Perioperative Morbidity and Mortality of Patients With COVID-19 Who Undergo Urgent and Emergent Surgical Procedures

Anne Knisely, MD; Zhen Ni Zhou, MD, PhD; Jenny Wu, BS; Yongmei Huang, MD, MPH; Kevin Holcomb, MD; Alexander Melamed, MD, MPH; Arnold P. Advincula, MD; Anil Lalwani, MD; Fady Khoury-Collado, MD; Ana I. Tergas, MD, MPH; Caryn M. St. Clair, MD; June Y. Hou, MD; Dawn L. Hershman, MD; Mary E. D'Alton, MD; Yolanda Ya-Chin Huang, MD, PhD; Jason D. Wright, MD


Annals of Surgery. 2021;273(1):34-40. 

In This Article


Patients and Procedures

We performed a retrospective cohort study of patients who underwent surgery at New York Presbyterian Hospital Columbia University Irving Medical Center and Weill Cornell Medical Center during the period of widespread community infection of COVID-19 in the New York City region. Data on all patients who underwent surgery between March 17, 2020 and April 15, 2020 were abstracted from electronic medical records. Patients were classified as COVID-19 positive if they had preoperative documentation of a positive COVID-19 test or documentation of a positive COVID-19 test within 21 days (before or after) their surgical procedure.[11] Institutional Review Board approval was obtained from both Columbia University Vagelos College of Physicians and Surgeons and Weill Cornell Medical College.

On March 13, 2020, as the number of cases of COVD-19 in the region began increasing, hospital level guidance was put in place to limit the performance of elective surgical procedures that could be delayed for more than 2 weeks. As the number of confirmed COVID-19 cases continued to rise, revised policy guidance was implemented on March 23, 2020 to further limit procedures to only those cases that were urgent or emergent and needed to be performed within 48 hours. During this period, all surgical procedure requests were reviewed by clinical and administrative staff.

On April 1, 2020, a program to allow preoperative testing of patients for COVID-19 was initiated. The decision to perform COVID-19 testing was at the discretion of the attending surgeon. Testing for COVID-19 was performed by RT-PCR using the cobas SARS-CoV-2 test by Roche Diagnostics until April 10, 2020 after which time testing using the Xpert Xpress SARS-CoV-2 test by Cepheid became available. Patients with COVID-19 were categorized based on the date of the first laboratory confirmation of the disease in relation to the date of surgery. Among the COVID-19 positive patients, those with any of the following symptoms before surgery were classified as symptomatic: fever, myalgia, fatigue, rigor cough, dyspnea, sore throat, diarrhea, chest pain, anosmia, or loss of taste.

Clinical Characteristics and Outcomes

Demographic characteristics included date of the procedure, age, sex, race/ethnicity, body mass index, and tobacco use. Preoperative clinical characteristics including significant medical comorbidities and medications were recorded. The type of surgical procedure (ENT, cardiac, colorectal, general surgical, gynecologic, neurosurgery, orthopedic, skin/breast, thoracic, urologic, vascular, and other) and urgency (urgent vs emergent) of the procedure were noted. Emergent procedures were those operations that in general required performance within 12 hours to minimize the risk of serious morbidity or morality. Intraoperative variables included American Society of Anesthesiology (ASA) class, type of intubation (rapid sequence) and airway, type of anesthesia, vital signs, transfusion requirements, and operative time.

Perioperative complications were assessed using the criteria of the American College of Surgeon's National Surgical Quality Improvement Project (NSQIP) classification schema.[12] A composite of serious complications was created and included the occurrence of any of the following: myocardial infarction, cardiac arrest, shock, respiratory failure, stroke, pneumonia, or venous thromboembolism (either pulmonary embolism or deep venous thrombosis). In addition to surgical complications, we examined return to operating room, intensive care unit (ICU) admission, transfusion (both intraoperative and postoperative) and in-hospital mortality. Given that follow-up time was short, we only measured complications during the index hospitalization.

Statistical Analysis

The clinical and demographic characteristics and outcomes were compared between the COVID-19 positive and COVID-19 negative cohorts using χ 2 or Fisher exact tests. Marginal multivariable log-linear models with Poisson distributions were used to examine the association between COVID-19 status and ICU admission, serious complications, and death while controlling for other clinical and demographic variables and hospital clustering. Purposeful selection was used for model building. In the multivariable model, patient's age was included a priori in all models. Other potential confounders were investigated for their effect on the association between COVID-19 infection and each outcome. We found that sex, comorbidity, functional status, ASA class, and urgency of the procedure changed the risk estimates (>10%) for ICU admission and serious complications, and were therefore included in the multivariable models. However, for the outcome of death, we only included ASA class to avoid over-fitting the model. Parsimonious models in which only ASA classification and COVID-19 status were included are also reported. Results are reported as risk ratios (RR) with 95% confidence intervals.

Among COVID-19 positive patients, the severity of disease was classified as mild, severe or critical based on the criteria of the Chinese Center for Disease Control and Prevention.[2] Mild disease including patients without pneumonia or mild pneumonia, severe disease as those with tachypnea (respiratory frequency ≥30/min), hypoxia (oxygen saturation ≤93%), a partial pressure of arterial oxygen to fraction of inspired oxygen ratio <300, or lung infiltrates of >50% within 24 to 48 hours, and critical disease as those with respiratory failure, septic shock or multiple organ dysfunction.[2]

We performed a number of sensitivity analyses. First, we performed a nested 1:1 case-control analysis in which each COVID-19 positive patient was matched to one COVID-19 negative control who underwent a similar surgical procedure with similar ASA status and age (within 10 years). McNemar χ 2 tests were used to compare the discordant pairs on the outcomes of ICU admission, severe complications, and death. Conditional logistic regression models were developed to estimate the odds ratio (OR) of the outcomes of interest for the COVID-19 positive patients where OR were reported.

Second, we compared the perioperative outcomes of the COVID-19 positive patients to expected outcomes based on predicted probabilities of adverse events using the ACS NSQIP Surgical Risk Calculator.[13] The ACS NSQIP Surgical Risk Calculator uses patient-level clinical, demographic, and procedural characteristics to estimate the predicted probability of an adverse event. The risk calculator has been tested and validated for a variety of procedures.[14,15] Finally, our primary analysis allowed a window of 21 days between the date of surgery and the date of COVID-19 testing to maximize capture of asymptomatic or mildly symptomatic COVID-19 patients. We performed a sensitivity analysis in which only patients who tested positive for COVID-19 within 14 days (before or after) were considered in the COVID-19 positive cohort.

We estimated the predicted probability of adverse events among COVID-19 positive patients and compared this estimate to the observed rate of each event. All analyses were performed using SAS version 9.4 (SAS Institute Inc, Cary, North Carolina). All tests were 2-sided, and a P-value <0.05 was considered statistically significant.