select search filters
roundups & rapid reactions
before the headlines
Fiona fox's blog

expert reaction to preprint looking at intensive care occupancy and risk of death in intensive care patients with COVID-19 in the first wave

A preprint, an unpublished non-peer reviewed study, looks at availability of mechanical ventilators and mortality risk in intensive care patients with COVID-19 during the first wave.


Associate Professor James Doidge, Senior Statistician, Intensive Care National Audit and Research Centre (ICNARC), said:

“This study, which has not yet been peer reviewed, is a sophisticated analysis of a large, real-world database providing timely insights on the impact of ICU occupancy on mortality.  However, the database is missing some crucial information required to fairly assess the impact of occupancy and the way that results are reported exaggerates the scale of the impact.

“The study analyses data for patients admitted between 2 April 2020 (the peak of the first wave, when occupancy was high) and 1 June 2020 (the trough between the waves, when occupancy was low).  During the peak period, the patients admitted to ICU were more severely ill than patients admitted during the trough, so mortality would be expected to differ between these periods.  While the CHESS database did not include information on severity of illness (a limitation acknowledged by the authors), a study of the ICNARC Case Mix Programme database that does include this information found that mortality differed between these periods, even after accounting for severity of illness1.

“The estimated impact of occupancy is described as a 19% increase in mortality during periods of high occupancy and a 25% decrease during periods of low occupancy.  These statistics refer to estimated odds ratios of 1.19 and 0.75, respectively.  For common outcomes such as ICU mortality, odds differ from risk.  The absolute risk of death in hospital is only reported for the study sample as a whole (38.4%) and not for the occupancy subgroups.  Using overall mortality as reference, the absolute risk of death would be 42.6% (4.2% higher) in the high-occupancy group and 31.9% (6.5% lower) in the low-occupancy group.

“The supplementary analysis of the impact of occupancy “on the date of recorded outcome” should be interpreted with extreme caution.  Patients discharged from hospital after admission to ICU with COVID-19 remain in hospital on average 2 months longer than patients who die2.  Deaths are therefore much more likely to occur during times of high occupancy and discharges during times of low occupancy, merely because of differing lengths of hospital stay.

“Lastly, another possible explanation for these results is that clinical learning improved patient outcomes as the pandemic progressed and as occupancy reduced.  However, this interpretation is neither supported by the sensitivity analysis provided, in which the authors additionally control for week of admission, nor by the return to higher levels of (unadjusted) mortality seen in the second wave3.  Further research incorporating measures of severity of illness and processes of care is required to understand the impact of clinical learning on patient outcomes during the pandemic.”

1 Doidge JC, Gould DW, Ferrando-Vivas P, et al. Trends in intensive care for patients with COVID-19 in England, Wales and Northern Ireland. American journal of respiratory and critical care medicine. 2020 Dec 11.

2 Richards-Belle, A., Orzechowska, I., Gould, D.W. et al. COVID-19 in critical care: epidemiology of the first epidemic wave across England, Wales and Northern Ireland. Intensive Care Med 46, 2035–2047 (2020).

3 Intensive Care National Audit & Research Centre (ICNARC). ICNARC report on COVID-19 in critical care 8 January 2021.


Prof Duncan Young, Retired Professor of Intensive Care Medicine, University of Oxford, said:

“This paper is yet to be peer reviewed.  I was unable to access the electronic supplements.

“It reports on data collected by Public Health England on ICU admissions for COVID-19 disease in the period 2nd April to 1st June 2020.  This covers the period when ICU admissions rose to a peak and then declined quite markedly.

“The headline result suggests that as occupancy of ICU beds increases, so does the overall percentage mortality of patients admitted during the periods of high occupancy.

“The authors conclude these results “…..highlight a potential major impact of operational pressure on patient survival during the first wave of the COVID-19 pandemic” and “…..highlight a potential major impact of operational pressure on patient survival during the first wave of the COVID-19 pandemic”.  Whilst it may be the case that the occupancy is causing the increased mortality there are at least two other possible interpretations.

“The first is that during periods of high occupancy the ICUs would likely preferentially admit the patients who were sickest, with the highest risk of dying.  If an ICU contains more high-risk patients, it would be expected that the mortality, measured as a percentage of admissions, would increase.  This does not mean there is an increased risk to individual patients in the ICU.  This can be assessed by using severity scoring to determine how sick the patients are who are admitted.  Unfortunately, unlike some other data sources, the CHESS data do not contain information on how sick the patients were at ICU admission.

“Another possibility is that this outcome is due to the reducing ICU mortality seen during this period in other data.  As mortality was reducing over time, and for the majority of the study period ICU occupancy was also falling, there would be an apparent association between occupancy and outcome.”



Preprint (not a paper): ‘The association between mechanical ventilator availability and mortality risk in intensive care patients with COVID-19: A national retrospective cohort study’ by Harrison Wilde et al was posted on medRxiv on Wednesday 13 January 2021.  This work is not peer-reviewed.



All our previous output on this subject can be seen at this weblink:



Declared interests

None received.

in this section

filter RoundUps by year

search by tag