Department of Pediatrics, Alberta Research Centre for Health Evidence, University of Alberta, Edmonton, Canada.
Department of Biostatistics, Julius Centre for Health Sciences and Primary Care, University Medical Centre, Utrecht, Netherlands.
BMC Med Res Methodol. 2018 Jan 10;18(1):7. doi: 10.1186/s12874-017-0456-8.
We wished to compare the nuisance parameters of pediatric vs. adult randomized-trials (RCTs) and determine if the latter can be used in sample size computations of the former.
In this meta-epidemiologic empirical evaluation we examined meta-analyses from the Cochrane Database of Systematic-Reviews, with at least one pediatric-RCT and at least one adult-RCT. Within each meta-analysis of binary efficacy-outcomes, we calculated the pooled-control-group event-rate (CER) across separately all pediatric and adult-trials, using random-effect models and subsequently calculated the control-group event-rate risk-ratio (CER-RR) of the pooled-pediatric-CERs vs. adult-CERs. Within each meta-analysis with continuous outcomes we calculated the pooled-control-group effect standard deviation (CE-SD) across separately all pediatric and adult-trials and subsequently calculated the CE-SD-ratio of the pooled-pediatric-CE-SDs vs. adult-CE-SDs. We then calculated across all meta-analyses the pooled-CER-RRs and pooled-CE-SD-ratios (primary endpoints) and the pooled-magnitude of effect-sizes of CER-RRs and CE-SD-ratios using REMs. A ratio < 1 indicates that pediatric trials have smaller nuisance parameters than adult trials.
We analyzed 208 meta-analyses (135 for binary-outcomes, 73 for continuous-outcomes). For binary outcomes, pediatric-RCTs had on average 10% smaller CERs than adult-RCTs (summary-CE-RR: 0.90; 95% CI: 0.83, 0.98). For mortality outcomes the summary-CE-RR was 0.48 (95% CIs: 0.31, 0.74). For continuous outcomes, pediatric-RCTs had on average 26% smaller CE-SDs than adult-RCTs (summary-CE-SD-ratio: 0.74).
Clinically relevant differences in nuisance parameters between pediatric and adult trials were detected. These differences have implications for design of future studies. Extrapolation of nuisance parameters for sample-sizes calculations from adult-trials to pediatric-trials should be cautiously done.
我们希望比较儿科和成人随机对照试验(RCT)的干扰参数,并确定后者是否可以用于前者的样本量计算。
在这项荟萃流行病学实证评估中,我们检查了 Cochrane 系统评价数据库中的荟萃分析,其中至少有一项儿科 RCT 和至少一项成人 RCT。在每个二元疗效结局的荟萃分析中,我们使用随机效应模型分别计算了所有儿科和成人试验的汇总对照组事件率(CER),并随后计算了汇总儿科 CER 与成人 CER 的对照组事件率风险比(CER-RR)。在每个连续结局的荟萃分析中,我们分别计算了所有儿科和成人试验的汇总对照组效应标准差(CE-SD),并随后计算了汇总儿科 CE-SD 与成人 CE-SD 的 CE-SD 比值。然后,我们使用 REM 计算了所有荟萃分析的汇总 CER-RR 和汇总 CE-SD 比值(主要终点)以及 CER-RR 和 CE-SD 比值的效应大小的汇总幅度。比值<1 表示儿科试验的干扰参数小于成人试验。
我们分析了 208 项荟萃分析(135 项用于二元结局,73 项用于连续结局)。对于二元结局,儿科 RCT 的 CER 平均比成人 RCT 小 10%(汇总 CE-RR:0.90;95%CI:0.83,0.98)。对于死亡率结局,汇总 CE-RR 为 0.48(95%CI:0.31,0.74)。对于连续结局,儿科 RCT 的 CE-SD 平均比成人 RCT 小 26%(汇总 CE-SD 比值:0.74)。
在儿科和成人试验的干扰参数之间检测到了临床相关的差异。这些差异对未来研究的设计有影响。从成人试验外推到儿科试验的样本量计算的干扰参数应谨慎进行。