The effect of the food environment on fresh produce served in family child care homes

Abstract

Background:

Higher produce consumption in childhood decreases risks of short- and long-term malnutrition, obesity, and disease. Children in early care programs, including family child care homes (FCCHs), receive 50–67% of daily nutrition while in care. Procuring nutritious foods requires grocer access, which is absent in food deserts (FDs).

Aim:

To determine if FCCH food environment (FE) impacted distance to grocers and amount of fresh produce served.

Methods:

Using a cross-sectional design, Modified Retail Food Environment Index scores determined census tract FD status. FCCH and grocer addresses were geocoded and distance to the nearest grocers was calculated. Fresh produce was observed during two lunches.

Results:

FE did not influence distance to grocers or fresh produce served. Non-desert FCCHs tended to serve fresh produce more frequently. The amount of fresh produce served was overall low.

Conclusion:

Further studies are warranted to inform policies aimed to reduce provider barriers regarding service of fresh produce.

Introduction

Higher produce consumption, specifically fruits and vegetables, in childhood is associated with decreased risk of chronic disease (Hodder et al., 2018). Eating habits are established in early childhood and are influenced by several factors including early care and education programs (ECEs) (Johnson SL, 2016). Three in five United States preschool-aged children are enrolled in ECEs (Redford et al., 2017), including small, licensed programs operated in a provider’s home (family child care homes (FCCHs)), and receive one-half to two-thirds of meals in care (Oklahoma Department of Human Services, 2018).

The United States Child and Adult Care Food Program (CACFP) reimburses ECEs for qualifying foods, especially ECEs serving low-income children and areas (United States Department of Agriculture Food and Nutrition Service, 2019). ECEs must follow CACFP guidelines that encourage provision of nutritious foods such as produce (United States Department of Agriculture Food and Nutrition Service, 2019). Despite guidelines and financial assistance, children in FCCHs are not meeting recommendations (Tsai et al., 2017).

The food environment (FE) considers grocer availability and proximity, food cost, and scope of nutrition assistance programs (United States National Research Council, 2009); and can be a systemic barrier for provision of healthful foods (Walker et al., 2010). Food deserts (FDs) are FEs in which residents have limited access to grocers with affordable healthful food options, such as fresh produce, and are typically found in low-income neighborhoods (United States National Research Council, 2009; Walker et al., 2010). FD residence has been linked with poorer diet and health outcomes (Walker et al., 2010).

The FCCHs within FDs may be equally susceptible to these outcomes since these small, home-based programs also depend on nearby grocers (National Farm to School Network, 2016). Neighborhood studies demonstrate limited grocer access and decreased availability of fresh produce in FDs (Walker et al., 2010). It can be inferred from these studies that FCCH residence within FDs may affect children’s meals. The aim of this study was to determine if FCCH FD status impacts proximity to grocers and fresh produce served to children in care. We hypothesized that FCCHs in FDs would drive further to reach grocers and serve less fresh produce.

Materials and methods

Participants and study design

This cross-sectional spatial epidemiology study involved 49 FCCHs within 60 miles of a medium-sized metropolitan area in the central United States. FCCHs were eligible if they participated in the CACFP and actively cared for at least one 2-to-5 year old child. FCCHs were geocoded and the FE (desert/non-desert) was determined. Service of fresh produce was observed during two unannounced lunch visits conducted at least 48 hours apart but within two weeks (from October 2017–2018). The University of Oklahoma Health Sciences Center Institutional Review Board approved this study. Participants consented prior to participation.

Experimental protocol

Spatial analysis.

State businesses, located in the United States Business Data Directory (United States Business Data, 2018), were classified as grocers based on Standard Industrial Classification codes. FCCH and grocer addresses and Modified Retail Food Environment Index (mRFEI) scores of census tracts were layered onto a map using ArcGIS Desktop, 10.6.1 (Environmental Systems Research Institute, 2018).

Operationalization of FE.

The FD status of FCCH census tracts were classified using the mRFEI measure (Centers for Disease Control and Prevention, 2018). Scores, ranging from 0–100, were calculated by dividing the number of healthy food outlets by the total number of healthy and unhealthy food outlets. Healthy and unhealthy food outlets within half a mile from the tract boundary were also included (Centers for Disease Control and Prevention, 2018). Census tracts with a numerator or denominator of zero were considered FDs (Centers for Disease Control and Prevention, 2018). Census tracts with scores greater than zero were considered non-deserts with access to at least one healthy food outlet (Centers for Disease Control and Prevention, 2018).

Proximity to the nearest grocer.

The distance (in miles) from the FCCH to the nearest grocer was examined by self-report and ArcMap Closest Facility tool, which measured network distance from FCCHs to their nearest grocer using roadways with traffic conditions on Wednesday, February 27, 2019 at 5:00:00 PM.

Fresh produce served.

The Dietary Observations in Child Care (DOCC) tool assessed foods served at lunchtime for up to three children on two non-consecutive days (Ball et al., 2007; Sisson et al., 2017). Seven trained researchers observed meals with high inter-rater reliability (intraclass correlation coefficient (ICC) = 0.946–0.967). Unique identifiers were not assigned to each child; it is possible that the same three children were observed on both days, or three different children, or some combination. Food observations were averaged across observed children and both days to create FCCH-level nutrition variables. Since serving sizes vary between different foods, items were counted if offered to the child regardless of the volume (Sisson et al., 2017). Children’s food acceptance increases with exposure (Johnson SL, 2016). Therefore, service of multiple fresh produce items received higher sum values.

Statistical analysis

Descriptive sample characteristics (mean α standard deviation, range, and frequency) were calculated. Independent sample t-tests determined differences in fresh produce items and self-reported and geographic information system (GIS)-determined driving distance to the nearest grocer by FE. A paired t-test determined the difference in self-reported and GIS-determined driving distance to the nearest grocer. Data were analyzed using SPSS Version 22.0.

Results

All providers were women with a mean age of 45.2 ± 12.6 years, ranging from 23 to 71 years. Fourteen FCCHs (28.6%) resided in FDs and 35 (71.4%) in non-deserts. The mRFEI scores ranged from zero to 50 with a mean score of 8.3 ± 9.3. Figure 1 shows FD status of census tracts of both the state and sample.

Figure 1.

Food environment (desert and non-desert) status of census tracts in Oklahoma.

Driving distances (GIS or self-report) were not significantly different between FCCHs in FDs and non-deserts (Table 1). Self-reported driving distance was significantly higher than GIS-determined driving distance (self-reported 9.7 ± 9.8 vs. GIS-determined 6.6 ± 9.4; t (47) = −5.119, p ≤ 0.001). Fresh produce did not differ by FD status (Table 1). Table 2 shows the frequencies of days fresh produce was offered.

Table 1.

Driving distance to the nearest grocer and amount of fresh produce offered in family child care homes by food environment (desert, non-desert); n = 49.

		Mean ± standard deviation	Range	p-value
Geographic information system-determined distance	Desert	6.9 ± 8.2	1.2–26.1	0.847
	Non-desert	6.3 ± 9.9	0.8–48
Self-reported distance*	Desert	10.4 ± 10.2	1–35	0.775
	Non-desert	9.5 ± 9.8	1–50
Amount of fresh produce items offered	Desert	1.1 ± 1.1	0–4	0.785
	Non-desert	1.2 ± 0.9	0–3.5

Note:

^*self-reported distance was collected by survey responses. There was one missing value (n = 48).

Table 2.

Frequencies of days that fresh produce was offered by food environment; n = 49.

Family child care home food environment	0 days	1 day	2 days
Desert (n = 14)	4 (28.6%)	4 (28.6%)	6 (42.9%)
Non-desert (n = 35)	4 (11.4%)	13 (37.1%)	18 (51.4%)
Total (n = 49)	8 (16.3%)	17 (34.7%)	24 (49.0%)

Discussion

This was the first study to examine the relationship between FCCH FD status and provider mealtime practices. Our hypothesis was informed by neighborhood environment studies showing that households with limited grocer access have decreased availability of fresh produce and poorer meal quality (Walker et al., 2010). There was no difference in grocer driving distance or fresh produce service by FD status. However, FCCHs in non-deserts tended to serve fresh produce more frequently compared to those in FDs.

Previous neighborhood studies report that grocer driving distances are longer in FDs (Walker et al., 2010). These studies often describe neighborhood FEs by calculating healthy food outlet density within a specific radius, using buffers of various shapes and distances around individual homes (Charreire et al., 2010). Another approach is to consider healthy food outlet density within census tracts, which can be beneficial for decentralized cities with great sprawl. For instance, in this study, the average driving distance was 6.9 and 6.3 miles in FDs and non-deserts, respectively, from FCCHs to their nearest grocer. It was previously assumed that residents choose grocers closest to where they live; however, recent evidence indicates that residents often choose to shop at further grocers for primary food shopping due to cost or product selection (Aggarwal et al., 2014), which is congruent with our findings showing that self-reported distance was significantly longer than GIS-determined driving distance by a mean of 3.2 miles.

Similar to other studies, no difference in fresh produce provision was observed across FD status (Caspi et al., 2012). CACFP participation and requirements may protect FCCHs in FDs. Clearly, individual households do not have federal requirements on foods served. Regardless of FE, fresh produce served was low; on average, providers served 1.1 and 1.2 fresh produce items in FDs and non-deserts, respectively. While not significant, results indicated that FCCHs in non-deserts tended to offer fresh produce more frequently than FCCHs in FDs (Table 2).

This evidence points to a broader issue—FCCH providers encounter barriers to offering fresh produce beyond grocer access and FE. Barriers may include low knowledge about purchasing and preparation, self-efficacy, cost, and access; along with provider and child preferences (Institute of Medicine, 2012). A strength of this study was that service, not consumption, of fresh produce was measured. This allocated “credit” to providers who at least offered fresh produce despite fears of wastage. Again, CACFP participation may have acted as a buffer since providers are mandated to serve produce for reimbursement. Barriers would have been further validated by inclusion of qualitative interviews with providers, which have been summarized elsewhere (Sisson et al., in press).

Diet observation was another strength due to its lower participant burden and elimination of social desirability bias seen with survey and food frequency questionnaires. Limitations of the DOCC methodology include variability between observers and the limited number of observed children (Institute of Medicine, 2012). However, researchers completed rigorous training in both the laboratory and child care setting (Ball et al., 2007) with high inter-rater reliability (ICC = 0.946–0.967). Observation of three children across two days has been considered adequate to reflect typical child care mealtime practices within ECEs (Ball et al., 2007; Sisson et al., 2017).

Our study included only fresh produce based on the hypothesis that would better predict grocer access, which contain more fresh produce (United States National Research Council, 2009). Some convenience stores may offer fresh produce, often limited in variety and higher priced. Based on neighborhood study findings, FCCH providers are likely to shop at grocers for primary shopping due to lower cost and higher availability of foods (Aggarwal et al., 2014). Regardless, mRFEI classification of convenience stores as “unhealthy” may have led to an over-representation of FDs in our sample (Thornton et al., 2020), since ground truthing was not feasible in this study to assess food outlet inventory.

An mRFEI measure strength was that it was exhaustive. It identified healthy food outlets as supermarkets, larger grocery stores, supercenters, and produce stores; and unhealthy food outlets as fast food restaurants, small grocery stores, and convenience stores within half a mile of the census tract boundaries (Centers for Disease Control and Prevention, 2018). This considered that providers may drive further to nearby non-deserts to reach preferred grocers.

The mRFEI measure did have limitations. Use of the mRFEI measure to quantify food outlets within census tracts may not adequately reflect the nuanced nature of FEs (Thornton et al., 2020). The dichotomy of “healthy” and “unhealthy” may have excluded food outlets that did not quite meet the classification criteria (Thornton et al., 2020), likely under- or overestimating the number and type of food outlets in an area. Also, it does not indicate the degree of “healthy” or “unhealthy” (Thornton et al., 2020). For example, large supermarkets and small produce markets would have equal weight in mRFEI scoring but would dictate different shopping habits. Lastly, the mRFEI methodology became a reflection of food outlet density. For instance, a census tract with one healthy and two unhealthy food outlets received the same mRFEI score as a tract with four healthy and eight unhealthy food outlets (Thornton et al., 2020). Our approach included the division between FDs (mRFEI = 0) and non-deserts (mRFEI >0). These limitations were therefore reduced since we considered whether FCCHs had access to either zero or at least one healthy food outlet; we did not examine the “quality” of non-deserts.

Conclusion

Preschool-aged children are in a growth period when nutrient requirements are maximized. Behaviors established in childhood persist into adulthood (Johnson SL, 2016; Hodder et al., 2018). Over 60% of United States children under five spend their early years in FCCH-based ECEs (Redford et al., 2017). Decreased grocer access, common in FDs, is associated with reduced produce availability (Walker et al., 2010). FCCH residence in a FD may impact fresh produce served to children.

The CACFP participation may encourage providers to serve produce regardless of FE. Further research is needed to understand provider barriers regarding service of fresh produce. Potential interventions could include further investigation of provider perceived barriers, education/training, updating food program guidelines; and offering incentives to establish farmers markets, produce markets, and/or grocers within low-access areas.

Availability of data and materials

Data and materials are not publicly available due to upholding confidentiality of the participants but are available from the corresponding author on reasonable request.