Design and setting
After Research Ethics Board approval from all centres, we conducted a multicenter observational cohort study based on a waived consent model in nine university hospitals in the Province of Québec from March 13, 2020, to June 19, 2020. This report follows STROBE guidelines for reporting observational studies .
We included all consecutive patients undergoing surgery who had tested positive for SARS-CoV-2 preoperatively. We defined positive SARS-CoV-2 infection by any positive Polymerase Chain Reaction (PCR) test (from either an oronasopharyngeal swab or an endotracheal aspirate) either before surgery or within 72 h after surgery and defined being symptomatic by the presence of any patient-reported COVID-19-related symptoms (fever, respiratory distress, etc).
We also included all patients undergoing surgery during the same period who were suspected of having COVID-19 at time of surgery (but subsequently tested negative) and all patients who had recovered from COVID-19. We defined suspected patients by the presence of the same COVID-19-related symptoms with unknown SARS-CoV-2 infection status at time of surgery (but later confirmed negative for SARS-CoV-2) which prompted the operating room team to take specific COVID-19-related precautions. We defined recovery as a patient who previously had a positive PCR test and then had two negative PCR tests for SARS-CoV-2 before surgery (or one negative test performed at least 14 days before surgery and the absence of symptoms at time of surgery). Patients were identified through the electronic medical data system or the operating room database specific to each site.
To address our primary objective, we first reported data on COVID-19 patients. To address our secondary objectives, we used the presence of COVID-19-related symptoms as an exposure variable and reported stratified data based on this variable for COVID-19 patients. To address another secondary objective, we used the COVID-19 disease status (confirmed, suspected, or recovered) as an exposure variable.
We collected baseline characteristics of the COVID-19 presentation such as the presence of symptoms at time of surgery, number of days since first symptoms or diagnosis, preoperative need for oxygen or invasive mechanical ventilation, and received treatments (e.g., antiviral agents, steroids, antibiotics). We also recorded demographic characteristics, baseline comorbidities, type of surgery, urgency of surgery, baseline laboratory values, preoperative Sequential Organ Failure Assessment (SOFA) score and intraoperative variables.
Our primary outcome was 30-day survival after surgery. Our secondary outcomes were the postoperative occurrence of respiratory complications (atelectasis, pneumonia, Acute Respiratory Distress Syndrome [ARDS], and pulmonary aspiration), non-pulmonary infectious complications, acute kidney injury, thrombotic-associated complications (pulmonary embolism, myocardial infarction, stroke and cardiac arrest), hospital length of stay, 30-day mechanical ventilation-free days, 30-day organ dysfunction-free days, and any new ICU admission during the index admission. The endpoints we examined regarding the overall impact on surgical care were the number of cases performed during the inclusion period; the duration of the procedures; the required time to allow for operating room preparation, cleaning, and patient extubation; and other post-anesthesia care procedures. We collected similar data in all sites during a comparable period of observation (i.e., same dates) of the preceding year.
We classified procedures into the following categories: neurosurgical (head and spine), cardiac, thoracic, major vascular (intrathoracic and/or intra-abdominal), non-vascular abdominal (laparotomy or laparoscopy), urogenital (urology and/or gynecology), non-spine orthopedic, peripheral vascular, otolaryngology-head-neck and other. Major surgery was defined as any neurosurgical, cardiac, thoracic, major vascular (cervical, intrathoracic and/or intra-abdominal) or non-vascular abdominal surgery (any intra-abdominal surgery). We also collected data on the surgical approach (minimally invasive, including laparoscopy, or invasive). Urgency of surgery was defined as the need to be performed within 24 h as requested by surgeons.
We measured survival at 30 days (within hospital or later after discharge if data available) and censored patients at last visit seen alive or at 30 days if still alive. We used existing definitions for postoperative pulmonary complications outcome (atelectasis, pneumonia, ARDS, pneumothorax) . Non-pulmonary infectious complications were defined as any infection requiring antibiotics for > 72 h. Acute kidney injury was defined by the creatinine difference of the KDIGO-AKI criteria . We defined thrombotic-related complications as reported by the treating physicians including pulmonary embolism, myocardial infarction, stroke or cardiac arrest. Mechanical ventilation was defined as the need for mechanical support (non-invasive or invasive). Organ dysfunction was defined as the need for vasopressors, mechanical ventilation and the use of renal replacement therapy after surgery . The 30-day freedom from adverse outcomes included any day without the corresponding outcome.
Data sources and management
Data was either collected prospectively by the clinical and the research team or retrospectively collected the days following surgery. Data was entered at each site into a centralized electronic database following a manual of standard operating procedures and individual site training. Outcomes were adjudicated by a physician at each site. The surgical impact data was extracted as aggregate data from operating room administrative systems of each centre.
We included a convenience sample of all eligible patients who underwent surgery during the observation period. Due to the limited sample size, the reported analyses are primarily descriptive. We reported continuous variables as mean (standard deviations (SD)) or median [interquartile range [IQR]] for skewed distributions) and categorical variables as proportions. We reported 30-day survival using Kaplan-Meier estimates with 95% confidence intervals (CI) stratified by subgroups. To compare symptomatic COVID-19 patients to asymptomatic ones, and COVID-19 patients to suspected ones and those who had recovered, we performed log-rank tests. Finally, we conducted a post hoc sensitivity analysis by removing patients for whom a tracheotomy was the surgical procedure performed, since they may have a different postoperative trajectory. We set our alpha level at 0.05. All analyses were performed with R statistical software (R Core Team, version 4.0.2).