Procter Kimberley L, Rudolf Mary C, Feltbower Richard G, Levine Ronnie, Connor Anne, Robinson Michael, Clarke Graham P
The University of Leeds, Leeds, United Kingdom.
Soc Sci Med. 2008 Jul;67(2):341-9. doi: 10.1016/j.socscimed.2008.02.029. Epub 2008 Apr 24.
This article explores the impact that schools have on their pupils' obesity and so identify those where targeted input is most needed. A modelling process was developed using data that had been collected over 2 years on a socio-economically and ethnically representative sample of 2367 school pupils aged 5 and 9 years old attending 35 Leeds primary schools. The three steps in the model involved calculating the "Observed" level of obesity for each school using mean body mass index standard deviation (BMI SDS); adjusting this using ethnicity and census-derived deprivation data to calculate the "Expected" level; and calculating the "Value Added" by each school from differences in obesity at school entry and transfer. We found there was significant variance between the schools in terms of mean BMI SDS (range -0.07 to +0.78). Residential deprivation score and ethnicity accounted for only a small proportion of the variation. Expected levels of obesity therefore differed little from the Observed, but the Value Added step produced very different rankings. As such, there is variation between schools in terms of their levels of obesity. Our modelling process allowed us to identify schools whose levels differed from that expected given the socio-demographic make up of the pupils attending. The Value Added step suggests that there may be a significant school effect. If this is validated in extended studies, the methodology could allow for exploration of mechanisms contributing to the school effect, and identify schools with the highest unexpected prevalence. Resources could then be targeted towards those schools in greatest need.
本文探讨了学校对学生肥胖问题的影响，从而确定那些最需要针对性投入的学校。利用在两年时间里收集的关于利兹市35所小学中2367名5至9岁学生的社会经济和种族代表性样本数据，开发了一个建模过程。该模型的三个步骤包括：使用平均体重指数标准差（BMI SDS）计算每所学校的“观察到的”肥胖水平；利用种族和人口普查得出的贫困数据对其进行调整，以计算“预期的”水平；根据入学和转学阶段肥胖情况的差异计算每所学校的“附加值”。我们发现，各学校之间的平均BMI SDS存在显著差异（范围为-0.07至+0.78）。居住贫困得分和种族仅占差异的一小部分。因此，肥胖的预期水平与观察到的水平差异不大，但附加值步骤产生了非常不同的排名。也就是说，各学校在肥胖水平方面存在差异。我们的建模过程使我们能够识别出那些肥胖水平与根据就读学生的社会人口构成预期的水平不同的学校。附加值步骤表明可能存在显著的学校效应。如果在进一步的研究中得到验证，该方法可以用于探索导致学校效应的机制，并识别出意外患病率最高的学校。然后可以将资源投向那些最需要的学校。
