School Meals and BMI Percentile: Secondary Analyses of Non-Self-Report Data

描述（由申请人提供）：人们担心参加食品援助计划可能与肥胖有关，因为许多肥胖者都接受了这些计划的服务。2004年3月召开的一个专家小组会议发现，儿童肥胖与参加学校早餐计划（SBP）和国家学校午餐计划（NSLP）之间没有一致的联系。该小组建议的调查方法包括：1)利用自然机会进行对照试验；2)在现有研究基础上增加体重、身高和项目参与；3)针对特定人群开展小规模研究；4)使用现有的数据集来理解关系。方法4的数据集可从R01 HL063189获得，该数据于2005年8月结束。一项小型研究发表的结果表明，高bmi（第85百分位数）的四年级儿童在学校膳食中观察到的（但不是自我报告的）能量摄入高于低bmi（第15百分位数和<50百分位数）。另一项研究的初步数据分析未发表的结果提供了进一步的证据，表明学校膳食中观察到的能量摄入因四年级儿童的BMI百分位数而异，但当分析自我报告的能量摄入时，结果却有所不同。我们现在提出一个为期2年的项目，使用R01 HL063189的4项研究的4年级儿童数据。目的1是调查收缩压和NSLP参与与儿童肥胖之间的可能关系。我们有四年级1,542名儿童（51%黑人，49%白人，51%女孩）参加SBP和NSLP的姓名标签记录（平均随机10至34天）。目的1有两个研究问题（rq），分别针对这4项研究和合并研究：RQ1 -四年级BMI百分位数与收缩压和NSLP参与之间的关系是什么，这是否不同a)收缩压和NSLP参与单独与合并，b)性别，c)种族？RQ2 -与姓名标签记录相比，父母对儿童收缩压和/或NSLP参与同意书的回答有多准确？这是否因BMI百分比、性别和种族而有所不同？当使用父母反应时，RQ1有何不同？目的二是检查一部分儿童的行为和选择，以了解实际（观察到的）校餐摄入量的差异。对于这个亚组，我们观察到344名儿童（54%黑人，46%白人，50%女孩）吃1到3顿早餐，1到3顿午餐，共计1268顿学校餐（50%早餐）。对于这一子集，Aim 2为这4项研究单独和联合提供了2个rq: RQ3 - BMI百分位数与学校膳食中观察到的能量摄入之间的关系是什么，这是否不同？ a)早餐和午餐分别与联合，b)性别，c)种族？RQ4 - RQ3差异在多大程度上可以由以下因素解释：a)标准化份量的食用量，b)贸易，c)牛奶口味/脂肪的选择，d)冷（即即食谷物）与热（如香肠饼干）早餐的选择，以及e)“点单”冰淇淋？结果将为未来的对照试验和/或队列研究的设计提供信息，以了解参与SBP和NSLP的儿童肥胖的途径。公共卫生相关性：参与学校早餐计划（SBP）和国家学校午餐计划（NSLP）可能与儿童肥胖有关，但先前的研究无法提供一致关系的证据。该项目的二次分析是一个有效和及时的下一步，因为结果将增加新的知识，并为未来旨在了解参与SBP和NSLP的儿童肥胖途径的对照试验和/或队列研究的设计提供信息。我们现有的4个数据集的特点使我们能够使用非自我报告数据来调查身体质量指数（BMI）百分位数与收缩压和NSLP参与之间的关系，以及儿童在学校膳食中实际（观察到的）能量摄入的差异，BMI百分位数以及观察到的行为和学校膳食中食物/饮料选择的各个方面包括：对1,542名儿童平均随机抽取10至34天的SBP和NSLP参与记录；观察344名儿童在学校膳食中的摄入量（从观察者间信度得出可接受的结果）；1542名儿童的体重/身高测量（测量者之间的可靠性结果可接受）。

DESCRIPTION (provided by applicant): There is concern that participation in food assistance programs may be related to obesity because many obese individuals are served by such programs. An Expert Panel convened in 03/2004 found no consistent evidence of association between childhood obesity and participation in the School Breakfast Program (SBP) and National School Lunch Program (NSLP). That Panel recommended investigative approaches including 1) use natural opportunities for controlled trials; 2) add weight, height, and program participation to existing studies; 3) initiate smaller studies with specific populations; and 4) use existing data sets to understand relationships. Data sets for approach 4 are available from R01 HL063189, which ended in 08/2005. Published results from a small study indicated that observed (but not self-reported) energy intake at school meals was higher for 4th-grade children of high-BMI (e85th percentile) than low-BMI (e5th and <50th percentiles). Unpublished results from preliminary analyses of data from another study provide further evidence that observed energy intake at school meals varies by 4th-grade children's BMI percentile, but results differed when self-reported energy intake was analyzed. We now propose a 2-year project using data for 4th-grade children from 4 studies for R01 HL063189. Aim 1 is to investigate a possible relationship between SBP and NSLP participation and childhood obesity. We have nametag records (compiled on an average of 10 to 34 random days) of SBP and NSLP participation in 4th-grade for 1,542 children total (51% Black, 49% White; 51% girls). Aim 1 has 2 research questions (RQs) for the 4 studies separately and combined: RQ1 - What is the relationship between BMI percentile and SBP and NSLP participation in 4th grade, and does this differ a) for SBP and NSLP participation separately vs. combined, b) by sex, and c) by race? RQ2 - How accurate are parental responses on consent forms concerning children's SBP and/or NSLP participation compared to nametag records, and does this differ by BMI percentile, sex, and race? How does RQ1 differ when parental responses are used? Aim 2 is to examine behaviors and selections for a subset of children to understand differences in actual (observed) school-meal intake. For the subset, we observed 344 children (54% Black, 46% White; 50% girls) for 1 to 3 breakfasts and 1 to 3 lunches for 1,268 total school meals (50% breakfast). For the subset, Aim 2 has 2 RQs for the 4 studies separately and combined: RQ3 - What is the relationship between BMI percentile and observed energy intake at school meals, and does this differ a) for breakfast and lunch separately vs. combined, b) by sex, and c) by race? RQ4 - To what extent are RQ3 differences explained by a) amounts eaten of standardized portions, b) trades, c) milk flavor/fat selections, d) cold (i.e., ready-to-eat cereal) vs. hot (e.g., sausage biscuit) breakfast selection, and e) ` la carte ice cream? Results will inform the design of future controlled trials and/or cohort studies to understand pathways to obesity among children who participate in SBP and NSLP. PUBLIC HEALTH RELEVANCE: Participation in the School Breakfast Program (SBP) and National School Lunch Program (NSLP) may be related to childhood obesity, but previous research has been unable to provide evidence of a consistent relationship. The secondary analyses in this project are an efficient and timely next step because results will add new knowledge and inform the design of future controlled trials and/or cohort studies aimed at understanding pathways to obesity among children who participate in SBP and NSLP. Features of our 4 existing data sets that uniquely position us to use non-self-report data to investigate the relationship between body mass index (BMI) percentile and SBP and NSLP participation as well as differences in children's actual (observed) energy intake at school meals by BMI percentile along with various aspects of observed behaviors and food/beverage selections at school meals include the following: nametag records of SBP and NSLP participation on an average of 10 to 34 random days for each of 1,542 children; observed intake at school meals for a subset of 344 children (with acceptable results from inter- observer reliability); and weight/height measurements for 1,542 children (with acceptable results from inter- measurer reliability).