National Records of Scotland

Preserving the past, Recording the present, Informing the future

Probable Suicides and Accidental Deaths: how the statistics may have been affected by changes in the views of Procurators Fiscal on the nature of deaths

Probable Suicides and Accidental Deaths: how the statistics may have been affected by changes in the views of Procurators Fiscal on the nature of deaths

Since 2009, there has been an increase in the percentage of poisoning deaths which Procurators Fiscal(PFs) have described as being the result of accidents, and a corresponding fall in the proportion which they have described as being due to events of undetermined intent.

  • For 2009 and 2010, PFs were of the considered view that around a third of poisoning deaths were accidental, and roughly 30 per cent were due to events of undetermined intent;
  • These percentages changed over the next few years. By 2015, the proportion of poisoning deaths that PFs classified as accidental had risen, and, as a consequence, there had been a fall in the proportion that were classified as due to events of undetermined intent. Over the same period, there was little change in the number of poisoning deaths believed to be due to intentional self-harm or assault.

Because deaths due to events of undetermined intent are included in the number of probable suicides (as explained in the definition of those statistics), this has contributed to the reduction in the number of probable suicides. Had there been no variation in the proportion of poisonings counted as accidental:

  • recent years’ reductions in the numbers of probable suicides would have been smaller than are shown by the published figures (but there would still have been a decrease); and
  • the numbers of accidental deaths (based on the old coding rules) would have remained within the range that has contained the values for every year from 1995.

The remainder of this note provides some background, describes the estimated scale of the effect of the changes on the figures, and covers some other relevant points.

Background

With effect from mid-2009, as explained elsewhere on this website (on the Mid-2009 Change in the Procedure Used to Inform NRS About Suicides page), PFs should tell National Records of Scotland (NRS), for each traumatic or suspicious death, about the nature of the death, that is whether they believe that the death was due to (i) accident, (ii) intentional self-harm, (iii) assault or (iv) undetermined intent. The last category should be specified in cases 'where the evidence is insufficient for the PF to form a view, on the balance of probabilities, as to which of the other categories is appropriate'. For some deaths, of course, it is not possible for PFs to give NRS their views before the latter 'freezes' its statistical data for the relevant year.

At the start of 2011, NRS implemented new rules for coding the causes of death. How ‘drug abuse’ deaths from ‘acute intoxication’, and ‘alcohol intoxication’ deaths, are coded changed: in 2010 and earlier years, they were counted under ‘mental and behavioural disorders’; for 2011 onwards, they are classified under ‘poisoning’, so some of them will be counted as probable suicides, and some as accidental deaths. More information about this is available from the Death Certificates and Coding the Causes of Death section of this website. NRS has estimated how many of the deaths registered in 2011 and later years would have been counted under the old coding rules, in order that users of the statistics can see any underlying trends without the break in the series caused by the introduction of the new coding rules. (However, NRS has not estimated how many of the deaths which were registered in 2010 or earlier years would have been counted under the new coding rules, because NRS’s data for 2010 and earlier years were not coded in a way that would allow reliable figures on the basis of the new coding rules to be produced.)

Estimating roughly the scale of the resulting changes to the figures

NRS has estimated roughly what the figures for 2009 to 2014 (on the basis of the old coding rules) might have been, had those years’ 'accidental' and 'undetermined intent' poisoning deaths been split between those two natures of death in the same way as they were in 2015. On average, per year, there might have been about 50 more accidental deaths, and therefore around 50 fewer probable suicides, than shown in the published figures. The estimates for different years vary, but are all between roughly 20 and 90. It follows that:

  • the variation in the proportion of poisoning deaths counted as accidental appear not to have affected greatly the numbers of probable suicides (on the basis of the old coding rules) for 2009 to 2014: the estimated effect is a decrease of between two per cent and 11 per cent, depending upon the year.
  • the changes have definitely not altered the direction of the trend. For example, in that period, the highest numbers of probable suicides (on the basis of the old coding rules) were 781 in 2010, 772 in 2011 and 762 in 2012. Even if each of them was reduced by the largest estimate of the effect (which is for 2011, and is under 90), they would all still be larger than the figure of 656 for 2015. So, it would remain the case that the number of probable suicides (on the basis of the old coding rules) would be clearly lower in 2015 (and 2014) than in the previous five or more years. The latest year’s figure (656) would also be well below the level that was seen around the year 2000, when six consecutive years had values that were between 874 and 899. 

The estimated effects represent smaller percentages of the larger numbers of accidental deaths: using the figures on the old basis, and would have increased the total by between one per cent and seven per cent, depending upon the year. Even if the numbers of accidental deaths (on the basis of the old coding rules) for 2009 to 2014 were each increased by the largest estimate of the effect in any one year (under 90), they would all remain within the range of roughly 1,250 to 1,400 that has contained the values for every year from 1995.

Some other points to note

As well as changing over time, the balance between 'accident' and 'undetermined intent' in PFs’ views has varied between different parts of Scotland. As a result, the changes may have had different effects at different times on the figures for different areas.

NRS’s method of estimating the effect of the changes is a 'crude' one, intended solely to provide an indication of whether the scale of the effect might be large enough to alter the main messages from the statistics. NRS has no way of knowing which of (say) the 2009 deaths might have been classified differently, had those deaths been classified in 2015, and no intention of trying to produce estimates for, say, different areas of Scotland or different age-groups, because breaking the statistics down in such ways would mean that the numbers involved in the calculations would be much smaller and so the results could be far less reliable. For Scotland as a whole, the estimated scale of the effect of the change to figures based on the old coding rules is (on average) about 50 per year: breaking this down between several categories (for example parts of Scotland, or age-groups) could give results for individual categories that were small and subject to large percentage uncertainties.  

NRS has also estimated roughly what the figures for 2011 to 2014, on the basis of the new coding rules, might have been, had the percentages from 2015 applied in those years. The effects are larger (on average, about 100 per year), because figures based on the new coding rules include more 'undetermined intent' deaths. However, because NRS cannot produce figures based on the new coding rules for 2010 or earlier years, such estimates are of limited value, as they cannot be compared with earlier years’ numbers.

Finally, it should be emphasised that the figures for 2015 should be more accurate than those for the previous few years, because PFs informed NRS as to the likely intent behind, or accidental nature of, a higher percentage of poisoning deaths in 2015 than in the previous years.

All statistical publications