Humidified flow nasal cannula (HFNC) delivers heated and humidified gas mixture at a flow greater than patient’s inspiratory flow demand and can provide intermediate level of support between low-flow oxygen delivery and non-invasive ventilation (NIV) in critically ill children [1]. Retrospective studies have shown that HFNC is useful for conditions like bronchiolitis, asthma, pneumonia and congenital heart disease [2]. The evidence for its safety or usefulness in children is limited [3]. There is paucity of prospective clinical trials on the effectiveness of HFNC in respiratory failure (not due to bronchiolitis) in pediatric intensive care unit (PICU).

This study aimed at assessing the efficacy and safety of HFNC as a primary mode of treatment in respiratory distress in children.

This cross-sectional study was undertaken at an urban tertiary care hospital of Western India from 1 January, 2018, to 31 December, 2018. The study was approved by the institutional ethics committee and informed consent from parents was taken prior to enrollment. Consecutive patients with respiratory distress necessitating admission to PICU, in the age group of 1 month to 16 years of age were included. Children requiring immediate non-invasive (NIV) or invasive ventilation and those with contraindications to HFNC, altered sensorium (GCS <12), apnea and catecholamine resistant shock were excluded.

Respiratory distress was defined as hypoxia (SpO2 <94% in room air), tachypnea (as per age) and increased work of breathing (chest wall retractions, use of accesssory muscles of breathing and nasal flaring/grunting). HFNC was started as the first line treatment if all the above clinical signs were present. Primary outcome measure was need for ‘NIV’ or invasive ventilation.

Bronchiolitis was defined as a clinical syndrome of respiratory distress in children less than two years with rhinorrhea followed by lower respiratory infection resulting in wheezing and crepts. Children with fever, respiratory distress, tachypnea and infiltrates on chest radiograph were classified as pneumonia. Children with fever, respiratory distress, tachypnea and chest signs of wheezing and crepts but without infiltrates on chest radiograph were classified as LRTI with wheeze.

A respiratory clinical score with the following parameters was calculated: age specific respiratory rate scores 0 to 3, retractions 0 to 3, dyspnea 0 to 3, and wheeze 0 to 3. Total score ranged between 0 for normal and 12 at the extremes [4]. FiO2 was adjusted to keep arterial oxygen concentration between 92-97% to calculate saturation to FiO2 (SF) ratio. HFNC tolerance was assessed using modified COMFORT scale [5]. The scale estimates eight parameters with a 1 (low) to 5 (high) score: alertness, calmness, respiratory response, physical movement, mean arterial pressure, heart rate, muscle tone, and facial tension. The total score can range between 8-40 (score of 17-26 suggesting good comfort). Respiratory clinical score, SF ratio and modified COMFORT score were calculated before starting HFNC treatment, at 60 to 90 minutes and 12-24 hours afterward.

HFNC system (Fisher and Paykel Healthcare, New Zealand) with junior circuit 900PT501 was used. Infant OPT316 or Pediatric OPT318 nasal prongs were selected as per child’s age. Flow was initiated at 1-2 L/kg/min for infants and 1 L/kg/min for pediatric patients and adjusted according to patient response and tolerance (max 2 L/kg/min). Failure on HFNC was defined as need for NIV or invasive ventilation, when clinical deterioration was present. Criteria for intubation were respiratory arrest, refractory hypoxia (SpO2 <90% on 100% FiO2), exhaustion due to increased work of breathing and inability to protect airway. Criteria for switching to NIV were left to discretion of the attending intensivist.

For calculation of sample size, a baseline risk for need of ventilation as 16% was assumed in children with respiratory distress presenting to the emergency. We hypothesised that HFNC would reduce the risk by 50% (absolute reduction of 8 percentage points). Using alpha error of 0.05 and for 90% power, we calculated a sample size of 178. To allow for potential 10% recruitment failure rate, required sample size was increased to 200.

Statistical analyses were performed using IBM SPSS 23 version (IBM 2015), and significance was assessed at 0.05 level. Comparisons between two groups were made using independent sample Mann Whitney U test and Kruskall Wallis test for continuous measurements. Univariable and multivariable Cox regression models were used to assess the association of HFNC failure with various clinical parameters.