Reduced-Item Food Audits Based on the Nutrition Environment Measures Surveys.

INTRODUCTION: The community food environment may contribute to obesity by influencing food choice. Store and restaurant audits are increasingly common methods for assessing food environments, but are time consuming and costly. A valid, reliable brief measurement tool is needed. The purpose of this study was to develop and validate reduced-item food environment audit tools for stores and restaurants. METHODS: Nutrition Environment Measures Surveys for stores (NEMS-S) and restaurants (NEMS-R) were completed in 820 stores and 1,795 restaurants in West Virginia, San Diego, and Seattle. Data mining techniques (correlation-based feature selection and linear regression) were used to identify survey items highly correlated to total survey scores and produce reduced-item audit tools that were subsequently validated against full NEMS surveys. Regression coefficients were used as weights that were applied to reduced-item tool items to generate comparable scores to full NEMS surveys. Data were collected and analyzed in 2008-2013. RESULTS: The reduced-item tools included eight items for grocery, ten for convenience, seven for variety, and five for other stores; and 16 items for sit-down, 14 for fast casual, 19 for fast food, and 13 for specialty restaurants-10% of the full NEMS-S and 25% of the full NEMS-R. There were no significant differences in median scores for varying types of retail food outlets when compared to the full survey scores. Median in-store audit time was reduced 25%-50%. CONCLUSIONS: Reduced-item audit tools can reduce the burden and complexity of large-scale or repeated assessments of the retail food environment without compromising measurement quality.

Optimizing data collection for public health decisions: a data mining approach.

BACKGROUND: Collecting data can be cumbersome and expensive. Lack of relevant, accurate and timely data for research to inform policy may negatively impact public health. The aim of this study was to test if the careful removal of items from two community nutrition surveys guided by a data mining technique called feature selection, can (a) identify a reduced dataset, while (b) not damaging the signal inside that data. METHODS: The Nutrition Environment Measures Surveys for stores (NEMS-S) and restaurants (NEMS-R) were completed on 885 retail food outlets in two counties in West Virginia between May and November of 2011. A reduced dataset was identified for each outlet type using feature selection. Coefficients from linear regression modeling were used to weight items in the reduced datasets. Weighted item values were summed with the error term to compute reduced item survey scores. Scores produced by the full survey were compared to the reduced item scores using a Wilcoxon rank-sum test. RESULTS: Feature selection identified 9 store and 16 restaurant survey items as significant predictors of the score produced from the full survey. The linear regression models built from the reduced feature sets had R2 values of 92% and 94% for restaurant and grocery store data, respectively. CONCLUSIONS: While there are many potentially important variables in any domain, the most useful set may only be a small subset. The use of feature selection in the initial phase of data collection to identify the most influential variables may be a useful tool to greatly reduce the amount of data needed thereby reducing cost.

Second births to teenage mothers: risk factors for low birth weight and preterm birth.

CONTEXT: Teenagers are more likely than older women to have a low-birth-weight infant or a preterm birth, and the risks may be particularly high when they have a second birth. Identifying predictors of these outcomes in second teenage births is essential for developing preventive strategies. METHODS: Birth certificate data for 1993-2002 were linked to identify second births to Milwaukee teenagers. Predictors of having a low-birth-weight second infant or a preterm second birth were identified using logistic regression. RESULTS: The same proportion of first and second infants were low-birth-weight (12%), but second births were more likely than first births to be preterm (15% vs. 12%). In analyses that adjusted for demographic, pregnancy and behavioral characteristics, the odds that a second infant was low-birth-weight or preterm were elevated if the mother smoked during pregnancy (odds ratios, 2.2 and 1.9, respectively), had inadequate prenatal weight gain (1.8 and 1.4), had an interpregnancy interval of less than 18 months (1.6-2.9 and 1.4-2.3) or was black (2.7 and 1.7). Women who had received an adequate level of prenatal care had reduced odds of both outcomes (0.6 and 0.4). Women younger than 16 also had increased odds of having a low-birth-weight second infant. Further adjustment for socioeconomic characteristics yielded largely the same results. In addition, women who were unmarried or did not identify a father were at increased risk of both outcomes (1.5 for each), and poor women were at risk of having a low-birth-weight infant (1.3). CONCLUSIONS: Predictors of poor birth outcomes include modifiable behaviors. Prenatal interventions addressing these behaviors could help improve outcomes.

Prevalence of children with special health care needs in Milwaukee, Wisconsin: data from the Milwaukee metropolitan statistical area.

BACKGROUND: The Milwaukee Health Department and the Wisconsin Southeast Regional Center for Children with Special Health Care Needs (CSHCN) were interested in understanding the level of need and prevalence of CSHCN in the city of Milwaukee. It was determined that a survey of the Milwaukee area was needed to obtain a prevalence estimate. METHODS: A survey to identify children with special health care needs in the Milwaukee metropolitan statistical area (MSA) was conducted using the CSHCN Screener. The survey was administered as part of the Greater Milwaukee Survey in October and November of 2004. Results from this survey were compared to state and national data from the National Survey of Children with Special Health Care Needs, 2001. RESULTS: The prevalence of CSHCN in Milwaukee (23.5%) was found to be almost double that of the surrounding Milwaukee MSA counties (12.1%), the state (13.4%), and the nation (12.8%). The number and type of positive screener questions were not significantly different among the geographic areas. CONCLUSIONS: The prevalence of CSHCN has been consistently associated with poverty in numerous other studies. The survey conducted in Milwaukee MSA indicated increased prevalence in Milwaukee among poor and African American children. Improvement and augmentation of services available to urban poor is necessary to alleviate this excessive burden.