Can we learn from others? The Swedish intensive care unit database for COVID-19 cases.

Abstract

This Invited Commentary accompanies the following original article: Chew MS, Blixt PJ, Åhman R, et al. National outcomes and characteristics of patients admitted to Swedish intensive care units for COVID19: A registry-based cohort study. Eur J Anaesthesiol 2021; 38:335–343. The first cases of patients infected with severe acute respiratory syndrome-coronavirus-2 (SARS-CoV-2) admitted to the intensive care unit (ICU) for severe COVID-19 pneumonia were hastily published last spring. The large diffusion of the information was understandable as an initial warning to prepare clinicians around the world for what they would soon face. However, the quality of data was somewhat questionable. Most of these publications were retrospective single-centre analyses, including limited numbers of cases, frequently combining mixed populations of ICU and non-ICU patients. Except for demographic data and underlying diseases, the clinical, severity and therapeutic features were frequently poorly described, with numerous missing data. Due to the eagerness to communicate, many of these reports gave only a short overview of the outcome, with some studies reporting that almost half of their cohort was still in the ICU at the time of publication.1 It is now time to obtain more detailed analyses through structured approaches coordinated at a regional or national level. The paper from Chew et al. reports findings from a national Swedish cohort of ICU patients treated for COVID-19 collected during the first wave of the SARS-CoV-2 pandemic.2 Studies of large ICU cohorts treated for COVID-19 are still scarce and the Swedish experience is highly valuable. We are grateful to the authors, who analysed the ICU and 30-day mortality rates, tried to identify the risk factors for death and performed an exhaustive follow-up to better describe the epidemic in a large population of more than 1500 ICU patients throughout the country. From the perspective of possible subsequent waves of infection, the Swedish experience is of major relevance to evaluate the effects of political strategies. However, several points which suggest a cautious interpretation of the data should be considered. The current report provides only a snapshot of ICU management during the first weeks of the outbreak, whereas the first wave was prolonged until mid-July. During other emerging viral disease outbreaks reported recently across the world, such as the Ebola, Zika or Middle East respiratory syndrome, the early mortality rates observed in epidemiologic studies were frequently higher at the beginning of the outbreak. It can be attributed to several factors: the evolving knowledge over time on the diagnosis and treatment of the disease; a progressively improving organisation of the healthcare system; and a better information provision to the public, who come to the hospital sooner when they are aware of the clinical signs. This issue can be illustrated by the Swedish mortality rate of 483 per million by June 2020, which is in the low/intermediate average range for the European region (https://coronavirus.jhu.edu/map.html; accessed 2 December 2020). Chew et al. stress the low number of ICU beds per head of the Swedish population compared with other European countries. This remark is of importance, suggesting that a large number of ICU beds are not the solution to all problems. Explanations for the encouraging results observed in this report have to be found elsewhere. The high quality of medical resources in Sweden could be a valuable clue. However, a cautious interpretation/extrapolation of these results is required even for comparisons between countries with similar levels of gross domestic product and numbers of healthcare workers and/or medical equipment. As such, one cannot analyse an ICU epidemiologic study during the COVID-19 pandemic without questioning the ethical considerations that many physicians faced during that period including the limitations on ICU admission. Most countries with overwhelmed health capacities had to set up nonofficial boundaries to try to restrict ICU bed availability to patients with a high expected survival rate. In the Swedish cohort, although the global comorbidity rate was similar to that in previous epidemiologic reports, when analysed individually, the prevalence of some major comorbidities (such as diabetes or arterial hypertension) were lower. It is challenging to evaluate whether this finding reflects peculiarities of the Swedish population or ethical consequences of admission restrictions to ICU. Thus, the extrapolation of these results to another country is somewhat questionable. The data from a prospective cohort of French ICU patients, containing more than 4000 patients, have been recently published.3 Interestingly, the cohort of French and Swedish patients from these two studies share some characteristics, such as age, sex, the incidence of moderate to severe adult respiratory distress syndrome and mechanical ventilation requirements. However, although the Swedish cohort had a lower PaO2/FiO2 ratio at ICU admission despite a higher sepsis-related organ failure assessment score, adjunct measures were slightly different for prone positioning (40% versus 70%) and extracorporeal life support (>1% versus 11%) in the Swedish and French studies, respectively. These discrepancies highlight different strategies in different countries that should be taken into account in the interpretation of results. One of the striking points of the Swedish study is the lack of association between mortality and comorbidities. All studies conducted in countries with high gross domestic product have established this relationship. Indeed, arterial hypertension,1 chronic cardiac disease,4 chronic pulmonary disease,4–7 high body mass index,8 coronary artery disease,8 active cancer,7,8 immunocompromised status,6 type 2 diabetes5 and hypercholesterolemia5 were found to be associated with the severity of adult respiratory distress syndrome and/or mortality. This lack of correlation emphasises a possible confounding effect of patient selection through ethical considerations or a ‘country’ effect (either through healthcare system organisation with a high healthcare worker-to-patient ratio or through a genetic background). Regardless, the extrapolation of these results to other countries might be limited. Finally, the question of the therapeutic improvements ahead of the next SARS-CoV-2 waves remains unsolved by the snapshot taken by Chew et al. No specific therapy has yet emerged9 and the recent advances improving the prognosis for critically ill patients are based only on the use of anticoagulants and steroids.10 These first national reports could be the starting point of broader comparisons. In this setting, international meta-analyses would be of major interest to evaluate the national or ethnic peculiarities with a potential impact on the outcome. Genetic characterisation of the populations could be another approach for explaining discrepancies in the high variability in the incidence and prognosis of the disease, as reported in Latin America and sub-Saharan Africa.