Decision Tree for Admission of Infants Under 6 Months of Age Presenting with Non-Specific Complaints

Background: Young infants are brought to the Emergency Department (ED) with a variety of complaints. Some of these complaints can be non-specific to the carers and the clinicians attending these infants. They are assessed thoroughly after arrival and sometimes have to undergo a variety of investigations for serious illnesses. When the serious diseases are confirmed or cannot be excluded in ED, the infants need to be admitted to the hospital for further management. This study was proposed to examine a prediction model for the young infants with nonspecific complaints.

Objective: This research project aimed to derive a prediction model for admission from the pertaining features of the young infants presenting to the ED.

Methodology: This was a retrospective observational study examining the clinical features and the disposition outcomes of young infants under 6 months presenting to a mixed ED in the 2013-14 period. The infants deemed to have non-specific complaints or diagnoses by the clinicians in the Emergency Department were included. With recursive partitioning statistical technique, clinical data of eligible patients extracted from electronic medical record were used to construct a decision tree models of risks for admission.

Results: The admission rate for these 968 infants studied was 33.6% (95% CI 30.59 – 36.55 %). Some risks factors associated with higher chance of admission were identified through statistical modelling. A decision tree incorporating the most important risk factors was constructed, that has a sensitivity of 85.96%, specificity 97.99% and a negative prediction value 93.01%. Its misclassification rate was only 6.15%.

Conclusion: Prediction models with excellent accuracy can be built on the features of young infants presenting with non-specific complaints. They could be used to guide decision making and as a bench marking measure for institutional comparisons.