School of Agricultural Economics and Rural Development, Renmin University of China, Beijing, 100872, China.
Centre for Public Health/UKCRC Centre of Excellence for Public Health (NI), Queen's University Belfast, Belfast, UK.
Int J Obes (Lond). 2019 Jun;43(6):1135-1146. doi: 10.1038/s41366-018-0265-0. Epub 2018 Nov 27.
Previous studies have documented that high rates of delay discounting are associated with obesity. However, studies utilizing monetary reward experiments typically report no associations, as opposed to positive associations apparent in studies utilising food-reward experiments. Our objective was to investigate the reasons behind the mixed evidence from a methodological perspective using systematic review and meta-analytic methodologies.
Seven databases (EMBASE, MEDLINE, PsycINFO, Scopus, Web of Science, Econlit and IBSS) were systematically searched. Logistic meta-regression was applied to identify the determinants of a significant association and risk of bias was assessed using a modified form of the Newcastle Ottawa cohort scale.
A total of 59 studies were identified, among which 29 studies (49.2%) found a significant positive association and 29 (49.2%) reported no association. A higher proportion of significant and positive associations was reported in those studies utilizing 'best-practice' methods (i.e. appropriate measurement models) to estimate monetary delay discounting (15/27; 55.6%) and incentive-compatible experiments (10/16; 62.5%) than those using non-'best-practice' methods (14/34; 41.2%) and hypothetical experiments (19/43; 44.2%). All five studies utilizing both 'best-practice' methods and incentive-compatible experiments generated a positive and significant relationship. Results from a logistic meta-regression also suggested that studies employing incentive-compatible experiments (OR: 4.38, 95% CI = 1.05-18.33, p value: 0.04), 'best-practice' methods (OR: 4.40, 95% CI = 0.88-22.99, p value: 0.07), parametric methods (OR: 3.36, 95% CI = 0.83-13.57, p value: 0.04), those conducted in children/adolescent populations (OR: 3.90, 95% CI = 0.85-17.88, p value: 0.08), and those with larger sample size (OR: 1.91, 95% CI = 1.15-3.18, p value: 0.01) tended to show positive and significant associations between delay discounting and obesity.
This review suggests that the mixed evidence to date is a result of methodological heterogeneity, and that future studies should utilise 'best practice' methods.
先前的研究已经证明，较高的延迟折扣率与肥胖有关。然而，利用货币奖励实验的研究通常报告没有关联，而利用食物奖励实验的研究则显示出明显的正相关。我们的目的是从方法学的角度利用系统评价和荟萃分析方法来探讨混合证据的原因。
系统地搜索了七个数据库（EMBASE、MEDLINE、PsycINFO、Scopus、Web of Science、Econlit 和 IBSS）。应用逻辑回归来确定显著关联的决定因素，并使用修改后的纽卡斯尔-渥太华队列量表评估偏倚风险。
共确定了 59 项研究，其中 29 项（49.2%）发现了显著的正相关，29 项（49.2%）报告没有关联。在利用“最佳实践”方法（即适当的测量模型）估计货币延迟折扣的研究中，报告显著和阳性关联的比例更高（15/27；55.6%），以及使用激励相容实验（10/16；62.5%），而不是使用非“最佳实践”方法（14/34；41.2%）和假设实验（19/43；44.2%）。利用“最佳实践”方法和激励相容实验的所有五项研究均产生了阳性和显著的关系。逻辑回归的结果还表明，使用激励相容实验的研究（OR：4.38，95%CI=1.05-18.33，p 值：0.04）、“最佳实践”方法（OR：4.40，95%CI=0.88-22.99，p 值：0.07）、参数方法（OR：3.36，95%CI=0.83-13.57，p 值：0.04）、在儿童/青少年人群中进行的研究（OR：3.90，95%CI=0.85-17.88，p 值：0.08）和具有较大样本量的研究（OR：1.91，95%CI=1.15-3.18，p 值：0.01）往往显示出延迟折扣与肥胖之间的阳性和显著关联。
本综述表明，迄今为止的混合证据是方法学异质性的结果，未来的研究应利用“最佳实践”方法。
