Documenting COVID-19 screening before surgery during lockdown (COVID Screen): an audit with routinely collected health data

Compared with many developed countries, Australia has had a low incidence of SARS-CoV-2 infection and COVID-19 disease.¹ However, early in the pandemic the Australian Health Protection Principal Committee recommended restricting surgery to non-elective and more urgent elective surgery as part of the national lockdown strategy to reduce viral spread and manage the possible demands of the pandemic.² Most hospitals introduced evidence-based screening for COVID-19 for patients before surgery. This screening is important for several reasons, including patient safety, staff safety, public health and hospital resources.³

Little is known about the efficacy and results of these screening programs. However, a major concern is patients undergoing surgery with unrecognised concurrent SARS-CoV-2 infection. The most appropriate screening in Australia is likely to be systematic questionnaires about COVID-19 combined with measuring temperature without quarantine or testing for SARS-CoV-2 infection.³ One aspect of escalating surgical care after lockdown is continuing to minimise the risk of SARS-CoV-2 being brought in to the hospital.^3,4 We propose that analysing screening during the lockdown of the first wave in 2020 may inform ongoing and future screening.

As part of this study, we asked several questions: (1) what proportion of patients had a documented systematic questionnaire for COIVD-19 and temperature measured before surgery; (2) were there important differences between surgical subgroups; (3) of patients screened before surgery for COVID-19, what proportion screened negative for COVID-19; (4) of patients screening positive, what items were positive; and (5) did infection control detect any patients, particularly those who screened negative, who tested positive to SARS-CoV-2 during their first 14 postoperative days?.

This study was a retrospective observational audit using routinely collected health data at two large geographically separated Melbourne hospitals. These hospitals perform most types of adult surgery and interventional and diagnostic procedures, but neither has an obstetric service. We called this work the COVID Screen audit.

The human research ethics committees (HREC) of each hospital approved this project as a low-risk audit-type activity that did not require formal HREC review or informed consent from patients.

We searched the medical records of all consecutive patients undergoing procedures that were surgical, diagnostic and interventional listed with the operating suites. At one hospital this was between 1 April and 10 May 2020, whereas at the other this was 1–30 April 2020; therefore, data were collected at both hospital for all of April 2020, but we included the May data for overall analysis. These differences in collection periods were due to varying research staff availability. Investigators enquired with hospital infection control or infectious diseases departments to see whether any of the sample patients had tested positive or had been reported to the hospital through the Victorian Department of Health and Human Services contact tracing up to 14 days postoperatively. The search strategies differed between hospitals because one hospital has an electronic heath record (Cerner, Kansas City, MO, USA) and the other has a scanned medical record. Patients were identified from stored electronic surgical lists and included those undergoing elective or non-elective surgery. We anticipated data on approximately 2000 patients.

Files were searched for documented specific preoperative COVID-19 questionnaires (Box 1) and temperature recordings. The questionaries were consistent with the Australian definitions for COVID-19.^3,5 Both hospitals had specific COVID-19 questionaries as part of the electronic or scanned medical record. In the absence of questionaries, admission, medical and nursing notes were searched for evidence of COVID-19-related history, usually questions around flu-like symptoms. We did not test interrater reliability on interpreting these notes.

Data were stored on in database (Excel; Microsoft, Bellevue, WA, USA) files with security access. Each patient was allocated a unique identifier based on the hospital and a four-digit number (e.g. AA1012 or BB0033). We did not collect potentially identifying data, such as patient name, hospital record number, date of birth or specific operation.

The following information was collected: patient age, patient sex, surgical speciality, date of surgery and elective versus non-elective surgery. We noted whether the patient was febrile (>38°C), afebrile or temperature not recorded. Where there was a COVID-19 questionnaire, we noted whether the patient had answered ‘no’ or ‘yes’ and, if ‘yes’, to which questions. We noted any preoperative or postoperative SARS-CoV-2 testing and the results with viral RNA testing (sensitivity 70%).⁶

For patients undergoing multiple procedures during one in-patient stay, only the first procedure was included. This applied to 60 patients. Surgical and procedural diagnostic and interventional specialities (surgical specialities) were bundled together into: cardiothoracic; ear nose and throat (ENT), incorporating faciomaxillary; endoscopy; general, incorporating breast, upper gastrointestinal, colorectal and hepatobiliary; neurosurgery; orthopaedics; plastics; urology; and vascular. The category of ‘other’ included interventional radiology, interventional cardiology, gynaecology, ophthalmology and psychiatry (electro-convulsive therapy).

A proposed minimum acceptable screening rate of >85% was used, consistent with other audits in perioperative medicine.⁷ Because we were not testing hypotheses, we did not have a systematic statistical plan. Rather, we planned summary statistics and then used post hoc analyses to examine the absolute difference and 95% confidence intervals (CIs) to indicate the precision of estimates for important differences using online software (VassarStats; Website for Statistical Computation; www.vassarstats.net, accessed 20 June 2020). In particular, we analysed differences between elective and non-elective surgery patients, as well as patients in the first and last weeks of April 2020.

We used the REporting of studies Conducted using the Observational Routinely-collected health Data (RECORD) Statement,⁸ which is an extension of the STROBE reporting statement for observational studies, and the Standards for Quality Improvement Reporting Excellence guidelines (SQUIRE 2.0).⁹

Data were collected on 2197 consecutive patients (Table 1; Fig. 1) between 1 April and 10 May 2020. Patients were spread across surgical specialities, with 58% undergoing elective surgery. The median age was 60 years (interquartile range (IQR) 45–72 years; range 1–98 years). Of these patients, 912 (42%) were women, 1283 (58%) were men and one (<1%) was other.

Table 1.  Documenting COVID-19 screening using a questionnaire or temperature check before surgery in subgroups
Unless indicated otherwise, data are given as the number of patients in each group, with percentages of that surgical subgroup given in parentheses. ENT, ear nose and throat

Fig. 1.  Percentage of patients who had either any COVID-19-related questions or temperature documented (grey circles) and those who had a recorded systematic COVID-19 questionnaire and temperature documented (black squares) before surgery; whiskers indicate 95% confidence intervals. The dashed line is the 85% minimum acceptable documenting rate.

We found that 2037 (93%) patient histories had some documented evidence of questions about COVID-19 history and symptoms (usually flu-like symptoms) and/or a recorded temperature before surgery (Table 1; Fig. 1). Most patients (n = 1303; 59%) had documented answers to a systematic questionnaire, and most of these patients also had temperature measured (n = 1273; 58% of total). Some patients (n = 156; 7%) had no record of any questions about COVID-19 or temperature measured before surgery.

In all, 152 patients (7%) had 159 tests for SARS-CoV-2; all were negative. Of 1364 patients (62%) who had a record of no apparent COVID-19 symptoms and were afebrile before surgery, 38 were tested for SARS-CoV-2 infection within 2 weeks of surgery and all were negative.

Of those with documented screening, 125 reported potential COVID-19 symptoms or related problems or were febrile before surgery; the most common problems were fever (n = 60; 48%), cough (n = 33; 26%) and/or sore throat (n = 11; 9%). Of these 125 patients, 67 (54%) were tested before surgery and all were negative; this included 20 of 33 who were both symptomatic and febrile. No patient reported having had COVID-19, and only one patient reported having been a close contact of someone with COVID-19; that patient tested negative.

Infection control units at both hospitals did not have any of the patients in this audit cohort reported to them by the Victorian Department of Health and Human Services as developing SARS-CoV-2 infection within 2 weeks of surgery.

Although most patients had a documented temperature, there was marked variation between groups in the percentage of patients who had both a COIVD-19 screening questionnaire and temperature documented (Table 1; Fig. 1). The most marked difference was between patients undergoing elective (72%) and non-elective (38%) surgery. This was an absolute (rounded) decrease of 35% (95% CI 31–38%; relative risk (RR) 49%). The documented percentage of patients with a positive questionnaire or febrile was 3.1% and 9.3% among the elective and non-elective surgery groups respectively, even with less screening. This was an absolute difference of 6.2% (95% CI 4.2–8.4%; RR 3.0%). We found that screening may have improved from Week 1 to Week 4 of April (Table 1; Fig. 1) by 6.0% (95% CI –0.5%, 12.5%). This was a relative increase of 11% from Week 1 to Week 4.

We undertook an audit of documenting screening for COVID-19 using routine clinical records at two large Melbourne hospitals. We recognise that we cannot assume that ‘not documented’ necessarily meant ‘not measured’. We found that documenting screening varied markedly across surgical groups. In particular, we found that non-elective surgical patients had the lowest rate of documenting and the highest rate of COVID-19-related history and signs. Approximately two-thirds of patients had documented negative screening. No patient returned a positive test for SARS-CoV-2.

Although there are many guidelines, there has been limited study of surgical patients during the COVID-19 crisis and previous respiratory pandemics.¹⁰ Identifying surgical patients with SARS-CoV-2 and COVID-19 disease is important for several reasons.^3,4 First, patients with moderate to severe COVID-19 may have more complications and greater mortality after surgery.¹¹ COVID-19 may further increase the already increased risks for complications and mortality among non-elective surgical patients.^7,11,12 Second, patients with undetected SARS-CoV-2 infection may infect both patients and staff if admitted to hospital. Third, in Australia, patients with known or suspected SARS-CoV-2 require different levels of personal protective equipment (PPE) for perioperative care than those not suspected of SARS-CoV-2 infection.¹³ Not only is exposure without appropriate PPE a staff health risk, but it also means that staff have to consequently self-isolate and are often tested for SAR-CoV-2. Along with community prevalence, screening is important to help determine prior probability when interpreting a subsequent SARS-CoV-2 test.⁶ A patient who screens positive but is undergoing elective surgery should have surgery delayed and be tested, whereas a patient who requires non-elective surgery could be tested if time allows or if more urgent surgery proceeds with high (aerosol)-level precautions and PPE^3,13 with subsequent follow-up. During both first and subsequent pandemic waves, documenting COVID-19 screening facilitates the management of patients, particularly during handovers and preoperative assessment, both universal within perioperative care. This is true even in states with mandatory preoperative SARS-CoV-2 testing, and is particularly important for areas with low prevalence, as well as when high prevalence decreases and surgery is escalating.

This audit was undertaken during government-mandated restricted surgery that was part of the national lockdown to manage the SARS-CoV-2 pandemic. Despite this national imperative, documented COVID-19 screening fell short of our proposed lower acceptable limit of 85%⁷ in almost all surgical groups. There was an 11% relative increase in documented screening between the first and last weeks of April, but the CI ranged from substantially improved to slightly worse. There are likely to be many overlapping reasons for limited documenting of COVID-19 screening. Possibly the most important reason is that both hospitals had several stages of screening, including before admission, telephone calls and at hospital entrances. Clinical staff may have assumed that these screens were adequate. However, this other screening was not routinely documented in patient records. However, non-elective surgical patients often do not undergo these walk-in screens and instead are likely to be screened if they come through the emergency department. Unlike elective surgery, non-elective admissions occur around the clock, come from a variety of sources, are admitted to wards rather than surgery admission units and range from life-threatening emergencies to less urgent but still non-elective problems. These factors may also affect screening. Another factor may be that the Australian pandemic response during April 2020 was very successful compared with other countries and by the end of April 2020 the Australian Health Protection Principal Committee² recommended a gradual increase in the availability of elective surgery. This success may have reduced the perceived importance of the screening.

The strength of this audit is that it includes over 2000 consecutive patients across a broad range of procedural specialities at two large public hospitals that are geographically separated, belong to different health services and have limited staff and patient overlaps. The COVID-19 questionnaires were consistent with the recommendations of the Australian Commission on Safety and Quality in Health Care³ using Australian definitions for COVID-19⁵ with clinical and epidemiological criteria. We suspect that our results are likely to apply to other Australian hospitals.

Limitations of our audit include that data were collected retrospectively and that we did not investigate other COIVID-19 screening or the implementation approaches¹⁴ for COVID-19 screening or the enablers and barriers to routine screening. Without antibody testing, we could not confirm that no patient with SARS-CoV-2 received surgical care because we were unable to determine whether any patients who had completely asymptomatic infection and did not undergo testing received surgical care. Further, we cannot exclude that all patients were fully screened but only some had this documented.

Although we found that most surgical patients had a documented preoperative temperature, far fewer had a documented systematic COVID-19 screening before surgery. In particular, non-elective patients had a low documented screening rate. However, no patient was found to have SARS-CoV-2 infection within 2 weeks of surgery. Future research could examine how to apply implementation science¹⁴ to new but urgent measures, such as screening during a pandemic across elective and non-elective admissions.

The authors report no potential competing interests.