Further comments on findings

1. Could consistent differences between the countries be due to differences in coding rules or interpretations, data completeness or data analysis? The Republic of Ireland clearly differs from the other regions by continuing to use ICD9 coding until 2007. However, this may not particularly affect the current comparative analysis. See analysing trend data for further discussion. Possible coding differences have also been highlighted in relation to falls in Scotland and drowning in England. There also appear to be differences in the proportions recorded as undetermined intent. See also problems analysing intent in relation to the likely undercount of homicides in and over-count of undetermined cause in England and Wales, and the possible effect of differing legal approaches and attitudes on coding of suicide and undetermined cause. It is also interesting to note that Scotland has a significantly higher rate for intentional or undetermined poisoning and a significantly lower rate for unintentional poisoning compared to the other regions. The Northern Ireland count will also be less complete than the other countries due to particularly delayed registrations, with an estimated 5% of injury deaths excluded – see occurrences rather than registrations.

2. The Office for National Statistics paper ‘Trends in injury and poisoning mortality using the ICE on injury statistics matrix, England and Wales, 1979-2004’ available at
www.statistics.gov.uk/articles/hsq/hsq32-injury&poisoning.pdf
suggests on page 6 that some alteration to codes included in ICE categories might be helpful to gain a fuller picture for some injury categories, namely:

• For homicides: include ‘other specified events of undetermined intent’ (ICD10 code ‘Y33.9 and ICD9 code ‘E988.8) – see problems analysing intent

• For poisoning (currently split into unintentional poisoning and intentional or undetermined cause of poisoning): include deaths from ‘mental and behavioural disorders due to psychoactive substance use’ (ICD10 codes F10-F19). The following reason is given: ‘In England and Wales, over 90% of these deaths are in fact acute poisonings where the coroner has given a verdict of drug misuse or addiction, instead of one of accident, suicide or an open verdict. To leave out these deaths would underestimate mortality from poisoning; including them increases the number of deaths in this mechanism by …54% (in 2004)’

• For falls: include

i) deaths relating to fractures where the cause is unspecified (ICD10: underlying cause of X59 - unspecified accident, with a secondary case of S72 – fracture of femur and ICD9: E887 – fracture, cause unspecified). The following reason is given:

‘Previous analysis showed that these (fracture-related) codes in ICD10 give a good approximation to the E887 code in ICD9. Many deaths from falls simply state the resulting fracture on the death certificate and to leave these deaths out of the falls category would underestimate deaths from falls. Including them increases the number of deaths included in this mechanism by …67% (in 2004)’

ii) deaths from osteoporosis (ICD10: M80-M81). The following reason is given:
‘…it is possible, in both ICD9 in England and Wales and in ICD10 in most countries, for deaths from fractures following falls to be coded as due to osteoporosis. In 2004, over 80% of deaths from osteoporosis were stated to have involved a pathological fracture. (Including) osteoporosis deaths in our analysis (increases) the number of deaths included in this mechanism by …48% (in 2004). These deaths do not have an intent recorded, as they are not within the injury and poisoning chapter of the ICD, but they are most likely to be unintentional deaths.’

It would be useful to explore this in future IOBI analysis.