Analysing trend data

When analysing trend data, it is important to bear in mind that coding practices can change over time. As far as the period 1996 to 2003 is concerned, there was a significant change to coding with the introduction of ICD10 in 2001. This was introduced in all the countries included in the IOBI analyses except for the Republic of Ireland who continued to use ICD9 until 2007.

Detailed information on the effect of ICD10 on England and Wales injury trends is provided in Health Statistics Quarterly 2003 No.19 available via www.statistics.gov.uk/downloads/theme_health/HSQ19.pdf .

Background information on the change from ICD9 to ICD10 in the Republic of Ireland can be found via
www.cso.ie/surveysandmethodologies/documents/word_docs/ICD_10_guide.doc

The specific codes recommended by the ‘International Collaborative Effort (ICE) on Injury Statistics’ should mean that the groupings used in this analysis are comparable between the two coding systems. However, the paper on the effect of ICD10 in England includes the comment that ‘it is not…possible to obtain an exact equivalent in ICD10 to the motor vehicle traffic accidents group in ICD9’. This has arisen because the ICD10 classification focuses on the person involved in the accident, whereas the ICD9 classification focussed on the nature of the accident and type of vehicle. The ICD10 classification also requires the word ‘motor’ or type of vehicle to be listed on the death certificate, otherwise the cause of death will be coded as ‘other land transport accident.’ Looking at the country trends, there do not appear to be particularly odd changes in 2001. However, coding issues appear to have affected Scotland’s ability to provide reliable comparative data for years prior to 2000 for this classification.

Our analysis also suggests that there was a change in the coding of falls in Scotland in 2000.

There are not enough years included to be able to undertake a statistical trend analyses for the 12 regions. An assessment of variations over time will rely on a visual inspection of the trend charts (noting that chart scales may magnify small differences) and the regional data.

However, we have undertaken a statistical test of significance of the change between 1996 and 2003 rates. Using 1996 as a baseline may appear rather arbitrary and the resulting measure could be misleading in relation to more recent or longer term trends. There is an element of correspondence, however, with the UK target in the white paper Saving Lives: Our Healthier Nation to reduce death rates from accidents by at least one fifth by 2010 from a baseline of the average of the years 1995-1997. The 3 year average helps to smooth out random variations and provides a more robust baseline for smaller populations. Unfortunately as explained earlier, 1995 data for England has only just been released to the Association of Public Health Observatories so could not be included in time for this analysis.