Batterham M, Tapsell L C, Charlton K E
National Institute for Applied Statistics Research Australia, School of Mathematics and Applied Statistics, University of Wollongong, Wollongong, New South Wales, Australia.
School of Medicine/Faculty of Science, Medicine and Health, University of Wollongong, Wollongong, New South Wales, Australia.
Eur J Clin Nutr. 2016 Feb;70(2):207-11. doi: 10.1038/ejcn.2015.45. Epub 2015 Apr 1.
BACKGROUND/OBJECTIVES: Dietary weight loss interventions have heterogeneous outcomes in long-term studies, with many participants regaining part or all of the lost weight. Growth mixture modelling is a novel analytic approach that can be used to identify different trajectories of weight change during a trial rather than focussing on the total amount of weight lost.
SUBJECTS/METHODS: Data were pooled from two 12-month dietary weight loss studies where no significant difference was detected between the treatment and control arms, thus allowing analysis independent of treatment. The data set included 231 subjects (74.5% female), with a mean weight loss of 6.40 kg (4.96). Growth mixture models were used to identify participants with similar trajectories of change in body mass index (BMI).
Three subgroups were identified. A rapid and continuing BMI loss over the study period (rapid, n=53), a rapid initial weight loss in the first 3 months with a slowing rate over the remaining 9 months (maintainers, n=146) and those with an initial loss trajectory, which slowed and began to increase at 9 months (recidivists, n=53). Age (s.d.) and BMI (s.d.) were significantly different between the three groups (rapid 53 years (7), 28.99 kg/m(2) (3.30); maintainers 47 years (9), 30.90 kg/m(2) (2.95); recidivists 44 years (7), 34.84 kg/m(2) (1.92), both P<0.001).
Older subjects with lower BMIs were more likely to have a rapid and continuing weight loss in a 1-year dietary-based weight loss intervention. Different interventional approaches may be necessary for different ages and baseline BMIs and stratification prior to randomisation may be necessary to prevent confounding in weight loss trials.
背景/目的：在长期研究中，饮食减肥干预的结果存在异质性，许多参与者会部分或全部恢复已减掉的体重。生长混合模型是一种新颖的分析方法，可用于识别试验期间体重变化的不同轨迹，而不是关注体重减轻的总量。
对象/方法：数据来自两项为期12个月的饮食减肥研究，在治疗组和对照组之间未检测到显著差异，因此可以独立于治疗进行分析。数据集包括231名受试者（74.5%为女性），平均体重减轻6.40千克（4.96）。使用生长混合模型识别体重指数（BMI）变化轨迹相似的参与者。
识别出三个亚组。在研究期间BMI快速持续下降的（快速下降组，n = 53）；在最初3个月体重快速下降，在剩余9个月速度减缓的（维持组，n = 146）；以及最初体重下降，在9个月时减缓并开始增加的（复发组，n = 53）。三组之间的年龄（标准差）和BMI（标准差）存在显著差异（快速下降组53岁（7），28.99千克/米²（3.30）；维持组47岁（9），30.90千克/米²（2.95）；复发组44岁（7），34.84千克/米²（1.92），P均<0.001）。
较低BMI的老年受试者在为期1年的饮食减肥干预中更有可能快速持续减重。对于不同年龄和基线BMI可能需要不同的干预方法，随机分组前进行分层可能有必要以防止减肥试验中的混杂因素。
