An Assessment of Food Offerings and Marketing Strategies in the Food-Service
Venues at California Children's Hospitals
Lenard I Lesser, MD MSHS (Corresponding Author. Reprints not available.)
Robert Wood Johnson Foundation Clinical Scholars Program and Department of Family
Medicine, David Geffen School of Medicine, University of California, Los Angeles
911 Broxton Ave, 3rd Floor
Los Angeles, CA 90049
Phone: 310-794-2268 Fax: 310-794-3288
Dana E Hunnes RD MPH, Department of Nutrition, UCLA Medical Center and School of
Public Health, University of California, Los Angeles
Phedellee Reyes MPH, Department of Medicine, David Geffen School of Medicine,
University of California, Los Angeles
Lenore Arab PhD, Department of Medicine, David Geffen School of Medicine, University
of California, Los Angeles
Gery W Ryan PhD, RAND Corporation, Santa Monica, CA
Robert H Brook MD ScD, RAND Corporation, Santa Monica, CA
Deborah A Cohen MD MPH, RAND Corporation, Santa Monica, CA
NEMS-R: Nutrition Environment Measures Survey for Restaurants
NEMS-C: Nutrition Environment Measures Survey for Cafeterias
Word Count: Abstract: 249. Manuscript: 2552
Food Service, Hospital/organization & administration
Hospitals, Pediatric/organization & administration
The Robert Wood Johnson Foundation Clinical Scholars Program funded this project.
We have no conflicts of interest to report. Mrs. Hunnes works for the Department of
Nutrition at UCLA, but was not involved in the rating of the institutions.
Background: Marketing strategies and food offerings in hospital cafeterias can impact
Objectives: Using a survey adapted to assess food environments, assess the food
environment available to patients, staff, and visitors at the food-service venues in all
fourteen California children’s hospitals.
Design: Direct observation of hospital food-service venues; adaptation of the
Nutritional Environment Measures Survey for Cafeterias (NEMS-C) by partnering with a
hospital wellness committee.
Setting: All fourteen tertiary care children’s hospitals in California.
Main Outcome Measures: NEMS-C score, which summarizes the number of healthy
items offered, whether calorie labeling is present, if there is signage promoting healthy
or unhealthy foods, pricing structure, and the presence of unhealthy combination meals.
The range of possible scores is zero (unhealthy) to 37 (healthy).
Results: Inter-rater reliability showed 89% agreement on the assessed items. For the
fourteen hospitals, the mean score was 19.1 (sd= 4.2, range 13-30). Analysis revealed
that nearly all hospitals offered diet drinks, low-fat milk, and fruit. Fewer than one-third
had nutrition information at the point of purchase and 30% had signs promoting healthy
eating. Most venues displayed high calorie impulse items such as cookies and ice
cream at the registers. Eight percent of the 359 entrees served were classified as
healthy according to NEMS criteria.
Conclusions: Most children’s hospitals’ food venues received a mid-range score,
demonstrating there is considerable room for improvement. Many inexpensive options
are underused, such as providing nutritional information, incorporating signage that
promotes healthy choices, and not presenting unhealthy impulse items at the register.
Using an objective scale, this study directly observed the food-service venues of the 14
major California Children’s Hospitals. Most food venues achieved a mid-range score,
indicating there is room for improvement in the food offerings and the marketing
practices at these hospitals.
The epidemics of overweight and obesity are a subject of national concern.
Particular focus has been on the consequences of childhood obesity for the future
health of our country. Increasingly, dealing with these epidemics has shifted from the
individual level to the community and public health level.1, 2 While the evidence is still
emerging, it appears that the “food environment” where people work and live influences
what people eat, and therefore, how much weight they gain.3-5 For instance, students
develop healthier eating patterns in schools where the food environment has been
altered by increasing the nutritional quality of food served.2, 6
Children’s hospitals represent locations with great potential for influencing what
people eat. Marketing practices, such as pricing, food placement, and signage, can
change the way hospital staff and visitors eat. Improving the marketing practices and
food offered at hospital food venues can improve the work-site food environment for the
staff at the hospital as well as the children and families visiting the hospital.
Furthermore, hospital food venues can serve as exemplars of healthy food
environments and function as a place where nutrition education can be directed at
children and their families.
The potential of the food environment at children’s hospitals to influence
purchasing is indicated in a study that looked at three children’s hospitals in Chicago:
one with a McDonald’s, one with McDonald’s branding (without a restaurant), and one
with neither. Families surveyed at the hospital with McDonald’s were four times more
likely to consume McDonald’s on that day, and were twice as likely to think McDonald’s
was healthy, as compared to families at the hospitals without McDonald’s.7
The few studies of hospital food environments that exist have found poor nutrition
environments. In 2002, a study of the nation’s “top 16” hospitals found that 38% were
serving “fast food”.8 A larger study conducted at all academically affiliated hospitals in
the United States found that 42% of 234 hospitals were serving brand name “fast food,”
including Krispy Kreme™, Subway™, Burger King™, and McDonald’s™ (listed in order
of decreasing prevalence).9
Other investigators have graded pediatric hospital cafeterias by scoring and
ranking their performance based upon interviews with food service directors.10 While
this method enabled the collection of information from a large number of hospitals,
these investigators were not able to personally visit every hospital. Therefore, the
findings could be biased by self-report. We aimed to address these limitations by
substituting objective measures in food venue assessments.
The provision of food in a hospital setting is itself an implied message to patrons
and employees since it models acceptable meals and other food service behaviors.
Therefore, it is critical to assess the messages that are being conveyed. Therefore, the
main purpose of the study was to evaluate the food-service venues of California’s
tertiary children’s hospitals.
We evaluated the food-service venues (cafeterias and fast food restaurants) in all
fourteen tertiary-care hospitals that are members of the California Children’s Hospital
Association.15 These hospitals include the University of California Hospitals, as well as
private non-profit hospitals.
A recent review of methods assessing food environments revealed 137 tools, of
which only 18 had some psychometric testing.11 The most widely used is the Nutrition
Environment Measures Scale for Restaurants (NEMS-R)12 , a 25-item checklist that
requires systematic observation. Inter-rater and test-retest reliability is high, with most
kappa values greater than 0.80. Details on the NEMS-R can be found in the original
publications and online.13 The NEMS-R documents elements of the food environment
including pricing, availability of vegetables, nutrition labeling, combo promotions, and
healthy beverage items. The tool classifies healthy meals as entrees that provide point
of sale nutrition information and meet the following guidelines: 1) Entrees must not have
more than 800 calories and sandwiches must not have more than 650 calories. 2)
Entrees must also have ≤ 30% calories from fat, and ≤ 10% calories from saturated fat.
Entrees can also qualify as healthy if the food venue labels the food as a healthy
choice, even if no nutrition information is provided. Entrees that have no nutritional
information or healthy choice labeling were not considered to qualify as healthy meals
even if indeed they were, since the consumer may not be able to independently judge
the quality or quantity of the food.
We slightly modified the NEMS-R to be applicable to all hospital food-service
venues. We named the revised scale “NEMS-C” (“Cafeteria”), indicating that this new
version of NEMS could be used in cafeterias. To assure face validity of the instrument,
we created a partnership with the University of California, Los Angeles (UCLA) Medical
Center Wellness Committee, a group of nine dietitians and food service managers.
Using input from the Wellness Committee, we minimally changed the NEMS-R to make
the criteria and scoring appropriate for the cafeteria setting. We eliminated some
labeling items that were duplicates on the original NEMS-R. We also disregarded
information on take-away menus and internet nutrition information, as our partnership
deemed these not applicable to most hospital environments. We then added one item
to the tool: whether the cafeteria had high-calorie impulse items at the register (as
placement of foods can contribute to increased purchasing14). We did not add any
items regarding vending machines, as other tools exist for this purpose. We did not
score children’s menus, which were only available in one hospital. Since we only made
minor changes to the tool, the revised tool could still be used in fast food restaurants,
which were present in some of the hospitals.
The original NEMS-R scoring system (available from the authors of reference
12), gave most items on the scale +3 or -3 points, based on evidence as to whether that
item contributed to healthy eating. We assessed the face validity of this scoring by
surveying the Wellness Committee members and asking them to rate how important
they thought each item was in maintaining a healthy cafeteria. Since they gave all items
approximately equal importance, all NEMS-C items were given a score of +1 if they
promoted health, and -1 if they promoted an unhealthy behavior. Hospitals received a
zero for an item (healthy or unhealthy) if it was not present. Entrees and salads were
scored slightly differently (as in the original NEMS-R): 1 point for one healthy entrée or
salad, 2 points for 2-4 healthy entrees, and 3 points for 5 or more healthy entrees or
salads. Additionally, hospitals could get an additional 1 point if 1-24% of their entrees
were healthy, 2 points if 25-49% of their entrees were healthy, and 3 points if >50% of
their entrees were healthy. These “extra” points were added on because the Wellness
Committee thought the majority of entrees that hospitals serve should be healthy. This
led to total possible scores of zero (least healthy) to 37 (most healthy). The full scoring
sheet can be found in the appendix.
After completing NEMS online training provided by the University of
Pennsylvania’s Nutrition Environment Measures Survey website, two investigators (LIL
and PR) divided the fourteen children’s hospitals according to location: Northern or
Southern California. Each investigator visited each hospital during weekday hours of
11am and 2 pm during the month of July 2010 when lunch menus, including daily
entrees and specials, were served to hospital employees, patients, and visitors.
Investigators visited and evaluated each hospital unannounced, except where security
procedures required visitors to obtain clearance the day before (2 hospitals).
We evaluated all the food-service venues in each listed hospital. We excluded
“grab and go” items, except pre-made salads with nutrition information, as preliminary
observations indicated that most “grab and go” items were duplicates of items that could
be ordered at a food station. If a hospital had more than one food venue, we scored
each separately. We also scored-service venues that were not cafeterias (i.e. brand-
name fast-food restaurants) on hospital grounds. The investigator filled out the entire
NEMS-C evaluation for each venue, keeping subjective notes for any item for which the
scoring was not clear. After visiting the hospitals, the investigators discussed the
unclear items and decided on a consensus score. For venues that were scored by both
investigators, we used one investigators’ (LIL) scoring for the final results. We
calculated overall prevalence for each healthy and unhealthy NEMS-C criterion. We
then calculated each venue’s score.
Each investigator evaluated eight hospitals. Two of the hospitals were evaluated
by both investigators to both aid in training and for reliability testing. For the three
venues in two hospitals, there were 96 individual items on the NEMS-C that were
scored. The two raters agreed on 86 of 96 items (89% agreement). The overall
difference in score between the three venues that we jointly scored was two points or
less. The most frequent disagreement was being off by one in counting healthy
entrees. For instance, one investigator may have counted two healthy entrees, while
the other counted three.
In the fourteen hospitals we found sixteen food-service venues. Eight venues
were contained in a stand-alone children’s hospital; six were in an adult hospital that
served as a venue for the attached children’s hospital; one was in a children’s hospital
that was attached to the adult hospital; and one was in a physically separate adult
hospital that served as a food venue for a children’s hospital.
The Table shows the percent of venues that had healthy and unhealthy NEMS-C
items. The majority of venues had low-fat or skim milk, diet soda, baked chips, fruit
without sugar, a salad bar, and non-fried vegetables (outside of the salad bar). About
half of the venues had nutrition information on the menu or healthy item designations,
whereas about one-third had nutrition information at the point of purchase. Half the
venues had low-fat or fat-free salad dressing. Less than half of the venues had signs
encouraging healthy eating (e.g. “Make sure to eat your fruits and vegetables”).
Approximately one-third of the venues had signs highlighting healthy menu options, and
served 100% whole-wheat bread. Three venues had five or more healthy entrees. Two
venues offered reduced portion sizes at reduced prices. No venue had menu notations
that encouraged healthy requests.
The most common unhealthy finding was that 81% of hospitals had high-calorie
impulse items, such as ice cream freezers, cookies, and candy, at or near the checkout
register. Half of the venues had combo options (often with a side and a drink) where
the combo pricing gave a discount for purchasing more items. Thirty-eight percent
(38%) had signs encouraging unhealthy eating, such as those promoting high calorie
items or combo deals. In hospitals that served healthy entrees, 38% percent of the
venues priced these healthy entrees, on average, higher than the unhealthy entrees.
Across the 16 venues we recorded 384 entrees or sandwiches. According to NEMS
criteria, 7% of these entrees were categorized as healthy. The overall scores for the
venues varied with a range of 13-30 and a mean of 19.1 (s.d. 4.2) (Figure).
Children’s hospitals are intended to make the sick healthy. As multiple efforts
are being made in attempt to change the food environments of schools and
communities, hospitals are another place to initiate a healthy food environment.
California’s children’s hospitals’ food-service venues demonstrate a wide range
of healthy and unhealthy food practices. Most venues received a score in the mid-
range of the NEMS-C. Only a small minority (7%) of the entrees and sandwiches met
the rigorous NEMS standard for being “healthy.”
A few studies have assessed food offerings at hospitals. These studies have
mostly relied on telephone interviews of cafeteria directors, with a tool developed by a
particular investigator. For instance, one study called food service directors at
children’s hospitals in the US and Canada to grade them on the variety of food
availability.10 While this system can assess many hospitals in a relatively short time, it
suffers from two major weaknesses: 1) food service directors may give biased answers
about what they serve, and 2) marketing practices (pricing, signage, etc.) that contribute
to food purchases are not assessed.
In contrast, this study modified a widely used scale of the food environment,
NEMS-R, into a tool that could be used to objectively assess hospital food venues. The
new scale, NEMS-C, has increased legitimacy in that it was developed in partnership
with a group that makes the decisions on what to serve in hospitals.
Yet the current scale needs further testing. We only tested it in a small sample of
California children’s hospitals, most of which were large teaching hospitals.
Additionally, the scoring was based on one wellness committee’s input. However,
additional analyses (not shown) demonstrated that alternating the scoring of certain
items did not appreciably change the distribution.
The current tool takes only about 30 minutes to use, but the training takes a
significant amount of time. However, the strength of this tool is that it provides objective
standards that hospital food-service venues can strive to achieve.
The major limitation of this study is that we did not observe what people actually
ate. It is not known whether customers eating at a high scoring venue are any more
likely to consume fewer calories or healthier meals. Even if healthier cafeterias resulted
in healthier meals being consumed, people could compensate by consuming too many
calories during other parts of the day. Furthermore, there are other places in the
hospital where visitors and staff can eat food, such as candy shops and vending
machines. This study did not evaluate those areas, where people may be able to
purchase high caloric foods, such as sugar-sweetened beverages.
Despite these limitations, hospitals achieving a high score are likely to provide a
healthy food environment that, in turn, fosters healthy eating. In a healthy hospital
cafeteria, the majority of entrees would be less than 800 calories and be labeled as
such; there would be no impulse items sold at the cash register; there would be multiple
choices of vegetables and fresh fruits, side dishes with whole grains; and a salad bar
with low calorie dressings. Deep-fried dishes and sugar-sweetened beverages would
not be available. Overall, a higher NEMS-C score requires more fruits, vegetables, and
whole grains; smaller portions; and fewer low-nutrient choices. The results of this study
show that many inexpensive options for improving the NEMS-C score are underused,
such as providing nutritional information, incorporating signage that promotes healthy
choices, and not presenting unhealthy impulse items at the register. However, one of
the studied hospitals (seen as the outlier in the Figure) was able to score well on many
of the NEMS-C criteria, which indicates that improvements are likely possible
Beyond the individual choices of consumers at a particular cafeteria, hospital
food offerings represent an opportunity to model healthy food choices for patients,
employees, and visitors. Because visitors to hospitals believe that the foods the hospital
offers are more likely to be healthy,7 hospitals should make an effort to meet this
expectation. Hospitals can use the NEMS-C scale to assess and improve the food
environment. National hospital organizations should establish national standards for
labeling, pricing, marketing, and food quality in their facilities. This is one area where
health professionals have the direct opportunity to improve the food landscape in the
United States. Achieving ideal children’s hospital food environments would be one
small, but significant, step in changing the quality and quantity of what our children eat
and preventing the onset of dietary-related chronic diseases.
Acknowledgement: The authors would like to thank Patti Oliver, MS, RD, MBA and the
UCLA Medical Center Wellness Committee for their assistance in refining the scale
used in this study.
Financial Disclosure/Conflict of Interest: None of the authors have any financial
conflict of interest in the study. Dana Hunnes is associated with one of the cafeterias
studied in this project, but had no role in the assessment of any of the hospitals.
Funding: Support for this project was provided by a grant from the Robert Wood
Johnson Foundation Clinical Scholars Program. The Foundation had no role in the
study design or outcome.