# 3. Drug-related deaths: trends, causes of death, drugs reported, sex and age

### 3.1 Overall numbers

3.1.1 Based on the definition used for these statistics, there were 526 drug-related deaths in 2013, 55 (9 per cent) fewer than in 2012. This was the fifth highest number recorded since the series of figures began in 1996, and was 209 (66 per cent) more than in 2003.

3.1.2 The figures in Table 1 show that the past ten years have had five rises and five falls in the number of drug-related deaths. However, the rises have tended to be greater than the falls, so the trend in the number of drug-related deaths has been upwards. Because the statistics show some year to year fluctuations, moving annual averages are likely to provide a better guide to the long-term trend than the change between any two individual years. Figure 1 illustrates this:

• the black dots show the figures for each year;
• the continuous grey lines show two moving annual averages - a 3-year average (thin grey line) and a 5-year average (thick grey line). The latter should provide a better indication of the overall long-term trend; and
• the broken grey lines show the likely range of random statistical variation around the 5-year moving average. Statistical theory suggests that, if the number of deaths can be represented as the result of a Poisson process, for which the underlying rate at which the events (deaths) occur is given by the 5-year moving average, then random year to year variation would result in only about one year in 20 having a figure outwith this range (which is a '95% confidence interval', calculated thus: the underlying rate of occurrence plus or minus 1.96 times its standard deviation; for a Poisson process, the standard deviation is the square root of the underlying rate of occurrence).

3.1.3 Looking at the chart, it is clear that, for many years, the individual years' figures tended to fluctuate around a long-term upward trend, and were generally within the likely range for random statistical year to year variation about the trend. It also appears that:

• the figure for 2008 was unusually high (being above the upper end of the likely range of random statistical variation around the 5-year moving average);
• the figure for 2009 was broadly in line with the long-term trend (being close to the 5-year moving average value for 2009);
• the figure for 2010 was unusually low, relative to the long-term trend (being below the lower end of the likely range of random statistical variation);
• the figures for 2011 and 2012 are broadly in line with the long-term trend: they are both not far from what one would expect the 5-year moving average to be, if the trend over the previous decade were extrapolated to those years; and
• the figure for 2013 is clearly below what would be expected for that year, based on the long-term trend over the previous decade.

Until the figure for 2013 was obtained, the rise in the 5-year moving average that had been seen for many years suggested that there was a clear long-term upward trend, and that the figure for 2010 had been unusually low relative to that long-term trend. However, the latest value of the 5-year moving average shows a slight fall (from an average of 554 for 2008 to 2012, to an average of 544 for 2009 to 2013), because there were fewer deaths in the year which has now entered the calculation (2013, with 526 deaths) than in the year which has now dropped out of the calculation (2008, with 574 deaths). Also, the pattern of rises and falls in recent years means that there has not been much change lately in the 3-year moving average (its latest five values are 525, 535, 538, 550 and 564, suggesting at most only a slight upward trend). Therefore, it can be argued the annual number of deaths may be 'levelling off' - and, if so, that the large increase between 2010 (485 deaths) and 2011 (584 deaths), and the drop between 2012 (581 deaths) and 2013 (526 deaths), are both year-to-year fluctuations around the fairly steady annual level that is suggested by the latest five values of the 3-year moving average.

### 3.2 Underlying causes of death

3.2.1 As explained in paragraph 2.6, National Records of Scotland (NRS) implemented WHO updates to the coding rules at the start of 2011. This changed the classification of the underlying cause of many drug-related deaths.However, NRS has estimated what the figures for 2011 onwards would have been, had the data been coded using the old rules.

3.2.2 Table 2 shows the number of drug-related deaths categorised by the underlying cause, defined in terms of groupings of the ICD codes. The final row gives the figures for 2013 that were produced by applying the new coding rules: the majority of drug-related deaths (365, or 69 per cent) were coded to 'accidental poisoning'. This covers the relevant categories within the ICD's section for 'Accidental poisoning by and exposure to noxious substances' (for example, it includes ICD-10 code X42 which is defined as 'Accidental poisoning by and exposure to narcotics and psychodysleptics [hallucinogens] not elsewhere classified'). Most of the other drug-related deaths in 2013 (88, or 17 per cent of the total) were counted as 'undetermined intent', which covers a number of ICD categories whose titles are along these lines: 'poisoning by and exposure to [name/type of substance], undetermined intent'.

3.2.3 Table 2 also provides NRS's estimates of the figures that would have been produced for 2011 onwards, had the old coding rules been used. On that basis, the underlying cause for the majority of drug-related deaths (359, or 68 per cent) would have been 'drug abuse', which covers the relevant categories within the ICD's section for 'Mental and behavioural disorders due to psychoactive substance use'.

3.2.4 Because some of the figures can fluctuate markedly from year-to-year, a better indication of the longer-term changes should be obtained from a comparison of the averages for 5-year periods. These show increases in deaths for which the underlying cause (on the basis of the old coding rules) was ‘drug abuse’ (from an average of 234 per year in 1999-2003 to an average of 370 in 2009-2013), ‘accidental poisoning’ (from an average of 15 to an average of 66), and ‘undetermined intent’ (from an average of 42 to an average of 66). There was not as much change in deaths caused by intentional self-poisoning (averages of 31 per year in 1999-2003 and 43 per year in 2009-2013).

### 3.3 Selected drugs reported

3.3.1 The NRS database records a wide range of drug combinations (e.g. in 2006, diazepam was mentioned in almost a fifth of the deaths for which heroin or morphine were reported; and heroin, morphine or methadone were mentioned in over half of the deaths for which cocaine was reported). A complete list of all the substances which were reported to NRS for every death from poisoning (including deaths which are not counted as 'drug-related' for the purpose of these statistics) can be found in Table 6.12 of the Vital Events Reference Tables, which are available on the NRS website. 'Unspecified drug(s)' is recorded in only a small proportion of drug-related deaths (on average, only a couple of per cent per year). Table 3, Table 6 and Table 7 give information on the frequency of reporting of selected drugs, whether alone or in combination with other substances. The drugs listed in these tables are reported in the majority of drug-related deaths (for example, not counting alcohol, at least one of them was reported in 91 per cent of the drug-related deaths in 2000, and also in 91 per cent of cases in 2013). The tables show a combined figure for 'heroin/morphine' because it is believed that, in the overwhelming majority of cases where morphine has been identified in post-mortem toxicological tests, its presence is a result of heroin use.

3.3.2 Since these tables record individual mentions of particular drugs, there will be multiple-counting of some deaths (e.g. if both heroin and diazepam were implicated in, or potentially contributed to, the cause of a death in 2013, that death will be counted in three of the 'drug' columns of Table 3: 'heroin/morphine', 'benzodiazepines' and 'diazepam'). Therefore, these tables do not give the numbers of deaths that are attributable to each of the drugs mentioned. When more than one drug was reported for a particular death, it may not be possible to deduce, from the information held in the NRS database, which (if any) of them was thought to be the (main) cause of the death, except to the extent that, for 2008 onwards, the database distinguishes between (a) drugs which were implicated in, or which potentially contributed to, the cause of death and (b) any other drugs which were present, but which were not considered to have had any direct contribution to the death. NRS's database has no information about the amounts of each drug that were found, or the possible consequences of taking particular combinations of drugs.

3.3.3 For 2008 onwards, the standard basis for figures for individual drugs is ‘drugs which were implicated in, or which potentially contributed to, the cause of death’ (further information about this is given in Section 2). Table 3 shows that heroin/morphine was implicated in, or potentially contributed to, the cause of 221 (42 per cent) of the 526 deaths in 2013; methadone was implicated in, or potentially contributed to, 216 (41 per cent); and benzodiazepines were implicated in, or potentially contributed to, 149 (28 per cent). Cocaine, ecstasy and amphetamines were implicated in, or potentially contributed to, 45, 17 and 27 deaths respectively. Alcohol was implicated in, or potentially contributed to, the cause of 103 of the 526 drug-related deaths in 2013.

3.3.4 Table 3 also shows that, in 2013, heroin and/or morphine were implicated in, or potentially contributed to, the same number of deaths as in 2012, but far fewer deaths than in 2008 (324 in 2008, 221 in 2012 and 221 also in 2013). Methadone was implicated in, or potentially contributed to, fewer deaths than in 2012, but more than in 2008 (169 in 2008, 237 in 2012 and 216 in 2013), its numbers having been highest in 2011 (275). There was also a fall in the number of deaths in which benzodiazepines were implicated, or to which they potentially contributed, as this figure returned to the same level as in 2008 (149 in 2008, 196 in 2012 and 149 in 2013). There was little change in the number of deaths for which cocaine was implicated, or to which it potentially contributed (45 in 2013; 30-something in each of the previous five years), and some large percentage year-to-year fluctuations in the relatively small numbers for ecstasy and amphetamines.

3.3.5 It is not possible to make a direct comparison with the figures for earlier years because there is a break in the series between 2007 and 2008, due to the revision of the questionnaire which collects information about the drugs found in the body (as explained in paragraphs 2.3 to 2.5). The statistics may also be affected by other differences, between years or between areas, in the reporting of drugs found in the body (examples of which are given in paragraph 2.8). Therefore, apparent changes in the numbers of deaths for which particular drugs were reported must be interpreted with caution, and with the knowledge that there is a clear break in the figures between 2007 and 2008. The change in the method of data collection may have contributed to the apparent large percentage increases, between 2007 and 2008, in the figures for methadone, benzodiazepines generally and diazepam specifically.

3.3.6 Because some of the figures can fluctuate markedly from year to year, the main changes over time are best identified by comparing the averages for 1996-2000 and 2003-2007 (the latter being the final 5-year period before the break in the series). These show that there were marked increases in the numbers of deaths for which there were reports of:

• heroin and/or morphine - from an average of 128 per year in 1996-2000 to an average of 229 in 2003-2007;
• cocaine - from an average of 6 to an average of 38; and
• alcohol - from an average of 91 to an average of 129.

There was not much change in the numbers of deaths for which there were reports of:

• methadone (averages of 74 and 90);
• diazepam (averages of 116 and 103); and
• ecstasy (averages of 7 and 13).

There was a marked fall in the number of deaths for which temazepam was reported (from an average of 47 per year in 1996-2000 to an average of 12 in 2003-2007).

3.3.7 However, while comparing 5-year averages should reduce the effect of year-to-year fluctuations, it will not necessarily give the full picture. In this case, it does not reveal some marked changes during the period:

• the number of deaths for which diazepam was reported rose from under 100 in 1996 and 1997 to over 200 in 2002 and then fell back to under 100 in 2005, 2006 and 2007; and
• the number of deaths for which methadone was reported appeared to fall in the late 1990s, but then rose to 114 in 2007 - above the level recorded in 1996 (100).

3.3.8 As mentioned in Section 2, NRS can also produce, for 2008 onwards, figures on the basis of 'all drugs which were found to be present in the body', including any other drugs which were present, but which were not considered to have had any direct contribution to the death. The lower half of Table 6 shows figures for 2013 on this basis. The main differences between the two halves of the table are in the figures for benzodiazepines (and diazepam in particular): benzodiazepines were found to be present in the body in the case of 377 of the drug-related deaths in 2013, but had been implicated in, or potentially contributed to, only 149 of those deaths (for diazepam, the equivalent figures are 316 and 107). There are also large percentage differences between the figures in the two halves of the table for cocaine (found present in 65 cases; implicated in, or potentially contributed to, 45 deaths), amphetamines (for which the numbers are 37 and 27, respectively) and alcohol (220 and 103). The figures for heroin/morphine and methadone do not differ much (in percentage terms) between the two halves of the table, these drugs were believed to be implicated in, or to have contributed to, the death in almost every case in which they were found.

3.3.9 Most drug-related deaths are of people who took more than one drug. In such cases, it may not be possible to say which particular drug caused the death. Table 7 shows the numbers of drug-related deaths for which only one drug was reported, which are the minimum numbers of deaths which may be wholly attributable to the specified drugs. The top half of the table shows that there were 46 deaths for which only one drug (and, perhaps, alcohol) was found to be present in the body: with a few possible exceptions (the footnote to the table gives further datails), these deaths will be wholly attributable to the specified drug (or, perhaps, to that drug in combination with alcohol). These numbers are all small, when compared to the total number of drug-related deaths: there were eight deaths for which the only drug reported was heroin/morphine; ten deaths for which only methadone was mentioned; three deaths for which only a benzodiazepine was reported; one death for which only cocaine was reported; no deaths for which only ecstasy was reported; and three deaths for which only amphetamines were reported. In total, there were 20 deaths for which alcohol was mentioned along with only one drug.

3.3.10 The lower half of Table 7 shows deaths for which only one drug (and, perhaps, alcohol) was implicated in, or potentially contributed to, the death. The numbers here are larger, because this part of the table includes deaths for which other drugs were mentioned as being present but were not considered to have had any direct contribution to the death. So, for example, the figures for methadone are the numbers of deaths for which only methadone (and, perhaps, alcohol) was implicated in, or potentially contributed to, the death - any other drugs (such as diazepam) which were found to be present in the body were not considered to have had any direct contribution to the death. There were 79 deaths for which heroin/morphine was the only drug which was believed to have been implicated in, or to have contributed to, the death; 56 deaths for which methadone was the only such drug; and 50 deaths for which alcohol was implicated in, or potentially contributed to, the cause of death, along with one drug. The numbers for each of the other drugs shown are all in single figures, so there were very few deaths which were believed to be due solely to one of those drugs alone.

3.3.11 In the lower half of Table 7, the sum of the figures for heroin/morphine, methadone, benzodiazepines, cocaine, ecstasy and amphetamines is 143, or 27 per cent of the total of 526 drug-related deaths in 2013. This means that one of these drugs was the only drug which was implicated in, or potentially contributed to, the cause of over a quarter of all drug-related deaths in 2013. There were also 50 deaths for which a drug which is not shown in the table was the only drug which was implicated in, or potentially contributed to, the cause of death. Information from NRS's database (which does not appear in any of the tables) shows that they included 18 cases where the only drug was dihydrocodeine; 7 cases where it was tramadol; 5 cases where it was oxycodone; 4 cases where it was codeine; and 2 cases where it was 'unspecified drug' (alcohol was also implicated in some of these deaths). Therefore, there was a total of 193 cases (37 per cent of all drug-related deaths) where only one drug (plus, perhaps, alcohol) was believed to have been implicated in, or potentially contributed to, the cause of death.

### 3.4 Sex and age

Table 4 shows that males accounted for the vast majority (393, or 75 per cent) of the drug-related deaths in 2013. This was the case throughout the past decade, although the precise balance between the sexes has varied from year to year. For example, between 2008 and 2013, the number of male drug-related deaths dropped (from 461 to 393) whereas the number of female deaths rose (from 113 to 133, having fallen back slightly from 165 in 2012) so the male percentage fell from 80 per cent to 75 per cent. Comparing the averages for 1999-2003 and 2009-2013, to reduce the effects of year-to-year fluctuations on the figures, the percentage increase in the number of drug-related deaths was greater for females (139 per cent) than for males (53 per cent).

3.4.1 In recent years, of the age-groups shown, the largest number of drug-related deaths have been among 25-34 and 35-44 year olds: using the averages for 2009- 2013, 166 out of 544 deaths (31 per cent) were of 25-34 year olds and slightly more were in the 35-44 age-group (188, or 35 per cent). In 2013, there were 184 drug-related deaths of people aged 35-44 (representing 35 per cent of that year’s total number of drug-related deaths) and 137 among 25-34 year olds (26 per cent of the total). In addition, 32 people aged under 25 died (6 per cent), as did 125 who were aged 45-54 (24 per cent) and 48 people aged 55 and over (9 per cent). The table shows that the number of deaths in a particular age-group can fluctuate markedly over the years (for example, the number of under 25s who died was 100 in 2002, 48 in 2005, 94 in 2007, 65 in 2010, and 32 in 2013). However, some clear trends can be seen. Comparing the averages for 1999-2003 and 2009-2013 (to reduce the effects of year-to-year fluctuations on the figures), there have been large percentage increases in the number of deaths of 35-44 year olds (from an average of 75 per year in 1999-2003 to an average of 188 in 2009-2013) and people aged 45-54 (from an average of 21 to an average of 98); the number of deaths of 25-34 year olds rose less markedly (from an average of 132 to an average of 166), as did deaths of people aged 55 and over (from an average of 11 to an average of 38); and there was a fall in the number of people aged under 25 who died (from an average of 85 to an average of 54).

3.4.2 Changes in the ages of drug-related deaths can also be seen from the values of the lower quartile age at death (a quarter of drug-related deaths were of people of this age or under), the median age at death (half the deaths were of people of this age or under) and the upper quartile age at death (a quarter of the deaths were of people of this age or older), which appear in the table:

• the lower quartile age at death rose from 22 years in 1996 to 32 years in 2013;
• the median age at death increased from 28 years in 1996 to 40 years in 2013; and
• the upper quartile age at death rose from 34 years in 1996 to 47 years in 2013.

The median is used (rather than the average) because it should be affected less by any unusually high (or low) values.

3.4.3 The lower part of Table 5 shows that, when the underlying cause of death is determined using the old coding rules, 280 (71 per cent) of the male deaths in 2013 were of known or suspected drug abusers compared to 79 (59 per cent) of the female deaths. Of the 48 deaths aged 55 and over, only 12 (25 per cent) were of people who were known, or suspected, to be drug-dependent. The table also provides a more detailed breakdown of the numbers by age-group for each sex.

3.4.4 Table 6 provides information about the ages and sexes of people who died having taken various drugs (perhaps more than one of the substances listed in the table, and maybe other drugs as well). The top half of the table provides figures on the standard basis: 'drugs which were implicated in, or potentially contributed to, the cause of death'. As mentioned earlier, men accounted for 75 per cent of all drugrelated deaths in 2013. Where the drugs listed below were implicated in, or potentially contributed to, the cause of death, men accounted for the following percentages of the deaths:

• cocaine - 91 per cent (41 out of 45);
• alcohol - 82 per cent (84 out of 103);
• heroin/morphine - 79 per cent (174 out of 221);
• benzodiazepines - 74 per cent (110 out of 149); and
• methadone - 69 per cent (150 out of 216).

There was one particularly marked difference between the distributions by age of people for whom heroin/morphine, methadone, benzodiazepines, cocaine or alcohol were implicated in, or potentially contributed to, the cause of their deaths. The under 25s accounted for 22 per cent of deaths in which cocaine was implicated, or to which it potentially contributed, compared with only 6 per cent of all drug-related deaths. The difference was even starker for the smaller numbers of 'ecstasy' deaths (59 per cent of them were under 25) and 'amphetamine' deaths (44 per cent under 25).

3.4.5 The lower part of Table 6 provides figures for all drugs which were found present in the body, including those which were not considered to have had any direct contribution to the death. Women accounted for 25 per cent of all drug-related deaths in 2013, but for only 21 per cent of the deaths for which heroin/morphine were found, only 20 per cent of deaths for which alchohol was found, and just 9 per cent of deaths for which cocaine was found. Again, the main differences between the distributions by age of those who died having taken the different drugs was that people aged under 25 accounted for 20 per cent of deaths following the use of cocaine compared with 6 per cent of all drug-related deaths; and for 58 per cent of deaths for which ecstasy was found and 35 per cent of deaths for which amphetamine was found.

3.4.6 The top half of Table 7 gives the numbers of deaths for which only one drug (and, perhaps, alcohol) was found to be present in the body. The numbers are all relatively small, so there is little that can be said about the ages and sexes of the people involved. The bottom half of the table shows deaths for which only one drug (and, perhaps, alcohol) was implicated in, or potentially contributed to, the death. Paragraph 3.3.10 explained why these numbers are larger. However, only for heroin/morphine (79 deaths) and methadone (56 deaths) are the figures large enough for analysis of the ages and sexes of the people involved. The main points to note are that females accounted for only 10 per cent (8 out of 79) of the deaths for which heroin/morphine (and, perhaps, alcohol) was the only drug which was implicated in, or potentially contributed to, the cause of death, and for 20 per cent (11 out of 56) of the deaths for which methadone (and, perhaps, alcohol) was the only drug which was implicated in, or potentially contributed to, the cause of death, compared with 25 per cent of all drug-related deaths in 2013. The distributions by age were broadly similar to that of all drug-related deaths.

3.4.7 Table 8 provides drug-related death rates per 1,000 population for a number of age-groups, and shows how these have changed, from 2000 to 2013. For most of that period, the drug-related death rate per 1,000 population was highest for people aged 25-34, and it averaged 0.25 over the latest five years (from 2009 to 2013). However, the rate for 35-44 year olds was higher in 2011 (0.29 per 1,000 population, compared with 0.27 for 25-34 year olds), 2012 (0.28 compared with 0.25), and 2013 (0.27 compared with 0.20), and had a latest 5-year average of 0.26. For both the 15-24 and 45-54 age-groups, the rate per 1,000 population has been much lower: for 15-24 year olds, it was 0.05 in 2013 and averaged 0.08 over the latest five years; for 45-54 year olds, it was 0.16 in 2013 with a latest 5-year average of 0.12. The rate for 55-64 year olds is no more than 0.06 per 1,000 population. Since 2000, there have been increases in the rates for all the agegroups apart from 15-24 year olds, whose rates have tended to decline (with some year-to-year fluctuations).

### 3.5 Death rates for problem drug users

3.5.1 The drug-related death rates per 1,000 population (shown in Table 8) are based on the size of the whole population of each age-group, the vast majority of whom do not use drugs. Therefore, those figures do not indicate the likely death rate for people who use drugs. Drug-related death rates for the part of the population whose put their lives at risk by using drugs can be calculated using the numbers of problem drug users (age 15-64) that are estimated by the Information Services Division (ISD) of NHS National Services Scotland. The latest such estimates, for the 2009/10 financial year, are available via www.drugmisuse.isdscotland.org website. For the purpose of ISD's estimates,'problem drug use' is defined as the problematic use of opiates (including illicit and prescribed methadone use) and/or the illicit use of benzodiazepines, and implies routine and prolonged use (as opposed to recreational and occasional use). It follows that ISD's estimates will be smaller than the total number of people who used illicit drugs at some time during the year.

3.5.2 Table 9 shows the annual average number of drug-related deaths for 2009-2013 and ISD's estimates of the number of problem drug users in 2009/10. The first two figures on the first row show that Scotland had 544 drug-related deaths (of all ages) per year (on average) between 2009 and 2013, and an estimated 59,600 problem drug users (aged 15-64) in 2009/10. Using those two figures gives an estimate of an annual average of 9.1 drug-related deaths per 1,000 problem drug users. The difference between the coverage of the two figures (‘all ages’ for deaths; ‘15-64’ for problem drug users) should not matter much, as there are very few drug-related deaths of people aged 0-14 or 65+.

3.5.3 Using ISD's estimates of the numbers of problem drug users by age and by sex in the same way, it appears that the annual average drug-death rate (per 1,000 problem drug users) is higher for males (9.6) than for females (8.2), and increases with age (4.9 for problem drug users who are aged 15-24, 7.2 for 25-34 year olds, and 12.5 for those aged 35-64). For males, the death rate clearly rises with age; for females, the figures 'suggest' a similar pattern 'but' that may not actually be the case, as ISD did not consider the estimated numbers of female problem drug users broken down by age to be sufficiently reliable for publication.

3.5.4 The ISD publication explains that the estimates are produced by combining data from a number of sources, and provides '95% confidence intervals' to indicate the likely margins of error in some of the figures. For the estimated total number of problem drug users for 2009/10, the 95% confidence interval is from 58,300 to 61,000 (or roughly +/- 2%). The values of the lower and upper ends of the confidence intervals can be used to calculate a likely range for the drug-related death rate. Dividing the annual average of 544 drug-related deaths by the value at the upper end (61,000 problem drug users) givers a minimum for the drug-death rate of 8.9 per 1,000 problem drug users; dividing by the value at the lower end (58,300 problem drug users) gives a maximum for the drug-death rate of 9.3 per 1,000 problem drug users.

3.5.5 ISD did not calculate 95% confidence intervals for its estimates of problem drug users broken down by age and sex, but one would expect them to be wider (in percentage terms) for the smaller sub-groups of the population (that is generally the case for the 95% confidence intervals for NHS Board and Council areas in Tables HB5 and C5).