Faculty Bibliography

Background: The Veterans Administration (VA) cares for over 8 million U.S. veterans annually, approximately 20% of whom have prevalent diabetes. To foster research and intervention opportunities, we developed the VA Diabetes Risk (VADR) Cohort using the VA electronic health record, a national cohort of diabetes-free U.S. veterans receiving primary care at the VA since January 1, 2008. This cohort provides important opportunities to study community-level risk factors for diabetes, such as attributes of the food environment, via geospatial linkage to residence information. We describe here the cohort profile and diabetes incidence by sub-group.
Method(s): To be eligible, diabetes-free patients had to have at least 2 primary care visits at least 30 days apart prior to enrollment. Diabetes incidence was defined as having >=2 inpatient or outpatient encounters with diabetes ICD-9/10 codes, any prescription of diabetes medicine, or one encounter with diabetes ICD-9/10 codes and >=2 hemoglobin A1C >=6.5%. The incidence of diabetes was calculated as the number of new cases diagnosed per 1000 person-years (PY) through December 31, 2018. Demographic and comorbidities data were abstracted using diagnostic codes, labs, prescriptions, and vital signs.
Result(s): The VADR cohort consisted of 6.17 million veterans, the majority of whom were male (91.7%) and non- Hispanic (NH) white (75.7%). Nearly half were above 60 years of age at enrollment (48.8%). The diabetes incidence rate was 27.0 per 1000 PY, increasing with age from 13.3 per 1000 PY among adults <45 years old to 41.8 per 1000 PY among those 65 years and older. Incidence was higher among men than women (34.6 vs. 18.6 per 1000 PY) and higher among NH black patients compared to NH white patients (38 vs. 31.7 per 1000 PY).
Conclusion(s): The VADR cohort provides a novel infrastructure for examination of community-level risk factors for diabetes among veterans, and facilitates assessment of the impact of national or regional strategies to prevent or manage diabetes in veterans

OBJECTIVE:Conceptualisations of 'food deserts' (areas lacking healthful food/drink) and 'food swamps' (areas overwhelm by less-healthful fare) may be both inaccurate and incomplete. Our objective was to more accurately and completely characterise food/drink availability in urban areas. DESIGN/METHODS:Cross-sectional assessment of select healthful and less-healthful food/drink offerings from storefront businesses (stores, restaurants) and non-storefront businesses (street vendors). SETTING/METHODS:Two areas of New York City: the Bronx (higher-poverty, mostly minority) and the Upper East Side (UES; wealthier, predominantly white). PARTICIPANTS/METHODS:All businesses on 63 street segments in the Bronx (n 662) and on 46 street segments in the UES (n 330). RESULTS:Greater percentages of businesses offered any, any healthful, and only less-healthful food/drink in the Bronx (42·0 %, 37·5 %, 4·4 %, respectively) than in the UES (30 %, 27·9 %, 2·1 %, respectively). Differences were driven mostly by businesses (e.g. newsstands, gyms, laundromats) not primarily focused on selling food/drink - 'other storefront businesses' (OSBs). OSBs accounted for 36·0 % of all food/drink-offering businesses in the Bronx (more numerous than restaurants or so-called 'food stores') and 18·2 % in the UES (more numerous than 'food stores'). Differences also related to street vendors in both the Bronx and the UES. If street vendors and OSBs were not captured, the missed percentages of street segments offering food/drink would be 14·5 % in the Bronx and 21·9 % in the UES. CONCLUSIONS:Of businesses offering food/drink in communities, OSBs and street vendors can represent substantial percentages. Focusing on only 'food stores' and restaurants may miss or mischaracterise 'food deserts', 'food swamps', and food/drink-source disparities between communities.

Using data from the United States Behavioral Risk Factor Surveillance System (2003-2012; N = 3,397,124 adults), we estimated associations between prevalent diabetes and four county-level exposures (fast food restaurant density, convenience store density, unemployment, active commuting). All associations confirmed our a priori hypotheses in conventional multilevel analyses that pooled across years. In contrast, using a random-effects within-between model, we found weak, ambiguous evidence that within-county changes in exposures were associated with within-county change in odds of diabetes. Decomposition revealed that the pooled associations were largely driven by time-invariant, between-county factors that may be more susceptible to confounding versus within-county associations.

No studies have documented the prevalence of the food industry's funding of academic programs, which is problematic because such funding can create conflicts of interest in research and clinical practice. We aimed to quantify the publicly available information on the food industry's donations to academic programs by documenting the amount of donations given over time, categorizing the types of academic programs that receive food industry donations, cataloguing the source of the donation information, and identifying any stated reasons for donations. Researchers cataloged online data from publicly available sources (e.g., official press releases, news articles, tax documents) on the food industry's donations to academic programs from 2000 to 2016. Companies included 26 food and beverage corporations from the 2016 Fortune 500 list in the United States. Researchers recorded the: (1) monetary value of the donations; (2) years the donations were distributed; (3) the name and type of recipient; (4) source of donation information; and (5) reasons for donations. Adjusting for inflation, we identified $366 million in food industry donations (N = 3274) to academic programs. Universities received 45.2% (n = 1480) of donations but accounted for 67.9% of total dollars given in the sample. Community colleges, schools (i.e., preschool, elementary, middle, and high schools), and academic nonprofits, institutes, foundations, and research hospitals collectively received 54.8% of the donations, but made up less than one-third of the monetary value of donations. Half of the donations (49.0%) did not include a stated reason for the donation. In our sample, donations grew from $3 million in 2000 to $24 million in 2016. Food companies in our sample donated millions of dollars to universities and other academic programs but disclosed little information on the purpose of the donations. Achieving transparency in donation practices may only be possible if federal policies begin to require disclosures or if companies voluntarily disclose information.

This study used interviews with New York City Hispanic Caribbean (HC) restaurant owners, managers, and cooks/chefs (n=19) to examine perceptions concerning the healthfulness of the HC diet and diet-related disparities in the HC community, and document factors potentially influencing their engagement in community nutrition interventions. The interviews revealed high awareness of diet-related issues. Respondents had mixed notions concerning their role in improving community food environments, noting important barriers for collaboration to consider in future interventions. The study underscores the important role of ethnic restaurants, providing information to facilitate engagement with this largely untapped sector in immigrant/ethnic communities in the US.

OBJECTIVE:This study aimed to examine the relationship between proximity to healthy and unhealthy food outlets around children's homes and their weight outcomes. METHODS:A total of 3,507,542 student-year observations of height and weight data from the 2009-2013 annual FitnessGram assessment of New York City public school students were used. BMI z scores were calculated, student obesity or obesity/overweight was determined using Centers for Disease Control and Prevention growth charts, and these data were combined with the locations of four food outlet types (fast-food restaurants, wait-service restaurants, corner stores, and supermarkets) to calculate distance to the nearest outlet. Associations between weight status outcomes and distance to these food outlet types were examined using neighborhood (census tract) fixed effects. RESULTS:Living farther than 0.025 mile (about half of a city block) from the nearest fast-food restaurant was associated with lower obesity and obesity/overweight risk and lower BMI z scores. Results ranged from 2.5% to 4.4% decreased obesity. Beyond this distance, there were generally no impacts of the food environment and little to no impact of other food outlet types. CONCLUSIONS:Proximity to fast-food restaurants was inversely related to childhood obesity, but no relationships beyond that were seen. These findings can help better inform policies focused on food access, which could, in turn, reduce childhood obesity.

OBJECTIVE:Researchers have linked geographic disparities in obesity to community-level characteristics, yet many prior observational studies have ignored temporality and potential for bias. METHODS:Repeated cross-sectional data were used from the Behavioral Risk Factor Surveillance System (BRFSS) (2003-2012) to examine the influence of county-level characteristics (active commuting, unemployment, percentage of limited-service restaurants and convenience stores) on BMI. Each exposure was calculated using mean values over the 5-year period prior to BMI measurement; values were standardized; and then variables were decomposed into (1) county means from 2003 to 2012 and (2) county-mean-centered values for each year. Cross-sectional (between-county) and longitudinal (within-county) associations were estimated using a random-effects within-between model, adjusting for individual characteristics, survey method, and year, with nested random intercepts for county-years within counties within states. RESULTS:(95% CI: -0.72 to -0.31) decrease in BMI over time. CONCLUSIONS:These results suggest that community-level characteristics play an important role in shaping geographic disparities in BMI between and within communities over time.

Objective/UNASSIGNED:To adapt and apply the Nutrition Environment Measures Survey for Restaurants (NEMS-R) to Hispanic Caribbean (HC) restaurants and examine associations between restaurant characteristics and nutrition environment measures. Methods/UNASSIGNED:We adapted the NEMS-R for HC cuisines (Cuban, Puerto Rican, Dominican) and cardiovascular health-promoting factors, and applied the instrument (NEMS-HCR) to a random sample of HC restaurants in New York City (NYC) (N=89). Multivariable linear regression was used to assess independent associations between NEMS-HCR score and restaurant characteristics (cuisine, size, type [counter-style vs sit-down] and price). Results/UNASSIGNED:None of the menus in the restaurants studied listed any main dishes as "healthy" or "light." More than half (52%) offered mostly (>75%) nonfried main dishes, and 76% offered at least one vegetarian option. The most common facilitator to healthy eating was offering reduced portion sizes (21%) and the most common barrier was having salt shakers on tables (40%). NEMS-HCR scores (100-point scale) ranged from 24.1-55.2 (mean=39.7). In multivariable analyses, scores were significantly related to cuisine (with Puerto Rican cuisine scoring lower than Cuban and Dominican cuisines), and size (with small [<22 seats] restaurants scoring lower than larger restaurants). We found a significant quadratic association with midpoint price, suggesting that scores increased with increasing price in the lowest price range, did not vary in the middle range, and decreased with increasing price in the highest range. Conclusions/UNASSIGNED:Our application of the NEMS-R to HC restaurants in NYC revealed areas for potential future interventions to improve food offerings and environmental cues to encourage healthful choices.

Prior research has documented the ways in which financial hardships negatively impact health, particularly mental health. However, this association between financial hardships and mental health outcomes has rarely been examined in sexual minorities. The purpose of this study was to examine associations between financial hardships and mental health burdens among a sample of gay, bisexual and other men who have sex with men (MSM) in Paris, France. Participants (n = 580) completed a cross-sectional survey advertised on a geosocial networking application. Participants responded to measures of mental health, financial hardship, and socio-demographics. Modified Poisson models were used to estimate risk ratios (RRs) and 95% confidence intervals (CI) for the associations between financial hardship and the following outcomes: 1) depressive symptoms, 2) anxiety symptoms, and 3) psychological distress. After adjusting for socio-demographics, high financial hardships were associated with depressive symptoms (aRR: 1.48, 95% CI: 1.04, 2.11) and psychological distress (aRR: 1.56, 95% CI: 1.09, 2.23). Analyses also demonstrated that stress acts as a mediating variable. These preliminary results suggest that future interventions to reduce financial hardships may have positive effects on the mental health of such a population.

Background: Household food purchasing behavior has gained interest as an intervention to improve nutrition and nutrition-associated outcomes. However, evaluating food expenditures is challenging in epidemiological studies. Assessment methods that are both valid and feasible for use among diverse, low-income populations are needed. We therefore developed a novel simple annotated receipt method to assess household food purchasing. First, we describe and evaluate the extent to which the method captures food purchasing information. We then evaluate within- and between-household variation in weekly food purchasing to determine sample sizes and the number of weeks of data needed to measure household food purchasing with adequate precision. Methods: Four weeks of food purchase receipt data were collected from 260 low-income households in the Minneapolis-St. Paul metropolitan area. The proportion of receipt line items that could not be coded into one of 11 food categories (unidentified) was calculated, and a zero-inflated negative binomial regression was used to evaluate the association between unidentified receipt items and participant characteristics and store type. Within- and between-household coefficients of variation were calculated for total food expenditures and several food categories. Results: A low proportion of receipt line items (1.6%) could not be coded into a food category and the incidence of unidentified items did not appreciably vary by participant characteristics. Weekly expenditures on foods high in added sugar had higher within- and between-household coefficients of variation than weekly fruit and vegetable expenditures. To estimate mean weekly food expenditures within 20% of the group's usual ("true") expenditures, 72 households were required. Nine weeks of data were required to achieve an r = 0.90 between observed and usual weekly food expenditures. Conclusions: The simple annotated receipt method may be a feasible tool for use in assessing food expenditures of low-income, diverse populations. Within- and between-household coefficients of variation suggest that the number of weeks of data or group sizes required to precisely estimate usual household expenditures is higher for foods high in added sugar compared to fruits and vegetables.