评估死亡率作为重症期间营养干预随机对照研究主要结局事件率与增量提升：系统回顾

　　在对重症患者的营养干预效果开展随机对照研究时，死亡率是最常用的终点指标，检验效能可根据事件预估的基础发生率和预计效应量两方面进行估计。

　　澳大利亚阿德莱德皇家医院、路易斯维尔大学、国家健康医学研究理事会回顾了19篇探究重症患者营养干预效果的文献，发现在这些随机对照研究中，普遍存在效应量的高估，导致检验效能降低，从而导致阴性结果。因此，可通过适当降低预估效应量，提高检验效能，但需增加样本量。

JPEN J Parenter Enteral Nutr. 2016;40:134.

Event Rate and Delta Inflation When Evaluating Mortality as a Primary Outcome From Randomized Controlled Trials of Nutrition Interventions During Critical Illness: A Systematic Review.

Matthew J. Summers; Lee-anne S. Chapple; Stephen McClave; Adam Deane.

Royal Adelaide Hospital, Adelaide, Australia; University of Adelaide, Adelaide, Australia; University of Louisville, Louisville, KY, USA; Centre for Research Excellence in Translating Nutritional Science to Good Health, National Health and Medical Research Council of Australia, Adelaide, Australia.

Purpose: There is a lack of high-quality evidence to inform nutrition practices for critically ill patients. Accordingly, current guidelines for the provision of nutrition therapy in the critically ill include predominately lower grades of recommendations (B to E). Mortality is frequently chosen as the primary outcome for randomized controlled trials of nutrition therapy in the critically ill. However, a grade A recommendation requires that a statistically significant difference is observed with respect to the primary outcomes, and randomized controlled trials of nutrition interventions may be underpowered to detect a statistical difference in mortality, but this concept has been somewhat neglected. Power calculations use estimated baseline event rate (ie, mortality during control group) and predicted delta (ie, predicted difference between mortality during control and intervention) from previous data, which may be either inadvertently or systematically overestimated. Such overestimations reduce power of the study and increase the risk of a false-negative result. The objective of this review was to evaluate whether power calculations for studies of nutrition interventions that had mortality as the primary end point were realistic and, if overestimation was systematic in the studies identified, to determine if this was due to event rate inflation or delta inflation.

Methods: A systematic review of the literature was performed to identify all randomized controlled trials (published between January 1, 2005, to June 30, 2015, in any of 19 prominent peer-reviewed journals selected because of their focus on general medicine, critical care, or nutrition) of a nutrition intervention administered to critically ill adults that had mortality as the primary outcome. Two reviewers independently extracted data utilizing a standardized form. Predicted event rate (predicted mortality during control), predicted mortality during intervention, predicted delta (predicted difference between mortality during control and intervention), actual event rate (observed mortality during control), observed mortality during intervention, and actual delta (difference between observed mortality during control and intervention) were recorded. The event rate gap (predicted event rate minus observed event rate) and delta gap (predicted delta minus observed delta) and the predicted number needed to treat were calculated. Event rate inflation and delta inflation were defined as positive event rate gap and delta gap, respectively, that would lead to an underpowered study. Summary statistics are presented as median (range).

Results: Fourteen manuscripts were extracted, and of these a power calculation for mortality was provided for 10 studies. The predicted event rate was 29.9% (20.0%-52.4%), and the predicted mortality during intervention was 21.5% (12.5%-32.4%) with a predicted delta of 7.9% (3.0%-20.0). If the study hypothesis was proven correct, then based on the power calculations, the predicted number needed to treat would have been 12.7 (5.0-33.3) patients. The actual event rate was 25.3% (6.1%-50.0%), and the observed mortality during intervention was 24.4% (6.3%-39.7%), with an actual delta 0.5% (-10.2% to 10.3%), such that the event rate gap was 2.6% (-3.9% to 23.7%) and delta gap was 7.5% (3.2%-25.2%).

Conclusions: Delta inflation occurs frequently in randomized controlled trials of nutrition interventions in the critically ill that are powered to determine a mortality benefit, whereas the predicated event rate was closer to the actual event rate. It appears that delta inflation contributes considerably to the number of “negative” studies in this field of research. Future studies should be powered according to a smaller delta, but this will increase sample sizes.

Financial support: A.M.D. is supported by a National Health and Medical Research Council Early Career Fellowship.