Perioperative Outcomes Among Surgeons Who Operated the Night
Prior
Eric Sun, MD/PhD
Michelle M. Mello, JD/PhD
Michelle T. Vaughn, MPH
Sachin Kheterpal, MD/MBA
Mary T. Hawn, MD
Justin B. Dimick, MD/MPH
Anupam B. Jena, MD/PhD
Disclosures
• Research support from NIDA, NIA, and AHRQ
• Advisory Board of Lucid Lane, LLC
• Consulting fees from Analysis Group, Inc.
Motivation
• Fatigue has been shown to adversely affect performance across a variety of domains
(Neu et al., 2011; Goode 2003; Fullagar 2015)
• Concerns about fatigue and patient outcomes (Barger 2006; Gates 2018)
• While residents face duty hour restrictions, attendings do not
• Is this safe for patients?
3
Summary of Findings
• Operating overnight was not associated with an increased risk of in-hospital death
or major complication among the subsequent day’s cases
• Small decrease (?) among subsequent day’s cases
• Results generally robust across a variety of subgroup and sensitivity analyses
4
Outline
• Background
• Sample
• Statistical Approach
• Results
• Conclusion
5
Outline
• Background
• Sample
• Statistical Approach
• Results
• Conclusion
6
Background
• Previous work
• Gates (2018): Systematic review of 47 studies, generally found no association
between fatigue and poorer outcomes
• Govindarajan (2015): Retrospective analysis of ~40k patients undergoing surgery at
a Canadian province: no association between operating overnight and outcomes on
the next day’s cases
• Limitations of previous work
• Underpowered to study mortality/complications
• Subject to confounding (surgeons who work overnight are different from those who
don’t)
• Self-reported overnight operating
• Overly stringent definitions f operating overnight
7
Outline
• Background
• Sample
• Statistical Approach
• Results
• Conclusion
8
Sample
• Inclusion criteria
• Cases between January 1, 2010 and August 30, 2020
• Nonmissing data on comorbidities, mortality, surgery CPT code, and whether the
surgeon operated the previous night
• Nonmissing data on case times
• “Major therapeutic procedure” as defined by AHRQ
9
Sample
• Initial Sample 2,110,956 cases
• Exclusion criteria
• Surgeons who never operated overnight (n=1,332,069)
• >1 attending surgeon (n=54,838)
• Missing data: age (n=89), ASAPS (n=7,902)
• ASA 5/6 (n=1,351)
• Low risk/out of OR procedures
• Upper endoscopy (n=47,127)
• Lower endoscopy (n=22,262)
• Opthalmologic surgeries (n=41,851)
• CVC access (n=16,435)
• Surgery CPTs with <100 observations (n=58,753)
• Surgeons with <100 observations (n=30,295)
10
Sample
• Final Sample
• 498,234 cases
• 1,131 surgeons
• 20 US institutions
11
Outline
• Background
• Sample
• Statistical Approach
• Results
• Conclusion
12
Statistical Approach
• Primary Outcome: In-Hospital Death or Major Complication
• Major Complications: Thromboembolic events, MI, stroke, PNA, sepsis
• Identified by discharge diagnosis codes
• Secondary Outcomes
• Surgery length (surgery end-surgery start)
• Death
• Major complications
• Minor complications (UTI, SSI)
13
Statistical Approach
• Exposure
• Having operated overnight before the current day’ surgery
• “Overnight” defined as 2300-0700
• Measured using surgery start and end times
• Baseline model was yes/no for any overnight operating time
• Sensitivity analyses examined alternate definitions
14
Statistical Approach
• Multivariable Linear Regression
• Key innovation is use of ”surgeon fixed effects”
• Reduces confounding because we are not comparing surgeons who operate
overnight to those who do not
• Compare a given surgeon’s cases where the surgeon operated overnight
compared to the cases where the same surgeon did not
• Additional variables include
• Patient characteristics (Age, sex, comorbidities)
• Surgery CPT Fixed Effects
• Hour of Surgery
15
Statistical Analyses
• Subgroup Analysis
• “High risk” patientsupper 25
th
percentile of predicted mortality/major
complication
• Sensitivity Analysis
• Two alternate approaches to measuring intensity of operating overnight
• Five groups:
• no overnight operating
• 0-2h
• 2-4h
• 4-6h
• 6-8h
• Introduce hours operated overnight as a linear variable
16
Outline
• Background
• Sample
• Statistical Approach
• Results
• Conclusion
17
Surgeon Operated the Previous Night
(N=13,098)
Surgeon did Not Operate the Previous Night
(N=485,136)
p-value/g-value
Patient demographics
Age, mean (s.e)
53.91 (0.15) 55.32 (0.02) <0.001/0.09
Male, % (s.e.)
52.32 (0.44) 46.72 (0.07) <0.001/0.11
Comorbidities, % (s.e.)
Congestive Heart Failure
8.73 (0.25) 5.32 (0.03) <0.001/0.15
Arrhythmia
18.46 (0.34) 13.07 (0.05) <0.001/0.16
Valvular Disease
7.72 (0.23) 5.18 (0.03) <0.001/0.11
Pulmonary Circulatory Disorders
4.57 (0.18) 2.94 (0.02) <0.001/0.10
Peripheral Vascular Disease
9.04 (0.25) 5.36 (0.03) <0.001/0.16
Hypertension, Uncomplicated
32.75 (0.41) 29.38 (0.07) <0.001/0.07
Hypertension, Complicated
11.70 (0.28) 7.16 (0.04) <0.001/0.17
Paralysis
1.70 (0.11) 1.08 (0.01) <0.001/0.06
Neurological Disorder
6.66 (0.22) 4.55 (0.03) <0.001/0.10
Chronic Pulmonary Disease
14.05 (0.30) 11.61 (0.05) <0.001/0.08
Diabetes, Uncomplicated
11.66 (0.28) 9.95 (0.04) <0.001/0.06
Diabetes, Complicated
5.10 (0.19) 3.09 (0.02) <0.001/0.12
Hypothyroidism
7.66 (0.23) 8.29 (0.04) 0.010/0.02
Chronic Renal Failure
12.35 (0.29) 7.73 (0.04) <0.001/0.17
Liver Disease
5.47 (0.20) 4.00 (0.03) <0.001/0.07
Peptic Ulcer Disease
0.73 (0.07) 0.58 (0.01) 0.030/0.02
AIDS
0.41 (0.06) 0.24 (0.01) <0.001/0.03
Lymphoma
1.00 (0.09) 1.38 (0.02) <0.001/0.03
Metastatic Disease
6.06 (0.21) 13.99 (0.05) <0.001/0.23
Solid Tumor
13.83 (0.30) 24.65 (0.06) <0.001/0.25
Rheumatoid Arthritis
2.79 (0.14) 2.32 (0.02) <0.001/0.03
Coagulopathy
7.59 (0.23) 4.23 (0.03) <0.001/0.17
Obesity
16.44 (0.32) 14.96 (0.05) <0.001/0.04
Weight Loss
5.79 (0.20) 3.81 (0.03) <0.001/0.10
Fluid or Electrolyte Disorder
17.58 (0.33) 9.94 (0.04) <0.001/0.25
Blood Loss
1.21 (0.10) 0.85 (0.01) <0.001/0.04
Iron Deficiency Anemia
2.72 (0.14) 2.07 (0.02) <0.001/0.05
Alcohol Abuse
1.12 (0.09) 0.66 (0.01) <0.001/0.06
Drug Abuse
4.21 (0.18) 2.42 (0.02) <0.001/0.11
Psychosis
0.75 (0.08) 0.51 (0.01) <0.001/0.03
Depression
11.35 (0.28) 9.95 (0.04) <0.001/0.05
ASA Score
Physical Status Score, mean (s.e.)
2.62 (0.01) 2.53 (0.00) <0.001/0.13
Emergent Case, % (s.e.)
8.70 (0.25) 2.60 (0.02) <0.001/0.37
High-Risk Patients, % (s.e.) 35.9 (0.4) 24.7 (0.1)
<0.001/0.26
18
19
4.8
5
5.2
5.4
5.6
5.8
6
6.2
6.4
6.6
Overnight Work No Overnight Work
Incidence (%)
Summary of Outcomes
Operating
Overnight
No Operating
Overnight
p
Primary Outcome 5.89 5.87 0.93
Surgery Length
(min)
112.7 117.4 0.020
Death (%) 0.87 0.89 0.82
Major Cx (%) 5.40 5.34 0.83
Minor Cx (%) 2.34 2.27 0.67
20
21
Hours operated night prior modeled as a categorical variable
Amount of time worked between 11pm and 7am
the previous night
Death or major
complication, adjusted %
(95% CI)
Absolute adjusted
difference (95% CI)
None (N=485,136) 5.87 (5.85 to 5.88) -
>0 to <2 hours (N=9,117) 6.05 (5.50 to 6.60)
0.18%
(-0.38% to 0.75%)
p=0.52
2 to <4 hours (N=2,116) 5.47 (4.35 to 6.59)
-0.40%
(-1.52% to 0.72%)
p=0.49
4 to <6 hours (N=883) 6.01 (4.17 to 7.85)
0.15%
(-1.70% to 1.99%)
p=0.88
6-8 hours (N=982) 5.21 (3.60 to 6.83)
-0.65%
(-2.27% to 0.97%)
p=0.43
Hours worked night prior modeled linearly
Absolute adjusted change in death or major
complication per additional hour worked (95%
CI)
-0.05% (-0.22% to 0.12%)
p=0.57
Conclusions
• Surgeons who operated overnight—even for extensive periods of time—did not have
worse outcomes when the operated the next day
• Results were robust to a variety of subgroup and sensitivity analyses
• Allowing surgeons who operated overnight to work the next day appears to be safe—
at least as it is practiced in this sample of (large) institutions
22
