Blood Gas Analysis as A Predictor of Mortality in Critical COVID-19 Patients in Dr. Moewardi Hospital Surakarta

COVID-19 has been declared a global pandemic affecting many countries, including Indonesia. The worsening of COVID-19 patients may go unnoticed because one of the clinical manifestations of COVID-19 is silent hypoxia, where the patients' blood oxygen saturation drops with no visible symptoms. Blood gas analysis is needed to detect silent hypoxia. This study aims to confirm whether blood gas analysis results correlate with the mortality of critical COVID-19 patients. This was a retrospective cohort study, the samples were taken from the medical records of critical COVID-19 patients from May 2020-July 2021. The variables observed were gender, age, comorbidities, oxygen saturation (SaO ), partial pressure of 2 oxygen (PaO ), partial pressure of carbon dioxide (PaCO ), bicarbonate (HCO -), and blood pH. Descriptive statistics, 2 2 3 bivariate, and multivariate analyses were done. There were 210 samples, where 137 patients survived (65.24%) and 73 patients died (34.76%). The data analysis showed that comorbidities and blood pH significantly correlate with the mortality of critical COVID-19 patients, with p-values < 0.05; indicating that comorbidities and blood pH can be used as mortality predictors in critical COVID-19 patients. The RR for comorbidities and blood pH were 2.194 and 2>294, respectively, with CI 95% 1.202-4.833 for comorbidities and 1.151-5.295 for blood pH. The cut-off value used for blood pH was 7.310.


INTRODUCTION
Coronavirus Disease 2019 (COVID- 19) was first reported in 2019, and since then has spread around 1 the world. The disease primarily affects the respiratory system, with symptoms ranging from minimal symptoms to hypoxia with Acute 2 Respiratory Distress Syndrome (ARDS). COVID-19 patients are considered in a critical phase when ARDS occurs. Acute respiratory distress syndrome is defined as a condition when the PaO /FiO is 300 2 2 mmHg or lower. Patients with ARDS tend to deteriorate very quickly, usually within one week 3 after the onset of ARDS. This deterioration may be u n d e t e c t e d b e c a u s e o n e o f C O V I D -1 9 ' s manifestations is silent hypoxia, where the patient's oxygen saturation is lower than normal without any 4 dyspnea or other discomforts. Blood gas analysis is needed to detect silent hypoxia. Blood gas analysis measures oxygen saturation (SaO ), partial pressure 2 of oxygen (PaO ), partial pressure of carbon dioxide 2 (PaCO ), bicarbonate (HCO -), and blood pH, 2 3 therefore this test will provide information on the patient's oxygenation, ventilation, and acid-base 5 balance. This study aimed to confirm whether blood gas analysis correlates with critical COVID-19 patients' mortality during their stay in the hospital.

METHODS
This was an observational analytic study with a retrospective cohort design. The research took place in Dr. Moewardi Hospital, Surakarta, the data used in this study was secondary data taken from critical COVID-19 patients' medical records in Dr. Moewardi Hospital from April 2020 to August 2021. The data included gender, age, existing comorbidities, and blood gas analysis results, which consist of SaO , 2 PaO , PaCO , HCO -, and blood pH. Comorbidity is 2

RESULTS AND DISCUSSIONS
There were 210 subjects in total, with 137 subjects that survived (65.24%) and 73 patients who died (34.76%). In the descriptive analysis, the variables were divided into two groups, clinical characteristics, and laboratory characteristics. Based on the Kolmogorov-Smirnov test on variables age and blood gas analysis results, it was found that the data were not distributed normally. Then Mann-Whitney U test was done. The samples' characteristics and the results of the Mann-Whitney U test are shown in Table 1.
The results showed significant differences in the age and comorbidity groups. As for the gender, SaO , 2 PaO , PaCO , HCO -, and blood pH, there were no 2 2 3 significant differences. Bivariate analysis was done using Chi-Square test and the results are shown in If the patient had any comorbidities present then the value was 1, otherwise, the value was 0. If the patient has a blood pH of <7.310 then the value was 1, otherwise the value was 0. A higher probability value means a greater risk of mortality for the patient.
According to previous studies, comorbidities that may increase the risk of deterioration and mortality of COVID-19 patients are cardiovascular diseases, COPD, hypertension, diabetes, liver diseases, kidney disease, cancer, obesity, pregnancy, and 10,11 dementia.
Pregnant females infected with COVID-19, had increased risks of maternal and fetal complications, such as preeclampsia/eclampsia, 12 maternal death, preterm birth, and low birth weight. Patients with autoimmune diseases have a higher   13 increasing the risk of their mortality. In patients with hypertension, diabetes, Chronic Kidney Disease (CKD), and COPD, there was an increase in ACE-2 expression, which was the entry point of SARS-CoV-2, resulting in an increased risk of transmission and worsening of COVID-19. Preexisting comorbidities may also cause deterioration of immune function, such as in CKD where dysregulation of immune function and inflammation may occur, and in diabetes cases with decreased antibody production. These immune dysfunctions would increase both the risk of infection and the worsening of COVID-19. In addition, in patients with comorbidities, there is a possibility of adverse drug reactions due to routine 14 drug consumption. Hypoxia and inflammation occurring in COVID-19 patients may cause acidosis. These three things i n t e r a c t w i t h e a c h o t h e r a n d f o r m a pathophysiological cycle, which could worsen the 5 patient's condition, leading to mortality. Inflammation in COVID-19 causes an increase in metabolism in an oxygen-limited environment. Moreover, in COVID-19 patients, gas exchange disorders may occur and would further reduce the amount of oxygen in the blood. As a result, hypoxia, which was a condition in which cellular oxygen needs cannot be met, occurs. The insufficient amount of oxygen to carry out aerobic metabolism causes anaerobic metabolism to take place in the form of glycolysis, which produces lactate. High levels of lactate may cause metabolic acidosis. Lactate is a granulocyte chemoattractant. Higher lactate production means higher proinflammatory molecules are released. The combination of hypoxic and acidotic conditions also causes hyperactivation of dendritic cells and dysregulation of cytokine responses. This would exacerbate the inflammation that had already 15,16 occurred and may even cause hyperinflammation.
Acidosis may also be caused by systemic thrombosis, which is caused by hypercoagulation in 14 COVID-19. In addition, COVID-19 was thought to cause acidosis directly via the interference of the Renin-Angiotensin-Aldosterone System (RAAS) and inhibition of proton excretion. COVID-19 causes massive damage to cells expressing ACE2 receptors, where it is known that ACE2 is a RAAS antagonist. High RAAS activity would increase sodium 15,17 reabsorption and potassium excretion.
It is known that patients with comorbidities are susceptible to metabolic acidosis. Diabetic patients are at risk for ketoacidosis. COVID-19 may also cause ketoacidosis in patients without diabetes, as COVID-19 may directly damage pancreatic cells, 15 causing diabetes-like conditions. Acidotic conditions would reduce cardiac contractility and cardiac output. It would also induce arterial vasodilatation. This phenomenon could cause sudden cardiac death, making the mortality rate for critical COVID-19 patients with acidosis 18 significantly higher. It is also known that in cardiac arrest cases caused by non-cardiac factors, such as acidosis-, non-shockable rhythm is more prevalent.
Cardiopulmonary resuscitation (CPR) would be ineffective in these conditions, which would further 19,20 worsen the prognosis.

CONCLUSIONS AND SUGGESTIONS
The presence of comorbidities and low blood pH (<7.310) can be used as mortality predictors for critical COVID-19 patients. Further study in multiple health centers is needed to represent a larger population scale.