Evolving swabbing practices for COVID-19 in a New Zealand emergency department during the early stages of an emerging pandemic

Objective : To review if tests for suspected COVID-19 were performed according to the Ministry of Health (MoH) case de ﬁ nitions, identify patterns associated with testing outside of the case de ﬁ nition, and discuss the potential impacts on hospital services. Methods : This was a retrospective audit of patients presenting to the Wellington Hospital ED between 24 March 2020 and 27 April 2020 who were swabbed for COVID-19 in ED. Swabs were audited against the March 15th and April 8th MoH COVID-19 case de ﬁ nitions. Results : Five hundred and thirty-six COVID-19 swabs for 518 patients were taken during the study period. There was poor alignment of testing


Introduction
The global outbreak of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causing the disease COVID-19 initiated many procedural changes in EDs worldwide.COVID-19 was first detected in New Zealand (NZ) on 27 February 2020. 1 A government modelling report estimated there could be 8560 to 14 400 deaths from (11.6% met criteria), but testing was much more closely aligned with the wider April 8th case definition (88.2% met criteria).• Higher proportions of patients with specific comorbidities, symptoms or epidemiological exposures were swabbed outside the case definition, indicating that clinician perception of COVID-19 risk for certain patients may not have aligned with the case definition in use.• The intended use of a case definition may not always align with its implementation.This has potential implications for patient flow, quality of care, and resource management in an early pandemic, including testing capacity.
COVID-19 in NZ. 2 As an effective vaccine and therapeutics were not yet available, the main mechanisms to limit transmission were early case identification and movement restriction.Movement restriction was mandated through legal powers.NZ entered the highest level of restrictions, Level 4, between 25 March 2020 and 28 April 2020.Level 4 lockdown restrictions required people to remain at home other than for essential personal movement and called for the reprioritisation of healthcare services. 3ase definitions formed a vital part of COVID-19 case identification and testing guidance in NZ.Comprised of epidemiologic and clinical criteria, these definitions were issued by the Ministry of Health (MoH).As was seen internationally, case definitions needed to be adapted to include new information about the clinical presentation of COVID-19 alongside other factors, including local prevalence and testing capacity. 4 A key issue during the early pandemic was maximising case definition sensitivity although limiting testing (mis)use. 5ring the Level 4 lockdown, international guidance for diagnostic testing evolved rapidly. 6Initially, the March 15th case definition required both epidemiological exposure and symptom criteria to be met, which were based on overseas case descriptions. 7On 8 April, the case definition changed, removing the epidemiological criteria to reflect the established community transmission and enable early case detection.The clinical criteria were also modified to include coryza and anosmia. 8his audit aims to review whether COVID-19 tests undertaken in Wellington Hospital ED aligned with the relevant case definition over the Level 4 lockdown period (24 March 2020 to 27 April 2020).This is a unique opportunity to review the implementation of changing national case definitions during a pandemic.We aim to inform future pandemic planning in EDs by identifying and discussing factors associated with testing outside of the case definition and considering the impacts of acting outside of guidelines on hospital processes.

Methods
This was a retrospective audit of patients presenting to the Wellington Hospital ED between 24 March 2020 and 27 April 2020 who were swabbed for COVID-19 during their presentation.

Objectives
To review if nasal-pharyngeal swabs for suspected COVID-19 were performed according to the MoH guidelines, and to identify patterns in swabbing practices and discuss the potential impacts on hospital services.
MoH national case definition guidelines (Table 1) from the 15 March to 8 April were in place locally and used as the audit standards.Individuals swabbed between 24 March and 7 April 2020 were assessed against the March 15th case definition.Individuals swabbed between April 8 and April 27 were assessed against the April 8th case definition.

Setting
Wellington Regional Hospital is an urban tertiary teaching centre with a mixed paediatric and adult ED and an annual ED census of 67 000 patients annually.At the beginning of the study period, a total of 205 confirmed and probable COVID-19 cases had been reported in NZ. 9 By 8 April, the total confirmed and probable cases was 1210. 10 During the time of the study, the decision to swab patients for COVID-19 was made by specialist doctors in conjunction with other clinical staff.

Audit population
Patients who presented to Wellington Hospital ED between 24 March 2020 and 27 April 2020.This period covers the Alert Level 4 Lockdown. 11

Inclusion criteria
Any patient with a nasal-pharyngeal swab performed to test for COVID-19 in Wellington Hospital ED during the study period.

Exclusion criteria
Patients who were swabbed for COVID-19 before their presentation to the ED and were not re-swabbed for COVID-19 during their ED presentation.Patients who had never been swabbed for COVID-19.

Data collection and analysis
The clinical records of all patients who presented during the study period were reviewed for a test result of a swab associated with that presentation, and those who met the inclusion criteria were identified.Data was collected by the primary author and a specialist nurse, supported by another physician.All were trained medical professionals familiar with Wellington Hospital ED and the electronic records.Clinical notes were reviewed by an audit investigator using a structured questionnaire covering demographics (age, sex, prioritised ethnicity), referral method, comorbidities (respiratory disease, cardiovascular disease, immunosuppression, BMI >40, diabetes, chronic kidney disease, liver disease, cancer), disability alerts, COVID-19 exposure risk, test results, disposition, COVID Assessment and Management Unit (CAMU) attendance, and symptoms (fever, cough, sore throat, shortness of breath, GI symptoms, coryza and anosmia).Presentation-related information was taken from recorded vitals and ED notes; demographic information from patient NHI; and comorbidity information from ED notes, medical notes and GP information.Data were de-identified and collected in an Excel spreadsheet kept securely on Wellington Hospital drives with controlled access.Alignment with guidelines was reviewed according to the case definitions listed above.
Swab results were reported for all patients in the study period only.Stratification of swab results by demographic variables resulted in small numbers in some strata with the risk of individuals being identifiable.We measured the proportion of swabs taken by age, sex, and ethnicity (prioritised).
Patients were classified by exposure criteria such as 'close contact'.Exposure categories were defined as shown in Table 2.
Approval for this audit was received from Capital and Coast District Health Board QIPS Clinical Audit Review on 20 August 2020.This project complies with CCDHB research policies and procedures.

Statistical analysis
Data analysis was completed using STATA software (Version 14.2; StataCorp LLC, College Station, TX, USA) and Microsoft Excel (Version 2111).Frequency, proportions with 95% confidence intervals (CIs), and medians with interquartile ranges (IQRs) were calculated as appropriate to describe the data.Missing data were not included in statistical testing.

Testing events
Of the 536 swabs taken during the audit period, 183 were taken during the first case definition and 353 were taken during the second case definition (Table 3).Only 11.6% (n = 19/164) of tests in ED between the 24 March and the 7 April met the March 15th case definition.In contrast, 88.2% (n = 300/340) of swabs taken between 8 April and 27 April met the April 8th case definition (Fig. 1).
Of the 145 swabs that did not meet the March case definition, 82.1% (n = 119/145) met the symptom criteria but not the exposure criteria.Exposure criteria alone were met for 5.5% (n = 8/145) swabs taken in that period, the majority from healthcare workers who did not meet the March 'alternative criteria'.Neither exposure nor symptom criteria were met for 11.0% (n = 16/115) swabs.Of this subgroup, 56.3% (n = 9/16) had a fever higher than 38 and 18.8% (n = 3/16) had a low-grade fever (>37.5 ).
Table 4 shows the frequency of a range of variables disaggregated by whether the swab did or did not meet the relevant case definition.All positive swabs met the relevant case definition.Half the swabbed patients were admitted (n = 289/536, 53.9%).Those swabbed despite not meeting the April case definition were significantly more likely to be admitted to hospital (n = 30/40, 75.0%[95% CI 59.4-86.1]).A high proportion of the swabs (n = 9/19, 47.4%) which met the March case definition were taken in those who were discharged after triage assessment (i.e. received a swab at triage and never entered the ED for treatment).

Discussion
Reflecting the evolving national situation, testing behaviour for COVID-19 in Wellington Hospital ED changed during the lockdown period.Initially, Wellington Hospital ED was one of the limited locations where people in the region could access a diagnostic test. 12This was evident in the high proportion of those meeting the March 15th case definition who were discharged after triage assessment.Only three swabs (0.6%) were positive overall, consistent with low prevalence in the region at the time.
The wider case definition from April 8th resulted in more patients being tested.At this time, testing for COVID-19 in symptomatic patients was conducted in ED, the hospitalbased CAMU, and various community testing locations.Laboratory testing capacity was also increased from 60 tests a day in early March 2020 13 to over 4000 tests a day in early April. 14his audit shows that a lower proportion of swabs aligned with the With no formal management of returning travellers, there was widespread concern about undetected community cases. 15,16One in 10 patients tested outside of the March case definition were contacts of travellers.Between January and August 2020 in the Greater Wellington region, 69% of confirmed COVID-19 cases were in overseas travellers arriving via flights. 17The focus on travellers as a high-risk group would likely have influenced the testing of contacts of travellers who were not known COVID-19 cases.Despite being downgraded as a clinical criterion, fever was the leading reason for testing in patients not meeting the April case definition.
Other key factors associated with testing outside criteria were a cancer diagnosis and subsequent hospital admission.This suggests that clinicians may have had concerns regarding COVID-19 in their febrile patients and is supported by the continued inclusion of fever in some ED screening tools used in mid-April 2020.We hypothesise three issues that may have contributed to this association.First, concern over narrow testing although there was limited clinical experience with COVID-19 in the region; second, vague wording in the April case definition of 'with or without fever' provided insufficient clarity on whether to test; or third, in certain patients, a COVID-19 test was useful for bed placement and isolation precaution purposes.In addition, although fever was not a clinical criterion, the case definitions did allow for clinical judgement in testing.This association may represent exercise of that clinical judgement.
The high rate of testing in patients meeting symptom criteria but not exposure criteria for the March case definition may have reflected concern around undetected community COVID-19 cases. 16,18Initial COVID-19 symptoms are non-specific and similar to other circulating respiratory viruses.Following the lockdown, these other respiratory viruses virtually disappeared. 19alf of the swabs performed outside of guidelines were in immunosuppressed or cancer patients during the April case definition (n = 20/40, 50%).Possible reasons for this association could include concerns of heightened individual risk, the potential for atypical presentations of COVID-19, and the risk of nosocomial outbreaks.
There was a high level of concern that COVID-19 might overwhelm the health system and result in a high mortality rate, as seen overseas. 20,21atterns seen in swabs which did not meet guidelines indicate that assessment of exposure risk (e.g.contacts of travellers), symptomology (e.g.fever), and patients' personal risk (e.g.cancer diagnosis) were associated with testing outside of the case definition.

Evaluation of case definition use
During a pandemic, the case definition serves to identify individuals at highest risk of having the infectious disease.Poor alignment of testing with the case definition raises several potential issues.One consequence was the (mis) use and prioritisation of limited laboratory testing capacity for individuals at low risk of having COVID-19.
Testing outside of the case definition has implications for timeliness and quality of patient care.Patients with suspected COVID-19 were seen in isolation rooms and required full personal protective equipment (PPE).If an isolation bed was unavailable, the patient remained in ED, blocking the bedspace and slowing patient flow.
However, when there is capacity for high levels of testing as seen towards the end of the lockdown period, maximising case definition sensitivity also has benefits.This allows for as many cases to be identified as possible, which is important as delayed diagnosis risks infection of staff and other patients.
The present study highlights the difficulty in creating guidelines in an evolving situation whereas balancing case definition sensitivity with resource limitations.The low proportion of tests aligning with the March case definition indicates that a narrow guideline may not restrict testing to the intended degree.Conversely, when guidelines are broader, testing appears to align more closely with them.There may also be a lag in behaviour change of physicians to match guidelines.
An important consideration for policymakers is whether the intended use of a case definition is reflected in its implementation.The present study indicates that may not always be the case.Factors that could be important in this situation include clear two-way communication between policymakers and clinicians and proactively recognising and addressing the impacts of anxiety, fear and media narrative on clinicians.Without this communication, resource management in an early pandemic is likely to be challenging, and this can have unintended consequences on processes and services.

Subsequent developments and learnings
In the months following the lockdown period, in Wellington Hospital ED, adaptations were made in stratifying patients' COVID-19 risk to assist with decision-making.An algorithm was created in the form of a checklist which includes epidemiological criteria and clinical criteria.It gives a decision on when to test, what PPE will be required, and what type of bed is required on admission, for example, a single room.

Limitations
This study has several limitations.Retrospective data collection depends on the quality of data collection at the time of presentation.Particularly for patients discharged after triage assessment, there was likely underreporting of patient symptoms and exposures in the electronic notes.Some patients (n = 9) had no data available about any symptoms on presentation.Only electronic records were reviewed, and some comorbidities data may have only been available in paper notes which were not reviewed.Both issues would have under-represented any association between being swabbed and these variables.
This audit did not include patients who were not swabbed for COVID-19 during lockdown, so we cannot comment on whether any patients who should have been swabbed were missed.As a retrospective audit, the present study can only note associations between variables but cannot imply causation.

Conclusion
The present study provides a unique insight into the implementation of changing case definitions to guide testing during a pandemic.The rapidly evolving case definitions provided guidance to the ED clinicians but may not have aligned with their perception of risk, resulting in swabbing of people with specific comorbidities, symptoms, or epidemiological exposures outside the case definition.There are potential serious consequences of misalignment of testing with guidelines, including using limited resources on low-risk patients, and impacts on patient flow and care quality.Involvement of frontline clinicians in the development of case definitions, improved communication of the justification for narrow case definitions, as well as proactive recognition and response to the concerns of clinicians implementing them may improve adherence and mitigate these consequences.

TABLE 1 .
Case definitions as described by the Ministry of Health on March 15 and April 8 2020

TABLE 2 .
Definitions of exposure categories

TABLE 3 .
Patient demographic information for individuals who had one or more COVID-19 swabs done during the audit period Figure 1.Percentage of patients meeting the case definition for COVID-19 by March 15th and April 8th case definitions.Error bars represent 95% confidence intervals.

TABLE 4 .
Patient and presentation characteristics in COVID-19 swabbing events compared across the March 15th and April 8th case definitions and whether the swab met the case definition criteria