Threat of COVID-19 impacting on a quaternary healthcare service: a retrospective cohort study of administrative data

Abstract

Objectives
The threat of a pandemic, over and above the disease itself, may have significant and broad effects on a healthcare system. We aimed to describe the impact of the SARS-CoV-2 pandemic (during a relatively low transmission period) and associated societal restrictions on presentations, admissions and outpatient visits.

Design
We compared hospital activity in 2020 with the preceding 5 years, 2015–2019, using a retrospective cohort study design.

Setting
Quaternary hospital in Melbourne, Australia.

Participants
Emergency department presentations, hospital admissions and outpatient visits from 1 January 2015 to 30 June 2020, n=896 934 episodes of care.

Intervention
In Australia, the initial peak COVID-19 phase was March–April.

Primary and secondary outcome measures
Separate linear regression models were fitted to estimate the impact of the pandemic on the number, type and severity of emergency presentations, hospital admissions and outpatient visits.

Results
During the peak COVID-19 phase (March and April 2020), there were marked reductions in emergency presentations (10 389 observed vs 14 678 expected; 29% reduction; p<0.05) and hospital admissions (5972 observed vs 8368 expected; 28% reduction; p<0.05). Stroke (114 observed vs 177 expected; 35% reduction; p<0.05) and trauma (1336 observed vs 1764 expected; 24% reduction; p<0.05) presentations decreased; acute myocardial infarctions were unchanged. There was an increase in the proportion of hospital admissions requiring intensive care (7.0% observed vs 6.0% expected; p<0.05) or resulting in death (2.2% observed vs 1.5% expected; p<0.05). Outpatient attendances remained similar (30 267 observed vs 31 980 expected; 5% reduction; not significant) but telephone/telehealth consultations increased from 2.5% to 45% (p<0.05) of total consultations.

Conclusions
Although case numbers of COVID-19 were relatively low in Australia during the first 6 months of 2020, the impact on hospital activity was profound.

Full text

1
McNamaraE, etal. BMJ Open 2021;11:e045975. doi:10.1136/bmjopen-2020-045975
Open access
Threat of COVID-19 impacting on a
quaternary healthcare service: a
retrospective cohort study of
administrative data
Elissa McNamara,1 Leanne Saxon ,2 Katherine Bond,3 Bruce CV Campbell,4
Jo Douglass,5,6 Martin J Dutch,7,8 Leeanne Grigg,9 Douglas Johnson,1,10,11,12
Jonathan C Knott,6,13 Digsu N Koye,5,14 Mark Putland,13 David J Read,15
Benjamin Smith,11,12 Benjamin NJ Thomson,16 Deborah A Williamson,3,12
Steven YC Tong,11,12 Timothy N Fazio1,6,17
To cite: McNamaraE,
SaxonL, BondK, etal. Threat
of COVID-19 impacting on a
quaternary healthcare service:
a retrospective cohort study of
administrative data. BMJ Open
2021;11:e045975. doi:10.1136/
bmjopen-2020-045975
►Prepublication history and
supplemental material for this
paper is available online. To
view these les, please visit
the journal online (http:// dx. doi.
org/ 10. 1136/ bmjopen- 2020-
045975).
EM and LS are joint rst authors.
Received 30 October 2020
Accepted 28 May 2021
For numbered afliations see
end of article.
Correspondence to
Professor Steven YC Tong;
Steven. Tong@ mh. org. au
Original research
© Author(s) (or their
employer(s)) 2021. Re- use
permitted under CC BY- NC. No
commercial re- use. See rights
and permissions. Published by
BMJ.
ABSTRACT
Objectives The threat of a pandemic, over and above
the disease itself, may have signicant and broad effects
on a healthcare system. We aimed to describe the
impact of the SARS- CoV-2 pandemic (during a relatively
low transmission period) and associated societal
restrictions on presentations, admissions and outpatient
visits.
Design We compared hospital activity in 2020 with the
preceding 5 years, 2015–2019, using a retrospective
cohort study design.
Setting Quaternary hospital in Melbourne, Australia.
Participants Emergency department presentations,
hospital admissions and outpatient visits from 1 January
2015 to 30 June 2020, n=896 934 episodes of care.
Intervention In Australia, the initial peak COVID-19 phase
was March–April.
Primary and secondary outcome measures Separate
linear regression models were tted to estimate the
impact of the pandemic on the number, type and severity
of emergency presentations, hospital admissions and
outpatient visits.
Results During the peak COVID-19 phase (March and
April 2020), there were marked reductions in emergency
presentations (10 389 observed vs 14 678 expected;
29% reduction; p<0.05) and hospital admissions (5972
observed vs 8368 expected; 28% reduction; p<0.05).
Stroke (114 observed vs 177 expected; 35% reduction;
p<0.05) and trauma (1336 observed vs 1764 expected;
24% reduction; p<0.05) presentations decreased; acute
myocardial infarctions were unchanged. There was
an increase in the proportion of hospital admissions
requiring intensive care (7.0% observed vs 6.0%
expected; p<0.05) or resulting in death (2.2% observed
vs 1.5% expected; p<0.05). Outpatient attendances
remained similar (30 267 observed vs 31 980 expected;
5% reduction; not signicant) but telephone/telehealth
consultations increased from 2.5% to 45% (p<0.05) of
total consultations.
Conclusions Although case numbers of COVID-19 were
relatively low in Australia during the rst 6 months of
2020, the impact on hospital activity was profound.
INTRODUCTION
The COVID-19 pandemic has affected hospi-
tals in varied ways. The usual business of
providing care to patients without COVID-19
has altered, and the pattern of presentations
and admissions has changed. For instance,
as the number of COVID-19 cases rose, a
decrease in overall number of emergency
presentations has been reported, ranging
from 49% in the UK to 88% in Italy.1–4 Unan-
ticipated and indirect impacts on hospital
services occurred even in regions with rela-
tively few reported COVID-19 cases.
Before July 2020, Australia had been rela-
tively spared. As of 30 June 2020, there had
been 8023 COVID-19 cases and 104 deaths
among a population of 25 million. In the
second- largest jurisdiction, Victoria, there
Strengths and limitations of this study
►The impact of the threat of COVID-19 in 2020 was
observed at a quaternary referral hospital in Victoria
across three settings: the emergency department,
hospital admissions and hospital outpatient visits
and compared with the preceding 5 years, totalling
896 934 episodes of care.
►Modelling patient data over the last 5 years, rather
than last year alone, provides a stronger prediction
of what the numbers in 2020 should have been.
►Not only were the changes in number and type of
presentations explored, but also the impact on vul-
nerable populations.
►We explored whether admitted patients were more
unwell during the peak of COVID-19.
►It is not known if patients who avoided presenting to
the Royal Melbourne Hospital during the COVID-19
outbreak sought care elsewhere, such as communi-
ty general practitioners or local hospitals.
on June 24, 2021 by guest. Protected by copyright.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-045975 on 24 June 2021. Downloaded from

2McNamaraE, etal. BMJ Open 2021;11:e045975. doi:10.1136/bmjopen-2020-045975
Open access
had been 2159 cases. The Victorian government declared
a state of emergency on March 16 and shut down all non-
essential activities. Health services were instructed to
suspend non- urgent surgery to maintain surge capacity.
The federal government introduced widened criteria for
telehealth consultation rebates to encourage the use of
telehealth.
The Australian context provides a unique opportu-
nity to examine the effects of the threat of COVID-19
on healthcare utilisation. Using hospital administrative
data from the Royal Melbourne Hospital, one of Victo-
ria’s largest hospitals and designated hospital for treating
patients with COVID-19, we determined changes in the
number, type and severity of emergency presentations,
hospital admissions and hospital outpatient visits, during
the first half of 2020 compared with the preceding 5 years,
2015–2019. An understanding of changes can improve
planning, public health messaging and resource manage-
ment for future surges.
METHODS
Study design and population
The Royal Melbourne Hospital is a major metropolitan,
quaternary referral and teaching hospital, operating
approximately 800 beds. It is one of two major trauma
referral centres in Victoria and one of Australia’s leading
public hospitals. Patients are transferred to the Royal
Melbourne Hospital from tertiary organisations across
Victoria and Tasmania. Commencing 1 March 2020, the
hospital opened a fever clinic for the screening of patients
for SARS- CoV-2 and became a designated hospital for
treating patients with COVID-19 .
Administrative data are collected on all patients and
can be accessed through an integrated data warehouse,
which links data from source systems including the patient
administration system, electronic health records and offi-
cial diagnostic coding data. We obtained episode level
data on all emergency department (ED) presentations
and hospital outpatient visits from 1 January 2015 to 30
June 2020. Admissions data were collected from 1 January
2015 to 31 May 2020 (June was excluded as discharge
coding for admissions was incomplete at the time of
data acquisition on 15 July 2020). Data fields included
demographics, ED discharge coding of diagnoses, prin-
cipal International Classification of Diseases-10 diagnosis
at hospital discharge, length of stay, intensive care unit
(ICU) stay, triage category according to the Australasian
Triage Scale (ATS, range 1–5, where 1 is most critical),
COVID-19 diagnosis, in- hospital mortality; and for outpa-
tient visits the modality of visit (in person, telephone, tele-
health). We accessed publicly available data on Victorian
COVID-19 notifications.
Patient and public involvement
No patient was involved. The study includes deidenti-
fied patient data from the Royal Melbourne Hospital,
Australia. It was not appropriate or possible to involve
patients or the public in the design, or conduct, or
reporting, or dissemination plans of our research.
Statistical analysis
To determine whether there were changes in the number
of different hospital cases in 2020 compared with the
preceding 5 years, we used linear regression models to
test the null hypothesis that there was no difference in
the number between the prepandemic (2015–2019) and
postpandemic (2020) periods. Separate linear regres-
sion models were fitted to the number of (1) ‘emergency
presentations’, excluding the dedicated COVID-19 fever
clinic; (2) ‘admissions’ including healthcare provided in
the home, that is, intravenous treatment for patients with
pneumonia, but excluding single day admissions, statis-
tical separations (cessation of an episode of patient care),
organ procurement, maternity and birth episodes; (3)
‘outpatient visits’, excluding missed appointments; (4)
patients with one of eight primary diagnoses: ‘pneumonia’,
‘trauma’, ‘mental health or substance abuse’, ‘acute coro-
nary syndrome’, ‘stroke’, ‘appendicitis’ or ‘cellulitis’
(expressed as a total number for the 1 or 2- month period)
(see online supplemental eTables 1–7 for categorisation
codes). These eight diagnoses were prespecified for
detailed analyses as they were hypothesised to change
in numbers of presentations during the pandemic and
each comprised a significant proportion of all presen-
tations; (5) patients within each ‘triage category’ (ATS
1–3=high acuity, 4–5=low to moderate); (6) ‘ICU admis-
sions’ (number of patients admitted to ICU expressed as
a percentage of admitted patients); (7) ‘deaths’ (number
of admitted patients who died, expressed as a percentage
of admitted patients); (8) patients from different suburbs
across Melbourne (expressed as a proportion of admitted
patients) as well as (9) ‘length of stay’ (excluding short-
term stay ≤24 hours and presented as the average dura-
tion in hours for all patients), against ‘year of admission’
(prepandemic/postpandemic) and ‘date of admission’
(dd/mm/year) to adjust for seasonality. The number of
‘ICU admissions’, ‘number of deaths’ and ‘length of stay’
were adjusted for ‘age’ and ‘gender’ in addition to ‘year
of admission’ and ‘date of admission’ since we expected
these outcomes to vary depending on the patient demo-
graphic presenting to the hospital, that is, an older cohort
of patients in 2020 may explain a rise in deaths and there-
fore, may not be associated with the pandemic. All other
dependent variables were used to represent healthcare
utilisation, regardless of the patient demographic; they
were not adjusted for age and gender.
To calculate the mean difference (and 95% CI) between
expected and observed numbers for 2020, we fitted a
linear regression model of case numbers against 2015–
2019 and predicted numbers for 2020 and compared
these with actual numbers for 2020. We used Gaussian
regression models for continuous variables and Poisson
regression models for all count variables. A conservative
approach was used to check for statistical differences
(p<0.05) based on whether the 95% CIs of the difference
on June 24, 2021 by guest. Protected by copyright.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-045975 on 24 June 2021. Downloaded from

3
McNamaraE, etal. BMJ Open 2021;11:e045975. doi:10.1136/bmjopen-2020-045975
Open access
overlapped with zero. We used the same models to observe
the pandemic’s effects from January to February, March
to April and June to July. These 2- month windows repre-
sented the pandemic’s pre, peak and transition phases,
respectively, and were compared with the same periods in
2015–2019. Only the January–June period was considered
because this is the available data for 2020 and it captured
the first wave of the COVID-19 outbreak when it posed
more of a threat to overwhelming the health service than
a reality.
The regression models showed little autocorrela-
tion as indicated by the Durbin- Watson test and the
residuals showed only minor deviations from normality
when using the univariate kernel density estimation.
Overdispersion was not detected in any analysis as indi-
cated by the Pearson χ2 dispersion statistic. Because this
is an exploratory study and not a study that would lead
to a change in clinical practice, we did not correct for
multiple- hypothesis testing. All analyses were conducted
using Stata V.16 (StataCorp, Texas, USA) and R V.4.0.2 (R
Project for Statistical Computing).
RESULTS
Observations during January–Jun 2020
From 1 January to 30 June 2020 there were 2159 posi-
tive SARS- CoV-2 notifications in Victoria, Australia, of
which 138 were diagnosed and 28 admitted at the Royal
Melbourne Hospital. From 1 January to 30 June 2020 there
were 47 609 emergency presentations (36 188 excluding
fever clinic), 16 867 admissions (excluding June) and 96
722 outpatient visits. Notifications for SARS- CoV-2 peaked
in Victoria in early April, and a coincident reduction in ED
presentations to the Royal Melbourne Hospital occurred
in March and April (figure 1A). A concomitant increase
in patients screened in the fever clinic was observed in
March and April and peaked in the third week of March,
2 weeks before the peak in COVID-19 cases. Emergency
presentations remained below pre- COVID-19 activity in
May and June.
Similarly, both emergency and elective hospital
admissions were considerably reduced during the peak
COVID-19 period compared with the pre- COVID period
and began to recover in May (figure 1B). Outpatient
appointments dropped for 1 week during the peak
COVID-19 period but quickly recovered (figure 1C). Tele-
health and telephone appointments increased during the
peak COVID-19 period to compensate for the decrease in
face- to- face appointments. They continued to represent a
large proportion of appointments in May and June.
ED episodes
From 2015 to 2019, there was a year- on- year increase
in caseload for all ED presentations (figure 2). During
March–April 2020 when COVID-19 cases peaked, there
was a marked reduction in ED presentations (10 389
observed vs 14 678 expected; 29% reduction, p<0.05)
(table 1). There were fewer trauma (major and minor)
cases (1336 observed vs 1764 expected; 24% reduction,
p<0.05), stroke cases (114 observed vs 177 expected;
36% reduction, p<0.05), mental health and substance
abuse cases (221 observed vs 267 expected; 20% reduc-
tion, p<0.05) and appendicitis cases (54 observed vs 76
expected; 29% reduction, p<0.05) but an increase in
cellulitis cases (89 observed vs 69 expected; 29% increase,
p<0.05). There was no difference in the actual versus
expected number of acute myocardial infarction or pneu-
monia cases.
As Victorian COVID-19 cases began to decline during
the May–June transition phase, the ED continued to
observe fewer overall presentations (11 298 observed vs
13 729 expected; 18% reduction, p<0.05), and there were
fewer high acuity cases (63.9% observed vs 65% expected;
Figure 1 Weekly number of positive COVID-19 cases in VIC
and treated at the RMH and the corresponding changes in
number of ED presentations (A), hospital admissions (B) and
outpatient appointments (C) from 1 January to 30 June 2020.
Complete hospital admission data are available till 30 May
2020. ED, emergency department; RMH, Royal Melbourne
Hospital; VIC, Victoria.
on June 24, 2021 by guest. Protected by copyright.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-045975 on 24 June 2021. Downloaded from

4McNamaraE, etal. BMJ Open 2021;11:e045975. doi:10.1136/bmjopen-2020-045975
Open access
p<0.05) (table 1, online supplemental table). Fewer cases
of pneumonia (32 observed vs 52 expected; 39% reduc-
tion, p<0.05), trauma (1419 observed vs 1564 expected;
9% reduction, p<0.05), appendicitis (36 observed vs 59
expected; 39% reduction, p<0.05) were detected, but
an increase in cellulitis (131 observed vs 99 expected;
32% increase, p<0.05) and mental health and substance
abuse cases (260 observed vs 220 expected; 18% increase,
p<0.05) was found but no difference in stroke or acute
myocardial infarction cases.
The residential postcode of ED arrivals changed
significantly from March to June compared with the
pre- COVID-19 period (travel for medical attention was
an exemption during lockdown). Proportionally fewer
people arrived from outer suburbs compared with
patients from suburbs situated closer to the hospital
(p<0.05) (online supplemental etables 8–10). A smaller
proportion was born outside of Australia (p<0.05).
Inpatient episodes
From 2015–2019, the yearly number of admissions
increased (figure 2). During the peak of COVID-19 cases
in March–April 2020, there were fewer admissions (5972
observed vs 8368 expected; 28% reduction, p<0.05), both
emergency and planned, to the Royal Melbourne Hospital
compared with the equivalent period in 2015–2019
(table 2). This included fewer stroke (134 observed vs 177
expected; 24% reduction, p<0.05), trauma (624 observed
vs 900 expected; 31% reduction, p<0.05), mental health
and substance abuse cases (93 observed vs 166 expected;
19% reduction, p<0.05), cellulitis cases (43 observed vs 88
expected; 51% reduction, p<0.05), and appendicitis cases
(56 observed vs 75 expected; 25% reduction, p<0.05)
but an increase in the number of pneumonia cases (138
observed vs 95 expected; 45% increase, p<0.05), 12 of
whom tested positive for COVID-19. The number of
admissions with acute myocardial infarction were not
different from predicted. A higher proportion of admis-
sions required time in ICU (7% observed vs 6% expected;
17% increase, p<0.05) or died in hospital (2.2% observed
vs 1.5% expected; 47% increase, p<0.05). There was no
difference in the average length of stay.
The transition phase in May continued to show fewer
admissions (3343 observed vs 4616 expected; 28%
reduction, p<0.05) and more patients arriving from ED
required time in ICU (9% observed vs 7.7% expected;
17% increase, p<0.05) (table 2 and online supplemental
eTables 11–13). As with the peak COVID-19 period,
the number of deaths was higher from expected (2.1%
observed vs 1.5% expected; 40% increase, p<0.05).
Outpatient episodes
Monthly outpatient appointments gradually increased
from 2015 to 2019 (figure 2) and showed no significant
change in the total number of appointments during the
COVID-19 peak (30 267 observed vs 31 980 expected; 5%
reduction, not significant) and transition phase (36 656
observed vs 36 878 expected; 6% reduction, not signifi-
cant). During the peak COVID-19 period, telehealth and
telephone appointments made up 45% of all appoint-
ments (45% observed vs 9% expected; p<0.05), and in the
transition phase 56% (56% observed vs 3.4% expected;
Figure 2 Time series of the monthly number of presentations to emergency, inpatient admissions and outpatient appointments
to the Royal Melbourne Hospital from January 2015 to June 2020. The solid line shows the actual numbers recorded for the
month, while the hashed line indicates the predicted numbers based on the underlying trend. Year- by- year trend shows the
case loads are increasing, except in 2020, where in March and April the numbers decreased dramatically and deviate from the
predicted line.
on June 24, 2021 by guest. Protected by copyright.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-045975 on 24 June 2021. Downloaded from

5
McNamaraE, etal. BMJ Open 2021;11:e045975. doi:10.1136/bmjopen-2020-045975
Open access
p<0.05) (table 3 and online supplemental eTables 14–16).
Those attending outpatients in the peak COVID-19
period were younger than expected, and the proportion
of patients born outside of Australia decreased (online
supplemental etables 14–16). The transition from face- to-
face to telehealth was also observed in potentially vulner-
able groups, such as the elderly and those born outside of
Australia (making up 44%–61% of all appointments from
March to June, p<0.05).
DISCUSSION
During an initial mild wave of COVID-19 in Victoria,
there was a marked reduction in the use of hospital
services at a major quaternary, level 1 trauma hospital.
We have provided a broad overview of the changes that
occurred across the hospital services of ED care, hospital
admissions (planned and emergency) and ambulatory
outpatient clinics. Some of these changes were planned
and predictable (eg, deferral of non- urgent elective
surgery cases), some surprising (eg, considerable reduc-
tions in emergency presentations and admissions) and
others forced but timely adaptations (eg, increased use
of telehealth).
We observed a 29% reduction in ED presentations
during March and April 2020 compared with predicted
presentations modelled on data from the previous 5
years. A similar phenomenon has been reported from
countries with high levels of COVID-19 infection rates4–7
and emerging data suggest this may also have occurred in
settings with a low burden of COVID-19.8 With less social
interaction, community surveillance data demonstrated
historically low levels of influenza- like symptoms in
Victoria,9 which may have resulted in fewer presentations
with influenza- like illnesses and also fewer viral exacer-
bations of chronic conditions. Psychological components
may have also been an important factor. Analysis from
previous pandemics has demonstrated that fear of
contracting disease in hospital or concern that hospital
resources are overwhelmed can lead to avoidance of
seeking medical care for patients with non- pandemic-
related illnesses.10 11
There was an apparent reduction in trauma presen-
tations, probably related to less population mobility
including fewer cars on the roads, cancellation of sporting
activities and reduced industrial activities. There were
also fewer stroke presentations, a trend observed else-
where.5 12 13 In contrast to other settings, but similar to
another major Melbourne hospital, we noted no change
in presentations of acute myocardial infarction.14–16
The proportion of people presenting to the ED who
were born outside of Australia, significantly reduced
during March–June 2020. It is concerning that this group
of patients has been accessing less hospital care during
the COVID-19 pandemic, given the health vulnerabilities
of some subsets of the culturally and linguistically diverse
population17 and that approximately 30% of people
living near the hospital have a non- English- speaking
Table 1 Characteristics of people presenting to the emergency department at the Royal Melbourne Hospital from 1 March to 30 April 2015–2020
March–April Mean difference between predicted and observed in 2020 (95% CI)
2015 2016 2017 2018 2019 2020
March–April
Peak COVID-19
January–February
Pre- COVID-19
May–June
transition
Presentations, N 11 011 11 746 12 590 13 012 13 878 10 389 −4289 (−3700 to −4880)* 680 (431 to 929)* −2431 (−2680 to −2180)*
ATS, %
High acuity, ATS 1–3 56.7 56.0 58.4 59.3 62.0 62.6 0.7 (−0.9 to 2.3) −2.5 (−3.3 to −1.7)* −1.1 (−1.8 to −0.3)*
Low to moderate acuity, ATS 4–5 43.3 44.0 41.6 40.7 38.0 37.4 −0.7 (−2.3 to 1.9) 4.2 (1.7 to 3.3)* 1.1 (0.3 to 1.8)*
Conditions presenting, N
Stroke 82 109 110 113 167 114 −63 (−84 to −42)* 19 (−2 to 40) 3 (−30 to 24)
Acute myocardial infarction 80 54 73 62 92 83 1 (−15 to 15) 21 (3 to 39)* 15 (−31 to 1)
Pneumonia 55 34 31 38 41 40 7 (−6 to 20) 17 (1 to 33)* −20 (−32 to −8)*
Cellulitis 74 82 96 76 62 89 20 (1 to 40)* 60 (37 to 83)* 32 (12 to 52)*
Mental health/substance abuse 197 229 231 243 249 211 −56 (−87 to −27)* −17 (−50 to 16) 40 (7 to 74)*
Appendicitis 69 56 76 75 71 54 −22 (−38 to −6)* −11 (−8 to 31) −23 (−36 to −10)*
Trauma 1591 1655 1689 1830 1644 1336 −428 (−527 to −329)* 36 (158 to 316)* −145 (−216 to −74)*
Percentages=(number/daily total) ×100 for the 2- month period. Numbers exclude attendees to the fever clinic in March–April (n=5471) and May–June (n=5925). Additional information is available in online supplemental tables.
Two- month differences between observed versus predicted emergency presentations are provided for March and April 2020 (peak COVID-19), January–February (pre- COVID-19) and May–June (transition period).
*Regression analysis: signicance based on 95% CI (p<0.05).
ATS, Australasian Triage Scale.
on June 24, 2021 by guest. Protected by copyright.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-045975 on 24 June 2021. Downloaded from

6McNamaraE, etal. BMJ Open 2021;11:e045975. doi:10.1136/bmjopen-2020-045975
Open access
Table 2 Characteristics of patients admitted to the Royal Melbourne Hospital from 1 March to 30 April 2015–2020
March–April Mean difference between predicted and observed in 2020 (95% CI)
2015 2016 2017 2018 2019 2020
March–April
Peak COVID-19
January–February
Pre- COVID-19
May
transition period
Admissions, N 5888 6505 7246 7550 7565 5972 −2396 (−2810 to −1981)* −533 (−929 to −135)* −1273 (−1570 to −976)*
Emergency admission, N (%) 3970 (67.4) 4253 (65.4) 4604 (63.5) 4840 (64.1) 4921 (65.1) 3801 (63.7) −1511 (−1709 to −1314)* −222 (−347 to −98)* −627 (−717 to −536)*
Planned admission, N (%) 1918 (32.6) 2252 (34.6) 2642 (36.5) 2710 (35.9) 2664 (34.9) 2171 (36.4) −888 (−1183 to −592)* −318 (−635 to −2)* −648 (−876 to −420)*
Length of stay, hours mean (SD)
Any admission 150.2 (236.0) 149.3 (242.9) 137.7 (239.5) 133.7 (233.4) 144.9 (287.8) 133.5 (211.6) −1.6 (−8.6 to 5.5) −6.0 (−11 to −1.0)* 1.0 (−6.4 to 8.4)
Emergency admission 126.7 (195.9) 124.4 (180.0) 119.7 (197.9) 114.3 (175.5) 122.8 (210.1) 110.9 (164.5) −6.0 (−0.3 to 12.2) −4.6 (−9.1 to −0.1)* 3.4 (−4.0 to 10.9)
Planned admission 198.8 (296.8) 196.3 (325.5) 169.1 (295.7) 168.3 (308.2) 186.0 (390.3) 173.2 (271.0) 19.7 (3.1 to 36.2)* 12.3 (−1.2 to 25.9) 12.5 (−4.6 to 29.7)
Requiring ICU (%)
Any admission 5.7 5.3 5.8 6.3 5.6 7.0 1.0 (0.5 to 1.6)* −0.2 (−0.7 to 0.4) 0.8 (−0.1 to 1.6)
Emergency admission 6.0 5.9 6.7 7.6 6.6 8.1 0.5 (−0.1 to 1.0) −0.5 (−1.0 to 0.1) 1.3 (0.4 to 2.2)*
Planned admission 5.2 4.0 4.2 4.1 3.7 5.0 0.6 (0.2 to 0.9)* 0.3 (0.0 to 0.6) −0.5 (−0.9 to −0.2)*
Died (%)
Any admission 1.9 1.6 1.8 1.4 1.6 2.2 0.7 (0.3 to 1.0)* 0.0 (−0.3 to 0.3) 0.6 (0.2 to 1.0)*
Emergency admission 2.4 2.1 2.0 1.7 2.0 2.3 0.4 (0.1 to 0.7)* 0.0 (−0.3 to 0.2) 0.5 (0.1 to 0.9)*
Planned admission 0.7 0.8 1.4 0.9 0.9 2.0 2.3 (2.1 to 2.5)* 0.0 (−0.1 to 0.1) 0.1 (−0.1 to 0.3)
Conditions presenting, N
Stroke 74 132 157 118 154 134 −43 (−64 to −22)* 9 (−19 to 37) −5 (−25 to 16)
Acute myocardial infarction 106 80 89 101 104 96 −5.5 (−22 to 11) 6 (−12 to 25) −3.6 (−17 to 9)
Pneumonia 88 74 97 84 95 138 43 (20 to 66)* 18 (0 to 38) −34 (−44 to −23)*
Cellulitis 16 63 75 46 71 43 −45 (−58 to −32)* −43 (−59 to −27)* −47 (−55 to −39)*
Mental health/substance abuse 63 111 96 108 96 93 −23 (−36 to −8)* −22 (−42 to −1)* −11 (−24 to 1.8)
Appendicitis 43 55 57 60 68 56 −19 (−32 to −6)* 1 (−15 to 17) −18 (−25 to −11)*
Trauma 683 799 786 880 815 624 −276 (−333 to −220)* −60 (−118 to 0) −117 (−159 to −74)*
Percentages=(number/daily total) × 100 for the 2- month period. Additional information is available in online supplemental tables.
Differences between observed versus predicted admissions are provided for March and April 2020 (peak COVID-19), January–February (pre- COVID-19) and May (transition period).
*Signicance based on 95% condence interval (p<0.05).
on June 24, 2021 by guest. Protected by copyright.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-045975 on 24 June 2021. Downloaded from

7
McNamaraE, etal. BMJ Open 2021;11:e045975. doi:10.1136/bmjopen-2020-045975
Open access
background.18 This highlights the importance of public
health messaging that hospitals continue to be opera-
tional and are safe places to access necessary care.19
During March–May, while the total number of admis-
sions decreased, we observed significant increased
mortality or need for ICU support among both planned
and emergency admissions. These findings suggest that
the patients presenting to the hospital during the threat
of COVID-19 and immediately after were more unwell
than before the pandemic began.16 Local and interna-
tional data suggest a delay in seeking medical care during
the COVID-19 outbreak,3 20–22 resulting in more severe
pathology at presentation. In Victoria, there was a 2%–3%
increase in the number of people dying in the community
during May compared with other months.23 Although
further examination is required, it broadly supports our
assertion that people were avoiding or delaying care
during this time. An overwhelming influx of COVID-19
cases, as seen elsewhere,24 can be discounted as a reason
for worse outcomes at the Royal Melbourne Hospital. The
underlying causes merit further exploration.
Admissions with pneumonia increased from March to
April, which could not be explained by the 12 cases of
COVID-19. We were surprised by this increase as there
was less community transmission of influenza in Victoria
during the stay- at- home order period. It is possible that
pneumonia cases, that might have otherwise been treated
in the community, were instead managed in the hospital
due to concern of a COVID-19 diagnosis.
There was a remarkable and rapid increase in telehealth
and telephone outpatient appointments during the peak
March–April period, with changes persisting after the
threat of COVID-19 had diminished. Patients accessing
telehealth appointments appeared to be younger and less
likely to be born outside of Australia. Using the telephone
or computer was not a significant barrier for 44%–61% of
older patients, including those born outside Australia, who
used these options instead of face- to- face appointments.
The severity of illness of the older patients who used tele-
health is unknown, nor is it known what their motivation
was for using it. While it is likely they feared contracting
COVID-19 at the hospital, we cannot discount their shift
in behaviour was simply because the clinicians promoted
it. A recent survey of clinicians and patients from the
Royal Melbourne Hospital reported that the standard of
outpatient care was not compromised by using telehealth
compared with on- site appointments.25 Although access
to care mostly continued during the pandemic, it will be
important to ensure that services cater to any disadvan-
taged patient groups.
Our study has some limitations. Coding lags for inpa-
tient admissions resulted in only 1 month of data to
represent the transition period after the initial peak of
COVID-19. Although the Royal Melbourne Hospital
is one of the largest hospitals state- wide, our data are
restricted to a single site. Future studies could explore
linking datasets between hospitals, general practice and
community health databases to examine whether there
Table 3 Characteristics of patients attending outpatient appointments at the Royal Melbourne Hospital from 1 March to 30 April 2015–2020
March–April Mean difference between predicted and observed in 2020 (95% CI)
2015 2016 2017 2018 2019 2020
March–April
Peak COVID-19
January–February
Pre- COVID-19
May–June
transition
Attended appointments 29 851 30 666 30 890 33 025 35 160 30 267 −1713 (−6483 to 3055) −4609 (−10 135 to 917) −222 (−5897 to 5421)
Face- to- face, N (%) 29 851 (100) 30 666 (100) 30 190 (97.8) 31 536 (95.5) 33 276 (94.6) 16 532 (54.6) −13,219 (−17 068 to −9370)* 3871 (−9137 to 1395) −18 292 (−21 093 to −15 492)*
Telephone, N (%) – – 628 (2.0) 1370 (4.2) 1664 (4.7) 8586 (28.4) 5876 (3448 to 8305)* −2558 (−2,863 to −2253)* 7815 (5739 to 9890)*
Telehealth, N (%) – – 72 (0.2) 119 (0.4) 220 (0.6) 5157 (17.0) 6029 (4366 to 7694)* −160 (−299 to −23)* 7596 (6176 to 9015)*
Age (years) mean (SD) 53.4 (18.4) 53.4 (18.6) 53.1 (18.6) 52.9 (18.6) 53.1 (18.4) 50.7 (18.3) −3.7 (−5.1 to −2.3)* 0.0 (−1.6 to 2.1) −1.7 (−3.4 to 0.1)
Born in Australia, % 55.6 56.0 56.6 57.9 59.3 64.1 4.8 (0.6 to 8.9)* 5.9 (0.2 to 11.6)* 9.2 (5.7 to 12.7)*
Age≥65 years, N 9660 9899 9675 10 165 10 939 8011 −1250 (−2157 to −344)* −1258 (−2688 to 172) −2276 (−3543 to −989)*
Face- to- face, % 100 100 98.0 96.1 96.0 54.3 −36.0 (−45.2 to −26.9)* 1.2 (0.5 to 1.9)* −45.6 (−52.6 to −38.6)*
Telephone, % – – 1.9 3.6 3.5 32.6 24.0 (17.2 to 30.7)* −0.5 (−1.4 to 0.4) 33.3 (27.9 to 38.6)*
Telehealth, % – – 0.1 0.3 0.5 13.0 12.1 (9.3 to 14.8)* 0.0 (−0.1 to 0.1) 12.3 (10.4 to 14.3)*
Two- month differences between observed versus predicted outpatient appointments are provided for March and April 2020 (peak COVID-19), January–February (pre- COVID-19) and May–June (transition period). Additional information is available in
online supplemental tables.
*Signicance based on 95% CI (p<0.05).
on June 24, 2021 by guest. Protected by copyright.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-045975 on 24 June 2021. Downloaded from

8McNamaraE, etal. BMJ Open 2021;11:e045975. doi:10.1136/bmjopen-2020-045975
Open access
is an overall reduction of care- seeking behaviour across
all services or if the drop is limited to specific hospital
services. Although changes in the population of the
primary catchment area of the Royal Melbourne Hospital
could influence the caseload, these are unlikely to
entirely explain the observed decreased numbers during
2020. Occupied bed days increased from a mean of 650
in 2015 to a mean of 780 early 2020 and was reflected
in the regression models that demonstrated an increase
in activity from 2015 to early 2020. The population of
the primary catchment area also increased by 2.5% per
annum from 2015 to 2020.26 On the other hand, there
would have been fewer people inhabiting the local
suburbs due to restricted international travel, difficulties
in residents returning home from overseas, fewer inter-
national students and a decline in tourist numbers. This
may also explain, in part, why fewer people born outside
Australia presented to the hospital during 2020. Never-
theless, it is likely that multiple factors associated with the
pandemic contributed to the abrupt changes in health
service utilisation, and it is beyond the scope of this paper
to explore these in detail. Other explanations cited in the
literature for a decrease in patients without COVID -19
in countries overwhelmed by positive infections do not
apply to our study, where we examine a unique time when
COVID-19 was only a threat. For instance, there was no
general hospital policy to shorten hospital stay or to keep
beds free in case of an influx of patients with COVID-19
or to divert ambulances. Nor was the Royal Melbourne
Hospital overwhelmed by furloughed or redeployed staff
at this stage of the pandemic.27
Our findings raise concern that during the initial threat
of COVID-19, and even after it abated, there has been
a marked reduction in hospital presentations and indi-
cators of increased severity in those presenting. It will
be imperative for public health authorities to improve
community messaging regarding the importance of
seeking timely care. Targeting vulnerable groups who
already have barriers to accessing care will be especially
important. This may require increased investment in
interpreting and community- based outreach services.
Hospitals should prepare for a potential increase in work-
load, not only from patients who had elective procedures
deferred but from patients who avoided care during the
initial threat of COVID-19. At the time of writing, Victoria
is experiencing a second surge in COVID-19 cases.
Ongoing monitoring and analysis of health outcomes
will help inform responses to this and future COVID-19
upsurges or other pandemics.
Author afliations
1Department of General Medicine, The Royal Melbourne Hospital, Melbourne,
Victoria, Australia
2Melbourne Academic Centre for Health, Parkville, Victoria, Australia
3Department of Microbiology, The Royal Melbourne Hospital, Melbourne, Victoria,
Australia
4Department of Medicine and Neurology, Melbourne Brain Centre at The Royal
Melbourne Hospital, University of Melbourne, Melbourne, Victoria, Australia
5The Royal Melbourne Hospital, Melbourne, Victoria, Australia
6Melbourne Medical School, University of Melbourne, Melbourne, Victoria, Australia
7Centre for Integrated Critical Care Research, University of Melbourne, Melbourne,
Victoria, Australia
8Department of Medicine and Radiology, The Royal Melbourne Hospital, Melbourne,
Victoria, Australia
9Department of Cardiology, The Royal Melbourne Hospital, Melbourne, Victoria,
Australia
10Department of Medicine, The Royal Melbourne Hospital, Melbourne, Victoria,
Australia
11Victorian Infectious Diseases Service, The Royal Melbourne Hospital, at the Peter
Doherty Institute for Infection and Immunity, Melbourne, Victoria, Australia
12Department of Infectious Diseases, The University of Melbourne at the Peter
Doherty Institute for Infection and Immunity, Melbourne, Victoria, Australia
13Department of Emergency Medicine, The Royal Melbourne Hospital, Melbourne,
Victoria, Australia
14Melbourne Clinical and Translation Science, University of Melbourne, Melbourne,
Victoria, Australia
15Trauma and Colorectal Units, The Royal Melbourne Hospital, Melbourne, Victoria,
Australia
16University of Melbourne Department of Surgery, The Royal Melbourne Hospital,
Melbourne, Victoria, Australia
17Health Intelligence, The Royal Melbourne Hospital, Melbourne, Victoria, Australia
Contributors EM wrote the discussion. LS wrote methods, results and conducted
data analysis. KB, BCVC, JD, MJD, LG, DJ, JCK, MP, DJR, BS, BNJT and DAW were
responsible for editorial and clinical input. DNK gave statistical advice. SYCT helped
in study design, editorial and clinical input. TNF helped in data extraction, statistical
analysis, editorial and clinical input.
Funding The authors have not declared a specic grant for this research from any
funding agency in the public, commercial or not- for- prot sectors.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval The study was approved as a Quality Assurance Project
(QA2020087) by the Melbourne Health Ofce for Research Ethics & Governance and
Human Research Ethics Committee.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request. The
study includes deidentied patient data from the Royal Melbourne Hospital,
Australia.
Supplemental material This content has been supplied by the author(s). It has
not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been
peer- reviewed. Any opinions or recommendations discussed are solely those
of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and
responsibility arising from any reliance placed on the content. Where the content
includes any translated material, BMJ does not warrant the accuracy and reliability
of the translations (including but not limited to local regulations, clinical guidelines,
terminology, drug names and drug dosages), and is not responsible for any error
and/or omissions arising from translation and adaptation or otherwise.
Open access This is an open access article distributed in accordance with the
Creative Commons Attribution Non Commercial (CC BY- NC 4.0) license, which
permits others to distribute, remix, adapt, build upon this work non- commercially,
and license their derivative works on different terms, provided the original work is
properly cited, appropriate credit is given, any changes made indicated, and the use
is non- commercial. See:http:// creativecommons. org/ licenses/ by- nc/ 4. 0/.
ORCID iD
LeanneSaxon http:// orcid. org/ 0000- 0003- 0666- 0571
REFERENCES
1 Mantica G, Riccardi N, Terrone C, etal. Non- COVID-19 visits to
emergency departments during the pandemic: the impact of fear.
Public Health 2020;183:40–1.
2 Thornton J. COVID-19: A&E visits in England fall by 25% in week
after lockdown. BMJ 2020;369:m1401.
3 Lazzerini M, Barbi E, Apicella A, etal. Delayed access or provision of
care in Italy resulting from fear of COVID-19. Lancet Child Adolesc
Health 2020;4:e10–11.
on June 24, 2021 by guest. Protected by copyright.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-045975 on 24 June 2021. Downloaded from

9
McNamaraE, etal. BMJ Open 2021;11:e045975. doi:10.1136/bmjopen-2020-045975
Open access
4 Mauro V, Lorenzo M, Paolo C, etal. Treat all COVID 19- positive
patients, but do not forget those negative with chronic diseases.
Intern Emerg Med 2020;15:787-790.
5 Kansagra AP, Goyal MS, Hamilton S, etal. Collateral effect of
COVID-19 on stroke evaluation in the United States. N Engl J Med
2020;383:400–1.
6 Park C, Sugand K, Nathwani D, etal. Impact of the COVID-19
pandemic on orthopedic trauma workload in a London level 1 trauma
center: the "golden month". Acta Orthop 2020:1–6.
7 Nuñez JH, Sallent A, Lakhani K, etal. Impact of the COVID-19
pandemic on an emergency traumatology service: experience at a
tertiary trauma centre in Spain. Injury 2020;51:1414–8.
8 Bjornsen LP, Naess- Pleym LE, Dale J, etal. Patient visits to an
emergency department in anticipation of the COVID-19 pandemic.
Tidsskr Nor Laegeforen 2020;140.
9 Dalton CB, Carlson SJ, Butler MT, etal. Flutracking weekly online
community survey of inuenza- like illness annual report, 2010.
Commun Dis Intell Q Rep 2011;35:288–93.
10 Chen T- A, Lai K- H, Chang H- T. Impact of a severe acute respiratory
syndrome outbreak in the emergency department: an experience in
Taiwan. Emerg Med J 2004;21:660–2.
11 Schull MJ, Stukel TA, Vermeulen MJ, etal. Effect of widespread
restrictions on the use of hospital services during an outbreak of
severe acute respiratory syndrome. CMAJ 2007;176:1827–32.
12 Rudilosso S, Laredo C, Vera V, etal. Acute stroke care is at risk in
the era of COVID-19: experience at a comprehensive stroke center in
Barcelona. Stroke 2020;51:1991–5.
13 Zhao J, Li H, Kung D, etal. Impact of the COVID-19 epidemic on
stroke care and potential solutions. Stroke 2020;51:1996–2001.
14 Braiteh N, Rehman WU, Alom M, etal. Decrease in acute coronary
syndrome presentations during the COVID-19 pandemic in upstate
New York. Am Heart J 2020;226:147–51.
15 Hauguel- Moreau M, Pilliere R, Prati G, etal. Impact of coronavirus
disease 2019 outbreak on acute coronary syndrome admissions: four
weeks to reverse the trend. J Thromb Thrombolysis 2020.
16 Toner L, Koshy AN, Hamilton GW, etal. Acute coronary syndromes
undergoing percutaneous coronary intervention in the COVID-19 era:
comparable case volumes but delayed symptom onset to hospital
presentation. Eur Heart J Qual Care Clin Outcomes 2020;6:225–6.
17 Chan DKY, Ong B, Zhang K, etal. Hospitalisation, care plans and not
for resuscitation orders in older people in the last year of life. Age
Ageing 2003;32:445–9.
18 Australian Bureau of Statistics. 2016 census community
proleshttps://www. abs. gov. au/ websitedbs/ D3310114. nsf/ Home/
2016% 20Census% 20Community% 20Proles [Accessed 15 Jul
2020].
19 Smith JA, Judd J. COVID-19: vulnerability and the power of privilege
in a pandemic. Health Promot J Austr 2020;31:158–60.
20 Teo K- C, Leung WCY, Wong Y- K, etal. Delays in stroke onset to
hospital arrival time during COVID-19. Stroke 2020;51:2228–31.
21 Bromage DI, Cannatà A, Rind IA, etal. The impact of COVID-19 on
heart failure hospitalization and management: report from a heart
failure unit in London during the peak of the pandemic. Eur J Heart
Fail 2020;22:978–84.
22 Lange PW, Gazzard M, Walker S, etal. Where are our patients?
retrospective cohort study of acute medical unit admissions during
and prior to the COVID-19 pandemic. Intern Med J 2020;50:1132–4.
23 Births, deaths and marriages Victoria. Available: https://www. bdm.
vic. gov. au/ research- and- family- history/ research- and- data- services/
death- statistics [Accessed 05 Oct 2020].
24 Carenzo L, Costantini E, Greco M, etal. Hospital surge capacity in a
tertiary emergency referral centre during the COVID-19 outbreak in
Italy. Anaesthesia 2020;75:928–34.
25 Schulz T, Long K, Kanhutu K. Telehealth during the coronavirus
disease 2019 pandemic: rapid expansion of telehealth outpatient
use during a pandemic is possible if the programme is previously
established. J Telemed Telecare 2020;1357633X:20942045:1–7.
26 World Population Review. Melbourne population 2021https://
worl dpop ulat ionr eview. com/ world- cities/ melbourne- population
[Accessed 05 Feb 2021].
27 Muhi S, Irving LB, Buising KL. COVID-19 in Australian health
care workers: early experience of the Royal Melbourne Hospital
emphasises the importance of community acquisition. Med J Aust
2020;213:44–44.e1.
on June 24, 2021 by guest. Protected by copyright.http://bmjopen.bmj.com/BMJ Open: first published as 10.1136/bmjopen-2020-045975 on 24 June 2021. Downloaded from