A1. The definition of a ‘drug-related death’ is not straightforward. Useful discussions on definitional problems may be found in articles in the Office for National Statistics publication "Population Trends" and in the journal "Drugs and Alcohol Today" (see the References). A report by the Advisory Council on the Misuse of Drugs (ACMD - see the References) considered current systems used in the United Kingdom to collect and analyse data on drug related deaths. In its report, the ACMD recommended that 'a short life technical working group should be brought together to reach agreement on a consistent coding framework to be used in future across England, Wales, Scotland and Northern Ireland'. GROS was represented on this group, and this paper presents information on drug-related deaths using the approach that was agreed, on the basis of the definition as it was implemented by GROS.
A2. The "baseline" definition for the UK Drugs Strategy covers the following cause of death categories (the relevant codes from the International Classification of Diseases, Tenth Revision [ICD10], are given in brackets):
a) deaths where the underlying cause of death has been coded to the following sub-categories of ‘mental and behavioural disorders due to psychoactive substance use’:
(i) opioids (F11);
(ii) cannabinoids (F12);
(iii) sedatives or hypnotics (F13);
(iv) cocaine (F14);
(v) other stimulants, including caffeine (F15);
(vi) hallucinogens (F16); and
(vii) multiple drug use and use of other psychoactive substances (F19).
b) deaths coded to the following categories and where a drug listed under the Misuse of Drugs Act (1971) was known to be present in the body at the time of death:
(i) accidental poisoning (X40 – X44);
(ii) intentional self-poisoning by drugs, medicaments and biological substances (X60 – X64);
(iii) assault by drugs, medicaments and biological substances (X85); and
(iv) event of undetermined intent, poisoning (Y10 – Y14).
A3. A number of categories of what may be regarded as "drug-related" deaths are excluded from the definition because the underlying cause of death was not coded to one of the ICD10 codes listed above. These include:
GROS also excluded from its implementation of the definition a small proportion of the deaths which were coded to one of the ICD10 codes listed in paragraph A2, specifically:
All three of these compound analgesics, particularly co-proxamol, have commonly been used in suicidal overdoses. As it is believed that dextropropoxyphene has rarely, if ever, been available other than as a constituent of a paracetamol compound, deaths caused by dextropropoxyphene have been excluded even if there is no mention of a compound analgesic or paracetamol. However, deaths for which codeine or dihydrocodeine were reported without any mention of paracetamol have been included, as these drugs are available on their own and are known to be abused in that form.
A4. From time to time, there may be minor discrepancies between the figures that were published previously and those which are produced henceforth. This is due to a change in the way in which "drug-related" deaths are identified using the data held by GROS. This process has two stages:
The new method was tested by using it to prepare figures for each year for 2000 to 2006, inclusive. The results were the same as, or within just 1-2 of, the figures which had been published previously. After examining the cases which were being counted differently by the old and the new methods, it was concluded that any flaws in the new method were not significant, and that it should be used henceforth. However, to avoid confusing users of these statistics, the tables which appeared in editions of this publication which were produced before the method was changed give figures for 2006 and earlier years which were extracted from the database produced by the old method, and so are as published previously. However, any subsequent new analyses of the data for 2000 onwards are likely to use the database produced by the new method, and so may include some totals or sub-totals (for the years from 2000 to 2006, inclusive) that differ slightly from the figures which were published previously, because the new method was used to produce the database of relevant cases for those years.