Worldwide, it is estimated that 10% of school-age children are currently overweight or obese with rates continuing to rise1. In the US, the most recent data suggest that 34% of children are overweight (>85^th percentile) and 17% are obese (>95^th percentile)2. Given these alarming trends, increased attention has been focused on understanding the etiology of obesity and possible management strategies, particularly the contribution of the environment 1345. The current environment has been described as "toxic" or "obesogenic'" due to the presence of an almost unlimited, convenient supply of highly palatable, energy dense foods, coupled with conditions that encourage sedentary behaviors and discourage physical activity 146. Most environmental research to date has focused on the impact of macro-level environments such as neighborhoods and communities on healthy weight behaviors. Such research has launched a wave of 'active living' initiatives that promote building, neighborhood and road network designs that encourage communities to become more active 789.

Micro-level, or 'home' environments are also likely to correlate to overweight in children, but there has been less attention in this area of research 1610. Physical and social parameters of the home environment hypothesized to influence children's diets include the foods available in the home, parents' own eating habits, and feeding practices parents employ 11121314. For example, the availability and accessibility of foods in the home may be a major factor influence on childhood diet, since children's food intake is largely dependent upon provision of food by others 1516. Studies in this area have found that an increase in the availability of foods is related to the consumption of those foods 16171819. Many studies have considered fruits and vegetables specifically, finding that when fruits and vegetables were available, children were more likely to eat them.

In addition to making healthy foods available and accessible in the home, most researchers would agree that parents can encourage their children to eat more healthfully by adopting appropriate feeding behaviors and practices. Research has shown that what parents eat themselves (i.e., parental modeling) has a strong influence on children's food preferences and intake 1120212223. Also, parental restriction has been positively associated with overeating, or eating in the absence of hunger, especially in girls 24. The few studies examining parental disinhibition 25 and neglect 26 suggest that both may lead to poorer dietary habits in children. Family meals are also likely to influence diet in children with evidence showing that skipping breakfast (or eating breakfast less often) is related to increased body mass index (BMI) in children 27282930 and eating family dinners is associated with more healthful dietary intake patterns, including greater consumption of fruits and vegetables 31.

Access to and provision of environments which encourage or discourage physical activity are also likely to influence childhood body composition, although this area has received somewhat less attention compared to food and eating behaviors. As with diet, a child's activity may be influenced by both social and physical parameters in the home, such as parents' own physical activity habits, parents' rules and policies around play, encouragement to play, restriction of sedentary activities and provision of areas for active play. Environmental determinants of children's physical activity have been examined in a recent review by Ferreira et al32. Parental physical activity (i.e. role modeling) has been studied extensively, but with mixed results. Father's activity was most consistently associated with child physical activity. Time spent outdoors, although less studied, was an area consistently associated with higher activity levels in children. Another major area of focus has been the impact of TV watching, with many studies finding positive correlations between the amount of time spent watching TV and BMI 33. Discrepancies in findings regarding the TV-childhood obesity relationship may be due to low quality measurements of TV exposure 3435. Given the challenges in measuring the amount of time spent watching TV, assessment of factors such as the number and location of TVs and parental restriction of access to TVs should be explored as possible predictors of sedentary behaviors.

Interventions that focus on improving the physical and social environments in the homes of children need to be tested. One impediment to this research is the lack of tools with sound reliability and validity evidence that can be used to measure factors within the home that may relate to healthy weight behaviors in children. The Healthy Home Survey (HHS) was developed to address this measurement gap. The purpose of this paper is to describe the development of the HHS, as well as the reliability and validity testing of this instrument.

To our knowledge, this study was the first to develop and evaluate a tool which measures both physical and social factors within the home setting that influence diet, physical activity, and sedentary behaviors of children. The instrument was found to be feasible and the majority of items demonstrated substantial to almost perfect agreement between the two phone interviews and between the first phone interview and the home assessment.

Assessment of the home environment did have some challenges. One area in particular that needs to be revisited is the assessment of food availability in the home, especially the quantity or number of servings available. Calculating servings required a number of assumptions. As described previously, package sizes were captured as small, medium, or large. In order to calculate the number of servings, these general sizes had to be translated into more exact volume or weight quantities. This methodology likely affected the reliability and validity of these items. Although the HHS was not intended to provide a complete food inventory, it may be valuable to capture more detailed package size information in future work with this instrument.

Although the majority of items in the survey had moderate to high reliability, somewhat lower scores were noted for variety and quantity of fresh fruit, 'fruit in view', 'seconds policy', 'set meal times', 'parent eats healthy', and 'restrict outdoor play'. Low reliability scores for variety and quantity of fresh fruit (and possibly fruit in view) may reflect actual changes due to consumption or purchase during those 7 days between the first and second telephone interviews, since fresh fruits usually perish within this time frame. While current data do not allow separation of the variance attributed to natural change, it will be important to determine so that data collection methods and questionnaire items could be modified to improve the quality of data collected. We believe that measurement of the variety and quantity of foods may be a better indicator of the quality of foods in the home compared to if presence or absence alone is measured. For example, a dictomous measure could report that fruit is available in a home when just one apple is present. On the other hand, a measure of the quantity of different fruits would distinguish whether a home contained just one apple or a variety of different fruits. For the purpose of this tool evaluation study, we did not take into account the variables which are likely to influence quantity or quality, including the number of days since shopping, or the number of people living in the home. Such factors would be considered when using the measure as a means to describe a family's home environment.

Low reliability for the three questions about eating and media policies may be the result of confusion about the meaning of these questions. For example, data collectors noted that a number of participants needed additional clarification about the 'seconds policy' question before being able to answer, suggesting that some questions need to be revised. Other items with lower scores for reliability were of a more subjective nature (e.g. parent self-report of whether they eat healthy).

In general, moderate to high validity was observed; although slightly lower scores were noted for some items on variety and the food environment. The variety items with the lowest validity were salty and sweet snacks and fresh fruit. Lower validity for items measuring snack variety may have been a function of how individual participants versus trained data collectors reported similar items either together or separately. For example, if a participant had 1 bag of potato chips and 1 bag of tortilla chips in their pantry, they may have reported "2 bags of chips." Trained data collectors would have distinguished between the two and recorded 1 bag of each. As a result, calculated variety based on participant report and direct observation would be different. Food environment items that demonstrated lower validity included 'fruit in view', 'adequate play space inside' and items ascertaining the degree of accessibility of candy, soda, and snacks. The subjective nature of these items likely impacted validity estimates. For example, when interpreting what is meant by 'accessible', data collectors considered foods to accessible if children could reach the item without assistance, including if there was a chair nearby to help them reach higher items. It is possible that parents' did not take this into consideration or that they believed that their child did not have access because they never attempted or were not permitted to retrieve them. In an attempt to reduce such error, parents were told that accessibility was not related to permission. A further item asked the parent to report whether the child was allowed to "help themselves to snacks", which we specified was related to permission.

Quantity scores (number of servings) for vegetables were of lower validity compared to food variety. Serving sizes were based on the original package size, rather than the amount that remained in a package, except for foods that are sold as loose items (e.g. fresh vegetables). We reported package size by category (e.g. small, medium or large). Consequently, this variable was more subject to interpretation. Additionally, vegetables are perishable and an estimated bag size will change with consumption. Thus, similar to reliability, assessment of some food items may be measuring actual change rather than validity of the item scores.

An area for future work is the development of items associated with the physical activity environment, including both social and physical factors. Compared to the diet area, less is known about the role of the home environment on physical activity behavior. Future work with this instrument should identify other environmental characteristics that could be related to this behavior and to BMI levels.

Assessment of the family home environment has been popular for many years in child development research. The HOME (Home Observations for Measurement of the Environment) instrument was developed over three decades ago to measure the quantity and quality of stimulation and support available to children within their homes to facilitate optimal development. Multiple versions of this questionnaire for specific age ranges now exist, and it is used both as a measurement tool and as a means to evaluate the intervention effectiveness 42. Recent data collected using the HOME found positive associations between items within HOME that relate to opportunity for productive activity (e.g., availability of books and games) and weight gain in children43. More specifically, those children who gained weight had fewer opportunities for productive activity and watched more television compared to those who did not gain weight. This study also found that children who gained weight were more likely to have parents with more controlling parenting styles. While such relationships with weight status have been observed by using the HOME tool, that instrument does not specifically target factors that influence healthy weight behaviors. More recently, Gattshall et al 44developed and examined a survey to assess home environments in overweight children. Results demonstrated generally good reliability (physical activity items ICC = 0.43–0.96; food/diet items 0.01–0.90), but it is not possible to compare validity scores to the HHS, since Gattshall did not assess criterion validity.

Other researchers have focused more specifically on assessing the home food environment as a proxy for diet intake. This is a growing area of research, but there remains to be a lack of published manuscripts detailing the evaluation of such measures 162145, and few measures exist with adequate evidence for reliability and validity of scores. Recently, Campbell et al., 22 measured predictors of childhood diet within family homes using an adolescent- and parental-report questionnaire that included assessments of food availability, child feeding practices, parental modeling and television exposure. This questionnaire was completed by a large sample and demonstrated good internal reliability; however, test re-test reliability and criterion validity were not assessed.

Measures of the home physical activity environment are even less common than those of food. Hume et al 46 recently developed and assessed an adolescent self-report questionnaire to measure perception of home and neighborhood physical activity environments. Items were both physical (e.g., presence and size of yard) and social (e.g., encouragement to be active). Test re-test reliability (9-day interval) was examined in a small sample of children and most items were found to have moderate to good agreement (physical environment ICC = 0.8–0.94; social environment 0.16–1.00). This tool may be a useful measure of youth perceptions of the physical activity environment and as a predictor of healthy weight behaviors in the home; however, the validity of the assessments is not known.

Measurement of the home media environment has received attention due to interest in the relationship between media and sedentary behaviors4748. Many questionnaires have measured the duration of watching TV, and some have reasonable reliability and validity evidence. However, few of these also considered and evaluated social environments related to screen time. Salmon et. al. 48 examined the sedentary environment, including the physical media environment as well as social factors like rules and restrictions, surrounding media behaviors from both the parents' and the children's perspectives. Scores from these instruments showed moderate to good test re-test reliability, but again, validity was not assessed.

Compared to earlier questionnaires, the Healthy Home Survey takes a more holistic approach, examining a variety of physical and social factors that might influence diet, physical activity, and sedentary behaviors in children. The instrument includes many items that have not been measured or evaluated in previous studies. For some of these new items, an open response approach was needed. This approach will enable us to further develop the tool with evidence based identification of valid and discriminate items. One downside is that this approach required a lot of time for coding and analysis; however we believe that, at this stage, open response within categories added to the richness of the data. Many