Problematic Child Media Use During the COVID-19 Pandemic

Participants were a nationally representative sample of 1000 parents living in the United States with at least 1 child between the ages of 6 and 17. Recruitment was facilitated by YouGov (a market research company). Detail on YouGov’s procedure to generate nationally representative samples using weighted propensity scores and sample matching has been described elsewhere.²  In brief, 547 parents with a reference child aged 6 to 10, and 535 parents with a reference child aged 11 to 17 completed an online survey between October 22, 2020 and November 2, 2020. Based on the 2017 American Community Survey, propensity scores were used to construct a sample of 500 in each reference child age category that was nationally representative in terms of parent age, race, and education, resulting in a total sample size of 1000 parents. The Seattle Children’s Hospital Institutional Review Board approved all research procedures.

Parents with multiple children between ages 6 and 18 were asked to respond to child-specific questions thinking about the child whose birthday is coming up next. We refer to this as the “reference child.”

Using questions from the Neighborhood Quality of Life Study,²³  which have been adapted to reflect recommendations in the American Academy of Pediatrics Media Use Plan, parents were asked to think about their family’s rules related to media use in 2 periods: before COVID-19 and now. Separately for each period, they indicated whether seven statements were true for their family: (1) mobile devices and other screens or TVs were not allowed in the child’s bedroom; (2) mobile devices were charged outside the child’s bedroom; (3) family members did not use screens or TVs during meal time; (4) child turned off screens at least 1 hour before bedtime; (5) restricted what shows and apps child has access to; (6) rules were enforced about how much screen time child was allowed on weekdays; and (7) rules were enforced about how much screen time child was allowed on weekends. Response options were yes (1) and no (0), and they were summed to create 2 separate 7-item indices of media rule implementation.

Reference child problematic media use was assessed using the 9-item Problematic Media Use Measure- short form,¹⁴  which measures elements of additive media use among children, including preoccupation, withdrawal, and unsuccessful attempts to control use by parents. Responses were on a 5-point Likert scale, with a possible range of 9 to 40, with higher scores indicating more problematic media use. This scale has evidence of reliability and validity when completed by mothers of school-aged children; in our data, the Cronbach α for the “problematic media use” scale is 0.927 with 95% CI (0.919 to 0.935). Although there are no established clinical cut-offs, others have used media use that “most of the time” or “always” interferes with daily functioning (item-level mean of ≥3 out of 5, or a total scale score of ≥27 out of 40) as a threshold for problematic use.²⁴ 

Parents completed the Patient Health Questionnaire-4 (PHQ-4),²⁵  a commonly used 4-item scale that assesses symptoms of depression or anxiety over the past 2 weeks. Responses are summed to create a scale with a possible range of 0 to 12, with higher scores indicating more frequent symptoms. The Cronbach α for the PHQ-4 in our data are 0.895 with 95% CI (0.878 to 0.910).

Parents completed the COVID-19 Exposure and Family Impact Survey²⁶  and indicated how COVID-19 impacted their own sleep, exercise, and eating. Response options for each item were: made it a lot better (1); made it a little better (2); made it a little worse (3); and made it a lot worse (4). Responses were summed to create an index with possible range of 3 to 12, where higher scores indicate a more negative impact of COVID-19 on self-care behaviors. The Cronbach α for this scale in our data are 0.829 with 95% CI (0.805 to 0.852).

Parents indicated whether their reference child is currently attending school in person, remotely, or in a hybrid modality.

Parents indicated the employment status of the reference child’s parent(s) or guardians(s), with options of full time, part time, and not employed. To reflect the family’s employment status on aggregate and to be inclusive of 1- and 2-parent families, we combined responses into 3 categories: no parents employed, 1 or more parents employed part-time, and all parents employed full time. Parents also reported whether they worked from home (WFH) (none of the time, some of the time, or all of the time). For analytic purposes, these 2 variables (employment and work from home status) were combined to generate a variable reflecting parent home presence. The parent home presence is considered “high” if (1) both parents were unemployed or (2) WFH status is “yes, all of the time;” parent home presence is “moderate” if WFH status is “yes, some of the time” or (WFH is missing but 1 or both parents were working part-time; or home presence is “low” if WFH status is “no” regardless both parents were working full-time or “part-time.”

Parents indicated their age, gender, race, and ethnicity using United States Census categories. Parents indicated the highest level of formal education obtained and their insurance status (as a proxy for family income²⁷ ), with options of public or private insurance. Parents indicated the reference child’s gender and age, which we subsequently grouped into 2 categories reflecting elementary (ages 6 to 10), secondary school (ages 11 to 17).

Descriptive statistics (means and frequency tables) were reported with sampling weights to be nationally representative in terms of parent demographic characteristics. Implementation of media rules was examined separately for children in the 6 to 10 and 11 to 17 year age ranges. McNemar’s χ² tests with continuity corrections were used to compare implementation of each of the 7 media-related rules, and a paired t test was used to compare the total number of rules, before COVID-19 and during the present period.

Poisson regression was used to examine the association between structural and parent- level variables and media rules implemented during the pandemic, controlling for prepandemic rules and reporting rate ratios (RR). Linear regression was used to assess the association between level of family- and structural conditions as well screen rule implementation and problematic media use. In this model both absolute number of rules and parent-reported change in rules (calculated as prepandemic rule score minus during-pandemic rule score) were included as independent variables. Regression analyses were stratified by age group (6 to 10, and 11 to 17), given potentially different school or peer contexts in primary as compared with secondary school, and potentially different parenting practices in younger children as opposed to adolescents. Posthoc analyses tested interactions between parent mental health and family characteristics (education and employment); these are detailed in electronic Supplemental Tables 5–7. All P values are adjusted using the Benjamini-Yekutieli method to control the overall false discovery rate at .05.

Weighted descriptive statistics (means and frequency tables) are presented in Table 1. In 36% of families, all parents were employed full-time. The mean parent PHQ-4 score was 3.15 (standard deviation [SD] = 3.26). Twenty-two percent of reference children were attending school in-person. On average, problematic media use scores were 22.95 (SD = 8.58). In this sample, 32.6% 6 to 10 years and 38.8% of 11 to 17 years had problematic media use scores of 27 or more (the previously used threshold for problematic use²⁴ ).

Among children ages 6 to 10, there was a small decline in the total number of rules implemented during the pandemic (prepandemic mean = 2.82, SD = 1.90, during pandemic mean = 2.55, SD = 1.86), a statistically significant but small effect (Cohen’s d = 0.14, P <.001). This was driven by change in 4 of the 7 rules: mobile devices charged outside child’s bedroom (49% prepandemic, 45% during, P = .016), no screens during mealtime (46% prepandemic, 43% during, P = .005), weekday screen time limits (29% prepandemic, 21% during, P <.001), and weekend limits (21% prepandemic, 17% during, P = .017). Among children ages 11 to 17, there was a small decline in the total number of rules implemented during the pandemic (mean = 1.74, SD = 1.68) as compared with prepandemic (mean = 2.00, SD = 1.77), a statistically significant but small effect (Cohen’s d = 0.15, P <.001). This was driven by change in 5 of the 7 rules: no screens during meal time (43% prepandemic, 41% during, P = .005), no screens at least 1 hour before bedtime (prepandemic = 36%, during = 28%, P = .012), restrictions on what shows and apps the child has access to (43% prepandemic, 41% during, P = .037), weekday screen limits (17% prepandemic, 11% during, P <.001), and weekend screen limits (10% prepandemic, 7% during, P = .002). All comparisons are presented in Table 2.

Poisson regression was used to examine the association between structural and parent variables and the number of media rules implemented during the pandemic (Table 3). In both age groups, the strongest association with during-pandemic rules was prepandemic rules. Among 11 to 17-year-olds, greater pandemic impact on parent self-care (RR = 0.97, 95% CI = 0.94 to 1.00) was associated with implementing fewer rules.

Linear regression was used to assess the association between structural and family factors and problematic media use (Table 4). The absolute number of rules a family implemented was not significantly associated with problematic use in either age group. Among younger children only, a greater decline in media rules was significantly associated with more problematic media use (0.99, 95% CI = 0.33 to 1.66, P = .004). For each 1-point increase in parent PHQ-4 score, problematic media use scores increased by 0.46 points among younger children (95% CI = 0.19 to 0.72, P = .001) and 1.27 points among older children (95% CI = 1.02 to 1.53, P <.001). There was a significant association between decline in parent self-care behaviors and higher problematic media scores in younger children only [B] = 0.71, 95% CI= 0.36 to 1.06). Child gender was significantly associated with problematic media use in only the younger age group, with girls having scores 1.71 points lower than boys (95% CI = −3.30 to −0.12). Among older children only, parent part time employment (B = 3.16, 95% CI = 0.16 to 6.15), full time employment (B = 5.17, 95% CI = 2.25 to 8.09) and high parent presence in home (B = 3.02, 95% CI = 0.60 to 5.44) were associated with higher problematic media use scores. Among older children only, as compared with those whose parents had formal educational attainment of a high school diploma or less, those whose parents with a 2-year degree (B = −4.05, 95% CI = −6.23 to −1.88), 4-year degree (B = −2.93, 95% CI = −5.28 to −0.58), or postgraduate degree (B = −3.01, 95% CI = −5.82 to −0.20) had lower problematic media use scores. As compared with 11 to 17 year olds whose schooling was fully remote, those in a hybrid modality had higher problematic media use scores (B = 2.14, 95% CI = 0.35 to 3.93). Exploratory tests for interactions were not significant and are reported in detail in Supplemental Information.

In this sample of parents of school-aged children in the United States, around one-third of children were engaging in media use classified as problematic. Consistent with the premise of the Interactional Theory of Childhood Problematic Media Use,¹³  problematic use was greater in families experiencing structural stressors and worse parent wellbeing and was unrelated to the number of media-related rules they implemented. Among younger children a greater decline in rules implemented during the pandemic as compared with earlier periods was associated with more problematic use, however this estimate was small and imprecise. Although fewer media rules were implemented during the pandemic, this change was explained almost entirely by prepandemic rules, and not by the structural and family-level stressors that were associated with problematic use.

Although others have documented benefits of having media rules for limiting screen time, at least at younger age,^16–18  results of this study raise questions about the sufficiency of such an approach to media-related parenting if the goal is limiting problematic media use. The Interactional Theory of Childhood Problematic Media Use¹³  emphasizes the importance of focusing on how media functions for a family and child, rather than focusing on the absolute level of use. Across both age groups, children whose parents were experiencing more psychological distress had notably higher problematic media use scores, and among younger children there was an additional association between parents experiencing a greater decline in self-care behaviors and problematic media use. Prepandemic data finds that among young children there is an association between greater parenting stress and less parent mediation of child media use,²⁰  and more child screen exposure. Further research is needed to understand how media functions in families when parents are experiencing challenges to their own wellbeing.

Among older children only, problematic media use was higher when parents were employed, and when they were present in the home (eg, working from home). This association raises questions about how media use functioned for families in home environments with competing parental demands. It is possible that among younger children, when parents had competing demands (eg, work), someone else was providing childcare, or parents were more closely involved in monitoring child activities (eg, assisting with logging into remote schooling or supervising leisure activities). Parents of older children may have been more likely to leave them to their own devices when faced with competing demands. As remote work seems likely to continue for many families, learning more about how families manage child media use when children are in the home and parents are working can help inform approaches to supporting families use of media in a way that is functional and not problematic.

There are many reasons why families may have changed their media use practices during the pandemic, and all need to be understood as occurring in structural, social, and family contexts that reinforce and constrain use. Nonetheless, problematic media use is something that many families will be dealing with as the pandemic resolves. Pediatricians and others counseling families about media use or designing behavioral interventions related to media use must balance addressing problematic use with the reality that media use may be helping families function in a range of ways and may be influenced by structural constraints outside of direct parenting control. A pragmatic approach to addressing problematic media use may be viewing it as 1 of many potential indicators of family functioning and parent wellbeing. A family’s change in media use practices could be used as an entry point for checking in about parenting and familial stress and connecting parents to available social services and resources that support their wellbeing and their family’s functioning.

A primary limitation of this study is that is it cross-sectional, with parents recalling prepandemic media practices. Recall may have been inaccurate in unbiased ways, making measurement imprecise. It is also possible recall was biased. Because of social desirability bias, parents may have systematically reported less problematic media use and more media-related rules. It is also possible that families with children with more problematic media use at the time of the study may have been more likely than families with less problematic use to favorably recall prepandemic media use. There were no prepandemic measures of parenting stress or problematic media use, further limiting any inferences related to causality. Second, 1 parent responded about 1 reference child. However, families are complex systems, often with more than 1 parent and child, or blended families across 2 households. Such dynamics (eg, number and age of children in the family or discordant parent perspectives on media use²⁸ ) undoubtedly impact how families approach rule implementation, how media use functions, and child risk of problematic use. Qualitative research and network-based survey research is needed to explore the complexities of media rule implementation across diverse family structures. Third, we did not include a measure of screen time in these analyses given problems with reliability and validity of parent-report of child screen use. Future research using a reliable, objective, means of measuring child media use would be useful to understand how changes in implementation of media use practices relate to amount of screen use. Additional variables that were unmeasured but are potentially important in explaining problematic media use include child-level mental health, behavior or functioning, social norms related to screen use, parent stress, more nuanced measurement of family employment-related stressors (eg, parent remote work), and more detail on family socioeconomic status, including a direct measure of family income.

Family stressors, but not media rules, were associated with more problematic media use among school-aged children during the COVID-19 pandemic. Harm reduction efforts related to problematic media use should consider the inequitable stressors to which families are exposed and how these impact the ways in which media is used by families to support functioning.