Introduction

Excess adiposity in childhood is a major public health issue, it is associated with a wide range of negative physical and psychological health outcomes [1, 2]. Socio-ecological models provide a useful framework for understanding the different factors contributing to childhood obesity risk [3], theorizing that children are shaped by the environments they interact with most often. The home environment and family context are where children spend a significant proportion of their time during key developmental years [4]. Around 70% of a child’s food (for children < 12 years old) is consumed at home and importantly, it is where children observe and learn from others’ behaviour [5,6,7,8]. Consequently, it is hypothesised that the home environment is a major factor in shaping children’s weight trajectories.

Numerous models have been developed to conceptualise how different aspects of the home environment may influence children’s growth and development [5, 9, 10]. Yet the multifaceted and complex nature of the home environment (HE) complicates attempts to characterise and measure its contribution to excess weight development in childhood. A variety of measures have been developed to capture different aspects of the obesogenic HE, such as the types and frequency of foods available in the home [11] or the availability of electronic devices in a child’s bedroom [12]. Relationships have been observed between these measures and children’s energy-balance behaviours, including dietary intake [13, 14], activity levels [15], and screen-based sedentary behaviours [16, 17]. However, the extent to which the HE is directly associated with child adiposity is less clear and no previous systematic reviews have examined this.

For the purpose of this review, the obesogenic HE has been partitioned into three domains hypothesised to influence children’s food intake, activity levels and sedentary behaviours [9, 10, 14, 18]; the food (e.g. availability of sugar sweetened beverages [SSB]), physical activity (e.g. access to a garden) and media-related (e.g. caregiver rules around electronic devices) domains within the home. As shown in Fig. 1, each domain of the HE can be sub-divided into both physical aspects and social aspects that can either deter or promote health behaviours. These are all hypothesised to influence child energy balance and, ultimately, body weight.

Fig. 1
figure 1

Conceptual model used to define the ‘obesogenic’ home environment [10]

Full size image

Despite the wealth of literature in this field, previous reviews have largely focussed on only one domain of the HE, for example, the home food environment [19, 20], the home media environment [12], or the home physical activity environment [21]. Most of these reviews also explored relationships between the HE and behavioural outcomes such as children’s diet, activity and/or sedentary behaviour, meaning direct associations with adiposity have not been explored by previous reviews. Only one review from a decade ago included all three HE domains, but it primarily explored the psychometric properties of existing measures, rather than evaluating relationships with child weight [18]. No previous review has synthesised the evidence to investigate relationships between the obesogenic HE in its entirety, and child adiposity outcomes.

It is important to understand how the different aspects of the home environment relate to excess weight gain in childhood in order to inform effective targeted child obesity prevention and intervention strategies. Therefore, the purpose of this review is to examine associations between physical and social aspects of the food, activity and media domains of the HE with measures of adiposity in childhood (≤12 years).

Methods

This review followed the PRISMA guidelines (Additional file 1 - PRISMA checklist) and was registered on PROSPERO (CRD42018115139).

Eligibility criteria

Manuscripts were included if they reported on at least one of the three domains of the HE (food, activity and/or media). Each domain was required to be assessed in terms of either the physical (availability of, and access to; foods, media or PA equipment) and/or social aspects (caregiver modelling/support or caregiver policies and rules around energy balance behaviours). Studies were also required to provide a quantitative estimate of the association between the chosen HE domain(s) and a measure of adiposity (e.g. BMI z-score). Studies were eligible for inclusion if they were peer-reviewed original observational research studies and recruited from non-clinical, non-intervention populations. The population of interest was children aged ≤12 years. This age range was chosen to broadly capture the upper age of primary school children, which can vary both between and within countries (for example children in the U.K. and Australia typically start secondary school at age 11, whereas in Singapore and the Netherlands children tend to start secondary school at age 12). Additionally, the upper limit of 12 years was selected to focus this review on pre-teenage years, before children’s autonomy over their environment increases and they spend more time outside the home. Studies were excluded if they were not published in English and no translation was available (n = 41).

Family mealtimes were excluded from the definition of the social home food environment as a recently published review examined the relationship between family mealtimes and child weight [22]. This meta-analysis found higher family meal frequency was associated with better overall diet quality, greater consumption of nutrient dense foods and fewer energy-dense foods and lower child BMI.

Literature search strategy

Six electronic databases were originally searched in November 2018, and this was then updated in October 2020: Medline (OVID from 1946 to Oct 2020), EBSCO CINAHL, EMBASE (OVID), Web of Science, PubMed, and PsycInfo (OVID). The search strategy (see Additional file 2) was informed by search terms from a relevant review [18]. Database searches were supplemented by reading the reference list of eligible studies and relevant reviews in the area [18].

Identification of relevant studies and data extraction

Study eligibility was assessed independently by 2 reviewers; with 5% of title and abstracts and 10% of full texts screened in duplicate. There was 96% agreement between the reviewers, with any disagreement resolved via discussion. A standardised format for extraction was developed to ensure detailed data were obtained from each included study. Data extracted included key study and sample characteristics (e.g. study design, sample size, demographics), aspects of the HE examined (e.g. availability of physical activity equipment, caregiver modelling, etc.) and details of the child weight-related outcomes (e.g. measures used, population reference data, obesity cut-off criteria). The strength and direction of relationships between HE aspects and adiposity measures were also extracted.

Assessment of study quality

Risk of bias was completed for each included study using an adapted version of the validated Newcastle Ottawa Scale (NOS) for cohort studies [23]. The tool was used to evaluate studies based on the research design, representativeness of the sample, appropriateness of the statistical analysis, recruitment strategy, measurement of exposure, and use of power calculation. A NOS score ≥ 7 was considered indicative of high study quality. The maximum score that could be awarded for study quality was 10. Full details are described in Additional file 3.

Results

Overall, the search strategy identified 21,747 independent publications. Following title and abstract review, 12,257 were excluded and a further 367 papers were excluded after assessment of the full texts. An additional seven papers were identified through searching relevant manuscript reference lists during the screening stage. In total, 62 studies met the inclusion criteria. Figure 2 outlines each stage of the study selection process.

Fig. 2
figure 2

PRISMA Flow Diagram of the systematic review literature search

Full size image

Characteristics of included studies

Table 1 summarises the key characteristics and findings from the 62 included studies. Overall, most studies were undertaken in North America (n = 20), Europe (n = 17) or Australia (n = 8), with only few studies undertaken in Asia (n = 7). Across the 62 included studies, there were a total of 105,268 children ranging in age from 18 months to 12 years, but the majority of studies (n = 45) focused on primary school aged children aged between 5 and 12 years. Seven studies were exclusively in pre-school aged children (< 5 years) [24, 40, 70, 80, 82], while 10 studies involved children spanning a broader age range (from pre-school up to 12 years) [34, 42, 47, 56, 61, 63, 64, 81]. Most studies examined associations between home environments and child body weight cross-sectionally (n = 51), with fewer prospective studies (n = 11). Most studies examined a single domain (n = 41) or two domains (n = 16), with only five studies assessing all three pre-defined domains of the HE [78,79,80,81,82].

Table 1 Characteristics and results table for included studies (n = 62)
Full size table

Most studies were published in the 5 years (n = 23; 37.1%) or 5–10 years (n = 30; 48.4%) prior to this review. Fewer studies had been published before 2009 (n = 9; 14.5%), with the earliest study published in 1985. A summary of the association between the food, PA and media domains and child adiposity are presented in Table 2 and Additional files 4, 5 and 6.

Table 2 Summary of cross-sectional associations between food, physical activity and media domains and child adiposity outcomes (n = 58a)
Full size table

Characteristics of HE and adiposity measures

There was substantial heterogeneity in the measurement tools used to examine the HE. Measurement methods varied, including comprehensive measures of all three domains of the HE such as the Home Environment Interview [82] and Family Nutrition & Physical Activity Screening Tool [79], and measures of one or two domains of the HE such as the Family Eating and Activity Habits Questionnaire [47], the Family Food Environment Questionnaire [52], and The Home Self-Administered Tool for Environmental Assessment of Activity and Diet (HomeSTEAD) [63]. Other measures were less comprehensive, using single items or a limited number of items to measure only one aspect of the HE (e.g. availability of television in the bedroom). Most studies exclusively used caregiver and/or child self-completed (n = 49) or interviewer-delivered (n = 4) questionnaire methodology. Few studies (n = 4) used in-home observation, and this was generally used in addition to questionnaire-based measures.

Measures of adiposity were taken via trained researcher (n = 48), parent-report (n = 5) or a combination of both (n = 2). Most studies (n = 61) used height and weight measurements to derive BMI, BMI z-score or BMI percentile. These measurements were taken using a calibrated weighing scale and stadiometer (n = 53) or with parents own scale and measuring tape (n = 9). Only nine studies supplemented height and weights with additional adiposity measures, such as body fat percentage (BF%) (n = 1), skinfold thickness (n = 1), Fat Mass Index (n = 1), waist circumference (n = 3), waist-to-height ratio (WtHR) (n = 1), or a combination of these (n = 2). Only one study used BF% as the primary outcome (n = 1).

Relationship between home environment and adiposity outcomes

Media domain

Twenty-nine studies examined physical aspects of the home media environment, with most studies (21/29; 72.4%) demonstrating positive associations between availability and access to electronic media equipment in the home and measures of child adiposity. This association was observed across children aged 3–12 in both cross-sectional (n = 19) [16, 26,27,28,29,30,31,32, 34,35,36,37,38,39,40, 44, 67, 68, 76, 79] and longitudinal studies (n = 2) [25, 35]. One large prospective study of 12,556 children from the U.K. reported that having a TV in the bedroom at age 7 was associated with increased risk of children developing overweight at ages 7 and 11, compared to those without a TV in their bedroom [35]. Similar results were reported in another prospective study following UK children from ages 9/10 to 10/11 (n = 2064) [25]. Over the past decade, there have been considerable changes in children’s use of screens, with a decline in TV viewing and an increase in use of other devices (e.g. tablets, mobile phones, laptops) to access video content online [83]. As such, more recent studies have expanded the scope of their measurement tools to capture a broader range of electronic devices in the home (e.g. games console, mobile phones, tablets, laptops etc.). These studies suggest that the number of electronic devices, as well as the types of devices available may have implications for child weight, with a greater number of electronic devices present in the home associated with higher BMI z-scores in children aged between 9 and 12 [29,30,31,32, 37, 38]. A study of children aged 9–11 (n = 502) reported that those with multiple devices (2–3 or more) in their bedroom had higher body fat percentage than children with no devices in their bedroom [29]. The remaining studies (8/29; 27.6%) reported null associations between availability of media equipment and child adiposity [33, 42, 45, 63, 64, 69, 70].

Sixteen studies examined relationships between social aspects of the home media environment and child adiposity [15, 24, 34, 41,42,43, 63,64,65,66, 68, 69, 76, 79,80,81]. Of these, five studies measured caregiver modelling of media use [43, 64, 65, 77, 80]. Two studies conducted in children aged 2–5 years [80] and 3–12 years [77] reported positive associations between caregiver modelling of media use and child BMI. The remaining three studies, involving children aged 2–6 years [64], and 7–12 years from Australia [65] and Taiwan [43], found no associations between caregiver modelling of media use and child adiposity. Sixteen studies examined relationships between caregiver rules and limit setting around media use and child adiposity, with inconsistent findings [15, 24, 34, 41,42,43, 63,64,65,66, 68, 69, 76, 79,80,81]. Four cross-sectional studies reported associations between fewer caregiver limits and less monitoring of media use with higher risk for abdominal obesity in Australian primary school aged children (aged 6–10 years, n = 3884) [68] and higher BMI scores in Malaysian children aged 10–12 (n = 802) [66], and preschool aged children from the US (n = 8550) [24] and South Korea (n = 241) [80]. One longitudinal study from the US (n = 213) reported similar findings with less caregiver monitoring of media use predicting steeper increases in children’s BMI z-scores from ages 5 to 9 [41]. However, two prospective studies from the Netherlands which pooled data from the Dutch KOALA birth cohort (n = 1694 and n = 1819) found the reverse relationship, with limits on electronic media use associated with greater increases in child BMI z-scores from ages 5 to 7 [15, 81]. Nine studies (56.3%) reported no relationships between caregiver rules and limits around media use and child weight [34, 42, 43, 56, 65, 69, 76, 79, 84]. Findings are summarised visually in Additional file 4.

Food domain

Fifteen studies examined relationships between physical aspects of the home food environment and child adiposity, demonstrating inconsistent findings. Six studies reported associations between food or drink availability and access, and child adiposity [46, 47, 49, 51, 70, 75]. Greater availability of nutrient dense foods (e.g. fruits and vegetables) were associated with lower BMI- z-scores in Czech 10 to 11 year olds (n = 97) [49], while higher availability and access to energy-dense foods (e.g. SSBs, sweets, etc.) predicted higher BMI among Chinese 3–6 year olds (n = 222) [47] and Greek children aged 9 to 12 years (n = 335) [46]. But conversely, one study of Mexican children (n = 684) aged 10–11 reported the opposite relationship; children with OW/OB had greater access to fruits and vegetables and less access to energy-dense foods (e.g. confectionary items, cookies, SSB and salted snacks) in the home [51]. Eight studies exploring food availability and accessibility in the home with measures of child adiposity found no relationship between them [13, 48, 50, 52, 53, 55, 71].

In total, eleven studies assessed relationships between social aspects of the home food environment and child adiposity, but these varied in their scope. Eight assessed caregiver modelling and/or support of eating [54, 57, 69, 71, 72, 79,80,81], one assessed caregiver rules and limit setting around food [68] and two assessed both caregiver modelling of eating and caregiver rules/limit setting [13, 56]. Of the ten studies examining the role of modelling, three demonstrated associations between caregiver modelling and/or support of healthy food intake and lower child BMI z-scores; this was observed both cross-sectionally in U.S. children aged 6–11 (n = 699) and aged 6–7 (n = 854) [13, 79] and longitudinally in Dutch children followed from ages 5 to 7 [81]. Contrastingly, one cross-sectional study conducted in South Korean preschool children (n = 241; 2–5 years) found greater modelling of healthy eating was associated with higher child BMI [80]. Null associations between caregiver modelling and/or support of food intake and child adiposity were observed in the remaining six studies [54, 56, 57, 69, 71, 72, 76]. A further three studies assessed caregiver rules and policies around unhealthy eating and found null associations with child adiposity across all studies [13, 56, 68]. The findings are summarised in Additional file 5.

Physical activity domain

Seven studies examined relationships between the home PA environment and measures of child adiposity. Two cross-sectional studies reported negative associations between availability and access to PA equipment and child adiposity outcomes [61, 63]. A large UK-based study (n = 6467) of 3–7 year olds reported children with access to garden space had lower odds of OW/OB compared to those without garden access [61]. These findings are consistent with a US study of children aged 3 to 12 years which found both the amount and condition of the PA equipment available in the HE was associated with lower BMI percentile [63]. Conversely, one study (n = 114) of 12 year olds in Puerto Rico reported greater access and availability of recreational and sports equipment at home was associated with higher child BMI [75]. The remaining three cross-sectional [62, 67, 76] and one longitudinal [58] study all reported null associations between access and availability of PA equipment and child BMI.

Thirteen studies explored social PA environments in the home in relation to child adiposity. Two studies, a cross-sectional study of 854 children aged 6–7 from the U.S. and a longitudinal study of 879 Polish children aged 6–11 reported caregiver modelling and support of PA were associated with lower child BMI z-scores [60, 79]. In contrast, a study of pre-schoolers from the Netherlands (n = 1554) found relationships with child adiposity varied according to the type of activity modelled by the parent. Higher levels of active travel (e.g. commuting to work via walking or bike) were associated with lower BMI z-scores in children aged 3–4 years old, but there were no associations for other types of modelled vigorous activity (e.g. running) [59]. One cross-sectional study examined caregiver modelling, support and encouragement of PA as separate constructs and found variation in the association with adiposity outcome; with null associations observed for caregiver modelling of PA, while encouragement and support of PA were associated with lower BMI percentiles [77]. The remaining nine studies reported null associations between caregiver modelling and/or support of children’s PA and measures of adiposity [15, 58, 65, 66, 71, 75, 76, 80, 81]. Findings are summarised in Additional file 6.

Composite scores of multiple domains of home environment

Composite scores of the HE reflect the combined contribution of more than one of the pre-defined domains – food, PA or media - within the HE. In this review, we identified four studies which generated composite scores across two (n = 2) or three (n = 2) domains of the HE and examined associations with child adiposity [73, 74, 78, 82], with mixed findings. One cross-sectional study of U.K. children aged 4 years (n = 1096) developed a composite score for the overall physical and social aspects of the HE, and for each of the pre-defined domains – the food, PA and media environment. No relationship with child BMI z-scores were observed for either the overall HE composite score, or any of the food, activity and media domains [82]. A second study, conducted in children aged 8–12 in Bangkok (n = 280), created a composite of the physical food and PA environments. Composite scores indicating a ‘lower quality’ (i.e. more obesogenic) HE were associated with 2.8 times higher risk of child obesity [74]. A third cross-sectional study of U.S. children aged 5–12 (n = 233) took a slightly different approach, developing a composite score called ‘Parent behaviours’ which assessed caregiver modelling of healthy eating and caregiver modelling and support of PA. No associations with child BMI z-scores were observed [73]. The fourth cross-sectional study of children aged 8–12 from the Netherlands (n = 1480), examined physical and social aspects of the HE by combining items into clusters capturing caregiver practices relating to food, PA and media-related energy balance behaviours. The study reported a weak positive association between ‘diet- and activity-related modelling’ and child BMI z-scores, suggesting that children with parents who exhibit greater modelling of healthy eating, less sedentary behaviour and who live in a home with greater availability of PA equipment actually had higher BMI z-scores [78]. Null associations were observed for the remaining four constructs (‘low availability of unhealthy food’, ‘High visibility and accessibility of screens and unhealthy food’, ‘diet and activity related positive modelling’, ‘positive modelling on sports and fruit’) and BMI z-scores. As these studies utilised a composite scoring system, it was not possible to establish the independent contribution of individual aspects of the HE on weight.

Risk of bias

Overall, 38/62 (61.3%) of the identified studies were rated as high quality based on the NOS quality assessment criteria. The most common methodological weaknesses were in selection and comparability of studies; with 37/62 of studies (59.7%) providing inadequate justification of sample size, 49/62 (79%) providing an inadequate description of response rate or lack of comparison between respondents/non-respondents, and 22/62 (35.5%) failing to control for important confounding factors such as age, sex, SES, energy balance behaviours or parental adiposity. Full details and individual study scores are described in Additional file 3.

Discussion

This is the first systematic review to appraise and synthesise the evidence for associations between the physical and social aspects of the food, PA and media domains of the HE with measures of adiposity in childhood (≤12 years). The most consistent associations were observed for the physical aspects of the home media environment, with greater availability and access to electronic media devices in the home, and specifically in the child’s bedroom, associated with higher risk of adiposity (21/29 studies). Findings were less consistent for the smaller number of studies examining physical aspects of the home food or PA environments. Half (8/15) of the studies examining physical food environments reported null associations, while similar numbers (6/15) demonstrated positive associations between more obesogenic food environments and higher child adiposity. Findings were similarly mixed for PA environments; with 4/7 reporting null associations, 2/7 reporting negative associations and 1/7 study reporting positive associations between access to physical activity equipment/garden space and adiposity. Fewer studies assessed social aspects (e.g. caregiver modelling or limit setting) of the home environment in relation to child adiposity and findings were again mixed; 9/16 media environment, 7/11 food environment and 9/13 physical activity environment studies reported null associations with child adiposity outcomes.

Although research has shown that children learn behaviour from those around them [7, 19, 85], we found limited evidence that behaviours learned at home translate into child weight outcomes. Caregiver modelling of behaviours, across the food, activity and media domains of the HE, were not consistently associated with child adiposity measures in expected directions. The variation in findings may in part result from a lack of consensus in how these constructs are defined and measured. Some studies defined modelling simply as how often parents consumed specific foods or beverages, or the length of time parents spent engaging in activities (e.g. PA or screen-based activities). This approach fails to consider a fundamental aspect of modelling - the frequency with which a child observes these behaviours. Additionally, associations with adiposity outcomes were largely explored cross-sectionally [64], hindering understanding of the directionality of associations, and failing to capture variations over time depending on children’s age, family circumstances. One of the few studies to examine the prospective relationship between social aspects in the HE (e.g. caregiver modelling, caregiver rules/limit setting) and adiposity, revealed that caregiver encouragement of ‘healthy eating’ at age 5 was associated with lower BMI z-scores at age 7 and caregiver restriction of sedentary time at age 5 were associated with higher increases in BMI from age 5 to 7 [81]. Importantly, this study was one of the first to simultaneously examine the influence of multiple aspects of the social HE. Such holistic approaches are important to incorporate a range of factors potentially contributing to child weight development.

Over half of the studies included at least one measure of the home media domain and it was the aspect of the HE most consistently linked to child weight outcomes. This is perhaps unsurprising as greater availability of media in the home has been shown to be associated with weight-related energy balance behaviours; increased sedentary behaviour, decreased activity levels and increased snacking [86]. The more consistent relationship between child adiposity and the physical home media environment may partly result from the fact it is more stable, less complex, and therefore easier to characterise and measure than the home food environment. Unlike the food environment, the media environment is unlikely to fluctuate from day-to-day or vary with seasonal changes to the same extent. It is also arguably easier to report the number and location of electronic devices in the home, than of food and beverage products. Drawing on this point, studies examining availability of foods and beverages in the home were cross-sectional, collecting data at a single time point. This approach fails to capture fluctuations in the types and/or amount of foods and beverages available in the home over the course of days, weeks and months. Thus, the foods available at data collection may not reflect the foods that are typically available within the home [11]. It is important for studies to account for this variation in the measurement. The general lack of longitudinal studies identified in this review also means we cannot conclude if features of the HE are driving excess weight gain in children or whether any observed associations result from parents modifying HEs in response to their child’s weight status (or weight related behaviour). For example, parents of children with higher adiposity may reduce the availability of energy dense foods at home in an attempt to improve their child’s dietary intake and achieve a healthier weight status.

Another possible explanation for the heterogeneity in findings across studies may be due to variation in the degree of adjustment in statistical models. Seven out of fifteen studies examining associations between food availability and child adiposity failed to include important potential confounding variables (e,g, SES or parental adiposity) in the statistical model. Controlling for such confounding variables is important to understand the true association between aspects of the HE and adiposity. This fact is highlighted in the results of a large study of 6–10 year olds in Australia (n = 3884) which reported results based on unadjusted and adjusted models. In the unadjusted model, greater availability of SSBs was associated with higher odds of OW or OB and abdominal obesity [68]. Conversely, in the statistical model adjusted for age, sex, SES and meeting recommended PA levels, no association with adiposity was observed. Varying degrees of adjustment in the included studies may partially account for variation in findings between studies.

Despite the arguably more straightforward composition of the home media environment, the measures utilised by studies included in this review were limited in scope and rarely captured the diversity of electronic devices currently available to children [87]. Most studies (n = 16/29) focussed on availability and access to televisions and/or computers within the home, perhaps reflecting the fact that growth in commercially available electronic devices is relatively recent. Ofcom figures reveal U.K. electronic device ownership and use has increased substantially over the last decade, with tablet ownership among 5–15 year olds, rising from 2% in 2011 to around 50% in 2018 [83]. The most comprehensive measure of the media environment was utilised by Canadian researchers, Dube et al. [28] and Chahal [30], who collected data in a cohort of children aged 10 to 11 (n = 2334 and n = 3398 respectively). Home availability of multiple electronic devices, including TVs, DVD players, computers, video game consoles, tablets and cell phones, were positively associated with higher child weight status [28, 30]. As home media use continues to evolve, it is important for future research to capture the increasing diversity of electronic devices available, along with use of different media platforms, when exploring the impact of the home media environment on children’s weight development.

Evidence for relationships between the home PA environment and child adiposity was very inconsistent. Most studies reported null associations with adiposity for the social PA environment (caregiver modelling and support of PA), while findings for associations between the physical PA environment (availability of and access to PA equipment) and adiposity varied. It is possible that the home PA environment is less important than the home media environment for influencing energy balance and thus weight in childhood. The home media environment has been found to contribute significantly towards sedentary behaviour [12], and energy intake [86], and was more consistently associated with weight in this review. The relationship between PA and child adiposity is complex, reviews have generally found an absence of convincing evidence for the contribution of PA to child adiposity [88, 89], which may in part be due to methodological weaknesses and imprecise measurement of PA levels. Excess weight gain has also been linked to a reduction in physical activity levels further complicating the relationship [90,91,92]. In addition, there may be age-related variation in the relationship between the home PA environment and child adiposity. The home PA environment may be more influential for younger children, who spend more of their time in the home setting. A large UK-based study (n = 6467) of 3–7 year olds reported that children with access to garden space had lower odds of OW/OB compared to those without garden access [61]. In contrast, as children reach secondary school age and gain independence, the neighbourhood and school activity environments may play a greater role in shaping energy balance behaviours and adiposity. More research measuring multiple components within the HE and in different age groups is needed to establish how and in what ways the home food, PA and media environments may interact to influence a child’s weight development.

Across all HE domains there was a notable lack of consensus on how to define and measure the HE. As a result, very few studies used validated measures (n = 20/62). Additionally, few studies attempted to characterise the holistic obesogenic HE, instead focusing on individual aspects of a single domain and often using one or two items to measure a single construct (n = 13). Studies that utilised validated measures (n = 20) tended to be more comprehensive (e.g. 15,36,96), however it should be noted this did not always result in clearer relationships with child weight outcomes [81, 82]. For example, Schrempft et al. [82] used a validated measure to comprehensively examine the three pre-defined domains of the HE but failed to ascertain associations with adiposity.

Limitations and recommendations for future research

This review is not without limitations. It was restricted to studies published in English-language, and non-clinical, non-intervention studies. As such we excluded studies in which the population received some type of intervention, for example federal support schemes (e.g. Head Start), which likely limited the number of studies included in the review and limited the number of low-income populations. Although interventions are important for determining causal relationships, this review focussed on observational studies as it is important to understand the effect of ‘real-world’ home environments before deciding how and where to intervene.

There are several problems with the current evidence base, limiting the conclusions of this review. The majority of included studies (n = 51/62) were cross-sectional. As discussed, the lack of longitudinal research (11/62 studies) means it is not possible to uncover the direction of relationships between the HE and child adiposity or infer causality. Prospective studies from birth with measurement at multiple time points are needed to identify which aspects of the HE promote or protect against excess weight gain in childhood. Ultimately robust randomized controlled trials of intervention studies will be necessary to determine which aspects of the home environment can be effectively modified to reduce excess weight gain in childhood.

The majority of included studies also predominantly relied on caregiver report and were thus susceptible to social desirability biases [93]. This may present a particular problem when exploring associations with child adiposity if reporting bias varies by weight status. For example, parents of children with overweight or obesity may be more likely to underreport availability of energy dense food in the home. However, caregiver report measures remain the best method for collecting HE information at scale, and they have been shown to be validated using objective measures (e.g. wearable cameras) [94].

There is a lack of research conducted in minority ethnic groups and low socioeconomic status (SES) families (only 15/62 studies (24.2%) considered differences in HE by SES). SES may well confound or moderate the relationship between the HE and child adiposity. For example, in economically developed countries, low SES households are more likely to have electronic devices available in the child’s bedroom compared with higher SES households [16, 95]. Little is currently known about the HE in low-income countries, or about how social inequalities influence the overall obesogenic nature of the home and how this in turn may influence children’s weight development. Alongside differences by age-groups and SES, future research should also consider individual variation in susceptibility to obesity. Not everyone interacts with the obesogenic environment in the same way [96]; for example, availability of energy dense foods may only be associated with increased adiposity in children with an avid appetite. Individual differences in susceptibility to an obesogenic environment likely influence associations between the HE and child adiposity.

Finally, heterogeneity in the measures used and a lack of consensus in both language and definitions of constructs means comparison of findings across studies is impeded. This review emphasises the need to harmonize definitions and measurement of the HE, in order to gain a reliable understanding of how factors within the home contribute to adiposity in childhood, and ultimately inform targeted family obesity prevention and treatment programs.

Conclusion

This review suggests that the most robust associations between the HE and child adiposity are observed within the physical home media environment. It is not clear whether this is due to a stronger relationship between the media environment and child weight development, compared to the food or PA environments, or whether it is an artefact of it being the HE domain most frequently investigated, and most accurately captured, in current research. This review also highlights that despite the large number of studies identified, there is a lack of agreement on how to conceptualise and measure salient aspects of the HE hypothesised to relate to health outcomes. Consensus is needed for a ‘gold standard’ measurement of the multidimensional HE. Future research should focus on utilising comprehensive measures of multiple HE domains in order to understand how, and to what extent, the different aspects of the HE interact to influence children’s weight development. Such efforts would facilitate the development of evidence-based guidance on how best to modify the HE to reduce childhood obesity risk.