a 1:1 ratio to either the intervention arm, which receives the LightUP app with active behaviour change features, or the control arm, which receives a placebo version of the app without behavioural components. The intervention period lasts 12 weeks and is followed by a 12-week post-intervention follow-up period, including a 2-week monitoring phase. Study staff conducting outcome assessments are blinded to group allocation throughout the study.

All assessments and app-based interventions are coordinated by TUMCREATE. Participants are expected to attend assessments at the respective Active Ageing Centres, while the intervention itself is delivered in participants’ home environments using smartphones and wearable devices. Participating Active Ageing Centres provide accessible facilities for participant recruitment and assessment. Personnel involved in delivering the intervention and conducting assessments are trained research staff with specific expertise in cognitive, mood, frailty, and physical performance testing, ensuring consistency and quality in data collection.

The trial design therefore allows for evaluation of both short-term (during intervention) and sustained (post-intervention) effects on light exposure, circadian rhythms, sleep, mood, cognition, frailty, physical activity and vitamin D status. This protocol adheres to the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) 2025 guidelines to ensure complete and transparent reporting [18].

Public involvement

Members of the public were involved in the design and planning of this study. Feedback from older adults recruited through community partners in Singapore helped to shape the study aims, ensure the intervention content was accessible and adapt the study procedures for cultural and linguistic appropriateness. Input was obtained on the usability of the mHealth app, the acceptability of wearing devices (light logger and activity tracker) and the burden of participant assessments. Community organisations and social service agencies such as Lions Befrienders also contributed to the development of recruitment strategies at Active Ageing Centres and will assist in the dissemination of study findings to participants and the wider community. The process of obtaining feedback from older adults and community partners has been reported in the published co-design protocol [19]. Participants and public representatives were not involved in drafting the present protocol.

Participants

Participants are community-dwelling older adults aged 60 years and above, residing in Singapore. At the screening visit, participants provide sociodemographic and background information using structured questionnaires. This includes age, sex, highest level of education completed, occupational history, marital status, living arrangement, and smartphone ownership and use, including the smartphone model used during the study. Health history and comorbidity status are assessed using the Self-Administered Comorbidity Questionnaire (SCQ) [20]. Additionally, we will administer the Morningness-Eveningness Questionnaire (MEQ) [21] and the Munich ChronoType Questionnaire (MCTQ) [22] to assess participants' chronotypes. These screening data are collected to describe the study sample and to support secondary and exploratory analyses examining factors that may influence adherence and intervention outcomes. Eligibility criteria are outlined below.

Eligibility criteria

Participants must be capable of independent mobility, with or without assistive devices and demonstrate functional independence as assessed by the Lawton Instrumental Activities of Daily Living (IADL) scale [23], with a score greater than 8 for females and greater than 5 for males. Participants are required to own a smartphone, be proficient in English reading and communication and not be currently enrolled in another mHealth interventional study.

Exclusion criteria include the presence of cognitive impairment, defined as a Montreal Cognitive Assessment (MoCA) [24] score below 26 after education adjustment, or depressive symptoms, defined as a score greater than 6 on the short form of the Geriatric Depression Scale (GDS) [25]. Individuals with significant physical impairments and/or medical diagnoses that interfere with daily activities, severe terminal illness, or psychiatric conditions affecting daily function will be excluded. Participants with visual impairment, diagnosed eye diseases, ocular abnormalities (as assessed by a licensed optometrist), or uncorrected hearing impairment are not eligible.

Recruitment, randomization, blinding and treatment allocation

Recruitment

Participants will be recruited from Lions Befrienders Active Ageing Centres across Singapore, which provide access to a diverse population of community-dwelling older adults. Recruitment strategies include on-site information sessions, flyers and outreach via community staff. Interested individuals will be screened for eligibility by trained research staff.

Enrolment will continue until the target sample size of 90 participants is reached. Research staff will maintain regular contact with participants during the baseline monitoring period to encourage retention and to address technical issues promptly. The collaboration with established community organisations is expected to facilitate efficient recruitment and reduce participant burden.

Randomisation

Sequence generation

Randomisation will be conducted using computer-generated block randomisation with the R package blockTools, stratified by age, sex and recruitment location to ensure balanced allocation across key demographic variables.

Allocation concealment and implementation

To guarantee allocation concealment, the allocation sequence will be securely implemented by a research team member who is not in direct contact with the participants.

Participant enrolment is conducted by trained research staff at the partner Active Ageing Centres. Group assignment is performed only after eligibility confirmation and completion of the baseline observation phase. The random allocation sequence is generated by an independent statistician and stored within a secure, central, web-based randomisation module managed by the data management team. Enrolling staff request assignment through this system; the system releases the allocation only at the point of randomisation. Personnel responsible for enrolment and those delivering study procedures have no access to the randomisation list or block details. The master sequence file is access-restricted (statistician and data manager only) and kept on an encrypted server with audit logs.

Blinding

Who is blinded

This study follows a double-blind outcome-assessor design with participant blinding procedures in place. Both participants and researchers will be blinded to the group assignments to prevent bias. Any technical or logistical queries regarding the use of light loggers and the LightUP app will be addressed by a designated technical support team blinded to the group assignments. This arrangement will reduce the risk of unintentional unblinding through routine interactions and ensure that study integrity is preserved.

How blinding is achieved and maintained

Blinding is maintained by using two apps with identical appearance, naming, iconography, onboarding flow and data entry screens. Both apps collect the same light, sleep and mood data and interact identically with the ActLumus logger and Garmin tracker; only the intervention app contains the active behaviour-change features (feedback, goals, prompts), which are implemented via server-side logic not visible to participants or assessors. App store/package identifiers and version numbers are masked in participant-facing materials; installation is performed via a study link that deploys the correct configuration automatically. Staff who perform outcome assessments have no role in app provisioning. To evaluate blinding integrity, participants and assessors will be asked at the endpoint to guess the assigned group and rate their confidence.

Unblinding procedures

Unblinding will only be conducted in exceptional cases where it is necessary for safety or technical reasons:

• Medical emergencies: In the event of a medical emergency where knowledge of the group assignment is essential for participant safety, the principal investigator (PI) will be notified. If deemed necessary, the PI will authorize a designated unmasked study personnel to access the randomization code by referring to the secure manual randomization list
• Technical issues: Should significant technical issues arise with the LightUP app or light loggers, unblinding may be required to resolve the problem. Only the technical team, authorized by the PI, will be granted access to the randomization code to address the issue. Participants will remain blinded throughout this process and no data will be compromised as a result of technical unblinding

Informed consent

Written informed consent will be obtained from all participants prior to any study procedures. Trained research staff at the Active Ageing Centres will explain the purpose, procedures, risks and potential benefits of the study using a standardised participant information sheet. Consent will be obtained in English or local dialects, depending on participant preference, and adequate time will be provided for participants to ask questions. Only participants who provide written, signed informed consent will be enrolled in the study.

Consent will cover the collection and use of self-report data, wearable device–derived light and activity data, cognitive performance measures and dried blood spot samples. Participants will be asked for permission for their anonymised data to be used in future ancillary studies and for sharing under open-access conditions (CC-BY). No biological samples will be stored beyond the planned vitamin D analyses.

Intervention

Experimental intervention

The experimental intervention is the LightUP mHealth smartphone application (app), which is downloaded onto participants’ own smartphones and used for 12 weeks, following a 4-week baseline monitoring phase. Participants in the intervention group will also wear a pendant-worn ActLumus light logger and a wrist-worn Garmin Vivosmart 5 activity tracker throughout the intervention period.

The LightUP app delivers evidence-based behaviour change techniques, including personalised feedback, self-monitoring, goal setting, and educational messages related to healthy light exposure. Participants are able to view their daily light exposure levels directly within the app, supporting self-monitoring and awareness of light-related behaviours. The app was co-designed with older adults and community service providers using a participatory, user-centred approach. The co-design process and development framework have been described in a previously published protocol paper [19].

Data from the ActLumus light logger are integrated into the app to tailor recommendations and to support participants in meeting daily and weekly light exposure targets. The app incorporates simple reward features, such as the awarding of trophies, to reinforce engagement and goal attainment. To support sustained engagement, participants receive a mid-intervention drop-in session during the intervention period.

Control intervention

The comparator is a placebo version of the LightUP app, which collects light, sleep, and mood data but does not provide behaviour change components such as personalised feedback, prompts, or goal-setting features. Instead, the placebo version retains the same data logging functions and presents neutral content and generic notifications to maintain engagement without directing behaviour.

Both the intervention and placebo versions share a similar interface, navigation structure, and notification timing, ensuring a comparable user experience across groups. Participants in the comparator group also wear the ActLumus light logger and Garmin Vivosmart 5 activity tracker, ensuring equivalent monitoring between groups and isolating the behavioural components of the intervention. Following the 12-week intervention period, participants in both groups enter a 12-week follow-up phase, which includes 2 weeks of extended monitoring.

Rationale for comparator

We selected a placebo version of the LightUP app as the comparator to ensure blinding and to isolate the specific effects of the behavioural intervention. Both groups use the same wearable devices (light logger and activity tracker) to monitor light, activity and sleep, but only the intervention group receives active features including personalised feedback, prompts and education. This design allows us to control potential confounding effects of self-monitoring and study participation while testing whether the behavioural features of the LightUP app confer additional benefit. As no standard of care currently exists for promoting light exposure in community-dwelling older adults, a placebo app was chosen over usual care to enhance methodological rigour.

Criteria for discontinuing or modifying allocated interventions

Participants may withdraw from the allocated intervention at any time, either at their own request or if adverse events, discomfort, or technical difficulties occur. In such cases, participants are invited to continue follow-up assessments unless they withdraw consent entirely. The intervention protocol is standardised and no modifications to the app are anticipated beyond its automated, data-driven feedback features. All instances of discontinuation or protocol deviation will be recorded.

Strategies to improve adherence

Adherence is supported through multiple strategies. The LightUP app incorporates reminders, progress-tracking features and personalised prompts to encourage consistent use. Research staff provide technical support for device setup, troubleshooting and ongoing use, including in-person onboarding with the wearable devices and LightUP installation, in-app tutorial, and a printed step-by-step participant booklet. Ongoing assistance is available via WhatsApp or phone, with in-person support provided on a needs basis. In addition, cloud-based data streams are monitored to identify issues such as failed syncing, battery depletion or limited app interaction, enabling timely follow-up. A mid-intervention drop-in session is conducted to maintain engagement. Overall, adherence and quality control of device and app-related components is monitored through app usage logs, wearable device data and completion of scheduled assessments.

Concomitant care

Participants are permitted to continue all aspects of their usual medical care, daily routines and lifestyle activities throughout the trial. However, concurrent participation in other mHealth intervention studies or trials targeting circadian rest-activity rhythms, sleep, mood, cognition, frailty, physical function, or vitamin D is not permitted, to avoid contamination and ensure that behavioural changes can be attributed to the LightUP app. All relevant changes in health status or treatments during the study will be documented.

Outcomes

Primary outcome

The primary outcome is the effectiveness of the LightUP app in improving light exposure behaviour in older adults. This will be assessed using data from the ActLumus wearable light logger (ActLumus, Condor Instruments; São Paolo, Brazil) across the 4-week baseline, the 12-week intervention and the 2-week follow-up monitoring period. The primary analysis metric is the mean daily time (minutes per day) above 250 lx melanopic equivalent daylight illuminance (melanopic EDI) during daytime hours. Changes will be calculated relative to baseline and both short-term effects (during the 12-week intervention) and longer-term effects (during follow-up) will be examined.

Secondary outcomes

Secondary outcomes span multiple domains.

Sleep

Quantity and quality: total sleep time, sleep efficiency, sleep onset latency and wake after sleep onset, measured via Garmin Vivosmart 5 (Garmin Ltd., Olathe, KS, USA) actigraphy and sleep diaries throughout the study.

Subjective quality: assessed using the Pittsburgh Sleep Quality Index (PSQI) [26] at weeks 4 (baseline), 10 (midpoint), 16 (endpoint) and 28 (follow-up).

Circadian rhythms

Interdaily stability, intradaily variability and relative amplitude of rest-activity cycles, measured continuously with Garmin data across the study period.

Mood

Mood will be assessed using the Brief Mood Introspection Scale (BMIS) [27] at baseline, midpoint, endpoint and follow-up. Daily mood will be captured using the Single-Item Mood Scale (SIMS) [28]. Weekly mood ratings will be assessed using the Visual Analogue Mood Scale (VAMS) during the intervention.

Cognition

Cognitive performance will be assessed using the NIH Toolbox^®^ battery [29] at baseline, midpoint, endpoint and follow-up, including the Dimensional Change Card Sort, Flanker Inhibitory Control and Attention, Oral Symbol Digit, Picture Sequence Memory, Rey Auditory Verbal Learning and Pattern Comparison Processing Speed tests.

Frailty and physical activity

Frailty will be assessed using standard criteria (weight loss, exhaustion, low activity via Physical Activity Scale for the Elderly (PASE), slowness via 5-metre walk and grip strength [30]). Physical activity outcomes including daily step count, sedentary behaviour, moderate-to-vigorous activity will be assessed using Garmin wearable data.

Body composition

Body composition outcomes, including weight, body fat percentage, muscle mass, bone mass and body mass index (BMI), will be assessed at baseline, midpoint, endpoint and follow-up using a bioelectrical impedance analysis (BIA) machine (TANITA DC-430MA, Tanita Corporation, Tokyo, Japan).

Vitamin D status

Vitamin D status will be assessed from dried blood spot samples at the same timepoints. 25-hydroxyvitamin D [25(OH)D] concentrations will be measured with a dried blood spot test kit (Vitamin D Test, NeoVos by SureScreen Health, Newcastle Upon Tyne, United Kingdom).

Feasibility, usability and acceptability

Usability of the LightUP app will be measured by the mHealth App Usability Questionnaire (MAUQ) [31] at endpoint. Acceptability of the app and wearable devices will be assessed using adapted versions of the Technology Acceptance Model (TAM) [32] and the Theoretical Framework of Acceptability (TFA) [33] questionnaires at endpoint. Qualitative feedback on participant experience, app usage behaviour and overall acceptability of the app will be obtained through a focus group discussion at endpoint assessment.

Time points for analysis

Primary and secondary outcomes will be analysed at baseline (week 4), mid-intervention (week 10), end of intervention (week 16) and follow-up (week 28), depending on the measure. Continuous monitoring outcomes (e.g. light exposure, actigraphy, mood diaries) will be aggregated into daily or weekly metrics as appropriate.

Participant timeline

The participant timeline and flow through the study are visualised in Supplementary Fig. 1. Following eligibility screening, obtaining informed consent, and documenting sociodemographic and health characteristics at Week 1, participants undergo a 4-week baseline monitoring period (weeks 1–4). During this period, they wear the ActLumus light logger and Garmin Vivosmart 5 activity tracker.

Following the baseline monitoring period, participants undergo baseline assessments (week 4) and are randomised in a 1:1 ratio to either the intervention or comparator arm.

The intervention phase lasts 12 weeks (weeks 4–16). Participants in the intervention group use the LightUP mHealth app, which delivers personalised feedback, education and goal setting, while participants in the comparator group use the placebo version of the app that records but does not provide feedback. Both groups continue to wear the ActLumus logger and Garmin device throughout this period. An additional drop-in session is delivered at week 10 to reinforce adherence.

A follow-up period of 12 weeks (weeks 17–28) follows the intervention. During this time, participants no longer use the app but continue their usual monitoring and assessments. The follow-up includes a 2-week monitoring phase with both the ActLumus and Garmin devices to assess sustained changes in light exposure, sleep and circadian rhythms.

Assessments are scheduled at four key time points:

• Week 4 – Baseline assessment (BL): sleep quality (PSQI), mood (BMIS), cognition (NIH Toolbox), frailty (five criteria), physical activity (PASE), body composition, vitamin D (25(OH)D)
• Week 10 – Midpoint assessment (MP): interim assessments of sleep, mood, cognition, frailty, physical activity, body composition and vitamin D
• Week 16 – Endpoint assessment (EP): repeat assessments including usability (MAUQ), acceptability (TAM, TFA) questionnaires and qualitative post-trial discussion
• Week 28 – Follow-up assessment (FU): repeat full assessments, including 2 weeks of device-based monitoring

Data collection

Outcome data will be collected using a combination of wearable devices, validated questionnaires, cognitive tests and clinical assessments as outlined in the Outcomes section of this protocol. Light exposure is measured continuously using the ActLumus pendant logger, while rest-activity cycles, sleep and physical activity are recorded using the Garmin Vivosmart 5 tracker. As outlined in Outcomes, validated self-report instruments include the PSQI, BMIS, VAMS, NIH Toolbox cognitive tests and PASE. Frailty measures include grip strength, gait speed, weight loss, exhaustion and activity level. Body composition is assessed using BIA and vitamin D is measured using dried blood spot analysis. Usability and acceptability of the intervention are assessed using the MAUQ, adapted TAM and TFA survey.

To ensure quality, all assessors are trained and certified in administering the questionnaires and performance-based measures. Standard operating procedures are in place for device calibration, biosample handling and data transfer. Data is collected electronically using secure, tablet-based entry systems, with automatic checks for out-of-range or missing values.

Participant retention and completeness of data collection will be supported through regular check-ins, technical support for device use and reminder notifications via the app in both trial arms. Participants who discontinue the intervention will still be invited to complete outcome assessments at scheduled visits unless they withdraw consent entirely. Reasons for non-adherence (e.g. device discomfort, loss of interest) or non-retention (e.g. withdrawal, loss to follow-up) will be systematically documented.

Data management

All study data will be entered into a secure, password-protected electronic database hosted on institutional servers. Data entry will be performed electronically, with built-in range checks, mandatory fields and double verification procedures for critical variables. Data from wearable devices and the mobile application will be automatically synchronised with the study server via encrypted data transfer. Bio sample results will be entered by laboratory staff following validation procedures.

Participant confidentiality will be preserved throughout the trial. Personal identifiers (e.g. name, contact details) will be collected separately from research data and stored in encrypted, access-restricted files accessible only to authorised study staff. Each participant will be assigned a unique study identification number, which will be used for all data collection and analysis. Wearable device and app data will be encrypted at rest and in transit and stored in compliance with the Singapore Personal Data Protection Act. All data shared with collaborators or made publicly available will be fully anonymised.

Data security will be ensured through role-based access controls, password-protected user accounts and regular system backups. The final locked dataset will be archived on secure institutional servers for a minimum of 10 years following study completion.

Sample size

Based on power calculations conducted using existing light exposure data, anchored to the primary outcome of time above threshold (TAT), defined as time spent above 250 lx melanopic EDI during daytime (TAT_250_), we estimate that a sample size of 25 participants per group (i.e., 50 participants in total) is sufficient to detect a 20% change in light exposure metrics with 80% power, assuming a standard deviation of 0.2 and a two-sided alpha of 0.05.

These calculations were based on an existing dataset from Malaysia, derived from a previous study comparing light exposure patterns in Switzerland and Malaysia [34]. This dataset is the closest available approximation to our target study population in terms of geographic and lifestyle similarities. The sample size estimation is documented in full in the project GitHub repository (https://github.com/tscnlab/MurukesuEtAl_BMCGeriatr_2026; 10.5281/zenodo.17914996 archived on Zenodo).

In the Malaysian dataset, daily time spent above 250 lx melanopic EDI (TAT_250_) averaged approximately 98 min per day, with 490 daily observations nested within 19 participants. This hierarchical data structure was retained in the power analysis using a mixed-effects model specification of the form outcome ~ intervention + (intervention | participant ID). As the original dataset was modelled using a Poisson distribution, the same distributional assumption was applied when generating simulation data for the power analysis via bootstrapping.

The power analysis focused on three light exposure metrics derived from the Malaysia dataset, as these represent conceptually relevant aspects of daytime and evening light exposure targeted by the LightSPAN intervention:

• Time above 1000 lx melanopic EDI during the daytime (TAT_1000_)
• Time above 250 lx melanopic EDI during the daytime (TAT_250_)
• Time below 10 lx melanopic EDI in the evening (TBT_10_)

All three metrics are relevant in the context of the project, with TAT_250_ designated as the primary outcome. Importantly, the required sample sizes were identical for all three metrics when evaluated in increments of five participants per group, and therefore the same sample size estimate applies to the primary outcome.

A 20% relative change, which defines the minimum detectable effect for the primary outcome (TAT_250_), in each of these metrics corresponds to approximately:

• ~ 9 min/day increase above 1000 lx melanopic EDI
• ~ 20 min/day increase above 250 lx melanopic EDI
• ~ 30 min/day reduction below 10 lx melanopic EDI in the evening

However, to account for:

• Additional outcome domains such as sleep, mood and cognitive function, which may show greater variability or require larger sample sizes to detect meaningful effects
• Considering the intensive nature of the trial and to effectively account for potential dropouts, an expected attrition rate of 25–30%, which is typical in studies involving older adults
• The need for more robust, generalizable results across a wider population

We plan to recruit a total of 90 participants (45 per group). This allows for approximately 63–68 participants to complete the study, corresponding to 31–34 participants per group, even after accounting for attrition. The recruitment strategy is guided by findings from similar mHealth interventions involving older adults, where dropout rates tend to be higher due to factors such as technology challenges, health status and varying levels of digital literacy. Studies in this demographic [35, 36] have demonstrated that over-recruiting by approximately 25–30% is a best practice to ensure adequate sample sizes for meaningful data analysis. By aiming to recruit at least 90 participants, we align with research standards that suggest this number is sufficient for exploratory analysis of the intervention’s effectiveness and feasibility.

Statistical analysis

Analysis population

Analyses will follow the intention-to-treat principle, including all randomised participants in their assigned groups. A per-protocol sensitivity analysis will be performed including only participants with high adherence (≥ 80% app engagement and valid device wear).

Primary outcome analysis

The primary endpoint is light exposure behaviour, operationalised as the mean daily time (minutes per day) spent above 250 lx melanopic EDI during daytime hours. Light exposure patterns will be analysed per individual using generalised linear mixed models (GLMMs). Daily TAT values will be entered into models including group assignment and assessment block/day-in-trial as fixed effects, with participant-level random effects to account for repeated daily measurements.

Based on prior analyses of TAT in a Malaysia–Switzerland light exposure study [34], Poisson and Gamma error distributions were identified as an appropriate fit for mixed-effects modelling and are therefore used for all analyses with TAT as the dependent variable.

The full model in Wilkinson notation will follow this specification:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned}&\:\mathrm{Outcome\:variable}\:\sim\:\mathrm{group}*\mathrm{assessment\:block}\\&+\mathrm{sex}+\mathrm{age}+\left(\mathrm{assessment\:block}\right|\mathrm{participant})\:\end{aligned}$$\end{document}

Model diagnostics will include visual checks for homoscedasticity, normality of residuals and random effects, and their location at zero. Baseline characteristics, including sociodemographic variables, health status, and baseline outcome measures, will be summarised descriptively by study group, and selected variables may be included as covariates in secondary or exploratory analyses where appropriate.

Secondary outcomes analyses

Secondary endpoints include outcomes spanning sleep, circadian rhythms, mood, cognition, frailty, physical activity, body composition, vitamin D status, and feasibility-related measures, assessed at BL, MP, EP, and FU, depending on the measure.

Quantitative differences across assessment time points will be examined using GLMMs or linear mixed-effects models, depending on the distributional properties of each outcome. Models will include group, assessment block, and group-by-assessment block interaction as fixed effects, with participant included as a random effect to account for repeated measures.

Statistical significance will be assessed using a two-sided alpha level of 0.05. p-values will be calculated using likelihood ratio tests comparing models with and without the parameter of interest. Unstandardised effect sizes (beta coefficients) will be reported. A false discovery rate correction will be applied within each branch of secondary outcome analyses, based on the number of models or tests performed within that branch.

Handling of missing data

Missing outcome data will be examined for patterns and assumed to be missing at random. Mixed-effects models inherently accommodate unbalanced repeated measures. For questionnaire outcomes, multiple imputations using chained equations will be considered. Sensitivity analyses (best-case/worst-case) will be conducted to assess robustness of findings. Missing or irregular data from the wearable light loggers will not be imputed. Irregular data may be regularized to a consistent time series within a participant and collection period. Participant days missing more than 2 h of data will be excluded from metric calculations and thus further analysis. Biller et al. (Malaysia/Switzerland) found that 6 h of missing data led to non-significant differences in a month-long data collection effort per participant [34]. As the follow-up period only contains 2 weeks of data collection, we choose this more conservative threshold.

Additional and exploratory analyses

Exploratory analyses will incorporate baseline, midpoint, endpoint, and follow-up as covariates or additional outcomes to characterise temporal patterns beyond the primary and secondary endpoints. Bayesian analyses may be conducted as a complementary approach, providing posterior estimates with corresponding credible intervals.

Continuously collected measurements, such as light exposure, might further be explored non-linearly through generalised additive mixed models (GAMMs) [37], which have been shown to be a good fit for personal light exposure data [38]. Additional exploratory analyses will be pre-registered on the Open Science Framework prior to execution to ensure transparency.

Interim analyses and stopping guidelines

No formal interim analyses are planned. Stopping criteria will be limited to safety concerns identified by the Steering Committee (e.g., unforeseen adverse events related to device use). Any decision to terminate or modify the trial will rest with the PI in consultation with the Steering Committee and the Institutional Review Board (IRB).

Statistical software

Statistical analyses will be conducted using R. Particular R packages that are planned to be used, among others, include LightLogR [39] for processing and calculating light data and variables, and tidyverse [40] for a stringent, tidy analysis framework.

Adverse events

The LightSPAN trial is considered minimal risk, as the intervention involves the use of a mHealth application and wearable monitoring devices. There are no known serious harms associated with these technologies. Possible risks include minor discomfort from wearing the ActLumus light logger or Garmin Vivosmart 5 activity tracker. These risks are considered low and are comparable to wearing commercially available wearable devices. No invasive procedures are involved except for dried blood spot collection for vitamin D analysis, which may cause minor discomfort.

Systematically assessed adverse events

Participants excluded following ocular health screening will receive a referral letter outlining the findings and recommending follow-up with an appropriate healthcare provider or specialist.

Non-systematically assessed adverse events

Participants are encouraged to report any problems or adverse experiences spontaneously to the study team throughout the trial. Reports will be collected via phone, email, or in person during centre visits.

Coding and grading of adverse events

All adverse events will be documented and coded by study staff according to severity (mild, moderate, severe) and relatedness to the intervention (unrelated, possibly related, probably related). Coding will be performed by personnel blinded to group allocation. No formal clinical grading system is planned, as no serious medical harms are anticipated.

Grouping of harms and reporting of adverse events

Harms will be grouped by seriousness (serious versus non-serious), severity and type (e.g., device-related discomfort, app-related difficulties). Any adverse events leading to discontinuation of the intervention will be recorded separately. If a serious adverse event is reported, it will be immediately reviewed by the PI and reported to the Parkway Independent Ethics Committee (PIEC), in line with ethical approval requirements.

Ancillary and post-trial care

As the intervention involves a digital health app and wearable devices, no specific ancillary or post-trial medical care is anticipated. Any discomfort related to device wear (e.g. skin irritation) will be addressed by study staff. Participants will be covered by institutional insurance policies for study-related harm, in accordance with local regulations. No compensation is planned for non-study-related health events. Participants will receive a lay summary of the study findings at trial completion.

Audits and inspections

Composition and role

Given the minimal-risk nature of the intervention (digital health app and wearable devices), no independent Data Monitoring Committee (DMC) will be established. Trial oversight will be managed by the study’s Steering Committee, which includes investigators from TUMCREATE and collaborating institutions. The Steering Committee meets monthly to review recruitment, adherence, data quality and any adverse events.

Trial monitoring

Trial conduct will be monitored by the institutional research governance office, which will conduct regular remote monitoring of recruitment, consent forms and data quality. Monitoring will include verification of consent, adherence to protocol and reporting of adverse events. No on-site audits are anticipated unless concerns arise.

Protocol amendments

All substantial protocol amendments (e.g., changes to eligibility criteria, outcomes, analyses) will be reviewed and approved by the relevant IRBs prior to implementation. Updates will be made in the trial registry (ISRCTN12391932, registered 5 September 2025) and communicated to study sites, funders and participants as appropriate. Minor administrative amendments (e.g., corrections of typographical errors) will be documented in the trial master file.

Discussion

Strengths

This study has several strengths. It represents a novel digital behavioural intervention that specifically targets environmental light exposure as a modifiable, lifestyle-based strategy for preventive health and health promotion among older adults. To our knowledge, this is the first study to operationalise light exposure optimisation through a mHealth platform in this population. The trial is underpinned by a rigorous methodological design, employing a double-blind randomised controlled framework with allocation concealment and stratified balancing by sex, age, and recruitment site. These design features minimise selection and performance bias and enhance internal validity. A further strength is the use of a multi-modal assessment framework that combines objective measures of light exposure, physical activity, and physiological outcomes, captured using the ActLumus light logger and Garmin Vivosmart 5, with validated subjective instruments. This approach enables comprehensive outcome assessment and robust triangulation across measurement modalities. The inclusion of a pilot and feasibility phase allows refinement of study procedures, identification of implementation challenges, and optimisation of intervention delivery prior to the trial, thereby reducing operational risks and strengthening overall feasibility. Finally, recruitment through Lions Befrienders’ Active Ageing Centres situates the intervention within a real-world community context, enhancing ecological validity and supporting the potential translation of findings into community-based and public health programmes.

Limitations

Several limitations should be considered. The digital nature of the intervention may impose a technology-related burden for some older participants, particularly with respect to device wear, data synchronisation, and mobile application use, which could affect adherence and data completeness. Furthermore, sample size assumptions were informed by effect size estimates derived from a Malaysian dataset, which may not fully reflect variability within the Singaporean older adult population and may introduce uncertainty in power estimation. Although over-recruitment has been planned, attrition related to digital literacy, participant burden, or device fatigue remains a potential risk and could reduce effective statistical power. In addition, the exclusion of individuals with cognitive impairment, ocular disease, or clinically significant depressive symptoms was necessary to ensure participant safety and reduce clinical and behavioural confounding. However, this may limit generalisability, resulting in a study sample that is comparatively healthier and more digitally capable than the broader older adult population. While blinding is strengthened by the identical visual design and navigation of the intervention and placebo applications, functional differences between versions may allow some participants to infer group allocation. Although outcome assessors remain blinded, the possibility of partial participant unblinding and residual performance bias cannot be entirely excluded.

Publications and dissemination policy

The findings of this study will be primarily disseminated through scientific channels. Results will be published in peer-reviewed open access journals and preprints will be deposited on servers such as medRxiv to allow rapid access by the research community. In addition, anonymised participant-level data, statistical code and supporting documentation will be deposited in an open repository to ensure transparency and facilitate secondary analyses.

Authorship will follow International Committee of Medical Journal Editors (ICMJE) guidelines and no professional medical writers will be involved in the preparation of manuscripts. Author contributions will be attributed using the Contributor Role Taxonomy (CRediT).

Protocol and statistical analysis plan

The full trial protocol will be published in BMC Geriatrics. Any protocol amendments will be updated in the ISRCTN record to ensure accuracy and transparency. A separate statistical analysis plan will be finalised before data analysis and will be made openly available on the Open Science Framework (OSF). This approach ensures that analytical and statistical procedures are defined in advance and can be independently scrutinised.

Supplementary Information

Supplementary Material 1.

Supplementary Material 2.