see description for an explanation of this category.
Dataset Name:
PD400TI
Type of Dataset:
Cross-sectional
Description:
Estimated numbers of deaths assigned to suicide, 1997-2000.
Extracted from Chapter 3.6 of Annual Reference Volume DH4 no.25, 2000.
Suicides
As for homicides, accurate statistics on numbers of suicides occurring in a year, based on death registrations, are also difficult to produce, but the problems are rather different. Here they relate more to definitions of suicide than to delays in registration.
The procedures of referral of a possible suicide to the coroner and subsequent inquest are unlikely to result in further legal proceedings, and there should thus be less delay than in recording homicides. Devis and Rooney (see Reference 1 below) found that in 1994 the median time from occurrence to registration of a suicide (ICD9 code E950-E959) was 1.9 months, compared with the median of 2.6 months for all deaths from external causes. Nearly all suicides are registered within six months of their occurrence. For deaths from undetermined causes (E980-E989) the median was 2.7 months, but these include some deaths likely to be assigned to homicide.
Certification and registration of suicides
If a death was violent, or unnatural, or suspicious, the registrar is bound to refer the case to the coroner; such cases may also be referred directly by the police. The coroner has then a number of options, which could include holding a post-mortem and/or an inquest. If an inquest is necessary, the death can usually be registered only after the inquest has been adjourned or completed.
On completion of the inquest the registrar is duly informed of the results by the coroner (using Form 99(REV)). Based on a suicide verdict ONS will assign an underlying cause of death in the ICD9 code range E950-E959, while deaths given an open verdict will be assigned to the range E980-E989. For some years it has been considered that many of these open verdict cases are deaths where the death was self-inflicted but suicide was not proven, beyond a reasonable doubt. This assumption is now rather dated and should be reviewed.
Where the death resulted in an adjourned inquest and subsequent legal proceedings, the latter may reveal in a few cases that the death was not a homicide but instead an accident or misadventure, or a likely suicide or a case leading to an open verdict. ONS uses details on the notification sent by the coroner (Form 121) in coding these cases. Thus, from Table 27 in Annual Reference Volume DH4 no.25, 2000 there were 46 deaths coded to E988.8 with open verdict in 2000, and at the time of the late extract this number had increased to 47.
Late registrations
Dataset - PD499TG shows the effect of late registrations on suicide data for 1999. In the period between the standard and late extracts (September 2000 to April 2002) a further 102 deaths were coded to E950-E959, an increase of 2.8 per cent; for deaths coded to E980-E989 excluding E988.8 the increase was 89 (5.9 per cent). The overall increase was 191 (3.7 per cent), confirming that delays in registration add little to the numbers.
Numbers of suicides
Using the information above it is thus possible to make three separate estimates of the number of deaths annually from suicide:
(a) The number coded to ICD9 E950-E959 (suicide). This is the basic ICD9 classification, to which all definite suicide verdicts are assigned. Numbers classified in this way appear in Annual Reference Volume DH4 no.24, 1999 Tables 2-4, 6-8, 18-20, 23-24 and 26-28.
(b) The number coded to E950-E959, plus those coded to E980-E989 excluding E988.8. This takes account of deaths where an open inquest verdict was returned. It is simple to make this adjustment, using published figures. This classification was used in the DH4 volume up to 1996.
(c) The number coded to E950-E959, plus those coded to E980-E989 excluding E988.8, plus those coded to E988.8 with an open verdict. This allows for those deaths coded to E988.8 which were probably suicides, using information on inquest verdicts. It is now widely used in the DH4 volume, in Tables 2-4, 6-8, 18-20, 23-24 and 26-28.
The references to tables above concern numbers from the 'standard' annual deaths dataset. In addition, Tables 27 and 28 in Annual Reference Volume DH4 no.25, 2000 provide, between them, numbers classified in (a) and (b) for the late extract data described in dataset pd400tf.
The figures for suicides used in tables in Annual Reference Volume DH4 no.25, 2000 are summarised in this dataset. This illustrates the advantages of including the E980-E989 category to estimate suicide deaths, as well as of taking late extracts from the deaths database to make best use of all information which becomes available.
Reference 1. Devis T and Rooney C. The time taken to register a death. 'Population Trends' 88 (1997), pp 48-55.
See Product details