Association of Anesthesiologist Handovers With Short-term Outcomes for Patients Undergoing Cardiac Surgery : Anesthesia & Analgesia

Secondary Logo

Journal Logo

Original Research Articles: Original Clinical Research Report

Association of Anesthesiologist Handovers With Short-term Outcomes for Patients Undergoing Cardiac Surgery

Hannan, Edward L. PhD*; Samadashvili, Zaza MD*; Sundt, Thoralf M. III MD; Girardi, Leonard MD; Chikwe, Joanna MD§; Wechsler, Andrew MD; Adams, David H. MD; Smith, Craig R. MD#; Gold, Jeffrey P. MD**; Lahey, Stephen J. MD††; Jordan, Desmond MD‡‡

Author Information
Anesthesia & Analgesia 131(6):p 1883-1889, December 2020. | DOI: 10.1213/ANE.0000000000005221



  • Question: Are complete anesthesia handovers during cardiac surgery associated with higher adjusted adverse outcome rates?
  • Findings: Cardiac surgery patients experiencing anesthesia handovers have higher short-term mortality rates, but other adverse outcome rates are not clinically different.
  • Meaning: Handovers should be carefully monitored, and unnecessary handovers studied more comprehensively to identify handover characteristics associated with higher adverse outcome rates.

The handover of anesthesia care during surgical procedures is a practice that can occur due to illness, personal or professional commitments, or an attempt to avoid the detrimental effects of sleep-deprived care.1 However, there is a potential danger of a handover of care from one anesthesiologist to another during a surgical procedure because important details of the procedure and the patient may be lost during the communication from one anesthesiologist to the other that has to happen while still monitoring and caring for the patient.

Several recent studies have examined the association of complete handovers of anesthesia care during surgery with adverse outcome rates, and most, but not all, of them have concluded that there is a deleterious effect of complete handovers.2–7 These studies have examined a variety of procedures, but most of them have been restricted to a single hospital and/or a single procedure.

The purposes of this study are to identify the extent to which complete handovers of anesthesia care for cardiac surgery occur in New York State, to compare the characteristics of patients who do and do not experience complete handovers, and to compare short-term outcomes of cardiac surgery patients in New York who did and did not experience complete anesthesia handovers during their cardiac procedure. The study includes all nonfederal hospitals in the state in which cardiac surgery is performed.


The requirement for written informed consent was waived by the institutional review board (IRB) of the University at Albany. This article adheres to the applicable STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) guidelines.


Patients in the study were identified from the New York State Department of Health’s Cardiac Surgery Reporting System (CSRS) registry. CSRS contains all patients who undergo cardiac surgery in nonfederal hospitals in the state. It contains detailed information on vessels diseased, ventricular function, comorbidities, dates of admission and discharge, date of procedure, complications of care, discharge status, and providers (including anesthesiologists). The registry was linked to New York’s acute care administrative database (the Statewide Planning and Research Cooperative System [SPARCS]) using patient identifiers, hospital identifiers, and dates of admission, procedures, and discharge to obtain in-hospital complications, and linked to the New York State Vital Statistics death registry to obtain deaths that occurred after discharge.

Study Population

Patients were initially eligible for the study if they underwent cardiac surgery in nonfederal hospitals in New York between 2010 and 2016. After exclusions for missing/invalid Social Security numbers (6249), out-of-state residence (8523), and missing/invalid surgery times (163), the remaining 103,102 patients were subjects of the study. Patients whose residence was outside the state were excluded because deaths of non–New York patients after discharge are not contained in the New York vital statistics database, and patients with missing social security numbers were excluded because they could not be matched to vital statistics data. Patients with complete anesthesia handovers were defined as patients whose anesthesiologists at the beginning and end of the procedure were different. The cardiac surgery groups that were studied included isolated coronary artery bypass graft (CABG) surgery, isolated valve surgery, valve/CABG surgery, and aortic surgery.


The primary outcome was in-hospital/30 day mortality. Other outcomes were major complications during the index admission or within 30 days from the index surgery, readmissions within 30 days of the index admission, and length of stay. Complications of care were identified using SPARCS data based on the complications used by Jones et al.2 Supplemental Digital Content, Appendix 1,, identifies the individual complications and provides their relative frequencies. All outcomes were specified a priori.

Statistical Analysis

Since this was an observational study in which patients could not be randomly assigned to anesthesia handovers versus no handovers, there was potential for selection bias whereby large differences in patient risk factors could occur between patients with and without anesthesia handovers. Consequently, inverse probability weighting based on propensity scores (inverse probability treatment weighting [IPTW]) was used to minimize selection bias.

The propensity score was derived by developing a nonparsimonious logistic regression model that predicted the probability that a given patient would experience an anesthesia handover based on all risk factors available in the registry.8–10 The dependent variable in the model was the presence or absence of a complete anesthesia handover. The independent variables included surgery duration, surgery start time, numerous demographics, type and priority of cardiac surgery, specific coronary vessels diseased, comorbidities, ventricular function, hemodynamic status, and previous myocardial infarction. A list of the variables used is presented in Table 1. Age and surgery duration were represented as continuous variables in the model but were later split into categories when exploring interactions.

Table 1. - Baseline Characteristics of the Patients Before and After IPTW
Characteristic Observed Data IPTW Data
No Handover Handover ASD No Handover Handover ASD
Surgery duration 4.6 (±1.6) 5.0 (±1.8) 23.6 4.6 (±1.7) 4.6 (±1.8) 4.8
Age, y 66.9 (±11.8) 66.9 (±11.9) 0.3 66.9 (±11.8) 66.8 (±11.9) 1
Female gender 30,845 (32.7) 2758 (31.3) 2.9 30,774.7 (32.6) 2837.2 (32.5) 0.2
Body surface area 2.0 (±0.3) 2.0 (±0.3) 3.3 2.0 (±0.3) 2.0 (±0.3) 0.5
Type of surgery
 Isolated CABG 48,368 (51.3) 4890 (55.6) 8.6 48,757.0 (51.6) 4527.6 (51.9) 0.5
 AVR 13,700 (14.5) 916 (10.4) 12.5 13,369.9 (14.2) 1224.9 (14.0) 0.4
 MVR 3374 (3.6) 227 (2.6) 5.8 3295.5 (3.5) 298.0 (3.4) 0.4
 MVV 4500 (4.8) 265 (3.0) 9.1 4359.8 (4.6) 428.7 (4.9) 1.4
 MULT 4849 (5.1) 405 (4.6) 2.5 4830.1 (5.1) 466.1 (5.3) 1
 AVR/CABG 8495 (9.0) 782 (8.9) 0.4 8490.7 (9.0) 740.4 (8.5) 1.8
 MULT/CABG 1545 (1.6) 181 (2.1) 3.1 1586.9 (1.7) 149.8 (1.7) 0.3
 MVR/CABG 1387 (1.5) 160 (1.8) 2.7 1421.2 (1.5) 132.3 (1.5) 0.1
 MVV/CABG 1944 (2.1) 242 (2.8) 4.5 1998.1 (2.1) 176.0 (2.0) 0.7
 AORTIC 6142 (6.5) 730 (8.3) 6.8 6333.0 (6.7) 587.7 (6.7) 0.1
Nonteaching hospital 42,577 (45.1) 4352 (49.5) 8.7 42,998.0 (45.5) 3874.5 (44.4) 2.3
Surgery start time
 6 am–12 pm 73,821 (78.3) 3177 (36.1) 94.2 70,421.0 (74.6) 6484.2 (74.3) 0.7
 12 pm–3 pm 15,976 (16.9) 4138 (47.0) 68.2 18,408.5 (19.5) 1712.6 (19.6) 0.3
 4 pm–7 pm 3830 (4.1) 1343 (15.3) 38.6 4814.0 (5.1) 454.8 (5.2) 0.5
 7 pm–12 am 463 (0.5) 65 (0.7) 3.2 529.9 (0.6) 54.7 (0.6) 0.8
 12 am–6 am 214 (0.2) 75 (0.9) 8.5 268.7 (0.3) 25.3 (0.3) 0.1
Priority of surgery
 Elective 42,994 (45.6) 3126 (35.5) 20.6 42,151.6 (44.6) 3332.1 (38.2) 13.2
 Urgent 48,501 (51.4) 5123 (58.2) 13.7 49,183.5 (52.1) 5024.7 (57.5) 11
 Emergency 2809 (3.0) 549 (6.2) 15.6 3107.0 (3.3) 374.7 (4.3) 5.2
Left ventricular ejection fraction (%)
 <20 1186 (1.3) 111 (1.3) 0 1189.4 (1.3) 112.2 (1.3) 0.2
 20–29 4887 (5.2) 507 (5.8) 2.6 4935.3 (5.2) 456.4 (5.2) 0
 30–39 8319 (8.8) 850 (9.7) 2.9 8429.8 (8.9) 786.1 (9.0) 0.3
 40–49 13,527 (14.3) 1291 (14.7) 0.9 13,562.6 (14.4) 1283.6 (14.7) 1
Unstable hemodynamics
 Refractory shock 207 (0.2) 48 (0.5) 5.3 236.2 (0.3) 31.6 (0.4) 2
 Nonrefractory shock 607 (0.6) 108 (1.2) 6.1 663.3 (0.7) 74.2 (0.8) 1.7
Previous myocardial infarction
 0–23 h 1423 (1.5) 237 (2.7) 8.3 1544.1 (1.6) 195.0 (2.2) 4.3
 1–7 d 11,378 (12.1) 1292 (14.7) 7.7 11,605.9 (12.3) 1198.5 (13.7) 4.3
 8–14 d 3153 (3.3) 289 (3.3) 0.3 3147.5 (3.3) 272.6 (3.1) 1.2
 15–20 d 647 (0.7) 81 (0.9) 2.6 666.1 (0.7) 59.2 (0.7) 0.3
 ≥3 wk 14,756 (15.6) 1396 (15.9) 0.6 14,803.7 (15.7) 1303.0 (14.9) 2.1
Vessels diseased
 Left main 18,755 (19.9) 2085 (23.7) 9.2 19,077.0 (20.2) 1731.7 (19.8) 0.9
 Proximal LAD 28,770 (30.5) 3002 (34.1) 7.7 29,096.7 (30.8) 2700.0 (30.9) 0.2
 1-vessel 14,075 (14.9) 1354 (15.4) 1.3 14,116.7 (14.9) 1350.4 (15.5) 1.4
 2-vessel 19,812 (21.0) 1976 (22.5) 3.5 19,937.5 (21.1) 1814.4 (20.8) 0.8
 3-vessel 28,051 (29.7) 3011 (34.2) 9.6 28,464.2 (30.1) 2638.6 (30.2) 0.2
 Peripheral vascular disease 13,243 (14.0) 1413 (16.1) 5.6 13,476.7 (14.3) 1225.2 (14.0) 0.7
 Cerebrovascular disease 14,530 (15.4) 1402 (15.9) 1.5 14,582.1 (15.4) 1257.1 (14.4) 2.9
 CHF, this admission 20,344 (21.6) 2235 (25.4) 9 20,700.2 (21.9) 1898.7 (21.7) 0.4
 CHF, past admission 6201 (6.6) 473 (5.4) 5.1 6105.4 (6.5) 532.1 (6.1) 1.5
 Ventricular arrhythmia 576 (0.6) 77 (0.9) 3.1 603.2 (0.6) 70.6 (0.8) 2
 COPD 20,049 (21.3) 1988 (22.6) 3.2 20,212.6 (21.4) 1802.8 (20.6) 1.9
 Aortic atherosclerosis 3861 (4.1) 328 (3.7) 1.9 3829.8 (4.1) 327.3 (3.7) 1.6
 Diabetes 29,725 (31.5) 2768 (31.5) 0.1 29,752.9 (31.5) 2798.5 (32.1) 1.2
 Hepatic failure 63 (0.1) 4 (0.0) 0.9 61.8 (0.1) 5.7 (0.1) 0
Renal dysfunction
 Creatinine, 1.3–1.5 11,022 (11.7) 1058 (12.0) 1 11,078.2 (11.7) 1040.2 (11.9) 0.6
 Creatinine, 1.6–2.0 5532 (5.9) 524 (6.0) 0.4 5542.2 (5.9) 474.9 (5.4) 1.9
 Creatinine, >2.0 2585 (2.7) 285 (3.2) 2.9 2640.5 (2.8) 243.2 (2.8) 0.1
 Renal dialysis 2939 (3.1) 302 (3.4) 1.8 2964.6 (3.1) 255.9 (2.9) 1.2
Endocarditis 1641 (1.7) 161 (1.8) 0.7 1678.5 (1.8) 145.6 (1.7) 0.8
Previous cardiac procedures
 PCI, this admission 1265 (1.3) 177 (2.0) 5.2 1320.3 (1.4) 128.2 (1.5) 0.6
 PCI, before this admission 18,961 (20.1) 1899 (21.6) 3.6 19,103.5 (20.2) 1698.8 (19.5) 1.9
 Open heart surgery 6812 (7.2) 684 (7.8) 2.1 6904.6 (7.3) 646.6 (7.4) 0.4
Abbreviations: AORTIC, other aortic valve; ASD, absolute standardized difference; AVR, aortic valve replacement; CABG, coronary artery bypass graft; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disorder; IPTW, inverse probability of treatment weighting; LAD, left anterior descending; MULT, multiple valve replacement; MVR, mitral valve replacement; MVV, mitral valve repair; PCI, percutaneous coronary intervention.

Each patient who experienced a handover was weighted by the inverse of his/her propensity (probability) of experiencing a handover, and each patient not experiencing a handover was weighted by his/her probability of not experiencing a handover (1 − his/her propensity of experiencing a handover). These weights were then stabilized to obtain appropriate estimates of the variance of the main effect. The ability of the propensity model to create weights that substantially modified the selection bias was tested by comparing the distributions of all patient risk factors for patients with and without handovers before and after IPTW.11–14 A standardized difference of <10% was regarded as ideal, but differences of <15% were deemed to be acceptable. Outcomes of interest in the study (short-term mortality, complications, 30 day readmission, length of stay) were then compared for patients with and without complete handovers after adjustment. Adjusted risk ratios (ARR) were used for mortality, readmissions, and complications, and an adjusted difference in means was used for the continuous measure (length of stay). Bonferroni corrections were used to account for multiple comparisons among the nonprimary outcomes (significance criterion of 0.05/3 = 0.017).

We determined a priori that a 10% relative difference in outcomes for patients with and without handovers would be required to constitute clinical importance regardless of statistical significance. The sample sizes after weighting (8732 for patients with handovers and 94,442 for patients without handovers) were sufficiently large to assure a power of 80% with a type I error of 0.05 to detect a 10% difference for all outcomes except mortality (see bottom of Table 2 for more detail). A P value of <.05 before the Bonferroni correction was considered to be statistically significant for all hypotheses except interactions, for which a P value of .15 was used.

Table 2. - Association of Mortality and Morbidity With Handover Status Before and After IPTW
Outcomes Before IPTW After IPTW
Handover (N = 8798)
Proportion (95% CI)
No Handover (N = 94,304)
Proportion (95% CI)
Risk Ratio (95% CI) P Handover (N = 8732)
Proportion (95% CI)
No Handover (N = 94,442)
Proportion (95% CI)
Risk Ratio (95% CI) P
In-hospital/30 d mortality 3.60 (3.21–3.99) 2.38 (2.29–2.48) 1.51 (1.35–1.70) <.001 2.86 (2.51–3.21) 2.48 (2.38–2.58) 1.15 (1.01–1.31) .03
Complications 27.4 (26.4–28.3) 22.0 (21.8–22.3) 1.24 (1.20–1.29) <.001 24.1 (23.2–25.0) 22.3 (22.1–22.6) 1.08 (1.04–1.12) <.001
30 d readmissiona 14.3 (13.5–15.0) 14.3 (14.1–14.5) 0.99 (0.93–1.04) .96 13.7 (13.0–14.4) 14.4 (14.1–14.6) 0.95 (0.90–1.00) .09
Length of stay (d)b 9.40 (9.24–9.57) 8.37 (8.32–8.42) 1.0 (0.8–1.2) <.001 8.79 (8.64–8.95) 8.46 (8.41–8.50) 0.3 (0.1–0.5) <.001
Sample size needed to detect 10% difference (RR = 1.1) between 2 proportions with group weights 1 and 11, reference proportion 0.025 (mortality), and power 0.8 is 426,552. Sample size needed to detect 10% difference (RR = 1.1) between 2 proportions with group weights 1 and 11, reference proportion 0.22 (complication), and power 0.8 is 38,112. Sample size needed to detect 10% difference (RR = 1.1) between 2 proportions with group weights 1 and 11, reference proportion 0.14 (readmission), and power 0.8 is 66,564. Sample size needed to detect 5% (0.4 d) difference between 2 means with group weights 1 and 11, reference mean 8.4 (LOS), and power 0.8 is 45,120.
Abbreviations: CI, confidence interval; IPTW, inverse probability of treatment weighting; LOS, length of stay; RR, risk ratio.
aAnalysis excludes death at the index surgery.
bInstead of risk ratio, risk difference is reported.

Analyses of interactions between levels of each baseline variable and handover (versus nonhandover) were undertaken for type of surgery, surgical priority, hospital teaching status, surgery start time, surgery duration, age, sex, body mass index, acute myocardial infarction, and a few comorbidities to test for homogeneity of the handover effect. All hypotheses tests were 2 sided, and all analyses were conducted in SAS version 9.4 (SAS Institute, Cary, NC).


Of the 103,102 patients in the study, 8798 (8.5%) experienced complete handovers, and 94,304 (91.5%) did not experience a complete handover. The percent of patients receiving a handover varied from 6.4% for isolated valve surgery to 10.6% for valve/CABG surgery. There was no significant trend in the percent of handovers during the 7-year period (P = .88 for trend).

Baseline Characteristics Before and After Weighting

Table 1 presents the baseline characteristics for patients with and without handovers before and after IPTW. Before weighting, patients with handovers had longer surgical procedures, were less likely to have undergone mitral valve replacements, were less likely to be elective/more likely to be urgent or emergency patients, were more likely to have 3-vessel or left main disease, and were significantly less likely to have a surgery start time in the morning hours. After IPTW, these differences were largely removed (with only 2 absolute standardized differences >10%).

Primary Outcome

Table 2 and the Figure contain outcomes for the primary outcome (in-hospital/30 day mortality), complications, 30 day readmissions, and length of stay for patients with and without handovers. For in-hospital/30 day mortality, the overall rate for the population of interest was 2.5%, and patients who experienced handovers had a higher unadjusted in-hospital/30 day mortality rate.

Hazard ratios for outcomes before and after inverse probability of treatment weighting. CI indicates confidence interval; IPTW, inverse probability of treatment weighting; OR, odds ratio; RR, risk ratio.

After adjustment for confounding using IPTW, we found a difference in in-hospital/30 day mortality by handover status (2.86% vs 2.48%, ARR = 1.15 [1.01–1.31]). The variables for which heterogeneity of the handover effect was observed were surgery duration, age, and chronic obstructive pulmonary disorder (COPD) (Supplemental Digital Content, Table 1, Patients who had shorter surgery durations were younger (≤65) or who did not have COPD had higher mortality when they had handovers.


Complication rates before and after adjustment are presented in Table 2 and the Figure. After adjustment, the complication rate for patients with handovers was statistically higher than the rate for patients without handovers (24.1% vs 22.3%, ARR = 1.08 [1.04–1.12]), but this difference (24.1%–22.3% = 1.8%) was <10% higher than the lower rate (24.1%), and was not judged to be clinically significant. The variables for which heterogeneity of the handover effect was observed were surgery priority, surgery duration, age, body mass index, and diabetes duration, age, and COPD (Supplemental Digital Content, Table 2,


Results for readmission within 30 days are also presented in Table 2 and the Figure. The overall rate of readmission was 14.3%. After adjustment, there were no differences in readmission rates (13.7% vs 14.4%, ARR = 0.95 [0.90–1.00]). The variables for which heterogeneity of the handover effect was observed were age, congestive heart failure (CHF), and cerebrovascular disease (Supplemental Digital Content, Table 3, Patients with handovers who had lower readmission rates than patients without handovers were of intermediate age (65–75) or did not have CHF (Supplemental Digital Content, Table 3,

Length of Stay

There were statistically lower lengths of stay for patients without handovers both before and after adjustment (8.8 vs 8.5 days, adjusted risk difference [ARD] = 0.3 [0.1–0.5]) using IPTW (Table 2). However, the adjusted difference was within 10% of the lower value and was not judged to be clinically significant. The variables for which heterogeneity of the handover effect was observed were surgery duration, age, and sex (Supplemental Digital Content, Table 4,


Handover of care from a sleep-deprived anesthesiologist to one who is well-rested can have beneficial effects on the quality of care. On the other hand, handovers can possibly jeopardize patient outcomes due to the loss of continuity of care and the inability to fully and accurately communicate details about the operation and patient.2–7 The purpose of our study was to examine differences in outcomes based on whether complete handovers of care from one anesthesiologist to another occurred.

There are several earlier studies on the topic. For example, Jones et al2 used multi-institutional data from Ontario, Canada, between 2009 and 2015 to examine differences in short-term outcomes for surgical patients (including cardiac surgery) who did and did not experience anesthesiologist handovers. They found that a total of 1.9% (1.4% for cardiac surgery) patients underwent surgery with complete anesthesia handovers. After adjustment for factors, such as surgery duration and surgery start time, as well as numerous patient risk factors for adverse outcomes, patients with complete handovers had a higher rate of their primary outcome (death, complications, or readmission within 30 days), all-cause death, and major complication, but not readmission within 30 days. For cardiac surgery, there was no difference in the primary outcome or in readmission within 30 days. However, patients with complete handovers had higher all-cause mortality rates within 30 days.2

Hudson et al3 studied the relationship between anesthesia handovers and adverse outcomes for cardiac surgery patients in a single hospital in Ottawa, Canada, between 1999 and 2009. Findings based on 7137 propensity score–matched patients were that in-hospital mortality was higher in the handover group, as was the major morbidity rate.3 In a single-institution study in Minnesota, Hyder et al4 studied outcomes of 927 patients undergoing colorectal surgery with and without anesthesia handovers. The authors found that the occurrence of death or any one of several major complications was significantly associated with the number of attending anesthesiologists.

Our study used clinical data from all 35 nonfederal cardiac surgery hospitals in New York during a 7-year period to examine short-term outcomes (mortality, major complications, readmissions, and length of hospital stay) for >100,000 cardiac surgery patients. Although it would have been possible to create a composite outcome, despite the fact that major complications were associated with, by far, the highest adverse outcome rate, and mortality was the outcome with clearly the highest severity, the statistical power was sufficient to examine each of the outcomes separately.15

After adjustment using IPTW to reduce selection bias, we found that there was a significant difference in in-hospital/30 day mortality (2.86% vs 2.48%, ARR =1.15 [1.01–1.31]). Thus, for every 1000 anesthesia handovers that occur, it is estimated that there would be 4 more deaths than there would have been if there had been no handovers. There was no statistically significant difference in readmissions, and the differences in complications and length of stay were not clinically meaningful. These findings are similar to those of Jones et al,2 who also found lower mortality for patients without handovers, but no differences in other outcomes. It is also notable that our study found that some groups of patients with lower risk (shorter surgery durations, younger, or without COPD) had lower mortality when they did not experience handovers.

There are many reasons why adverse outcomes can occur during and after cardiac surgery. There are numerous intraoperative processes of care that are essential to avoid mortality, complications, readmissions, and prolonged length of stay. These include insulin treatment for high glucose levels, temperature control, maintenance of acceptable hematocrit/hemoglobin levels, and monitoring of urine output during and after surgery. For example, the absence of appropriate antibiotic prophylaxis leads to postoperative complications like sternal wound infections, major disruption of surgical wounds, postoperative respiratory problems, and possibly 30 day readmissions.

All of these processes of care can be compromised during the course of anesthesia handovers. In general, misleading information, absence of a safety culture, ineffective communication methods, lack of time, poor timing, interruptions, distractions, lack of standardized procedures, and insufficient staffing can all be impediments to optimal information transfer during anesthesia handovers. Information on medication initiation and redosing, like antibiotics, heparin, or information on the patient’s lack of sensitivity or exaggerated responses to medications, including the response to pressors, amnestic, and analgesic agents, may easily be lost during care transitions.

It is also true that an anesthesiologist who is fatigued may also have difficulty in managing the processes of care just mentioned. So there is a trade-off between fatigue on the one hand and the dangers of inadequate communication on the other hand. Our study found that mortality was higher among patients experiencing handovers, but other outcomes (complications, readmissions, length of stay) were not clinically higher. However, as noted below, there are caveats to the study that could have affected these relative outcomes.


There are several limitations to this study. It is an observational study and is, therefore, subject to selection bias because of its nonrandomized nature. This could bias the study against cases with handovers. We attempted to minimize this bias by using IPTW, and we were able to demonstrate that the resulting patients were well matched with respect to numerous important correlates of adverse outcomes. Nevertheless, there are inevitably factors related to outcomes that were not available to us, and unmeasured confounding is undoubtedly present. We were unable to measure the individual competence and experience level of the anesthesiologists and were also unable to measure postoperative risk, blood loss, and pressor use. We could not determine when handovers occurred, and this may have been associated with adverse outcomes. Also, we had no information about the number of handovers that occurred for each patient experiencing handovers, or the quality or nature of the handovers. There was no information available about the presence of anesthesia trainees or certified registered nurse anesthetists (CRNAs) during the procedures, or on exchanges that occurred among them. Thus, there could have been more or less continuity of care than there appeared to be. Centers with more patients and more handovers receive more weight in the study, and this could introduce a bias. The complications (Supplemental Digital Content, Appendix 1, were obtained from administrative data (SPARCS), and the quality of those data is not monitored as carefully as clinical registry data. The results apply to New York State and may not be generalized to other settings, particularly settings with vastly different handover rates.


Cardiac surgery patients in New York who had complete anesthesia handovers experienced higher short-term mortality rates than patients without handovers, but there were no clinically important differences in 30-day readmission rates, complications, or length of stay. Efforts to minimize unnecessary handovers and improve the quality of communications during handovers when they are necessary based on fatigue and other concerns will likely benefit patients undergoing cardiac surgery. Future studies should examine factors such as when handovers occur and when they should occur, whether residents/fellows are involved and what their impact is, and how information is transmitted during the course of handovers.


Name: Edward L. Hannan, PhD.

Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.

Conflicts of Interest: None.

Name: Zaza Samadashvili, MD.

Contribution: This author helped design the study, conduct the study, analyze the data, and write the manuscript.

Conflicts of Interest: None.

Name: Thoralf M. Sundt III, MD.

Contribution: This author helped review the analysis of the data and approved the final manuscript.

Conflicts of Interest: None.

Name: Leonard Girardi, MD.

Contribution: This author helped review the analysis of the data and approved the final manuscript.

Conflicts of Interest: None.

Name: Joanna Chikwe, MD.

Contribution: This author helped review the analysis of the data and approved the final manuscript.

Conflicts of Interest: J. Chikwe received speaker honoraria from Edwards Lifesciences.

Name: Andrew Wechsler, MD.

Contribution: This author helped review the analysis of the data and approved the final manuscript.

Conflicts of Interest: None.

Name: David H. Adams, MD.

Contribution: This author helped review the analysis of the data and approved the final manuscript.

Conflicts of Interest: The Icahn School of Medicine at Mount Sinai (affiliation of D. H. Adams) receives royalty agreements for intellectual property related to the development of valve repair products from Edward Lifesciences and Medtronic.

Name: Craig R. Smith, MD.

Contribution: This author helped review the analysis of the data and approved the final manuscript.

Conflicts of Interest: None.

Name: Jeffrey P. Gold, MD.

Contribution: This author helped review the analysis of the data and approved the final manuscript.

Conflicts of Interest: None.

Name: Stephen J. Lahey, MD.

Contribution: This author helped review the analysis of the data and approved the final manuscript.

Conflicts of Interest: None.

Name: Desmond Jordan, MD.

Contribution: This author helped design the study, reviewed the analysis of the data, and approved the final manuscript.

Conflicts of Interest: None.

This manuscript was handled by: Stefan G. De Hert, MD.


1. Howard SK, Rosekind MR, Katz JD, Berry AJ. Fatigue in anesthesia: implications and strategies for patient and provider safety. Anesthesiology. 2002;97:1281–1294.
2. Jones PM, Cherry RA, Allen BN, et al. Association between handover of anesthesia care and adverse postoperative outcomes among patients undergoing major surgery. JAMA. 2018;319:143–153.
3. Hudson CC, McDonald B, Hudson JK, Tran D, Boodhwani M. Impact of anesthetic handover on mortality and morbidity in cardiac surgery: a cohort study. J Cardiothorac Vasc Anesth. 2015;29:11–16.
4. Hyder JA, Bohman JK, Kor DJ, et al. Anesthesia care transitions and risk of postoperative complications. Anesth Analg. 2016;122:134–144.
5. Terekhov MA, Ehrenfeld JM, Dutton RP, Guillamondegui OD, Martin BJ, Wanderer JP. Intraoperative care transitions are not associated with postoperative adverse outcomes. Anesthesiology. 2016;125:690–699.
6. Saager L, Hesler BD, You J, et al. Intraoperative transitions of anesthesia care and postoperative adverse outcomes. Anesthesiology. 2014;121:695–706.
7. O’Reilly-Shah VN, Melanson VG, Sullivan CL, Jabeley CS, Lynde GC. Lack of association between introperative handover of care and postoperative complications: a retrospective operational study. BMC Anesthesiol. 2019;19:182.
8. Mascha EJ, Sessler DJ. Design and analysis of studies with binary-event composite endpoints: guidelines for anesthesia research. Anesth Analg. 2011;112:1461–1471.
9. Rosenbaum PR, Rubin D. The central role of propensity score in observation studies for causal effects. Biometrika. 1983;70:41–55.
10. Rosenbaum PR. Model-based direct adjustment. J Am Stat Assoc. 1987;82:387–394.
11. Austin PC. The performance of different propensity-score methods for estimating differences in proportions (risk differences or absolute risk reductions) in observational studies. Stat Med. 2010;29:2137–2148.
12. Haukoos JS, Lewis RJ. The propensity score. JAMA. 2015;314:1637–1638.
13. Ukoumunne OC, Williamson E, Forbes AB, Gulliford MC, Carlin JB. Confounder-adjusted estimates of the risk difference using propensity score-based weighting. Stat Med. 2010;29:3126–3136.
14. Austin PC, Stuart EA. Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Stat Med. 2015;34:3661–3679.
15. Curtis LH, Hammill BG, Eisenstein EL, Kramer JM, Anstrom KJ. Using inverse probability-weighted estimators in comparative effectiveness analyses with observational databases. Med Care. 2007;45:S103–S107.

Supplemental Digital Content

Copyright © 2020 International Anesthesia Research Society