Changes in paediatric Ambulatory Care Sensitive Conditions in Victoria, 2018–20: the COVID-19 effect?

Keywords: Ambulatory Care Sensitive Conditions, Australia, COVID-19, emergency department, hospital admissions, linked data, paediatric, pandemic, SARS‑CoV‑2, Victoria.

Ambulatory Care Sensitive Conditions (ACSCs) are a defined group of conditions whereby adequate and timely primary level care may lead to avoidance of hospitalisation.¹ Concerns about infection transmission during the coronavirus disease 2019 (COVID-19) pandemic led to significant societal changes worldwide in 2020. In Australia, there was a shift away from in-person consultations towards remote or telehealth consults. Visits to general practitioners declined,² limiting the capacity to examine patients or offer preventive care at the primary care level. These changes also extended to dental care,³ where access to care was restricted to emergency situations only. There were concerns that these changes in access to traditional primary and preventative health care would lead to worsening health outcomes in children.

This study aimed to examine the associations between the timing of the COVID-19 pandemic and its associated public health restrictions, and to the characteristics of paediatric ACSCs in Victoria, Australia betweenJanuary 2018 and October 2020. Specifically, we sought to describe the number and urgency of emergency department (ED) presentations with paediatric ACSCs during the study period (overall and condition-specific), the proportion of those presenting to the ED who required admission to hospital, and the absolute number of hospital admissions.

We built a linear prediction model using routinely collected data from Victorian hospitals and EDs to examine the impact of the initial months of the COVID-19 pandemic on health service use for paediatric ACSCs in Victoria.

We obtained ED presentation data from the Victorian Emergency Minimum Dataset (VEMD) and hospital separation data from the Victorian Admitted Episodes Dataset (VAED). The latter comprises de-identified demographic, clinical and administrative details for all admitted episodes occurring in both public and private Victorian hospitals.⁴ The VEMD data comprises de-identified demographic, administrative and clinical data, which is mandatorily collected from all 39 Victorian public hospitals with a designated ED.⁵ Each presentation has a single principal diagnosis recorded by an ED clinician using the International Statistical Classification of Diseases and Related Health Problems (Tenth Revision, Australian Modification (ICD-10-AM) codes).⁶

All episodes with presentations or admissions for ACSC in patients aged <18 years between 1 January 2018 and 31 October 2020, in Victoria, Australia, were included. We applied the 2020 definitions of a list of potentially preventable hospitalisations maintained by the Australian Institute of Health and Welfare⁷ to capture the conditions of interest for both VAED and VEMD. The following conditions were identified as relevant to children: urinary tract infection (UTI, including pyelonephritis); iron deficiency anaemia; diabetes complications; dental conditions; ear, nose and throat (ENT) infections; convulsions and epilepsy; and asthma. Cellulitis and vaccine-preventable conditions are also included within a traditional definition of ACSCs; however, they are not included in the data analysis. This was first due to the fact that the ACSC definition of cellulitis requires exclusion of certain procedures,⁸ the codes for which are not available in the VEMD. Second, we were unable to build a meaningful prediction model using the data for vaccine-preventable conditions as a result of the unusual spike in 2019 caused by an outbreak of influenza, which significantly skewed the pre-COVID-19 data. A full list of included and excluded ACSCs, along with the relevant ICD-10 codes used for data extraction, is available in supplementary online material (Supplementary File S1).

The ED triage category in the VEMD records the Australasian Triage Scale, stratified according to urgency of need for medical and nursing care: 1 (resuscitation), 2 (emergency), 3 (urgent), 4 (semi-urgent), 5 (non-urgent).⁹ ED presentations that had a triage 1 or 2 classification were defined as ‘urgent’ episodes, whereas the rest were classified as ‘non-urgent’. Child socioeconomic status (SES) was derived from the patient residential postcode using the Australian Bureau of Statistics (ABS) Socio-Economic Indexes for Areas (SEIFA) and its derived Index of Relative Socio-economic Disadvantage (IRSD).¹⁰ The IRSD is an index of economic and social conditions of people and households in an area, based on census data; lower scores correspond to greater disadvantage. The IRSD was dichotomised for each patient presentation; ‘lower’ SES incorporated IRSD decile areas 1–5, whereas ‘higher’ SES incorporated IRSD deciles 6–10. Remoteness was defined based on the Remoteness Areas Structure within the Australian statistical geography standard (ASGS) of the ABS,¹¹ dividing Australia into five classes of remoteness based on a measure of relative access to services: major cities, inner regional, outer regional, remote, and very remote. Inner regional, outer regional, remote, and very remote areas were aggregated and classified as regional Victoria, which covers all areas of Victoria, except metropolitan Melbourne.

The number of ED presentations and hospital admissions were converted into a monthly time series format and the entire observational period was divided into a pre-COVID-19 period and a COVID-19 period. The COVID-19 period in this Victorian context was defined from 31 March 2020 (start of stage 3 lockdown in Victoria when children could no longer attend school in person, community and indoor sports were closed, and employees could not attend the workplace if reasonable and practicable to work from home) to 31 October 2020 (end of the lockdown in 2020).¹² Thus, the study period comprised 27 ‘pre-COVID-19’ and 7 ‘COVID-19’ months of observations, giving a total of 34 observation points.

Based on the observations occurring in the pre-COVID-19 period, a Poisson prediction model (with the number of presentations as the outcome variable and the time passed from the start of the observational period as the predictor) was built for each disorder group to predict what the number of monthly presentations/admissions in the COVID-19 period would have been, if the COVID-19 pandemic and associated restrictions had not happened and the pre-COVID-19 trend had continued. To quantify the uncertainty in the randomness associated with the point being predicted as well as in the coefficient estimates, 95% prediction intervals (PI) of the predicted ED presentation/admission numbers are reported. The difference between the observed and the predicted number was computed. Relative difference was then calculated by subtracting the predicted from the observed and dividing this by the predicted number. For each condition, a mean difference was then computed by averaging the sum of the monthly difference, and a mean relative difference was computed by averaging the sum of the relative difference of each month in the COVID-19 period.

Subgroup analyses by remoteness, socioeconomic status, presentation urgency, admitted proportion, sex, and age were conducted using a chi-squared test to investigate any characteristic change associated with ED presentations/hospital admissions before and during the COVID-19 restrictions.

Data were examined for distribution. Significantly skewed data (length of stay, LOS) was log transformed prior to analysis, based on the assumption that the distribution of LOS is log normal. The VEMD data were also assessed in terms of the ratio of presentations per individual to allow for an assessment of repeat presentations by the same child. All analyses were performed in R version 4.0.3 (R Development Core Team).

The study obtained organisational approval from The Royal Children’s Hospital for an exemption from ethical review (Project ID: 70069).

In total, there were 108 104 paediatric ACSC ED presentations in Victoria between 1 January 2018 and 31 October 2020, resulting in 31 371 admissions (28.9% of total presentations). A significant reduction in absolute presentation numbers was seen in the COVID-19 period compared to previous years (41 319 in 2018; 44 978 in 2019; and 21 807 until October 2020), with a decline in the proportion of presentations resulting in admission in 2020; 12 521 in 2018 (30.3%); 12 959 in 2019 (28.8%); and 5791 in 2020 (26.6%). This represents data for 70 813 individuals (33 069 in 2018; 22 383 in 2019; and 15 361 in 2020), 74 595 (69%) of whom were residing in metropolitan areas. Females accounted for 51 462/108 104 (47.6%) of all presentations, with a median [interquartile range; IQR] age of 3 years [1, 8]. Overall, in 2020, a significant reduction in presentations with ACSCs was detected at the peak of public health restrictions (March and September), with rebound spikes when lockdown measures were eased (June and October)consistently remaining below pre-pandemic levels (Fig. 1).

Fig. 1.  Trend in emergency department presentations with paediatric Ambulatory Care Sensitive Conditions, in Victoria, Australia from January 2018 to October 2020.

The largest decline in the absolute number of presentations was observed relating to asthma, followed by ENT infections and convulsions/epilepsy (Table 1), which, when the data were analysed in the context of the number of observed versus predicted presentations, were noted to be <95% PI at all data timepoints (Fig. 2). An increase in the presentation: individual ratio was seen for convulsions/epilepsy and asthma during the COVID-19 period, whereas this measure was stable for all other conditions over time. Metropolitan Melbourne experienced a larger decrease in ED presentations relating to UTI, ENT infections, convulsions/epilepsy and asthma than regional Victoria.

Table 1.  Emergency department presentations with paediatric Ambulatory Care Sensitive Conditions in Victoria, Australia from January 2018 to October 2020.

Fig. 2.  Observed versus predicted emergency department presentations with paediatric ACSCs, in Victoria, Australia from January 2018 to October 2020.

A decrease in the number of low urgency ED presentations was noted from February 2020, rebounding during periods when lockdown measures were eased (Fig. 3). Higher urgency (triage category 4 and 5) presentations were stable over 2020, albeit at a slightly lower rate compared to either 2018 or 2019.

Fig. 3.  Emergency department presentations for paediatric ACSCs by severity, in Victoria, Australia from January 2018 to October 2020.

Similarly, although the number of children presenting to the ED who were admitted to hospital reduced in line with the reductions in overall numbers (Fig. 4), the proportion who were admitted to hospital was stable over time for UTIs, iron deficiency anaemia, and dental conditions (Table 2). The proportion of children who attended with presentations relating to asthma and ENT infections who were admitted in the COVID-19 period reduced, whereas an increase in the proportion admitted was seen for those presenting with convulsions/epilepsy and diabetes complications (Table 3). In keeping with the above, a significant reduction was also seen in the absolute numbers of admissions for paediatric ACSC during COVID-19 lockdown periods (Table 4). There was no remarkable increase in LOS in hospital in 2020, and the overall number of intensive care unit (ICU) admissions slightly decreased during COVID-19 lockdown periods for children with ACSCs, with the exception of those who presented with diabetes complications.

Fig. 4.  Number and proportion of children presenting with ACSCs who were admitted from the emergency department, in Victoria, Australia from January 2018 to October 2020.

Table 2.  Demographic and outcome data relating to emergency department presentations with paediatric Ambulatory Care Sensitive Conditions, in Victoria, Australia from January 2018 to October 2020.

Table 3.  Observed versus predicted emergency department presentations with paediatric Ambulatory Care Sensitive Conditions, in Victoria, Australia from January to October 2020.

Table 4.  Admissions with paediatric Ambulatory Care Sensitive Conditions with ACSCs, in Victoria, Australia from January to –October 2020.

In this first Australian report on the trends in paediatric ACSC across the first year of the COVID-19 pandemic, reduced presentations are shown in Victoria for conditions that are typically mediated or moderated by transmission of viral infections, notably ENT infections, seizures/epilepsy and asthma. The proportion of children presenting with ACSCs who required hospital admission was stable or reduced over time, with the exception of presentations relating to convulsions/epilepsy and diabetes complications, indicating that severity of presentation was not increased for the majority of children.

This reduction in viral-mediated conditions echoes international experience^13–16 and likely occurred as a combined result of increased handwashing, increased use of hand sanitisers, social distancing, school closures and reduced access to childcare facilities. In addition to the reductions in viral transmission,¹⁷ this may also reflect increased parental supervision and hence compliance with medications, reduction in air pollution due to reduced road traffic and an increase in preventative prescribing.¹⁸

Dental care provision effectively ceased in Victoria, except in emergency situations. Despite this, there was no observed increase in the number of dental presentations, which is of concern given that it would be expected that dental illness was still developing. Our data indicate that this has not been detected and in is contrast with the reported experience in Europe.¹⁹ No changes were seen in terms of the numbers or urgency of presentations relating to UTIs or anaemia, which has not previously been reported, but is expected given that these conditions are less likely to be affected by the environmental and societal changes that were experienced in Victoria in 2020.

Although we found no change in the numbers of presentations for diabetes complications, we did find an increase in the proportion of those who were admitted to hospital and also to ICU in 2020 compared to previous years. A rise in diabetic ketoacidosis (DKA) was seen during the first waves of COVID-19 in the quartile of countries with the highest COVID-19 mortality,²⁰ but within Australia, there have been conflicting reports regarding changes to patterns of severity of presentation at diagnosis of type 1 diabetes.^21,22 Given that we were unable to distinguish whether DKA was arising as part of a new presentation or from complications of existing diabetes, interpretation is limited to an interrogation of trends in presentation numbers over time, which did not increase in 2020. In the absence of clinical data (including acid–base status), we were limited to using admission rates and ICU admission as surrogate markers of severity. The vast majority of children with new-onset diabetes in Victoria are admitted to hospital for education and initial management, which makes hospital admission less relevant as an outcome measure in this population. The rates of admission to ICU may reflect increased bed availability (due to reduced surgical procedures and by the inherent reductions in overall ED presentations with viral infections) and, therefore, a more appropriate management pathway for children with severe DKA (who often require a prolonged period of ED management when ICU beds are not available) rather than a change in severity, per se.

Due to a large second wave of infections, those living in Victoria, particularly metropolitan Melbourne, endured longer public health restrictions than any other Australian state in 2020.²³ These state-based population data provide a powerful insight into the function of the Victorian healthcare system during 2020. However, there are a number of factors that limit interpretation and further analyses. Analysis of ED data by diagnosis is complicated by the limited diagnostic information extracted from ED records and the quality of data entered,²⁴ resulting in potential data gaps. In the Victorian datasets, a single diagnostic code is recorded and so it is not possible to assess presentation in terms of complexity (e.g. child with cerebral palsy and seizures) or to distinguish new-onset from established conditions (e.g. presentations with diabetes). As discussed above, in the absence of granular clinical data, admission to hospital and/or ICU was used a surrogate marker of severity. The latter in itself is subject to criticism, as the decision to admit a child to ICU may have been impacted by increased capacity of paediatric ICUs, as described above. Our findings suggest that health service planning during future pandemic events should take into account the likely fluctuations in ACSC presentations if similar lockdown measures are implemented, with a focus on ensuring that the most vulnerable people in our communities retain timely access to appropriate health care.

The societal changes and restrictions during the COVID-19 pandemic in Victoria resulted in an overall reduction in the number of viral-mediated paediatric ACSCs, without any short-term impact on the urgency of presentation of these conditions. Presentation numbers rebounded quickly between lockdown periods, and it is to be expected that further spikes in ED visits will occur as restrictions are eased further. As expected, there were no changes in the number or urgency of presentations for conditions that would not typically be affected by common viruses or interactions with the environment such as diabetes, iron deficiency anaemia and urinary tract infections.

The public health measures that were implemented in 2020 in Victoria were not only successful at avoiding widespread dissemination of COVID-19, but also led to a significant reduction in the transmission of endemic childhood viral illnesses, providing evidence for consistent messaging in terms of the benefits of handwashing and social exclusion during illness. This resulted in an increased capacity of the EDs to respond to any potential increases in SARS‑CoV‑2 presentations, which in this instance was avoided.

Supplementary material is available online.

The data that support this study were obtained from the Centre for Victorian Data Linkage (CVDL) by permission. Data will be shared upon reasonable request to the corresponding author subject to permission from the CVDL data custodians.

The authors declare no conflicts of interests.

This project was supported by the Victorian Department of Health’s Victorian Collaborative Healthcare Recovery Initiative. The Murdoch Children’s Research Institute is supported by the Victorian Government’s Operational Infrastructure Support Program. HH is supported by a National Health and Medical Research Council (NHMRC) Practitioner Fellowship (1136222). SC’s work on this project has been supported by a grant from the Monash Health Foundation COVID-19 Research Fund.