Discussion
In this study of patients aged 70 and above admitted with COVID-19 disease, we found a decrease in thirty-day survival from 57% in the first surge to 49% in the second surge, even after adjustment for important co-factors such as age, gender, SOFA score, comorbidities and frailty. The major differences between the two groups besides mortality were the reduction in the use of intubation and mechanical ventilation and its early use, reduced use of vasoactive drugs, increased use of non-invasive ventilation (NIV) and an increased use of corticosteroids during the second surge. Although management of patients changed, we cannot clearly attribute the change in outcome to a specific change in practice.
These findings are surprising, as the ICU community had gained experience in treating these patients during the first surge. Here, the initial very high reported [
16] mortality was soon followed with a reduction in mortality towards the end of the first surge [
17]. It was thought that in the second surge, the use of steroids in patients with severe respiratory distress, and the delay in intubation, following the use of NIV to its full potential would translate to better outcomes. However, our detailed analysis revealed that besides the treatment during the second surge, older age, male sex, increased frailty, increased SOFA score and chronic kidney disease were associated with poor outcome especially 15 days after ICU admission.
To date, there are only very few reports comparing the two surges in COVID-19 hospitalised patients. In a study from 955 US hospitals, researchers compared a trend analysis for the first surge: from 1 January to 30 April 2020 and 1 May to 30 June 2020. The overall hospital event rate for 30 day mortality or referral to hospice within 30 days fell from 16.56 to 9.29%, indicating improved outcome in the last period of the first surge suggesting a steep learning curve [
18]. This has also been confirmed by a study from the UK in > 21,000 critical care patients with COVID-19 showing improved survival from March to June 2020 [
19]. In another study looking at COVID-19 outcomes in hospitalised patients with rheumatic disease, outcomes from the first 90 days were compared to the following 90 days [
20]. They found a reduced risk of hospitalisation and admission to an ICU in the late cohort, and also a fall in the risk of death (9.3% versus 4.5%).
Our patients were all treated in the ICU during both surges, and therefore, a comparison with these initial experiences from other patient groups is difficult. However, reports from Intensive Care National Audit and Research Centre (ICNARC) in the UK reveal valuable information in this respect. This national registry compared patients admitted before and after 1 September, 2020, and found a small increase in mortality in patients ventilated within the first 24 h from 46.6% in the first to 48.7% in the second cohort and a similar reduction in all patients discharged alive from the ICU [
21]. Their results were also similar to our patient cohort in other respects, with considerably fewer patients receiving mechanical ventilation, (down from 72 to 50%) and more patients given basic respiratory support (from 25 to 47%). In addition, the duration of mechanical ventilation was shorter. Analogous to our results, they found that more patients had a low PaO
2/FiO
2 ratio at admission.
There are several possible reasons for the increased mortality seen in our study although we can only describe associations in this kind of study setting. While our data do not give a satisfactory explanation, it allows for several potential contributing factors to be discussed and to guide specific attention to differences occurring during the disease course, such as the increasing relevance of kidney failure and potential gender differences after day 15.
A worse outcome might have been caused by the increased length of time spent in other departments before ICU admission, resulting in patients deteriorating prior to eventual admission. This is supported by a decreased PaO
2/FiO
2 ratio seen at ICU admission in the second period, possibly suggesting more severe respiratory failure. This combined with a trend towards a reduction in the use of mechanical ventilation may not have been beneficial in this group of elderly critically ill patients, although this remains speculative. Although several studies and a meta-analysis suggested that timing of intubation may have no effect on mortality and morbidity in COVID-19 [
22], this remains to be confirmed in elderly patients. Also failure of non-invasive ventilation with delayed intubation needs to better defined, with especially high mortality rates [
23].
A similar rate of limitation of life-sustaining therapy was seen in the two cohorts, so this is unlikely to account for the difference in mortality. Two additional differences are a slight increase in age and frailty score in the second cohort, which could explain an increase in mortality; however, the difference in mortality remained even after adjustment for these factors.
Another possible explanation for the increased mortality, which cannot be ruled out, could be a reduction in quality of care, despite all dedicated efforts on the part of the staff, in the second compared to the first surge. When the second surge started, many hospitals and in particular ICUs had already been overstretched for half a year and were running well above their usual capacity. This had consequences for both the permanent staff who had been working increased hours over a long period of time, and also the continuous dilution of expertise as non-ICU personnel, both physicians and nurses were being brought in to work in ICUs. There has been great concern about the burden of work on the health of ICU workers [
24], leading to fatigue and physical and mental health problems, which ultimately may affect quality of care. In the current survival analysis, survival differs from day 15 onwards and it is tempting to speculate that quality of care in particular had consequences for elderly patients with prolonged treatment duration.
Another important factor may be related to use of corticosteroids. In the second surge, 93% of our patients received corticosteroids, which is more than twice that found in the first cohort. It is well documented that steroids have potentially serious side effects in ICU patients. Steroids increase infection rate and hence mortality in patients admitted with influenza pneumonia [
25] and thus could also increase the number of patients acquiring sepsis in the ICU. It was of interest to study the details in the supplementary appendix from the RECOVERY study where COVID-19 patients were randomised to receive corticosteroids [
7]. Only a small number of patients requiring mechanical ventilation were over 70 years old. In a pre-specified analysis of the RECOVERY trial, there was no difference in mortality in patients above 70 years. Despite this, that landmark trial—among others—changed guidelines [
26] and practice independent of age, in severely ill COVID-19 patients. Although the RECOVERY trial is a major achievement in these difficult times, some unanswered questions remained. For example, it was unclear whether patients with uncontrolled diabetes, acute delirium, underlying malignancy, immunosuppression, or other conditions in which corticosteroids might have harmful effects were included [
8,
27].
Finally, development of COVID-19 mutations may change virulence and hence potentially lead to worse outcomes. It is well known that during the pandemic a new mutant virus emerged in Europe [
28]. The clinical properties of this new strain are largely unknown as whole genome sequence studies have not been performed in large scale and ordinary COVID-19 testing does not differentiate mutant viruses from the original one. Such a cause for worse outcomes is uncertain but remains a theoretical possibility.
Another possible explanation could be selection bias creating differences between the cohort in the first and second surge. That patients are already selected before ICU admission is common, and many undergo formal and informal triage. Criteria for triage were extensively discussed from the beginning of the pandemic when some ICUs experienced a rapid overflow of patients. We do not have detailed insight in what happened before ICU admission in our study, as this was not a research question. There are, however, some casemix differences between the first and second surge, mainly connected to longer time in hospital before ICU admission during the second surge, and a decreased oxygen ratio at ICU admission. This indicates potential differences in initial treatment leading to changes in selection maybe connected with increased knowledge of the feasibility to treat some severe covid-19 patients outside the ICU [
29].
Our study has further weaknesses mainly concerning the absence of details of variables that might account for our differences in outcome. There was no information about how steroids were administered, no control group of younger COVID-19 patients for comparison and there was no information about quality of care and the nurse-to-patient ratios as well as a measure of stress for personnel. Also, admission policies or local guideline changes were not recorded. The recording of treatment limitations is not without difficulties but important in ICU studies in elderly patients [
30]. The reception of this might differ across study sites reflecting a wide heterogenicity. In addition, although we provided definition of the comorbidities, their characterisation do not provide in-detail characterisation and differ across studies in the literature. Also, data on anticoagulation, sedation practices and on lung-protective ventilation were not collected which might account for outcome differences. Another limitation is that we did not ask centres to monitor consecutive inclusion with a screening log, serving as proofs of consecutive recruitment and allowing generalizability. We have no proofs that all eligible patients in all centres have been included into the study. Our observations can only describe associations without ascribing causality; however, we have observed that an untargeted but consistent change in practice has changed the outcomes between the cohorts in the two surges.
Acknowledgements
The authors want to thank all investigators and study personal for their great support of the study. The COVIP study group consists of the authors and the following persons: Philipp Eller, Michael Joannidis, Dieter Mesotten, Pascal Reper, Sandra Oeyen, Walter Swinnen, Helene Brix, Jens Brushoej, Maja Villefrance, Helene Korvenius Nedergaard, Anders Thais Bjerregaard, Ida Riise Balleby, Kasper Andersen, Maria Aagaard Hansen, Stine Uhrenholt, Helle Bundgaard, Jesper Fjølner, Aliae AR Mohamed Hussein, Rehab Salah, Yasmin Khairy NasrEldin Mohamed Ali, Kyrillos Wassim, Yumna A. Elgazzar, Samar Tharwat, Ahmed Y. Azzam, Ayman abdelmawgoad Habib, Hazem Maarouf Abosheaishaa, Mohammed A Azab, Susannah Leaver, Arnaud Galbois, Bertrand Guidet, Cyril Charron, Emmanuel Guerot, Guillaume Besch, Jean-Philippe Rigaud, Julien Maizel, Michel Djibré, Philippe Burtin, Pierre Garcon, Saad Nseir, Xavier Valette, Nica Alexandru, Nathalie Marin, Marie Vaissiere, Gaëtan Plantefeve, Thierry Vanderlinden, Igor Jurcisin, Buno Megarbane, Anais Caillard, Arnaud Valent, Marc Garnier, Sebastien Besset, Johanna Oziel, Jean-herlé Raphaelen, Stéphane Dauger, Guillaume Dumas, Bruno Goncalves, Gaël Piton, Christian Jung, Raphael Romano Bruno, Malte Kelm, Georg Wolff, Eberhard Barth, Ulrich Goebel, Eberhard Barth, Anselm Kunstein, Michael Schuster, Martin Welte, Matthias Lutz, Patrick Meybohm, Stephan Steiner, Tudor Poerner, Hendrik Haake, Stefan Schaller, Detlef Kindgen-Milles, Christian Meyer, Muhammed Kurt, Karl Friedrich Kuhn, Winfried Randerath, Jakob Wollborn, Zouhir Dindane, Hans-Joachim Kabitz, Ingo Voigt, Gonxhe Shala, Andreas Faltlhauser, Nikoletta Rovina, Zoi Aidoni, Evangelia Chrisanthopoulou, Antonios Papadogoulas, Mohan Gurjar, Ata Mahmoodpoor, Abdullah khudhur Ahmed, Brian Marsh, Ahmed Elsaka, Sigal Sviri, Vittoria Comellini, Ahmed Rabha, Hazem Ahmed, Silvio a Namendys-Silva, Abdelilah Ghannam, Martijn Groenendijk, Marieke Zegers, Dylan de Lange, Alex Cornet, Mirjam Evers, Lenneke Haas, Tom Dormans, Willem Dieperink, Luis Romundstad, Britt Sjøbø, Finn H. Andersen, Hans Frank Strietzel, Theresa Olasveengen, Michael Hahn, Miroslaw Czuczwar, Ryszard Gawda, Jakub Klimkiewicz, Maria de Lurdes Campos Santos, André Gordinho, Henrique Santos, Rui Assis, Ana Isabel Pinho Oliveira, Mohamed Raafat Badawy, David Perez-Torres, Gemma Gomà, Mercedes Ibarz Villamayor, Angela Prado Mira, Patricia Jimeno Cubero, Susana Arias Rivera, Teresa Tomasa, David Iglesias, Eric Mayor Vázquez, Cesar Aldecoa, Aida Fernández Ferreira, Begoña Zalba-Etayo, Isabel Canas-Perez, Luis Tamayo-Lomas, Cristina Diaz-Rodriguez, Susana Sancho, Jesús Priego, Enas M.Y. Abualqumboz, Momin Majed Yousuf Hilles, Mahmoud Saleh, Nawfel Ben-HAmouda, Andrea Roberti, Alexander Dullenkopf, Yvan Fleury, Bernardo Bollen Pinto, Joerg C. Schefold, Mohammed Al-Sadawi, Nicolas Serck, Elisabeth Dewaele, Pritpal Kumar, Camilla Bundesen, Richard Innes, James Gooch, Lenka Cagova, Elizabeth Potter, Michael Reay, Miriam Davey, Sally Humphreys, Caroline Hauw Berlemont, Benjamin Glenn Chousterman, François Dépret, Alexis Ferre, Lucie Vettoretti, Didier Thevenin, Andreas Faltlhauser, Milena Milovanovic, Philipp Simon, Marco Lorenz, Sandra Emily Stoll, Simon Dubler, Kristina Fuest, Francesk Mulita, Eumorifa Kondili, Ioannis Andrianopoulos, Iwan Meynaar, Alexander Daniel Cornet, Britt Sjøbøe, Anna Kluzik, Paweł Zatorski, Tomasz Drygalski, Wojciech Szczeklik, Joanna Solek-pastuszka, Dariusz Onichimowski, Jan Stefaniak, Karina Stefanska-Wronka, Ewa Zabul, Filipe Sousa Cardoso, Maria José Arche Banzo, Teresa Maria Tomasa-Irriguible, Ángela Prado Mira, Susana Arias-Rivera, Fernando Frutos-Vivar, Sonia Lopez-Cuenca, Pablo Ruiz de Gopegui, Nour Abidi, Ivan Chau, Richard Pugh, Sara Smuts.
Philipp Eller, Michael Joannidis, Dieter Mesotten, Pascal Reper, Sandra Oeyen, Walter Swinnen, Helene Brix, Jens Brushoej, Maja Villefrance, Helene Korvenius Nedergaard, Anders Thais Bjerregaard, Ida Riise Balleby, Kasper Andersen, Maria Aagaard Hansen, Stine Uhrenholt, Helle Bundgaard, Jesper Fjølner, Aliae AR Mohamed Hussein, Rehab Salah, Yasmin Khairy NasrEldin Mohamed Ali, Kyrillos Wassim, Yumna A. Elgazzar, Samar Tharwat, Ahmed Y. Azzam, Ayman abdelmawgoad Habib, Hazem Maarouf Abosheaishaa, Mohammed A Azab, Susannah Leaver, Arnaud Galbois, Bertrand Guidet, Cyril Charron, Emmanuel Guerot, Guillaume Besch, Jean-Philippe Rigaud, Julien Maizel, Michel Djibré, Philippe Burtin, Pierre Garcon, Saad Nseir, Xavier Valette, Nica Alexandru, Nathalie Marin, Marie Vaissiere, Gaëtan Plantefeve, Thierry Vanderlinden, Igor Jurcisin, Buno Megarbane, Anais Caillard, Arnaud Valent, Marc Garnier, Sebastien Besset, Johanna Oziel, Jean-herlé Raphaelen, Stéphane Dauger, Guillaume Dumas, Bruno Goncalves, Gaël Piton, Christian Jung, Raphael Romano Bruno, Malte Kelm, Georg Wolff, Eberhard Barth, Ulrich Goebel, Eberhard Barth, Anselm Kunstein, Michael Schuster, Martin Welte, Matthias Lutz, Patrick Meybohm, Stephan Steiner, Tudor Poerner, Hendrik Haake, Stefan Schaller, Detlef Kindgen-Milles, Christian Meyer, Muhammed Kurt, Karl Friedrich Kuhn, Winfried Randerath, Jakob Wollborn, Zouhir Dindane, Hans-Joachim Kabitz, Ingo Voigt, Gonxhe Shala, Andreas Faltlhauser, Nikoletta Rovina, Zoi Aidoni, Evangelia Chrisanthopoulou, Antonios Papadogoulas, Mohan Gurjar, Ata Mahmoodpoor, Abdullah Khudhur Ahmed, Brian Marsh, Ahmed Elsaka, Vittoria Comellini, Ahmed Rabha, Hazem Ahmed, Abdelilah Ghannam, Martijn Groenendijk, Marieke Zegers, Dylan de Lange, Alex Cornet, Mirjam Evers, Lenneke Haas, Tom Dormans, Willem Dieperink, Luis Romundstad, Britt Sjøbø, Finn H. Andersen, Hans Frank Strietzel, Theresa Olasveengen, Michael Hahn, Miroslaw Czuczwar, Ryszard Gawda, Jakub Klimkiewicz, Maria de Lurdes Campos Santos, André Gordinho, Henrique Santos, Rui Assis, Ana Isabel Pinho Oliveira, Mohamed Raafat Badawy, David Perez-Torres, Gemma Gomà, Mercedes Ibarz Villamayor, Angela Prado Mira, Patricia Jimeno Cubero, Susana Arias Rivera, Teresa Tomasa, David Iglesias, Eric Mayor Vázquez, Cesar Aldecoa, Aida Fernández Ferreira, Begoña Zalba-Etayo, Isabel Canas-Perez, Luis Tamayo-Lomas, Cristina Diaz-Rodriguez, Susana Sancho, Jesús Priego, Enas M.Y. Abualqumboz, Momin Majed Yousuf Hilles, Mahmoud Saleh, Nawfel Ben-HAmouda, Andrea Roberti, Alexander Dullenkopf, Yvan Fleury, Bernardo Bollen Pinto, Joerg C. Schefold, Mohammed Al-Sadawi, Nicolas Serck, Elisabeth Dewaele, Pritpal Kumar, Camilla Bundesen, Richard Innes, James Gooch, Lenka Cagova, Elizabeth Potter, Michael Reay, Miriam Davey, Sally Humphreys, Caroline Hauw Berlemont, Benjamin Glenn Chousterman, François Dépret, Alexis Ferre, Lucie Vettoretti, Didier Thevenin, Andreas Faltlhauser, Milena Milovanovic, Philipp Simon, Marco Lorenz, Sandra Emily Stoll, Simon Dubler, Kristina Fuest, Francesk Mulita, Eumorifa Kondili, Ioannis Andrianopoulos, Iwan Meynaar, Alexander Daniel Cornet, Britt Sjøbøe, Anna Kluzik, Paweł Zatorski, Tomasz Drygalski, Wojciech Szczeklik, Joanna Solek-pastuszka, Dariusz Onichimowski, Jan Stefaniak, Karina Stefanska-Wronka, Ewa Zabul, Filipe Sousa Cardoso, Maria José Arche Banzo, Teresa Maria Tomasa-Irriguible, Ángela Prado Mira, Susana Arias-Rivera, Fernando Frutos-Vivar, Sonia Lopez-Cuenca, Pablo Ruiz de Gopegui, Nour Abidi, Ivan Chau, Richard Pugh, Sara Smuts.