In 1965, theAmericanMedical Association declared that “An intern’s duties and responsibilities are not discharged on a ‘nine-to-five’ basis. While an acceptable internship provides for a reasonable amount of free time, [the intern’s] thought for andcontact with his patients should be on a ‘round-the-clock’ basis.”1 In the intervening 45 years, the advent of thehospitalistmovement fragmented inpatient and outpatient care, payment by diagnosis reduced length of stay, hospitalization rates per capita increased by 15%,2 and more than 1000 new drug applications were approved.3 In short, inpatient care is now more fragmented, more frantic, andmore complicated than in the 1960s. At the same time, the scienceof sleepandcognitionmaturedenough tomaketherisksofworkingwhile fatiguedunequivocallyclear, and the influxofwomen intomedical trainingmade it increasingly difficult to sustain the fiction that house staff have no obligations outside the hospital. The medical establishment reacted accordingly by restricting work hours of house staff. Thus, in 2013 not only is the “round the clock” internship a relic of thepast, but inEurope internships are approaching the “nine to five” standard,withwork-hours restricted to48hours a week.4 Despite an anticipated reduction in fatigue-related errors,work-hour regulationshavebeenassociatedwith little to no improvement inpatientoutcomeseither in theUnitedStates or in Europe.5 Many commentators ascribe this disappointing outcome to an offsetting increase in discontinuity and in handoff-related errors. Numerous anecdotal and systematic studies have demonstrated that handoffs are associatedwith medical errors. There are also numerous reports of programs andtools to increase thesafetyandeffectivenessofhandoffs.6-8 To date, however, almost no evidence suggests that improvements in handoffs reduce the rate of subsequent errors.6 The report by Starmer et al9 in this issue of JAMA is by far the most comprehensive study of the direct effects of handoff interventions on outcomes within the context of existing work-hour regulations and is the first to demonstrate an associated significantdecrease inmedical errors ona large scale. Theauthorsconductedanuncontrolled,before-and-afterstudy of the effectiveness of a multifaceted “handoff bundle” on medical errors and preventable adverse events. They studied 642 admissions cared for by42house staff on 2pediatric units at a single hospital fromJuly throughSeptember 2009, implemented ahandoff bundle inOctober, and then studied613 admissions cared for by 42 house staff on the same units from November 2009 through January 2010. The authors found a decline inmedical errors from 33.8 per 100 admissions in the preinterventionperiod to 18.3 per 100 admissions in thepostintervention period (P < .001). Although far from a gold-standard design, this study has numerousstrengths.Thehandoffbundlewasmultifacetedand includedteamtraining (which isnotusuallyconsideredahandoff intervention per se but has been shown to improve outcomes10), structural changes in teamorganizationandelectronic documentation (which have been shown to improve handoff quality8,11), anddidactic training inhandoff standardization. Adverse events were identified prospectively in real time andwere later characterized according to standard definitions. The authors assessed a variety of important outcomesbeyondadverse events, includingdirect observationof time spent with patients and assessment of written handoff materials. Among the most provocative and intriguing findings of the study is that time spentwith patients significantly increased in the postintervention period, even though time spent creating computerized handoff documents and conducting verbal handoffs did not change. Nevertheless, results of this study should be considered preliminary because of numerous limitations. First, the authors used an uncontrolled, before-and-after design, making the results vulnerable to the effects of the steep early year intern learning curve; changes inpatient demographics, comorbidity, and diagnoses over time; the Hawthorne effect of observation;andotherquality improvement initiatives.Moreover, thenurses collecting adverse event data in real time couldnot be blinded to preintervention or postintervention period and were thereforeanother importantpotential sourceofbias.Even thoughall eventswereadjudicatedbyblindedphysicians, their agreementoncategorizationandpreventabilitywasonlymoderate and they were reliant on the study nurses to bring adverse events to their attention. The brief and reasonably light touchnature of the intervention seemsdisproportional to the large effect size (46%reduction inmedical error rate), increasing concern for unmeasured confounding or random variation. The authors did not attempt to characterize events as handoff related or not and largely did notmeasure theuptake of the intervention, such as improved teamwork, use of unified team handoff, supervision by senior clinicians, or quality of verbal handoffs. The follow-up time period of 3months was relatively brief, raising questions as to long-term sustainability of improved outcomes. Moreover, the authors implemented the entire bundle wholesale, making it impossible to determine which, if any, elements were more effective than others. Althoughquality improvement studies aredifficult toperformintruly randomizedfashion,optionsareavailable tomake Viewpoint page 2247