Stop the Presses (and the Data Collection)! Aggressive Fluids Harm Patients With Acute Pancreatitis: May 2023 Annals of Emergency Medicine Journal Club.

de-Madaria E, Buxbaum JL, Maisonneuve P, et al. Aggressive or moderate fluid resuscitation in acute pancreatitis (WATERFALL Study). N Engl J Med. 2022;387:989-1000. In patients with acute, mild pancreatitis, does the use of moderate fluid resuscitation result in a lower rate of progression to moderate/severe pancreatitis in comparison with aggressive fluid resuscitation? Design: Multicenter, multinational, open-label, parallel-group, randomized, controlled, superiority trial. Setting: Eighteen acute care hospitals in India, Italy, Mexico, and Spain. Population: Two hundred forty-nine adult patients with acute pancreatitis. Intervention: Moderate bolus and continuous fluid resuscitation with lactated ringer’s solution at 1.5 cc/kg per hour as compared with an aggressive bolus and continuous resuscitation at 3.0 cc/kg per hour. Primary and Secondary Outcome: The primary outcome was progression to moderately severe or severe acute pancreatitis. Secondary outcomes included organ failure, local complications, persistent organ failure, respiratory failure, hospital length of stay, ICU admission, and ICU length of stay. Sponsors: Supported by grants from Instituto de Salud Carlos III, the Spanish Association of Gastroenterology, and Instituto de Investigación Sanitaria y Biomédica de Alicante (ISABIAL). ClinicalTrials.gov: NCT04381169. Among the 249 patients included in the intention-to-treat analysis, progression was seen in 22.1% of patients in the aggressive fluid group and 17.3% of patients in the moderate fluid group (adjusted Relative Risk [ARR], 1.30; 95% confidence interval (CI), 0.78-2.18).The secondary endpoints favored the moderate fluid group, but none reached statistical significance. However, the study was terminated early because of a prespecified stopping rule for safety owing to higher risk of fluid overload in the aggressive fluid resuscitation group: 20.5% versus 6.3% (aRR 2.85; 95% CI, 1.36-5.94). The authors conclude that aggressive fluid administration in mild pancreatitis resulted in a higher incidence of fluid overload without a corresponding signal of clinical benefit. This high-quality study should reinforce practice migration toward reduced fluid administration in the treatment of acute pancreatitis. Although moderate fluid administration was not found to be superior to aggressive fluid administration, moderate fluid administration is more likely beneficial than not. The increased risk of fluid overload, although a subjective measurement in this trial, warrants recognition. An individualized approach to volume status correction, if necessary, is likely appropriate. 1.This trial was stopped prior to full data collection. What are the reasons for stopping a trial early? Studies may be appropriately halted prior to completion because of safety or ethical concerns. These decisions should ideally be made by independent trial data and safety monitoring boards (DSMBs), not the trial investigators, who are an essential component of most clinical trials. DSMBs are comprised of experts who independently and objectively review patient safety data, trial integrity measures, and outcomes data to assure the safety of trial participants. DSMBs should be staffed prior to trial initiation and typically have authority to recommend continuation, pausing, or early cessation of a trial. If a study reveals unintended harms related to one arm or the other, it may be unethical to continue enrolling patients when equipoise no longer exists. Alternatively, a significant benefit may be detected, raising the argument of the ethics of withholding an effective treatment or quickly disseminating evidence of a treatment’s benefits to the wider community. However, data collection is often inappropriately stopped outside of pre-established stopping points. This often occurs in industry-funded research, in which finding a benefit carries financial incentives.1Lièvre M. Ménard J. Bruckert E. et al.Premature discontinuation of clinical trial for reasons not related to efficacy, safety, or feasibility.BMJ. 2001; 322: 603-605Crossref PubMed Google Scholar, 2Psaty B.M. Rennie D. Stopping medical research to save money: a broken pact with researchers and patients.JAMA. 2003; 289: 2128-2131Crossref PubMed Scopus (96) Google Scholar, 3Chopra S.S. MSJAMA: industry funding of clinical trials: benefit or bias?.JAMA. 2003; 290: 113-114Crossref PubMed Scopus (79) Google Scholar Without a prespecified stopping point, collection can be halted and data can be analyzed at any point. These analyses may incorporate financial and professional concerns, weakening the reliability and integrity of the research conducted. As such, studies should have unambiguous stopping rules for data collection with preset thresholds. The WATERFALL study was stopped early after a prespecified interim analysis demonstrated harm to the aggressive fluid group. At the time of trial stoppage, there was a 4.8% absolute difference in the primary outcome. Unfortunately, due to the early termination, this observation lacks sufficient precision to rely upon as evidence favoring moderate fluid administration. The further enrollment necessary to demonstrate superiority would unethically expose study participants to a greater risk of fluid overload.2.What systematic effect on the evidence base occurs due to early trial termination? Early termination may affect both the direction of the results and the magnitude of the effect size. Wide CIs around the point estimate for the effect size provide some indication regarding the reliability of the observations. A particular bias occurs regarding novel or unexpectedly large effect sizes, in which early termination leads to publication of an overestimation of the point estimate. However, this benefit often dissipates with full data collection. Regression to the mean is the tendency of extreme results, possibly due to chance, to move closer to the null as a trial progresses and more events occur.4Montori V.M. Devereaux P.J. Adhikari N.K. et al.Randomized trials stopped early for benefit: a systematic review.JAMA. 2005; 294: 2203-2209Crossref PubMed Scopus (578) Google Scholar A review of 143 randomized controlled trials stopped early for benefit found that treatment effects were often implausibly large.4Montori V.M. Devereaux P.J. Adhikari N.K. et al.Randomized trials stopped early for benefit: a systematic review.JAMA. 2005; 294: 2203-2209Crossref PubMed Scopus (578) Google Scholar Small studies are at a higher risk of this overestimation of benefit.5Bassler D. Briel M. Montori V.M. et al.Stopping randomized trials early for benefit and estimation of treatment effects: systematic review and meta-regression analysis.JAMA. 2010; 303: 1180-1187Crossref PubMed Scopus (466) Google Scholar Favorable results increase public attention, resulting in dissemination of incomplete information to the clinical and scientific community.6Briel M. Bassler D. Wang A.T. et al.The dangers of stopping a trial too early.J Bone Joint Surg Am. 2012; 94: 56-60Crossref PubMed Scopus (26) Google Scholar Additionally, investigators may overestimate the efficacy of their results by conducting repeated interim analyses without adjusting the level of significance. Designating a prespecified point for interim analysis in the study design and implementing stricter levels of significance during early interim analysis can help to address these issues.6Briel M. Bassler D. Wang A.T. et al.The dangers of stopping a trial too early.J Bone Joint Surg Am. 2012; 94: 56-60Crossref PubMed Scopus (26) Google Scholar,7Viele K. McGlothlin A. Broglio K. Interpretation of clinical trials that stopped early.JAMA. 2016; 315: 1646-1647Crossref PubMed Scopus (28) Google Scholar The effects of stopping early can be compounded when data are incorporated into systematic reviews and meta-analyses. Prematurely stopped trials have been found to contribute significantly to meta-analyses (>20% of the data) and artificially inflate the conclusions.6Briel M. Bassler D. Wang A.T. et al.The dangers of stopping a trial too early.J Bone Joint Surg Am. 2012; 94: 56-60Crossref PubMed Scopus (26) Google Scholar,8Guyatt G.H. Briel M. Glaszio P. et al.Problems of stopping trials early.BMJ. 2012; 344e3863Crossref Scopus (86) Google Scholar Ending studies prematurely can also have a negative impact on the conduct of future trials, known as the “freezing effect.”9Walter S.D. Han H. Guyatt G.H. et al.A systematic survey of randomised trials that stopped early for reasons of futility.BMC Med Res Methodol. 2020; 20: 10Crossref Scopus (20) Google Scholar This can occur when other studies investigating the same question are stopped because it could be considered unethical to continue considering existing data. This effect can be problematic because this may induce negative publication bias toward other relevant data, delaying its potential influence on clinical practice.