In December 2019, the first cases of Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2) pneumonia, which causes Coronavirus Disease-2019 (COVID-19), were reported in Wuhan city (Hubei province, China). Since then, this virus has spread worldwide in pandemic form.

COVID-19 patients usually present with fever, dry cough, upper respiratory congestion, and shortness of breath. Cases of headache, hemoptysis and diarrhea, loss of smell (anosmia) and loss of taste (ageusia) have also been reported1,2.

The pandemic has caused an unprecedented situation resulting in a significant number of reported cases, deaths and, ultimately, in a major social and economic upheaval. Science has had to mobilize all its resources to provide an urgent response to the need for evidence. Today, many unknowns remain, and there is a massive and pressing demand for evidence on the treatment of COVID-19. Aware of this urgency, the Spanish Society of Hospital Pharmacists (SEFH) launched, back in March 2020, the Spanish COVID 19 Drug therapy Outcomes Registry (RERFAR). The aim of this study was to determine the baseline characteristics associated with increased 42-day mortality in patients hospitalized for COVID-19 in Spain.

The present study included a cohort of patients with PCR confirmation of COVID-19 admitted to 174 Spanish hospitals between March 20 and July 15, 2020. A maximum of 200 patients were included from each hospital to avoid overrepresentation of hospitals with a larger number of patients. A simple random sampling procedure produced a total sample of 15,628 patients. Patients with suspected nosocomial infection with COVID-19 (n = 227) were excluded if symptoms began after admission.

A total of 15,628 patients were included in the study; 2,806 of them (18.2%) died within 42 days. Tables 1–3 show the baseline characteristics of the patients included in the registry, both from a demographic and a clinical point of view, as well as pre-existing comorbidities, their baseline analytical profile, and the different (pharmacological and/or supportive) interventions initiated within 48 hours from admission.

As regards the demographic characteristics of the sample, mean age was 66.29 years (SD: 15.74); 57.2% were men, 15.5% were obese, 11.4% came from nursing homes and 4.5% were healthcare providers. A total of 19.6% of subjects had a respiratory rate > 24 rpm, 47.1% had an oxygen saturation rate below 93%, 40.1% had fever and 11.6% had severe pneumonia (CURB-65).

The most frequent baseline diseases were (Table 1): hypertension (50.1%), diabetes mellitus (21.8%), previous neurological disorders (14.6%), renal failure (9.8%) and ischemic heart disease (9%). At admission, 373% and 14.3% of subjects were receiving treatment with ACEI/ARA-II and NSAIDs, respectively.

A total of 65.5% of the patients received oxygen therapy at baseline, while 5% received noninvasive ventilation and 4.8% were subjected to mechanical ventilation. As for pharmacological treatments, 664 patients (4.3%) did not receive any treatment. The most frequently prescribed drugs in the first 48 hours after admission included hydroxychloroquine (81.3%); low molecular weight heparins at prophylactic doses (74.5%); antimicrobials (68.7%), of which the most frequent was azithromycin (in 55.6% of all patients); lopinavir/ritonavir (55%): and corticosteroids (16.7%).

Tables 1–3 show the results of a bivariate analysis of the relationship between each of the variables and 42-day mortality. Table 4 shows the results of the multivariate analysis. With regard to the baseline demographic and clinical characteristics, coming from a nursing home (OR 1.938; 95%CI: 1.686-2.227); presenting with a higher respiratory rate (OR 1.511; 95%CI: 1.330-1.717); and having diabetes (OR 1.221; 95%CI: 1.089-1.368), heart failure (OR 1.477; 95%CI: 1.256-1.736) and moderate (OR 1.738; 95%CI: 1.492-2.025) or severe (OR 2.940; 95%CI: 2.746-3.491) pneumonia were associated with higher mortality. In contrast, being female (OR 0.769; 95%CI: 0.684-0.863) and young (18-29 years: OR 0.097; 95%CI: 0.039-0.24; 30-64 years OR 0.162; 95%CI: 0.136-0.194; 65-79 years: OR 0.428; 95%CI: 0.379-0.483), having asthma (OR 0.770; 95%CI: 0.618-0.961) and being a healthcare provider (OR 0.433; 95%CI: 0.238-0.787) were associated with lower mortality. As shown in table 4, baseline alterations in various analytical values (leukocytes, lymphocytes, platelets, AST, LDH, procalcitonin, CPK, D-dimer, hemoglobin, C-reactive protein and creatinine) were associated with increased mortality, with values ranging from an OR of 1.200 (95%CI 1.040-1.384) for D-dimer levels between 500 and 3,000 ng/mL to an OR of 2.175 (95%CI 1.601-2.954) for AST levels > 100 U/L. A platelet count > 450 x 10 3platelets/mm3 was associated with a reduction in mortality (OR 0.593; 95%CI 0.391-0.897), as was an ALT value between 40-80 U/L (OR 0.726, 95%CI 0.617-0.854) and > 80 U/L (OR 0.490, 95%CI 0.373-0.643) and creatinine values < 0.5 mg/dL (OR 0.676, 95%CI 0.468-0.976). Finally, as regards treatments initiated within 48 h from hospital admission, the different modes of respiratory support were associated with greater mortality (oxygen therapy: OR 1.987, 95%CI 1.739-2.271; noninvasive ventilation: OR 2.877, 95%CI 2.348-3.480; and mechanical ventilation: OR 3.471, 95%CI 2.825-4.266). Pharmacotherapy with interferon beta (OR 1.507; 95%CI: 1.266-1.794), corticosteroids (by continuous [OR 1.425; 95%CI: 1.248-1.627] and bolus [OR 1.421; 95%CI: 1.179-.712] infusion), and lopinavir/ritonavir (OR 1.158; 95%CI: 1.033-1.300) were associated with higher mortality; conversely treatments associated with lower mortality included hydroxychloroquine (OR 0.707; 95% CI: 0.615-0.813), tocilizumab (OR 0.761; 95%CI: 0.603-0.960), low-molecular-weight heparin at prophylactic doses (OR 0.836; 95%CI: 0.741-0.942) and azithromycin (OR 0.870; 95%CI: 0.779-0.970).

This article analyzes the association between 42-day mortality and a wide range of clinical and demographic variables and analytical parameters related to COVID-19 patients, collected at hospital admission. The study of the factors associated with mortality was intended to gain a better understanding of the disease and coming up with a better stratification of patients so as to allow for a more efficiently management of the resources dedicated to treatment. The use of 42-day mortality as an outcome variable extends beyond in-hospital mortality and includes potential mortality after discharge. SEFH's Spanish COVID 19 Drug therapy Outcomes Registry includes a large number of hospitals from the entire Spanish territory as well as a high number of patients, 80% of whom were included in March 2020 (first COVID-19 wave). This offers a significant degree of homogeneity across the different cases included. For logistic simplification, the number of included patients per center was limited to 200, which may lead to an overrepresentation of small hospitals. In large hospitals, on the other hand, only the first patients admitted were included, which means that the level of experience with and evidence on the treatments in those cases is probably lower. To reduce this selection bias, all hospitals with more than 200 patients hospitalized for COVID-19 that participated in the registry were randomly sampled.

In addition to the ones mentioned above, the work presents several limitations that should be taken into consideration for the interpretation of results. Given the observational nature of the study and its design as a retrospective collection of data, the associations described between each variable and mortality may be subject to biases or confounding factors. This design is not suitable for testing the effect of treatments. Firstly, as patient inclusion started at the beginning of the pandemic, there was a lack of evidence about the treatments selected and most patients received multiple treatments during their hospitalization. On the other hand, as pharmacotherapy was one of the different baseline characteristics included in the analysis, a 48- hour period from admission was set to determine which patient had received a treatment. This complicated interpretation of the results of the multivariate pharmacotherapeutic analysis as patients who did not receive a given treatment within 48 h from admission could have received it later. Moreover, by the time they received it, their final health status may have changed, which introduced a difficult-to-eliminate source of bias.

The scarcity of available evidence at the beginning of the pandemic resulted in the main treatment for COVID-19 in our cohort being hydroxychloroquine, used in more than 80% of patients6. This explains discrepancies with currently available evidence, which has shown hydroxychloroquine to lead to no benefits regarding mortality and to increased adverse events7,8. Dexamethasone is currently the standard treatment for patients requiring respiratory support9,10, who accounted for 75.3% of the patients in this registry. Even so, only 16.7% received corticosteroid treatment in the first 48 hours. In the present analysis, patients who received corticosteroids after the first 48 hours were classified as patients not treated with corticosteroids. However, in light of the existing evidence, it is very plausible that these patients may have obtained a benefit from corticosteroid therapy which means that classifying them as not treated with corticosteroids in the statistical analysis could constitute a contradiction with respect to the results of RECOVERY11. In addition, it is also likely that patients who received corticosteroids were the ones with most severe forms of the disease, implying a selection bias12. Subsequent evidence has shown the use of corticosteroids to be indicated in severe and critically ill patients but to provide no benefit in mild10,13 and moderate cases14. The other treatment that has shown some effect on mortality reduction is tocilizumab15. According to the SEFH registry data, it was administered to 4.5% of patients within the first 48 hours. The initial bivariate analysis showed an OR of 1.414 (95%CI 1.184-1.689) while, when adjusted for the remaining variables, the analysis pointed to a protective effect of tocilizumab on 42-day mortality, with an of OR 0.768 (95%CI 0.609-0.969). This is consistent with the results of the only two studies where tocilizumab has been shown to reduce mortality, the REMAPCAP study16 and the RECOVERY study11, and could be indicative of the use of tocilizumab in patients with progressive COVID-19 (C-reactive protein ≥ 75 mg/L and O2 Sat < 92%) or in critically ill patients within the first 48 hours of admission11.

The adjusted multivariate model found diabetes and male sex to be associated with mortality. According to the literature, this association may be explained by the increased expression of angiotensin-converting enzyme peptidase 2 (ACE2), which is the gateway for the SARS-CoV-2 virus17. Specifically, patients with diabetes-derived hyperglycemia present with an increased expression of ACE2, which seems to favor penetration of the virus into immune cells; patients with asthma, in contrast, have a decreased expression of this enzyme18. As for sex differences, in addition to the anatomical, hormonal and lifestyle differences observed during the SARS epidemic of 200319, it seems that differences also exist with respect to the immune response, whereby men would seem to be more vulnerable to infection with COVID-1919.

Regarding the association between mortality and the analytical parameters of our analysis, it is well known that lymphocytes play an essential role in controlling viral infection, specifically the inflammatory response and homeostasis. Lymphopenia due to lymphocyte destruction (particularly T lymphocytes) and depletion caused by virus invasion has consistently been shown to be a poor prognostic factor20,21. The findings of our study coincide with those of the literature in that the clinical parameters associated with the increased severity and mortality of COVID-19 include elevated levels of LDH, C-reactive protein, procalcitonin, CPK, platelets and D-dimer22,23.

An expected finding from the registry was an increase in mortality among patients who required externally supplied oxygen (oxygen therapy, noninvasive ventilation, or mechanical ventilation) to maintain adequate saturation, as well as among patients with higher CURB-65 scores. These findings are consistent with those reported by other authors24,25.

The 42 day-mortality follow-up applied in this study, aimed at identifying potential increases in residual mortality after discharge, constitutes a more comprehensive follow-up than that found in other observational studies6,26,27 and clinical trials11,28, which typically analyze mortality at 28 days. In our study, mortality after hospital discharge was 0.7%, which increased overall mortality from 17.5% to 18.2%. This higher mortality highlights the need to follow up the outcomes of the different studies on COVID-19 over longer periods of time.

The results of SEFH's Spanish COVID 19 Drug therapy Outcomes Registry, which contains a broad representation of the population admitted to hospital for COVID-19 during the first wave of the pandemic and which followed up patient mortality for longer than most studies in the literature, show the possible prognostic usefulness of the baseline analytical and clinical parameters used, which may contribute to improving the management of a disease.

No funding.

The authors would like to thank María del Carmen Valcárcel Cabrera from the Andalusian School of Public Health as well as all who contributed to putting together the RERFAR registry (Annex 1).

No conflict of interest.

The protocol was published on the website of the European Network of Centres for Pharmacoepidemiology and Pharmacovigilance. It is available at: http://www.encepp.eu/encepp/viewResource.htm?id=34344

The preliminary results of this article was presented in the XXXIX Annual Meeting of the Spanish Epidemiology Society (SEE) – XVI congress of the Portuguese Epidemiology Association – IX SESPAS congress. LEÓN, SEPTEMBER 7-10, 2021.