To our knowledge, this is the largest cohort of patients with COVID-19 living in a city located at an altitude of more than 2000 masl and one of the largest in a single centre.
The overall mean age in this cohort was 66 years (IQR, 53–77), which was one of the oldest reported [28–33], and 69 years (IQR, 59-76) in patients admitted to the ICU, which was the highest age reported in patients admitted to this service [33–37].
In all, 25% of the patients in this study died. If we excluded patients who were managed as outpatients and lost to follow-up, the mortality of patients requiring hospitalization was 30%, which is similar to other series reported in Italy (25%), the United States (US) (21%), Iran (24%), and Germany (27%) [30, 32, 38, 39] and higher than the series from China (5% to 16%) [28, 31, 40].
In this cohort, 1223 patients (24%) required ICU management, which was a larger proportion than the 7% found by Yang et al. in China  but similar to previous reports from the USA ranging from 22 to 32% [29, 37]. IMV was given to 898 patients (17%), which was a value higher than the Chinese studies, in which the use of IMV ranged from 1.4 to 8% [31, 40], and in an Italian study (8.7%)  but similar to studies from the USA in which 12.2%, 21%, and 22% of patients, respectively, required IMV [32, 37, 41]. This suggests that, in China, in-hospital management was given to patients with less severe disease.
Out of the 898 patients who received IMV, 613 of them died (68%), which is a figure lower than the studies from the onset of the pandemic in China (81%)  and the USA (88.1%) , which is similar to reports from the UK in March and April 2020 (66.3%)  and at the Brazilian National Registry of Intensive Care (66.9%) . In later studies in high-income countries, the mortality reported in patients with IMV was lower, and among those, an Italian study showed mortality of 52% in ventilated patient s, and a multicentre prospective cohort of patients admitted to ICUs in France, Belgium, and Switzerland showed a 36% mortality rate .
Roedl et al. , in a multicentre study conducted in Germany, showed a mortality rate of 44% in patients with IMV. Lim et al. , in a review of 69 studies with more than 57,420 patients ventilated for severe COVID-19, found a mortality rate of 45% (95% CI, 39–52%); however, at the time of publication, many patients were still hospitalized. In this same study, the estimated mortality for patients aged 61 to 70 years was 77%.
Oxygenation indices have been broadly studied around the world as predictors of mortality and ICU admission in patients with COVID-19, leading to useful clinical practices to prioritize patients at hospital and ICU admission. Within those, a lower P/F ratio at ICU admission has been found to be a predictor of mortality among these patients, but few studies have been carried out regionally where this is stated.
In this study, the P/F ratio at ICU admission is one of the lowest when compared to other cohorts around the world, 118 vs. 129 in the USA; 160 in Italy; 154 in France, Belgium, and Switzerland; and 172 in China [29, 34, 35, 46]. This may be explained by the hypobaric hypoxaemia expected with altitude in Bogotá, Colombia. Although a study developed in critically ill patients in the USA at sea level  described a median P/F ratio of 103 at ICU admission, we analysed their parameters and found that the median FiO2 at ICU admission in these patients was 90%, compared to ours, in which the median FiO2 at ICU admission was 50%.
The P/F ratio on admission to the ICU in patients who underwent IMV was 112 vs. 154 in those who did not, thus indicating that the decision to start IMV was based not only on the oxygenation indices but also on the clinical presence of respiratory failure, and that a higher level of hypoxaemia was tolerated for the start of this therapy in relation to cities located at altitudes less than 1000 masl.
In patients admitted to the ICU, the P/F ratio was also significantly lower in patients who died (105) than in those who survived (137) compared to other studies where the P/F ratios in deceased patients and survivors were 134 and 163 , respectively.
There is insufficient clinical evidence to change the standard management of patients with COVID-19 based on oxygenation indices. The 2012 Berlin consensus  recommends an adjustment of the P/F ratio based on the BP in places over 1000 masl (BP/760); however, this recommendation does not have references on how it was derived . In daily practice, based on experience and institutional management guidelines, patients with a P/F ratio greater than 250 without supplemental oxygen and without clinical dyspnoea may receive outpatient management. Patients with a P/F ratio of 200 to 250 are admitted to the hospital for surveillance, and supplemental oxygen is used for oxygen saturation greater than 88% and PaO2 greater than 60 mmHg. Patients with a P/F ratio less than 200 require surveillance in intermediate care with a high risk of requiring admission to the ICU. The decision to initiate IMV should not only be based on oxygenation indices but also on the presence of respiratory distress and tolerance and response to other methods of oxygen administration, such as high-flow cannula and noninvasive mechanical ventilation. Clinical trials are required to evaluate which oxygenation level once the patient is on mechanical ventilation is optimal for the initiation of neuromuscular relaxants and prone use. In the institutional guidelines of the Hospital Universitario Mayor, we recommend P/F levels of between 120 and 150 for these interventions at the discretion of the clinician.
Based on the current evidence, the oxygenation level at which it is best to start IMV is unknown. It must also be taken into account that during peaks of the pandemic, hospital occupancy was close to 100%, which interferes with the quality of care and may delay the start of IMV.
There is little information about the clinical characteristics of patients in high-altitude areas. Chen et al. , in a retrospective court of 67 patients who live in the Tibetan and Qiang Autonomous Prefecture of Ngawa, China, located at 2600 masl, found 4 severe cases (6%), 39 nonsevere cases (58.2%), and 24 (35.8%) asymptomatic cases, suggesting, as in the work of Seclén et al. , that altitude can protect against severe diseases caused by SARS-CoV-2. These results differ greatly from our findings, where only 17% of the patients were nonsevere cases with outpatient management, mortality was 25%, 24% of the patients were admitted to the ICU, and in 17% of the cases, IMV was used.
Abdelsalam et al.  analysed the clinical characteristics and laboratory findings of patients infected with COVID-19 in the city of Taif, Saudi Arabia, located at 1879 masl. Of 790 patients included in the analysis, 91.5% recovered without ICU admission, 6.8% recovered after ICU admission, and 1.7% died. In addition, a paired analysis was performed by age and comorbidities with 208 patients from the city of Jeddah, Saudi Arabia, located at sea level, finding a mortality of 14.4% and observing a lower mortality in people living at a high altitude. This reported mortality is much lower than that found in this study, which is explained by the age of the patients, since the total mean age was 41.4 years and 57.3 years among those who died, which is in great contrast with our population where the mean age of the entire population was 63.5 years and 74.3 years in the patients who died.
Several authors have documented alterations in C-reactive protein, D-dimer, and LDH that are associated with severity and mortality from COVID-19 . Variations in the haemogram, coagulation tests, markers of myocardial injury, creatinine, and factors associated with systemic inflammation have been identified . In the present study, the paraclinics that had a greater relationship with clinical outcomes in the bivariate analysis were LDH, D-dimer, and the neutrophil-lymphocyte ratio (NLR).
An uncontrolled inflammatory response is associated with worse clinical outcomes in SARS-CoV-2-infected patients . Elevated serum concentrations of some inflammatory mediators have been described in patients with severe COVID-19, in particular, cytokines such as IL-1 and IL-6 . However, in countries with limited resources, its clinical use is rare. More widely available tools, such as the NLR, can be used as markers of hyperinflammatory states [55, 56]. In COVID-19, it has been found that elevated NLR values are associated with a worse prognosis . In addition, its ability to predict the severity and risk of death has been evaluated, as summarized in a meta-analysis that found an acceptable performance (mortality AUC 0.90 and severity AUC 0.85) with a cut-off of 6.5 .
Similar to the data described, our findings show a higher NLR among the deceased (NLR 11.78) than among the survivors (NLR 7.58). Likewise, the NLR was higher in patients admitted to the ICU (outpatient NLR 5.64, hospitalized NLR 7.82, and ICU NLR 11.58) and those who required IMV.
Moreover, the NLR among surviving patients was higher than those previously reported . It remains to be confirmed whether high altitude can influence the baseline expression of inflammatory markers  or if it is a manifestation of disease severity at altitude. No previous articles have reported this ratio at high altitude.
LDH is the enzyme responsible for the passage from pyruvate to lactate, and its elevation has been associated with severity and death in infectious and neoplastic diseases [61, 62]. Different studies have established the performance of LDH as a prognostic biomarker in patients with SARS-CoV-2 infection [51, 63–65]. It has been documented that its elevation is associated with a risk up to six times higher for developing severe forms of COVID-19 (OR 6.7, 95% CI: 2.4–18.9)  and four times higher for mortality (OR 4.22, 95% CI: 2.49–7.14) . In our study, the data were consistent with those previously reported. LDH was higher among the dead patients (471 U/L) than among the living patients (351 U/L). Likewise, its value was higher among ventilated patients (510 U/L) than among those who did not require this support (354 U/L).
Ballaz et al.  described the findings of the haemogram upon admission in Quito, Ecuador, at 2850 masl, finding expected inflammatory changes for patients with COVID-19, with a high NLR, and lymphopenia in the most severe cases. The study by Ballaz showed mean haemoglobin values similar to those in our study, 14.9 g/dL vs. 14.7 g/dL in nonsevere cases and 15.4 g/dL vs. 14.3 g/dL in severe cases. These findings suggest that haemoglobin values associated with exposure to high altitude at altitudes between 2500 and 3000 masl may not be relevant to the severity of COVID-19 presentation.
In the prediction model presented by the CART methodology, oxygenation indices, LDH level (as an inflammatory marker), and COPD (as a comorbidity) increase the probability of IMV, findings similar to other studies [37, 44].
In the prediction model for mortality, it is not surprising that the requirement for IMV is the main factor associated with death, since this indicates which patients are sicker; however, the model allows us to observe that being older than 79 years, hypoxaemia and elevated levels of LDH (as an inflammatory marker) increase the probability of death, even in nonventilated patients. These findings are similar to other studies [31, 33, 44, 46]. The advantage of this modelling is that it allows us to discover effect modifiers and complex interactions between variables.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.