Infant television watching predicts toddler television watching in a low-income population

Abstract

OBJECTIVES:

This study examines the development of active television (TV) watching behaviors across the first 2 years of life in a racially and ethnically diverse, low-income cohort and identifies caregiver and child predictors of early TV watching.

METHODS:

We used longitudinal data from infants enrolled in the active control group (N = 235; 39% Latino; 29% black; 15% white) of Greenlight, a cluster randomized multi-site trial to prevent childhood obesity. At preventive health visits from 2 months to 2 years, caregivers were asked: “How much time does [child’s first name] spend watching television each day?” Proportional odds models and linear regression analyses were used to assess associations among TV introduction age, active TV watching amount at 2 years, and sociodemographic factors.

RESULTS:

68% of children watched TV by 6 months, and 88% by 2 years. Age of TV introduction predicted amount of daily active TV watching at 2 years, with a mean time of 93 minutes if starting at 2 months; 64 minutes if starting at 4 or 6 months; and 42 minutes if starting after 6 months. Factors predicting earlier introduction included lower income, fewer children in household, care away from home, male sex, and non-Latino ethnicity of child.

CONCLUSIONS:

Many caregivers report that their infants actively watch TV in the first 6 months of life. Earlier TV watching is related to sociodemographic factors yet predicts more daily TV watching at 2 years even controlling those factors. Interventions to limit early TV watching should be initiated in infancy.

INTRODUCTION

The 2016 policy statement on screen media by the American Academy of Pediatrics discourages television (TV) watching and media use for infants before the age of 18–24 months.¹ Nevertheless, introduction of TV watching before 18 months old is common internationally.^2,3 Over the last few decades, screen media use has become more common at earlier ages—even active TV watching now often begins as early as 2 months old.^4–6

High levels of early TV are concerning, because it has been associated with obesity,⁶ sleep problems,⁷ developmental problems,⁸ and behavioral problems.⁹ Additionally, watching patterns established in early life appear to continue through childhood and adolescence.¹⁰ However, less is known about whether active TV watching as early as the first year of life predicts watching practices later in life, and very few studies have examined trends in the first two years of life.¹¹ Finally, given reduced outdoor play and social distancing, higher levels of screen media are likely to continue through the COVID-19 pandemic and subsequent changes in social interaction and schooling. ^12,13

We sought to examine the development of active TV watching behaviors across the first two years of life in a group of racially and ethnically diverse children, to improve the understanding of trends in TV watching in infancy and early childhood. Specifically, we aimed to (1) identify child and caregiver factors associated with earlier introduction of TV watching, and (2) identify factors associated with more TV watching at two years of age. We hypothesized that earlier active TV introduction would lead to more TV watching at two years of age.

METHODS

Sites and Sample

We analyzed longitudinal data from caregiver-infant dyads enrolled in the active control group of the Greenlight Intervention Study. The Greenlight Study is a cluster randomized multi-site trial of an obesity prevention intervention delivered to children seen in pediatric resident continuity clinics during their first two years of life.¹⁴ Four university-affiliated pediatric clinics (University of North Carolina at Chapel Hill, New York University/Bellevue Hospital Center, Vanderbilt University, and University of Miami/Jackson Memorial Medical Center) participated in the study, with two sites randomized to using a literacy- and numeracy-sensitive approach to obesity prevention and two sites randomized to address injury prevention as an active control. We enrolled caregivers of healthy infants at their infants’ 2-month-old well visits, and followed them at each subsequent well visit up to the two year well visit.

Caregiver-infant dyads were enrolled if they met the following eligibility criteria: caregiver fluency in English or Spanish; no plans to leave the clinic within the following 2 years; child presenting for a 2-month well visit with a resident; and child between 6 weeks and 16 weeks of age. Dyads were excluded if they met any of the following criteria: caregiver < 18 years old; caregiver had serious mental or neurologic illness likely to impair ability to consent or participate; poor visual acuity; infant born before 34 weeks gestational age or birthweight < 1500 g; weight < third percentile at enrollment; or infant had any chronic medical problem impacting weight gain patterns. All caregivers enrolled provided consent; this study was reviewed and approved by the Institutional Review Board at each site. At each well visit from 2 months to 2 years, the resident provider delivered the intervention (obesity prevention or active control, which was The Injury Prevention Program, depending on site). At each time point, caregivers completed questionnaires with trained research assistants fluent in English and Spanish. Questionnaires included information on sociodemographic characteristics and information about behaviors including TV watching.

Measures

TV Watching

At each well visit, caregivers reported the answers to two questions related to TV time. First they were asked, “How much is the television on each day with [child’s first name] is in the room, even if [child’s first name] is “not watching?” The second question, which we defined as “active TV watching” asked “How much time does [child’s first name] spend watching television each day?” to query active TV watching on an average day. Our study concerns the answer to the second question because we believe that active TV watching may be more directly connected with obesogenic behaviors.

Statistical Methods

We chose to analyze dyads only enrolled into the active control arm, as the Greenlight study intervention group received materials targeting behavior change in diet and physical activity domains, including limiting screen media use. All participants in this study were observed at scheduled 2, 4, and 6 month well visits, so we defined age at TV introduction using a three-level, ordinal variable with values: 2 months, 4 or 6 months, and after 6 months; 4 and 6 months were combined because few participants (N = 21) reported TV introduction at 6 months. We limited analyses of TV introduction age to the first six months because few studies have examined how TV behaviors very early in life affect TV watching later in life.

We examined the association between age at TV introduction and a pre-specified set of risk factors using a three-level proportional odds model with the response levels (i.e., 2, 4 or 6, and >6 months) ordered so that odds ratios greater than (less than) one correspond to a later (earlier) TV introduction. Covariates in the analysis, chosen based on literature review of factors associated with screen time behaviors, included the caregiver’s education level (<high school/GED, high school or GED, at least some college), annual household family income (<10K, 10–39K, and 40K+), language (Spanish and English), adults in household (one or greater than one), child’s sex, child’s race/ethnicity (Latino, non-Latino white, non-Latino black, and non-Latino other race), number of children in household (continuous), child care outside of the home (yes or no), and site (UNC and Miami). To examine the relationship between TV time at two years and the age at introduction and other risk factors, we used linear regression analyses. We regressed minutes of TV time at 2 years on the three-level age at introduction (i.e., 2, 4 or 6, and >6 months) and adjusted for the covariates described above, and one additional covariate: TV in the room where the child sleeps (yes or no). Because the distribution of TV time at 2 years was non-standard (skewed and with digit preferences), we used Huber-White robust standard error estimates of uncertainty.^15,16 TV time was missing in approximately 30% of our sample at 2 years; however, the Greenlight Study collected TV time and other data at the 2, 4, 6, 9, 12, 15, 18, and 24 month well visits, which allowed us to multiply impute missing TV times at 2 years. We used predictive mean matching to construct 25 imputation datasets, and we combined the results using the standard Rubin’s Rules.^15,17 All analyses were completed using statistical software: R version 3.6.2 with the rms() package and hypothesis tests were conducted using two-sided, 0.05 significance levels.^18,19

RESULTS

Of the 865 parent-child dyads enrolled in the study, 404 were enrolled in the active control. To ensure that active TV watching introduction age was known for all children in this subset, we excluded an additional 169 dyads with missing data at any well visits at 2, 4, and 6 months, leaving a sample of 235 parent-child dyads for analysis. In this sample, the proportion of children not actively watching any amount of TV decreased from 53.6% at 2 months to 12.0% at 2 years (Fig 1). The proportion of children actively watching >60 minutes of TV increased from 9.4% at 2 months to 30.2% at 2 years (Fig 1).

FIGURE 1.

Trends in child television (TV) watching from 2 months to 24 months of age. The percentage of children in each of the TV watching groups at each well visit is presented.

Potential Risk Factors and Age of TV Introduction

Table 1 shows characteristics of children and caregivers gathered at the initial study visit (2 months), separated based on age of TV introduction. Table 2 shows the associations between child and caregiver characteristics and age of TV introduction, adjusted for caregiver’s education level, family income, language, number of adults in household, child’s sex, child’s race/ethnicity, number of children in household, child care outside of the home, and site. After adjustment, age of TV introduction was significantly associated with income level, child’s sex, Latino ethnicity, number of children in the household, and child care outside of the home. There was a stepwise relationship between income and age of TV introduction; as income increased, active TV watching was more likely to be introduced at later ages. Compared with a family income level of <$10,000 per year, an income of $10,000–39,999 was associated with later age of TV introduction (adjusted odds ratio [aOR] = 2.41; 95% CI: 1.16–4.99), and an income level of $40,000 or more was associated with even higher odds of later introduction (aOR = 6.96; 95% CI: 2.65–18.29). Female sex was also associated with later age of introduction compared to male sex (aOR = 1.97; 95% CI: 1.14–3.42). More children in the household increased the odds that the participating infant had active TV introduced later; for each additional child in the household, there was a 29% higher odds of later introduction (aOR = 1.29; 95% CI: 1.02–1.63). Non-Latino infants tended to be introduced to TV earlier than Latino children. For example, the aOR for later TV introduction was 0.32 (95% CI: 0.11–0.92) for white, non-Latino children compared to Latino children, and it was 0.23 (95% CI: 0.09–0.59) for black, non-Latino children versus Latino children. Child care outside of the home was also associated with earlier introduction (aOR = 0.19; 95% CI: 0.05–0.69).

Table 1:

Baseline characteristics of children and caregivers by age of TV introduction (N = 235)

Covariates	Age 2 months	Age 4 or 6 months	After age 6 months
Caregiver factors
Education
Less than high school	27 (25%)	17 (19%)	9 (24%)
GED/high school	67 (61%)	48 (55%)	12 (32%)
Some college	15 (14%)	23 (26%)	16 (43%)
Income
Less than $10,000	39 (36%)	14 (16%)	4 (11%)
$10,000–39,999	49 (45%)	47 (54%)	13 (35%)
$40,000 or more	14 (13%)	18 (21%)	18 (49%)
Caregiver did not know	6 (6%)	8 (9%)	2 (5%)
Language
English	82 (75%)	55 (62%)	24 (65%)
Spanish	27 (25%)	34 (38%)	13 (35%)
Number of adults in household
1	12 (11%)	11 (12%)	3 (8%)
2	66 (61%)	62 (70%)	29 (78%)
3 or more	31 (28%)	16 (18%)	5 (14%)
Child Factors
Child’s sex
Male	60 (55%)	50 (56%)	13 (35%)
Female	49 (45%)	39 (44%)	24 (65%)
Race/Ethnicity
Latino	30 (31%)	42 (52%)	20 (59%)
White Non-Latino	16 (16%)	12 (15%)	8 (24%)
Black Non-Latino	42 (43%)	21 (26%)	4 (12%)
Other Non-Latino	10 (10%)	6 (7%)	2 (6%)
Environmental factors
# of children in household
1	53 (49%)	26 (29%)	15 (41%)
2	30 (28%)	30 (34%)	12 (32%)
3 or more	26 (24%)	33 (37%)	10 (27%)
Away from home care at 2 months	14 (13%)	4 (4%)	0 (0%)
Away from home care at 2 years	19 (26%)	19 (32%)	7 (24%)
TV in a room where the child sleeps at 2 years	34 (46%)	28 (46%)	8 (27%)

Note: Categorical variables were presented as count with percentages.

Table 2:

Associations of risk factors and later age of TV introduction

Covariates	Adjusted odds ratio with 95% CI	P Value
Caregiver factors
Education (vs < High school)		0.11
GED/High school	1.38 (0.65 to 2.92)
Some college	2.79 (1.05 to 7.41)
Income (vs < $10,000)		0.003
$10,000–39,999	2.41 (1.16 to 4.99)
$40,000 or more	6.96 (2.65 to 18.29)
Caregiver did not know	2.42 (0.76 to 7.68)
Language (vs English)		0.85
Spanish	0.88 (0.35 to 2.24)
Number of adults in household (vs More than one adult)		0.06
Single adult	2.22 (0.93 to 5.34)
Child factors
Child’s sex (vs Male)		0.02
Female	1.97 (1.14 to 3.42)
Race/Ethnicity (vs Latino)		0.02
White Non-Latino	0.32 (0.11 to 0.92)
Black Non-Latino	0.23 (0.09 to 0.59)
Other Non-Latino	0.30 (0.09 to 1.05)
Environmental factors
# of children in household		0.04
For each additional child	1.29 (1.02 to 1.63)
Away from home care (vs No)		0.02
Yes	0.19 (0.05 to 0.69)

Note: Proportional odds model was used to examine associations between risk factors and TV introduction with adjustment for caregiver’s education level, income, language, number of adults in household, child’s sex, child’s race/ethnicity, number of children in household, and away from home care. Greater odds ratio indicates higher odds of later TV introduction.

Associations of Potential Risk Factors and TV Watching Time at 2 Years Old

In unadjusted analyses, earlier age of active TV introduction predicted greater amounts of active TV watching at age 2 years in a dose-dependent manner, with a mean time of 93 minutes for children starting at 2 months; 64 minutes for children starting at 4 or 6 months; and 42 minutes for children starting after 6 months (P < .05). Despite adjusting for covariates affecting the relationship (including caregiver’s education level, family income, language, number of adults in household, child’s sex, child’s race/ethnicity, number of children in household, child care outside of the home, TV in the room where the child sleeps, and site), there remained a significant association between active TV introduction time and the amount of time spent watching TV at 2 years (P < .05). Compared to children starting at 2 months, children starting after 6 months watched an average of 36 minutes less TV per day (95% CI: −64 to −8; see Table 3), and children starting at 4 or 6 months watched an average of 11 minutes less TV per day, though this second difference was not statistically significant (95% CI: −36 to 15; see Table 3).

Table 3:

Associations of risk factors and TV watching time (in minutes) at 2 years of age

Covariates	Adjusted mean difference (minutes) with 95% CI
Caregiver factors
Education (vs < High school)		0.53
GED/High school	6 (−21 to 33)
Some college	−11 (−50 to 27)
Income (vs < $10,000)		0.80
$10,000–39,999	12 (−21 to 45)
$40,000 or more	21 (−28 to 70)
Caregiver did not know	13 (−30 to 55)
Language (vs English)		0.95
Spanish	0 (−42 to 41)
Number of adults in household (vs More than one adult)		0.47
Single adult	−1 (−40 to 39)
Child factors
Child’s sex (vs Male)		0.03
Female	27 (3 to 52)
Race/Ethnicity (vs Latino)		0.09
White Non-Latino	36 (−15 to 86)
Black Non-Latino	55 (11 to 98)
Other Non-Latino	26 (−27 to 79)
Environmental factors
# of children in household		0.0002
For each additional child	−17 (−26 to −8)
Away from home care (vs No)		0.01
Yes	−43 (−75 to −11)
TV in room where child sleeps (vs No)		0.02
Yes	33 (6 to 59)
Age of TV introduction (vs 2 months)		0.04
4 or 6 months	−11 (−36 to 15)
After 6 months	−36 (−64 to −8)

Note: Linear regression was used to examine associations between risk factors and TV watching time at 2 years with adjustment for caregiver’s education level, income, language, number of adults in household, child’s sex, child’s race/ethnicity, number of children in household, away from home care, TV in room where child sleeps, and age of TV introduction. Mean number of minutes relative to the comparator is reported; for example, for “Education”, those children with caregivers with a GED/High school education watched on average 6 more minutes (with 95% CI of −21 to 33 minutes) than children with caregivers with < High school education.

Table 3 shows the adjusted associations between caregiver, child, and environmental factors and TV watching time at 2 years. After adjustment, TV watching time at 2 years was significantly associated with child’s sex, Latino ethnicity (among Black children), number of children in the household, child care outside of the home, having a TV in the room where the child sleeps, and age of TV introduction. Compared to male children, on average, female children actively watched 27 minutes more TV per day at 2 years (95% CI: 3 to 52). Non-Latino Black children watched an average of 55 minutes more than Latino Black children (95% CI: 11 to 98); this association was not significant when comparing non-Latino White children to Latino White children (95% CI: −15 to 86). For each additional child in the household, the child of interest watched, on average, 17 minutes less (95% CI: −26 to −8) of TV per day. Children with child care outside of the home watched 43 minutes less than children without child care outside of the home (95% CI: −75 to −11), whereas children with a TV in the room where they sleep watched 33 minutes more than children without a TV in that room (95% CI: 6 to 59).

DISCUSSION

At the time of enrollment for this study, the AAP recommended that “Pediatricians should urge parents to avoid television viewing for children under the age of 2 years.”²⁰ In this multisite sample of racially and ethnically diverse, low-income parents followed from their child’s 2-month-old well visit to 2-year-old well visit, we found that a high proportion of caregivers (46.4%) reported that their children actively watched TV daily as early as two months of age. By 2 years of age, 88% of children in our cohort were actively watching some amount of TV, with 30% watching >60 minutes daily. Furthermore, there was a dose-dependent relationship of early TV introduction to the amount of later TV watching, with earlier introduction predicting a greater amount of TV watching at 2 years. Our findings are similar to other studies showing that children are exposed to high amounts of screen time early in life.^2,3 Our study adds to this literature by showing that introduction of active TV watching in infancy predicts more TV watching at 2 years, which could disrupt healthy development during preschool years.²¹ Our findings reinforce the AAP’s current Bright Futures guidelines, which recommend that providers discuss avoidance of TV and other digital media at the first postnatal visit. However, our data suggests that message should be strong and clear, since there is a breadth of literature identifying problems associated with early and excess screen time, ranging from obesity,⁶ to sleep problems,⁷ to developmental and behavioral problems.^8,9

Independent of sociodemographic background, infants in our sample who watched TV earlier in life became toddlers who watched more TV than their same-age peers. However, we identified in adjusted analyses certain factors that stood out as more predictive of earlier TV introduction, and/or more TV time at 2 years old. Some of these factors have been previously connected to screen time behaviors, and lend themselves to clear recommendations; for example, having a TV in a child’s bedroom has been linked to greater amounts of screen time, along with health outcomes such as obesity and emotional/behavioral problems.²² These findings reinforce the recommendation to maintain a TV-free environment where the child sleeps, and turning off the TV as much as possible when the child is present in a room with a TV. Compared to children with siblings, only children are more likely to have less physical activity, and our findings add that only children may also have more TV time.²³ Physicians should be aware of this phenomenon and help parents to identify strategies to keep only children physically active and off screens, despite not having siblings to play with. Parents with lower incomes may be less able to implement these strategies to limit their children’s screen time, perhaps due to greater life pressures including household stress, busy schedules, and having one or fewer rooms.²⁴ Unfortunately, for these families, it is more difficult to give universal recommendations for reducing TV watching; instead, physicians should recognize these children as high-risk and work with families on an individual basis to identify low-cost ways to increase physical activity and limit TV’s negative effects.

Our analysis yielded some novel associations. First, regarding race and ethnicity, while prior studies have documented that Latino children are less likely to meet screen time recommendations relative to non-Latino White children,²⁵ our study revealed that non-Latino Black and White children are more likely to start watching TV earlier than Latino children, and Latino children watched less TV at 2 years than non-Latino Black children. Second, number of adults (presumably caregivers) in the household did not have a significant impact on either of our TV watching outcomes, despite the fact that parental limit-setting is known to influence the screen time habits of older children, and despite the assumption that having two or more caregivers in the household would make it easier to enforce limits and avoid the “TV as babysitter” phenomenon.²⁶ Nevertheless, previous research has not shown a significant role of family structure on TV time in younger children, and, given the wide confidence interval of our own results, further research should continue to explore this counter-intuitive finding.²⁷ Third, although males were more likely to start watching TV earlier, females were on average watching more TV than males at 2 years old, contradicting the overall sample’s association between earlier introduction and greater TV time at 2 years. Potential factors contributing to the finding that males watch less TV at 2 years old include parental restriction of TV watching for boys more so than girls;²⁸ and male children getting involved in other digital media by 2 years old—such as interactive tablets or mobile phones—that parents do not consider “TV watching”.² However, a systematic review reported no difference in screen media use between young female and male children.²⁹ It may be that, at this early stage in development (i.e., less than two years old), sex does not play as important of a role in how parents organize their child’s TV watching activities, which may partially account for our mixed results.³⁰ In conclusion, the nature of the association between sex and screen media usage appears to be complex, and deserves further exploration. Fourth, although children participating in child care outside of the home were more likely to watch TV earlier in life, they watched less TV at 2 years old. The reasons for this are unclear; one observational study demonstrated that child care centers are generally compliant with recommendations to limit infant screen time, so one would expect that infants receiving care in these centers would receive less TV exposure in general.³¹ However, we do not know the varieties of child care outside of the home that children received, and whether screen time was allowed at that location; prior evidence demonstrates that children watch more TV at home if their school or day care allows screen time.³²

The interpretation of our results should be considered in the context of some limitations. First, our measure of infant TV watching relies on caregiver recall, subject to problems related to response bias and participant insight. However, one advantage over previous studies is that our study distinguished between exposure to background television versus exposure to television for the purpose of having a child actively watch television. Despite the fact that the survey question whose results we analyzed came after a question about TV exposure (i.e., how much time the TV is on when the child is in the room, even if the child is not watching), we cannot say for sure how caregivers interpreted the question: “On a typical day, how much time does [child’s first name] spend watching television?”¹⁰ However, our dose dependent results related to time of introduction seem valid since the question was kept constant. Still, future research could benefit from more objective measures of TV watching and exposure, rather than relying on caregiver report.³³ Second, bias related to missing data may have affected the external and internal validity of our conclusions, and we note that TV time data was missing at the 24 month timepoint in 30% of children.³⁴ Besides individual motivation factors among patients and their caregivers, attrition in pediatric behavior change programs has been consistently linked to income levels, or related factors such as parents’ inability to miss work or find transportation.³⁵ Considering studies showing that life pressures (e.g., low income) affect parents’ ability to limit their children’s screen exposure, sample attrition bias may have affected results related to screen time.²³ To address this challenge, we conducted imputation analyses that exploit the longitudinal data collection for the Greenlight Study, allowing us to use TV time at all timepoints they were available (2, 4, 6, 9, 12, 15, and 18 months) to impute missing values at 24 months. Even though we cannot eliminate the possibility of informative missingness, we believe the imputation analyses conducted herein exploit all meaningful data to arrive at our study results. Third, our study only gathered data on TV watching, while additional forms of screen media, such as smart phones and tablets, have recently proliferated. However, despite large increases in mobile device use in this century, TV is still the primary mode of screen media for young children, accounting for 86% of screen time among children 0 to 2 years old in one analysis.³⁶ Finally, the population enrolled in Greenlight is a racially and ethnically diverse and low-SES sample. While not generalizable to the nation at large, this group is at high risk for obesity and other health relationships to screen time, and, as such, their early TV watching patterns are important.³⁷ In fact, the sample’s diversity, the longitudinal assessment of TV time, and the breadth of available covariates (i.e., demographics, household size, and child care) are all significant strengths of the study as these are children who are less commonly studied.

CONCLUSION

The AAP’s 2016 statement on screen media recommends that providers “Start the conversation early. Ask parents of infants and young children about family media use, their children’s use habits, and media use locations.”¹ Our study adds to the growing body of literature identifying children being exposed to high amounts of screen time early in life and emphasizes the need for early screening and counseling regarding limiting screen habits for infants and to encourage healthy alternative to screen time (e.g. reading, singing, playing). Clinicians who provide care to children should be aware of both the prevalent very early introduction of TV and the fact that early introduction predicts later patterns. Future studies should test whether TV watching in toddlerhood predicts even later screen time behavior, and whether interventions aimed at educating caregivers on limiting screen exposure in the first two years of life (per current AAP recommendations) or even earlier and interventions that account for newer forms of screen exposure are helpful in mitigating later childhood screen time and its concomitant health effects.

WHAT’S NEW:

This study demonstrates that, regardless of individual caregiver and infant characteristics, introduction to TV in the first six months of life is associated with more TV watching at age two years, reinforcing the need for early anticipatory guidance.