National Records of Scotland

Preserving the past, Recording the present, Informing the future

Accidental Deaths

Accidental Deaths

The Definition of the Statistics

The different parts of this page provide information on:

(a)

how accidental deaths are identified;

(b)

changes to the collection and coding of the data that were made in 2011 and 2009;

(c)

how the statistics for 1979 to 1999 were produced, and why they differ from previously - published figures for those years;

(d)

5-year moving annual averages, and the likely ranges of values around the moving annual average;

The International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10), which National Records of Scotland (NRS) uses to code the causes of deaths, has many different categories of ‘external’ causes of morbidity and mortality (i.e. causes which are ‘external’ to the person - e.g. accidents and acts of violence). ICD-10 groups these as follows: (i) accidents; (ii) intentional self-harm; (iii) assault; (iv) events of undetermined intent; (v) legal intervention and operations of war; (vi) complications of medical and surgical care; (vii) sequelae of external causes of morbidity and mortality; and (viii) supplementary factors related to causes of morbidity and mortality classified elsewhere.

The deaths which are counted, for the purpose of these statistics, as ‘accidental’ are the ones for which the underlying cause of the death was given one of the following ICD-10 codes:

  • V01 to X59 - accidents;
  • Y85 - sequelae of transport accidents; and
  • Y86 - sequelae of other accidents.

The cause of a death is classified, for statistical purposes, as 'accidental' in cases where NRS believes this to be appropriate, on the basis of what was recorded on the death certificate or of information from other official sources (such as the Crown Office and Procurator Fiscal Service [COPFS], a pathologist, or the doctor who certified the death). For example, COPFS may indicate that a death which it has investigated is believed to have been the result of an accident. However, if NRS has no information about (what is believed to be) any intent behind the act that caused a death, NRS will count it as an accident (rather than as an 'event of undetermined intent') if it was the result of (e.g.) a transport vehicle crash, a fall in a place that seems unlikely to be a venue for suicide (such as down stairs), a fracture for which no cause was specified but which seems likely to be due to a fall (in cases where the person was aged 75 or over), drowning while diving or having fallen from a boat, or hanging (in cases where the person was aged 15 or under).

In order to be of use for the purpose of statistical classification, information about any intent behind a death must reach NRS before it 'freezes' its statistical database for the calendar year, around the middle of the following year (e.g. NRS 'froze' its statistical database for 2011 at the end of May 2012). There will be deaths for which information that subsequently comes to light would have resulted in a different classification, had NRS been informed before it 'froze' the statistical database. NRS does not revise its statistical classification of deaths after it has 'frozen' its database, because that would change the figures whenever new evidence came to light, possibly many years later. These statistics will underestimate the true number of accidental deaths because, at the time that NRS 'froze' the data for the year, some deaths that were, in fact, accidental would have been classified as due to events of undetermined intent as NRS did not then know whether they were the result of, say, an accident, an assault or intentional self-harm.

The change to the coding of the data with effect from the start of 2011

At the start of 2011, NRS implemented a World Health Organisation update to the ICD, as a result of which ‘drug abuse’ deaths from ‘acute intoxication’ that would previously have been counted under ‘mental and behavioural disorders’ are now counted under ‘poisoning’ - so some of them will be counted as accidental deaths. Examples of these cases are people who were known or suspected habitual drug abusers, for whom the cause of death was certified as ‘adverse effects of heroin’, ‘methadone toxicity’ or ‘morphine intoxication’. Under the new coding rules, if NRS was informed that the overdose was believed to be accidental, such deaths are coded as being due to ‘accidental poisoning by …’, whereas under the old coding rules they would have been counted as being due to ‘mental and behavioural disorders due to use of ….’. Therefore, it is assumed that (for 2011 onwards) deaths which have both (i) the underlying cause coded as ‘X40-X49 Accidental poisoning …’ and (ii) a ‘drug abuse’ (F11 to F19) code used for one of the factors that contributed to the death would not have been counted as accidental under the old coding rules.

The change to the coding rules also affected deaths for which the cause was given as ‘acute alcohol intoxication’. Such deaths are now coded as being due to ‘accidental poisoning by alcohol’ whereas under the old coding rules they would have been counted as being due to ‘mental and behavioural disorders due to use of alcohol’. As there were hardly any deaths for which the underlying cause was coded ‘X45 – ‘accidental poisoning by alcohol’ in the years from 2000 to 2010, inclusive, it is assumed that (for 2011 onwards) deaths for which the underlying cause is coded ‘X45’ would not have been counted as accidental under the old coding rules.

A note on the changes to the way in which causes of death are coded is available on the Death Certificates and Coding Causes of Death page of our website.

The change to the collection of the data during 2009

During 2009, there was a change in the procedure used by COPFS to tell NRS which deaths should be counted as suicides, and this could also have affected the number of deaths which NRS counts as accidental. However, analysis of the numbers of deaths from various causes in 2008 and 2009 suggested that most of the rise, between the two years, in the number of accidental deaths was due to year-to-year fluctuation (rather than the introduction of the new procedure). Notes providing further information on this are available from our website via the following links:

(a)

on the change to the procedure - The Change, Made in mid-2009, in the Procedure Used by the Crown Office and Procurator Fiscal Service to Inform the National Records of Scotland (NRS) About Suicides;

(b)

on estimating the effects of the change (including analysis of the numbers of accidental deaths from various causes) - Likely Effect of the Procedural Change, Made in mid-2009, on the Total Number of Probable Suicides.

How the statistics for 1979 to 1999 were produced, and why they differ from previously-published figures for those years

NRS has used ICD-10 codes for the causes of all deaths that have been registered since the start of 2000. ICD-9 was used for deaths registered from 1979 to 1999. There are some marked differences between the structures of ICD-9 and ICD-10, and between the categories that have been included under the heading of ‘accidental deaths’ in the statistics that have been published at different times.

In order to provide a consistent series, the deaths which are counted as ‘accidental’ (for the purpose of the figures for 1979 to 1999 that are given here) are those for which the underlying cause had one of the following ICD-9 codes:

  • E800 to E869; and
  • E880 to E929.

These are the ICD-9 codes which cover the same categories of causes of death as are included in the ‘accidental deaths’ part of ICD-10. Therefore, using these ICD-9 codes gives figures which are on the same basis as the ones that NRS (and, previously, GROS) has produced for 2000 onwards.

Previously, other bases have been used to produce what were described, at the time, as statistics of accidental deaths for years in the period from 1979 to 1999. Those figures actually cover different categories of the causes of death. For example:

  • several tables in the Registrar General's Annual Report for 1999 gave figures for deaths from ‘accidents and adverse effects’, which covered ICD-9 codes E800-E949. They included all the causes of death that ICD-10 has under ‘complications of medical and surgical care’ (the ones whose ICD-9 codes are E870-E879 and E930-E949, and whose ICD-10 codes are Y40-Y84), so their coverage was much wider than that of the ‘accidental death’ figures for 2000 onwards.
  • Table 6.1 in the Registrar General's Annual Report for 2000 gave death figures for 1999 (and some earlier years) which were described as ‘accidents’ and which covered ICD-9 codes E800-E929. These included some of the causes of death that ICD-10 has under ‘complications of medical and surgical care’ (the ones whose ICD-9 codes are E870-E879, and whose ICD-10 codes are Y60-Y84), so their coverage was wider than that of the ‘accidental death’ figures for 2000 onwards. 

It follows that the statistics of ‘accidental deaths’ for 1999 and earlier years that were published previously were on a different basis from those that are given here.

5-year moving annual averages and likely ranges of values around the moving annual average

The number of accidental deaths can fluctuate markedly from year to year, particularly for the smaller Health Board and Council areas. Therefore, all the tables include 5-year moving annual averages, as these should provide a better indication of the overall long-term trend than the figures for the individual years.

As well as the figures for Scotland as a whole and the 5-year moving annual average, Chart 1 also shows the likely range of values around the moving annual average. This likely range of statistical variability in the figures is estimated by assuming that the numbers represent the outcome of a Poisson process, with the underlying rate of occurrence in each year being the same as the value of the 5-year moving annual average which is centred on that year. 'Upper' and 'lower' boundaries of an approximate '95% confidence interval' around the moving annual average are calculated by adding/subtracting twice the standard deviation. (For a Poisson distribution, the mean and the variance are the same, so the standard deviation is simply the square root of the moving average.)

In the years up to the mid-1990s, the long-term trend was clearly downwards, so the moving annual average fell fairly steadily. As a result, the moving annual average would usually be higher than the underlying rate of occurrence of the time, so many of the figures for individual years were around the lower boundary of the likely range of values. However, with effect from the mid-1990s (when the moving annual average ‘levelled off’), there appears to have been very little subsequent change in the underlying rate of occurrence of accidental deaths. As a result, in the period from 1995 to 2011 (inclusive), almost all the year-to-year fluctuations in the numbers have been within the likely range of values: only one of the 17 years had a figure which was outwith that range, which is consistent with what one would predict, based on statistical theory (as only about 5% of observations would be expected to fall outwith an approximate 95% confidence interval).

All statistical publications