Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 15
2
Alignment of Young Children’s Dietary
Intake with Current Dietary Guidance
T
here are very few data on the quality of meals and snacks served to
children in the Child and Adult Care Food Program (CACFP) and
the contributions of those meals and snacks to children’s overall di-
etary intakes. Thus, a major goal of future research identified in the CACFP
report is to assess the food and nutrient content of meals and snacks served
to and consumed by children and the impact of these meals and snacks on
children’s overall diets. This chapter summarizes presentations and discus-
sion on existing methodologies that might be useful in implementing a study
that would address these and related issues.
Mary Kay Fox began the presentation by describing potentially relevant
methodologies used in previous national studies of child nutrition. These
included the School Nutrition Dietary Assessment (SNDA) studies, the
Feeding Infants and Toddlers Study (FITS) studies, and two previous studies
of CACFP. She explored ways that the methods used in these studies could
provide “some starting ground” for thinking about optimal approaches to
collecting data in a nationally representative study of child day care, includ-
ing CACFP. She also pointed out how the child day care setting presents
unique challenges that need to be addressed.
In addition to national surveys, smaller studies in the published litera-
ture serve as another source of information for potentially relevant meth-
odologies. Dianne Ward presented an overview of the scientific literature
on methods for assessing foods served in child care settings. She argued
that, regardless of the method(s) chosen, researchers should consider using
multiple methods to ensure accuracy. She also encouraged testing protocols
before widespread implementation.
15
OCR for page 16
16 RESEARCH METHODS TO ASSESS DIETARY INTAKE
While Ward focused on foods served, Sara Benjamin Neelon focused
on foods consumed. She explored the various methods that have been used
in the past, as well as some potentially new methods, to assess food and
nutrient intake both in and out of child care. Based on comments made at
various times throughout the workshop, other speakers and participants
seemed to generally agree with Benjamin Neelon’s assessment that the
preferred method for collecting data to assess nutrient intake is direct ob-
servation in the child care setting, coupled with 24-hour dietary recalls1 of
intake both inside and outside the care facility.
Finally, Beth Dixon considered the different types of dietary data that
can be collected in child care settings and elaborated on the trade-off be-
tween project scope and data detail (i.e., with fixed funding, the larger the
scope of a study, the less detailed the dietary data collection). The more
detailed the data, the greater the opportunity for accurate comparisons with
recommended dietary intakes or meal pattern recommendations. While di-
rect observation provides the greatest level of detail, it is an expensive data
collection method, especially because of the labor involved in collecting the
data on site by trained observers.
A recurrent theme over the course of the 1-day workshop was the need
to be very clear about the desired outcome(s) of a study before developing
the methodology—addressing the “what” before the “how.” This is because
the “best” method depends on the desired outcome(s). Different methods
yield different types of information. This theme was especially prominent
in the dialogue summarized in this chapter. For example, the preferred
methodology for assessing what children are being served is not necessar-
ily the same as the preferred methodology for assessing what children are
actually consuming. Nor are the best methods in one setting necessarily the
best methods for another setting. All of the speakers featured in this chap-
ter emphasized the importance of formulating the research question(s) and
desired outcome(s) before deciding on which methods, tools, or research
design to use.
ADAPTING METHODOLOGY FROM PREVIOUS
NATIONAL STUDIES TO ASSESS THE CHILD
AND ADULT CARE FOOD PROGRAM2
A useful starting point for considering how to move forward with a
nationally representative study of child day care, including CACFP, is to
1 The 24-hour recall method involves collecting data on everything consumed by the partici-
pant over the previous 24 hours.
2 This section summarizes the presentation Adapting Methodology from SNDA and FITS
Studies to CACFP by Mary Kay Fox of Mathematica Policy Research.
OCR for page 17
17
ALIGNMENT OF YOUNG CHILDREN’S DIETARY INTAKE
examine methodologies used in similar large national studies. Mary Kay
Fox identified two series of studies, conducted by Mathematica Policy Re-
search, Inc. for the U.S. Department of Agriculture (USDA), which might
be especially helpful: (1) the SNDA studies and (2) the FITS studies. USDA
has been conducting the SNDA studies since the early 1990s to assess the
food and nutrient content of meals offered and served to students in schools
as well as the contribution of these meals to students’ total diets. The 2002
and 2008 FITS studies assessed the usual dietary intakes of infants and tod-
dlers and included special procedures for collecting and processing dietary
intake data for these age groups. In addition, the SNDA and FITS studies
included national samples and provided data that informed previous Insti-
tute of Medicine (IOM) Food and Nutrition Board committee work. While
components of the SNDA and FITS studies data collection methodologies
may be useful in designing the approach for a study of child care, includ-
ing the CACFP, Fox emphasized that the methodologies would need to be
tailored to the CACFP child care setting(s). She indicated that she would
also discuss data collection approaches used in two previous studies of the
CACFP, which were modeled on the approaches used in the SNDA studies.
Relevance of the SNDA Studies
USDA has relied on the SNDA studies since the early 1990s to monitor
the quality of school meals and contributions of school meals to children’s
overall dietary intakes. Ideally, a national study of child care would pro-
vide “SNDA-like” data for the CACFP. The two most recent rounds of the
SNDA studies, SNDA-III and SNDA-IV,3 were conducted in school years
2004–2005 (Gordon and Fox, 2007) and 2009–2010, respectively. Both
studies included large national samples of school districts and schools—
SNDA-III included 129 school districts and 398 schools in 36 states, and
SNDA-IV included 578 school districts and 884 schools in 48 states—and
collected data on meals offered and served in schools. SNDA-III also col-
lected detailed information on students’ dietary intakes both in school and
outside of school (2,314 students in 287 schools).
Relevance of the FITS Studies
The FITS studies, which were sponsored by the Nestlé Nutrition In-
stitute in 2008 and Gerber Products Company in 2002, included a com-
prehensive assessment of food and nutrient intakes of infants, toddlers,
and preschoolers. FITS 2002 included 3,022 infants and toddlers 4 to 24
months of age. FITS 2008 had a slightly larger sample (3,273) and included
3 Data will be published in 2012.
OCR for page 18
18 RESEARCH METHODS TO ASSESS DIETARY INTAKE
infants, toddlers, and preschoolers from birth to 48 months of age. FITS
researchers developed special procedures for collecting and processing di-
etary intake data for infants and young toddlers. Given that CACFP feeds
infants and toddlers and the unique challenges of collecting dietary intake
data on those age groups, Fox suggested that some of the procedures de-
veloped by FITS researchers may be useful in a national study of child care,
including CACFP.
Previous Food and Nutrition Service (FNS)-Sponsored Studies of CACFP
Dating back to 1979, USDA has conducted at least three studies of
meal quality in CACFP. Fox suggested that the data collection methodolo-
gies used in the two most recent studies—the Early Childhood and Child-
care Study (Fox et al., 1997) and the Family Child Care Homes Legislative
Changes Study (Crepinsek et al., 2002)—should also be reviewed in plan-
ning a national study. The Early Childhood and Childcare Study collected
data on the foods included in meals and snacks offered by 1,962 CACFP
providers, including both child care centers and family child care homes. To
obtain information about the amounts of food served, the study included
observations of 1,347 children. The CACFP Legislative Changes Study col-
lected data on a smaller sample of Tier 24 CACFP family child care homes
to see if changes in the way homes were categorized for reimbursement pur-
poses (i.e., dividing homes into Tier 1 and Tier 2 categories) had any impact
on the quality of meals provided. The study adapted and used most of the
data collection methods used in the Early Childhood and Childcare Study.
Modifying Existing Methodologies for Use in a National Study of CACFP
When considering how existing methodologies might be useful for a
nationally representative study of CACFP, Fox suggested that a good first
step is to identify key research question(s). With the desired outcome(s)
in mind, one can then evaluate whether and how the methodologies can
be used or adjusted to accommodate the unique challenges that must be
addressed when collecting data in child care settings. She identified two
potential overarching research questions for the study. First, how do the
meals and snacks offered in child day care centers and homes, including
4 Family child care homes that participate in the CACFP are categorized as either Tier 1 or
Tier 2 for reimbursement purposes. Tier I homes are those that are located in low-income
areas, or those in which the provider’s household income is at or below 185 percent of the
federal income poverty guidelines. Tier II homes are those that do not meet the location or pro-
vider income criteria for a Tier I home. However, Tier II providers may elect to have their spon-
soring organizations identify income-eligible children, so that meals served to those children
who qualify for free and reduced-price meals could be reimbursed at the higher Tier I rates.
OCR for page 19
19
ALIGNMENT OF YOUNG CHILDREN’S DIETARY INTAKE
those participating in CACFP, align with required meal patterns and the
underlying requirements for calories and nutrients? Second, how do the
meals and snacks consumed in these settings contribute to children’s overall
dietary intakes, and how do children’s overall dietary intakes align with
current dietary recommendations?
To answer the first question, Fox pointed to the menu survey used in
SNDA-IV as a potentially useful tool. She described the SNDA-IV menu
survey as a sophisticated and well-organized data collection tool that has
been continually tested, refined, and improved since the early 1990s. The
original survey was a blank form on which participants simply recorded
what they served at lunch and breakfast for a full school week. Over time,
SNDA researchers found ways to decrease the burden on respondents while
increasing the completeness and quality of the data they provided. The
menu survey used in SNDA-IV is now precoded with foods and beverages
commonly offered in school meals and includes check boxes and prompts
to elicit descriptive information about the foods needed for nutrient analysis
(for example, the fat content of milk and whether canned fruit was packed
in heavy syrup, light syrup, juice, or water). Menu survey respondents re-
ceive an instruction booklet that includes simple but detailed instructions
about how to complete each form. In addition, respondents receive in-depth
training (by telephone) before they start completing the menu survey, as
well as assistance throughout the course of data collection. There is also a
detailed editing process that occurs after the surveys are completed (e.g., if
there is no mention of condiments, does that mean that no condiments were
served or that the respondent forgot to record the condiments?). Finally,
respondents in SNDA-IV were offered a $40 to $50 incentive to complete
the menu survey.
According to Fox, the menu survey methodology used in SNDA-IV
could be used as a starting point in developing a menu survey for a study
of CACFP, but the materials would need to be modified to accommodate
important differences between schools and CACFP settings. Although the
meals offered in CACFP settings are generally simpler than meals offered
in schools (i.e., they tend to be “set” menus with few choices), CACFP
providers typically have less technical food service skills than school food
service operators. Additionally, they serve a broad range of age groups,
from infants through school-age children, and may serve different foods
and beverages to children of different ages. Finally, obtaining information
about portion sizes is a particular challenge in CACFP settings. CACFP pro-
viders do not typically serve standardized portions the way schools do (e.g.,
No. 8- or 4-ounce scoops). Many child care providers use family-style meal
service at tables where one provider sits with multiple children; providers
do not formally track how much food any individual child is receiving.
The Early Childhood and Child Care Study used a menu survey that
OCR for page 20
20 RESEARCH METHODS TO ASSESS DIETARY INTAKE
was adapted from the tools used in the early SNDA studies. It included
fewer forms and simpler step-by-step instructions. Fox suggested that this
tool could be improved by incorporating precoded lists of foods and check
boxes and prompts like those used in the SNDA-IV menu survey. These
changes would make it easier for respondents to complete the survey and
make it less likely that they would leave out valuable descriptive informa-
tion about the foods offered.
With respect to the second key research question, which pertains to
children’s dietary intakes and contribution of CACFP meals and snacks,
useful insights about potential data collection methodologies can be ob-
tained from SNDA, FITS, and previous CACFP studies. SNDA-III and both
rounds of FITS studies (2002 and 2008) collected data on dietary intakes
using 24-hour recalls. Each study included a second 24-hour recall on a
subset of the study population in order to estimate usual nutrient intake
distributions (IOM, 2000). The FITS studies also collected data on vitamin
and mineral supplements, which Fox noted would be important to include
in a national study of CACFP if the goal is to assess total nutrient intake.
She also emphasized the importance of incentives in gaining cooperation
from parents and caregivers. Both SNDA-III and the FITS studies offered
incentives, SNDA-III provided $5 to $10 for each recall, and the FITS stud-
ies provided $20 for the first recall and $10 for the second.
Fox identified two unique challenges that need to be considered in
thinking about adapting SNDA-III and FITS methodologies to a study of
child day care. The first is how to collect data on the foods consumed while
the child is in care. In SNDA-III, data collectors interviewed children about
their in-school food consumption shortly after meal times. But the CACFP
serves younger children who cannot provide reliable information about
their food intakes. In the FITS studies, data on what a child consumed while
in care were communicated by the provider either to the parent or to study
data collection staff. Theoretically, this approach could work in a child care
setting. However, Fox questioned whether it would be realistic to expect a
provider to report on multiple children.
The second challenge identified by Fox was collecting data on out-
of-care intakes for infants and toddlers. Portion sizes for infants and tod-
dlers are a unique challenge because of spillage, under- or overestimating
amounts, and other factors. For example, portion sizes were overreported
in the 2002 FITS, primarily as a result of overreporting of beverage portion
sizes. Before the 2008 FITS was initiated, great effort was spent on investi-
gating and revising visual aids for estimating portion sizes (i.e., researchers
developed visual aids that included age-appropriate measuring cups and
bowls; see Figure 2-1). The data collection protocol also needs to include
procedures for probing respondents about spillage and waste when estimat-
ing the amounts that children actually consumed. Fox also commented that
OCR for page 21
21
ALIGNMENT OF YOUNG CHILDREN’S DIETARY INTAKE
FIGURE 2-1 Age-appropriate measuring cups used to measure portion size in the
2008 FITS.
SOURCE: Nestlé, 2011.
Figure 2-1
Bitmapped
because of the number of different age groups served in the CACFP, total
sample sizes for the dietary intake component of the study may need to be
quite large if the goal is to assess usual nutrient intakes for each age group.
HOW TO ASSESS FOOD SERVED IN CHILD CARE SETTINGS5
The published literature describes several methodologies for assessing
foods served in child care or similar settings: plate waste, direct observa-
tion, provider self-report, child care menu analysis, and food purchase
receipts. Dianne Ward provided an overview of each methodology and
discussed the strengths and limitations of each. To ensure accurate data
collection, Ward urged the use of multiple methods (e.g., validating menu
analysis with actual observation data; validating self-report with direct
observation data; and using a combination of self-report, menu, and direct
observation data to get a full picture of types and amounts of foods and
beverages served). Regardless of the method(s), she urged that protocols be
tested and results verified before large-scale implementation.
5 This section summarizes the presentation of Dianne Ward from the University of North
Carolina.
OCR for page 22
22 RESEARCH METHODS TO ASSESS DIETARY INTAKE
Plate Waste Studies
Plate waste studies involve weighing portions provided at a meal or
snack prior to service; weighing waste after eating; and then, using a
database and software program, translating specific foods and amounts
consumed into macro- and micronutrients (e.g., Buzby and Guthrie, 2002;
Graves and Shannon, 1983). Plate waste studies have two major strengths,
according to Ward. First, they provide a precise estimate of food served and
consumed. Second, foods served and consumed can easily be translated into
nutrients using the appropriate database or software program. They are
limited by the fact that they work better with plated meals (if not plated,
the researcher must take self-served meals from the children, then weigh
and return them); they require handling foods, which may not be permitted
in some settings; and they are very labor intensive and costly.
Diet (Direct) Observation
Diet, or direct, observation is a visual estimation of the types and
amounts of foods and beverages served. There is considerable variability
in methodology among published studies with respect to specific proto-
col, amount of training required, and precision obtained. Most published
observational studies are based on relatively small numbers of centers or
homes, with the number of days of intakes observed varying from half a
day to multiple days (e.g., half-days at 40 centers, Erinosho et al., 2011; 2
consecutive days at 20 centers, Ball et al., 2008; 3 consecutive days at 12
centers, Briley et al., 1999; 14 days, 7 in the fall and 7 in the spring, at two
centers, Bruening et al. 1999).
Ball et al. (2007) developed the first published protocol for using direct
observation to assess dietary intake among young children in child care.
The goal was to train five observers to visually estimate at a predetermined
level of precision the amount of food types being served. The training re-
quired a total of 56 hours over the course of a month. During the training,
after practicing with measuring cups and spoons, the observers were tested
in a laboratory setting with 20 common child care foods (e.g., applesauce,
animal crackers, spaghetti). The researchers found good agreement between
the observer food-quantity estimates and the 20 measured portions of com-
mon child care foods (with a mean intraclass correlation coefficient value,
or ICC, of 0.99). However, there was considerable variability across foods,
with portion sizes of spaghetti being more difficult to estimate than other
foods. At the end of the training, the observers were tested again, but in
a child care center setting. With foods that were not easy to discern with
visual observation, observers were instructed to ask the classroom staff or
cook about preparation of the food in question. The child care setting cer-
OCR for page 23
23
ALIGNMENT OF YOUNG CHILDREN’S DIETARY INTAKE
tification demonstrated good agreement between the observer food-quantity
estimates and the 56 foods and beverages tested (ICC = 0.88).
The strengths of direct observation are that it is less intrusive than food
weighing and provides a replicable dietary observation approach that is po-
tentially more usable by researchers. However, the methodology is limited
by its requirement for extensive training to ensure interrater reliability, the
small number of children that can be observed by one observer, and the
need for observers to clarify with food service staff portions for foods that
are not easy to decipher by vision.
Provider Self-Report
Provider self-report involves asking providers (e.g., directors, teachers,
food staff) what foods were served to children and, sometimes, the amount
of food served. As part of an effort to develop an effective self-report
instrument for assessing comprehensive nutrition and physical activity en-
vironments in child care, Ward discussed evaluating the reliability and va-
lidity of self-report data collection in child care centers. In 2008 Ward and
colleagues used an observational instrument known as the Environmental
Policy and Assessment Observation tool, which had been developed for the
Nutrition and Physical Activity Self Assessment for Child Care program.
She described the methodology as being similar to 24-hour recall except
that the assessment was conducted in real time (more like a food diary),
with both trained observers and care providers recording food intake (Ward
et al., 2008). More recently, trained observers visited 50 centers every day
for 4 days and assessed the nutrition environment, including foods served
(not portion size, except for juice) in target classrooms. Teachers of the
target classrooms (i.e., the “providers”) were asked to report on foods and
beverages served to the children on each of the same 4 days. The research-
ers assessed both reliability (i.e., repeatability of the providers’ reported
food lists from one day to the next) and validity of the provider reports
(i.e., how closely the providers’ food lists matched the trained observers’
food lists). The researchers concluded that providers can report what was
served, but multiple days of reporting might yield more accurate reporting
of foods served.
Provider self-report is typically used in combination with direct obser-
vation on a subsample of centers or homes. For example, both provider
self-report and direct observation were used in the two CACFP studies de-
scribed by Mary Kay Fox, the Early Childhood and Child Care study (Fox
et al., 1997) and the Effects of Lower Meal Reimbursement study (Crepin-
sek et al., 2002) (see the previous section summarizing Fox’s presentation).
In the 1997 study, which aimed to describe the food and nutrient content
of meals and snacks offered by CACFP providers and consumed by par-
OCR for page 24
24 RESEARCH METHODS TO ASSESS DIETARY INTAKE
ticipating children, the researchers used a provider self-report menu survey
of all foods and beverages in all meals and snacks served over a specified
5-day period. Because pretesting showed that most providers were unable
to report sufficient detail on portion size, the researchers did not collect
portion size data. In addition to provider self-report of the types of foods
and beverages offered, trained field staff conducted meal observation in a
subset of child care facilities on all meals and snacks consumed by children
while in child care on 2 nonconsecutive days. Each observer was responsible
for collecting data on six or fewer children. Prior to the meal or snack pro-
vision, the observers weighed or measured five reference portions of each
food that was going to be offered. During meal time, observers used visual
methods to estimate the amount of food received and the amount of food
left over by each child. The observation data were used to calculate aver-
age portion sizes, which were then used to estimate the nutrient content of
the foods offered.
The strengths of provider self-report include its lower cost compared
to direct observation and other methodologies; the potential to be imple-
mented in large numbers of centers and homes; and that it can be supple-
mented with observation. However, the methodology is limited by a risk
of misreporting, possible inaccuracies, and the multiple days of reporting
required.
Child Care Menus
Child care menu methodology involves assessing detailed lists of foods
and beverages served to children for meals and snacks. Such lists are typi-
cally offered as a service to parents and, in some states, as a mandate to
ensure compliance with program requirements. Ward opined that child
care menus could be used to monitor CACFP guidelines. However, foods
actually served do not always match planned menus. Fleischhacker et al.
(2006) observed and recorded foods served throughout the school day and
then compared the food records against monthly menus for 77 days in one
Head Start center and found very poor agreement. Of 269 meals and snacks
observed, only three breakfast meals and one “ethnic day” matched the
menu. Benjamin Neelon et al. (2010), on the other hand, compared 1 day
of direct observation with menus at 84 child care centers and found good
agreement, with 52 percent of 254 meals and snacks served matching the
menu and 87 percent of 710 individual items served matching the menu.
Other limitations of child care menus include their lack of detail regarding
specific types of foods and beverages provided (e.g., fruit juice versus 100
percent fruit juice), lack of specification about how foods are prepared
(e.g., no indication whether the chicken in a sandwich is baked or fried),
OCR for page 25
25
ALIGNMENT OF YOUNG CHILDREN’S DIETARY INTAKE
the absence of reported condiments used with foods, and difficulty in deci-
phering specific ingredients in some dishes (e.g., casseroles). A key strength
of child care menus is that they provide a quick and easy way of collecting
information about foods served to children.
Food Purchase Receipts
Food purchase receipts have been used to evaluate associations between
food cost and quality of food served. For example, Monsivais and Johnson
(2012) collected receipts and detailed menus from 60 home child care pro-
viders and found that greater cost was associated with a higher nutritional
quality of foods (based on servings of whole grains, fresh whole fruits, and
vegetables; energy density [kJ/g]; and mean nutrient adequacy for seven
nutrients of concern for child health). Menu details were obtained from
forms given to the providers by the research staff. Food purchase receipts
provide a quick, easy, and cheap way to collect data about types of foods
and beverages offered to children. The methodology places little to no
burden on participants, especially if receipts are collected over only a short
period of time. However, they are limited as a methodology by the impact
of missing or lost receipts; social desirability bias (i.e., during the period of
examination, the providers might change the foods they purchase); and that
they provide information only about foods purchased, not foods prepared,
served, or consumed.
DIETARY ASSESSMENT IN YOUNG CHILDREN: TOTAL
DAILY INTAKE OF FOOD AND NUTRIENTS6
Sara Benjamin Neelon described assessing dietary intake as a “daunt-
ing task,” especially in young children (i.e., children under 5 years of age).
She identified several methods for collecting dietary intake data: 24-hour
recall (i.e., retrospectively asking someone about everything that the person
ate or drank over the previous 24-hour period), food record or diary (i.e.,
prospectively asking participants to write down everything they consume
in a day), food frequency questionnaire (i.e., measuring usual intake over
a given amount of time), direct observation (i.e., visually observing what
individuals eat), indirect observation (i.e., electronically documenting what
individuals eat via videography or photography), and biomarkers (e.g., as-
sessing iron status through a blood test, selenium with a toenail clipping, or
6 This section summarizes the presentation of Sara Benjamin Neelon from Duke University.
It also describes the discussion that occurred at the end of the first session, when audience
members asked questions about one of the methods that Benjamin Neelon described (indirect
observation via videography or digital photography).
OCR for page 26
26 RESEARCH METHODS TO ASSESS DIETARY INTAKE
TABLE 2-1 Three Potential Methods for Assessing Dietary Intake of
Children in Child Care
Type of Assessment Assessment Method Respondents
Direct observation Researcher observation of N/A
child
Food record or diary Self-administered by adult Parent and child care provider
Indirect observation Photography or videography N/A
of child
SOURCE: Benjamin Neelon, 2012.
carotenoid intake as a proxy for fruit and vegetable intake via a palm scan).
She focused on three of these methods: direct observation, food record or
diary, and indirect observation (see Table 2-1).
According to Benjamin Neelon, direct observation is the most common
method for assessing dietary intake in child care and is likely to provide
the most specific information about foods and beverages consumed in child
care. Regardless of the method employed, Benjamin Neelon emphasized
the importance of consistency in the use of dietary assessment methodol-
ogy to allow for comparison across studies. For example, Ball et al. (2008)
assessed foods served to and consumed by children in 20 child care centers
and compared both amounts to CACFP and MyPyramid recommended
portion sizes for children. By using the same observation method in a recent
study of dietary intake of children in child care in Mexico City, Benjamin
Neelon et al. (unpublished) were able to compare the Ball dietary intake
data to that of the Mexican children.
Most researchers couple whichever method they use for in-care data
collection with a second method for foods consumed outside of child care.
Choices for at-home data collection include 24-hour recall by the parent,
food record by the parent, direct observation, and indirect observation (e.g.,
setting up a video camera in the area of the home where meals are typically
eaten). Direct observation in a home setting is usually not practical, and
indirect observation is limited by the fact that people often eat in multiple
areas of the home (e.g., while watching TV or while “grazing”), outside of
the range of the video camera. In Benjamin Neelon’s opinion, 24-hour recall
is likely to provide the most specific information about foods and beverages
consumed outside of care.
Direct Observation in Child Care
With direct observation, trained observers are present in the classroom
or family child care home for all meals and snacks. Observers document in
OCR for page 27
27
ALIGNMENT OF YOUNG CHILDREN’S DIETARY INTAKE
detail, often using prepopulated forms, all foods and beverages, including
condiments, provided to and consumed by children. Data for both foods
and beverages are usually reduced to food groups (e.g., milk) and nutrients
(e.g., grams of fat). The preferred ratio is one observer to three children.
Benjamin Neelon agreed with Dianne Ward and other workshop par-
ticipants who emphasized the importance of using well-trained observers,
given that effective direct observation requires practice and skill. She also
emphasized the importance of collecting data over multiple days, typically
2 or 3.
Limitations of direct observation data collection include the intensive
training required of observers, the large amount of time required of observ-
ers (both training and observation time), the need for target children to sit
near each other (i.e., if an observer is watching multiple children), the small
number of children that can be observed (i.e., three per observer in one
classroom/area), and the need for some onsite interaction to record specific
information about foods and beverages served (e.g., preparation method).
Benefits of direct observation in child care include its appropriateness for
documenting intake in children of all ages, including infants; its potential
to be used in both child care centers and family child care homes; its rela-
tively accurate portrayal of foods and beverages and nutrients consumed,
with proper training of observers; and the opportunity to collect detailed
information about foods and beverages.
Many researchers have coupled direct observation in child care with a
second method to assess dietary intake at home (i.e., in the morning before
child care and in the evening after child care). For example, Bruening et
al. (1999) combined direct observation in child care with 24-hour recall
from parents (i.e., the 24-hour recall excluded time in child care). Briley
et al. (1999) combined direct observation in child care with food records
from parents (i.e., the food records excluded time in child care). Bollela et
al. (1999) combined direct observation with both 24-hour recall and food
records.
Food Records in Child Care
Food records in child care involve providers recording the amount of
all foods and beverages consumed by children throughout the day. Data
are often reduced only to food groups. Estimating nutrient content is more
challenging (e.g., if a provider does not record condiment consumption, any
estimate of fat or sodium consumption would be inaccurate). While it is
possible for providers to record both foods provided and foods consumed,
researchers typically ask providers to focus on one or the other. The method
requires clear instructions and visual aids to help the provider determine
portion sizes and other details (e.g., method of food preparation). Research-
OCR for page 28
28 RESEARCH METHODS TO ASSESS DIETARY INTAKE
ers have not had enough experience with food records to know how many
children a single provider can account for at any given time. As an example
of food record data collection, Hoyo et al. (2011) asked child care provid-
ers to record foods consumed by specific children in child care and asked
parents of those children to conduct a 24-hour recall for foods consumed
outside of child care.
Food record data collection is challenging. The method is limited by the
small number of children that can be covered by a provider (one or two);
its high risk for inaccuracy, which makes it difficult to calculate nutrient
intake; the burden it imposes on providers, given that providers must pay
attention to all children during meals and snacks; and limited opportunity
to correct mistakes at a later time. Its strengths include its relatively low
cost and little researcher time; its lesser invasiveness and disruption for both
providers and children, compared to direct observation; its appropriateness
for documenting intake in children of all ages, including infants (Benjamin
Neelon remarked that food records are “probably one of the better ways
to report usual intake” for infants); its potential to be used in multiple
classrooms; and its appropriateness for both child care centers and homes.
Indirect Observation in Child Care
Benjamin Neelon described indirect observation as an “interesting al-
ternative” to assessing dietary intake in child care. She was unable to find
any published studies that rely on the method but described work that
she and her colleagues have been conducting. The staff member places a
video camera in the classroom to record all meals and snacks; the camera
must be positioned properly; otherwise, some consumption can be missed
(e.g., if children move around while eating or drinking). Once the data are
collected, trained observers view the videotapes and, as with direct obser-
vation, document all foods and beverages provided to and consumed by
children. The data are then reduced to foods and beverages by food groups
(e.g., milk) and nutrients (e.g., grams of fat).
The limitations of indirect observation in child care include the need
for target children to sit in a specified location and not move; the inability
to adapt to changes in the setting unless the observer stays with the video
camera; limited opportunity to correct mistakes; its requirement for some
onsite interaction to record certain types of information about foods and
beverages (e.g., preparation method); and its limitations for documenting
intake in infants (unless the video camera scans the entire room). Benefits of
the methodology are the ability to assess multiple children at once; its rela-
tively low cost and little researcher time; its lesser invasiveness and disrup-
tion compared to direct observation (i.e., children typically become adapted
to the presence of video cameras and eventually ignore them); the potential
OCR for page 29
29
ALIGNMENT OF YOUNG CHILDREN’S DIETARY INTAKE
for multiple observers to review and record dietary intake at a later time,
building some quality control into the method; the potential for data to be
collected in multiple classrooms (i.e., using multiple video cameras); and its
applicability in both child care centers and family child care homes.
The question-and-answer period at the end of the first session included
some discussion about indirect observation. First, Joanne Guthrie asked
about the use of digital photography as a data collection method. Benjamin
Neelon replied that food consumption in child care is a dynamic process,
with children moving, trading food, throwing food, etc., which can be dif-
ficult for a still photo to capture. Dixon mentioned a colleague who worked
on a large school food project in New York City that involved photography.
In addition to trying to capture what is a dynamic process, it was difficult to
get detailed information about type and amount of food. For example, one
may see juice but not know whether it is 100 percent juice. With respect to
amounts, it was difficult to get the camera angled properly. The research-
ers eventually abandoned the effort. Moderator Karen Cullen added that it
would be very difficult to capture in a photograph the amount of trading
that goes on with elementary children. Second, Jay Hirschman asked about
the types of clearances and consents required for videography data collec-
tion, as it is not a method that FNS has used in any of its studies. Benjamin
Neelon responded that, in her work, she and colleagues ask parents to sign
a video consent release form in addition to the standard consent form. So
far, they have not received any resistance from parents.
ANALYSIS OF DIETARY DATA COLLECTED
FROM CHILD CARE SETTINGS7
Beth Dixon elaborated on what emerged as a major overarching theme
of the workshop: the choice of method depends on the level of detailed
information and outcome(s) desired. She said, “You have to begin with the
end in mind.” She identified two sets of “big picture” questions:
1. What do you want to measure? Is the goal of the study to measure
what is served, what the children are eating, or both?
2. How much detail do you want to measure? For example, do you
want to know “generic” types of foods being served or consumed
(e.g., that a child is drinking milk) or do you want to know “spe-
cific” types and amounts of foods being served or consumed (e.g.,
that a child is drinking 1/2 cup of 1 percent milk)?
7 This section summarizes the presentation of Beth Dixon, from New York University.
OCR for page 30
30 RESEARCH METHODS TO ASSESS DIETARY INTAKE
TABLE 2-2 Suitability of Different Types of Data for Collecting Generic
Versus Specific Information About Foods and Beverages, as Well as
Amounts
Generic Information Specific Information
Data Collection Method (e.g., “milk”) (e.g., “1% milk”) Amounts
Receipts Yes Yes No
Menus Yes Possibly Possibly
Staff reports Yes Possibly Likely
Direct observation Yes Likely Yes
SOURCE: Dixon, 2012.
The more detailed the information, the greater the accuracy when
comparing data to quantitative dietary recommendations and assessing
overall quality of foods served or consumed. While there are a variety of
sources of dietary data (see Table 2-2), Dixon said that direct observation
provides the greatest level of detail about types and amounts of foods either
served or consumed, with menu and staff-reported data being helpful for
confirmation.
The Challenge of Data Variability
Regardless of the source of data, Dixon cautioned that there is consid-
erable cross-center variability in the level of detail recorded. This is espe-
cially true of menu data collection. For example, a menu from one center
may simply list “milk” or “peaches,” while another may list “one-half cup
of 1 percent milk” or “4 ounces of canned peaches in light syrup.” Some-
times there is considerable cross-center variability in detail even with the
more detailed data collection methods, including direct observation.
Dixon described two “tiers” of variability in direct observation data
collection. First is the actual visual estimate of the type and amount of food
being served or consumed, with details varying not just from study to study
(e.g., depending on how much instruction is provided), but also from ob-
server to observer. With respect to amount, one of the challenges with direct
observation is that most estimates of amounts of foods served or consumed
are visual “guesstimates” based on observations or staff reports. Most
observers are trained to look at foods from a distance and count pieces or
mounds of something, which is very difficult given how much activity goes
on around food (e.g., children drop food; they share food; sometimes they
eat very quickly).
The second source of variability in direct observation data collection is
data entry into a dietary assessment software system. Researchers rely on
OCR for page 31
31
ALIGNMENT OF YOUNG CHILDREN’S DIETARY INTAKE
both public and private dietary assessment software systems for estimating
food group and nutrient composition of foods served or consumed. Dixon
mentioned the National Cancer Institute Automated Self-Administered 24-
hour Dietary Recall, which is based on the USDA Automated Multiple Pass
Method, the USDA Food and Nutrient Database for Dietary Studies, and
the USDA MyPyramid Equivalents Database; the University of Minnesota’s
Nutrition Data System for Research; and the Elizabeth Stewart Hands and
Associates Food Processor. Irrespective of the software system used, when
data are entered in real time, questions about types and amounts of foods
recorded on forms can be clarified as necessary. But when entered later,
and particularly if they are entered by someone other than the observer, it
becomes more difficult to confirm the information. When information is
not clear, data enterers have to use their best judgment, rely on defaults in
the software system, and, when possible, confirm information with staff at
the centers.
Scope of the Study Versus Detail of the Data
The best-quality data will be those that can be reduced to a level of
detail that allows for comparison with both CACFP-recommended meal
patterns and Dietary Reference Intakes (i.e., either Recommended Dietary
Allowance or Adequate Intake, depending on nutrient). This is true regard-
less of whether data are collected only while the child is in child care (and
regardless of type of child care facility) or both in care and out of care (i.e.,
total dietary intake over the course of an entire day), and regardless of
whether one is seeking information on what is being served or consumed.
Additionally, it would be helpful to collect age-dependent data so that com-
parisons can be made to age-based recommendations, as both meal pattern
and nutrient intake recommendations are different for different age groups.
One of the challenges for the child care setting is that children are often
observed randomly and their exact ages are unknown.
In addition to detail, an accurate comparison to CACFP-recommended
meal patterns and Dietary Reference Intakes requires accurate collection
of both types and amounts of foods. Reiterating what previous speakers
said, Dixon emphasized the importance of observer training. Research staff
members need to be trained to observe consistently, record consistently
(e.g., “cookie” versus “chocolate chip cookie” versus “Chips Ahoy choco-
late chip cookie”), accurately estimate common portion sizes, and enter
data into the data assessment software system consistently. To improve
identification of foods, observers can confirm with teachers or food service
staff and check the kitchen. To improve estimation of amounts, researchers
should provide observers with visual aids of portion sizes and list options
for proportions of portion sizes on forms (e.g., 1/4 cup, 1/2 cup, 3/4 cup).
OCR for page 32
32 RESEARCH METHODS TO ASSESS DIETARY INTAKE
Which Method(s)?
Importance in Study
less more
Number of Dietary Factors
less more
Distribution of
Dietary Factors
less more
Time
less more
Money
less more
Food Menus, Direct
Receipts Staff Reports Observation
FIGURE 2-2 Different dietary data collection methods provide varying levels of
detailed information and require varying amounts of time and money.
SOURCE: Dixon, 2012.
Because the collection of high-quality comprehensive dietary informa-
tion takes time and money (see Figure 2-2), Dixon cautioned that a nation-
Figure 2-2
ally representative survey will need to be either small(er) in scope and focus
on detailed dietary data collection from all centers or large(r) in scope with
a subsample that includes detailed dietary data collection.
REFERENCES
Ball, S. C., S. E. Benjamin, and D. S. Ward. 2007. Development and reliability of an observa-
tion method to assess food intake of young children in child care. Journal of the Ameri-
can Dietetic Association 107(4):656-661.
Ball, S. C., S. E. Benjamin, and D. S. Ward. 2008. Dietary intakes in North Carolina child-care
centers: Are children meeting current recommendations? Journal of the American Dietetic
Association 108(4):718-721.
Benjamin Neelon, S. E. 2012. Dietary assessment in young children: Total daily intake of food
and nutrients. Presented at the Institute of Medicine Workshop on Review of the Child
and Adult Care Food Program: Future Research Needs, Washington, DC, February 7.
OCR for page 33
33
ALIGNMENT OF YOUNG CHILDREN’S DIETARY INTAKE
Benjamin Neelon, S. E., K. A. Copeland, S. C. Ball, L. Bradley, and D. S. Ward. 2010. Com-
parison of menus to actual foods and beverages served in North Carolina child-care
centers. Journal of the American Dietetic Association 110(12):1890-1895.
Benjamin Neelon, S. E., H. Reyes-Morales, J. Haines, M. W. Gillman, and E. M. Taveras. Un-
published. Nutritional quality of foods and beverages on child-care center menus in
Mexico.
Bollella, M. C., A. Spark, L. A. Boccia, T. A. Nicklas, B. P. Pittman, and C. L. Williams. 1999.
Nutrient intake of Head Start children: Home vs. school. Journal of the American College
of Nutrition 18(2):108-114.
Briley, M. E., S. Jastrow, J. Vickers, and C. Roberts-Gray. 1999. Dietary intake at child-care
centers and away: Are parents and care providers working as partners or at cross-
purposes? Journal of the American Dietetic Association 99(8):950-954.
Bruening, K. S., J. A. Gilbride, M. R. Passannante, and S. McClowry. 1999. Dietary intake
and health outcomes among young children attending 2 urban day-care centers. Journal
of the American Dietetic Association 99(12):1529-1535.
Buzby, J. C., and J. F. Guthrie. 2002. Plate waste in school nutrition programs: Final report
to Congress. E-FAN-02-009: USDA/ERS.
Crepinsek, M. K., N. R. Burstein, E. B. Lee, S. D. Kennedy, and W. L. Hamilton. 2002. Meals
Offered by Tier 2 CACFP Family Child Care Providers—Effects of Lower Meal Reim-
bursements. Report submitted by Abt Associates, Inc. to U.S. Department of Agriculture,
Economic Research Service, Washington, DC. http://webarchives.cdlib.org/sw1s17tt5t/
http://ers.usda.gov/Publications/efan02006/ (accessed May 4, 2002).
Dixon, L. B. 2012. Analysis of dietary data collected from childcare settings. Presented at the
Institute of Medicine Workshop on Review of the Child and Adult Care Food Program:
Future Research Needs, Washington, DC, February 7.
Erinosho, T., L. B. Dixon, C. Young, L. M. Brotman, and L. L. Hayman. 2011. Nutrition
practices and children’s dietary intakes at 40 child-care centers in New York City. Journal
of the American Dietetic Association 111(9):1391-1397.
Fleishhacker, S., K. L. Cason, and C. Achterberg. 2006. “You had peas today?”: A pilot study
comparing a Head Start child-care center’s menu with the actual food served. Journal of
the American Dietetic Association 106(2):277-280.
Fox, M. K., F. B. Glanz, L. Geitz, and N. Burstein. 1997. Early Childhood and Child Care
Study: Nutritional Assessment of the CACFP Final Report, Volume II. Report submitted
by Abt Associates, Inc. to U.S. Department of Agriculture, Food and Consumer Service,
Alexandria, VA. http://www.fns.usda.gov/ora/menu/published/CNP/FILES/ChldCare
Vol2Part1.pdf (accessed May 4, 2012).
Gordon, A., and M. K. Fox. 2007. School Nutrition Dietary Assessment Study-III: Summary
of Findings. Report submitted by Mathematica Policy Research, Inc. to U.S. Depart-
ment of Agriculture, Food and Nutrition Service, Alexandria, VA. http://www.fns.usda.
gov/ora/menu/published/cnp/FILES/SNDAIII-SummaryofFindings.pdf (accessed May 4,
2012).
Graves, K., and B. Shannon. 1983. Using visual plate waste measurement to assess school
lunch food behavior. Journal of the American Dietetic Association 82(2):163-165.
Hoyo, C., A. P. Murtha, J. M. Schildkraut, M. R. Forman, B. Calingaert, W. Demark-
Wahnefried, J. Kurtzberg, R. L. Jirtle, and S. K. Murphy. 2011. Folic acid supplemen-
tation before and during pregnancy in the Newborn Epigenetics STudy (NEST). BMC
Public Health 11(1):46.
IOM (Institute of Medicine). 2000. Dietary Reference Intakes: Applications in dietary assess-
ment. Washington, DC: National Academy Press.
Monsivais, P., and D. B. Johnson. 2012. Improving nutrition in home child care: Are food
costs a barrier? Public Health Nutrition 15(2):370-376.
OCR for page 34
34 RESEARCH METHODS TO ASSESS DIETARY INTAKE
Nestlé. 2011. FITS 2008 food measurement guide. http://medical.gerber.com/nestlescience/fits.
aspx (accessed May 7, 2012).
Ward, D. S., D. Hales, K. Haverly, J. Marks, S. Benjamin, S. Ball, and S. G. Trost. 2008. An
instrument to assess the obesogenic environment of child care centers. American Journal
of Health Behavior 32(4):380-386.
