Faculty Bibliography

BACKGROUND:Tobacco remains a leading cause of preventable death in the U.S., responsible for more than 440,000 deaths each year. Approximately 10% of these deaths are attributable to exposure of non-smokers to secondhand smoke (SHS). Residents living in public multi-unit housing (MUH) are at excess risk for SHS exposure compared to the general population. On November 30, 2016, the U.S. Department of Housing and Urban Development (HUD) passed a rule requiring all public housing agencies to implement smoke-free housing (SFH) policies in their housing developments by July 30, 2018. METHODS:As part of a larger natural experiment study, we designed a protocol to evaluate indoor SHS levels before and after policy implementation through collection of repeat indoor air samples in non-smoking apartments and common areas of select high-rise NYCHA buildings subject to the HUD SFH rule, and also from socio-demographically matched private-sector high-rise control buildings not subject to the rule. A baseline telephone survey was conducted in all selected buildings to facilitate rapid recruitment into the longitudinal study and assess smoking prevalence, behaviors, and attitudes regarding the SFH policy prior to implementation. Data collection began in early 2018 and will continue through 2021. DISCUSSION/CONCLUSIONS:The baseline survey was completed by 559 NYCHA residents and 471 comparison building residents (response rates, 35, and 32%, respectively). Smoking prevalence was comparable between study arms (15.7% among NYCHA residents and 15.2% among comparison residents). The majority of residents reported supporting a building-wide smoke-free policy (63.0 and 59.9%, respectively). We enrolled 157 NYCHA and 118 comparison non-smoking households into the longitudinal air monitoring study and performed air monitoring in common areas. Follow up surveys and air monitoring in participant households occur every 6 months for 2.5 years. Capitalizing on the opportunity of this federal policy rollout, the large and diverse public housing population in NYC, and robust municipal data sources, this study offers a unique opportunity to evaluate the policy's direct impacts on SHS exposure. Methods in this protocol can inform similar SFH policy evaluations elsewhere.

Demographic and income disparities may impact food accessibility. Research has not yet well documented the precise location of healthy and unhealthy food resources around children's homes and schools. The objective of this study was to examine the food environment around homes and schools for all public school children, stratified by race/ethnicity and poverty status. This cross-sectional study linked data on the exact home and school addresses of a population-based sample of public school children in New York City from 2013 to all corner stores, supermarkets, fast-food restaurants, and wait-service restaurants. Two measures were created around these addresses for all children: 1) distance to the nearest outlet, and 2) count of outlets within 0.25 miles. The total analytic sample included 789,520 K-12 graders. The average age was 11.78 years (SD ± 4.0 years). Black, Hispanic, and Asian students live and attend schools closer to nearly all food outlet types than White students, regardless of poverty status. Among not low-income students, Black, Hispanic, and Asian students were closer from home and school to corner stores and supermarkets, and had more supermarkets around school than White students. The context in which children live matters, and more nuanced data is important for development of appropriate solutions for childhood obesity. Future research should examine disparities in the food environment in other geographies and by other demographic characteristics, and then link these differences to health outcomes like body mass index. These findings can be used to better understand disparities in food access and to help design policies intended to promote healthy eating among children.

INTRODUCTION/BACKGROUND:The present study measures how racially-targeted food and beverage ads affect adolescents' attitudes toward ads and brands, purchase intentions for advertised products, and willingness to engage with brands on social media. METHODS:Black and White adolescents were recruited through Survey Sampling International in 2016. Participants completed an online survey in which they were randomized to view either four food and beverage ads (e.g., soda, candy commercials) featuring Black actors or four food and beverage ads featuring White actors. RESULTS:For the two components of the attitudinal outcome, Black participants were more likely to report a positive affective response toward racially-similar ads compared to Whites. However, White participants were more likely to like ads that were racially-dissimilar compared to Black participants. Data were analyzed in 2016-2017, and we used an alpha level of 0.05 to denote statistical significance. CONCLUSIONS:Both Black and White adolescents reported more positive affective responses to ads that featured Blacks compared to ads that featured Whites. Because there were no differences on two outcomes, future research should examine the influence of racially-targeted marketing in real-world contexts (e.g., social media) and longitudinal exposure to targeted advertising on dietary behavior.

BACKGROUND:Because of the strong link between childhood obesity and adulthood obesity comorbidities, and the difficulty in decreasing body mass index (BMI) later in life, effective strategies are needed to address this condition in early childhood. The ability to predict obesity before age five could be a useful tool, allowing prevention strategies to focus on high risk children. The few existing prediction models for obesity in childhood have primarily employed data from longitudinal cohort studies, relying on difficult to collect data that are not readily available to all practitioners. Instead, we utilized real-world unaugmented electronic health record (EHR) data from the first two years of life to predict obesity status at age five, an approach not yet taken in pediatric obesity research. METHODS AND FINDINGS/RESULTS:We trained a variety of machine learning algorithms to perform both binary classification and regression. Following previous studies demonstrating different obesity determinants for boys and girls, we similarly developed separate models for both groups. In each of the separate models for boys and girls we found that weight for length z-score, BMI between 19 and 24 months, and the last BMI measure recorded before age two were the most important features for prediction. The best performing models were able to predict obesity with an Area Under the Receiver Operator Characteristic Curve (AUC) of 81.7% for girls and 76.1% for boys. CONCLUSIONS:We were able to predict obesity at age five using EHR data with an AUC comparable to cohort-based studies, reducing the need for investment in additional data collection. Our results suggest that machine learning approaches for predicting future childhood obesity using EHR data could improve the ability of clinicians and researchers to drive future policy, intervention design, and the decision-making process in a clinical setting.

BACKGROUND:The decisions that individuals make about the food and beverage products they purchase and consume directly influence their energy intake and dietary quality and may lead to excess weight gain and obesity. However, gathering and interpreting data on food and beverage purchase patterns can be difficult. Leveraging novel sources of data on food and beverage purchase behavior can provide us with a more objective understanding of food consumption behaviors. OBJECTIVE:Food and beverage purchase receipts often include time-stamped location information, which, when associated with product purchase details, can provide a useful behavioral measurement tool. The purpose of this study was to assess the feasibility, reliability, and validity of processing data from fast-food restaurant receipts using crowdsourcing via Amazon Mechanical Turk (MTurk). METHODS:Between 2013 and 2014, receipts (N=12,165) from consumer purchases were collected at 60 different locations of five fast-food restaurant chains in New Jersey and New York City, USA (ie, Burger King, KFC, McDonald's, Subway, and Wendy's). Data containing the restaurant name, location, receipt ID, food items purchased, price, and other information were manually entered into an MS Access database and checked for accuracy by a second reviewer; this was considered the gold standard. To assess the feasibility of coding receipt data via MTurk, a prototype set of receipts (N=196) was selected. For each receipt, 5 turkers were asked to (1) identify the receipt identifier and the name of the restaurant and (2) indicate whether a beverage was listed in the receipt; if yes, they were to categorize the beverage as cold (eg, soda or energy drink) or hot (eg, coffee or tea). Interturker agreement for specific questions (eg, restaurant name and beverage inclusion) and agreement between turker consensus responses and the gold standard values in the manually entered dataset were calculated. RESULTS:Among the 196 receipts completed by turkers, the interturker agreement was 100% (196/196) for restaurant names (eg, Burger King, McDonald's, and Subway), 98.5% (193/196) for beverage inclusion (ie, hot, cold, or none), 92.3% (181/196) for types of hot beverage (eg, hot coffee or hot tea), and 87.2% (171/196) for types of cold beverage (eg, Coke or bottled water). When compared with the gold standard data, the agreement level was 100% (196/196) for restaurant name, 99.5% (195/196) for beverage inclusion, and 99.5% (195/196) for beverage types. CONCLUSIONS:Our findings indicated high interrater agreement for questions across difficulty levels (eg, single- vs binary- vs multiple-choice items). Compared with traditional methods for coding receipt data, MTurk can produce excellent-quality data in a lower-cost, more time-efficient manner.

BACKGROUND:Policy makers are considering changes to the Supplemental Nutrition Assistance Program (SNAP). Proposed changes include financially incentivizing the purchase of healthier foods and prohibiting the use of funds for purchasing foods high in added sugars. SNAP participant perspectives may be useful in understanding the consequences of these proposed changes. OBJECTIVE:To determine whether food restrictions and/or incentives are acceptable to food benefit program participants. DESIGN/METHODS:Data were collected as part of an experimental trial in which lower-income adults were randomly assigned to one of four financial food benefit conditions: (1) Incentive: 30% financial incentive on eligible fruits and vegetables purchased using food benefits; (2) Restriction: not allowed to buy sugar-sweetened beverages, sweet baked goods, or candies with food benefits; (3) Incentive plus Restriction; or (4) Control: no incentive/restriction. Participants completed closed- and open-ended questions about their perceptions on completion of the 12-week program. PARTICIPANTS/SETTING/METHODS:Adults eligible or nearly eligible for SNAP were recruited between 2013 and 2015 by means of events or flyers in the Minneapolis/St Paul, MN, metropolitan area. Of the 279 individuals who completed baseline measures, 265 completed follow-up measures and are included in these analyses. STATISTICAL ANALYSIS/METHODS:analyses were conducted to assess differences in program satisfaction. Responses to open-ended questions were qualitatively analyzed using principles of content analysis. RESULTS:There were no statistically significant or meaningful differences between experimental groups in satisfaction with the program elements evaluated in the study. Most participants in all conditions found the food program helpful in buying nutritious foods (94.1% to 98.5%) and in buying the kinds of foods they wanted (85.9% to 95.6%). Qualitative data suggested that most were supportive of restrictions, although a few were dissatisfied. Participants were uniformly supportive of incentives. CONCLUSIONS:Findings suggest a food benefit program that includes incentives for purchasing fruits and vegetables and/or restrictions on the use of program funds for purchasing foods high in added sugars appears to be acceptable to most participants.

Research has examined how the food environment affects the risk of cardiovascular disease (CVD). Many studies have focused on residential neighbourhoods, neglecting the activity spaces of individuals. The objective of this study was to investigate whether food environments in both residential and global positioning system (GPS)-defined activity space buffers are associated with body mass index (BMI) and blood pressure (BP) among low-income adults. Data came from the New York City Low Income Housing, Neighborhoods and Health Study, including BMI and BP data (n=102, age=39.3±14.1 years), and one week of GPS data. Five food environment variables around residential and GPS buffers included: fast-food restaurants, wait-service restaurants, corner stores, grocery stores, and supermarkets. We examined associations between food environments and BMI, systolic and diastolic BP, controlling for individual- and neighbourhood-level sociodemographics and population density. Within residential buffers, a higher grocery store density was associated with lower BMI (β=- 0.20 kg/m2, P<0.05), and systolic and diastolic BP (β =-1.16 mm Hg; and β=-1.02 mm Hg, P<0.01, respectively). In contrast, a higher supermarket density was associated with higher systolic and diastolic BP (β=1.74 mm Hg, P<0.05; and β=1.68, P<0.01, respectively) within residential buffers. In GPS neighbourhoods, no associations were documented. Examining how food environments are associated with CVD risk and how differences in relationships vary by buffer types have the potential to shed light on determinants of CVD risk. Further research is needed to investigate these relationships, including refined measures of spatial accessibility/exposure, considering individual's mobility.

Sedentary behavior and lack of physical activity are key modifiable behavioral risk factors for chronic health problems, such as obesity and diabetes. Little is known about how sedentary behavior and physical activity among adolescents spatially cluster. The objective was to detect spatial clustering of sedentary behavior and physical activity among Boston adolescents. Data were used from the 2008 Boston Youth Survey Geospatial Dataset, a sample of public high school students who responded to a sedentary behavior and physical activity questionnaire. Four binary variables were created: 1) TV watching (>2 hours/day), 2) video games (>2 hours/day), 3) total screen time (>2 hours/day); and 4) 20 minutes/day of physical activity (≥5 days/week). A spatial scan statistic was utilized to detect clustering of sedentary behavior and physical activity. One statistically significant cluster of TV watching emerged among Boston adolescents in the unadjusted model. Students inside the cluster were more than twice as likely to report > 2 hours/day of TV watching compared to respondents outside the cluster. No significant clusters of sedentary behavior and physical activity emerged. Findings suggest that TV watching is spatially clustered among Boston adolescents. Such findings may serve to inform public health policymakers by identifying specific locations in Boston that could provide opportunities for policy intervention. Future research should examine what is linked to the clusters, such as neighborhood environments and network effects.

BACKGROUND:Local food environments include food stores (eg, supermarkets, grocery stores, bakeries) and restaurants. However, the extent to which other storefront businesses offer food/drink is not well described, nor is the extent to which food/drink availability through a full range of storefront businesses might change over time. OBJECTIVES/OBJECTIVE:This study aimed to assess food/drink availability from a full range of storefront businesses and the change over time and to consider implications for food-environment research. DESIGN/METHODS:Investigators compared direct observations from 2010 and 2015. PARTICIPANTS/SETTING/METHODS:Included were all storefront businesses offering foods/drinks on 153 street segments in the Bronx, NY. MAIN OUTCOME MEASURES/METHODS:The main outcome was change between 2010 and 2015 as determined by matches between businesses. Matches could be strict (businesses with the same name on the same street segment in both years) or lenient (similar businesses on the same street segment in both years). Investigators categorized businesses as general grocers, specialty food stores, restaurants, or other storefront businesses (eg, barber shops/beauty salons, clothing outlets, hardware stores, laundromats, and newsstands). STATISTICAL ANALYSES PERFORMED/METHODS:Investigators quantified change, specifically calculating how often businesses in 2015 were present in 2010 and vice versa. RESULTS:Strict matches for businesses in 2015 present in 2010 ranged from 29% to 52%, depending on business category; lenient matches ranged from 43% to 72%. Strict matches for businesses in 2010 present in 2015 ranged from 34% to 63%; lenient matches ranged from 72% to 83%. In 2015 compared with 2010, on 22% more of the sampled street segments, 30% more businesses were offering food/drink: 66 vs 46 general grocers, 22 vs 19 specialty food stores, 99 vs 99 restaurants, 98 vs 56 other storefront businesses. CONCLUSIONS:Over 5 years, an urban food environment changed substantially, even by lenient standards, particularly among "other storefront businesses" and in the direction of markedly greater food availability (more businesses offering food on more streets). Failure to consider a full range of food/drink sources and change in food/drink sources could result in erroneous food-environment conclusions.

The purpose of this research was to examine whether the local food environment, specifically the distance to the nearest sugar sweetened beverage (SSB) vendor, a measure of SSB availability and accessibility, was correlated with the likelihood of self-reported SSB consumption among a sample of fast food consumers. As part of a broader SSB behavior study in 2013-2014, respondents were surveyed outside of major chain fast food restaurants in New York City (NYC). Respondents were asked for the intersection closest to their home and how frequently they consume SSBs. Comprehensive, administrative food outlet databases were used to geo-locate the SSB vendor closest to the respondents' home intersections. We then used a logistic regression model to estimate the association between the distance to the nearest SSB vendor (overall and by type) and the likelihood of daily SSB consumption. Our results show that proximity to the nearest SSB vendor was not statistically significantly associated with the likelihood of daily SSB consumption, regardless of type of vendor. Our results are robust to alternative model specifications, including replacing the linear minimum distance measure with count of the total number of SSB vendors or presence of a SSB vendor within a buffer around respondents' home intersections. We conclude that there is not a strong relationship between proximity to nearest SSB vendor, or proximity to a specific type of SSB vendor, and frequency of self-reported SSB consumption among fast food consumers in NYC. This suggests that policymakers focus on alternative strategies to curtail SSB consumption, such as improving the within-store food environment or taxing SSBs.