INTRODUCTION

In December 2019, an outbreak caused by the novel human severe acute respiratory syndrome type 2 (SARS-CoV-2) coronavirus was first detected in Wuhan city, Hubei province, China ¹, since then it has spread worldwide. Because of its rapid spread, the World Health Organization declared the 2019 coronavirus disease (COVID-19) pandemic on March 11, 2020 ².

The COVID-19 pandemic, a little more than a year after its appearance in Peru, has caused more than one million infections among Peruvians and has caused the death of more than 40,000. According to the General Comptroller of the Republic of Peru, in 2017, only 57% of a total of 251 hospitals possessed the adequate number of health personnel for proper functioning and only 58% could perform basic laboratory tests in addition to having the minimum equipment for the analysis of samples ^{3 , 4}. The Peruvian health system collapsed; it had the highest mortality rate in the world during the first wave of contagion, with a shortage of health personnel of approximately 24,000 workers ⁵.

The main risk factors associated with death from COVID-19 are being ≥65 years of age, being male, suffering from hypertension, cardiovascular disease, diabetes, chronic obstructive pulmonary disease, or cancer ⁶. However, this information was obtained from high-income countries, which differs from low- to middle-income countries. It is known that the burden of critical illness is greater in middle- and low-income countries, which added to the precariousness of the health system ⁷, could have a greater impact on mortality due to COVID-19 in Peru. Due to this situation and the different behavior of the disease in each region of Peru, it is crucial to know the clinical or laboratory factors associated with mortality in order to be able to identify in a timely manner those patients at greater risk of death.

This study reports the results of an analysis of all hospitalized cases with confirmed diagnosis of SARS-CoV-2 pneumonia at Hospital III Daniel Alcides Carrión, in Tacna. The aim of the study was to identify demographic, clinical, laboratory and treatment characteristics associated with hospital death in patients with COVID-19.

MATERIALS AND METHODS

RESULTS

Of the 450 medical records reviewed, 99 did not meet the inclusion criteria (80 patients had a negative confirmatory test and 19 were hospitalized for other diagnoses). We included 351 medical records of patients hospitalized with a diagnosis of SARS-CoV-2 pneumonia. The median age was 61 (IQR: 50-69) years, the patients were predominantly male (74.1%) and 12.8% had more than three comorbidities, the most important being obesity (31.6%), arterial hypertension (27%) and diabetes (24.5%). The median hospital follow-up was 8 (IQR: 4-15) days.

Of the patients admitted, 70.1% had an oxygen saturation of less than 90%, of these, 39% had a saturation between 80% and 90%. There were 64 (18.2%) patients who required mechanical ventilation, 47 (73.4%) of them were admitted to the ICU, the median length of stay in the ICU and stay on mechanical ventilation was 13 (IQR: 8-18) and 8.5 (IQR: 5-14.5) days, respectively. Regarding ancillary tests, leukocytosis greater than 10,000 cells/mm³ was observed in 56.1% of patients, and the median percentage of lymphocytes was 10% (IQR 5-16) (Table 1).

During hospital stay, 119 (32.9%) patients died, representing a mortality rate of 3.2 per 100 person-days; 38 (59.4%) patients on mechanical ventilation died. Statistically significant differences in mortality were observed in patients older than 50 years and in patients with comorbidities such as hypertension, diabetes and chronic kidney disease. Oxygen saturation was significantly lower in the group of patients who died (p < 0.001). In addition, we observed that admission to mechanical ventilator, having more than three comorbidities, leukocytosis, lymphopenia, and having increased C-reactive protein, D-dimer and lactate dehydrogenase were significantly more frequent in patients who died. Colchicine administration was significantly more frequent in patients who survived (Table 2).

Predictors of mortality were evaluated by Cox proportional hazard regression. The crude analysis found that the factors associated with mortality were age older than 50 years, and all oxygen saturation categories less than 90%. Comorbidities associated with mortality in the crude analysis were hypertension, diabetes mellitus, chronic kidney disease, cancer, and having more than three comorbidities. Altered laboratory tests associated with mortality were elevated TGP and LDH. The treatments associated with decreased mortality in the crude analysis were colchicine and tocilizumab. In the adjusted analysis, oxygen saturation less than 90% was independently associated with in-hospital mortality, presenting HRa = 2.20 (95% CI: 1.07-4.51), HRa = 2.54 (95% CI: 1.14-5.70) and HRa = 4.07 (95% CI: 2.10-7.88) times higher risk of death for oxygen saturation of 89% to 85%, 84% to 81% and ≤80%, respectively. Likewise, being older than 65 years was independently associated with mortality with HRa = 3.55 (95% CI 1.70-7.40) as was having lactate dehydrogenase greater than 720 U/L with HRa = 2.08 (95% CI 1.34-3.22). Treatment with colchicine was shown to be a protective factor for death in the multivariate analysis with HRa = 0.46 (95% CI 0.23-0.91) (Table 3).

The risk proportionality assumption of the Cox model was evaluated using Schoenfeld residuals, and the assumption was considered to be complied with if the p value of the test was greater than 0.05. In this evaluation, the variable type 2 diabetes mellitus was excluded from the adjusted analysis because it had a p value = 0.023. Subsequently, a p value = 0.093 was obtained for the evaluation of the multivariate model. Figure 1 represents the survival function using Kaplan-Meier curves for the variables that were associated in the adjusted regression analysis.

Figure 1 Kaplan-Meier survival curves according to age category (A), oxygen saturation (B), lactate dehydrogenase (C) and colchicine treatment (D). 

DISCUSSION

One third of the participants died during the study period and approximately one fifth of them required mechanical ventilation. The risk factors with the greatest impact on in-hospital mortality were age over 65 years, oxygen saturation less than 90% and elevated lactate dehydrogenase. Colchicine administration resulted in a lower risk of death.

The mortality found in this study (32.9%) was higher than that observed in other reports from middle- and high-income countries (6.3% - 21%) ^{11 - 12}. This is probably due to the fact that in these countries the health systems were better prepared to face the pandemic, with a greater number of intensive care beds, health personnel and medical treatment (mainly medical oxygen), as well as standard and updated care protocols. Likewise, the mortality reported in two hospitals from the Peruvian capital (36.9% - 52.3%) exceeded that of this study, as well as the total mortality of three hospitals in the north of the country (60.2%) ^{9 , 13 - 14}. Another important difference was the mortality of patients admitted to intensive care in one hospital in Lima (73.9%) and in three hospitals in Lambayeque (81.3%) compared to that of this study (36.2%) ^{9 , 13 - 14}.

These differences, even at the departmental level, are probably due to the preparation level of the regional health system, as well as the idiosyncrasy of each region of Peru, which led to different responses to the quarantine measures applied by the central government, so that, although the spread of the virus occurred throughout the Peruvian territory, each region had different infection curves, being a little delayed in the southern part of the country. This provided more time to prepare, as well as to develop and update protocols for the management of patients with COVID-19, in addition to the fact that many patients received therapies that were still being evaluated for COVID-19 (corticosteroids, tocilizumab, early pronation) and this may not have been similar in all Peruvian hospitals. These measures are probably reflected in the low mortality rate reported.

On the other hand, the proportion of patients admitted to the ICU was lower than what was described in other studies (14.2% in New York; 19.4% in Detroit; 20.5% in Genoa), but higher than what was observed in a hospital in Lima (10.2%) ^{12 , 14 - 16}. These differences are probably due to the limited resources available in Peruvian hospitals. By March 8, 2020, the date in which patient zero was detected in Peru, there were only approximately 300 ICU beds ¹⁷. Although efforts were made to cover this large deficit, they were not sufficient to deal with the large volume of patients requiring intensive care.

We observed that the proportion of patients admitted with oxygen saturation less than 90% was higher than what was described in two level III public referral hospitals in Lima, Peru (64.5% in the Cayetano Heredia Hospital and 34.4% in the Edgardo Rebagliati Martins Hospital) ^{14 , 18}. As in other studies, oxygen saturation less than 90% was associated with an increased risk of death, mainly when oxygen saturation was less than 80%. This high proportion of hypoxemia on admission reflects the precarious prehospital monitoring system, and the lack of adherence to the COVID-19 treatment algorithm at the first level of healthcare, where the measurement of oxygen saturation is most useful ¹⁹.

In this study, colchicine treatment was found to be a factor associated with greater survival, and although the small number of participants does not allow definitive conclusions, the initial clinical experience in this hospital showed that the evolution and prognosis improved in those who received this treatment; these data are similar to the results of cohort studies and clinical trials that show that treatment with colchicine could improve the results in patients with COVID-19 ^{20 - 21}. It is proposed that colchicine could possess an inhibitory effect on inflammasome activation, destabilization and degradation, and a potential antiviral effect that could be secondary to inhibition of microtubule polymerization ^{22 - 23}. A systematic review and meta-analysis of four published studies reports an odds ratio for mortality of 0.28 (95% CI 0.18-0.44) suggesting a significant survival benefit ²³. Although this systematic review has serious methodological deficiencies and its results should be interpreted carefully, it proposes the need for randomized clinical trials with good methodological design in order to demonstrate this potential effect.

Despite conflicting results in some randomized clinical trials of tocilizumab for COVID-19 ^{24 - 27}, tocilizumab treatment in this study showed a protective effect, but not statistically significant in the adjusted regression. Although the number of subjects who received it was small (30 patients), these findings coincide with the preliminary report of the RECOVERY clinical trial that included 4,116 participants, which showed that those patients who received tocilizumab had a lower risk of death: rate ratio: 0.86 (95% CI: 0.77-0.99), this benefit was observed in all subgroups and mainly in those who additionally received corticosteroids ²⁶. These findings are similar to those reported in the REMAP-CAP clinical trial, where treatment with tocilizumab in critically ill patients with COVID-19 improved survival ²⁷.

These findings should be interpreted with caution. The main limitation of this study was its retrospective nature, so it was not possible to examine all clinical characteristics (weight, body mass index, nutritional status prior to hospitalization, etc.) or history (smoking, alcohol consumption, physical activity, etc.), because in most patients these data were not recorded at hospital admission. On the other hand, there were unmeasured laboratory data (troponin, interleukin 6, procalcitonin, total and myocardial creatine phosphokinase). It is necessary that future reports should have a single registry card for patients diagnosed with COVID-19 in which all these data are included. Furthermore, the study was carried out in only one health institution, so the results cannot be extrapolated to the general population, and this is reflected in the fact that the results differ from other investigations carried out in Peru, so prospective studies that include a larger number of patients, that are multicenter and that have more detailed information are needed. Finally, it is possible that many of the participants may have self-medicated before admission to the hospital, which could alter the natural history of the disease and some laboratory values due to the adverse effects of self-medication.

In conclusion, this study describes that most of the hospitalized patients were male, the most frequent comorbidities were obesity, hypertension and diabetes, and one third of the hospitalized patients died. The main risk factors for mortality due to COVID-19 were age, being older than 65 years, and clinical alterations at hospital admission, such as elevated lactate dehydrogenase and oxygen saturation less than 90%, especially less than 80%. Early recognition of these risk factors may help to identify patients with a poor prognosis early in the course of the disease and implement better strategies that could reduce COVID-19-related mortality. It should be considered that the behavioral pattern of COVID-19 in Peru is probably different in each region, so the public health strategies should be adapted to the reality of each of these regions.