Respiratory support technology is the main means of treating children with critical and severe conditions. Because invasive respiratory support has been noticed to have many disadvantages, such as the presence of ventilator-induced lung injury (VILI), in recent years, more attention has been paid to non-invasive respiratory support, which could alleviate the pain of endotracheal intubation and reduce complications. However, studies have confirmed that the failure of non-invasive respiratory support is related to an increase in intubation rate and mortality「10」. Therefore, if the focus is only on non-invasive respiratory support, invasive respiratory support via endotracheal intubation may be delayed, accompanied with poor prognosis for patients. As a new type of non-invasive ventilation oxygen therapy device, high-flow nasal cannula oxygen therapy (HFNC) is even more comfortable than traditional non-invasive mechanical ventilation (NIV), with a lower incidence of complications such as head shaping, nasal injury, pneumothorax, and abdominal distension [11–12]. Several clinical studies have shown that HFNC has achieved good clinical effects in the treatment of respiratory diseases, such as respiratory failure, respiratory distress syndrome (RDS), sleep paroxysmal apnoea symptoms, and reduced the rate of endotracheal intubation in children [13–15]. The study by Wing et al[16] also found that early HFNC used in children with acute respiratory insufficiency would reduce their possibility of endotracheal intubation and invasive mechanical ventilation. However, there is no uniform standard for HFNC indications and contraindications in pediatric applications, and there are few reports on its related risks. Furthermore, the study by Gaunt et al[17–18] found that if the patient's condition did not improve within 48 h after HFNC, the respiratory support mode should be upgraded; otherwise, it led to further deterioration in respiratory function and increased mortality. Therefore, we suspected that the failure to HFNC in critically ill children would also have adverse consequences.
Considering that the earlier the failure time was, the lower the possibility was that failure was caused by disease progression, we compared the HFNC success group with the 48 h failure group, 24 h failure group and 2 h failure group, and the results were almost consistent. Among the 349 pediatric patients who were included in this study, the GCS score, PRISM III score, pH value, PaCO2/PaO2 ratio and oxygenation index were significantly different between the HFNC success and failure groups (P < 0.05). By multivariate logistic regression analysis, PRISM III score and PaCO2/PaO2 ratio were considered as risk factors for HFNC failure.
Neither PCT nor the CRP, which suggested inflammatory reactions exist, were significantly different in the 48 h failure group or 2 h failure group compared with success group, but their differences between the 24 h failure group and the success group were statistically significant(P < 0.05). Multivariate logistic regression showed that when PCT > 0.67 ng/ml, the risk of HFNC failure within 24 h increased by more than 2 times. Normally when PCT and CRP were significantly increased, inflammation or organ damage was often more severe in pediatric patients. Therefore, PCT and CRP are supposed to be a risk factor for HFNC failure. Further study is needed to expand the sample size and conduct hierarchical analysis for the inconsistent results. At same time when PCT or CRP significantly increase, HFNC as a support treatment should be closely monitored.
The pediatric risk of mortality (PRISM III) is currently the most widely used as pediatric critical assessment tool worldwide,which was positively correlated with organ failure「19」. This study found that the PRISM III score was higher in each failure group than in the success group (P < 0.05). Multivariate logistic regression showed that when the PRISM III score was > 4 points, the risk of HFNC failure within 48 h was over 4 times higher, while a PRISM III score > 6.5 points was associated with an over 27 times higher risk of HFNC failure within 2 h. The higher the score, the higher the risk of failure was. Therefore, for critically ill children, especially those with PRISM III scores > 6.5 points, HFNC should be closely monitored. Because PRISM III score involves 14 physiological parameters and 23 parameter ranges「20」, such as blood gas, blood sugar, electrolytes, liver and kidney function, and coagulation function, it would take longer time to get than some warning scores for example PEWS score, which limits the value of PRISM III to early predict and need more research to find more valuable scoring system.
HFNC can provide not only a constant oxygen concentration but also a certain positive end-expiratory pressure in the airway with high-flow gas, thus improving oxygenation「21」.PaO2 is an index of respiratory function, while PaCO2 is a better index to pulmonary ventilation function [22]. Therefore, the PaCO2/PaO2 ratio could indicate pulmonary ventilation and diffusion function. In this study, it was found that the difference of PaCO2 between the 24 h or 48 h failure group and success group was statistically significant (P < 0.05), while PaCO2/PaO2 ratio was statistically different between all the failure groups and the success group (P < 0.05). Moreover, the higher PaCO2 and the PaCO2/PaO2 ratio, the shorter failure time is. Multivariate logistic regression showed that when PaCO2 ༞43 mmHg, the risk of HFNC failure within 24 h and 48 h was over 3 times and 4 times higher, respectively. When the ratio of PaCO2/PaO2 was ༞0.67, the risk of HFNC failure within 2 h was over 64 times higher. Therefore, for pediatric patients with abnormal ventilation, HFNC should be carefully selected. More research on how to combine PaCO2 and the PaCO2/PaO2 ratio to judge the failure risk of HFNC is needed.
Expert guidelines for clinical practice with non-invasive positive pressure ventilation[9] indicated that evaluation about the effectiveness of non-invasive respiratory support therapy after 2 h initial treatment played an important role in subsequent treatment decisions; therefore, in this study, we found that the decline in PEWS scores for the success group was significantly greater than that for the failed groups. Furthermore, the changes in pH value, oxygen saturation index and oxygenation index before and after HFNC showed an upward trend in the success group and a downward trend in the failed group (P < 0.05). Multivariate logistic regression analysis found that when the oxygenation index decreased by > 28% after 2 h of HFNC treatment, the risk of early HFNC failure increased by more than 5 times. Given that, the change trend of the oxygenation index was a warning for early HFNC failure, if which showed a decreasing trend after 2 h treatment with HFNC, there was a high possibility of failure to HFNC and close monitoring was required, and the respiratory support mode should be upgraded in time to avoid further deterioration of respiratory function.
To further analyse the possibility of increased HFNC failure due to improper choice about respiratory support modes by inexperienced physicians, we compared the HFNC failure rates between the 1st half and the 2nd half of each month and found that there was no significant difference (P > 0.05), which means no reason to attributes the failure to inexperienced new residents, suggesting that the current training and working modes were reasonable.
The PICU stay time, hospitalization cost and in-hospital mortality were greater in the group that respiratory support upgraded from HFNC to invasive than the group that was synchronously admitted to the PICU and directly received invasive respiratory support, but not statistically different, while the invasive mechanical ventilation time was statistically longer (P < 0.05). The failure of HFNC in critically ill children had adverse consequences, which at minimum, might prolong the invasive mechanical ventilation time.