Skip to main content
NCBI home page
JAMA Network logoLink to JAMA Network
. 2023 Mar 30;6(3):e235681. doi: 10.1001/jamanetworkopen.2023.5681

Association of Demographic and Socioeconomic Indicators With the Use of Wearable Devices Among Children

Ethan H Kim 1, Jessica L Jenness 2, Adam Bryant Miller 3,4, Ramzi Halabi 1, Massimiliano de Zambotti 5, Kara S Bagot 6, Fiona C Baker 5, Abhishek Pratap 1,7,8,9,10,✉
PMCID: PMC10064258  PMID: 36995714

Key Points

Question

Does the large-scale usage of wearable devices in children vary based on demographic and socioeconomic indicators?

Findings

In this cohort study of 10 414 children, there was a statistically significant association between participants’ sociodemographic characteristics and willingness to enroll and engage in a wearable device study. Black children and those from lower socioeconomic status households were less likely to participate and wore devices for significantly less time than White children and those from higher socioeconomic status households, respectively.

Meaning

The findings of this study suggest that without factoring in the broader social determinants of health that may affect individual and group experiences and participation in research, inequities in data collection using wearable technologies may continue to exist, especially for youths belonging to racial and ethnic minority groups.

This cohort study examines whether demographic and socioeconomic indicators are associated with willingness to join a wearable device study and adherence to wearable data collection in children.

Abstract

Importance

The use of consumer-grade wearable devices for collecting data for biomedical research may be associated with social determinants of health (SDoHs) linked to people’s understanding of and willingness to join and remain engaged in remote health studies.

Objective

To examine whether demographic and socioeconomic indicators are associated with willingness to join a wearable device study and adherence to wearable data collection in children.

Design, Setting, and Participants

This cohort study used wearable device usage data collected from 10 414 participants (aged 11-13 years) at the year-2 follow-up (2018-2020) of the ongoing Adolescent Brain and Cognitive Development (ABCD) Study, performed at 21 sites across the United States. Data were analyzed from November 2021 to July 2022.

Main Outcomes and Measures

The 2 primary outcomes were (1) participant retention in the wearable device substudy and (2) total device wear time during the 21-day observation period. Associations between the primary end points and sociodemographic and economic indicators were examined.

Results

The mean (SD) age of the 10 414 participants was 12.00 (0.72) years, with 5444 (52.3%) male participants. Overall, 1424 participants (13.7%) were Black; 2048 (19.7%), Hispanic; and 5615 (53.9%) White. Substantial differences were observed between the cohort that participated and shared wearable device data (wearable device cohort [WDC]; 7424 participants [71.3%]) compared with those who did not participate or share data (no wearable device cohort [NWDC]; 2900 participants [28.7%]). Black children were significantly underrepresented (−59%) in the WDC (847 [11.4%]) compared with the NWDC (577 [19.3%]; P < .001). In contrast, White children were overrepresented (+132%) in the WDC (4301 [57.9%]) vs the NWDC (1314 [43.9%]; P < .001). Children from low-income households (<$24 999) were significantly underrepresented in WDC (638 [8.6%]) compared with NWDC (492 [16.5%]; P < .001). Overall, Black children were retained for a substantially shorter duration (16 days; 95% CI, 14-17 days) compared with White children (21 days; 95% CI, 21-21 days; P < .001) in the wearable device substudy. In addition, total device wear time during the observation was notably different between Black vs White children (β = −43.00 hours; 95% CI, −55.11 to −30.88 hours; P < .001).

Conclusions and Relevance

In this cohort study, large-scale wearable device data collected from children showed considerable differences between White and Black children in terms of enrollment and daily wear time. While wearable devices provide an opportunity for real-time, high-frequency contextual monitoring of individuals’ health, future studies should account for and address considerable representational bias in wearable data collection associated with demographic and SDoH factors.

Introduction

Over the last decade, consumer-grade wearable devices have gained popularity, with as many as 1 in 5 adults in the United States wearing smartwatches or fitness trackers.1 With lower costs and increasingly high-fidelity multimodal sensing, wearable devices can help assess the impact of frequent changes in physiological and behavioral patterns on personalized health outcomes.2

Researchers have used wearable devices across various health conditions and contexts, including monitoring patients with advanced cancer,3 daily stress,4 depression,5 and autism6; patient care management7,8; and aiding early detection of COVID-19.9 Given the high-frequency daily usage of mobile technologies, with smartphone ownership increasing from 69% to 91% between the ages of 11 and 18 years,10 wearable devices are particularly well-matched to better understanding youth health outcomes. Specifically, wearable devices could help to provide in-depth data on sleep and physical activity—important transdiagnostic risk factors for pediatric mental health disorders.11,12,13

However, despite the promise of wearable devices in health research, questions remain regarding potential biases in device deployment, data collection, and interpretation.14,15,16,17 There is limited evidence on whether long-term health monitoring using wearable devices will yield reliable and equitable real-world data18 across the diverse socioeconomic and demographic19,20 population spectrum,21,22 particularly among racial and ethnic minority groups.23 Research shows that social determinants of health (SDoHs)24 could potentially affect the acceptability of technology usage among racial and ethnic minority populations. The use of wearable devices in health research should also be evaluated in the context in which devices are worn daily. Fewer in-school resources,25 less safe housing and venues for physical activity,26 and lower socioeconomic status (SES) may lead to racial and ethnic minority youths spending more time on screen or media activities10 compared with their White counterparts. Racial and ethnic minority parents are also more likely to monitor their children’s device usage,10 so parental background and views on devices may affect the youth’s willingness or ability to engage.

Prior research also shows multiple structural and systemic barriers27 that may significantly affect the self-initiation of personal use of wearable devices among racial and ethnic minority groups. These include, but are not limited to, inequitable allocation of resources and access to health care technology,26,28,29 concerns about data accuracy30 and privacy,25 and language barriers.31 In addition, nonuniformity in willingness to join wearable device studies and share health data continually with frequent charging requirements and variation in sensor accuracy may also influence the equitable and representative collection of wearable device data from a large population.32,33,34,35 With research using consumer-grade wearable devices still in the early stages, there is a critical need to examine whether data from such devices can be collected equitably across large populations with varying age groups, SES, SDoH factors, and racial and ethnic backgrounds.

To empirically assess some of the challenges associated with large-scale wearable device data collection in children, we used the data collected from devices in the ongoing Adolescent Brain and Cognitive Development (ABCD) Study. The ABCD Study is the most extensive study of brain development and child health in the United States, with more than 11 000 children aged 9 to 10 years old recruited from 21 sites across the United States.36 In addition to episodic assessments consisting of neuroimaging, clinical interviews, and neuropsychological tests, the study used wearable devices during the year 2 (Y2) time point, collecting as much as 3 weeks of physiological data. To further understand factors potentially affecting wearable device data collection for children in naturalistic settings, we investigated the following specific questions: (1) can demographic and socioeconomic factors affect large-scale deployment and collection of wearable device data from children? and (2) is the difference in device wear time associated with children’s and their parents’ demographic and socioeconomic characteristics?

Methods

Study Design

We used the 4.0 data release from the ABCD Study36; specifically, data from 10 414 children enrolled at Y2 follow-up were included in the present analysis. As part of the main ABCD Study, parents provided written consent and children provided assent to participate, with centralized institutional review board approval obtained from the University of California San Diego’s institutional review board.37 The study recruited children from schools throughout the United States based on stratified probability38 to match the demographic and socioeconomic diversity of the United States.39 Further details on the ABCD Study protocol can be found in Karcher et al.38 This secondary cohort study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines and was approved by the research ethics board of the Center of Addiction and Mental Health (Toronto, Canada).

Primary Measures

Sociodemographic Characteristics

During onboarding and Y2 follow-up, children and their parents completed questions related to demographic and socioeconomic indicators (eg, race and ethnicity, age, gender, height, weight, income, education, marital and employment status). We also used the national percentile value of the area deprivation index (ADI)40 to account for variation in participants’ SDoH. ADI is a composite metric that helps compare US neighborhoods based on multiple socioeconomic indicators, including income, education, employment, and housing quality. ADI has been used to compare variations in health delivery and services, especially for the most disadvantaged neighborhood groups.41 The study recruitment sites were divided into 5 key geographical regions in the United States, depending on their location. eAppendix 1.1 in Supplement 1 includes more details on demographic and socioeconomic variables.

Wearable Device Data

As part of the Y2 follow-up, all participants were given the option to wear a device (Charge HR 2 [Fitbit]). Children who gave their assent and whose caregivers consented were given a device and instructed to wear it continuously, excluding bathing or water activities, over 3 weeks. Automated alerts were sent to participants if 3 days had passed without a sync or if 2 days had passed and the device was not sufficiently charged. If data-sharing issues persisted, research assistants reached out to participants. eAppendix 1.2 in Supplement 1 provides further details on device deployment and data collection.

Statistical Analysis

Demographic Analysis

Statistically significant differences in participants’ sociodemographic characteristics in different subcohorts were assessed using univariate χ2 and Kruskal-Wallis tests for categorical and continuous variables, respectively. Rank-based (Spearman) method was used to evaluate correlations between variables of interest.

Retention Analysis

Participant retention in the wearable device substudy was assessed based on the last date participants’ device data were shared within the 21-day study observation period. Kaplan-Meier (KM) survival curves42 to assess retention differences across sociodemographic and economic factors. The statistically significant differences in retention KM curves were evaluated using nonparametric log-rank test.43 A stratified version of the log-rank test was used to adjust for potential heterogeneity across study sites. With participant retention assessed in a short observation period (21 days) based on passively collected wearable data, we used the time to retain 75% of the cohort for comparing retention differences in variables of interest. eAppendix 1.3 in Supplement 1 includes further details on retention analysis.

Wear-Time Analysis

Per-minute heart rate (HR) data obtained from devices were used to approximate the total daily wear time (in hours) by calculating the time points for which the device recorded per-minute HR. Previous studies have used HR as a proxy for wear time22 such that time intervals with no acquired HR data were regarded as nonwear periods.44 We determined the first date of available data as the start of the 3-week observation period per participant. For all participants, day 1 was excluded due to varying start times of device wear, resulting in as many as 20 days of data being used for the present wear-time analysis. Given the multilevel structure of the data, we used a mixed-effects regression model45 to investigate potential associations between total wear time and participants’ demographic and socioeconomic factors. The key variables of interest, such as participants’ race and ethnicity, gender, parent’s marital status, education level, ADI percentile, and household income, were considered primary covariates (referred to as fixed effects) in the mixed-effects model. We used random intercepts for recruitment sites and study enrollment quarters to account for the potential clustering of participants within the same recruitment site or enrollment patterns during the COVID-19 pandemic. Mixed-effect models were implemented using lme4 package version 1.1-28.46 Statistical significance of the association of variables of interest with wear time was assessed using Satterthwaite approximation method, available from lmerTest package version 3.1-3.45 The best model fit was determined based on the Akaike information criterion. Variation inflation factor was used to assess model collinearity for all model covariates and met acceptable thresholds (<5).47 For sensitivity analysis, we compared the inference drawn from the retention and wear-time data collected from participants recruited before and during the COVID-19 pandemic. All analyses were performed using either R version 4.0.5 (R Project for Statistical Computing)48 or Python version 3.9.7 (Python Software Foundation). Statistical significance was assumed when with a 2-sided P < .05.

Results

Cohort Characteristics

The mean age (SD) of the 10 414 children enrolled when wearable devices were offered to all active participants was 12.00 (0.72) years. Of these, 5444 (52.3%) were male, and 5615 (53.9%) were White, with the largest racial and ethnic minority groups identifying as Hispanic or Latino (2048 [19.7%]), followed by Black (1424 [13.7%]). A notable proportion of parents (2349 [22.6%]) reported their annual household income to be less than $49 999, with nearly a majority (4599 [44.2%]) earning more than $100 000. Table 1 and eAppendix 2.1 in Supplement 1 further summarize the demographic characteristics of children and parents, respectively. No significant differences in sociodemographic characteristics of the participants were observed between those enrolled in the wearable device substudy during the COVID-19 pandemic (694 of 7424 participants [9.4%]) compared with participants enrolling in the study before COVID-19 (6546 participants [88.2%]) (eAppendix 2.2 in Supplement 1).

Table 1. Adolescent Brain and Cognitive Development Study Year 2 Cohort Demographic Characteristics.

Characteristic Participants, No. (%) P value Test performed (test value)
Overall Wearable data availability
No device Device
No. 10 414 2990 7424 NA NA
Age, mean (SD), y 12.00 (0.72) 12.11 (0.72) 11.96 (0.72) <.001 Kruskal-Wallis (88.453; df = 1)
Race and ethnicity
American Indian, Alaska Native, or Pacific Islander 208 (2.0) 50 (1.7) 158 (2.1) <.001 χ2 (208.37; df = 6)
Asian 201 (1.9) 64 (2.1) 137 (1.8)
Black 1424 (13.7) 577 (19.3) 847 (11.4)
Hispanic 2048 (19.7) 694 (23.2) 1354 (18.2)
Multiple racial and/or ethnic groups 807 (7.7) 255 (8.5) 552 (7.4)
White 5615 (53.9) 1314 (43.9) 4301 (57.9)
Othera 111 (1.1) 36 (1.2) 75 (1.0)
Gender
Male 5444 (52.3) 1605 (53.7) 3839 (51.7) .08 χ2 (6.7031; df = 3)
Female 4955 (47.6) 1384 (46.3) 3571 (48.1)
Otherb 9 (0.1) 0 (0.0) 9 (0.1)
Do not know or refused to answer 4 (0.0) 1 (0.0) 3 (0.0)
ADI percentile, mean (SD) 39.46 (26.49) 42.45 (28.33) 38.30 (25.65) <.001 Kruskal-Wallis (33.692; df = 1)
Open in a new tab

Abbreviations: ADI, area deprivation index; NA, not applicable.

a

Includes participants who did not or refused to provide an answer or indicated that they did not know.

b

Includes participants who indicated “Different,” “Gender queer,” or “Trans” for their gender.

Wearable Device Data Collection

Significant differences were seen between participants (7424 [71.3%]) who enrolled in the wearable device substudy (wearable device cohort [WDC]) compared with the 2990 (28.7%) who did not participate or share data (no wearable device cohort [NWDC]) (Figure 1 and Table 1). There was a significantly lower (−59%) relative proportion of Black children in the WDC (847 [11.4%]) than in the NWDC (577 [19.3%]; P < .001). In contrast, the WDC had a significantly higher (+132%) relative proportion of White children (4301 [57.9%]) than the NWDC (1314 [43.9%]; P < .001). Similar significant differences in relative proportion for Hispanic or Latino children were observed across the 2 cohorts (WDC, 1354 [18.2%]; NWDC, 694 [23.2%]; P < .001). Furthermore, a significantly lower proportion of children in the WDC (638 [8.6%]) were from households with income less than $24 999 compared with the NWDC (492 [16.5%]). Similarly, a notably lower proportion of parents whose children were in the WDC (1093 [14.7%]) had reached higher than an International Standard Classification of Education level of 3 (equivalent to grades 10-12 of high school) compared with the NWDC (705 [23.6%]). Table 1 and eAppendix 2.1 in Supplement 1 provide a complete comparison of differences in participant and parental demographic characteristics across WDC and NWDC.

Figure 1. Flowchart of Wearable Device Data Availability From the Overall Adolescent Brain and Cognitive Development (ABCD) Study Cohort.

Figure 1.

Open in a new tab

Participant Wearable Device–Based Retention

Marked differences in participant retention (last day of device wear) were observed in the WDC subcohort that shared data (6546 participants). At 75% cohort retention, Black children shared their data for a significantly shorter duration (16 days, 95% CI, 14-17 days; P < .001) compared with children from other racial and ethnic groups (White children: 21 days; 95% CI, 21-21 days; P < .001; range across groups, 18-21 days) (Figure 2A). Parents’ SES varied notably between retention levels. Children living in households with incomes less than $25 000 had the lowest retention, at 15 days (95% CI, 14-17 days) (Figure 2B). Children whose parents completed education level was between ISCED 1 to 3 (at most, upper high school) were retained for a significantly shorter period (17 days; 95% CI, 15-18 days; P < .001) compared with children whose parents had completed an ISCED 5 level of education (associate’s degree equivalent in the United States) (20 days; 95% CI, 19-20 days). Sensitivity analysis showed similar differences in retention for participants recruited during the COVID-19 pandemic (eAppendix 2.3 in Supplement 1).

Figure 2. Kaplan-Meier Curves of Participant Retention Based on Device Wear in the 21-Day Observation Period, by Sociodemographic and Socioeconomic Factors.

Figure 2.

Open in a new tab

Kaplan-Meier curves showed significant variation in participant retention based on device wear in the 21-day observation period by (A) participants’ race, (B) household income, and (C) parental education (based on International Standard Classification of Education [ISCED] levels). AIAN/P indicates American Indian, Alaska Native, and Pacific Islander.

Association Between Device Wear Time and Socioeconomic Indicators

We investigated factors associated with participants’ total device wear time during the observation period using a mixed-effects model. The median (IQR) total device wear time within the 20-day observation period was 400.7 hours (286.4-446.0 hours). However, participants from racial and ethnic minority groups wore their devices significantly less than their White counterparts (Black vs White children: β = −43.00 hours [95% CI, −55.11 to −30.88 hours]; P < .001) (Figure 3A and Table 2). Notably, study sites in the southwestern region of the United States had lower wear times (Southwest vs Midwest: β = −24.1 hours [95% CI: −43.77 to −4.43 hours]; P = .02), whereas sites in the western region of the US had the highest wear times (West vs Midwest: β = 4.07 hours [95% CI, −16.46 to 24.59 hours]; P < .001) (Figure 3B and Table 2). eAppendix 2.4 in Supplement 1 includes more details.

Figure 3. Bar Plots of Total Wear Time by Sociodemographic and Study-Related Factors.

Figure 3.

Open in a new tab

Total wear time was by (A) race and ethnicity, (B) Adolescent Brain and Cognitive Development Study site location, and (C) participant enrollment period. C, Shading indicates months affected by the COVID-19 pandemic, with the dotted blue and orange lines showing the median wear time during and before the pandemic. CUB indicates University of Colorado Boulder; FIU, Florida International University; LIBR, Laureate Institute for Brain Research; MUSC, Medical University of South Carolina; OHSU, Oregon Health & Science University; Q, quarter; ROC, University of Rochester; SRI, SRI International; UCLA, University of California, Los Angeles; UFL, University of Florida; UMB, University of Maryland at Baltimore; UMICH, University of Michigan; UMN, University of Minnesota; UPMC, University of Pittsburgh; UTAH, University of Utah; UVM, University of Vermont; UWM, University of Wisconsin-Milwaukee; VCU, Virginia Commonwealth University; WUSTL, Washington University in St Louis; and YALE, Yale University.

Table 2. Mixed-Effects Model Examining Confidence Intervals for Linear Mixed-Effect Model.

Factor Differential wear time (95% CI), h P value
Fixed effects
Intercept 361.36 (336.79 to 385.92) <.001
Race/ethnicity
White 1 [Reference] NA
American Indian, Alaska Native, or Pacific Islander −9.84 (−30.76 to 11.08) .04
Asian 23.78 (−0.29 to 47.85) .05
Black −43.00 (−55.11 to −30.88) <.001
Hispanic −4.39 (−14.61 to 5.83) .40
Multiple racial and/or ethnic groups −1.99 (−13.97 to 9.99) .75
Other −22.61 (−53.30 to 8.08) .15
Gender
Male 1 [Reference] NA
Different, gender queer, or trans 44.96 (−51.77 to 141.70) .36
Do not know or refused to answer 29.80 (−106.79 to 166.40) .67
Female 11.92 (5.85 to 17.99) <.001
Weight category (BMI percentile)
Healthy (5th to <85th) 1 [Reference] NA
Missing −9.95 (−30.39 to 10.49) .34
Obesity (≥95th) −8.97 (−17.80 to −0.15) .046
Overweight (85th to ≤95th) −9.35 (−18.16 to −0.54) .04
Underweight (<5th) 1.64 (−14.10 to 17.37) .84
Parental marital status
Married 1 [Reference] NA
Divorced, separated, widowed −26.93 (−36.36 to −17.49) <.001
Living with partner −27.64 (−41.82 to −13.47) <.001
Never married −25.22 (−38.48 to −11.97) <.001
Household income, $
50 000-74 999 1 [Reference] NA
<25 000 −25.51 (−40.73 to −10.28) .001
25 000-49 999 −3.45 (−16.21 to 9.30) .60
75 000-99 999 −5.49 (−17.35 to 6.36) .36
>100 000 −11.04 (−21.43 to −0.65) .04
Do not know −8.80 (−29.00 to 11.41) .39
Refused to answer −29.89 (−48.66 to −11.13) .002
Parental education
ISCED 8 1 [Reference] NA
ISCED 1-3 −29.55 (−45.95 to −13.15) <.001
ISCED 5 −19.19 (−33.24 to −5.13) .007
ISCED 6 −9.16 (−22.30 to 3.98) .17
ISCED 7 −5.23 (−18.64 to 8.19) .45
ADI quartile (percentiles)
4th (75th to 100th) 1 [Reference] NA
1st (1st to 24th) 9.67 (−3.68 to 23.03) .16
2nd (25th to 49th) 16.38 (4.34 to 28.41) .008
3rd (50th to 74th) 12.34 (0.27 to 24.41) .045
Geographical region
Midwest 1 [Reference] NA
Northeast −7.94 (−32.09 to 16.21) .52
Southeast −24.1 (−43.77 to −4.43) .02
Southwest −8.54 (−41.41 to 24.33) .61
West 4.07 (−16.46 to 24.59) .70
Random effects
σ2 14456.37 NA
τ00 Site location 192.31 NA
τ00 Enrollment quarter 442.34 NA
ICC 0.04 NA
Site locations, No. 19 NA
Enrollment quarters, No. 9 NA
Total observations, No. 6114 NA
Marginal R2 (conditional R2) 0.070 (0.109) NA
Open in a new tab

Abbreviations: ADI, area deprivation index; ICC, intraclass correlation; ISCED, International Standard Classification of Education; NA, not applicable.

In addition, device wear time was substantially lower for the population recruited during the COVID-19 pandemic than before the pandemic period (Figure 3C). eAppendix 2.2.2 in Supplement 1 includes more details regarding enrollment during the COVID-19 pandemic. We also observed 4 distinct longitudinal device wear patterns from the total daily wear time (eAppendix 2.5 in Supplement 1).

Discussion

Our findings using more than 100 000 days of wearable data from one of the largest and most diverse wearable device deployments among more than 10 000 children nationwide show significant differences in data collection association with demographic and socioeconomic indicators. Specifically, (1) a significantly lower proportion of racial and ethnic minority youths enrolled into the wearable device substudy compared with White youths; (2) racial and ethnic minority youths, in particular Black children, and those from lower SES households, shared data for fewer days (ie, lower retention in the 21-day observation period) and had significantly less total wear time compared with White children or those from higher SES households; (3) wear time was associated with external factors, such as recruitment site; and (4) the COVID-19 pandemic was associated with device wear time. As consumer-grade wearable devices become more common in health research, these data-driven findings could help to guide the development of strategies for reaching, recruiting, and retaining participants from racial and ethnic minority groups and lower SES populations.

It is well known that racial and ethnic minority participants are underrepresented in clinical trials.49 Our analyses reflected this underrepresentation, with the relative proportion of the total Black children (59% lower WDC vs NWDC) enrolled having wearable device data available being significantly lower than that of White children (132% higher in WDC vs NWDC) (Table 1). As the use of smart devices widens in health research and care, there is an urgent need to address known technical,30 structural, and systemic barriers27 to enable equitable and representative use of technology for health data collection. Structural racism is also deeply rooted in societal structures that control power and resources, leading to discriminatory and inequitable systems that can reinforce discriminatory beliefs and values. The inequitable allocation of resources (eg, the education system)50 could be due to multiple factors, including reduced neighborhood safety; access to green space, health care, and education; and diminished opportunities for upward mobility, resulting in individuals residing in disadvantaged neighborhoods for longer.32 These factors can negatively affect physical activity levels, particularly for Black communities, which have historically lower trust in health care and research51,52 due to historical discrimination from significant events such as Sims, Lacks, and Tuskegee.53

There was also poorer retention of Black compared with White families, which likely contributed to lower overall wear time. Future wearable device research should address various sociotechnical and human factors to improve wearable device–based data collection. These can range from participants’ and their families’ understanding of the study, consent and assent language, the potential influence of personal data sharing, its secondary usage, and their overall trust in the study team.32 Characteristics of the study team may also inform whether participants opt in to studies, as racial and ethnic minority communities are more likely to trust those that are like them and understand their experiences.54,55,56,57 Failure to do so may result in findings that reinforce negative stereotypes in racial and ethnic minority youths,58 becoming more detrimental than beneficial, specifically for their communities. Finally, the present wearable technology is known to be less reliable for people with darker skin tones and higher body mass index (eg, inaccurate HR measurement).30 The interplay of many such sociotechnical factors can lead to nonuniform participant engagement and data collection among racial and ethnic minority populations, which may result in the learning of treatment paradigms that are not appropriate or inadequate for certain underrepresented populations and may further contribute to health disparities in disease and treatment outcomes and access to care.

Despite some of the above challenges, wearable devices may still provide an avenue for large-scale health monitoring in real-world settings guiding early detection and intervention opportunities. Personalized just-in-time adaptive interventions59 (JITAIs) are one promising digital approach that processes in-the-moment wearable data and provides timely and appropriate interventions when support is most needed. JITAIs may improve the effectiveness of evidence-based interventions by identifying vulnerable states with passive measures of treatment targets (eg, sleep, physical activity) and providing in-the-moment intervention prompts to affect proximal treatment targets and improve distal outcomes (eg, psychopathology symptoms). These have been used to treat a variety of adolescent health issues60,61,62,63,64 effectively. While less studied, JITAIs in the context of mental health care is an area of interest identified by the National Institute of Mental Health63 and holds promise for changing the landscape of child and adolescent mental health care. However, as evidenced in the present data set, there is an urgent need to understand and address SDoH factors, particularly for racial and ethnic minority communities or those with lower SES, such that they are recruited and retained equitably to their White counterparts.

Limitations

The analysis of retention and wear time of wearable devices should be viewed in the context of certain limitations linked to the deployment of these devices. First, our approximation of wear time was based on available per-minute HR values, which have been used in previous studies22; however, wearable device data are affected by motion artifacts.64,65 Thus, our measure of wear time may be an imperfect proxy. The HR sensor can also be affected by skin tone, which may bias wearable devices’ data against racial and ethnic minority youths who may have darker skin tones.64 In addition, racial and ethnic minority groups are disproportionately affected by diabetes and obesity,66 which can affect HR measurement.67 Other variables, such as gender affecting cutaneous blood flow,68 may also affect data collected from wearable devices. Further research is needed to assess wear time accurately by fusing multimodal sensor data. Second, the wearable devices used in this study were not certified to be worn by those younger than 13 years, and most participants were 11 to 12 years of age at the time of data collection. While more research is necessary to understand how these devices perform with younger children, they have been found to have adequate performance.20 Third, nearly one-third (28.7%) of the Y2 cohort did not have wearable device data (eAppendix 2.6.2 in Supplement 1); it was unclear how many opted out or were unable to participate in the substudy due to factors, such as participants not being offered a wearable device due to site-specific availability or not being mailed a device for remote assessment due to the COVID-19 pandemic. The latter may have affected wear time, especially for racial and ethnic minority youths who may have been disproportionately affected by psychosocial, economic, and health issues, thus having a diminished capacity to engage with the ABCD Study.

Conclusions

In one of the largest deployments of wearable devices in children, we found evidence that children from lower SES backgrounds and Black children were less likely to participate, share data, and engage in wearable device–based health research than those with higher SES backgrounds and White children. With the use of wearable devices being an emerging area of health research, it is critical to understand and address factors that can significantly affect wearable device data collection for robust and generalizable evidence generation.

Supplement 1.

eAppendix 1. Supplementary Methods

eAppendix 2. Supplementary Results

eReferences.

Click here for additional data file. (876.2KB, pdf)
Supplement 2.

Data Sharing Statement

Click here for additional data file. (18.3KB, pdf)

References

  • 1.Vogels EA. About one-in-five Americans use a smart watch or fitness tracker. Pew Research Center. Published January 9, 2020. Accessed May 18, 2022. https://www.pewresearch.org/fact-tank/2020/01/09/about-one-in-five-americans-use-a-smart-watch-or-fitness-tracker/
  • 2.de Zambotti M, Cellini N, Menghini L, Sarlo M, Baker FC. Sensors capabilities, performance, and use of consumer sleep technology. Sleep Med Clin. 2020;15(1):1-30. doi: 10.1016/j.jsmc.2019.11.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Gresham G, Hendifar AE, Spiegel B, et al. Wearable activity monitors to assess performance status and predict clinical outcomes in advanced cancer patients. NPJ Digit Med. 2018;1(1):27. doi: 10.1038/s41746-018-0032-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Choi Y, Jeon YM, Wang L, Kim K. A biological signal-based stress monitoring framework for children using wearable devices. Sensors (Basel). 2017;17(9):E1936. doi: 10.3390/s17091936 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Rykov Y, Thach TQ, Bojic I, Christopoulos G, Car J. Digital biomarkers for depression screening with wearable devices: cross-sectional study with machine learning modeling. JMIR Mhealth Uhealth. 2021;9(10):e24872. doi: 10.2196/24872 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Black MH, Milbourn B, Chen NTM, et al. The use of wearable technology to measure and support abilities, disabilities and functional skills in autistic youth: a scoping review. Scand J Child Adolesc Psychiatr Psychol. 2020;8:48-69. doi: 10.21307/sjcapp-2020-006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Heintzman ND. A digital ecosystem of diabetes data and technology: services, systems, and tools enabled by wearables, sensors, and apps. J Diabetes Sci Technol. 2015;10(1):35-41. doi: 10.1177/1932296815622453 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Hernandez-Silveira M, Ahmed K, Ang SS, et al. Assessment of the feasibility of an ultra-low power, wireless digital patch for the continuous ambulatory monitoring of vital signs. BMJ Open. 2015;5(5):e006606. doi: 10.1136/bmjopen-2014-006606 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Mason AE, Hecht FM, Davis SK, et al. Detection of COVID-19 using multimodal data from a wearable device: results from the first TemPredict Study. Sci Rep. 2022;12(1):3463. doi: 10.1038/s41598-022-07314-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Rideout V, Robb MB. The Common Sense census: Media use by tweens and teens. October 28, 2019. Accessed February 21, 2023. https://www.commonsensemedia.org/research/the-common-sense-census-media-use-by-tweens-and-teens-2019
  • 11.Ahn S, Fedewa AL. A meta-analysis of the relationship between children’s physical activity and mental health. J Pediatr Psychol. 2011;36(4):385-397. doi: 10.1093/jpepsy/jsq107 [DOI] [PubMed] [Google Scholar]
  • 12.Harvey AG, Murray G, Chandler RA, Soehner A. Sleep disturbance as transdiagnostic: consideration of neurobiological mechanisms. Clin Psychol Rev. 2011;31(2):225-235. doi: 10.1016/j.cpr.2010.04.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Rodriguez-Ayllon M, Cadenas-Sánchez C, Estévez-López F, et al. Role of physical activity and sedentary behavior in the mental health of preschoolers, children and adolescents: a systematic review and meta-analysis. Sports Med. 2019;49(9):1383-1410. doi: 10.1007/s40279-019-01099-5 [DOI] [PubMed] [Google Scholar]
  • 14.de Arriba-Pérez F, Caeiro-Rodríguez M, Santos-Gago JM. Collection and processing of data from wrist wearable devices in heterogeneous and multiple-user scenarios. Sensors (Basel). 2016;16(9):1538. doi: 10.3390/s16091538 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Ometov A, Shubina V, Klus L, et al. A survey on wearable technology: history, state-of-the-art and current challenges. Comput Netw. 2021;193:108074. doi: 10.1016/j.comnet.2021.108074 [DOI] [Google Scholar]
  • 16.Pitt L, Kietzmann J, Robson K, et al. Understanding the opportunities and challenges of wearable technology. In: Stieler M, ed. Creating Marketing Magic and Innovative Future Marketing Trends. Springer; 2017:139-141. doi: 10.1007/978-3-319-45596-9_29 [DOI] [Google Scholar]
  • 17.Balbim GM, Marques IG, Marquez DX, et al. Using Fitbit as an mHealth intervention tool to promote physical activity: potential challenges and solutions. JMIR Mhealth Uhealth. 2021;9(3):e25289. doi: 10.2196/25289 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.US Food and Drug Administration. Real-world evidence. December 12, 2022. Accessed December 14, 2022. https://www.fda.gov/science-research/science-and-research-special-topics/real-world-evidence
  • 19.Huang J, Galal G, Etemadi M, Vaidyanathan M. Evaluation and mitigation of racial bias in clinical machine learning models: scoping review. JMIR Med Inform. 2022;10(5):e36388. doi: 10.2196/36388 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Godino JG, Wing D, de Zambotti M, et al. Performance of a commercial multi-sensor wearable (Fitbit Charge HR) in measuring physical activity and sleep in healthy children. PLoS One. 2020;15(9):e0237719. doi: 10.1371/journal.pone.0237719 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.US Centers for Disease Control and Prevention. Growth charts. January 31, 2019. Accessed June 13, 2022. https://www.cdc.gov/growthcharts/index.htm
  • 22.Gorny AW, Liew SJ, Tan CS, Müller-Riemenschneider F. Fitbit Charge HR wireless heart rate monitor: validation study conducted under free-living conditions. JMIR Mhealth Uhealth. 2017;5(10):e157. doi: 10.2196/mhealth.8233 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Colvonen PJ, DeYoung PN, Bosompra NA, Owens RL. Limiting racial disparities and bias for wearable devices in health science research. Sleep. 2020;43(10):zsaa159. doi: 10.1093/sleep/zsaa159 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Sieck CJ, Sheon A, Ancker JS, Castek J, Callahan B, Siefer A. Digital inclusion as a social determinant of health. NPJ Digit Med. 2021;4(1):52. doi: 10.1038/s41746-021-00413-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Kuhn AW, Grusky AZ, Cash CR, Churchwell AL, Diamond AB. Disparities and inequities in youth sports. Curr Sports Med Rep. 2021;20(9):494-498. doi: 10.1249/JSR.0000000000000881 [DOI] [PubMed] [Google Scholar]
  • 26.Hacker K, Houry D. Social needs and social determinants: the role of the Centers for Disease Control and Prevention and public health. Public Health Rep. 2022;137(6):1049-1052. doi: 10.1177/00333549221120244 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Nagata JM, Ganson KT, Iyer P, et al. Sociodemographic correlates of contemporary screen time use among 9- and 10-year-old children. J Pediatr. 2022;240:213-220.e2. doi: 10.1016/j.jpeds.2021.08.077 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Riley WJ. Health disparities: gaps in access, quality and affordability of medical care. Trans Am Clin Climatol Assoc. 2012;123:167-172. [PMC free article] [PubMed] [Google Scholar]
  • 29.Holko M, Litwin TR, Munoz F, et al. Wearable fitness tracker use in federally qualified health center patients: strategies to improve the health of all of us using digital health devices. NPJ Digit Med. 2022;5(1):53. doi: 10.1038/s41746-022-00593-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Shcherbina A, Mattsson CM, Waggott D, et al. Accuracy in wrist-worn, sensor-based measurements of heart rate and energy expenditure in a diverse cohort. J Pers Med. 2017;7(2):3. doi: 10.3390/jpm7020003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Greenwood S. Facts on US immigrants, 2018. Pew Research Center. August 20, 2020. Accessed February 12, 2023. https://www.pewresearch.org/hispanic/2020/08/20/facts-on-u-s-immigrants/
  • 32.Pratap A, Allred R, Duffy J, et al. Contemporary views of research participant willingness to participate and share digital data in biomedical research. JAMA Netw Open. 2019;2(11):e1915717. doi: 10.1001/jamanetworkopen.2019.15717 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Yeager CM, Benight CC. If we build it, will they come? issues of engagement with digital health interventions for trauma recovery. Mhealth. 2018;4:37. doi: 10.21037/mhealth.2018.08.04 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Minen MT, Stieglitz EJ. Wearables for neurologic conditions: considerations for our patients and research limitations. Neurol Clin Pract. 2021;11(4):e537-e543. doi: 10.1212/CPJ.0000000000000971 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.O’Connor S, Hanlon P, O’Donnell CA, Garcia S, Glanville J, Mair FS. Understanding factors affecting patient and public engagement and recruitment to digital health interventions: a systematic review of qualitative studies. BMC Med Inform Decis Mak. 2016;16(1):120. doi: 10.1186/s12911-016-0359-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Jernigan T, Brown SA, Dale AM, et al. Adolescent Brain Cognitive Development Study (ABCD)—annual release 4.0 #1299. doi: 10.15154/1523041 [DOI]
  • 37.Auchter AM, Hernandez Mejia M, Heyser CJ, et al. A description of the ABCD organizational structure and communication framework. Dev Cogn Neurosci. 2018;32:8-15. doi: 10.1016/j.dcn.2018.04.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Karcher NR, Barch DM. The ABCD study: understanding the development of risk for mental and physical health outcomes. Neuropsychopharmacology. 2021;46(1):131-142. doi: 10.1038/s41386-020-0736-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Thompson WK, Barch DM, Bjork JM, et al. The structure of cognition in 9 and 10 year-old children and associations with problem behaviors: findings from the ABCD study’s baseline neurocognitive battery. Dev Cogn Neurosci. 2019;36:100606. doi: 10.1016/j.dcn.2018.12.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Kind AJH, Buckingham WR. Making neighborhood-disadvantage metrics accessible—The Neighborhood Atlas. N Engl J Med. 2018;378(26):2456-2458. doi: 10.1056/NEJMp1802313 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Knighton AJ, Savitz L, Belnap T, Stephenson B, VanDerslice J. Introduction of an Area Deprivation Index measuring patient socioeconomic status in an integrated health system: implications for population health. EGEMS (Wash DC). 2016;4(3):1238. doi: 10.13063/2327-9214.1238 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Rich JT, Neely JG, Paniello RC, Voelker CCJ, Nussenbaum B, Wang EW. A practical guide to understanding Kaplan-Meier curves. Otolaryngol Head Neck Surg. 2010;143(3):331-336. doi: 10.1016/j.otohns.2010.05.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Bland JM, Altman DG. The logrank test. BMJ. 2004;328(7447):1073. doi: 10.1136/bmj.328.7447.1073 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Collins JE, Yang HY, Trentadue TP, Gong Y, Losina E. Validation of the Fitbit Charge 2 compared to the ActiGraph GT3X+ in older adults with knee osteoarthritis in free-living conditions. PLoS One. 2019;14(1):e0211231. doi: 10.1371/journal.pone.0211231 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Kuznetsova A, Brockhoff PB, Christensen RHB. lmerTest package: tests in linear mixed effects models. J Stat Softw. 2017;82:1-26. doi: 10.18637/jss.v082.i13 [DOI] [Google Scholar]
  • 46.Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed-effects models using lme4. J Stat Softw. 2015;67:1-48. doi: 10.18637/jss.v067.i01 [DOI] [Google Scholar]
  • 47.Fox J, Monette G. Generalized collinearity diagnostics. J Am Stat Assoc. 1992;87(417):178-183. doi: 10.1080/01621459.1992.10475190 [DOI] [Google Scholar]
  • 48.R Core Team . R: a language and environment for statistical computing. Accessed February 24, 2023. https://www.R-project.org/
  • 49.US Food and Drug Administration. FDA issues summary report on global participation in clinical trials: 2015-2019—drug information update. November 9, 2020. Accessed June 19, 2022. https://content.govdelivery.com/accounts/USFDA/bulletins/2aacc22
  • 50.Aasland E, Engelsrud G. Structural discrimination in physical education: the “encounter” between the (White) Norwegian teaching content in physical education lessons and female students of color’s movements and expressions. Frontiers in Sports and Active Living. November 15, 2021. Accessed January 26, 2023. https://www.frontiersin.org/articles/10.3389/fspor.2021.769756 [DOI] [PMC free article] [PubMed]
  • 51.Boulware LE, Cooper LA, Ratner LE, LaVeist TA, Powe NR. Race and trust in the health care system. Public Health Rep. 2003;118(4):358-365. doi: 10.1016/S0033-3549(04)50262-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Jacobs EA, Rolle I, Ferrans CE, Whitaker EE, Warnecke RB. Understanding African Americans’ views of the trustworthiness of physicians. J Gen Intern Med. 2006;21(6):642-647. doi: 10.1111/j.1525-1497.2006.00485.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Baptiste DL, Caviness-Ashe N, Josiah N, et al. Henrietta Lacks and America’s dark history of research involving African Americans. Nurs Open. 2022;9(5):2236-2238. doi: 10.1002/nop2.1257 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Benkert R, Peters RM, Clark R, Keves-Foster K. Effects of perceived racism, cultural mistrust and trust in providers on satisfaction with care. J Natl Med Assoc. 2006;98(9):1532-1540. [PMC free article] [PubMed] [Google Scholar]
  • 55.Pugh M Jr, Perrin PB, Rybarczyk B, Tan J. Racism, mental health, healthcare provider trust, and medication adherence among Black patients in safety-net primary care. J Clin Psychol Med Settings. 2021;28(1):181-190. doi: 10.1007/s10880-020-09702-y [DOI] [PubMed] [Google Scholar]
  • 56.Schwei RJ, Kadunc K, Nguyen AL, Jacobs EA. Impact of sociodemographic factors and previous interactions with the health care system on institutional trust in three racial/ethnic groups. Patient Educ Couns. 2014;96(3):333-338. doi: 10.1016/j.pec.2014.06.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Jacobs EA, Mendenhall E, Mcalearney AS, et al. An exploratory study of how trust in health care institutions varies across African American, Hispanic and White populations. Commun Med. 2011;8(1):89-98. doi: 10.1558/cam.v8i1.89 [DOI] [PubMed] [Google Scholar]
  • 58.Corbie-Smith G, Moody-Ayers S, Thrasher AD. Closing the circle between minority inclusion in research and health disparities. Arch Intern Med. 2004;164(13):1362-1364. doi: 10.1001/archinte.164.13.1362 [DOI] [PubMed] [Google Scholar]
  • 59.Nahum-Shani I, Smith SN, Spring BJ, et al. Just-in-time adaptive interventions (JITAIs) in mobile health: key components and design principles for ongoing health behavior support. Ann Behav Med. 2018;52(6):446-462. doi: 10.1007/s12160-016-9830-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Haug S, Paz Castro R, Scholz U, Kowatsch T, Schaub MP, Radtke T. Assessment of the efficacy of a mobile phone-delivered just-in-time planning intervention to reduce alcohol use in adolescents: randomized controlled crossover trial. JMIR Mhealth Uhealth. 2020;8(5):e16937. doi: 10.2196/16937 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Cerrada CJ, Dzubur E, Blackman KCA, Mays V, Shoptaw S, Huh J. Development of a just-in-time adaptive intervention for smoking cessation among Korean American emerging adults. Int J Behav Med. 2017;24(5):665-672. doi: 10.1007/s12529-016-9628-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Partridge SR, Redfern J. Strategies to engage adolescents in digital health interventions for obesity prevention and management. Healthcare (Basel). 2018;6(3):70. doi: 10.3390/healthcare6030070 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.National Institute of Mental Health. Just-in-time adaptive interventions to optimize adolescent mental health treatments . May 18, 2021. Accessed July 4, 2022. https://www.nimh.nih.gov/funding/grant-writing-and-application-process/concept-clearances/2021/just-in-time-adaptive-interventions-to-optimize-adolescent-mental-health-treatments
  • 64.Bent B, Goldstein BA, Kibbe WA, Dunn JP. Investigating sources of inaccuracy in wearable optical heart rate sensors. NPJ Digit Med. 2020;3(1):18. doi: 10.1038/s41746-020-0226-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65.Lee J, Kim M, Park HK, Kim IY. Motion artifact reduction in wearable photoplethysmography based on multi-channel sensors with multiple wavelengths. Sensors (Basel). 2020;20(5):1493. doi: 10.3390/s20051493 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 66.Walker RJ, Strom Williams J, Egede LE. Influence of race, ethnicity and social determinants of health on diabetes outcomes. Am J Med Sci. 2016;351(4):366-373. doi: 10.1016/j.amjms.2016.01.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.Ajmal B-AT, Boonya-Ananta T, Rodriguez AJ, Du Le VN, Ramella-Roman JC. Monte Carlo analysis of optical heart rate sensors in commercial wearables: the effect of skin tone and obesity on the photoplethysmography (PPG) signal. Biomed Opt Express. 2021;12(12):7445-7457. doi: 10.1364/BOE.439893 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Chin LC, Huang TY, Yu CL, Wu CH, Hsu CC, Yu HS. Increased cutaneous blood flow but impaired post-ischemic response of nutritional flow in obese children. Atherosclerosis. 1999;146(1):179-185. doi: 10.1016/S0021-9150(99)00135-5 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement 1.

eAppendix 1. Supplementary Methods

eAppendix 2. Supplementary Results

eReferences.

Click here for additional data file. (876.2KB, pdf)
Supplement 2.

Data Sharing Statement

Click here for additional data file. (18.3KB, pdf)

Articles from JAMA Network Open are provided here courtesy of American Medical Association

RESOURCES