The Scottish Government

« Previous | Contents | Next »

Listen

3. FINDINGS

3.1 Location of deaths to intentional self harm and undetermined categories

Overall 72% of suicide deaths during 1989-2002 were classified as intentional self harm. This average conceals massive variation by method (table 3.1) . Virtually all deaths by hanging were coded as intentional self harm, whereas under 40% of deaths by drowning and less than a third of deaths by 'other' methods were coded as intentional self harm. This finding is in line with previous research relating to an earlier time period in Scotland (Platt et al 1988) .

Table 3.1 Percent of deaths classified as intentional self harm or undetermined, by method, Scotland, 1989-2002

Table 3.2 shows that there was also variation in the classification of suicide deaths across different local authority areas, albeit on a smaller scale. The most anomalous area is Eilean Siar, where nearly two thirds of suicide deaths are classified as undetermined. The most likely explanation is the frequency of deaths by drowning (40 out of 83 suicides) and the difficulty of meeting the burden of proof necessary for the classification of such deaths as suicide. In fact, all but three of the 40 drowning deaths were assigned to the undetermined category.

Table 3.2 Percent of deaths classified as intentional self harm or undetermined, by local authority, Scotland, 1989-2002

3.2 National suicide trends over time (objective 1)

Male rates of death by suicide increased by 22 percent and female rates increased by 6 percent from 1989 to 2004 (single years) (table 3.3) . Among males the peak rate during the period under review occurred in 2002 (34.1 per 100,000 aged 15+ years) ; among females the peak rate was in 2001 (10.9) . It would be premature, however, to assume that 2001/2002 marks a turning point in terms of suicide incidence in Scotland. It should be noted that the rate in 2004 was higher than that in 2003 for both men and women.

Table 3.3 Suicide among people aged 15 and over, by sex and year, Scotland, 1989-2004

Figures 3.1 and 3.2 illustrate the temporal trends in national suicide incidence, based on crude rates and a three year moving average, respectively.

Figure 3.1 Crude suicide rates, by sex, Scotland, 1989-2004

Figure 3.2 Three year moving average of suicide rates centred on the middle year, by sex, Scotland, 1989-2004

3.3 Local suicide trends (three year moving average) (objective 1)

Annex 1 provides a full set of graphs illustrating trends in suicide incidence, based on a three year moving average over the period 1989-2004, for each local authority in Scotland. Annex 2 contains a similar set of graphs for each health board.

Examples of different patterns of three-year moving average suicide rates over the period 1989-2004 are shown in figures 3.3-3.9 (all examples refer to local authority areas) . There are two types of patterns, which also interact with each other. The first concerns the relationship between trends in a local area and in Scotland as a whole. In one local authority (Glasgow) the suicide rate was (significantly) higher than the Scottish average in all years among both men and women (see figures 3.3 and 3.4) . In a further three areas (Western Dunbartonshire, Dundee City and Highland) the male suicide rate was consistently (but not always significantly) higher than the Scottish average ( see annex 1) . In several other areas (including Dundee City (females) , Edinburgh City (females) and Eilean Siar (males) ) the suicide rate was mostly above the Scottish average ( see annex 1) . On the other hand, in six areas (Aberdeenshire (males) , E Dunbartonshire (males) , East Lothian (males) , East Renfrewshire (males and females) , North Lanarkshire (males) and South Lanarkshire (males) , the local suicide rate was consistently (but not always significantly) below the Scottish average. Trends in male suicide in North and South Lanarkshire are presented in figures 3.5 and 3.6, respectively. (Note the overlapping confidence intervals in some years. These indicate that we cannot be confident that local rates are significantly different from the rate for Scotland as a whole.) ( See annex 1 for graphs of trends in the other areas.) In several other areas (including East Lothian (females) , Fife (males) , Midlothian (males) and Perth & Kinross (males) ) the suicide rate was mostly below the Scottish average ( see annex 1) . Elsewhere there was less consistency in the relationship between suicide trends in the local area and national trends. Examples relating to Aberdeen City (females) and East Ayrshire (males) are shown in figures 3.7 and 3.8, respectively. (For other examples, see annex 1.)

The second pattern concerns the temporal trend in suicide incidence in the local area. In two areas (Angus (females) and Dundee City (females) ) there was a (fluctuating) downward trend in suicide over time. Figure 3.9 shows the trend among females in Dundee City. The next most common trend is fluctuating but upward. Figure 3.5 provides an example (North Lanarkshire (males) ) . However, the overwhelmingly predominant (typical) trend is highly fluctuating, irregular or stable ( i.e. no trend) . Figures 3.3 and 3.4 provide examples relating to Glasgow City. More extreme versions of fluctuating/irregular trends are shown in annex 1. The local areas with most irregular trends in suicide incidence tend to be those with the fewest deaths, as evidenced by very wide confidence intervals. In some areas the difference between highest and lowest rates over the period is very substantial indeed. The most extreme example is found in Shetland, where the difference is more than sixfold among men and more than fivefold among women (figures 3.10 and 3.11) .

Figure 3.3 Three year moving average of male suicide rates, Glasgow City, 1989-2004

Figure 3.4 Three year moving average of female suicide rates, Glasgow City, 1989-2004

Figure 3.5 Three year moving average of male suicide rates, North Lanarkshire, 1989-2004

Figure 3.6 Three year moving average of male suicide rates, South Lanarkshire, 1989-2004

Figure 3.7 Three year moving average of female suicide rates, Aberdeen City, 1989-2004

Figure 3.8 Three year moving average of male suicide rates, East Ayrshire, 1989-2004

Figure 3.9 Three year moving average of female suicide rates, Dundee City, 1989-2004

Figure 3.10 Three year moving average of male suicide rates, Shetland Islands, 1989-2004

Figure 3.11 Three year moving average of female suicide rates, Shetland Islands, 1989-2004

3.4 Suicide rates by social class (objective 2)

At the national level there was a marked variation in male suicide rates by social class in both 1989-95 and 1996-2002 (figure 3.12) . Differences between rates in the non-manual groups were not statistically significant. However, there were significant differences between rates in the non-manual groups and social class IIIM, between IIIM and IV, and between IV and V.

Charts showing variations in male suicide rates by social class in local authority and health board areas are presented in annexes 3 and 4, respectively.

In 1996-2002 there was an inverse linear relationship between social class and suicide (the lower the social class, the higher the rate) in 12 local authorities and an inverse trend in the other 20 local authorities. In 27 areas the relationship was more pronounced in 1996-2002 than in 1989-1995. In seven local authorities the suicide rate in classes IV and V combined was significantly higher than the rates in the other social classes. In 12 local authorities the suicide rate in classes IV and V was significantly higher than the rates in the non-manual social classes, while in 11 local authorities the suicide rate in classes IV and V was higher than the rate in classes I and II.

In 1996-2002 there was an inverse linear relationship between social class and suicide in six health boards and an inverse trend in nine other health boards. In all areas, with the exception of Fife, the relationship was more pronounced in 1996-2002 than in 1989-1995. In nine health boards the suicide rate in classes IV and V combined was significantly higher than the rates in the other social classes. In two health boards, the suicide rate in classes IV and V was significantly higher than the rates in the non-manual social classes, while in three health boards the suicide rate in classes IV and V was higher than the rate in classes I and II.

Figure 3.12 Suicide rates by social class, males, Scotland, 1989-95 and 1996-2002

3.5 Ratio of male: female age-adjusted rates (objective 2)

Suicide rates of males were consistently higher (approximately threefold across the whole country [see table 3.3]) than suicide rates of females across the period 1989-2002. Variation in the male: female suicide ratio across local authorities during 1989-2002 is shown in figure 3.13. The excess of male deaths was particularly marked in Eilean Siar, Clackmannanshire, West Dunbartonshire and Moray. However, estimates in these areas are not very reliable, as evidenced by the large confidence intervals. (To take an extreme example, we cannot state with certainty that the male: female ratio was significantly higher in Eilean Siar than any other local authority because of overlapping confidence intervals.) Some change in the ratio of male: female suicide ratio between the two halves of the time period is evident (figure 3.14) , but there is no consistent pattern and none reaches statistical significance.

Variation in the male: female suicide ratio across health boards during 1989-2002 is shown in figure 3.15. The excess of male deaths was particularly marked in the Western Isles, Highland, Argyll & Clyde, Grampian and Borders. Again, some change in the ratio of male: female suicide ratio between the two halves of the time period is evident (figure 3.16) , but there is no consistent pattern and none reaches statistical significance.

This analysis suggests a higher level of vulnerability to suicide among men in the more rural and remote areas of the country.

Figure 3.13 Ratio of male to female age-adjusted suicide rate by local authority, Scotland, 1989-2002

Figure 3.14 Ratio of male to female age-adjusted suicide rate by local authority, Scotland, 1989-95 and 1996-2002

Figure 3.15 Ratio of male to female age-adjusted suicide rate by health board, Scotland, 1989-2002

Figure 3.16 Ratio of male to female age-adjusted suicide rate by health board, Scotland, 1989-95 and 1996-2002

3.6 Methods of suicide

During the years 1989-2002, the most common methods of suicide among males were hanging (7.9 per 100,000) , self-poisoning (6.1) , drowning (3.1) and gassing (3.0) . Among females the most common method of suicide was self-poisoning, with a suicide death rate of 4.4 per 100,000. Figures 3.17 and 3.18 illustrate suicide rates by different methods in 1989-95 and 1996-2002 among men and women, respectively. Figures 3.19 and 3.20 present the proportions of suicide by different methods in the two time periods among men and women, respectively. Both sets of analyses reveal that hanging death rates have significantly increased over time for both men and women, while death rates by gassing (mainly carbon monoxide [car exhaust] poisoning) have significantly decreased.

Rates of suicide by different methods have been calculated for each local authority and health board area (annexes 5 and 6, respectively) , for 1989-95 and 1996-2002. In the majority of areas the rank ordering of methods and trends over time are similar to what has been reported for Scotland as a whole. The main difference is the greater popularity of drowning as a method of suicide in Highland and the islands, especially among males in Eilean Sear (see figure 3.21) . In 1996-2002 the male suicide rate by drowning was significantly higher than male suicide rate by hanging.

Figure 3.17 Suicide rate by method, males, Scotland, 1989-95 and 1996-2002

Figure 3.18 Suicide rate by method, females, Scotland, 1989-95 and 1996-2002

Figure 3.19 Proportion of suicides by method, males, Scotland, 1989-95 and 1996-2002

Figure 3.20 Proportion of suicides by method, females, Scotland, 1989-95 and 1996-2002

Figure 3.21 Suicide rate by method, males, Eilean Siar, 1989-95 and 1996-2002

3.7 Variation in suicide rates by local area (objective 1)

There was a substantial geographical variation in suicide rates. Suicide numbers and rates by sex over the period 1989-2002 are presented for each local authority (table 3.4) and health board (table 3.5) .

Across local authority areas (table 3.4) , the highest male suicide death rates occurred in Eilean Siar, Highland, Glasgow City and West Dunbartonshire (all over 40 per 100,000) . The lowest male death rates occurred in East Renfrewshire, East Dunbartonshire, East Lothian and South Lanarkshire (all under 25 per 100,000) . Among women, the highest rates occurred in Orkney Islands, Glasgow City, Dundee City, and Shetland Islands (all over 12 per 100,000) . The lowest death rates occurred in Clackmannanshire, East Renfrewshire, East Lothian and Eilean Siar (under 7 per 100,000) .

Table 3.4 Suicide among people aged 15 and over, by sex and local authority, 1989-2002

Across the health boards (table 3.5) , the highest male suicide death rates occurred in Western Isles and Highland, both over 40 per 100,000. High male suicide rates also occurred in Greater Glasgow (37.1) , Shetland (36.1) and Argyll and Clyde (34.5) . Lowest male suicide rates were found in Lanarkshire, Lothian and Fife (under 28 per 100,000) .

Among women, the highest rates were recorded in Orkney, Shetland, Greater Glasgow, Tayside and Highland (all over 10 per 100,000) and the lowest rates in Western Isles (under 7 per 100,000) .

Table 3.5 Suicide among people aged 15 and over, by sex and health board, 1989-2002

It should be noted that there were few suicide deaths in the island councils/health boards. However, rates were high as a result of the small population denominators. For example, in Western Isles, where the highest male death rate was recorded, the average number of deaths each year was 5.1. In Orkney, where the highest female death rate was recorded, the average number of deaths each year was 1.3.

Figure 3.22 shows standardised mortality ratios ( SMRs) by local authority area across the whole period for all persons, while figure 3.23 shows SMRs separately for males and females. All-person SMRs were significantly elevated (compared to Scotland as a whole) in West Dunbartonshire (also male SMR) , Highland (also male SMR) , Glasgow City (also male and female SMRs) , Eilean Siar (also male SMR) , Dundee City (also male and female SMRs) and Argyll & Bute (also male SMR) . All-person SMRs were significantly lower in West Lothian (also male SMR) , South Lanarkshire (also male SMR) , North Lanarkshire (also male SMR) , Fife (also male SMR) , Falkirk, East Renfrewshire (also male and female SMRs) , East Lothian (also male and female SMRs) , East Dunbartonshire (also male SMR) , Angus (also male SMR) and Aberdeenshire (also female SMR) . In Edinburgh female SMR was significantly elevated, while male SMR was significantly lower than expected. In Clackmannanshire female SMR was significantly lower than expected. Figures 3.24 and 3.25 show SMRs in 1989-95 and 1996-2002 for males and females, respectively, in each local authority. Little change is apparent over time in the relative status of areas compared to Scotland as a whole, apart from the island local authorities, where SMR estimates are markedly more imprecise due to the small numbers of deaths.

Figure 3.26 shows standardised mortality ratios ( SMRs) by health board across the whole period for all persons, while figure 3.27 shows SMRs separately for males and females. All-person SMRs were significantly elevated (compared to Scotland as a whole) in Western Isles (also male SMR) , Highland (also male SMR) and Greater Glasgow (also male and female SMRs) All-person SMRs were significantly lower in Lothian (also male SMR) , Lanarkshire (also male and female SMRs) , Grampian (also female SMR) , Forth Valley and Fife (also male SMR) . In Argyll & Clyde male SMR was significantly elevated. Figures 3.28 and 3.29 show SMRs in 1989-95 and 1996-2002 for males and females, respectively, in each health board area. As for local authorities, little change is apparent over time in the relative status of health boards compared to Scotland as a whole, apart from the island health boards, where SMR estimates have wide confidence intervals due to the small numbers of deaths.

Figure 3.22 Standardised mortality ratio of suicide by local authority area, males and females, 1989-2002

Figure 3.23 Standardised mortality ratio of suicide by local authority area and sex, 1989-2002

Figure 3.24 Standardised mortality ratio of suicide by health board area, males, 1989-95 and 1996-2002

Figure 3.25 Standardised mortality ratio of suicide by health board area, females, 1989-95 and 1996-2002

Figure 3.26 Standardised mortality ratio of suicide by health board, males and females, 1989-2002

Figure 3.27 Standardised mortality ratio of suicide by health board area and sex, 1989-2002

Figure 3.28 Standardised mortality ratio of suicide by health board area, males, 1989-95 and 1996-2002

Figure 3.29 Standardised mortality ratio of suicide by health board area, females, 1989-95 and 1996-2002

3.8 Suicide rates by age and sex (objective 2)

At a national level, over the period 1989-2002, the highest male suicide rate of 40.8 per 100,000 occurred in 25-34 year age group (table 3.6) . The rate in men aged 35-44 was second highest, at 36.0 per 100,000, and the rate in men aged 45-54 third highest, at 31.0 per 100,000. For women, the highest suicide rate of 11.6 per 100,000 occurred in 45-54 year age group while the second highest rate of 10.9 per 100,000 occurred in the 25-44 year age groups. The youngest age group (15-24) of men had a higher suicide rate than three old age groups of 55-64, 65-74 and 75 years and over. By contrast, the youngest age group (15-24) of women had a lower suicide rate than the three oldest age groups (55 years and over) . The excess of suicide deaths among males was particularly marked in the younger age groups (15-34 years) .

Table 3.6 Suicide among people aged 15 and over by sex and age, Scotland, 1989-2002

The age-related pattern of suicide in 1989-95 and 1996-2002 is shown in figures 3.30 and 3.31. Among men (figure 3.30) the inverse relationship with age was more marked in the later years, with significantly higher rates in the 15-44 age groups. Among women (figure 3.31) the pattern is less clear and the only significant difference between the two time periods was a higher rate among 15-24 year olds in 1996-2002.

Annexes 7 and 8 present suicide rates by age group in 1989-95 and 1996-2002 among men and women across local authorities ( annex 7) and health boards ( annex 8) . Broadly speaking, the age-related pattern found at national level is replicated at local level, although there are some anomalous patterns ( e.g. highest rate in the oldest age group in several areas) . There appears to be a trend towards a more pronounced inverse relationship between age group and suicide (highest rates in the youngest age groups) in 1996-2002 compared to the earlier period. Care needs to be taken, however, when making comparisons between local areas because of small numbers of deaths and wide confidence intervals around estimates of rates.

Figure 3.30 Suicide rates among men, by age, Scotland, 1989-95 and 1996-2002

Figure 3.31 Suicide rates among women, by age, Scotland, 1989-95 and 1996-2002

3.9 Area deprivation analysis

Figure 3.32 shows suicide SMRs for all persons aged 15+ years in 1989-1995 and 1996-2002, by population weighted deprivation quintile. There was a clear social gradient ('suicide gap') during both periods, whereby suicide increased with increasing levels of area deprivation. This gradient was steeper in the later period, indicating that the association between suicide and deprivation became more pronounced over time. Differences between the two time periods in SMRs were not statistically significant in the three least deprived quintiles. However, the SMRs in the two most deprived quintiles were significantly higher in 1996-2002. The same pattern of linear trend and widening gap over time between SMRs of least deprived and most deprived areas is found among both males (figure 3.33) and females (figure 3.34) , and in persons aged 15-44 years (figure 3.35) and 45+ years (figure 3.36) .

Figure 3.32 SMRs by population weighted deprivation quintile, all persons, Scotland, 1989-95 to 1996-2002

The widening suicide gap in Scotland, (1989-1995 to 1996-2002)
among all persons aged 15 years and older

Figure 3.33 SMRs by population weighted deprivation quintile, males, Scotland, 1989-95 to 1996-2002

The widening suicide gap in Scotland, (1989-1995 to 1996-2002)
among males aged 15 years and older

Figure 3.34 SMRs by population weighted deprivation quintile, females, Scotland, 1989-95 to 1996-2002

The widening suicide gap in Scotland, (1989-1995 to 1996-2002)
among females aged 15 years and older

Figure 3.35 SMRs by population weighted deprivation quintile, persons aged 15-44 years, Scotland, 1989-95 to 1996-2002

The widening suicide gap in Scotland, (1989-1995 to 1996-2002)
among all persons aged 15-44 years

Figure 3.36 SMRs by population weighted deprvation quintile, persons aged 45+ years, Scotland, 1989-95 to 1996-2002

The widening suicide gap in Scotland, (1989-1995 to 1996-2002)
among all persons aged 45 years and older

We formally calculated the 'relative gap' in suicide between the most and least deprived quintiles, for each period separately, by dividing the SMR for the most deprived quintile by the SMR for the least deprived quintile. The change in the relative gaps over time, calculated by dividing the relative gap in 1996-2002 by the relative gap in 1989-95, is referred to as the 'widening gap'.

Table 3.7 uses data shown in figures 3.32-3.36 to calculate the relative suicide gap between the suicide rate in deprivation quintile 5 and the suicide rate in deprivation quintile 1 in each time period. The 'widening gap' is the ratio of the relative gap in 1996-2002 to the relative gap in 1989-1995. To illustrate the calculation of the relative suicide gap we can give as an example the data relating to all persons, all ages (top two rows of table 3.7) . In 1989-1995 the suicide SMR in the most deprived quintile was 136.28, which was 2.16 times higher ('relative gap') than the SMR in the least deprived quintile (63.04) . In the later period the relative gap increased to 2.97 (165.73/55.73) . The widening gap was therefore 2.97/2.16, or 1.38. The magnitude of the widening gap was similar for men and women (table 3.7) . Although the relative gap was higher among people aged 15-44 years (2.95 in 1989-95 and 3.70 in 1996-2002) than among people aged 45+ years (1.44 and 2.26, respectively) , the widening gap was more pronounced in the older age group (1.57) than in the younger age group (1.25) .

Table 3.7 The widening suicide gap in Scotland, 1989-1995 to 1996-2002

In addition to examining trends in the relationship between deprivation and suicide at the national level, we have undertaken a similar (but not identical) analysis at local level using population weighted terciles (rather than quintiles) due to the small number of suicides in some areas. Annexes 9 and 10 present graphs relating to the widening suicide gap in local authorities and health boards, respectively. Tables 3.8 and 3.9 summarise the main findings. In 1989-95, evidence of a suicide gap (higher SMRs in tercile 3 compared to tercile 1) could be found in 28 out of 32 local authorities and 12 out of 15 health boards. In the later time period there was a suicide gap in 31 local authorities and 14 health boards. (It should be noted that not all differences were statistically significant due to small numbers of deaths (hence wide confidence intervals) .) A widening suicide gap over time (final column of each table) was found in 24 local authorities and 12 health boards.

Table 3.8 The widening suicide gap, 1989-95 to 1996-2002, by local authority

* No suicide deaths were recorded. Therefore SMR, 'suicide gap' and 'widening gap' could not be calculated.

Table 3.9 The widening suicide gap, 1989-95 to 1996-2002, by health board

* No suicide deaths were recorded. Therefore SMR, 'suicide gap' and 'widening gap' could not be calculated.

3.10 Relationship between area deprivation and social class (objective 5)

The relationship between area-level socioeconomic deprivation and individual-level social class position is explored in figure 3.37 (relating to 1989-95) and figure 3.38 (1996-2002) . In both time periods there are no significant differences in suicide rates between the non-manual classes in the different deprivation categories. There is a gradient, however, between the non-manual classes, on the one hand, and classes IIM, IV and V, on the other, in each deprivation category. Thus, the suicide rate is significantly higher in class V than in class IV (and all the other social classes) in all areas, irrespective of the degree of socio-economic deprivation in the areas in which people live.

In the earlier time period the patterning of social class differences does not differ markedly between categories of socioeconomic deprivation. This suggests a strong compositional effect and a very weak or non-existent area effect: that is to say, the main influence on suicide rates is at the individual, rather than area, level.

The situation changes in 1996-2002. Figure 3.38 suggests that there is an area effect as well as a compositional effect, as evidenced by the trend towards an increase in the social class gradient as the level of socioeconomic deprivation worsens: that is to say, the gap between suicide rates in the highest and lowest social classes increases as the level of socio-economic deprivation worsens. However, the compositional effect (the influence of individual social class) is undoubtedly far stronger than the area effect (the influence of the level of socio-economic deprivation in the locality) .

Figure 3.37 Suicide rates by deprivation quintile and social class, 1989-1995, Scotland, males

Figure 3.38 Suicide rates by deprivation quintile and social class, 1996-2002, Scotland, males

« Previous | Contents | Next »