Continuous Positive Airway Pressure (CPAP) is a common form of support for patients admitted to Intensive Care Units (ICUs) of industrialized countries with respiratory distress (1). Nasal CPAP (NCPAP) is effective in correcting hypoxemia and contributes to reducing the number of children requiring endo-tracheal intubation and mechanical ventilation (2). CPAP is most frequently delivered to neonates using conventional mechanical ventilators, and thus there is minimal or no cost saving. There are other ways of delivering CPAP, such as Bubble-CPAP, which requires a source of gas flow (typically 6-8 L/ minute in a neonate), an air-oxygen blender, a humidifier and a T-piece.(3). The expiratory arm is inserted in a bottle of water and the level of CPAP delivered is equivalent to the length of the expiratory tubing that remains under water. Robust equipment is now available at a fraction of the cost of mechanical ventilators. Bubble-CPAP has potential advantages over the mechanical ventilation, such as lower cost, ease of application by nursing staff, lower risk of complications, and has been proposed as an inexpensive method of delivering CPAP in developing countries (3).

High flow air/oxygen mix is useful in reducing the indication of mechanical ventilation (4); however, there is a lack of randomized studies comparing it with bubble CPAP or with standard flow O2 supplementation by nasal prongs. High flow air/oxygen mix uses flows of 2 litre per kg per minute of blended air/oxygen mix, usually with a low fraction of inspired oxygen (say 25-40%). It is easy to apply, but requires additional equipment to standard oxygen therapy, and closer monitoring. "High flow" delivers uncertain levels of CPAP, so it is not clearly superior to bubble-CPAP, and there have been no controlled comparative trials of these two techniques.

Pneumonia and malnutrition are two of the most common co-morbidities in children in developing countries (5). In hospitals in resource-poor settings, children with severe malnutrition and pneumonia often present with respiratory distress with or without severe hypoxaemia and impending respiratory failure (6). They initially receive O2 supplementation through nasal prong or face mask. Support from bubble CPAP might help to effectively treat hypoxaemia, improve respiratory function, avoid the need for mechanical ventilation and its complications, and reduce mortality.

Almost half of the patients admitted in the intensive care unit of the Dhaka hospital of ICDDR,B present with hypoxaemia, many with impending respiratory failure. Children with pneumonia also invariably have severe malnutrition with or without diarrhoea (Chisti MJ, MMed thesis, unpublished data). They often need mechanical ventilation, with attendant costs, complications and high mortality rates. However, no published data are available about the use of bubble-CPAP in children with pneumonia and malnutrition and there have been no controlled trials of CPAP in developing countries.

The Hypothesis is:

In children with severe pneumonia and hypoxaemia the probability of treatment failure (see definition below) will be significantly lower when respiratory support is initially provided by bubble-CPAP or high-flow, humidified air/O2 mix by nasal prongs, compared to standard oxygen flow.

1. Design and Methods

1.1 Study design: This will be a randomised, controlled study in which the investigators will prospectively provide intervention in children under the age of 5 years admitted to the Dhaka Hospital of ICDDR,B and fulfilling the inclusion criteria (January 2011 and December 2012), subject to the obtaining written informed consent from respective parents/care-givers.

1.3 Intervention and comparators

Children under five admitted to the SCU of ICDDR, B with pneumonia and hypoxaemia will be studied. This study will evaluate the proportions of deaths in children under five in three treatment arms.

The investigators will use sealed envelopes to randomly assign the children to one of three treatment arms. In first arm, children will receive bubble CPAP; in 2nd arm, children will receive standard O2 supplementation by nasal cannula at 0.5 - 2 l/min; in the 3rd arm, children will receive humidified high flow air/O2 mix at 2 l/kg/min through nasal cannula. The three arms will be compared for rates of treatment failure (see below).

During the study period, children (whether under ventilation or not) in all three study arms will also receive standardized hospital management for pneumonia and malnutrition.

"Treatment failure" will be standardized, based on clinical and monitoring data.

During the study period any child from either study group who develops the features of "treatment failure" (any two of three criteria in the definition below) will be managed according to best practice in the clinical context. "Treatment failure" is the criteria to be followed to decide intubation and mechanical ventilation.

Laboratory investigations:

Routine: Blood for total and differential count of white blood cell (WBC), haematocrit, culture and sensitivity, serum electrolytes and creatinine, stool culture and urine culture, chest radiograph (CXR).

Protocol-specific: Arterial blood gas analysis, tuberculin skin test (TST), gastric lavage for acid fast bacilli (AFB) and culture for MTB.

1.4 Measurements

All children will be monitored for by pulse oximetry for arterial O2 saturation, respiratory rate, lower chest wall in-drawing, intercostal retraction, head nodding, cyanosis, tracheal tug, heart failure (defined by the presence of tachypnea, tachycardia, gallop rhythm, hepatomegaly, pedal oedema, basal crackles). Arterial or capillary blood gas analyses will be done for children failing to maintain saturation (>90% with allocated treatment), or if there is concern about hypercarbia or acidosis.

1.5 Criteria for primary outcome

.

The primary outcome, by intention to treat analysis will be reached if a child:

A. Has any two or more of the following three criteria of treatment failure (after at least one hour of intervention)

1. Severe hypoxaemia (SpO2<85%) after being on one of the study arm treatments for > 1 hour

2. Clinical signs of exhaustion, including active contraction of respiratory muscles with paradoxical abdominal and thoracic motion, gasping, severe chest wall in-drawing

3. PCO2 > 80mm Hg and pH < 7.2 on capillary blood gas OR B. Receives intubation and/or mechanical ventilation OR C. Dies while in hospital or within 30 days of discharge OR D. Absconds * while still on the allocated respiratory support.

   • If they abscond the day before they are due to be discharged and they are off oxygen and are well, then that's just a mother's choice to go home early, such children should not be put in the category of equivalent of treatment failure.

     2 Sample Size Calculation and Outcome (Primary and Secondary) Variable(s) Background, recent data (1st January until 31st October, 2010) from the ICU of the Dhaka Hospital of the ICDDR,B revealed that the approximate treatment failure from the management of hypoxaemia in children with pneumonia and severe malnutrition was 30%. The investigators assumed that the new intervention would result in a reduction of treatment failure from 30% to 18% (i.e. a 40% absolute reduction). Thus to detect a difference of 40% in the death from hypoxaemia with 90% power and type 1 error 0.03 (the investigators are intending to do a mid-term analysis and for that the investigators are considering type 1 error 0.03 instead of 0.05 ), the sample would be 295 children in each group {sample size = [(p1×q1 + p2q2)/ (p2-p1)2]×factor for α, β; where is the p1 is the percentage of the "treatment failure", q1 is the 1- p1, p2 is the percentage of the expected "treatment failure" from intervention; q2 is the 1- p2, α is the type 1 error and β is the type 2 error and factor for α, β with 90% power is 11.9}. Considering 10% drop out after admission in hospital, the total sample size is at least 325 in each group. So, our total sample size will be 325×3=975. The patients will be enrolled over a period of two years.

Inclusion Criteria:

Children of either sex, aged 0-4 years, with severe/very severe pneumonia (per WHO guidelines) and hypoxaemia (SpO2 < 90%) will be included in our study in the ARI Unit of the Longer Stay Ward (LSW), High Dependency Unit HDU, and ICU unit of Dhaka Hospital of ICDDR,B.

Exclusion Criteria:

Children with uncorrected cyanotic CHD, hypercapnoea (PCO2 > 65 mm of Hg), status asthmaticus and upper-airway obstruction, and those for whom inform consent can't be secured from their parents/ care-givers. Children with features (any two of three criteria given below) of "treatment failure" on admission (before enrollment into the study) will also be excluded from the study. Arterial blood gas analysis will be performed to check the exclusion criteria