The ladder of evidence in clinical medicine starts with observation and hypothesis. Hypotheses with merit move through a series of clinical trials with increasing scientific rigor, culminating in randomized controlled trials, which, more often than not, fail to support the observations of less-rigorous testing.1 Although we appreciate our colleagues' review of our article,2 there appears to be an implicit assertion that failure to observe the preferred result (that enhanced recovery after surgery would shorten hospitalization) stems from problems with design or execution, rather than a potential flaw in the premise.
Unfortunately, the “trove” of data on enhanced recover after surgery to which our colleagues refer consists almost exclusively of sequential-cohort or case series studies (often retrospective). Given that these designs are particularly prone to observer bias and Hawthorne effect (that might best be revealed in a more formal, prospective, randomized study such as ours), it is unsurprising that, despite the recent Society for Gynecologic Oncology guidelines, the 2015 Cochrane Review concludes “there is no evidence from high-quality studies to support or refute the use of perioperative enhanced recovery programmes for gynaecologic cancer patients.”3
Our study tests the question, does introducing a formal enhanced recovery algorithm on a gynecologic oncology service improve outcomes over current best care, which includes informal use of these elements. Blinded randomization and intention-to-treat analyses make moot the question of surgical complexity and cancer status, because the study population was defined a priori and the arms appear balanced. Similarly, although we did not formally analyze total fluid balance, we observed a reduction in hospitalization from the prestudy baseline in both study arms—suggesting that neither group was impeded by fluid imbalance.
Though including more or different enhanced recovery after surgery elements might have altered our findings, we reject the notion that the choice of elements was somehow suboptimal. Each was selected based on our previous, successful sequential-cohorts trial of abdominal hysterectomies, which is still among the largest series published.4 Furthermore, the current literature includes markedly conflicting data on which elements are essential to reducing hospitalization, extensive but unvalidated extrapolation from the colorectal surgery literature, and virtually no head-to-head, contemporaneous comparisons of different enhanced recovery after surgery strategies to support the conclusion that one is optimal or even superior to any other.
Although the enhanced recovery after surgery philosophy remains attractive, we look forward to further high-quality studies that demonstrate efficacy.