Protocol for a single-arm, multi-component behavior change technique (BCT) intervention to develop a walking habit among caregivers for persons with Alzheimer disease and related dementias (ADRD)

Abstract

Even low to moderate physical activity is critical in improving and maintaining physical health and well-being. Caregivers of people living with Alzheimer disease and related dementias (ADRD) are a burdened population and, as such, can experience challenges with managing even modest increases in physical activity while caring for others. While some interventions have been proposed to increase physical activity, many fail to consider the unique needs of caregivers of people with ADRD.

The purpose of this 12-week decentralized behavioral trial is to test the efficacy of a multi-component, personalized text-message delivered BCT intervention to encourage the formation of a daily walking habit among caregivers of persons with ADRD assessed by Fitbit activity trackers via the key mechanism of behavior change (MoBC) of behavioral automaticity. Formation of a daily walking habit will be defined as attainment of walking 1000 or more additional steps during the same one-hour period on 7 consecutive days as set up in a personalized walking plan. We will also evaluate the association of habit formation attainment with changes in behavioral automaticity, association between longitudinal behavioral automaticity and habit formation attainment over time, and the heterogeneity of treatment effects between participants.

Results will advance science about behavioral habit formation among caregivers for persons with ADRD and determine whether behavioral automaticity acts as the primary MoBC for the effect on this BCT intervention on daily habitual walking.

This trial is registered on www.ClinicalTrials.gov (https://clinicaltrials.gov/study/NCT06803797); NCT #: NCT06803797.

1. Introduction

In the United States, rates of Alzheimer disease and related dementias (ADRD) are increasing [1,2]. Currently, 1 in 9 people aged 65 years and older have ADRD [3]. It is projected that ADRD will affect 8.4 million older Americans by 2030 and 14 million by 2060 [2,4]. Caring for this rising population of individuals with ADRD is time intensive and demanding for both paid and unpaid caregivers [4-7]. In 2022, the Alzheimer's Association estimated over 11 million unpaid caregivers (family members and others) provided approximately 18 billion hours of care to people with ADRD [3].

It is well understood that caregivers of people with ADRD experience adverse effects on their mental health, physical health, quality of life, and financial well-being [5-13]. ADRD caregivers experience more difficulty taking care of their own personal health than non-ADRD caregivers, with nearly twice as many ADRD caregivers reporting that caregiving made their health worse [5]. This burden disproportionately affects dementia caregivers who are women, as well as those who are Hispanic, Black, and Asian American [14-17]. These levels of burden often interfere with positive health behaviors, including physical activity [18-21].

Physical activity, even low to moderate levels, has been shown to improve physical health and wellbeing and prevent numerous chronic diseases and conditions that impact overall quality of life [22]. Physical activity can even prevent or delay the onset of functional impairment and prolong the ability to live independently in older adults [23]. As a means of increasing physical activity, prior interventions have shown that increasing daily step counts can have significant benefits to overall health, cardiometabolic health, and psychological functioning [24-31]. Although most trials argue that more steps per day are associated with greater improvements in health, prolonged periods of activity may not always be feasible for populations like caregivers. Several trials have shown benefits of taking as little as an additional 1000 steps per day on both physical and psychological health [25,27,30]. With one-third of caregivers of people with ADRD being aged 65 years or older themselves, and 60% of caregivers not engaging in any regular physical activity, this population is particularly poised to benefit from engaging in even small amounts of physical activity to prevent physical decline and is well-primed for physical activity interventions [5,32]. However, the burdensome nature of caregiving for people with ADRD must be considered. This further emphasizes the need for interventions that are uniquely personalized to caregiver's circumstances.

Behavior change technique (BCT) interventions are one of many types of interventions that have been used to promote positive health behaviors like physical activity and may present an opportunity for such personalization [33,34]. BCTs are observable, replicable, irreducible intervention components that are thought to influence mechanisms of behavior change (MoBCs) that causally change behavior [35,37]. Researchers have applied habit formation theory to the design of behavior change interventions, including physical activity interventions [33,36,38,39]. Using habit formation theory in a remotely delivered trial could be effective in that this model relies on context-dependent repetition of a behavior, and the remote delivery of that repetition may be easily integrated into an individual's day-to-day life (for example, walking a lap around the office at the start of each hour during a person's unique workday schedule) [33]. Further, habit has been conceptualized as a form of behavioral automaticity. Behavioral automaticity is believed to be a possible mechanism of behavior change (MoBC), or the way BCT(s) affect behavior [33,34,36]. Whether personalized BCT interventions for physical activity that are rooted in habit formation via the MoBC of behavioral automaticity have the potential for effectiveness requires additional investigation.

Therefore, the current trial examines the efficacy of a multi-component BCT intervention to form a habit of personalized daily walking among caregivers. The primary goal of this study is to is to identify whether a significant proportion of caregivers of individuals with ADRD will be able to develop daily habitual walking. We hypothesize that this BCT intervention to increase a walking habit of 1000 steps per day will lead to successful development of habitual walking among 60% of caregivers enrolled. Secondary hypotheses include that 1) changes in the MoBC behavioral automaticity will be associated with an increased likelihood of forming a walking habit; 2) as the intervention increases levels of the MoBC behavioral automaticity increases longitudinally, the likelihood of developing a walking habit will increase; 3) there will be heterogeneity in the effects of BCT intervention on the analyses described above.

2. Methods

2.1. Study design

This fully remote, NIH Stage II decentralized behavioral trial employs a single-arm, multi-component, personalized BCT intervention to caregivers of individuals with ADRD [37]. This is a 12-week study comprised of a 2-week baseline period and a 10-week intervention period (Fig. 1). We aim to enroll 100 caregivers to have an analysis sample of 60 caregivers at study completion. The trial is informed by habit formation theory, and tests if behavioral automaticity is the MoBC for habitual physical activity (Fig. 2) [33]. All procedures will be performed in compliance with relevant laws and institutional guidelines. The privacy rights of human subjects will be observed and informed consent will be obtained for human subjects. This study was approved by the Northwell Health Institutional Review Board (IRB; IRB# 24–0045 first approved 2/20/2024).

Fig. 1.

Participant timeline. Timeline visual of 12-week trial participation.

Fig. 2.

Habit formation theory applied to this trial. Adapted from Lally and Gardener, 2013.

2.2. Study population

Participants must identify as a formal (e.g. paid professional) or informal (e.g. unpaid) caregiver of individual(s) with ADRD, aged 18–85 years, who have self-reported low levels of physical activity or walking, and whose primary language is English or Spanish. Further inclusion and exclusion criteria are detailed in Table 1.

Table 1.

Eligibility criteria. Inclusion and exclusion criteria for eligibility.

Inclusion Criteria	Exclusion Criteria
Identify as a caregiver (formal/paid or informal/unpaid) for persons with Alzheimer's Disease or Alzheimer's Disease Related Dementias (AD/ADRD)	Individuals who self-report having been informed by a clinician it is medically or physically unsafe to engage in a walking intervention
Age ≥18 and ≤85	Does not own or cannot regularly access a smartphone capable of receiving text messages or accessing the internet
Speak English or Spanish as primary language	Does not own or have access to an email address
Self-report low levels of physical activity or walking	Lives outside the United States

2.3. Recruitment

Caregivers will be recruited using a combination of online and inperson methods. Within the Northwell Health system, recruitment efforts will be made via collaborations with ADRD-specific institutions and the Northwell Health Caregiver Business Employee Resource Group, and via the electronic health record to identify individuals with ADRD caregiver status. External digital channels including social media platforms like Facebook and Reddit will also be utilized. Recruitment efforts will extend to community settings within and outside the Northwell network, leveraging electronic communications and events. All recruitment and advertising materials will maintain ethical standards, avoiding coercion and undue incentives to participation. These materials, along with any other forms approved by the IRB, were professionally translated into Spanish by a third-party service to accommodate participants whose primary or preferred language is Spanish.

Interested individuals will be directed to a secure REDCap (Research Electronic Data Capture) form in their preferred language to begin screening. REDCap is a secure, web-based software platform designed to support data capture for research studies [40,41]. Table 2 outlines all survey assessments during the screening and the rest of the trial.

Table 2.

Survey measures and time points.

		Time Point
		Screening	Consent	Pre-Baseline	Baseline	Pre-Intervention	Intervention	Post-Intervention
Survey Measure	Authorization/Contact Information	X
	NIA CROMS Demographic Survey	X
	Eligibility Survey	X
	Consent Form		X
	Onboarding Survey			X
	Zarit Burden Interview			X				X
	Perceived Stress Scale-10			X				X
	Patient Health Questionnaire-8			X				X
	State-Trait Anxiety Inventory			X				X
	EuroQol EQ-5D-5L			X				X
	Katz Activities of Daily Living			X
	Neuropsychiatric Inventory Questionnaire			X
	Self-Reported Behavioral Automaticity Inventory + item from EuroQol EQ-5D-5L*				X		X
	Intervention Comprehension Survey					X
	System Usability Scale							X
	Satisfaction Survey							X

Note: *Indicates a survey measured at a weekly cadence during the length of the timepoint.

After completing a HIPAA authorization form, providing contact information, and completing a demographic survey required by the funder (National Institute on Aging [NIA] for the Clinical Research Operations & Management System [CROMS]), potential participants will complete a screening survey to assess eligibility based on inclusion and exclusion criteria. Individuals deemed ineligible from their survey responses will be directed to a page indicating they do not qualify and thanking them for their interest. Individuals meeting criteria will be provided with the study consent form.

2.4. Consent and pre-baseline

An IRB-approved REDCap form will be used to obtain consent electronically. Eligible participants will be sent a secure link to electronically review and sign the consent form. After consenting, participants will answer an onboarding survey with questions, including those on their relationship to the cared-for individual with ADRD, duration, and type of caregiving (e.g. primary or not primary caregiving), which are potential moderators of the intervention's effect. Additionally, the onboarding survey will have participants select three possible study start dates. Participants will be asked to fill out additional self-assessment tools to measure caregiver burden (12-item Zarit Burden Interview), stress (10-item Perceived Stress Scale; PSS-10), anxiety (10-item StateTrait Anxiety Inventory; STAI-S), depressive symptoms (8-item adapted Patient Health Questionnaire excluding the 9th question on having thoughts that one would be better off dead, or thoughts of hurting oneself; adapted PHQ-9), and quality of life (6-item EuroQol EQ-5D-5L) [42-47]. All surveys will be completed in REDCap and sent electronically via SMS text message.

Following the completion of the onboarding survey and self-assessment tools mentioned above, the study team will contact the participant to assign a baseline period start date. The study team will ship eligible participants a Fitbit activity tracker and will electronically send a set up guide for the device. At this time, participants will complete electronic measures of care recipient functional independence (6-item Katz Activities of Daily Living; Katz ADL) and care recipient symptom severity and associated caregiver distress (12-item Neuropsychiatric Inventory Questionnaire; NPIQ), which may also be moderators of the intervention's effect [47-50]. Pre-baseline surveys are sent in batches to minimize perceived burden.

2.5. Baseline

On their start date, participants will begin a 2-week baseline period where they are encouraged to maintain their regular activity levels and complete weekly measures sent via SMS text once per week that include the 4-item Self-Report Behavioral Automaticity Index (SRBAI) [51] and the single-item overall health assessment question from the EuroQol EQ-5D-5L [46]. During this baseline period, participants will be asked to wear their Fitbit activity tracker for a minimum of 10 h daily on at least 12 of the 14 days and fill out at least one of the two weekly surveys to be eligible for entry into the 10-week intervention period of the study.

Upon completion of the baseline period, the study team will ask the participant if they want to continue to the next phase of the study. Up to 3 days before beginning the intervention period of the study, participants will be electronically sent a survey assessing comprehension of key details of the intervention period. Participants will be asked to select a preferred 1-h window to take an additional 1000 steps daily for the next 10 weeks. Participants will also select times where they prefer to engage in goal setting, action planning, and self-monitoring behavior (the BCT components of the intervention, described in detail below).

2.6. Intervention period

During the 10-week intervention period, participants will be sent 4 BCTs daily via 4 SMS text messages before and after each participan's preferred walking time. The selection of BCTs was informed by habit formation theory; the sequential engagement of these 4 BCTs are meant to initiate and learn the behavioral habit of daily walking [52-55]. Each day, the text message BCTs will alert participants to set a goal to walk (1.1 Goal Setting (behavior)), sent before their preferred walking time, develop a plan to walk (1.4 Action Planning), sent before their preferred walking time, engage in walking (7.1 Prompts/Cues), sent at their preferred walking time, and self-monitor their walking success (2.3 Self-Monitoring of behavior), sent 30 min after their preferred walking time [36]. Table 3 includes the language of each BCT intervention component in English and Spanish. Fig. 3 provides a sample personalized schedule of the multi-component BCT intervention for a caregiver.

Table 3.

BCT language. Language of BCT intervention components in English and Spanish.

BCT	Language
	English	Spanish
Goal setting (behavior)	Take a moment to set a goal to walk at least 1000 extra steps during your next walking time between [walking period start time] and [walking period end time].	Tómese un momento para la fijación de un objetivo para caminar al menos 1000 pasos más durante su próxima caminata entre [walking period start time] y [walking period end time].
Action planning	Think about where, when, and for how long you will walk to take at least 1000 extra steps during your next walking time between [walking period start time] and [walking period end time].	Piense en dónde, cuándo y durante cuánto tiempo caminará para dar al menos 1000 pasos más durante su próxima caminata entre [walking period start time] y [walking period end time].
Self-monitoring of behavior	Take a moment and reflect: did you walk at least 1000 extra steps between [walking period start time] and [walking period end time] today? You do NOT need to respond to this message.	Tómese un momento y reflexione: ¿caminó al menos 1000 pasos más entre [walking period start time] y [walking period end time] hoy? NO es necesario que responda a este mensaje.
Prompts/Cues	It's time to take at least 1000 extra steps this hour.	Es hora de dar al menos 1000 pasos adicionales esta hora.

Fig. 3.

Sample BCT Schedule. Sample personalized daily BCT package schedule for a caregiver participant with a preferred 1-h walking time of 1–2 PM.

Participants will continue to wear their Fitbit and complete weekly measures of the SRBAI and a single-item health assessment from the EuroQol EQ-5D-5L during the 10-week intervention period. We do not expect adverse events (serious or otherwise) to occur as part of participation in this no-greater-than-minimal risk research. We will use a non-systematic collection method of reported events to determine if an adverse event has occurred, and to determine next steps, if any, for reporting. Events reported by participants that meet the definition of an adverse event from the start of intervention through the end of the study will be collected in electronic format using REDCap. The severity, expectedness, and potential relatedness to the study intervention of each adverse event will be classified. Adverse events will be reported to the local safety monitor, funder, and IRB according to policy.

After the 10-week intervention period concludes, participants will be asked to fill out surveys assessing caregiving burden, stress, anxiety, depressive symptoms, and quality of life using the same pre-intervention self-assessment measures. Participants will also be asked to complete a survey assessing trial usability (System Usability Scale; SUS) and their satisfaction with the trial and BCTs [56-58]. Participants will be reminded that this research is voluntary, and they may choose to withdraw from the trial at any point. Participants will be given an option to partially withdraw from the trial; in this scenario participants will not receive BCTs but will give permission to the study team to continue to track their Fitbit data until their 10-week intervention period would have ended.

2.7. Compensation

Upon completion of the trial (defined as completion of the end of study surveys), participants will be compensated $200 via ClinCard (reloadable pay card) that will be shipped to them. Participants will have a chance for additional compensation via a weekly lottery that takes place during the intervention period. Each week, one winner will be selected from a pool of participants in the intervention phase of the study for the lottery. Selected participants who wear their Fitbit tracker for at least 10 h per day 6 out of 7 days that week and complete their weekly survey that week are eligible to win the lottery prize. Winners of the lottery prize will be compensated with a $150 ClinCard and are eligible to win up to 10 times during their intervention period (which is highly unlikely). Participants not selected as the winner of the lottery each week will receive notification that they were not selected, details on whether they would have been eligible or ineligible if they had been selected, and how they can become eligible in future weeks.

3. Outcomes and statistical analyses

3.1. General statistical analytical approach

Baseline measurements will be summarized with descriptive statistics (mean, median, and standard deviation for continuous variables [e. g., age] and proportions for categorical variables [e.g., sex]). We will estimate all coefficients with 95% confidence intervals in regression models. For point estimates, two-sided p values will be used.

3.2. Primary outcome and analysis

The primary outcome is a binary indicator of habit formation, defined as first attainment of daily walking of 1000 additional steps or more during each participan's preferred 1-h walking time for 7 consecutive days during the 70-day intervention period. This preferred 1-h walking time was identified before the intervention period. Fig. 4 demonstrates successful and unsuccessful attainment of formation of a daily walking habit in a given week during the trial. Defining habitual activity based on a week of consistent behavior is based on a previous trial from this group [59]. Hourly steps will be objectively measured using the Fitbit device. We will assess the efficacy of the personalized BCT intervention by testing the null hypothesis of habit formation rate equal to 40% using a 1-sample binomial test at the 5% level 2-sided. Rates of achieving habit formation in the current study will be reported as an observed proportion with a 95% confidence interval estimate.

Fig. 4.

Examples of successful and unsuccessful attainment of formation of a daily walking habit.

3.3. Secondary outcomes and analyses

There will be 3 secondary outcome analyses completed in the current trial: (1) the association between habit formation attainment and changes in behavioral automaticity, (2) the association between longitudinal behavioral automaticity and habitual formation attainment over time, and (3) a reporting of the amount of heterogeneity of treatment effects (HTEs) for the intervention's effects on habit formation and on changes in behavioral automaticity.

Behavioral automaticity of habitual daily walking will be measured using the Self-Report Behavioral Automaticity Index (SRBAI) [51]. The SRBAI is a 4-item measure with a Likert scale rating of 1 (Strongly Disagree) to 7 (Strongly Agree) with a higher score indicating higher levels of automaticity. Behavioral automaticity will be measured 12 times (2 times during baseline, and 10 times during the intervention period) and utilized as the key MoBC for this trial. To examine whether habit formation attainment is associated with positive changes in behavioral automaticity, Fisher's exact test (5% 2-sided) will be used. Specifically, for each participant, the difference between average behavioral automaticity during the last 2 weeks of intervention and average baseline behavioral automaticity will be calculated. The rate of habit formation between the group of participants whose behavioral automaticity increased and the participants whose behavioral automaticity did not increase will be compared using a Fisher's exact test. Finally, logistic regression will be used to assess the effects of behavioral automaticity on the development of a daily walking habit with adjustment for other factors, such as caregiver demographic characteristics and factors related to caregiving. A significant test result may indicate behavioral automaticity as one of the potential underlying MoBCs of habit formation.

To examine the longitudinal association among behavioral automaticity and habitual walking over time, the change of behavioral automaticity during a given week from baseline behavioral automaticity will be used as an independent covariate in a logistic regression model, at first, individually. The univariate logistic regression will provide preliminary estimate of how behavioral automaticity change at a given week will attainment of habit formation, thus informing how the timing of behavioral automaticity change will impact habit formation. A multivariate Bayesian monotone regression model trained using iPIPE will be used to explore these weekly behavioral automaticity changes together. iPIPE is a novel statistical learning method developed by our research team that can be used to train and improve precision of Bayesian models with a goal to assess the joint effect of multiple behavioral automaticity changes on habit formation [60]. Briefly, iPIPE models the rate of habit formation as a monotone nonparametric multivariate function of weekly behavioral automaticity changes during the intervention. This is a completely data-driven approach to explore the association of the underlying mechanism that is measured longitudinally with habit formation attainment.

To characterize the HTEs for habit formation and changes in behavioral automaticity, analyses of HTEs will be conducted across participants. This will involve examining the heterogeneity in time to achieving habitual daily walking due to the BCT intervention using methods developed for N-of-1 behavioral clinical trials [61]. Briefly, the analyses will compare a random intercept model of the effect of the intervention on habit formation versus a random slope model using likelihood ratio test. If a model with a random slope fits better, this will suggest that the relationship between the intervention and habit formation differs between participants. We will also calculate an index of heterogeneity called Cheung-Mitsumoto Index [60].

3.4. Exploratory outcomes and analyses

This trial will conduct exploratory analyses to identify potential moderators of the effect of the BCT intervention on habit formation. Stratified analyses and examination of interaction effects between each moderating variable and the intervention on habitual physical activity will be conducted for the following variables: caregiver sociodemographic factors from the NIA CROMS survey required by the funder, caregiver's relationship to the person they provide care to with ADRD, whether they are the primary caregiver (defined as providing all or most of the care that the person receives), how long they've been providing care, number of hours per week they provide care, whether they live with their care recipient, when their care recipient was diagnosed with ADRD, the race, ethnicity, gender, and age of the person they provide care for, and pre- baseline measurements of the Katz ADL (with a higher score indicating more independence) and NPIQ (with a higher symptom rating indicating more severe symptoms and a higher caregiver distress rating indicating more distress).

To identify whether the development of habitual physical activity is associated with important elements of caregiver well-being, pre-post paired samples t-tests will be conducted to identify whether the intervention is associated with reductions in caregiver self-reported burden, stress, anxiety symptoms, and depressive symptoms using the ZBI, PSS-10, STAI-S, and adapted PHQ-9 described previously, respectively, and with increases in quality of life using the EuroQol EQ-5D-5L. Subgroup analyses will be performed by the baseline demographics by sex-defined subgroup.

3.5. Sample size calculation

Assuming a null habit formation rate of 40% with a sample size of n = 60, a one-sided, one-sample binomial test at 5% significance will have about 92% power to declare the BCT intervention effective if the true habit formation rate is 60% or above. To anticipate the variability due to the proposed total sample size of 60 for the secondary analysis, the statistical power to detect a difference between 30% habit formation (in the group of participants without increased behavioral automaticity) vs 70% habit formation (in the group of participants with increased behavioral automaticity) under different scenarios was calculated. Specifically, when 30% of the participants experience an increase in behavioral automaticity, there is 83% power to detect the difference. When 50% of the participants experience an increase in behavioral automaticity, the power increases to about 88%.

With plans to enroll a sample of 100, and assuming a 20% attrition following baseline (N = 20) and an additional 25% attrition after the intervention (N = 20), the current trial still will have a sufficient sample (N = 60) to provide adequate power.

4. Conclusions

This study protocol describes the aims of a proposed NIH Stage II decentralized behavioral trial to test the efficacy of a multi-component, personalized BCT intervention to form a daily walking habit among care providers of individuals with ADRD via the key MoBC of behavioral automaticity. The long-term goal of this research program is to engage in theory-based behavioral interventions informed by habit formation theory and utilize BCT components and measurement of putative MoBCs to improve the physical activity and health of caregivers for individuals living with ADRD. We hope to employ our fully digital, remote, behavioral intervention to begin a larger research program improving the health of caregivers and the recipients of care.

This will provide data to support larger NIH stage model clinical trials designed for rapid progression through other NIH Stages to assess real-world efficacy (Stage III), effectiveness (Stage IV) and community effectiveness (Stage V).

Glossary

ADLs

activities of daily living

ADRD

Alzheimer disease and related dementias (ADRD)

BCT

Behavior change technique

CROMS

Clinical Research Operations & Management System

HTEs

Heterogeneity of treatment effects

IADLs

instrumental ADLs

IRB

Institutional Review Board

Katz ADL

Katz Activities of Daily Living

MoBCs

mechanisms of behavior change

NIA

National Institute on Aging

NIH

National Institutes of Health

NPIQ

Neuropsychiatric Inventory-Questionnaire

PHQ-9

Patient Health Questionnaire-9

PSS

Perceived Stress Scale

SBRAI

Self-Report Behavioral Automaticity Index

STAI-S

State-Trait Anxiety Short Form

ZBI

Zarit Burden Interview

Data availability

No data was used for the research described in the article.