Clinical Trial Evidence to Advance the Science of Cholecystectomy

Abstract

Randomized trials evaluating drugs are frequently performed, both for purposes of initial regulatory approval and, after approval, to examine theutility of thedrug inother clinical settings. No drug can get tomarketwithout being fully vetted.This isnot thecase for surgical procedures. Randomizedtrialsevaluatingsurgical interventions are uncommon, even for someof themost commonlyperformedoperations. In this issueof JAMA, 2 randomizedclinical trials examineoneof themostcommonlyperformedgeneral surgical operation—cholecystectomy for acute cholecystitis. To assess the need for postoperative antibiotics, Regimbeau and colleagues1 conducted an open-label, multicenter, noninferiority, randomized clinical trial involving 414 patientswithmildormoderatecalculouscholecystitis. In thetrial, patients received a regimen of antibiotics before and during surgery. The investigators then compared patients who receivednopostoperative antibioticswithpatientswho continued receiving the same regimen of antibiotics for 5 days after surgery. In an imputed intention-to-treat analysis, the rates of postoperative surgical site infections and remote infections were 17% in thenontreatment group and 15% in the antibiotic group. In the per protocol analysis, the corresponding rates were 13% and 13%. The absence of postoperative antibiotic treatment was not associated with worse outcomes in terms of serious adverse events or length of stay. To determine the best management strategy for suspectedcommonduct stones, Iranmaneshandcolleagues2 randomized100patientswithpossiblecommonduct stones toundergo immediate laparoscopic cholecystectomyorendoscopic commonduct evaluationpreoperatively. Patientswhounderwent cholecystectomyfirst comparedwith thosewhohadpreoperative common bile duct evaluation had a significantly shortermedian length of hospital stay ( 5 days vs 8 days), and fewer common bile duct investigations (25 vs 71). There were no significant differences in morbidity or quality of life between the 2 groups. These 2 clinical trials provide important new evidence to better inform surgeons performing cholecystectomy; however, both studieshave a frequently encountered limitationof surgical trials—lack of blinding. Inmany surgical trials, blinding is not possible, or in some cases, as with sham procedures, blinding may raise important ethical considerations. Blinding in randomized trials avoids physician and patient perceptions of efficacy from influencingprotocol adherence or outcome assessment.3,4 Blinding is usually associated with random treatment assignment, and the treating physician, thepatient, and theoutcomeassessor donot know what intervention thepatient received. Blinding and randomization minimize the risk of conscious and unconscious bias in clinical trials (performance bias) and interpretation of outcomes (ascertainment bias). Empirical evidence suggests that lack of blindingmay inflate treatmenteffect estimates.5-7 Similarly,whennotblinded, physiciansandotherhealthcareprofessionalsmaydeliver care differently to the groups (performance bias). For example, if a treatingphysician is aware that a certainmedication is being given for infections andbelieves thatmedication is less effective, heor shemaymonitor thepatientmore frequently byobtainingwhite blood cell counts and imaging studies. Furthermore, lack of blinding of outcome assessors can result in systematic differences in outcome assessment (ascertainment bias) for similar reasons. Although the postoperative antibiotic trial by Regimbeau et alwasnotblinded, the investigators took steps tohelpminimizebias. Patientswere randomized inablinded fashionprior to operation and definition of pathology. This limited the biasing that could occur if the surgeon believed any particular patientwould receive inadequate treatment ifonlygiven1dose of antibiotics, or, conversely, if the surgeon believed a patient wouldbe inappropriately treatedwith 5daysof antibiotics. To assess potential bias in patient enrollment (resulting in a selected, not general, patient population), the reasons for patients not participating in the studywere recorded. However, this information was limited because reasons for noninclusion were only recorded at 1 site. Outcomeassessments alsowerenotblinded.Toassess the validity of the primary outcome assessment, 2 independent surgeons not involved in the study and blinded to the randomization groups reviewed40 sets (10%) of themedical records. This relatively small sample would only identify about 4 treatment failures. Ideally, all of the outcome assessments should have been evaluated this way. Despite these limitations, the study findings support not administering postoperative antibiotic therapy to cholecystectomy patients with mild to moderate clinical severity of illness and is an important addition to the evidence base for future guidelines regarding postoperative antibiotic use in the absence of established infections. As illustrated in the study by Iranmanesh et al blinding in trials investigating invasive treatments iscumbersomeandmay notbepossible. In this clinical trial, surgeons caring for thepaRelated articles pages 137 and 145 Opinion