American Journal of Respiratory and Critical Care Medicine

Rationale: Cross-coverage is associated with medical errors caused by miscommunication during handoffs. However, no direct evidence links handoffs to outcomes, or explains the mechanisms leading to outcomes. Furthermore, the previous literature may overestimate the impact of handoffs because of hindsight bias.

Objectives: To explore the effects of nighttime cross-coverage on mortality and decision making in critically ill patients.

Methods: Observational cohort of 629 consecutive critically ill admissions, admitted for at least 48 hours, and critical care fellows in an academic hospital.

Measurements and Main Results: Intensive care unit (ICU) mortality and nighttime decisions. Our exposure variable was cross-covering status of fellows. We observed a decrease in ICU mortality (odds ratio, 0.77 per 1 d; 0.60–0.99; P = 0.04), a higher number of nighttime decisions (19.3 vs. 10.4%; odds ratio, 2.02; 95% confidence interval [CI], 1.03–3.95; P = 0.04), an increase in fentanyl equivalents administered to patients at night (difference, +10.2 μg/h; 95% CI, +1.4 to +19.0; P = 0.02), and an increase in transfusions at night (difference, +465 ml; 95% CI, +98 to +832; P = 0.01) when fellows were cross-covering.

Conclusions: In this single-center study exposure to cross-covering fellows was associated with a decrease in ICU mortality and with more nighttime decisions. Our findings contradict the dominant hypothesis that cross-coverage is associated with worse outcomes, and suggest that a “second look” by cross-covering fellows may mitigate cognitive errors. Future interventions to improve patient safety in ICUs should focus both on the quality of handoffs and on strategies to decrease cognitive errors.

Scientific Knowledge on the Subject

Nighttime coverage is associated with medical errors in a variety of healthcare settings. However, these data are limited by hindsight bias.

What This Study Adds to the Field

This is the first study to observe an association between greater nighttime cross-coverage and decreased intensive care unit mortality. The findings challenge the concept that cross-coverage is harmful, and suggest that a “second look” may be beneficial in mitigating cognitive errors.

Handoffs and cross-coverage are essential elements of modern critical care. A handoff encompasses the exchange of information between clinicians to facilitate continuity of care. When handoffs are used to facilitate continuity for a limited period of time, as when clinicians take over care for only the nighttime, the nighttime clinician is providing cross-coverage (1). The number of handoffs has increased considerably because of changes in work hour regulations for residents (2), which makes these handoffs an important target for quality improvement interventions. Strategies to increase continuity of care include changing trainee’s scheduling (3, 4) and improving the handoff process (5). However, to date the only interventions on handoffs shown to decrease medical errors are the use of an electronic system (6, 7) and handoff bundles (8). Currently, no data exist investigating the association of cross-coverage and mortality.

Cross-coverage is associated with an increased risk of medical errors (9) and malpractice claims (10). These errors are related to incomplete or incorrect clinical information relayed to the cross-covering clinician (11). Trainees interviewed after their on-call period commonly report “surprises” or unexpected changes (12), and do not feel adequately prepared for 80% of nighttime events, 75% of which can be anticipated during handoffs (13). Although a large body of medical literature suggests that handoffs are a potential threat to safety, most of the data are subject to hindsight bias (14), because the perception that a handoff leads to an error is biased by identification of gaps in communication after the fact.

Therefore, more compelling and unbiased evidence linking handoffs to patient-centered outcomes, and demonstrating how decision making differs at night, is required to advance knowledge in this field. We hypothesized that increased exposure to cross-covering fellows is associated with worse outcomes and designed this study to investigate the effects of nighttime cross-coverage on mortality and nighttime decision making in critically ill patients.

Study Design and Setting

Our study comprised an observational cohort of critically ill patients in a 20-bed intensive care unit (ICU) in an academic hospital in Ontario, Canada. This ICU works with a closed-model of physician staffing, and with two separate medical teams providing care for patients. Each medical team is composed of one staff intensivist, one critical care fellow (a trainee in critical care who has finished training in anesthesia, surgery, or internal medicine), and three to five residents. Each team has separate rounds once a day. On weekdays, at 4:00 p.m. both teams handoff to the night team, composed of one fellow and two residents. The nighttime fellow may be the fellow responsible for one of the daytime teams or may be a fellow who was working in a different unit during the day. The call schedule for residents is independent of the call schedule from the fellows and not nested within a team. On weekends residents and fellows take 24-hour shifts. An electronic sign-out system is available to support the process of handoffs.

Variables

The primary outcome variable was ICU mortality and our unit of analysis was each ICU admission. Our secondary outcome variable was nighttime decisions, defined as new interventions or diagnostic procedures ordered after 4:00 p.m. and before 7:00 a.m. of the following day, or a change in the treatments or plans that were established before handoff. We selected interventions and diagnostic procedures that were deemed a priori to be feasible for objective collection from medical charts without the need for interview from clinicians, to avoid recall bias. We presented the proposed variables to one focus group of critical care researchers for face validity and to explore other elements that would merit data collection. The final set of data elements for nighttime decisions included eight variables (Table 1): (1) initiation of new antibiotics, (2) ordering of computed tomography (CT) scans, (3) changes in mechanical ventilation, (4) changes in extubation plan, (5) changes in doses of sedatives, (6) changes in doses of opioids, (7) changes in fluid boluses, and (8) changes in blood transfusion. Further definitions for nighttime decisions are included in the online supplement.

Table 1: Definitions for Nighttime Decisions (Decisions between 4:00 p.m. and 7:00 a.m.)

DecisionDefinition
AntibioticOrders written for a new antibiotic
CT scanOrders written for a new CT scan
Changes in mechanical ventilationAn increase in mechanical ventilation requirements, defined as one of the following:
 1. Intubation
 2. Change from breathing through a tracheal mask to ventilation in either pressure support or an assisted-controlled mode
 3. Change from ventilation in pressure support mode to an assisted-controlled mode
Changes in extubation planUnplanned extubation, defined as an extubation that was not documented as part of the goals for the day
Self-extubation, as noted from nursing notes
Failure to proceed with extubation, defined as the failure to extubate a patient when extubation was documented as part of the goals of the day
SedationChanges in the average hourly infusion of sedatives at night, transformed into hourly equivalents of midazolam
AnalgesiaChanges in the average hourly infusion of opioids at night, transformed into hourly equivalents of fentanyl
Fluid bolusChanges in the volume of fluids infused as boluses at night
TransfusionChanges in the volume of packed red blood cells transfused at night

Definition of abbreviation: CT = computed tomography.

Our primary exposure variable was the cross-covering status of fellows. We defined a continuity-of-care fellow as a person that was responsible for the care of patients at night who attended the morning interdisciplinary rounds related to that patient. We defined a cross-covering fellow as a person that was responsible for the care of patients at night without having attended the morning interdisciplinary rounds related to that patient.

Patient Selection

We enrolled all consecutive patients admitted in a 12-month period from January 2010 to December 2010 that were admitted for a minimum of 48 hours in the ICU and had at least one episode of an interdisciplinary round on a weekday. For our secondary outcomes we limited data collection to the 3-month period between July 2011 and September 2011. For our secondary outcomes we also excluded observations that occurred on Fridays and weekends because the fellows and residents are always present for the entire 24-hour period, to exclude any weekend effects (15) and for feasibility reasons.

Data Collection

For the primary analysis we collected data on the exposure variable from the call schedules and abstracted the outcomes and covariates from medical charts during the period of interest. For the secondary outcomes two investigators (B.S.B. and C.C.P.P.B.) collected data into an electronic database that was previously piloted. Data were collected each morning from the preceding call night. We assigned individual codes for fellows in the unit and collected data on who were the members and patients of each team for that day and for the previous night. Because there was a risk of ascertainment bias if data collectors were aware of cross-coverage status, we defined the outcome variables by protocol and we used different data collectors for the assessment of outcomes and cross-coverage status.

We collected demographic variables for each patient, including date of admission, age, sex, reason for ICU admission, admission date, and Acute Physiology and Chronic Health Evaluation (APACHE) II scores at admission.

Statistical Analysis

We described the baseline characteristics of patients with means and standard deviations, medians and interquartile ranges, or proportions, as appropriate. We performed univariate analysis using chi-square or Fisher exact test for categorical variables and t test or Wilcoxon rank sum test for continuous variables.

Our primary analysis used logistic regression to estimate the effects of the initial 7-day cumulative exposure to cross-coverage on mortality. Because the exposure is linked to the time-at-risk for cross-coverage, we also adjusted for ICU length of stay truncated at 7 days. Because the potential effects of nighttime cross-coverage on mortality may vary over time we did six sensitivity analyses to properly account for the time-varying characteristics of the exposure and to support the veracity of our findings: (1) we included a Cox proportional hazards model to estimate the effects of the cumulative exposure to cross-coverage until Day 7 on ICU survival (cumulative cross-coverage and number of weekend days were entered as time-dependent variables); (2) we repeated a similar Cox proportional hazards analysis with the exposure coded as percentage of cross-coverage in the first 7 days; (3) we used McNemar test to compare the highest and lowest quartiles of exposure to cross-coverage; (4) we used a time-dependent Cox proportional hazards model to measure the effects of the immediate exposure to cross-coverage on survival, recoding cross-coverage as a time-dependent binary variable; and we used a landmark analysis (16, 17) with (5) logistic regression and (6) Cox-proportional hazards to estimate the effects of cumulative exposure to cross-coverage on Days 2 through 7 of ICU admission. Landmark analysis is commonly used in cancer research to adjust for immortal-time bias, with recoding of the exposure status up to the landmark day. For example, in this study, a patient that has no cross-coverage on Day 2 may have 3 days of cross-coverage by Day 5; therefore, the landmark analysis shifts the exposure on daily basis. For all patients alive and in ICU at a given landmark day we used Cox proportional hazards and logistic regression to estimate the effects of cumulative exposure to cross-coverage up to and inclusive of the landmark day on subsequent ICU survival. We adjusted all models for variables collected at ICU admission, APACHE II, sex, age, type of admission, source of admission, admission on weekends, and for the subsequent exposure to weekend days. We chose an arbitrary cut-off of 7 days for the exposure variable because after a certain period of time all patients converge to a similar amount of cross-coverage (see Figure E1 in the online supplement). Also, by choosing 7 days, we allowed all patients that have a length of stay longer than 7 days to have an equal number of nights at risk for cross-coverage, independent of admission on weekends or weekdays. For the secondary outcomes, we used generalized estimating equations for negative binomial and continuous variables with clustering at the patient level to provide estimates of effect size, thus avoiding underestimating the standardized errors caused by correlated observations on the same patients. We adjusted for age, sex, APACHE II score, and daily Sequential Organ Failure Assessment score. For the categorical variables (antibiotics, CT scans, changes in ventilation, and changes in extubation plan) we expected a low number of decisions in each category and therefore we combined the four variables to create a single categorical variable, presence of nighttime decisions. For the continuous variables we were interested in changes from daytime to nighttime. For sedation and analgesia we averaged doses per hour of infusion for both periods and for fluid boluses and transfusions we added the total amount for each period. The analytical approach thus compares differences between groups in the within-patient day with night change in average doses or median volumes administered.

All aspects of the study were reviewed and approved by the research ethics board of Sunnybrook Health Sciences Centre (Project Identifier 076–2011). Because this research was deemed to pose minimal risk to participants, the need for individual informed consent was waived.

Patient Characteristics

There were 882 admissions for the period of interest, and after the exclusion of patients with missing data or length of stay equal or less than 2 days, we analyzed data for 629 admissions, from 569 unique patients (Figure 1). Patient characteristics are presented in Table 2. Reflecting the patient population at our center, most patients were male and were admitted with trauma, respiratory failure, or a primary neurologic condition as their primary diagnosis. The median length of stay was 6 days (interquartile range, 4–10 d) and the median APACHE II score was 23 (interquartile range, 19–28). During this period, there were 29 different fellows and the median proportion of nights cross-covered through the 7th day was 50% (interquartile range, 29–60%).

Table 2: Demographics of the 629 Admissions

Age, yr, mean ± SD59.8 ± 20.7
Sex, % male66
APACHE II, median (IQR)23 (19–28)
ICU length of stay, d, median (IQR)6 (4–10)
Reason for admission, % (n) 
 Trauma29 (182)
 Respiratory failure32 (202)
 Neurologic13 (80)
 Cardiovascular6 (37)
 Gastrointestinal7 (42)
 Sepsis4 (24)
 Other10 (62)
Mortality, % (n)18 (116)

Definition of abbreviations: APACHE = Acute Physiology and Chronic Health Evaluation; ICU = intensive care unit; IQR = interquartile range.

For our secondary outcomes we collected data from 121 consecutive admissions, with a total of 525 patient-nights for assessment. During this period, we observed 16 different fellows. There were 400 nights of cross-coverage by fellows and 125 of continuity of care.

ICU Mortality

Univariate comparisons between patients exposed to cross-coverage higher or lower than the median show no significant differences between the groups, except for ICU length-of-stay (Table 3). In the primary logistic regression analysis, cumulative exposure to cross-coverage was associated with reduced mortality (odds ratio [OR], 0.77 per 1 d; 0.60–0.99; P = 0.04) and APACHE II score was associated with increased mortality (OR, 1.15 per 1 point; 1.11–1.19; P < 0.0001). Results from the sensitivity analysis show the same directionality of effects, with a varying effect size, caused by the different definitions of exposure and outcome (Table 4, Figure 2A). The 28-day difference in predicted mortality for patients exposed to 2 versus 5 days of cross-coverage was 20% (Figure 2B).

Table 3: Comparisons between Admissions Exposed to Higher and Lower Levels of Cross-Coverage

 Cross-Coverage Less Than or Equal to the Median (n = 386)Cross-Coverage Greater Than the Median (n = 243)
Age, yr, mean ± SD59.9 ± 20.459.6 ± 21.2
Sex, % male6664
APACHE II, median (IQR)23 (18–28)23 (19–29)
ICU length of stay, d, median (IQR)5 (4–9)*7 (4–13)
Reason for admission, % (n)  
 Trauma29 (113)28 (69)
 Respiratory failure32 (124)32 (78)
 Neurologic13 (50)12 (30)
 Cardiovascular5 (20)7 (17)
 Gastrointestinal6 (24)7 (18)
 Sepsis4 (15)4 (9)
 Other10 (40)9 (22)
Mortality, % (n)19.4 (75)16.9 (41)
Cross-coverage, % (IQR)33.3 (25–42.9)66.7 (57.1–83.3)

Definition of abbreviations: APACHE = Acute Physiology and Chronic Health Evaluation; ICU = intensive care unit; IQR = interquartile range.

*P < 0.01.

Table 4: Primary and Sensitivity Analysis of the Effects of Nighttime Cross-Coverage on Mortality

Statistical AnalysisDefinition of ExposurePopulationOutcome MeasuredEstimated Benefit of Additional Cross-Coverage (95% CI)P Value
Logistic regressionCounts of cross-coverage on first 7 dAllICU mortalityOR, 0.77 (0.60–0.99)0.040
Time-dependent Cox modelCounts of cross-coverage on first 7 dAllTime to ICU deathHR, 0.78 (0.63–0.96)0.018
Landmark analysis with logistic regressionCounts of cross-coverage on first 7 dAlive at Day 7ICU mortalityOR, 0.86 (0.62–1.18)0.345
Landmark analysis with Cox modelCounts of cross-coverage on first 7 dAlive at Day 7Time to ICU deathHR, 0.74 (0.56–0.99)0.042
Time-dependent Cox modelPercentage cross-coverage on first 7 d, per 10%AllTime to ICU deathHR, 0.91 (0.84–0.99)0.048
Unadjusted comparison of highest and lowest quartilesQuartile of exposure to cross-coverageHighest and lowest quartilesICU mortalityRR, 0.83 (0.56–1.24)0.240
Time-dependent Cox modelCross-coverage status until discharge (binary, noncumulative)AllTime to ICU deathHR, 0.82 (0.54–1.27)0.387

Definition of abbreviations: CI = confidence interval; HR = hazard ratio; ICU = intensive care unit; OR = odds ratio; RR = relative risk.

Nighttime Decisions

There were no differences between patients, on the nights they were assessed for nighttime decisions, when fellows were cross-covering or providing continuity care (Table 5). There were a higher number of nighttime decisions when fellows were cross-covering (19.3% vs. 10.4%; OR, 2.02; 95% confidence interval [CI], 1.03–3.95; P = 0.04). We observed an increase in fentanyl equivalents administered to patients at night (difference, +10.2 μg/h; 95% CI, +1.4 to +19.0; P = 0.02) and an increase in transfusions at night (difference, +465 ml; 95% CI, +98 to +832; P = 0.01) when fellows were cross-covering. Results were robust to a post hoc sensitivity analysis, using clusters both at the patient and provider levels (see Table E8).

Table 5: Comparisons between Cross-Covering and Continuity Nights

 Fellow Cross-Covering
Yes (n = 400)No (n = 125)
Age, yr, mean (SD)54.7 (22.5)56.6 (20.6)
APACHE II, median (IQR)24 (21 to 29)24 (21 to 28.5)
ICU length of stay, median (IQR)7 (3 to 19)7 (3 to 14)
SOFA score at day of assessment, mean (SD)6.5 (4.1)6.2 (4.1)
Nighttime decisions, n (%)77 (19.3)13 (10.4)*
 Computed tomography scan20 (5)1 (0.8)
 Antibiotics31 (7.8)6 (4.8)
 Changes in mechanical ventilation17 (4.2)4 (3.2)
 Changes in extubation planning13 (3.2)2 (1.6)
Sedation and analgesia  
 Sedatives (changes in mg/h of midazolam equivalents), mean (SD)−0.3 (8.8)−0.1 (11.2)
 Opioids (changes in μg/h of fentanyl equivalents), mean (SD)3.3 (33.2)−6.6 (34.7)*
Resuscitation  
 Fluid bolus (changes in ml), median (IQR)+233 (−778 to +933)−583 (−778 to +622)
 Transfusion (changes in ml), median (IQR)−78 (−415 to +266)−544 (−724 to −417)*

Definition of abbreviations: APACHE = Acute Physiology and Chronic Health Evaluation; ICU = intensive care unit; IQR = interquartile range; SOFA = Sequential Organ Failure Assessment.

*P < 0.05.

In this single-center study of cross-covering physicians we observed that the odds of dying in the ICU were decreased by 23% for each additional day of nighttime cross-coverage in the first 7 days of ICU admission. Cross-covering fellows at night made more nighttime decisions (defined as CT scans, antibiotic initiation, changes in mechanical ventilation, and changes in extubation plans), which happened twice as frequently when compared with continuity-of-care fellows. This was particularly noted in an increase in CT scans, which were ordered by cross-covering fellows on 5% of patient-nights, but only in 0.8% of patient-nights by continuity fellows.

Based on the available literature, we expected that increased exposure to cross-covering fellows would lead to worse outcomes. However, our results contradict this hypothesis. The analysis was robust to several sensitivity analyses, including the use of a time-dependent variable and a landmark analysis. The landmark analysis was only significant at Day 7, which we speculate may be caused by the study being underpowered for the other landmarks, or by the cross-coverage effects being more important after the first ICU days, when daytime clinicians may become more comfortable with the current working diagnosis and may be less attentive to new diagnosis or to the lack of response to treatment.

Most previous reports on handoffs demonstrate associations between handoffs and medical errors (18). Thus, the dominant hypothesis is that increases in handoffs and cross-coverage should lead to worse patient outcomes given loss or corruption of relevant clinical information. However, the countervailing hypothesis is that a different clinician may be more prone to reevaluate patients and act on it. Indeed, this mechanism was suggested in one previous report from handoffs in anesthesia, because they observed “favorable incidents” in which the cross-covering anesthetist discovered an error that was previously missed (19). These cognitive errors may be caused by two potential mechanisms: anchoring biases, where clinicians are locked on earlier diagnosis, despite disconfirming evidence, which was the most common cause of cognitive errors by internal medicine residents (20); and clinical inertia, where clinicians observe a problem, but fail to act on it. In a cohort of patients with dyslipidemia with ischemic heart disease, clinical inertia, represented by a failure to adjust lipid-lowering therapy, occurred in 43% of consultations (21).

In fact, we believe our results may not be that surprising given two important considerations. First, we believe the current literature assessing handoffs is biased by using medical errors as the outcome. Because clinicians assess medical errors conditional on patient outcomes (14), the assessment of medical errors made unmasked to whether a handover event occurred and with knowledge of the outcome is subject to bias. This may inflate the association linking handover to medical errors. We addressed this limitation by using two objective measures, clinical decisions at night and mortality, assessed by strict definitions independently of handover status. Second, the association between medical errors and patient outcomes is overinflated. In a careful review of hospital deaths, only 0.5% of deaths would be prevented had optimal care been provided (22). Thus, the hypothesis that reducing or improving handover would lead to substantial improvements in mortality by preventing clinically significant medical error may not be justified. Although it is plausible that some of the medical errors associated with handoffs are true, our data suggest that the balance between medical errors from handoffs and the benefit of a “second look” by another clinician seems to favor the latter. This second look may act as a control for cognitive errors. Therefore, strategies to improve handoffs are necessary, but interventions that decrease cognitive errors may be even more important. Examples of the latter include shared rounds with other intensivists, mandatory consultation by a second intensivist in complex cases, creating discontinuity in the nighttime schedule, diagnostic checklists (23), and decision-support systems.

Our study has several limitations. First, this is a single-center study; therefore, the results may not be generalizabile. Of particular relevance are the culture and organization of this unit, which provide the cross-covering fellow with freedom to change plans of care. Second, we did not focus our study on the actions of residents at night. Much of the previous literature focuses on residents; however, in our center the primary responsibility for patient care at night falls on the fellows, who are more experienced clinicians, and thus our findings may not be generalizable to units that do not have more experienced trainees at night or that are not academic. Third, we also did not study the effects of cross-coverage by staff intensivists. In fact, one study of weekend coverage by experienced intensivists did not observe differences in clinical outcomes between a continuity-of-care weekend schedule and a cross-coverage weekend schedule (24) and in a randomized trial the addition of a nighttime intensivist did not lead to better outcomes (25). However, our study is fundamentally different from the latter trial, because we did not study the addition of another provider at nighttime, but rather the differences between continuity and cross-covering fellows. Furthermore, we do not have information on the cross-covering status of trainees in the latter trial. Fourth, the high-intensivist staffing model used in this unit further limits generalizability to other settings, because the effects of nighttime staffing are different in low- versus high-intensivist staffed units (26). Fifth, to avoid ascertainment biases for nighttime decisions we decided to focus on items that could be objectively abstracted from charts. Although this certainly increases the internal validity of our study, it is possible that many other decisions at night could be different between cross-covering and continuity fellows. Finally, observational studies are more prone to residual confounding that could potentially explain the observed association; however, given that the design of the call schedule has no association with the outcome it is unlikely that an important residual confounder is present.

Our observations have direct implications for patient safety because the differences observed between cross-covering and continuity-of-care fellows may explain errors, especially cognitive errors. It is possible that anchoring biases and clinical inertia led continuity fellows to perform fewer actions at night, such as ordering diagnostic imaging studies and antimicrobials, therefore delaying proper diagnosis and treatment. Because our study was conducted in a setting where an electronic tool for handoffs is available, we were surprised to observe differences in nighttime decisions. However, previous reports observed frequent failures in electronic tools for handoffs, such as discrepancies in medications in use (27), goals of care, diagnosis (28), and the presence of contradictory content (29).

Conclusions

Our findings contradict the dominant hypothesis that cross-coverage is associated with worse outcomes, because we observed a decreased mortality with greater cross-coverage. This important and novel finding may be mediated by differences in clinical decisions at night between cross-covering and continuity-of-care fellows. Our findings suggest that a “second look” by cross-covering fellows may mitigate cognitive errors. Future interventions to improve patient safety in ICUs should focus both on the quality of handoffs and on strategies to decrease cognitive errors.

1. Cohen MD, Hilligoss B, Kajdacsy-Balla Amaral AC. A handoff is not a telegram: an understanding of the patient is co-constructed. Crit Care 2012;16:303.
2. Nasca TJ, Day SH, Amis ES Jr; ACGME Duty Hour Task Force. The new recommendations on duty hours from the ACGME Task Force. N Engl J Med 2010;363:e3.
3. Emlet LL, Al-Khafaji A, Kim YH, Venkataraman R, Rogers PL, Angus DC. Trial of shift scheduling with standardized sign-out to improve continuity of care in intensive care units. Crit Care Med 2012;40:31293134.
4. Smalley HK, Keskinocak P, Vats A. Development of a handoff continuity score to improve pediatric ICU physician schedule design for enhanced physician and patient continuity. Crit Care 2011;15:R246.
5. Riesenberg LA, Leitzsch J, Little BW. Systematic review of handoff mnemonics literature. Am J Med Qual 2009;24:196204.
6. Petersen LA, Orav EJ, Teich JM, O’Neil AC, Brennan TA. Using a computerized sign-out program to improve continuity of inpatient care and prevent adverse events. Jt Comm J Qual Improv 1998;24:7787.
7. Van Eaton EG, Horvath KD, Lober WB, Rossini AJ, Pellegrini CA. A randomized, controlled trial evaluating the impact of a computerized rounding and sign-out system on continuity of care and resident work hours. J Am Coll Surg 2005;200:538545.
8. Starmer AJ, Sectish TC, Simon DW, Keohane C, McSweeney ME, Chung EY, Yoon CS, Lipsitz SR, Wassner AJ, Harper MB, et al. Rates of medical errors and preventable adverse events among hospitalized children following implementation of a resident handoff bundle. JAMA 2013;310:22622270.
9. Petersen LA, Brennan TA, O’Neil AC, Cook EF, Lee TH. Does housestaff discontinuity of care increase the risk for preventable adverse events? Ann Intern Med 1994;121:866872.
10. Kachalia A, Gandhi TK, Puopolo AL, Yoon C, Thomas EJ, Griffey R, Brennan TA, Studdert DM. Missed and delayed diagnoses in the emergency department: a study of closed malpractice claims from 4 liability insurers. Ann Emerg Med 2007;49:196205.
11. Arora V, Johnson J, Lovinger D, Humphrey HJ, Meltzer DO. Communication failures in patient sign-out and suggestions for improvement: a critical incident analysis. Qual Saf Health Care 2005;14:401407.
12. Philibert I. Use of strategies from high-reliability organisations to the patient hand-off by resident physicians: practical implications. Qual Saf Health Care 2009;18:261266.
13. Borowitz SM, Waggoner-Fountain LA, Bass EJ, Sledd RM. Adequacy of information transferred at resident sign-out (in-hospital handover of care): a prospective survey. Qual Saf Health Care 2008;17:610.
14. Caplan RA, Posner KL, Cheney FW. Effect of outcome on physician judgments of appropriateness of care. JAMA 1991;265:19571960.
15. Laupland KB, Shahpori R, Kirkpatrick AW, Stelfox HT. Hospital mortality among adults admitted to and discharged from intensive care on weekends and evenings. J Crit Care 2008;23:317324.
16. Anderson JR, Cain KC, Gelber RD. Analysis of survival by tumor response. J Clin Oncol 1983;1:710–719.
17. Rao SV, Jollis JG, Harrington RA, Granger CB, Newby LK, Armstrong PW, Moliterno DJ, Lindblad L, Pieper K, Topol EJ, et al. Relationship of blood transfusion and clinical outcomes in patients with acute coronary syndromes. JAMA 2004;292:15551562.
18. Cohen MD, Hilligoss PB. The published literature on handoffs in hospitals: deficiencies identified in an extensive review. Qual Saf Health Care 2010;19:493497.
19. Cooper JB, Long CD, Newbower RS, Philip JH. Critical incidents associated with intraoperative exchanges of anesthesia personnel. Anesthesiology 1982;56:456461.
20. Ogdie AR, Reilly JB, Pang WG, Keddem S, Barg FK, Von Feldt JM, Myers JS. Seen through their eyes: residents’ reflections on the cognitive and contextual components of diagnostic errors in medicine. Acad Med 2012;87:13611367.
21. Lázaro P, Murga N, Aguilar D, Hernández-Presa MA; INERTIA Study Investigators. Therapeutic inertia in the outpatient management of dyslipidemia in patients with ischemic heart disease. The INERTIA study. Rev Esp Cardiol 2010;63:14281437.
22. Hayward RA, Hofer TP. Estimating hospital deaths due to medical errors: preventability is in the eye of the reviewer. JAMA 2001;286:415420.
23. Ely JW, Graber ML, Croskerry P. Checklists to reduce diagnostic errors. Acad Med 2011;86:307313.
24. Ali NA, Hammersley J, Hoffmann SP, O’Brien JM Jr, Phillips GS, Rashkin M, Warren E, Garland A; Midwest Critical Care Consortium. Continuity of care in intensive care units: a cluster-randomized trial of intensivist staffing. Am J Respir Crit Care Med 2011;184:803808.
25. Kerlin MP, Small DS, Cooney E, Fuchs BD, Bellini LM, Mikkelsen ME, Schweickert WD, Bakhru RN, Gabler NB, Harhay MO, et al. A randomized trial of nighttime physician staffing in an intensive care unit. N Engl J Med 2013;368:22012209.
26. Wallace DJ, Angus DC, Barnato AE, Kramer AA, Kahn JM. Nighttime intensivist staffing and mortality among critically ill patients. N Engl J Med 2012;366:20932101.
27. Arora V, Kao J, Lovinger D, Seiden SC, Meltzer D. Medication discrepancies in resident sign-outs and their potential to harm. J Gen Intern Med 2007;22:17511755.
28. Aylward MJ, Rogers T, Duane PG. Inaccuracy in patient handoffs: discrepancies between resident-generated reports and the medical record. Minn Med 2011;94:3841.
29. Campion TR Jr, Denny JC, Weinberg ST, Lorenzi NM, Waitman LR. Analysis of a computerized sign-out tool: identification of unanticipated uses and contradictory content. AMIA Annu Symp Proc 2007;2007:99104.
Correspondence and requests for reprints should be addressed to Andre Carlos Kajdacsy-Balla Amaral, M.D., Sunnybrook Health Sciences Centre, 2075 Bayview Avenue, Office D1 08, Toronto, M4N 3M5 Canada. E-mail:

Author Contributions: Conception and design, all authors. Analysis and interpretation, A.C.K.-B.A., R.P., and G.D.R. Drafting the manuscript for important intellectual content, all authors.

Originally Published in Press as DOI: 10.1164/rccm.201312-2181OC on April 29, 2014

This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org

Author disclosures are available with the text of this article at www.atsjournals.org.

Comments Post a Comment




New User Registration

Not Yet Registered?
Benefits of Registration Include:
 •  A Unique User Profile that will allow you to manage your current subscriptions (including online access)
 •  The ability to create favorites lists down to the article level
 •  The ability to customize email alerts to receive specific notifications about the topics you care most about and special offers
American Journal of Respiratory and Critical Care Medicine
189
11

Click to see any corrections or updates and to confirm this is the authentic version of record