Injury Indicators Update

Following an email to the Injury ICE Listserv, I received four responses regarding indicators work amongst members of the ICE Injury Indicators Group (ICEIInG). These were presented in Slide 2.

The response from John Langley (New Zealand) included a description of the work that Colin Cryer, John Langley and Shaun Stephenson (Injury Prevention Research Unit, University of Otago, New Zealand) had carried out, aimed at identifying national injury outcome indicators for the New Zealand Injury Prevention Strategy (NZIPS). This required the identification / development of indicators for all injury, as well as for the six priority areas shown on Slide 3. It was felt relevant to share this with ICE since it builds on previous work of ICEIInG.

Our approach was to:

identify current New Zealand national indicators, develop generic indicators for fatal and non-fatal injury

apply them to ‘all injury’ and the six priority areas validate each of the identified indicators using our ICEIInG criteria (developed at the Washington DC meeting in 1991)

retain only those for which there was no significant threat to validity

The Generic indicators we developed are shown in Slide 5. There are two relating to each of fatal and non-fatal serious injury. For each, one relates to annual average risk of injury death or non-fatal serious injury, the other relates to the public health burden as measured by the number of cases.

For the non-fatal indicators, the definition of serious injury was an ICISS (ICD-based Injury Severity Score) < 0.941. A description of ICISS and the meaning of this definition are described in Slides 6 and 7. Our intention was to choose a high severity threshold (in this case, an estimated probability of death of at least 5.9% for each case selected) so that the cases chosen would have a very high probability of admission to hospital. (The face validity of this threshold was confirmed through inspection of the diagnoses of cases captured using this definition.) The chosen source of non-fatal injury data was hospital discharge data. A high severity threshold would mean that service utilization effects on the indicators would be minimized.

Our indicators satisfied our six validation criteria. The trends since 1994 in the age-adjusted rates for fatal and for non-fatal injury were presented. They showed contradictory trends – rates of fatal injury declined, whereas rates of serious non-fatal injury increased. My a priori hypothesis was that the trends would be similar, due to the high threat-to-life threshold. The possible reasons for the differences were presented in Slide 10 and discussed.

Colin Cryer, June 2004.

This page last reviewed January 11, 2007