Factors affecting the outcome of the usage of high-flow nasal cannula on severe or critically ill COVID-19 patients: a multicentric study from a developing nation

In resource-constrained settings, High-Flow Nasal Cannula (HFNC) can reduce the burden on mechanical ventilation in COVID-19 induced Acute Hypoxemic Respiratory Failure (AHRF). The aim was to observe the factors those might affect the outcome of the usage of HFNC on severe/critically ill COVID-19 patients. This is a multicentric prospective observational study. We observed rRT-PCR positive severe/critically ill ICU patients requiring HFNC for more than six hours. Statistical analysis was done to correlate between factors and outcome. Weaning from HFNC was successful in 47.5% of patients. The death rate was higher in ≥ 50 years older (56.50%), and patients with asthma (60.57%), COPD (60.00%), and CKD (68.42%). Fever (91.67%), cough (72.5%), and dyspnea (67.5%) were the most common symptoms. Mortality rates were higher for patients with raised blood sugar, creatinine levels. Severely systemic inflammatory response was seen very high for the expired patients. On HFNC, percent saturation of oxygen (SpO2) and partial pressure of oxygen (PaO2) progression was significantly high for the surviving patients requiring less inspired fraction of oxygen (FiO2%). The survival rate was higher for the patients using both HFNC and non-rebreather mask (NRM) concomitantly. While after HFNCSpO2% and FiO2% were significantly related with outcome of the HFNC only treated patients, duration of hospital stay and on HFNCFiO2% affected the HFNC + NRM treated patients’ outcome. HFNC could save more lives of critically ill AHRF patients who otherwise might need invasive or noninvasive ventilation. Some biochemical tests were observed to have association with the prognosis of the disease though HFNC was given to all. Survival benefit of dual HFNC and NRM therapy needs future study.


Introduction
Since its outbreak in China in December 2019, COVID-19's rapid spread worldwide has forced the global population to live in a pandemic situation with dread for life (Zhu et al., 2020). Among the previously identified six human CoVs (coronaviruses) (Tang et al., 2015), severe acute respiratory syndrome coronavirus (SARS-CoV) and Middle East Respiratory Syndrome Coronavirus (MERS-CoV) were considered to be highly infectious (Cui et al., 2019). However, the COVID-19 contagion, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was in no way behind its beta coronavirus counterparts in terms of infection rates and manifesting severe health complications. As of now, SARS-CoV-2 has affected 220 countries and territories drastically (Worldometers, 2021a). As of September 1, 2021, SARS-CoV-2 has infected over 210 million people and has caused 4,542,691 deaths, most of which have been identified in countries other than China (Worldometers, 2021a). COVID-19's first instance was detected in Bangladesh on March 8, 2020 (Islam et al., 2020). To date, about 1.5 million COVID-19 positive cases have been identified and more than 25,000 deaths have been reported in Bangladesh (Worldometers, 2021b). As SARS-CoV-2's entry route involves various organs of the respiratory tract and its localization is majorly in the lung cells, a range of symptoms mutual to other diseases of the respiratory tract is triggered. Clinical manifestations often range from fever and can take an even severe form, such as pneumonia. A significant number of cases become critical (EWGNER, 2020;Wu and McGoogan, 2020). The burden of severe cases has created an unusual freight on the health sector, presenting the urgency of worthwhile therapies for . Several studies reported that severe COVID-19 leads to acute hypoxemic respiratory failure (AHRF), which requires a high fractional concentration of inspired oxygen (FiO2) and noninvasive ventilation (NIV) procedures (Antonelli et al., 1998;Carrillo et al., 2012;Delclaux et al., 2000). Several devices are being used in intensive care units (ICUs) to deliver supplemental oxygen to COVID-19 patients who have developed hypoxemic respiratory failure including simple face mask, face mask with a nonrebreathing reservoir bag, venturi masks, etc. HFNC, on the other hand, tends to be more successful than others because it can reach 100% humidification at 37°C and has a positive end-expiratory pressure (PEEP) effect while patients breathe with the mouth closed (Roca et al., 2010;Sztrymf et al., 2011). It acts as a noninvasive setup to improve oxygen level and decrease dead space carbon dioxide level. Relative to other oxygen delivery devices, HFNC foremost balances the patients' inspiratory demands by delivering up to 60 L/min of air flow with a FiO 2 up to 1.0. In this way, it reduces the adverse effects more as there is no risk of tooth avulsion, intubation of the esophagus, Ventilator-Associated Pneumonia (VAP), Ventilator Induced Lung Injury (VILI), and use of neuromuscular blocker and sedatives are not required in HFNC in contrast to endotracheal intubation (Huang et al., 2018;Lee et al., 2016;Maitra et al., 2016;Ni et al., 2017). Wake-prone positioning is a striking benefit of HFNC, a patient could eat and drink himself as well and there is very little need for maintenance as this is a very simple device. HFNC also decreases the burden of mechanical ventilation during a phase where mechanical ventilation is demanding owing to the rise in critical cases every minute. In previous studies, it was found that HFNC to some extent reduces the need for intubation in patients requiring supplemental oxygen and was also found to have contributed to reducing mortality rates (Frat et al., 2015;Rochwerg et al., 2019). The goal of this study is to evaluate the associated factors affecting the outcome of HFNC usage on severe/critically ill ICU admitted COVID-19 induced AHRF patients.

Patients and data collection
This multicentric prospective observational study was conducted on 240 COVID-19 patients admitted to ICUs in 250 bedded Chattogram General Hospital, Chittagong Medical College Hospital, Chattagram Maa-O-Shishu General Hospital, Parkview Hospital, andSurgiscope Hospital Ltd. between February 15, 2021, andJune 14, 2021. These are specialized hospitals authorized to manage the most critical COVID-19 patients with 50 ICU beds within the ICUs of the port city of Bangladesh, Chattogram. Epidemiological and demographic data for the cases were collected by designated researchers from the patients' treatment records and by interviewing the corresponding attendants.

Ethical clearance
The Institutional Review Board of the 250 bedded Chattogram General Hospital approved this study (Approval No.: 1724).

Inclusion and exclusion criteria
This survey included Real-time Reverse-Transcriptase Polymerase Chain Reaction (rRT-PCR) positive patients with AHRF admitted to COVID ICU and received HFNC oxygenation for more than six hours. Patients' having SpO 2 less than 90% persistently after providing 15 liters per minute oxygen was considered as an indication for HFNC. On the contrary, patients who needed MV or NIV from the beginning of ICU admission and those who did not want to participate in the study willingly were excluded. Additionally, a face mask with NRM along with HFNC was provided to the patients who failed to maintain target oxygen saturation (>90%) after high flow.

The criteria for ICU admission
Most of the severe and critically ill patients were admitted to the ICU. Patients having dyspnea, i.e., a respiratory rate of ≥ 30 beats per minute in rest and oxygen saturation (SpO 2 ) of ≤ 92%, were enlisted in the 'severe' category. The 'critical' group, on the other hand, included those who had respiratory failure, sepsis, or shock, necessitating mechanical ventilation, as well as those who had multiple organ failures requiring ICU support. Here, acute respiratory distress syndrome (ARDS) was defined according to the Berlin definition (Ranieri et al., 2012), and shock was defined according to the Sepsis-3 criteria (Singer et al., 2016).

rRT-PCR Test
Respiratory tract samples were used to do rRT-PCR to confirm whether the sample cases were positive for COVID-19 (Mannan et al., 2021). Throat swabs, nasopharyngeal swabs, and bronchial aspirates were collected from the patients in a tube containing a viral transport medium, and all of them were shifted to the laboratories later. According to the World Health Organization's (WHO) guideline, the SARS-CoV-2 RNA extraction for COVID-19 was conducted in the Molecular Biology laboratory of the Microbiology department of Chittagong Medical College (Rudra et al., 2021) and Chattagram Maa-Shishu O General Hospital.

Statistical analysis
IBM SPSS version-25 was used for data analysis and figure preparation. To check significant differences between categorical variables, Pearson's Chi-Square (χ2) evaluation method was used. Categorical and continuous variables were tested for difference by applying Independent-Sample T-Test (95% confidence interval) and compared 'means' with 'standard deviations'. P values less than 0.05 were considered statistically significant, and in the case of categorical and continuous variable correlation, a P value of "Equal variance not assumed" was considered. Factors having significant differences when comparing with the outcome of the patients were further analyzed by Simple bivariate logistic regression and multiple bivariate logistic regression to find the correlating factors. The variables strongly correlated (value> 0.7) with other variables (determined by Pearson's Correlation Coefficient analysis) were omitted from the regression analysis. Then the significant factors were further analyzed (Simple bivariate logistic regression and multiple bivariate logistic regression) by dividing into two groups HFNC only, and HFNC + NRM treated concerning the outcome of the patients. Results were shown in crude odds ratio (COR), adjusted odds ratio (AOR), and their ranges at 95% confidence interval. Omnibus tests of model coefficients' P value less than 0.05 and 'Hosmer and Lemeshow Goodness of Fit' test's P value greater than 0.05 were considered significant to assess, if the regression model had been fit for the data. The specificity and sensitivity of the data were also tested during regression analysis.

Relationships of Outcomes with Patient Demographics and Clinical biomarkers
In our study, 240 critically ill COVID patients treated with HFNC were included, and 47.5% (126) of them survived. The mortality rate among patients with less than 50 years (39.30%) was significantly (P = 0.024) lower than with ≥ 50 years (56.50%). Death among rural residents (54.20%) was higher than urban residents (51.10%). But the total number of cases was high in urban residents (Table 1). The death rate was higher among patients having a history of smoking (61.30%) than those who never smoked (49.70%). According to our study, 50 patients did not have any comorbidities and the mortality rate was significantly higher (55.80%) among comorbid patients. With P values of <0.001, the mean ± SD of hospital stay (14.07 ± 7.91), and ICU stay (9.65 ± 5.64) in days was significantly higher among surviving patients (Table 1). A total of 133 patients received HFNC with NRM and the survival rate (59.4%) was more than those who were only on HFNC (32.2%) ( Table  1). Correlations between investigations and patients' outcomes have been shown in Table 1. Random blood sugar level was significantly related to the outcome (P = 0.038). The mortality rate was 58.6% among patients with increased blood sugar levels, while the rates among normal and impaired glucose tolerance blood sugar levels were below 50%. We witnessed another significant correlation between serum Creatinine levels (P = 0.013) and the death rate. The death rate among patients with elevated Creatinine (64.8%) was higher than normal ones (47.3%) ( Table 1). D-dimer level among deceased patients (2.39 ± 2.90) was higher than survived patients (1.173 ± 1.45), and this relation was significant (P = <0.001). The mortality rate increased proportionally with the level (from 'no' to 'severe') of the systemic inflammatory response (P = 0.006). 93.8% of the patients with severe systemic inflammatory responses died, while 48.1% with no inflammatory responses expired in the ICU (Table 1).

Impact of Comorbidities and Clinical Features on Outcomes
Death and survival rates of patients with comorbidities are illustrated in Figure 1-A. Common comorbidities for HFNC treated patients were Hypertension (HTN) (60.4%), Diabetes Mellitus (DM) (56.7%), Ischemic Heart Disease (IHD) (19.6%), Asthma (9.6%), Chronic Kidney Disease (CKD) (7.9%), and Chronic Obstructive Pulmonary Disease (6.3%). DM and HTN constituted more than 50% of the patients, and among them, 52.21% and 55.17% of patients died in the hospital. The survival rate was less than 40% for the patients with CKD and Asthma (Figure 1-A). Relations between clinical features of HFNC treated COVID-19 induced AHRF patients and hospital outcome were also observed ( Table 2). Having a cough was significantly related to the outcome of the patients (P = 0.003). A total of 220 patients suffered from fever, and 51.4% of them died. Besides, 162 patients had difficulties breathing, and 54.9% of them expired. Other symptoms were sore throat (9.17%), loss of smell (7.92%), diarrhea (7.08%), weakness (35.42%), and confusion (6.25%) ( Table 2).

Immediate Complications Observed during HFNC Therapy
Immediate complications of HFNC have been shown in Figure 1-B. The percentages of non-visible nasal bleeding and nasal obstruction (evidenced by the removal of clotted blood) were 40% and 31.67% of the total cases for the death and survived ones, respectively (Figure 1-B). A total of 30% of the patients faced discomfort (irritation and burning sensation in the nose) because of HFNC. As per our finding, non-visible nasal bleeding and obstruction by clot were much more common than other HFNC complications. These led to more breathing difficulty for the patients. To protect nasal passage from drying out from high flow and prevent non-visible nose bleeds from culminating in partial/complete obstruction of nostrils, 'Liquid paraffin' and 'Normal saline' were used.

Common Medications Used in HFNC Treated Patients
We illustrated the post-admission medications administered in Figure 2. The frequently used medications were antibiotics (intravenous/oral), antiviral drugs (intravenous/oral), low molecular weight heparin (twice/once daily), and steroid (dexamethasone/methylprednisolone). According to our data, plasma therapy and interleukin-6 inhibitors were not common like other treatments along with HFNC in ICU for COVID-19 patients (Figure 2). About 55% of patients treated with IL-6 inhibitor died.

Factors affecting outcome of the patients
Two variables like duration of ICU stay and P/F ratio was removed from the analysis, because they were strongly correlated with other variables. Almost three times (2.9, 1.1-7.4) more of the patients who survived later in hospital/ICU due to COVID-29 related AHRF were given NRM too. Duration of hospital stay was slightly lower (1.1, 1.0-1.4) for the deceased patients (Table 4). Those who survived suffered less (0.1, 0.01-0.9) from severe systemic inflammatory response syndrome. The improvement of SpO2 % was 1.2 times (1.0-1.5) better for the survived patients, and they required less (0.9, 0.87-0.94) FiO2%. Another factor that was found significant with the outcome was On HFNC-PaO2 and this factor showed more progression (1.1, 1.0-1.1) for survived patients more than the dead ones (Table 4).

Factors affecting HFNC ± NRM concerning outcome of the patients
As HFNC ± NRM was observed as one of the significant factors relating to the outcome of the patients, the effects of other factors on patients treated with HFNC only and HFNC + NRM were also analyzed in this study. Those who were managed with only HFNC and also survived were seen to improve their SpO2 % almost two times (1.8, 1.3-2.6) more than the dead ones, and they needed less (0.9, 0.8-0.9) fraction of inspired oxygen (FiO2%) through HFNC (Table 5). Duration of hospital stay was more (1.1, 1.0-1.3) for the HFNC + NRM treated survived patients, and the HFNC + NRM treated expired patients needed more (0.9, 0.8-0.9) fraction of inspired oxygen (FiO2%) through a nasal cannula (Table 5). Further studies with larger sample regarding dual therapy are needed.

. Discussion
Symptomatic treatment protocols not being sufficient for ICU admitted COVID-19 patients had become a worldwide phenomenon for restoring their health (Demoule et al., 2020). Aligning to a previous study that recommends the use of HFNC for lessening invasive/mechanical ventilation use, our study found that HFNC substantially helped reduce mortality rates in cases having COVID-19 induced AHRF; by maintaining optimal oxygen saturation (Matthay et al., 2020). Heated and humidified oxygen (at high flow) was administered through high flow nasal cannula to 240 ICU admitted COVID-19 patients, and 47.5% of severely/critically ill AHRF patients' lives were saved by using HFNC who failed to maintain oxygenation by 15 L/min oxygen. We also found that in this study population of critically ill COVID-19 cases, the instance of fever and cough was higher as compared to similar populations of COVID-19 induced AHRF patients reported in a previous study . However, in that study, a higher instance of breathing difficulty was found in patients. Mortality was very high in patients with elevated body temperature in this study, which corresponds to the observations by  Patients feeling severe weakness were also typical in our study. But sore throat, diarrhea, and anosmia were rarely experienced by patients.
Previous studies have stated that comorbidities like HTN and DM can make the patients more vulnerable to COVID-19 (Akter et al., 2020;Rudra et al., 2021). In a study in the southern region of Bangladesh, the death rate of diabetic patients was 60.2%, which was 1.2 times higher than our study . Though the mortality percentage of hypertensive COVID-19 patients was almost equal to another study of Bangladesh, IHD patients died at about 1.5 times the higher frequency in their study . Leonardo Antonicelli et al. found out that Asthma seems to play a minimal role in clinical severity for COVID-19 patients, which contradicts our finding (Antonicelli et al., 2021). The mortality percentages for patients with CKD, Asthma, and COPD were very high in our study. Further studies are obligatory for the proper management of COVID-19 patients with these diseases. Additionally, the death rate of the COVID-19 affected was very high among patients with high RBS, serum Creatinine, and Procalcitonin levels. Another significant finding of this study was a number of patients with high RBS being more likely to die than those with standard RBS. The mortality rate substantially increased with the rise of the Procalcitonin levels, and almost 94% of patients with severe systemic inflammatory responses died which is significantly affected the death rate of the patients. The death rate of patients with elevated Ddimer was high in a study in Bangladesh, and our finding was D-dimer of the in-hospital death cases being double from those of survived patients . High levels of RBS, Creatinine, Procalcitonin and Ddimer could be denominators to HFNC treated COVID-19 patients' mortality. After HFNC only, the SpO 2 % for survived patients increased about two times more than those who died later in our study. Patients' death with decreased SpO 2 % in a study was higher than ours, which indicates the benefit of using HFNC . On HFNC, oxygen saturation among both dead (died later) and surviving patients increased compared to before starting-HFNC. But the SpO 2 % progression was comparatively better among patients who survived post COVID-19 treatment. Though the P/F ratio was more than 1.5 times higher for the surviving patients, the level of P/F ratio was below the standard level for both deaths and survived cases. Moreover, NRM was tested when only HFNC failed to meet the oxygen saturation to the target level. We found that the survival rate increased after providing both HFNC and NRM. HFNC with NRM also increased the SpO 2 % and PaO 2 of the patients more than those with only HFNC administration. Survived patients needed less fraction of inspired oxygen for both HFNC only and HFNC + NRM treatment which proved their better condition than the others. Convalescent plasma therapy is an effective treatment for COVID patients within the first week of the onset of the symptoms (Kesici et al., 2020). But in this study, the plasma exchange was a rare treatment protocol, and it did not affect the mortality of the patient it was administered upon. The survival rate after using dexamethasone was greater than methylprednisolone usage. A similar result on the administration of these drugs was found by Heidi Ledford (Ledford, 2020). A total of 55.4% of the patients who got IL-6 inhibitors (Tocilizumab) (N = 56) died. But it has become a very commonly used medication in our country. A recent RCT (recovery trial) found that tocilizumab improved survival and other clinical outcomes in hospitalized Covid-19 patients with hypoxia and systemic inflammation (RECOVERY Collaborative Group, 2021). It was done in a high resource setup, but further research is needed in the context of lower income countries. Additional study is also required to prevent complications like partial or complete obstruction of the nasal passage due to mucosal injury caused by high flow jet of air. Further insights and comparison could be made on the contribution of HFNC in surviving a patient if the effects of other modes of ventilation (e.g. CPAP or MV) could have been observed. We observed that locally practiced preventive measures to protect against high flow induced nasal mucosal injury were not sufficient in some cases. Though concomitant use of NRM seems to work, it is not an established practice. The duration of the study was short. Lastly, although our results were significant, larger sample size is needed for more accurate correlations.

Conclusions
According to the findings of our study, HFNC is indeed an effective mode of delivering supplemental oxygen and improving SpO 2 % to critically affected COVID-19 AHRF patients. About half of the patients' lives could have been saved as observed in this study. However, comorbidity and the levels of some specific biomarkers were found to be major caveats in controlling the progression of the disease, even after HFNC administration. It was also observed that NRM concomitant with HFNC improves survival rate in patients. To corroborate the usage of HFNC + NRM and reduce the complications of HFNC for COVID-19 affected AHRF patients, studies with larger samples are needed.