Matthew Reid, 2009
Abstract
Injury to children in New Zealand is a major public health problem. It is a leading cause of death and hospitalisation in under five year olds. Injury surveillance has the potential to identify important causes and risk factors, and inform activities aimed at the prevention of injury.
Compared to what is known about fatal and hospitalised cases there is less information available about injuries serious enough to need medical attention but not requiring admission. Injuries presenting to emergency departments represent a potential source of information about some of the injury occurring to children in the community. This dissertation reviews the sources of information on injury in New Zealand and examines international examples of injury surveillance focused on children.
A retrospective audit was conducted at Christchurch Hospital by the Paediatric Department to collect information on children under five presenting to the Emergency Department over the period of one year in 2004/5 (the CPD dataset). This dataset was compared in this study to routinely collected data from hospital discharges for the same age group of children over the same period (the NZHIS dataset).
The aim was to examine the two methods of injury data collection to assess the quality and utility of the information generated. The two datasets were refined in order to focus on cases fitting an operational definition of injury, leaving a CPD dataset of 900 cases and a NZHIS dataset of 404 cases. Of these, 237 cases could be linked between the datasets.
An analysis was done of a subgroup of non-admitted cases in the CPD dataset comparing them to the admitted cases that constituted the NZHIS dataset, to establish whether the pattern of injury might be different between the datasets and to examine features of the datasets themselves and how well they produced useful information on injury.
Analysis of the datasets revealed that, compared to non-admitted cases in the CPD dataset, NZHIS cases did show a different pattern of injury, being less likely to have lacerations and limb injuries and more likely to have fractures, burns/scalds and poisoning, and for their mechanism of injury to be a fall greater than one metre, burn/corrosion and poisoning. There was a similar pattern of sex distribution between the datasets, and slightly younger cases in the NZHIS dataset. Ethnicity data was very incomplete in the non-admitted CPD dataset, which precluded analysis of differences for ethnic groups between the datasets. Deprivation, as measured by NZDep score (although assigned in different ways in each dataset), tended to be higher in the NZHIS dataset.
The overall reliability and utility of the CPD dataset were less than necessary to inform prevention activities. This study therefore concluded that, without first carefully considering alternative sources of injury data and the international experience of injury surveillance in emergency departments, and especially in the case of retrospective data extraction, research as conducted by the Christchurch Paediatric Department may not contribute useful information about childhood injury to inform injury prevention.