National Records of Scotland

Preserving the past, Recording the present, Informing the future

Likely Effect of the Procedural Change, Made in mid-2009, on the Total Number of Probable Suicides

Likely Effect of the Procedural Change, Made in mid-2009, on the Total Number of Probable Suicides

During 2009, there was a change to the way in which the Crown Office and Procurator Fiscal Service notifies the National Records of Scotland (NRS) of deaths which should be counted as being due to suicides. Further information about this change, and how it altered the balance between the numbers of deaths counted as being due to 'events of undetermined intent' and to 'intentional self-harm', is available on the page 'The change, made in mid-2009, in the procedure used by the Crown Office and Procurator Fiscal Service to inform NRS about suicides.'

This note considers the likely effect of the change on the overall total number of probable suicides. It finds that the reduction in the number of probable suicides between 2008 and 2009 is not large enough for one to be confident that it was caused by, or even mainly due to, the change of procedure, and that the same applies in the case of the increase between 2008 and 2009 in the number of deaths which are counted as being due to accidents. It appears that any effect of the change is no greater than the likely range of year-to-year fluctuation in the numbers.

As the number of deaths counted as being due to assault fell in 2009, the change in procedure has clearly not caused that figure to rise. When the numbers of deaths from specific causes are examined, the main rises in accidental deaths do not correlate well with the main falls in deaths from events of undetermined intent, which suggests that the change in procedure has not caused many deaths which would previously have been counted as probable suicides to be classified now as accidents. Therefore, most of the fall in the number of probable suicides (and most of the rise in the number of accidental deaths) appears to be due to year-to-year fluctuation (rather than the introduction of the new procedure).

Background

As a result of the change, some deaths which would previously have been counted as being due to events of undetermined intent will now be counted as being due to intentional self-harm. Because it is thought that most of the deaths which are classified as being due to events of undetermined intent are likely to be suicides, the number of probable suicides is defined as the total of the numbers of deaths due to intentional self-harm and events of undetermined intent. Therefore, the introduction of the new procedure should not have increased the overall total, because the resulting rise in the number of deaths counted as due to intentional self-harm will have been offset by a fall in those counted as due to events of undetermined intent.

However, the change could have reduced slightly the total number of deaths counted as probable suicides. This is because some deaths which would previously have been counted as being due to events of undetermined intent may now be counted as being due to accidents or assault. Before the change was made, it was expected that there would be few such deaths (since most of the deaths which were classified as being due to events of undetermined intent were thought likely to be suicides), this note looks at the figures, to see if that is so.

Number of probable suicides

In the statistics which are available on the Probable Suicides page, Table 1 and Chart 1 show that the overall trend (as indicated by the 5-year moving average) has been generally downwards since around 2000 (although the numbers have tended to fluctuate from year to year). Given that overall trend, a drop in the total number of probable suicides between 2008 and 2009 was not unexpected. The extent of the fall (97) appears to be within the likely range of values around the moving average, so it could be due to year-to-year fluctuation (like the fall of 105 between 2002 and 2003). In recent years, the annual total number of probable suicides has been around 800-or-so. While the 2009 figure of 746 is smaller than any of the previous six years' numbers (which are, in ascending order, 763, 765, 794, 835, 838, 843), it is not excessively low (as, with an underlying rate of 800 per year, the likely range of values in any given year is from about 743 to around 857). So the reduction in 2009 in the total number of probable suicides could be the result of year-to-year fluctuation - it is not large enough for one to be confident that it was caused by, or even mainly due to, the change in procedure.

Numbers of deaths from various types of external cause

As mentioned earlier, the change in procedure could have led to some of the deaths which would previously have been counted as due to events of undetermined intent now being counted as due to accidents and assault. Information about any trends in these figures can be obtained from the Vital Events Reference Tables. Statistics of deaths from 'external causes' from Vital Events Reference Table 6.1 have been put into a separate workbook called Suicides – estimating any likely effect of procedural change (Excel - 53 Kb), together with charts which show some of the figures for 2000 onwards (as that is the first year for which Scottish cause of death statistics are based on ICD-10 codes).

The first chart shows the numbers of deaths from the specified types of external cause. In general, the figures have not changed markedly since 2000, although there have been some year-to-year fluctuations.  

The remaining charts give the numbers of deaths from (a) accidents, (b) intentional self-harm, and (c) events of undetermined intent. These charts include 5-year moving averages, which should show more clearly if there are any overall trends. The moving averages are plotted against the middle years for the periods (e.g. the average for 2000 to 2004 appears against 2002). The last 5-year average shown is the one for 2004 to 2008 (plotted against 2006), because the 5-year average for 2005 to 2009 could be affected by the change of procedure in 2009.

Deaths due to assault

The first chart shows that the number of deaths which were due to assault is relatively small (around 100 per year). The statistics from Table 6.1 show a fall in deaths from assault in 2009, so the change in procedure has not caused any more deaths to be counted as due to assault. In any case, because NRS cross-checked its data on deaths from assault against information from the Scottish Government's statistics of homicides, previous years' figures for deaths due to events of undetermined intent should have included very few cases which PFs would now indicate were likely to be due to assault.

Deaths due to accidents

The second chart shows that the number of deaths due to accidents has generally been falling over the years since 2000. The 5-year moving averages suggest a steady downward trend, the grey dashed lines show the likely lower and upper limits of year-to-year fluctuation around the trend, and the thin dashed line shows a rough extension of the trend to the other years. There are two years for which the number of deaths due to accidents departs markedly from the general trend: 2004 (when there was an increase of 65) and 2009 (a rise of 71).  

The increase between 2008 and 2009 could be (mainly) due to the introduction of the new procedure: if so, then previous years' figures for ‘undetermined intent’ deaths (and, hence, total numbers of probable suicides) might have included many deaths that, under the new procedure, PFs have indicated are likely to be due to accidents - perhaps 60-or-so per year (based on the difference between the actual and ‘trend’ values for 2009). In that case, the change in procedure would account for most of the fall of 97 in the total number of probable suicides. However, the increase between 2008 and 2009 could also be an example of the kind of large year-to-year fluctuation that occurs from time to time (like the rise between 2003 and 2004). With an underlying rate of roughly 1,300 deaths due to accidents per year, the likely range of values in any given year is from about 1,228 to around 1,372, so an increase from 1,261 in 2008 to 1,332 in 2009 is well within the likely range of year-to-year fluctuation. Therefore, while the increase between 2008 and 2009 in the number of deaths due to accidents could be a consequence of the change in procedure, it is not large enough for one to be confident that it was caused by, or was even mainly due to, that change.

For completeness, the third and fourth charts show ‘intentional self-harm’ and ‘undetermined intent’ deaths. For the former, the overall trend is generally downwards, but there are year-to-year fluctuations, for the latter, there is no clear trend, with some large percentage changes in other years in addition to the big fall between 2008 and 2009. 

‘Undetermined intent’ and ‘accidental’ deaths from specific causes

A more detailed analysis uses the numbers of deaths for each specific cause (which can be obtained from the 2008 and 2009 versions of Vital Events Reference Table 6.4). The main reasons for the number of ‘undetermined intent’ deaths dropping by 96 between 2008 and 2009 were falls in deaths from the following causes (which are identified by their ICD-10 codes):

  • Y10 - poisoning by and exposure to nonopioid analgesics, antipyretics and antirheumatics, undetermined intent - fall of 16 (from 25 to 9).
  • Y11 - poisoning by and exposure to antiepileptic, sedative-hypnotic, antiparkinsonism and psychotropic drugs not elsewhere classified, undetermined intent - fall of 21 (from 47 to 26).
  • Y12 - poisoning by and exposure to narcotics and psychodysleptics (hallucinogens) not elsewhere classified, undetermined intent - fall of 35 (from 99 to 64).
  • Y21 - drowning and submersion, undetermined intent - fall of 18 (from 37 to 19).

No other ‘undetermined intent’ cause of death had a fall of five or more. Together, the above four causes of death accounted for 90 of the overall net fall of 96 in ‘undetermined intent’ deaths.

If the change in procedure had had a significant effect on the number of probable suicides, one would expect to see large increases in the corresponding numbers of ‘accident’ deaths. The changes between 2008 and 2009 in the figures for accidental deaths from these four causes are as follows:

  • X40 - accidental poisoning by and exposure to nonopioid analgesics, antipyretics and antirheumatics - no change (2 in both years).
  • X41 - accidental poisoning by and exposure to antiepileptic, sedative-hypnotic, antiparkinsonism and psychotropic drugs not elsewhere classified - rise of 17 (from 3 to 20).
  • X42 - accidental poisoning by and exposure to narcotics and psychoysleptics (hallucinogens) not elsewhere classified - fall of 1 (from 56 to 55).
  • W65 to W74 - accidental drowning and submersion (there are different codes for different types of location) - rise of 33 (from 7 to 40).

If the introduction of the new procedure changed the classification of some deaths from ‘undetermined intent’ to ‘accident’, each fall in the ‘undetermined intent’ figures should have a corresponding rise in the ‘accident’ figures (which might not be as large, since some of the deaths previously counted as ‘undetermined intent’ could now be counted as ‘intentional self-harm’ rather than ‘accident’). In fact, the changes for each of the specified causes are as follows:

  • Y10/X40 - ‘undetermined intent’ falls by 16, ‘accident’ does not change at all.
  • Y11/X41 - ‘undetermined intent’ falls by 21, ‘accident’ rises by 17.
  • Y12/X42 - ‘undetermined intent’ falls by 35, ‘accident’ falls by 1.
  • Y21/W65-74 - ‘undetermined intent’ falls by 18, ‘accident’ rises by 33.

So, only one of the four causes has a fall in ‘undetermined intent’ deaths that is consistent with the increase in ‘accident’ deaths. That most of the changes do not correlate well suggests that much of the increase in the number of accidental deaths is due to year-to-year fluctuation (rather than the introduction of the new procedure).

One can also look at the causes for which there were large increases in the number of accidental deaths between 2008 and 2009, which are as follows:

  • W00 to W19 - falls - up by 51 (from 634 to 685).
  • W65 to W74 - accidental drowning and submersion - up by 33 (from 7 to 40).
  • X41 - accidental poisoning by and exposure to antiepileptic, sedative-hypnotic, antiparkinsonism and psychotropic drugs not elsewhere classified - up by 17 (from 3 to 20).

Together, these three causes accounted for 101 of the overall net rise of 71 in 'accident' deaths (there were reductions for some of the other causes). Clearly, ‘falls’ was the biggest single cause of the rise in accidental deaths, accounting for about two-thirds of the net increase - but there was reduction of only 7 in the number of ‘undetermined intent’ deaths from ‘falling’ (codes Y30 and Y31: down from 30 to 23). Again, a poor correlation between the increases in ‘accident’ deaths and the decreases in ‘undetermined intent’ deaths suggests that most of the rise in the number of accidental deaths is due to year-to-year fluctuation (rather than the introduction of the new procedure).

All statistical publications