Professional Documents
Culture Documents
3. 4th Department of Internal Medicine, Attikon University Hospital, Medical School, National
Corresponding author:
Email: t.lytras@euc.ac.cy
Abstract
Aims: While healthcare services have been expanding capacity during the COVID-19 pandemic,
quality of care under increasing patient loads has received less attention. We examined in-hospital
mortality of intubated COVID-19 patients in Greece, in relation to total intubated patient load,
Methods: Anonymized surveillance data were analyzed from all intubated COVID-19 patients in
Greece between 1 September 2020 and 6 May 2021. Poisson regression was used to estimate the
Results: Mortality was significantly increased above 400 patients, with an adjusted Hazard Ratio of
1.25, 95%CI: 1.03-1.51), rising progressively up to 1.57 (95%CI: 1.22-2.02) for 800+ patients.
Hospitalization outside an ICU or away from the capital region of Attica were also independently
Conclusions: Our results indicate that in-hospital mortality of severely ill COVID-19 patients is
adversely affected by high patient load even without exceeding capacity, as well as by regional
disparities. This highlights the need for more substantial strengthening of healthcare services, focusing
Keywords (MeSH)
COVID-19; pandemic; healthcare disparities; intensive care units; right to health; quality of care;
intubation; mortality
Word count
During the COVID-19 pandemic, healthcare systems’ capacity and resilience has been in the spotlight
[1]. However, less attention has focused on how the actual stress placed on healthcare services by
COVID-19 has affected their performance [2, 3]. Even without exceeding nominal capacity, a high
patient load could plausibly compromise quality of care, especially if occurring in a context of chronic
In this context, the aim of our study was to examine how in-hospital mortality of intubated COVID-19
patients in Greece is affected by patient load, intensive care unit (ICU) availability and regional
disparities.
Methods
In Greece the National Public Health Organization (NPHO) is responsible for COVID-19 surveillance,
and does active case finding and follow-up of all laboratory-confirmed COVID-19 cases in the country
that are intubated or hospitalized in an ICU. We obtained from NPHO anonymized patient data for all
intubated cases from 1 September 2020 to 6 May 2021, including dates of intubation, extubation, ICU
admission and discharge. As the study used only anonymous surveillance data from which no patient
can be identified, no ethical approval was necessary. The analyzed data and analysis code are available
For all cases follow-up time from intubation until extubation or death was split finely into days, and
Poisson regression was used to estimate the hazard of dying (fatality rate per day of hospitalization) as
a function of fixed and time-varying covariates [6]. To avoid bias, deaths occurring up to five days
after extubation were classified as deaths at the end of follow-up. The current total of intubated
COVID-19 patients at each time point during follow-up was used as an indicator of healthcare system
stress, and was included as a time-varying categorical variable in the model (reference: 0-199 patients,
vs 200-299, 300-399 etc up to 800+ patients). Age was modelled with a natural cubic spline with 1
internal knot. Other covariates included sex, a linear time trend, ICU hospitalization (vs non-ICU, for
each day of follow-up) and hospital region (the metropolitan regions of Attica and Thessaloniki, vs the
rest of Greece, Supplementary Figure S1). Model-based effect estimates (Hazard Ratios) were also
used to calculate Population Attributable Fractions [7]. Furthermore, we examined how the age
distribution of patients varied in relation to patient load, region and ICU hospitalization, in order to
better interpret the findings. All analyses were done in the R software environment version 4.0.4 [8].
Results
During the study period, two epidemic waves were observed in Greece; the first peaked in November
2020 and was focused in Central and Northern Greece, whereas the second was associated with the
B.1.1.7 variant and affected Athens and the entire country. The series of new intubations, deaths
among the study population, and total intubated COVID-19 patients are illustrated in Figure 1.
After excluding 296 cases with incomplete covariate information, 6282 cases were analyzed (Table 1)
of whom 3988 died (63.5%). Most patients spent part or all of their hospital stay in an ICU
(5971/6282, with 93951/97601 person-days total). Among those not admitted to an ICU, 275 (88.4%)
Model results, expressed as adjusted Hazard Ratios (HR) and 95% Confidence Intervals (CI), are
shown in Figure 2. There was a significant association between mortality and total intubated patients
above 400, with its magnitude increasing progressively: from 1.25 (95%CI: 1.03–1.51) for 400-499
patients, up to 1.57 (95%CI: 1.22–2.02) for 800+ patients. Being intubated outside an ICU was
strongly associated with mortality (HR 1.87, 95%CI: 1.65–2.12), as was age, particularly after 60
years old. Interestingly, being hospitalized outside the capital region of Attica was also associated with
increased in-hospital mortality, with HRs of 1.35 (95%CI: 1.24–1.47) for Thessaloniki and 1.40
(95%CI: 1.30–1.51) for the rest of the country. There was no association with sex, but there was a
small negative time trend (HR 0.97 per month, 95%CI 0.94–1.00, p=0.02) indicating gradually lower
Given the above associations, out of 3988 deaths reported, 947 (95%CI: 343–1460) were attributable
to the high load (≥200) of intubated COVID-19 patients, 133 (95%CI: 101–169) to being outside an
ICU, and 656 (95%CI: 526–790) to being hospitalized away from Attica. A combined total of 1535
There was no association between patient age and total patients intubated, offering no evidence of
patient selection with higher patient loads (Supplementary Figure S2). In contrast, patients who never
entered an ICU were older on average (median 73 vs 68 years, p<0.001). Also, patients hospitalized in
Thessaloniki were slightly younger than those in Attica and other regions (median 67 vs 69 years,
p<0.001).
Discussion
Our analysis provides national-level evidence that in-hospital mortality of severely ill COVID-19
patients is adversely affected by high patient load. Importantly, this occurs not only when healthcare
capacity is stressed to depletion, but also at intermediate stress levels, when the availability of
resources and care is not nominally restricted. This represents a major preventable factor to limit
avoidable deaths from COVID-19, and highlights the need for more extensive investment in healthcare
beyond the minimum to meet peak demand during the pandemic, thereby ensuring adequate quality of
care. Similar findings for COVID-19 patients have been recently reported from Veterans Affairs
hospitals in the United States [2], and a large referral hospital in Northern Italy [3].
Our study further demonstrated important regional disparities, as in-hospital mortality was
substantially lower for people hospitalized in Attica compared to the rest of the country. This
highlights the chronically uneven regional distribution of healthcare resources in Greece, with beds,
equipment and trained healthcare workers concentrated in metropolitan areas [9] (Supplementary
Figure S1). It also compounds other existing health inequities that define the rural-urban divide,
resulting in a higher COVID-19 case fatality ratio in underserved and rural areas despite higher
Our results must be interpreted with caution; as this is an observational study, there is a possibility that
patient selection (prioritizing for care those most in need and with a worse prognosis) might have
influenced our associations. However, the lack of association between total patient load and the age of
newly intubated patients suggests that those needing care did indeed receive it. In addition, patients in
Thessaloniki were younger than those in Attica, thus residual confounding by age cannot explain their
increased mortality. In contrast, patients treated outside an ICU were slightly older, thus some of them
might have been deemed “too ill to benefit” from the ICU. A clear limitation of our analysis is the lack
of information on comorbidities and baseline health status (at presentation to the hospital). However,
both these factors are unlikely to be distributed differently over time, and are both correlated with and
eclipsed by age, which is the major determinant of the risk of death and which we carefully adjusted
for [12, 13]. Therefore the observed associations most likely reflect real and avoidable differences in
the quality of care for COVID-19 patients, due to increased patient load, ICU availability and regional
disparities. The precise mechanisms by which patient load is associated with increased mortality need
Nevertheless, these findings already highlight the need for strengthening healthcare systems in the
such as the availability of trained healthcare staff. In Greece the increased healthcare worker needs due
to COVID-19 were largely met with staff redeployments, short-term hirings, and requisitioning of
private practitioners’ services [14, 15]. These might have been insufficient to counter the chronic
underfunding and understaffing of public healthcare services as a consequence of the long economic
crisis [16]. It is therefore essential to make long-term investments in health also for the post-COVID
We acknowledge the staff of the National Public Health Organization for their hard work in
Author contributions
Original idea: ST; Data analysis: TL; Data interpretation: TL, ST; First draft of the manuscript: TL;
Conflicts of Interest
None. ST is an advisor to the Greek Ministry of Health; views expressed here are his own.
Funding
[1] Sagan A, Thomas S, McKee M, et al. COVID-19 and health systems resilience: lessons going
forwards. Eurohealth; 26.
[2] Bravata DM, Perkins AJ, Myers LJ, et al. Association of Intensive Care Unit Patient Load and
Demand With Mortality Rates in US Department of Veterans Affairs Hospitals During the
COVID-19 Pandemic. JAMA Netw Open 2021; 4: e2034266.
[3] Soria A, Galimberti S, Lapadula G, et al. The high volume of patients admitted during the
SARS-CoV-2 pandemic has an independent harmful impact on in-hospital mortality from
COVID-19. PLoS One 2021; 16: e0246170.
[4] Oppenheim B, Gallivan M, Madhav NK, et al. Assessing global preparedness for the next
pandemic: development and application of an Epidemic Preparedness Index. BMJ Glob Health
2019; 4: e001157.
[5] Gabler NB, Ratcliffe SJ, Wagner J, et al. Mortality among patients admitted to strained
intensive care units. Am J Respir Crit Care Med 2013; 188: 800–806.
[7] Greenland S, Drescher K. Maximum likelihood estimation of the attributable fraction from
logistic models. Biometrics 1993; 49: 865–872.
[8] R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R
Foundation for Statistical Computing, http://www.R-project.org/ (2019).
[9] Economou C, Kaitelidou D, Karanikolos M, et al. Greece: health system review. Health
Systems in Transition 2017; 19: 1–192.
[10] Iyanda AE, Boakye KA, Lu Y, et al. Racial/Ethnic Heterogeneity and Rural-Urban Disparity of
COVID-19 Case Fatality Ratio in the USA: a Negative Binomial and GIS-Based Analysis. J
Racial Ethn Health Disparities. Epub ahead of print February 26, 2021. DOI: 10.1007/s40615-
021-01006-7.
[11] Karim SA, Chen H-F. Deaths From COVID-19 in Rural, Micropolitan, and Metropolitan Areas:
A County-Level Comparison. J Rural Health 2021; 37: 124–132.
[12] Petrilli CM, Jones SA, Yang J, et al. Factors associated with hospital admission and critical
illness among 5279 people with coronavirus disease 2019 in New York City: prospective cohort
study. BMJ 2020; 369: m1966.
[13] Williamson EJ, Walker AJ, Bhaskaran K, et al. Factors associated with COVID-19-related
death using OpenSAFELY. Nature 2020; 584: 430–436.
[14] European Observatory on Health Systems and Policies. COVID-19 Health System Response
Monitor: Greece, https://www.covid19healthsystem.org/countries/greece/countrypage.aspx
(accessed May 9, 2021).
[17] Paremoer L, Nandi S, Serag H, et al. Covid-19 pandemic and the social determinants of health.
BMJ 2021; 372: n129.
Table 1: Characteristics of the study population; laboratory-confirmed COVID-19 cases in
Figure 1: Distribution over time of total intubated COVID-19 patients, new intubations and deaths
patients and age, sex, hospitalization type, hospital region and patient load, Greece, 1 September 2020
to 6 May 2021.
Supplementary Figure 1: Study regions in Greece and distribution of ICU beds, as of 20 April 2021
Supplementary Figure 2: Patient age distribution in relation to patient load, hospitalization type
(inside/outside ICU) and hospital region, Greece, 1 September 2020 to 6 May 2021.
Total intubated
800
New intubations
80
New deaths
70
60
Total patients intubated
50
400
40
30
200
20
10
0
0
Sep 01 Oct 01 Nov 01 Dec 01 Jan 01 Feb 01 Mar 01 Apr 01 May 01
Date
Total intubated patients (vs: 0–199) HR (95% CI)
200-299 1.16 (0.94–1.43)
300-399 1.14 (0.93–1.39)
400-499 1.25 (1.03–1.51)
500-599 1.35 (1.13–1.62)
600-699 1.32 (1.09–1.61)
700-799 1.39 (1.10–1.77)
800+ 1.57 (1.22–2.02)
Patient sex (vs: female)
Male 0.99 (0.92–1.05)
Hospitalization type (vs: in ICU)
Outside an ICU 1.87 (1.65–2.12)
Hospital region (vs: Attica)
Thessaloniki 1.35 (1.24–1.47)
Rest of Greece 1.40 (1.30–1.51)
95% CI
6
Adj. Hazard Ratio
5
4
3
2
1
0 20 40 60 80 100
Age (years)
Kruskal-Wallis p=0.2
100
80
60
Age
40
20
0
100
80
80
60
60
Age
Age
40
40
20
20
0