These two papers are part of the multicentric WHO study on signs of severe illness in young infants. Hence, a lot of data especially related to multivariate analysis incorporating numerous predictors including low birth weight is not depicted in the papers. Similarly, the sensitivity and specificity data has been left out by the editors because of constraints of space. The main paper is being published in The Lancet soon and couple of supplementary papers shall follow.

At the Delhi site, 748 exclusions were made because of following reasons: needed immediate resuscitation 19, outside study area 259, hospitalized in previous two weeks 163, received prior treatment 184, previously participated in study 152, congenital malformations 2 and refused consent 25 (some infants excluded for more than one reason). This information is missing from the text box in the figure because of formatting error and this account for the discrepancy in numbers.

It is true that special arrangements were made to run the "OPD" till 9 pm for the study. This was done to imitate the ground realty of infants reporting sick any time of the day. This special "OPD" was physically located in the emergency for logistic reasons at Chandigarh while at Delhi site during 9-11.30 am , these infants presented in the OPD or emergency ward from casualty as per hospital existing policies. A logbook was maintained to register all eligible infants. However, the infants coming to emergency in a state needing cardio-pulmonary resuscitation were not included.

Ideally, one would have liked to conduct this study in the community itself. However, it was not possible to do such a large scale study at multiple sites in the community because of technical and logistic constraints of obtaining a gold standard assessment in the community, the risk of contamination of findings between the two observers performing clinical evaluations and the inability to validate their assessments in the community. So, a simulation was done by choosing a place as close to the community as possible –at first line health care facilities which work like First Referral Units and where parents have free access to walk in with any kind of complaints. This is also reflected in the pattern of morbidities seen in the infants reporting to both these sites (Tables II and III of Delhi paper and Tables I and III of Chandigarh paper) which mimics that expected in the community. The first contact person was a nurse with GNM or ANM qualifications who "had not worked in leading hospital". This is akin to the real life health worker who would be expected to see the infant in the community. A short period of training and re-orientation was done as requirement for a research study, to ensure uniformity and consistency in their assessment. However these ANM’s/GNM’s are expected to be fully trained in routine to perform these simple assessments.

Pulse oximetry was done by the study person B in all enrolled infants after completing the physical examination. He could use the information as an aid for making decisions. In the modern era of medicine, one would expect pulse oximeter to be ultimately available at all first referral units. This gadget was available at all sites and uniform methods were adopted by study persons B across all sites for making a decision regarding "need for admission". If indicated, initial laboratory investigations (serum bilirubin, glucose, chest x-ray) were done and a decision was taken within two hours by study person B for need of hospitalization. The quality of the diagnoses being made by study person B was ensured by an initial period of training, creating a manual of operations with standard definitions and an ongoing review of case records by a committee of senior investigators (pediatricians with more than 15 years experience) with regular feedbacks. For this purpose complete case records along with all relevant investigations were taken into consideration for providing feedback to study person B.

In addition, one of the useful secondary objectives of the study was to document in detail and precisely the possible range of specific diagnosis encountered at first referral units in infants <2 months of age. Hence, laboratory investigations were necessary to confirm the diagnoses and have reliable information.

Anil Narang,
AK Deorari*,
Department of Pediatrics,
PGIMER, Chandigrah and
*AIIMS, New Delhi, India.