Since the beginning of the COVID-19 outbreak, the overall number of SGH ED visits and admissions have dipped as compared to corresponding pre-COVID-19 period. There was also a proportional decline in visits for all diagnosis categories, except for complaints related to the respiratory system which more than doubled during the COVID-19 period. Notably, there were fewer visits and presentations for all diagnostic categories by SGH ED’s highest priority P1 group. This raises the concern whether there was any delay and failure in seeking necessary medical treatments which can translate to poorer health outcomes26,27. Amidst the decline, we noted a shift in ED patient case-mix from ‘Non-fever’ cases to ‘Fever’ cases. This likely gave rise to the two-phase ED attendance trend that we observed at SGH ED, where each phase has its distinct trough-to-peak visit pattern. The first and second phase corresponds with the pre-Circuit Breaker and Circuit Breaker period, respectively. Lastly, the Circuit Breaker period had a higher average ED admission rate as compared to the pre-Circuit Breaker period. Between these two periods, the average number of ED patients admitted per day remained relatively unchanged, except for P3F patients. The Circuit breaker period, however, saw a greater decline in average daily ED attendance as compared to the pre-Circuit Breaker period.
Possible explanations for the two-phase ED attendance trend at SGH
The general decline in ED attendance was widely reported in several countries, including fewer visits for non-respiratory complaints6,9 and certain medical emergencies such as stroke and acute myocardial infarction1,13-15. Here at SGH ED, we observed a similar trend with an overall decline in visits for non-fever and non-respiratory complaints, raising the concern that some patients needing emergency care were not seeking help at ED. With a large decrease in ED visits, one may expect the sickest patients (P1) to continue attending ED for unquestionably needed emergency care. Nevertheless, COVID-19’s impact on the ED attendance of P1-equivalent triage group seemed to vary by country. While two studies reported no change in ED attendance of their sickest patient group28,29, several other EDs observed a significant decline in visits of the same vulnerable group7,10,30. Like most studies, SGH ED saw fewer life-threatening presentations and a decrease in both the number and proportion of P1 visits throughout the COVID-19 period. We speculate the following reasons that could have contributed to the overall decline in SGH ED attendance, even amongst the most urgent cases.
Some patients may have avoided SGH ED due to fear of burdening the healthcare system22 or contracting the dreadful virus in a public hospital that is screening and treating for COVID-1926,31. Also, social distancing measures and a partial lockdown have likely led to lower incidences of common community infections and injuries related to work, sports, and road traffic accidents. It is also likely that the gradual expansion of COVID-19 screening and treatment capacity in Singapore have redirected patients from ED to alternative avenues. These avenues include community COVID-19 swab operations and isolation facilities, teleconsultation services and general practitioner (GP) clinics adequately equipped to manage patients with respiratory symptoms20. Furthermore, the close collaboration between Singapore’s public and private healthcare hospitals likely expanded treatment capacity for management of chronic medical ailments and non-life-threatening clinical conditions. Our data, however, cannot prove whether there has been a true decrease in the incidence of diseases, or whether patients are avoiding ED or seeking treatment elsewhere. Whether there was any delay and failure in seeking urgently needed treatments remains a major public health concern. Additional data and in-depth analysis are required to study the relative contributions of the aforementioned factors to the decline in ED attendance.
Unlike the surge in ED attendance and number of ‘Fever’ visits that we saw during pre-Circuit Breaker period (First phase), ED attendance declined even as daily infected cases hit record-breaking numbers during the Circuit Breaker (Second phase). During the first phase where daily counts of newly infected community cases were relatively low, the shift in ED patient case-mix from ‘Non-fever’ cases to ‘Fever’ cases may be contributed by suspected COVID-19 cases seeking testing and treatment at the ED, as well as patients seeking care for flu-like symptoms for fear that they have contracted the COVID-19 virus32. As compared to pre-COVID-19 period, the percentage change in ‘Fever’ and ‘Non-fever’ visits and respiratory-related diagnoses during the COVID-19 period was likely an overestimation due to modified triage thresholds at ED. Patients who met suspect case criteria for COVID-19 were likely triaged as ‘Fever’ cases and given a respiratory-related primary diagnosis, no matter how mild their respiratory symptoms were or even if their chief complaint was of a non-respiratory nature33. This was to facilitate the prompt placement of such patients in specific acute respiratory inpatient wards upon admission. This contrasts with the pre-COVID-19 period, where such patients with mild respiratory symptoms or non-respiratory presenting complaints were less likely to be given a respiratory-related diagnosis and the ‘Fever’ triage category mainly functioned to isolate patients with suspected conditions that were highly transmissible, such as pulmonary tuberculosis, measles, chicken pox and herpes zoster4. Amidst the rise in ‘Fever’ visits during the first phase, the overall number of ‘Non-Fever’ visits fell and this decline persisted into the Circuit Breaker period (Second phase). On the contrary, the surge in number of infected cases during Circuit Breaker (Second phase) was largely restricted to the densely occupied migrant worker dormitories, rather than the community4,34. The prompt gazetting of these dormitories as isolation areas, general compliance to social distancing measures, rapid contact tracing and systematic screening for COVID-19 have likely stemmed any widespread infection to the community. Moreover, the scaling up of community medical facilities and services during the Circuit Breaker (Second phase) allowed quicker testing and housing of clinically well COVID-19 patient34. These measures likely gained traction throughout Circuit Breaker and prevented an overwhelming number of ‘Fever’ visits to SGH ED, which partially explains the persistent nadir of ED attendance observed during the first half of Circuit Breaker.
In the second half of the Circuit Breaker, the recovery in ‘Non-Fever’ visits may be due to the delay in seeking treatment, resulting in more severe presentations that required emergent care and admission from ED. It may also be attributed to better public perception of the local COVID-19 situation and the declining number of new cases in the community. These may have quelled the public’s fear of contracting the virus in the community and hospital settings.
Possible explanations for ED admission rate trends at SGH
The ED admission rate is one of several ED key performance indicators at SGH and it is dependent on ED attendance, the number of patients admitted from the ED and the patient case-mix unique to SGH. It is worth noting that SGH ED is likely to have higher than average ED admission rates as compared to other typical EDs during COVID-19 period due to its position in the largest tertiary hospital in Singapore35. During peacetimes, a consistent trend in ED attendance allows the reliable use of ED admission rate to reflect inpatient bed needs of ED patients. This rate, however, can be misleading during a pandemic.
Existing few studies have reported temporal associations between declining ED attendance and rising ED admission rates during the COVID-19 period36,37. At SGH ED, we observed obvious variability in the average ED admission rates between the pre-Circuit Breaker period (First phase) and Circuit Breaker period (Second phase). Although the overall number of patients admitted from the ED were the same in both periods, the trough of ED attendance persisted longer during the Circuit Breaker period (Second phase) as compared to the pre-Circuit Breaker period (First phase). These numbers gave rise to a higher average ED admission rate during the Circuit Breaker period despite the need for the same number of hospital beds. Our findings suggest that the admission rate should be distilled down to its components for evaluation. It is possible that the average acuity of presenting illness has increased across triage groups, necessitating a higher proportion of patients admitted. Further analysis of ED resource utilisation and life-saving interventions may tell us the severity of conditions that these patients presented to ED with. Another possible implication could be related to the change in protocol to admit suspected COVID-19 cases or patients with respiratory symptoms, who would otherwise be typically discharged during the pre-COVID-19 period. This included admitting patients who had difficulty self-isolating to Acute Respiratory Infection and isolation wards to reduce community spread.
The first implication of our findings was the possible delay in P1 patients seeking life-saving interventions at ED. The decline in SGH ED attendance might also be attributed to reduced incidence of disease in the community or a diversion towards alternative treatment avenues. Additional data and further exploration of the factors above are needed to conclude whether there was a true delay in seeking necessary treatments and explore the reasons behind it. The delay of emergent care and chronic care might lead to increased morbidity and mortality in the community, and potentially increased the need for hospital care. Further exploration of the issues above may help to identify key targets of interventions for better health outcomes. Secondly, the sole use of ED admission rate may not reflect inpatient bed needs reliably during a pandemic. The analysis of ED admission trends should also include admitted ward types and a good understanding of ED workflow and protocols during the COVID-19 period. Thirdly, the pandemic presents timely opportunities for policymakers to understand how some ED patient groups can be managed appropriately outside of ED for better management of ED crowding in a post-COVID-19 era.
The COVID-19 pandemic is still showing no signs of abating globally and the resurgence of outbreaks, as seen in many other countries, is inevitable as Singapore reopens its borders. A deeper understanding on the pandemic’s indirect impact on patient outcomes and the overall demand for emergent care across the nation is needed. As Singapore eased from social interaction mitigation measures in a stepwise manner, further studies on patients presenting to the ED during this period are needed to fully understand the health implications of the COVID-19 outbreak. It is also essential to explore public perceptions on seeking treatment at ED during the COVID-19 period. Moreover, the prediction of a post-COVID-19 peak rebound in ‘Non-Fever’ visits will facilitate informed decision-makings in resource allocation. It will be useful to have a whole-system model in the future which estimates the healthcare resources needed for health systems to adapt rapidly in various outbreak scenarios. This tool will inform policy responses and minimise the impact of an evolving outbreak on Singapore’s health systems’ outcomes.
As this was a single-centre study based on SGH’s patient profile, and the analysis was performed using data collected from SGH’s eHINTS-ED database, our findings may not be generalisable to other EDs in Singapore and we do not know whether more or less patients are seeking treatment at other hospitals. Although this study was not able to prove causation due to the nature of a descriptive analysis, our findings provide first insights on how ED utilisation at SGH has changed across different subperiods of the COVID-19 pandemic, laying out important questions and implications that future studies should address.
Another limitation of our study lies in our diagnosis dataset. As our dataset did not allow differentiation between primary and secondary diagnoses for each ED visit, all documented diagnoses were studied, and the proportion of each diagnosis category was analysed as a percentage of total number of documented diagnoses. Consequently, the results may have under-represented some diagnosis categories as there were more documented diagnoses than ED visits, or over-represented some diagnosis categories if secondary diagnoses were concentrated in those categories. Nevertheless, the broad conclusion remains that the frequency of nearly every diagnosis category has declined and the number of respiratory-related presentations have gone up during the COVID-19 period. Also, we were not able to map 3000 SNOMED CT codes to ICD-10, which represent 232 unique SNOMED CT diagnosis. However, these were fairly distributed across multiple diagnosis categories. As such, their exclusion will not likely alter the general conclusions of this study.