key: cord-0879695-0w905c2x
authors: Abbott, T.E.F.; Fowler, A. J.; Dobbs, T. D.; Gibson, J.; Shahid, T.; Dias, P.; Akbari, A.; Whitaker, I. S.; Pearse, R. M.
title: Mortality after surgery with SARS-CoV-2 infection in England: A population-wide epidemiological study
date: 2021-06-11
journal: Br J Anaesth
DOI: 10.1016/j.bja.2021.05.018
sha: 2065ef5caa368e7363294a9420d38f0838d43f42
doc_id: 879695
cord_uid: 0w905c2x

Background The COVID-19 pandemic has heavily impacted elective and emergency surgery around the world. We aimed to confirm the incidence of perioperative SARS-CoV-2 infection and associated mortality after surgery. Methods Analysis of routine electronic health record data from National Health Service (NHS) hospitals in England. We extracted data from Hospital Episode Statistics in England describing adult patients undergoing surgery between 1st January 2020 and 28th February 2021. The exposure was SARS-CoV-2 infection defined by ICD-10 codes. The primary outcome measure was 90-day in-hospital mortality. Data were analysed using multivariable logistic regression adjusted for age, sex, Charlson co-morbidity index, index of multiple deprivation, presence of cancer, surgical procedure type and admission acuity. Results are presented as n (%) and odds ratios (OR) with 95% confidence intervals. Results We identified 2,666,978 patients undergoing surgery of whom 28,777 (1.1%) had SARS-CoV-2 infection. In total, 26,364 (1.0%) patients died in hospital. SARS-CoV-2 infection was associated with a much greater risk of death (SARS-CoV-2: 6,153/28,777 [21.4 %] vs No SARS-CoV-2: 20,211/2,638,201 [0.8%]; OR: 5.7 [95% CI: 5.5 – 5.9]; P<0.001). Amongst patients undergoing elective surgery 2,412/1,857,586 (0.1%) had SARS-CoV-2 of whom 172/2,412 (7.1%) died, compared with 1,414/1,857,586 (0.1%) patients without SARS-CoV-2 (OR: 25.8 [95% CI: 21.7 - 30.9]; P<0.001). Amongst patients undergoing emergency surgery 22,918/582,292 (3.9%) patients had SARS-CoV-2, of whom 5,752/22,918 (25.1%) died compared with 18,060/559,374(3.4%) patients without SARS-CoV-2 (OR: 5.5 [5.3 – 5.7]; P<0.001). Conclusions The low incidence of SARS-CoV-2 infection in NHS surgical pathways suggests current infection prevention and control policies are highly effective. However, the high mortality amongst patients with SARS-CoV-2 suggests these precautions cannot be safely relaxed.

Surgery is an essential treatment modality with more than 330 million surgical procedures performed worldwide every year. [1] [2] [3] However, the COVID-19 pandemic has led to substantial reductions in the volume of surgery performed. 4, 5 Recent estimates suggest that half of the 4.5 million expected surgical procedures in the English National Health Service (NHS) were cancelled or postponed in 2020. 4 This is partly driven by reallocation of resources to care for patients with COVID-19, but also by strict infection prevention and control procedures implemented to prevent patients becoming infected with SARS-CoV-2 whilst in hospital. 6 Patient self-isolation before surgery, pre-operative testing for SARS-CoV-2, creation of separate 'green zones' in hospitals to isolate non-infected patients, and personal protective equipment procedures all contribute to reductions in the efficiency of surgical care pathways, and substantial reductions in the volume of patients treated. 7 Early reports have suggested that patients with SARS-CoV-2 infection who undergo surgery are at much greater risk of postoperative pulmonary complications and death. 8, 9 The COVIDSurg collaborative undertook a large international study of outcomes for surgical patients infected with SARS-CoV-2 and reported mortality rates as high as 24% at the peak of the pandemic. 8 However, more recent data suggest the mortality risk for surgical patients with SARS-CoV-2 may be lower than originally thought, 9 especially for minor surgery in younger patients. 10 It remains unclear how SARS-CoV-2 infection affects outcomes after surgery, and whether high mortality rates reported in some studies relate to a change in surgical casemix during the pandemic, or to SARS-CoV-2 infection itself. At present there are no large studies describing surgical outcomes for contemporaneous patient groups with and without SARS-CoV-2 infection. Given the extensive disruption to surgical services and the likely excess mortality due to untreated cancer and other surgical diseases, the NHS is under significant pressure to relax infection prevention and control procedures to increase the volume of patients who can undergo surgery.

We used routine NHS electronic health record data to report the rate of SARS-CoV-2 infection amongst surgical patients during the pandemic, and the associated mortality and hospital stay.

Population-wide epidemiological study using routinely collected electronic health record data.

Setting NHS hospitals in England.

All patients aged 18 years or older who underwent a surgical procedure between 1 st January and 28 th February 2021.

We used pseudonymised record level Hospital Episode Statistics (HES) to identify eligible patients. This data source provides detailed data describing every episode of hospital care in England. Surgical procedures were identified using Office for Population Censuses Surveys version 4 (OPCS4) codes as described previously. 2, 4 These define procedures typically performed in an operating theatre or under general/regional anaesthesia (Appendix A).

Individuals entered the cohort on the date of their first operative procedure and were followed until the point of hospital discharge. Hospital discharge date was determined based on the discharge from continuous in-patient spells, which were constructed by mapping continuous in-patient episodes. 11 For patients remaining in hospital, their discharge date was right censored to the 28 th February 2021. The analysis was approved by the Health Research Authority (20/HRA/3121) and the NHS Digital Independent Group Advising on the Release of Data (DARS-NIC-375669-J7M7F).

The exposure of interest was a diagnosis of SARS-CoV-2 viral infection defined using the following ICD-10 codes: U07.1 (Virus identified), U07.2 (Virus not identified) or B97.2 (Other coronavirus as the cause of diseases classified elsewhere). Patients with SARS-CoV-2 were categorised as symptomatic if they had ICD-10 codes for concomitant respiratory illness or J o u r n a l P r e -p r o o f OPCS-4 codes indicating respiratory support (Supplementary table 1) . Patients with SARS-CoV-2 were categorised as not symptomatic if there were no associated ICD-10 codes for respiratory illness or OPCS-4 codes for respiratory support. We divided the exposure into three time points: Preoperative if SARS-CoV-2 codes were first recorded in an episode that finished within the 30 days prior to the index operation; Perioperative if a code was first recorded in the same episode as the index operation; and Postoperative if a code was first recorded in a subsequent healthcare episode prior to discharge within 30 days after the index operation.

The primary outcome was in-hospital death, censored at 90 days for patients remaining in hospital beyond this point. The secondary outcome was length of hospital stay, calculated as the number of days between initial operation date and discharge date of their continuous in-patient spell, censored at 90 days.

Age was defined as that recorded on the start date of the hospital episode including the first operative procedure. Charlson co-morbidity index was derived according to the Royal College of Surgeons mapping which includes a one-year look back file to capture diagnoses from prior admissions. 12 Classification of the type of surgical procedure was based on the first operative code. 4 Where multiple operative codes were associated with surgery on a single day, the highest ranked code was considered the principal surgical procedure.

Socioeconomic deprivation was defined according to lower super-output areas using the Index of Multiple Deprivation (IMD). 13 We classified missing IMD as its own category.

The incidence of SARS-CoV-2 infection among patients undergoing surgery was calculated by dividing the number of patients with an ICD-10 code indicating SARS-CoV-2 infection by the total number of patients in the cohort. Age-adjusted incidence was calculated using deciles of age. 14 To test for associations between SARS-CoV-2 infection and in-hospital mortality we used multivariable logistic regression analysis including the following covariates: age, sex, Charlson co-morbidity index, IMD, presence of cancer, surgical procedure type and J o u r n a l P r e -p r o o f admission acuity. [15] [16] [17] [18] Results are presented as n (%), mean (SD), median (IQR) or as odds ratios (OR) with 95% confidence intervals and p-values. The threshold for statistical significance was p<0.05. We undertook two pre-specified sub-group analyses. First, we assessed the risk of mortality among patients with SARS-CoV-2 infection stratified by urgency of surgery (elective surgery or emergency surgery). Second, we compared the risk of mortality among patients with SARS-CoV-2 infection between patients with and without respiratory symptoms. Analyses were conducted using Python (version 3.7.5) and graphs made in R (V4.0.2, R-project, Vienna).

To check the completeness of ICD-10 coding for a positive SARS-CoV-2 diagnosis, we 

Given the long incubation period of SARS-CoV-2 infection, we report the frequency of emergency hospital re-admission in England within 30 days of surgery, stratified by the presence of SARS-CoV-2 infection ICD-10 codes.

This was a population-wide cohort study including all patients that underwent surgery in England during the study period. A sample size of 2 million patients, with an allocation ratio of 0.01, and an alpha of 0.05 would give approximately 100% power to detect a 10% difference in the relative risk of mortality among patients with and without SARS-CoV-2 infection.

There were 5,336,438 surgical procedures among 2,917,480 patients in England between 1 st January 2020 and 28 th February 2021. After predefined exclusions (209,358; 7%) and exclusion of those with missing data (41,144; 1.4%), 2,666,978 patients undergoing were included in the primary analysis ( Figure 1 ). Demographic data are presented in Table 1 . Data are presented according to surgical specialty in Table 2 . Demographic data of patients with missing age data is in supplementary Table 2. *insert Table 1 , Figure 1 , and Table 2 here. Table 3 ). The adjusted odds of death among surgical patients with SARS-CoV-2 infection compared to surgical patients without infection is 5.7 (95% CI: 5.5 -5.9); P<0.001 ( Table 4 ). Length of hospital stay among patients with and without SARS-CoV-2 infection is presented in Table 3 . The survival curves of those with and without SARS-CoV-2 infection are presented in Figure 3 . Table 3 and Table 4 , and Figure 3 here (Table 3) . Therefore, the incidence of death from SARS-CoV-2 infection among elective surgical patients was 172/1,859,998 (0.009%, or 1 in 10,814). The adjusted odds of death among patients undergoing elective surgery with SARS-CoV-2 around the time of surgery is 25.8 (95% CI: 21.7-30.9); P <0.001 (Supplementary table 3 

Among patients with SARS-CoV-2 infection and respiratory symptoms 5,265/15,637 (33.7%) died compared to 888/13,140 (6.8%) of patients with SARS-CoV-2 infection but no respiratory symptoms recorded, OR 5.5 (95% CI: 5.1-6.0); P < 0.001 (Supplementary tables 5 and 6).

We identified 116,531 patients undergoing their first surgical procedure in Wales between 1 st January 2020 and 28 th February 2021. In Wales, a higher proportion of patients underwent emergency surgery (29.4%) than in England (21.8%). Using ICD-10 codes alone, 

The principal finding of this population-wide epidemiological study is that the incidence of SARS-CoV-2 infection among NHS surgical patients in England through 2020 and up to 28 include preoperative testing, household isolation and dedicated 'green' pathways for patients who are known to be SARS-CoV-2 negative. Patients and clinicians should be reassured that the population incidence of SARS-CoV-2 among surgical patients is low and the overall risk of death from SARS-CoV-2 for a patient undergoing elective surgery in a 'green' pathway is less than one in 11,000. 20 Our data confirm that the volume of surgical activity during 2020 is about 25% lower than expected compared to previous years. 1, 2 The predicted delays to surgical treatments as a result of the pandemic are substantial, with an estimated five million cases outstanding by March 2021. 4 Although the morbidity and mortality associated with cancelled or postponed surgery is unknown, it is clearly not desirable to have a large and rapidly expanding waiting list for treatments. 5, 31 Our data confirm that while it is possible to undertake surgery safely during the pandemic, the risk of mortality among surgical patients with SARS-CoV-2 is substantial and all efforts should be made to prevent nosocomial infection.

13 This study has several strengths. We included data from all patients undergoing surgery in England during the COVID-19 pandemic in England, thus our results represent the true population incidence of SARS-CoV-2 infection and associated mortality. We included data from more than two million surgical patients, which represents one of the largest observational cohorts of surgical patients during the COVID-19 pandemic. These data will be generalisable to other high-income countries. A major strength of our analysis is that we controlled for changing surgical case-mix during the pandemic, as well as potential factors that could confound association between SARS-CoV-2 infection and mortality. We used the contemporaneous comparator of patients undergoing surgery during the pandemic who did not have SARS-CoV-2 infection. This allowed us to control for unexpected changes in surgical case-mix associated with changes in population behaviour and healthcare delivery. We explored the potential for misclassification of SARS-CoV-2 status using a sample of patients from Wales with detailed SARS-CoV-2 testing data. The incidence of SARS-CoV-2 defined using ICD-10 codes in this cohort was higher than in England, possibly due a higher proportion of emergency surgical patients. The paradoxically lower incidence of SARS-CoV-2 recorded on PCR tests is likely due to the reduced availability of tests during the first wave, where clinical suspicion was required for diagnosis. Adding SARS-CoV-2 PCR tests increased the incidence compared to ICD-10 codes alone by around 15%. However, among elective surgical patients, the incidence of SARS-CoV-2 infection remained very low (<0.4%).

Our analysis also has some limitations. It possible that our data underestimate the true incidence of SARS-CoV-2 infection among surgical patients, particularly amongst those undergoing day case procedures, which would lead to under-representation of asymptomatic patients who are less likely to die. We determined the proportion of patients requiring emergency hospital admission with SARS-CoV-2 within 30 days of surgery in England and found this was very low, particularly among elective surgical patients. We only included in-hospital deaths, and our findings will therefore underestimate the true mortality risk. Early in the pandemic, testing may have only been available to patients with more severe disease leading to an over representation of these cases. SARS-CoV-2 may cause asymptomatic disease, and our case definition would miss these patients. Our sensitivity analysis indicates that ICD-10 codes underreport SARS-CoV-2 infections by around 15% compared to SARS-CoV-2 PCR results. However, the overall incidence of SARS-CoV-2 J o u r n a l P r e -p r o o f infection remained low, particularly amongst elective surgical patients. Administrative data sets are dependent on coding quality. Data are coded by trained professionals according to standardised methods. [33] [34] [35] Reports indicate Hospital Episode Statistics are accurate, particularly for procedures and important diagnoses. The influence of the pandemic on coding quality is unclear, but our sensitivity analysis results suggest that the quality of SARS-CoV-2 coding is good. At the date of our data extract, some patients will have remained in hospital so the true impact of the second wave will not be fully captured. 

TA affirms that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned have been explained. Agency (Northern Ireland), British Heart Foundation (BHF) and the Wellcome Trust. The funding sources had no role in the study design, data collection, analysis, interpretation, or writing the report.

The data used in this study were derived from two data sources. It is not possible to share the raw patient-level data provided by NHS Digital describing NHS patients in England.

Regarding data from NHS patients in Wales, the data used are available in the SAIL Databank at Swansea University, Swansea, UK, but as restrictions apply they are not publicly available.

All proposals to use SAIL data are subject to review by an independent Information Governance Review Panel (IGRP). Before any data can be accessed, approval must be given by the IGRP. The IGRP gives careful consideration to each project to ensure proper and appropriate use of SAIL data. When access has been granted, it is gained through a privacy protecting safe haven and remote access system referred to as the SAIL Gateway. SAIL has established an application process to be followed by anyone who would like to access data via SAIL at https://www.saildatabank.com/application-process.

This work uses data provided by patients and collected by the NHS as part of their care and support. We would like to acknowledge all data providers who make anonymised data available for research and the collaborative partnership that enabled acquisition and access to the de-identified data, which led to this output. 

A patient representative was consulted in the design of this study.

The results of this study have been shared with representatives from NHS England prior to publication, in order to inform patient care.

The Corresponding Author has the right to grant on behalf of all authors and does grant on J o u r n a l P r e -p r o o f 

Age of patients undergoing surgery

Frequency of surgical treatment and related hospital procedures in the UK: a national ecological study using hospital episode statistics

Estimate of the global volume of surgery in 2012: an assessment supporting improved health outcomes

Estimated surgical requirements in England after COVID-19: a modelling study using hospital episode statistics

Elective surgery cancellations due to the COVID-19 pandemic: global predictive modelling to inform surgical recovery plans. The British journal of surgery

Global guidance for surgical care during the COVID-19 pandemic. The British journal of surgery

Can we safely continue to offer surgical treatments during the COVID-19 pandemic?

Mortality and pulmonary complications in patients undergoing surgery with perioperative SARS-CoV-2 infection: an international cohort study

Factors Associated With Surgical Mortality and Complications Among Patients With and Without Coronavirus Disease 2019 (COVID-19) in Italy

Mortality and pulmonary complications in patients undergoing upper extremity surgery at the peak of the SARS-CoV-2 pandemic in the UK: a national cohort study

Calculating hospital length of stay using the Hospital Episode Statistics; a comparison of methodologies

Royal College of Surgeons Co-morbidity Consensus G. Identifying co-morbidity in surgical patients using administrative data with the Royal College of Surgeons Charlson Score

International Surgical Outcomes Study g. Global patient outcomes after elective surgery: prospective cohort study in 27 low-, middle-and high-income countries

Identification and characterisation of the high-risk surgical population in the United Kingdom

Mortality after surgery in Europe: a 7 day cohort study

Socioeconomic deprivation and long-term outcomes after elective surgery: analysis of prospective data from two observational studies

Understanding and responding to COVID-19 in Wales: protocol for a privacy-protecting data platform for enhanced epidemiology and evaluation of interventions

Results of COVID-minimal Surgical Pathway During Surge-phase of COVID-19 Pandemic

Elective Cancer Surgery in COVID-19-Free Surgical Pathways During the SARS-CoV-2 Pandemic: An International, Multicenter, Comparative Cohort Study

Early mortality after hip fracture surgery in COVID-19 patients: A systematic review and meta-analysis

Postoperative mortality among surgical patients with COVID-19: a systematic review and meta-analysis

Peri-operative Outcomes and Predictors of Mortality in COVID-19 Positive Patients with Hip Fractures: A Multicentre Study in the UK

Analysis of worldwide surgical outcomes in COVID-19-infected patients: a gynecological oncology perspective

Vascular Surgery During COVID-19 Emergency in Hub Hospitals of Lombardy: Experience on 305 Patients

IMPACT-Restart: the influence of COVID-19 on postoperative mortality and risk factors associated with SARS-CoV-2 infection after orthopaedic and trauma surgery

New geographic model of care to manage the post-COVID-19 elective surgery aftershock in England: a retrospective observational study

Reduction in emergency surgery activity during COVID-19 pandemic in three Spanish hospitals. The British journal of surgery 2020. 30. collaborative CCr. Colorectal cancer services during the COVID-19 pandemic

Delaying surgery for patients with a previous SARS-CoV-2 infection

Immediate and long-term impact of the COVID-19 pandemic on delivery of surgical services. The British journal of surgery

Data Resource Profile: Hospital Episode Statistics Admitted Patient Care (HES APC)

Colorectal cancer ascertainment through cancer registries, hospital episode statistics, and self-reporting compared to confirmation by clinician: A cohort study nested within the UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS)

Accuracy of general hospital dementia diagnoses in England: Sensitivity, specificity, and predictors of diagnostic accuracy

Admission category Elective 1859998 (69.7%) 1857586 (99.9%)

%) 2253 (1.1%)

SARS-CoV-2 Infection No SARS-CoV-2 Infection