Excessive screen time for children is linked with poor sleep, inactivity, and behavioral problems [1-4], and only few children meet the World Health Organization’s recommendation that children under 5 years receive ≤1 hour of screen time per day [5,6]. The ways in which families use screens have changed dramatically over the past 2 decades, in large part owing to the introduction of mobile and interactive digital media devices. Historically, children’s screen time comprised mostly television use, which took place mainly at home. With new forms of digital media such as mobile phones and tablets, families can now use screens across many different locations and contexts. The pervasive availability of mobile devices has resulted in a dramatic increase in children’s use of mobile media. Children’s mobile screen time increased from 5 minutes per day in 2011 to 55 minutes per day in 2020, with nearly half (46%) of the children aged 2 to 4 years and more than two-thirds (67%) of the children aged 5 to 8 years having their own mobile device (tablet or smartphone) [7]. Videos on the web now constitute two-thirds of children’s screen viewing (66%), supplanting traditional television, which now accounts for only 23% of the average daily video screen time [7]. Unfortunately, our understanding of mobile screen use and its impact on child health lags behind the accelerated adoption of mobile technology.

Measuring screen time is complicated and has historically relied upon time-consuming, expensive observational methods conducted by highly trained staff or upon parent recall survey methods [8-12]. In a 2021 systematic review of young (aged 0-6 years) children’s screen time, none of the 622 studies used an objective measure of screen time [13]. However, there was only a moderate correlation between self-reported and objectively observed media use (r=0.38). Of the 49 studies that examined adults’ digital media–use behaviors, only 3 were within 5% of their objectively measured mean [10]. Another important limitation of commonly used screen time measures is their provision of global estimates of screen use, which is often gathered from a single question that asks parents to estimate television use in a typical day [14,15]. This global measure of screen time precludes the ability to examine the context or timing of screen use within and across days, thus limiting the understanding of the complex and changing patterns of children’s daily screen use. Therefore, updated, objective, and low-burden measures of screen use and its context are needed.

Recent developments in the passive monitoring of mobile screen use have the potential to provide novel insights into how, when, where, and why children and families use digital media (ie, mobile phones and tablets) throughout the day [16-18]. This method of digital media monitoring is both less expensive than direct observation and less burdensome for participants than self-reporting measures [15]. Similarly, advances in ecological momentary assessments (EMAs) allow for the assessment of the context and timing of behaviors while minimizing recall bias. Finally, advances in accelerometry have made it possible to passively collect continuous 24-hour data on children’s sedentary behavior, physical activity, and sleep over extended periods (ie, 30 days). While these technologies have been tested in isolation [8,16,19,20], to date, no studies have leveraged multiple streams of data from passive mobile sensing, EMA, and accelerometry to specifically study young children’s screen time [13,21]. It is unclear whether the confluence of these methods can be successfully used to gather meaningful information about the digital media use of children and families. Furthermore, the emerging literature on passive mobile sensing has been conducted with high-income White families and may not be generalizable [8,16].

We aim to evaluate the feasibility and utility of a comprehensive multibehavior protocol to measure digital media use and screen time context among a racially and economically diverse sample of preschoolers and their families. Figure 1 depicts the conceptual model of the associations between screen time context and children’s behaviors.

This study includes 100 caregiver-child dyads and uses a longitudinal, observational, dyadic, case-crossover design [22]. Using a case-crossover design allows a dyad to serve as its own control to assess the within-day effects of immediate antecedents on a dependent variable measured multiple times throughout the day and week [22]. Caregivers and their children aged 3 to 5 years will participate in a 30-day assessment protocol, with 50% (50/100) of the sample invited to participate in a second wave of data collection 3 to 12 months later to evaluate the feasibility of retaining the sample over time and to assess the longer-term acceptability of the protocol.

Participants include racially and economically diverse caregivers and preschoolers in the greater Southeastern United States. The caregiver inclusion criteria are as follows: (1) being a primary caregiver of a child between the ages of 3 and 5 years, (2) owning a smartphone device, and (3) being able to read and speak English. The exclusion criteria for children include a diagnosis of a severe developmental or physical disorder that would prevent ambulation. This decision was made because of the inability to recruit a large enough sample of children to draw meaningful conclusions.

The study maintains rolling recruitment, with each dyad on their own assessment schedule, depending on the time of enrollment. Enrollment was scheduled to take place between September 2020 and May 2022.

We aim to recruit a nonrandom volunteer sample by posting fliers at daycare centers, pediatric clinics, and community centers such as food banks, as well as Facebook advertising in the form of “boosted” posts. To obtain a socioeconomically diverse sample, we will partner with daycares serving low-income families and prioritize the enrollment of low-income families. In addition, we will use snowball recruitment methods where participants are compensated for referring families who then successfully participate in the study.

Data collection is completely remote, in part driven by the necessary COVID-19 pandemic protocol adjustments [23]. Interested participants will be directed to an informational website through a QR code or hyperlink. The informational website describes the study procedures and research participant protections. The informational website includes a web-based consent form and directs interested participants to a short screener survey to assess initial study eligibility. A trained member of the research team will contact interested and eligible families by phone to answer any remaining questions and verbally confirm their desire to participate.

Following recruitment, the eligible child and caregiver dyads are texted a Qualtrics link to the baseline survey. The baseline questionnaire is designed to be completed within approximately 30 minutes. After caregivers complete the baseline survey, they are sent instructions to download a screen time monitoring app (Chronicle; Android devices) [18] or upload screenshots (iOS devices) depending on the make of their smartphone. Technical support is provided as required by the research team. Caregivers are then mailed 2 Axivity AX3 accelerometers (Axivity Ltd), one for themselves and one for their preschool-age child. Caregivers and their children are asked to wear the Axivity AX3 and monitor their screen time for 30 days. We will conduct a 30-day assessment to examine the higher limit of time that the families are willing to wear an activity watch and monitor their digital media use. The first week of the 30-day monitoring period includes 7 days of EMAs, which are texted to the primary caregivers’ smartphones. EMAs are limited to the first week (vs the entire monitoring period) to minimize participant burden, as this portion of the study requires active participation. The first week of the 30-day assessment was selected to allow for flexibility because of any potential issues that might prevent a participant from completing EMAs (eg, work conflicts). Participants who do not complete enough EMAs to earn a gift card during the first week (Compensation section) are offered an extension to complete additional EMAs. In line with the recent EMA studies [20], the Tots and Tech study allows flexibility in the survey completion window in an effort to retain families from diverse backgrounds who might experience additional barriers to completing measures within the time frame. Following the 30-day monitoring protocol, caregivers will complete a semistructured qualitative interview about their experiences. Participants who complete the study protocol (eg, do not drop out) in the first year (wave 1) are randomly selected (50/100, 50%) to be recontacted 3 to 12 months after the initial evaluation and invited to repeat the entire research protocol for additional compensation. Families that decline to participate in the second assessment are asked to complete a short web-based survey regarding their experiences and reasons for declining. The number of families that decline or are lost to follow-up is documented to assess protocol feasibility.

Caregivers will complete the EMA surveys in the first week of the 30-day protocol. EMA surveys are sent to caregivers via SMS text messages to their smartphones. The SMS text messages contain a Qualtrics survey link to assess the domains using prompts listed in Table 1. Additional items that are only included in the end-of-day survey are listed in Table 2.

Caregivers are informed that they will receive 4 “short surveys” per day between 8:30 AM and 9 PM and that they have 2 hours to complete each survey. For example, if a survey is sent at 8:30 AM, caregivers will have until 10:30 AM to complete the survey. After 2 hours, the links expire, and caregivers will no longer be able to access the survey. The schedule of assessments differs over the course of the 7 days and thus appears random to participants. The exact schedule is presented in Table 3. The timing protocol uses 4 signal-contingent prompts, including 1 end-of-day EMA message, which occurs at 9 PM every night. The EMA surveys are delivered in nonoverlapping time windows and refer to the previous 2 hours. This timing was selected based on previous research indicating that ≤5 prompts per day are acceptable to families [48].

In line with the recent guidelines for EMA use [49], caregivers are provided training on completing the EMAs and given a practice opportunity before the start of the assessment period. Training comprises 1 practice EMA and 1 practice end-of-day survey (with additional questions at the day level), and participants have the opportunity to seek clarity on the survey items and procedures before the assessment period.

First, caregivers report how much time they have spent with their child over the previous 2 hours. Caregivers who indicate that they have not been with their child are not presented with survey items regarding their child’s behaviors. EMA prompts regarding the frequency and intensity of child behavioral problems are based on EMA items previously tested among parents of young children [41]. Items regarding specific disruptive behavior are adapted from the Multidimensional Assessment of Preschool Disruptive Behavior [42]. Exposure to stressors is assessed using items adapted from the Daily Hassles Scale [43,50]. Degree of stress is assessed from a single item, “How stressed are you feeling right now?” Responses range from “not at all” to “extremely” [44]. Time Spent with Child and Limit Setting items are informed by existing protocols using EMA among parents [43]. Caregiver report of the overall screen time is assessed using questions adapted from the National Health and Nutrition Examination Survey screen time survey [51]. Daily assessment of screen time is presumed to improve recall bias compared with 30-day recall measures.

Several methodological considerations were made when designing the EMA protocol to balance the benefits of data richness with the drawbacks of potential participant burden and demand characteristics. Reduced-item EMA subscales are used instead of the full scales in order to limit survey fatigue. In addition, EMA surveys are prompted at seemingly random times within preset intervals (ie, a hybrid signal-interval contingent sampling schedule) to prevent anticipatory effects, such as pausing or changing current behavior in anticipation of a survey prompt at a known time [52]. Despite the use of repeated measures, reactivity is generally low with EMA procedures [53]. Furthermore, the combination of EMA with accelerometer data minimizes the weakness of using either instrument independently [54].

Both caregivers and children are asked to wear an Axivity AX3 (Axivity Ltd), a waterproof triaxial accelerometer, on their nondominant wrist for 30 days to assess physical activity, sedentary behavior, and sleep. The nondominant wrist placement improves compliance (compared with waist placement) [55]. Before data collection, the accelerometers are initialized using Open Movement (OMGui, version 1.0.0.43; Newcastle University) at a sample rate of 50 Hz with a range of +8g to −8g. Participants are mailed their accelerometers with reminders about wearing the device, a sticker chart for children to track the number of days worn, and a prestamped envelope to return the accelerometers following the 30-day protocol. The participants are instructed to wear the device at all times, including sleeping periods. In the event of a lost or damaged device, a replacement device is mailed to the participant. The number of lost or damaged devices is used to determine the utility and feasibility of deploying accelerometers over longer wear periods (ie, 30 days).

Axivity data are downloaded using Open Movement (OMGui, version 1.0.0.43; Newcastle University). Raw Axivity AX3 .cwa accelerometer files are processed in R (R Foundation for Statistical Computing) using the GGIR package (version 2.6-4) [56]. During processing, GGIR autocalibrates the signal using local gravity as a reference, identifies abnormally high values, detects nonwear periods, and calculates the magnitude of the acceleration corrected for gravity. Files are excluded if the postcalibration error is >0.02g (a measure of acceleration) and if the wear time is less than 16 hours during the 24-hour period. The nonwear time is determined based on the SD and the value range of each accelerometer axes raw data during 15-minute blocks within a 60-minute window. These blocks are classified as nonwear time if the SD of the 60-minute window is <13 milli-g, and the value range of the 60-minute window is <50 milli-g for 2 of the 3 accelerometer axes [57]. The default method for nonwear time is used, as the average acceleration at similar times on days of the week is imputed for invalid data. We determine the time spent in physical activity intensity categories (eg, inactive, light, and moderate to vigorous) using the intensity thresholds (milli-g) described by Roscoe et al [58] and Hildebrand et al [59,60] for preschoolers and caregivers, respectively. Sleep outcomes are estimated using methods developed by van Hees et al [61], which identify periods of sustained inactivity when the z-angle does not change by >5° for at least 5 minutes.

Following the 30-day monitoring period, caregivers are contacted by phone to complete a semistructured interview with a member of the research team (Multimedia Appendix 1). The semistructured interview is designed to address the following constructs of the study protocol: relevance, comprehensiveness, comprehensibility, satisfaction, and frequency or rationale of nonadherence [62-65]. Feedback provided by caregivers during the semistructured interviews is used to inform protocol changes for the future iterations of the study.

In total, caregivers will receive up to US $180 in Amazon gift cards for participating in each wave of the study. Participants will be compensated with electronic gift cards that are delivered after they complete each study task. Caregivers will receive a US $15 gift card for completing the baseline survey. Caregivers who complete at least 21 EMA prompts (75%) will be considered compliant and receive a US $40 gift card. Participants who fail to complete 21 EMA prompts within the 7-day period will be given the option to extend their EMA period beyond 7 days to have additional survey opportunities. Those who complete 21 EMA prompts with additional days of surveys will be compensated the full amount (US $40). Caregivers will receive a US $30 gift card for completing at least 21 days (70%) of screen time monitoring on their smartphones. Caregivers will receive a US $50 gift card upon the return of the Axivity accelerometers; caregiver-child dyads who both wear the Axivity accelerometers for at least 21 valid days (70%) will receive an additional US $25 gift card. Finally, caregivers will be compensated US $20 for completing the semistructured phone interview.

A total of 50% (50/100) of the families who complete the study protocol will be invited to participate in a second wave of data collection 3 to 12 months after their initial enrollment. The 9-month discrepancy in the timing of the second wave is due to rolling recruitment. The purpose of the second wave of data collection is to evaluate the feasibility of retaining the sample over time and to assess the longer-term acceptability of the protocol. The protocol will be deemed feasible and acceptable if 80% of those invited agree to re-enroll in wave 2 of the study. Furthermore, the wave 2 protocol will be adapted based on wave 1 participant feedback.

Data will be analyzed using IBM SPSS Statistics for Windows version 27. The descriptive statistics of the feasibility outcomes will be presented. As feasibility is the primary outcome of interest, we will use standard effect size estimates (ie, Cohen d and r) and minimal acceptable feasibility metrics in favor of significance testing [66]. Two-tailed independent sample t tests will be used to examine the differences between those who withdraw from the study and those who complete the protocol.

A sample size of 100 is adequate to evaluate the feasibility of the multibehavior protocol to measure digital media use and screen time context among preschoolers and their families, allowing for generalization to other families with children of similar age and demographics. Given that this is a pilot study, no power analysis is required [67].

The study protocol was approved by the institutional review board of the University of South Carolina (Approval Number: Pro00092634).

Excessive screen time has been linked to poor physical health outcomes (ie, obesity) [68] and mental health outcomes (ie, attention-deficit/hyperactivity disorder and externalizing behaviors) [69,70]. However, studies evaluating the relationship between screen time and health behaviors have significant limitations. The existing studies on screen time have focused on television viewing, largely neglected digital media use [71], relied on retrospective parental reports of average screen time [12], and not yet examined individual variability or temporality [72] of screen time within a day. This study protocol is designed to address these weaknesses by (1) leveraging data already collected by smartphones and tablets (ie, app use) to objectively measure smartphone screen time and (2) integrating intensive longitudinal data collected from multiple sources (ie, passive mobile screen time sensing, EMA, and accelerometry).

This protocol attempts to overcome the methodological limitations of previous studies by incorporating objective measures of mobile screen time and health behaviors (physical activity, sedentary time, and sleep) in conjunction with EMA. The intention of this protocol is to measure screen time behaviors among caregivers and their preschool-age children more accurately. The results will be used to evaluate the feasibility and utility of a comprehensive multibehavior protocol to measure digital media use and screen time context. If this protocol is deemed feasible and acceptable, it can be used in future observational and intervention studies to better measure screen time and its context. In addition, this protocol makes it possible to examine more complex predictors of child health behaviors, such as the within-dyad daily dynamics and between-dyad associations between parenting practices and child behavior problems.

However, our methodology is not without limitations. Objectively monitoring mobile devices (eg, smartphones and tablets) can only provide objective information on digital media use, which is only one of the sources of screen time. It is likely that children and their caregivers also use other types of screen media (ie, television viewing, computers, and shared tablets). This additional screen use is assessed using a combination of EMA measures and traditional questionnaire data. Although the use of parent-report measures of children’s additional screen time is less than ideal, the EMA method of intermittent assessment could reduce the recall bias present in many existing measures of typical screen use. Using this EMA method of screen assessment can provide temporal context to examine the within-day variability of screen time patterns. Future studies could leverage the emerging advances in facial recognition and machine-learning technology to passively and objectively measure additional forms of screen use among families.

The current methods of smartphone-based digital media–use assessment are also limited by the inability to determine who is using the smartphone. We attempt to disentangle smartphone use by determining whether the apps used are designed for children or adults. During qualitative interviews, caregivers indicate whether frequently used apps are used by themselves or by their children and provide information about who has access to the smartphone during peak time windows.

For the purpose of this feasibility study, all device use is considered screen time. However, the protocol for monitoring screen time on Android (but not iOS) devices provides information on how the device is being used (eg, app use) at specific times throughout the measurement period. Future studies can use this protocol to evaluate the differential effects of different types of screen time. For example, do children exhibit more behavioral problems following playing games on their tablet compared to video chatting? These studies are needed to advance science and inform guidelines for children’s screen time.

To avoid overlapping time points and minimize missing data due to noncompliance [43], participants are sent 4 EMAs per day; these EMAs do not assess every hour of the day and, therefore, likely miss important screen time context and resultant child behaviors. Nevertheless, the methodology used by the Tots and Tech study greatly improves upon common methods in child screen time literature.

This study is designed to evaluate the feasibility and utility of a comprehensive multibehavior protocol to measure digital media use and screen time context among a racially and economically diverse sample of preschoolers and their families. The findings will inform protocol adjustments in preparation for a future well-powered study. Ultimately, the Tots and Tech study aims to reveal the process-oriented science that underlies the association between screen time and physical (ie, sleep and activity) and mental health (ie, behavioral problems). Preliminary data from this study will inform model convergence statistics and provide informative priors for the intraclass correlation coefficient and path estimates. These are necessary to conduct a power analysis to inform a future well-powered observational cohort study to examine these connections in further detail over time. The study protocol uses EMAs and objectively measured methods that can reveal the temporal mechanisms of health behavior. Understanding individual-level behavioral patterns has the potential to advance the science of personalized intervention approaches and inform health behavior theories to improve the health and well-being of children and their families.