Correlates of screen time in the early years (0–5 years): A systematic review

Highlights • This review included 53 studies examining 91 correlates of screen time.• Studies should focus on sociocultural, environmental, behavioral and economic factors.• Improved methodology is recommended to ensure high-quality studies.• Interventions should focus on physical environment and sociocultural factors.


Introduction
Screens (e.g., television (TV) or mobile phone), are nowadays omnipresent. This has led to substantial increases in screen time in early childhood (children aged < 5 years) (Barber, 2017;Downing et al., 2015;Vanderloo, 2014;McNeill, 2019). The World Health Organization (WHO) screen time guidelines recommend no screen time for infants and toddlers up to age 2 years (World Health Organization, 2019), and<1 h/ day for children aged 2-4 years (World Health Organization, 2019). Currently, many young children are exceeding these screen time recommendations (Barber, 2017;Downing et al., 2015;Vanderloo, 2014;McNeill, 2019). A review from 2015 estimated 2.3% to 83% of children aged < 2 years are meeting the screen time recommendation . More recent studies estimate that 25% of children aged 12 months from the UK meet the guidelines (Barber, 2017), and that Australian preschoolers spent 2.4 h/day on electronic media (McNeill, 2019).
Given the high levels of screen time in young children, there is a growing concern about the impact of screen time on their health and development. Evidence suggest that higher levels of screen time in young children are associated with higher levels of adiposity, more sleep problems and lower scores on measures of psychosocial health, cognitive development and motor development (McNeill, 2019;Li, 2020;Poitras, 2017), highlighting the need to intervene at an early age.
Interventions aimed at reducing screen time in early childhood are effective (Downing, 2018), however, heterogeneity of studies (e.g., intervention setting) and a lack of clear trends in subgroup analyses (e. g., different age groups) have resulted in a need for more research. Examining correlates of screen time is an important step towards identifying effective strategies to reduce screen time. This is especially important as sedentary behavior, including screen time, tracks from early childhood to adolescence and into adulthood (Jones, 2013;Busschaert, 2015). A previous review on correlates of sedentary behaviors in preschoolers concluded that sex and outdoor play were not associated with time spent TV viewing (Hinkley, 2010). A review specifically examining the childcare setting revealed that higher staff education was associated with less screen-viewing activities in preschoolers while home-based care (compared to center-based childcare) was associated with more screen-viewing activities (Vanderloo, 2014). A review examining correlates in children aged < 7 years found several sociodemographic (e.g., higher child age, low socioeconomic position households, non-white ethnicities), sociocultural (e.g., living in higher TV-viewing households less parental screen-viewing rules, maternal depressive symptoms, higher parental body mass index) and environmental (e.g., greater access to media sources and less perceived safety in neighborhoods) correlates of higher screen time (Hoyos Cillero and Jago, 2010). Both reviews included all types of screen time as a total outcome measure. A review examining parental influences in children aged < 6 years concluded that parental screen time was positively associated with young children's screen time (Xu et al., 2015). Finally, a review in children aged < 3 years concluded that an older age, a minority race/ethnicity, a higher body mass index (BMI), maternal distress/depression, and maternal TV viewing time were associated with higher screen time (Duch, 2013). Additionally, cognitive stimulation in the home environment was associated with lower screen time (Duch, 2013).
Previous reviews have examined specific contexts (Vanderloo, 2014;Xu et al., 2015), age groups (Vanderloo, 2014;Hinkley, 2010;Duch, 2013), mostly or only examined TV viewing as type of screen time (Hinkley, 2010;Duch, 2013) and were all conducted several years ago (<2015). Therefore, and due to the continuous development of screenbased devices (e.g., smartphones) an updated overview of the correlates of (different types of) screen time in early childhood is required. This review expands on previous work by focusing on the entire range of early childhood (0-5 years), and including a broad focus regarding types of correlates and types of screen time. The aim of this systematic review is to identify correlates of screen time in typically developing children aged 0-5 years. Variables reported as potential correlates of screen time were categorized using an ecological perspective (Stokols, 1992;Sallis et al., 2008;Swinburn et al., 1999).

Protocol and registration
A review protocol was developed based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)-statement (Moher, 2009). This systematic review was registered with the International Prospective Register of Systematic Reviews (PROSPERO: CRD42020204330).

Eligibility criteria
Studies were included if they 1) examined the association between a potential correlate and any type of screen time (total screen time as well as specific types of screen time) in typically developing, apparently healthy children aged 0-5 years (operationalized as a mean age of < 5 years at the time of screen time assessment), 2) used a longitudinal or cross-sectional design, 3) assessed screen time quantitatively (duration and/or frequency) and 4) were written in English and published after 2000, in a peer reviewed scientific journal.
Exclusion criteria were: studies 1) solely presenting data on children born preterm (<37 weeks), 2) using adherence to guidelines (yes/no) as sole assessment of screen time, 3) examining a prenatal potential correlate, except for prenatally assessed socio-demographic factors as these contribute to identifying which groups should be targeted for interventions. The review was conducted with a view to informing interventions to limit screen time.

Literature search and study selection
A comprehensive search from inception up to 26 October 2021was performed in the bibliographic databases PubMed, Embase, PsycINFO and SPORTDiscus. The following terms were used (including synonyms and closely related words) as index terms or free-text words: "Infant", "Screen time", "correlate*". The search was performed without language or publication status restriction. The full search strategies for all databases can be found in supplementary File 1.
One reviewer (SLCV) individually screened all titles and abstracts after removal of duplicates, and a second (JSG) and third reviewer (TMA) each individually and independently screened 50%. Following screening, discrepancies were discussed until consensus was reached. Full text articles were screened by one reviewer (SLCV). Ten percent of the inclusion and each of the exclusion categories were checked by a second reviewer (JSG). Discrepancies (<5%) were discussed until consensus was reached.

Data extraction
Data on the following variables were extracted using a structured form: study methodology (e.g. design, study duration and points of data collection), participants (e.g. sample size, mean age, percentage girls), screen time (e.g. outcome variable, assessment tool used), correlate (e.g. type, assessment tool used) and results. Statistical significance of p < 0.05 was accepted as significant. Data extraction was individually conducted by two researchers per article (SLCV for all articles, JSG for 75% of articles and LV for 25% of articles). Results were compared and discrepancies were discussed until consensus was reached.
Variables reported as potential correlates of screen time were categorized. in seven categories that were divided into two main categories, using an ecological perspective (Stokols, 1992;Sallis et al., 2008;Swinburn et al., 1999): 1) Individual (including subcategories Biological, Behavioral attributes and skills, and Cognitive, emotional or psychological); and 2) Environmental (including subcategories Physical, Economic, Political, and Sociocultural).

Quality assessment
Methodological quality of all included studies was determined using an adjusted version of the 'Quality Assessment Tool for Quantitative Studies' (EPHPP) (Jackson and Waters, 2005;Thomas, 2004) (see supplementary file 2). Two reviewers each independently scored all included studies (SLCV and either TMA or MJMC). The EPHPP tool contains 19 items divided over eight quality criteria: selection bias, study design, confounders, blinding, data collection methods, withdrawals and drop-outs, intervention integrity, data-analysis. Per quality criterion, a quality score was provided: good, fair or poor. Discrepancies were discussed until consensus was reached. The overall methodological quality of a study was classified as 'high' when none of the quality criteria were scored as poor. A study was classified as 'moderate' when at most one quality criterion was scored as poor. The overall methodological quality of a study was classified as 'weak' when two or more quality criteria were scored as poor. If a study was the only one examining a particular correlate, the methodological quality was not assessed as a best evidence synthesis was not possible (see Synthesis of evidence).

Synthesis of evidence
A best evidence synthesis was applied to draw conclusions on the level of evidence for the association between a potential correlate and screen time in children aged 0-5 years. This synthesis was based on the number of studies, their methodological quality and the consistency of findings (Slavin, 1995): • Strong evidence: consistent findings in multiple studies (≥2) of high methodological quality. • Moderate evidence: consistent findings in one study of high methodological quality and at least one study of weak or moderate methodological quality, or consistent findings in multiple studies (≥2) of weak or moderate methodological quality. • No evidence: consistent findings for no significant association in multiple studies (≥2) of moderate or high methodological quality. • Inconsistent evidence: inconsistent findings in multiple studies (≥2).
• Insufficient evidence: only one study available.
Findings were considered consistent when ≥ 75% of studies demonstrated findings in the same direction, based on significance of p < 0.05 in the fully adjusted model. If two or more studies of high methodological quality were available, results of studies with weak methodological quality were ignored in determining level of evidence.
As there was a large variation in how correlates were defined and measured, we combined studies examining a similar construct. Additionally, if studies examined multiple associations reflecting associations with multiple measures of one potential correlate (e.g. house features: presence of a yard, fences or covered outdoor area), they were considered to add evidence when consistently demonstrating a significant association in > 50% of examined associations.
For type of screen use, we aimed to examine the correlates per type of screen time. However, given the large number of correlates for which we had to conclude insufficient or inconsistent evidence based on two studies, we decided to combine all types of screen time to examine correlates. If studies examined multiple associations for different types of screen time (e.g. analyzing the association with multiple assessments of screen time such as TV viewing, playing video games and computer use), they were considered to add evidence when consistently demonstrating a significant association in > 50% of examined associations. Besides total screen time, we also examined TV time (see summary tables in supplementary file 3).

Overview of studies
The search identified 6,618 hits after removal of duplicates. In total, 53 studies were included in this review. Fig. 1 presents the flow diagram. Table 1 presents the characteristics of included studies and the results on associations with screen time. Supplementary file 4 presents the methodological quality of included studies.

Biological correlates
Twenty-four studies examined biological correlates, of which sex was most frequently examined (n = 15 studies), followed by age and race/ethnicity (both n = 13 studies). The majority of studies found no significant association between sex and screen time (n = 12 studies) (Bernard, 2017;Abbott, 2016;Barr, 2010;Berglind and Tynelius, 2017;Bleakley et al., 2013;Carson and Kuzik, 2017;Carson et al., 2014;Morowatisharifabad et al., 2015;Nikken and Schols, 2015;. Results were inconsistent for age and race/ethnicity. Eight out of 13 studies examining age found that a higher age was associated with more screen time, while five studies found no significant association (Bernard, 2017;Carson and Janssen, 2012;Carson and Kuzik, 2017;Carson et al., 2014;Morowatisharifabad et al., 2015;. Eight studies examining race/ethnicity of either the child or parent found a significant association with screen time (Barber, 2017;Bernard, 2017;Hish, 2021;Xu, 2016;Carson and Kuzik, 2017; and five studies found no significant association Bleakley et al., 2013;Flores et al., 2005;Thompson and Christakis, 2007). Seven out of eight studies found no significant association between having siblings/the number of children at home and screen time (Barr, 2010;Bleakley et al., 2013;Carson and Kuzik, 2017;Nikken and Schols, 2015;Thompson and Christakis, 2007; and three out of four studies found no significant associations between being a first-born child and screen time (Bernard, 2017;. One study found that a lower BMI was associated with less screen time  whereas three studies found no significant association Morowatisharifabad et al., 2015).
In summary, we found no evidence for an association of child sex, BMI, the number of siblings/children at home and being a first-born child with screen time. We found inconsistent evidence for an Table 2 Summary of results regarding correlates of screen time in young children (<4 years).

Correlate
Included studies a,b Summary coding (n/N, (%)) a,b

Biological correlates
Sex + + 0 0 0 0 0 0 0 0 0 0 0 0 -0 (12/15, 80%) Based on consistent findings among studies with low-to-high methodological quality, there is no significant evidence for an association between sex and screen time. Age þ + + + + + + + 0 0 0 0 0 ? Based on inconsistent findings among studies with low-to-high methodological quality, there is insufficient evidence for an association between age and screen time. Ethnicity/race child or parent/ country of birth parent + + + + + 0 0 0 0 0 ---? Based on inconsistent findings among studies with low-to-high methodological quality, there is insufficient evidence for an association between ethnicity/race of the child or parent and screen time. Siblings/nr of children in the house 0 0 0 0 0 0 0 -0 (7/8, 88%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between having siblings/ the number of children in the house and screen time. First-born + 0 0 0 0 (3/4, 75%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between being first-born and screen time. BMI 0 0 0 -0 (3/4, 75%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between BMI and screen time.

Behavioral attributes and skills-related correlates
Sleep duration ---(2/2, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a negative association between sleep duration and screen time. Physical activity 0 0 0 (2/2, 100%) Based on consistent finding among studies with low-to-moderate methodological quality, there is no significant evidence for an association between physical activity and screen time.

Cognitive, emotional or psychological correlates
Temperament/personality 0 0 0 0 (3/3, 100%) Based on consistent findings among studies with low-moderate methodological quality, there is in no significant evidence for an association between temperament/ personality and screen time.

Physical environmental correlates
Electronic devices/computer/ TV in (bed)room where child sleeps + + + + + + + + + 0 0 0 þ (9/12, 75%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a positive association between having electronic devices in the (bed) room where the child sleeps and screen time. Electronic devices/screen-based/TVs at home + + + + 0 0 0 ? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between having electronic devices at home and screen time. House features (size, presence yard, fences, covered outdoor area) ---(2/2), 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a negative association between house features and screen time. Neighbourhood-related factors 0 0 0 0 -0 (4/5, 80%) Based on consistent findings among studies with low-to-moderate quality methodological quality, there is no significant evidence for an association between neighborhood-related factors and screen time. Toys availability + -? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between having toys and screen time.

Economic correlates
Parental education 0 0 0 0 0 0 ----------? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between parental education and screen time. Family income 0 0 0 0 0 0 0 0 --------? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between family income and screen time. Socio-economic variables 0 0 0 0 0 (4/4, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between socio-economic variables and screen time. Parental employment + + + 0 0 ---? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between parental employment and screen time. Rural vs urban living area ---(2/2, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a negative association between living area and screen time. Marital status/parents live together 0 0 0 0 0 0 0 (6/6, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between marital status/ parents living together and screen time.

Sociocultural correlates
Parental screen time/media use/ modelling + + + + + + + + + + + + + + + + + 0 0 0 þ (17/20, 85%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a positive association between parental screen time and children's screen time. Rules around screen time + 0 0 0 0 ---------? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between having rules around screen time and screen time. TV on at home + + + þ (3/3, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a positive association between having a TV at home and screen time.

Evidence synthesis
Monitoring screen time ---(2/2, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a negative association between monitoring screen time and children's screen time. Away from home care 0 0 ------(6/8, 75%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a negative association between being away from home care and screen time. Placing high importance and value on physical activity ---(2/2, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a negative association between placing high importance and value on physical activity and screen time. Descriptive norms + + + þ (3/3, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a positive association between descriptive norms and screen time. TV on during meals/snacks + + 0 0 ? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between having a TV on during meals and screen time. Parental physical activity 0 0 0 (2/2, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between parental physical activity and screen time. Park visits/active places 0 -? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between park visits and screen time. Parental weight status + 0 0 0 0 0 (4/5, 80%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between parental weight status and screen time. TV on to control behaviour + 0 ? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between having a TV on to control behavior and screen time. Support/reinforcement for physical activity from other adults + -? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between support for physical activity from other adults and screen time. Parental sex 0 0 0 (2/2, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between parental sex and screen time. Parent-child co-watching + -? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between parent-child cowatching and screen time. Parent-child interactions/time with child 0 -? Based on inconsistent findings among studies with low methodological quality, there is insufficient evidence for an association between parent-child interactions and screen time. Parental health and wellbeing 0 0 0 (2/2, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between parental health and well-being and screen time. Positive outcome expectations 0 0 0 (2/2, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between positive outcome expectations and screen time. Parental depression + 0 0 0 0 (3/4, 75%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between parental depression and screen time. Parental mental health: stress + + 0 0 ? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between parental mental health and screen time. Parental age + 0 0 0 0 0 0 (5/6,83%) Based on consistent findings among studies with low-to-moderate methodological quality, there is no significant evidence for an association between parental age and screen time. Parenting styles/strategies + -? Based on inconsistent findings among studies with low methodological quality, there is insufficient evidence for an association between parenting styles and screen time. Parental attitudes + + + + + + 0 0 0 ? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between parental attitudes and screen time. Negative outcome expectations + 0 ? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between negative outcome expectations and screen time. Confidence to limit TV time 0 -? Based on inconsistent findings among studies with low-to-moderate methodological quality, there is insufficient evidence for an association between the confidence to limit TV time and screen time. Parental self-efficacy 0 -------(6/7,86%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a negative association between parental selfefficacy and screen time. Country + + þ (2/2, 100%) Based on consistent findings among studies with low-to-moderate methodological quality, there is moderate evidence for a positive association between countries and screen time.
a bold indicates results from high quality study; b summary score: + = consistent positive association, -= consistent negative association, 0 = consistent no association,? = inconsistent findings; score is based on all studies unless 2 or more high quality studies were available. Abbreviations: BMI = body mass index, PA = physical activity, TV = television.
association of age and race/ethnicity with screen time.

Behavioral attributes and skills correlates
Three studies examined behavioral attributes and skills-related correlates. Two studies examining sleep duration found a negative association with screen time (Xu, 2016;. Two studies examining physical activity found no significant association with screen time . In summary, we found moderate evidence for a longer sleep duration being associated with less screen time and we found no evidence for an association of physical activity with screen time.

Cognitive, emotional and psychological correlates
Three studies examined the association between a child's temperament or personality and screen time . None of these found an association. In summary, we found no evidence for an association of a child's temperament or personality with screen time.

Physical environmental correlates
Sixteen studies examined physical environmental correlates of screen time, of which twelve studies examined the association between having an electronic device in the child's (bed)room. Nine of these twelve studies found that having an electronic device in the bedroom was associated with more screen time (Hish, 2021;Carson and Janssen, 2012;Thompson, 2018) whereas three studies found no significant association (Birken, 2011;Bleakley et al., 2013;. Four out of seven studies found that having an electronic device at home was associated with more screen time  whereas three studies found no significant association (Birken, 2011;Bleakley et al., 2013;. Four out of five studies examining the association between neighborhood-related factors (e.g., safety and living on a cul-de-sac) and screen time found no significant association  while one study found a perceived safer neighborhood associated with less screen time . Household features (e.g., having a yard, fences) were associated with less screen time in two studies Morowatisharifabad et al., 2015). Inconsistent results were found for the association between the availability of toys  and screen time.
In summary, we found moderate evidence for having an electronic device in the (bed)room where the child sleeps and household features being associated with screen time. We found no evidence for an association of neighborhood-related factors with screen time, and inconsistent evidence for an association of having electronic devices at home and the availability of toys with screen time.

Economic correlates
Twenty-nine studies examined economic correlates with household income and parental education (both n = 16 studies) mostly examined. Eight out of 16 studies found a negative association of household income with screen time (Brown, 2010;Carson and Janssen, 2012;Carson and Kuzik, 2017;Nikken and Schols, 2015;Tombeau Cost, 2020; while another eight found no significant association (Bernard, 2017;Hish, 2021;Krogh, et al., xxxx;Birken, 2011;Bleakley et al., 2013;Thompson and Christakis, 2007). Ten studies found a negative association between parental education and screen time (Bernard, 2017;Krogh, et al., xxxx;Xu, 2016;Bleakley et al., 2013;Carson et al., 2014;Nikken and Schols, 2015; and six studies found no significant association (Hish, 2021;Carson and Kuzik, 2017;Morowatisharifabad et al., 2015;Thompson and Christakis, 2007). Three out of seven studies examining parental employment found an association between (full-time) maternal employment and more screen time Birken, 2011;Bleakley et al., 2013), one study found maternal unemployment associated with more screen time , two studies found being (part-time) employed associated with less screen time (Xu, 2016;Brown, 2010), and two studies found no significant association Kim, et al., xxxx). None of the six studies examining parental marital status/living together (Bernard, 2017;Carson and Janssen, 2012;Carson and Kuzik, 2017;Tombeau Cost, 2020) or of the four studies examining socioeconomic level indicators (e.g., measures of position, levels of area deprivation) Barr, 2010; found a significant association with screen time. One study found that living in a rural area was associated with less screen time (Contreras, et al., xxxx) while another study found that living in an urban area was associated with less screen time .
In summary, we found no evidence for an association of parental marital status/living together or socio-economic level indicators with screen time. We found inconsistent evidence for an association of household income, parental education, parental employment, and living area with screen time.

Sociocultural correlates
Forty-two studies examined sociocultural correlates of screen time. When examining parental socio-demographic correlates of screen time, none of the included studies found an association between parental gender and screen time (Carson and Kuzik, 2017;Nikken and Schols, 2015). For parental age, five out of six studies found no significant association (Bernard, 2017;Krogh, et al., xxxx;Carson and Kuzik, 2017;. Two studies examined the country in which families lived and screen time (Álvarez, et al., xxxx;. One study found that children in Chile had more screen time compared to children in Colombia and Spain (Álvarez, et al., xxxx), while another study found that children from Germany had higher screen time compared to children from Spain, Belgium, Poland, Bulgaria and Greece .
Seventeen out of 20 studies found a positive association of parental screen time associated with children's screen time (Barber, 2017;Xu, 2016;Birken, 2011;Bleakley et al., 2013;Carson and Janssen, 2012;Carson et al., 2020;De Decker, 2015;Morowatisharifabad et al., 2015;Nikken and Schols, 2015;Wiseman et al., 2019), and three studies found no significant associations (Bernard, 2017;Abbott, 2016;Tang, 2018). Nine of the 14 studies found that having rules around screen time (e.g., regarding duration or content) was associated with less screen time (Thompson, 2015;Xu, 2016;Birken, 2011;De Decker, 2015;Sanders, 2016;Tang, 2018;Thompson, 2018;Wiseman et al., 2019). Four studies found no significant association (Barr, 2010;Bleakley et al., 2013; while one study found a positive association . Five studies found that being in childcare compared to being at home was associated with less screen time (Hish, 2021;Xu, 2016;Carson and Kuzik, 2017; whereas two studies found no significant association . Two out of four studies found that having the TV on during meals or snacks was associated with more screen time (Xu, 2016;Tang, 2018), while two studies found no significant association (Birken, 2011;Thompson, 2018). Having a TV on at home (Barber, 2017;Xu, 2016;Wiseman et al., 2019) or parents finding a higher amount of TV time suitable (descriptive norm) (Carson and Janssen, 2012; were both associated with more screen time in all included studies examining this correlate. Monitoring screen time was associated with lower screen time in two studies (Tang, 2018;Wiseman et al., 2019).
Two studies found no significant association between the level of parental physical activity and children's screen time . When examining parenting styles, children of authoritarian mothers or partners or permissive mothers had more screen time in one study , while nurturing authoritative parenting was associated with less screen time in another study . Other associations with inconsistent results among two studies were between parental/adult support for physical activity Wiseman et al., 2019), turning on the TV to control behavior (Tang, 2018;Wiseman et al., 2019), the number of visits to the park or active places , parent-child co-TV-watching (Birken, 2011;Bleakley et al., 2013) or parent-child interactions Morowatisharifabad et al., 2015) and screen time.
Parental attitudes were examined in nine studies of which six found a less concerned attitude around screen time associated with higher screen time (Barber, 2017;Carson and Janssen, 2012;De Decker, 2015; and three studies found no significant association (Hinkley, 2017;Nikken and Schols, 2015;Sanders, 2016). Parents placing high value and importance on physical activity was associated with less screen time in two studies Wiseman et al., 2019).
Twelve studies examined parental (mental or physical) health as a correlate of screen time. Five studies examined parental weight status of which four studies found no significant association between parental weight status and screen time (Bernard, 2017; and one study found a positive association (Brown, 2010). Three out of four studies found no significant association between parental depression and screen time Kim, et al., xxxx) and one study found more depressive symptoms associated with more screen time . Two studies examining parental stress found higher levels of parental stress associated with more screen time (Thompson and Christakis, 2007;Tombeau Cost, 2020) whereas two studies found no significant association (Barber, 2017;Kim, et al., xxxx). Parental wellbeing (including having a physical disability) was examined in two studies, which both demonstrated no significant association with screen time .
Six out of seven studies found a higher parental self-efficacy to limit screen time associated with lower screen time (Hnatiuk, 2015;Carson and Janssen, 2012;Sanders, 2016) while one study found no significant association . Two studies found no significant association between positive outcome expectations from screen time and screen time . When examining negative outcome expectations from screen time, one study found a positive association  while another study found no significant association . One study found higher parental confidence associated with less screen time  while another study found no significant association .
In summary, we found moderate evidence for an association of country, parental screen time, having a TV on at home, descriptive norms, placing high value and importance on physical activity, monitoring screen time, being away from home care and parental self-efficacy with screen time. We found no evidence for an association of parental sex, age, physical activity, weight status, depression, wellbeing, and positive outcome expectations with screen time. We found inconsistent evidence for an association of having rules around screen time, having the TV on during meals/snacks or to control behavior, parental support for physical activity, park or active place visits, co-watching, parent-child interactions, parenting styles, parental attitudes, concerns, mental or physical health, negative outcome expectations and confidence to limit TV time with screen time.

Overview of findings
This systematic review aimed to summarize the evidence on correlates of screen time in children aged 0-5 years. For a number of correlates we found moderate or no evidence for associations with screen time, however, we were unable to draw strong conclusions as for the majority of correlates we found inconsistent or insufficient evidence.
The methodological quality of included studies was rated as moderate or low for almost 95% of studies, with the majority of studies having a low methodological quality. Only two studies had a high methodological quality. Over half of the assessed studies used questionnaires that are not or have unknown validity and/or reliability to assess screen time in this young age group. There was a large variation in type of screen time measured (e.g., total screen time, TV time, hand-held devices). Many types of screen time (e.g. hand-held devices, electronic media use) were only studied in one to three studies, and only TV time and total screen time were studied sufficiently to report correlates separately. As correlates might differ for different types of screen time, future studies should explore various types of screen time, including using smartphones and tablets. The need to report and/or improve psychometric properties of screen time assessments was recently highlighted in a systematic review examining assessment in early childhood (Byrne et al., 2021;Arts, 2022). Additionally, studies scored low on the items 'selection bias' and 'participation rate'. For selection bias, the low scores are due to a combination of non-representative samples and lack of information on recruitment rates. For participation rate, this was mostly due to a lack of information. Additional information provided through references did not yield the required information. Finally, most studies had a cross-sectional design where as a longitudinal design provides stronger evidence.
Over 55% of the included studies were published in the last five years, confirming screen time in early childhood is a quickly developing area of research and the pertinence of our review. Overall, our results are in line with previous reviews (Hinkley, 2010;Hoyos Cillero and Jago, 2010;Duch, 2013). There is inconsistent or insufficient evidence of a large number of examined correlates, highlighting the need for more research in this area. Our results provided new evidence for correlates that were absent or undetermined in previous reviews (e.g., child's sleep duration and parental age) as well as confirmed results from previous reviews (e.g., child sex and parental employment).
The inconsistent or insufficient evidence (70% of examined correlates) might be explained by the large variation of correlates examined as well as the measures used to assess them. When possible, studies examining a similar construct but using different measures were therefore combined. Compared to previous reviews, studies included in our review focused more often on the different types of screen time, including handheld devices in addition to TV viewing (Vanderloo, 2014;Xu et al., 2015;Duch, 2013).
Most examined correlates were related to parents, underscoring the important role parents play in young children's lives through role modelling and creating a home environment (Rhee, 2008). This reviews builds on previous reviews when examining parental correlates by adding to the evidence-based in this area of research for several correlates (e.g., monitoring screen time and parental physical activity). Even though we found inconsistent or insufficient evidence for most included correlates, we were able to draw conclusions on 15 sociocultural correlates. Results were in line with previous reviews for parental screen time (Hoyos Cillero and Jago, 2010;Xu et al., 2015) and self-efficacy (Xu et al., 2015). We found different results compared to Hoyos et al. (2010) for parental BMI, parental depressive symptoms (no evidence in this review compared to a positive association in previous review) and rules around TV use (insufficient evidence in this review compared to a negative association in previous review). These differences can be explained by different inclusion criteria: Hoyos et al. (2010) included studies with a mean child's age up to 7 years and studies using adherence to guidelines (yes/no) as an assessment of screen time (Hoyos Cillero and Jago, 2010). The moderate association found between country and screen time are likely caused by underlying differences in cultural, socio-economic or other factors, and need further investigation.

Strengths and Limitations
Strengths of this review are the examination of correlates per type of screen time (although there was insufficient evidence to present the results per type), the methodological quality assessment, the best evidence synthesis and the inclusion of a large number of correlates and the full early childhood age range (0-5 years).
As there was a large variation in how correlates were defined and measured, we combined studies examining a similar construct, which might have led to more inconsistent results. Additional limitations are not contacting authors of the original studies for more information, due to practical reasons, and only including English publications. Furthermore, we could not evaluate potential publication bias or selective reporting of significant findings, which could have influenced our results.

Recommendations for future research
To further examine correlates of screen time in early childhood and to inform future interventions we recommend: Focusing on sociocultural correlates, especially parental factors, physical environmental factors, behavioral factors and economic factors, as there are various correlates with undetermined evidence in these categories.
Investing in better methodologies and improved reporting to ensure high quality studies, including: conducting longitudinal studies; the use of valid and reliable assessment of different types of screen time (e.g., smartphone use, television viewing) and potential correlates; recruiting more representative samples; reporting of study details, especially around recruitment and participation rates.
Exploring interactions between factors that could influence screen time, as currently few studies examined such interactions (e.g. gender differences, ethnicity and culture).

Conclusions
This systematic review summarized the correlates of screen time in typically developing children aged 0-5 years. Based on these results, we recommend future interventions to focus on factors related to the physical environment (e.g., the presence of electronic devices in the bedroom) as well as on sociocultural factors (e.g., parental self-efficacy). However, the evidence for most included correlates was inconsistent or insufficient and as such more high-quality research is needed to inform future interventions.

Declarations
Ethics approval and consent to participate: Ethics approval was not needed as this study was based on publicly available anonymized data.

Funding
This study was funded by the Netherlands Organization for Health Research and Development (ZonMw; Project No. 546003008) and the Bernard van Leer Foundation. The funding bodies had no role in the design of the study; the collection, analysis, and interpretation of data; or the writing of the manuscript.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

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