The pandemic caused by SARS-CoV-2 infection has resulted in the greatest healthcare crisis seen in recent times. The enormous number of patients with coronavirus disease 2019 (COVID-19) attended by the healthcare systems of those countries hardest hit by the pandemic caused these systems to become overwhelmed from the early stages. This made it necessary to implement organizational and care strategies at all levels: primary care, out- and in-hospital emergency care, conventional hospitalization, admission to the Intensive Care Unit (ICU), and hospitalization in unconventional alternative centers or units of different kinds.1–5 Another fundamental aspect of the healthcare systems that showed important weakening during the pandemic is referred to the provision of adequate equipment, devices and facilities for effective patient management, protection of the healthcare professionals, and prevention of the spread of the infection within the healthcare system itself – these being circumstances already seen in the past in other highly transmissible infectious processes such as influenza.6–8

In this scenario, one of the main problems in the context of an overwhelmed healthcare system was the assignment of ICU beds for those patients that suffered worsening of their clinical condition. Although during the pandemic up to 300% more critical care beds were habilitated in Spain in hospitals,9 this did not prevent the ICUs from also becoming overwhelmed in certain cities or regions (Autonomous Communities) of the country during a number of days in the course of the pandemic. Different societies protocolized a series of principles for the assignment of hospitalization beds in the ICU that were embodied in recommendations characterized by important consensus and in most cases with common positionings.10–12 However, debate has arisen in this scenario in both the population and in political circles regarding potential inadequacies in the decision to admit patients to the ICU13,14 and potential medical responsibilities in the case of patients with COVID-19 who died without having been admitted to the ICU. The present study explores the basal characteristics of a large and geographically diverse sample of COVID-19 patients that died during the peak of the pandemic, establishing comparisons between those with prior admission to the ICU and those who died without having been admitted to the ICU. The main objective was to determine whether application was made of the standard criteria for patient admission to the ICU, which are fundamentally based on patient age and prior comorbidity.15,16

The management of serious acute disease when the patient suffers gradual organ and system failure, using the most aggressive therapeutic measures available in medical practice, is carried out by specialists in intensive care medicine in the setting of the ICU. However, such therapies are not generally applicable to all patients or in all contexts. In this regard, a series of principles must be universally applied with the aim of ensuring that these treatments are adequately adjusted to each individualized patient – since their indiscriminate application can prove futile and is often associated to therapeutic excess. Adequate application is largely fundamented upon due consideration of the age and comorbidities of the patient – these being essential aspects, since they have a direct impact upon the clinical severity of the acute process causing decompensation, the concomitant involvement of other organs, and the potential reversibility of the clinical condition. The present study shows that, in the same way as in routine clinical practice,15,16 the age and global comorbidity burden of the patient (particularly dementia and active cancer) are independent factors associated to death without admission to the ICU. In fact, only 11% of those who died without admission to the ICU were under 70 years of age, and of these, the great majority had important comorbidity. Furthermore, 80.9% of the patients that died without admission to the ICU were hospitalized for at least 48h (95.1% for at least 24h) – this being long enough for consensus-based decision making among the patients, relatives, attending physicians and specialists in intensive care. Thus, it should be considered that in the great majority of cases there was sufficient time for a pondered and consensus-based decision.

With regard to age, it is clear that aging of the population has significant ethical implications for the management of elderly individuals in the ICU. In our study, most of the elderly patients in turn had important morbidity, and most died without having been admitted to the ICU. In effect, only 5% of the deceased patients aged 80 years or older in our series were admitted to the ICU. This is largely in line with the situation found in the ICUs of Lombardy (Italy), where only 1% of 1591 patients admitted to the ICU were over 80 years old, and less than half of them survived.20 However, age considered on an isolated basis should not be used in the assignment of resources in either the general population or in the population with COVID-19.21

In this sense, it is well known that comorbidity exerts a significant and negative influence upon different outcomes of hospital care, such as the duration of stay, complications, disability, chances for rehabilitation, resource utilization and the possibility of surviving the disorder leading to admission in the first place, and moreover results in worsened functional capacity and quality of life after discharge.22 Thus, comorbidity is the other key aspect in addition to age that must be considered in deciding the intensity of management in critical patients. We found dementia and active cancer to be the two comorbidities associated to non-admission to the ICU. Other conditions such as ischemic heart disease, cardiovascular disease and peripheral vascular disease could also be related, though in our study they failed to reach statistical significance. In any case, global comorbidity burden was clearly associated to a lesser probability of admission to intensive care. This comorbidity burden has been consistently more manifest in patients with COVID-19 that died in the larger series from other countries.23,24 Similarly, an analysis of 104 deceased patients in Wuhan (China) showed that those who died without admission to the ICU had greater comorbidity.25

On analyzing the results of our study, it cannot be ruled out that the decisions made in the Spanish ICUs may have been influenced by the poor outcomes reported in previous series (China, Italy) in patients with COVID-19 within a certain age range, or by the recommendations linked to age ranges made by the implicated scientific societies.10–12 All this could have led to exacerbated and premature pessimism, with age being focalized as the predominant factor, and with less consideration of other variables – something that does not happen in other serious diseases (sepsis or pneumonia, to mention just two examples). The fact that the deceased patients in our series had a high median age (80 years) and that the median Charlson score was not particularly high (2 points) would speak in favor of this hypothesis: patients without excessive comorbidity in which age possibly had a greater influence than normal in decision making, particularly in centers where demand at the time clearly outweighed the available resources. Although the ultimate cause underlying the decision was not analyzed in our study, Fig. 1 shows that comorbidity in a certain sense modulated the decision. Lastly, mention must be made of the importance of a practical ethical approach to the admission of these patients to the ICU, combining the classical survival probability factors considered in the our study (comorbidity, age, dependency), as a first approach, with social usefulness criteria in the event of an important disproportion between care offer and demand – in line with the recommendations of recent publications.26

The intensity of management of the patients in the emergency department, according to whether they were admitted to the ICU or not, did not differ in any major ways, beyond the greater use of lopinavir–ritonavir and tocilizumab in those patients that were finally admitted to the ICU. It should be noted that all these treatments were prescribed on an off-label basis, reflecting an attempt to always offer what was considered to possibly improve the condition of the patient, based on the data available at the time. This strategy, defined as living systematic review, has been defended by a number of authors for use during the COVID-19 pandemic, particularly in critical patients requiring admission to the ICU.27,28 It is quite possible that the increased use of lopinavir–ritonavir and of tocilizumab reflects the identification – from the emergency department – of those patients with a clinical profile indicating candidate status for more aggressive management. However, beyond the inherent aggressivity of treatment, it is also important to ensure the early identification of those patients likely to worsen and require intensive support measures.29 It is possible, however, that such early identification, which is routinely made in ordinary care situations, was not feasible during the resource-overwhelming period of the COVID-19 pandemic in the months of March and April 2020.

Lastly, our study does not address the potential effect of the patients admitted to the ICU with therapeutic ceiling upon the observed outcomes. Nevertheless, sensitivity analysis C seeks to offer an estimate in this regard (eliminating the patients that died in the ICU without having undergone OTI/IMV), and shows that the results obtained are very similar in all aspects to those obtained in the main analysis. On the other hand, we now know that certain noninvasive therapies, such as normobaric high-flow oxygen therapy, noninvasive mechanical ventilation, and continuous positive airway pressure (CPAP) devices were little used due to concern about the spread of aerosols in the environment, with secondary contamination of the healthcare staff (particularly at the start of the pandemic, and in the absence of adequate and supervised facilities for such therapies). However, recent studies have shown that these therapies are feasible in patients with COVID-19 and can avoid OTI/VMI in up to 30% of the cases.30,31 Although our study did not document whether these modalities were used in the 19.5% of patients that died in the ICU without OTI/IMV, they should be taken into account in new waves of elderly patients that may respond to such treatments, as a measure of support and mainly in the ICU due to the structural limitations of many centers, before resorting to OTI/IMV.

Our study has a number of limitations. Firstly, although a part of the patients of the SIESTA cohort were selected on a randomized basis (group A), another part (group B) corresponded to patients presenting some specific complication upon arrival in the emergency department. Nevertheless, the sensitivity analysis showed the main factors associated to death with admission to the ICU in both groups to be the same. Secondly, the tests made in the emergency department and RT-PCR confirmation of SARS-CoV-2 infection, the decision referred to admission, the place of admission, and the treatment of the patients were all adjusted to the changing protocols issued by the authorities and the infrastructure and availability of each center. Consequently, they were not homogeneous during the entire period. Thirdly, this is a secondary analysis of an existing cohort that included all the deceased patients, without prior sample size calculation. As a result, in some cases there may have been type II error, and some estimations of the associations between the basal characteristics and admission to the ICU might not have reached statistical significance due to the limited sample size involved. A fourth limitation is the fact that our study assessed age, gender, and comorbidity as basal characteristics. In this regard, other aspects such as frailty or functional dependency have been recommended by experts for better decision making.21,32,33 However, in view of the retrospective nature of our study, these variables could not be compiled. The same applies to certain severity indexes such as the APACHE II and SOFA, for which the patient case histories did not specify the scores, and for which not all the variables were available for calculating the scores – particularly in the patients that died without admission to the ICU. Likewise, we did not determine the degree of ICU occupation in each moment and in each of the hospitals (nor did we assess patient transfer from centers lacking an ICU). Consequently, we cannot determine the impact of the real-time availability of ICU beds upon the described outcomes. A fifth limitation of our study is the fact that among the different treatments applied (particularly in the patients not admitted to the ICU), the use of noninvasive ventilatory therapy was not documented – such therapy possibly having been employed in some cases in the emergency department or in conventional wards or areas habilitated for this purpose as therapeutic ceiling. A sixth limitation is that we intentionally avoided individual assessment of any patient in the series. We therefore cannot rule out the possibility that some of the deaths that occurred without admission to the ICU were sudden deaths with no chance for assessing or requesting admission to intensive care in view of the rapid and unexpected worsening of the patient. A seventh limitation is the fact that the complementary tests and treatments recorded were circumscribed to the emergency care setting, without exploring what happened from that time up until death. An eighth limitation was the fact that pressure upon the ICUs was not homogeneous among all the Autonomous Communities, and thus the application of limitations may have been more intense in some Communities than in others, and even in some concrete centers with respect to others. Although the aim of our study was to afford a global estimate without going into details referred to each Autonomous Community or center, we found the impact of comorbidity (adjusted for age) upon the probability of prior admission to the ICU to be similar in HEDs with a high care burden due to COVID-19 and in the rest of HEDs (p=0.33, for the interaction between comorbidity adjusted for age and care burden in the HED). Lastly, part of our conclusion that ICU admission of the patients with COVID-19 was consistent with routine practice in other disease processes was based on the unconfirmed assumption that patient age and the incidence of comorbidities (particularly dementia and cancer, which were those found to be significantly different) among the patients with COVID-19 that died admitted to the ICU were the same as in other disease conditions giving rise to admission to the ICU.

In conclusion, the profile of the patients with COVID-19 that died without admission to the ICU in Spain during the peak of the 2020 pandemic was consistent with that seen in routine practice before the pandemic, and the basal characteristics limiting admission were age and the global comorbidity burden. With regard to the latter, the comorbidities most closely related to non-admission to the ICU were dementia and active cancer. This is consistent with the observations in series of patients presenting other types of diseases, and with the opinion of the medical community in general.15,16,34 Therefore, this study, which included deceased patients in 62 Spanish hospitals, evidences proportionality in the use of the ICU as a management resource in patients with COVID-19 during the peak of the pandemic.

All the authors of the present manuscript contributed to the conception and design of the study, and to data acquisition, analysis, and interpretation. Dr. Òscar Miró wrote the article, and the rest of the authors conducted a critical review of the intellectual content, with final approval of the submitted manuscript.