Article Text
PDF
Abstract
Background This study investigates the degree to which a previous hospitalisation for injury of any intent is a risk of subsequent youth suicide and whether this association is influenced by family socioeconomic status or economic stress.
Methods A nationwide register-based cohort study was conducted covering all Swedish subjects born between January 1977 and December 1991 (N=1 616 342, male/female ratio=1.05). The cohort subjects were followed-up from January 1998 to December 2003, when aged 7–26 years. Poisson regression and the likelihood ratio test (95% CI) were used to assess the age-adjusted effect of hospitalisation for injuries of various intent on youth suicide and its effect once adjusted for family sociodemographic and social circumstances.
Results Each set of exposures was associated independently and significantly with suicide mortality. Being hospitalised for self-inflicted injuries or injuries of undetermined intent was associated with a risk of suicide 36 to 47 times, respectively, that of subjects never hospitalised in the period under study (95% CI 28.36 to 45.58 and 26.67 to 83.87 for self-inflicted injuries and for events of undetermined intent, respectively; overall p<0.01). Similarly, previous events of unintentional injury markedly increased the risk of suicide (RR 3.08; 95% CI 2.26 to 4.19). These effects were solid and not substantially altered after adjustment for family demographic and socioeconomic circumstances.
Conclusion A strong association exists between previous hospitalisation for injury of any intent and youth suicide. The association is robust and unaltered by family socioeconomic circumstances.
- Youth suicide
- attempted suicide
- unintentional injuries
- socioeconomic status
- assault
- adolescent
- public health
- suicide
- violence
Statistics from Altmetric.com
- Youth suicide
- attempted suicide
- unintentional injuries
- socioeconomic status
- assault
- adolescent
- public health
- suicide
- violence
Introduction
Every 30 s, someone in the world, irrespective of age or social, ethnic or religious group, commits suicide.1
Suicide is globally one of the leading causes of death in males and females aged 10–24 years, representing a massive loss of young people.1 2 European countries are no exception, and rising trends in youth suicides, in particular among males, are a considerable public health concern.3–6
It is established that those who have attempted suicide are at greater risk of eventually dying by suicide.7
There is also a consensus among health professionals that the classification of injuries as ‘undetermined’ intent may actually mask a self-infliction, but is used in situations where it is difficult to assess the actual intent behind an injury. The use of ‘undetermined intent’ is particularly common among young people, while this diagnosis is used to varying degrees across countries and population strata.8 9 Furthermore, as for previous self-inflicted injuries, previous injuries of undetermined intent may also form part of the pathway to completed suicide.
Research also suggests that even previous injuries due to interpersonal violence may be linked to suicide,10–12 and so may injuries of unintentional intent.13 14 In the latter case, the mechanisms are uncertain and include the possibility of an induced pathology following the occurrence of an unintentional event, as would be the case, for instance, for a post-traumatic stress syndrome.15 16
Besides the increasing rates of youth suicide, their unequal socioeconomic distribution has received attention.17–21 Whether the association between previous injuries and youth suicide is affected by socioeconomic status remains to be established. Studies in this area are scarce, and there is a need to enlighten policy and practice on preventive strategies to combat post-injury morbidity and mortality.22 23
This study addresses this knowledge gap and investigates (1) whether an association between previous episodes of injuries of various kinds, including those attributed to interpersonal violence and undetermined intent and unintentional injuries, and youth suicide exists, and (2) the extent to which any association is influenced by family socioeconomic status.
Materials and methods
Study design
This register-based nationwide cohort study encompassed all individuals born between 1 January 1977 and 31 December 1991, found in the Swedish Total Population Register (N=1 616 342). The subjects were followed-up for the period January 1998 to December 2003, when aged 7–26 years. The unique personal identification number assigned to each permanent resident in Sweden enabled cross-linkage of nationwide registers. Information on unintentional and intentional injuries was gathered from the Hospital Discharge Register and the Cause Of Death Register. Demographic and socioeconomic information on the subjects was taken from the Swedish Population and Housing Census (parental country of birth, mother's age at birth of cohort member, socioeconomic position in the household), the Swedish Register of Education (parental level of education), and the Total Enumeration Income Survey (income, social benefit).
Injury diagnoses
We obtained information about hospital admissions requiring at least an overnight stay for intentional or unintentional injuries according to the International Classification of Diseases-10th Revision. We considered history of hospitalisation as a time-varying explanatory variable whose level of risk of suicidal behaviour could vary over time during the follow-up period. The subjects contributed with their person-time and placed themselves accordingly into the following categories sorted by increasing level of risk of suicidal behaviour: time with no hospitalisation, which was referred to as time at no risk of suicidal behaviour; time after hospitalisation for unintentional injury (V01–X59); time after hospitalisation for assault (X85–Y09); time after hospitalisation for self-inflicted injury (X60–X84); and time after hospitalisation for injury of undetermined intent (Y10–Y36). Each subject could pass from a no-risk condition to a higher level of risk, but not backwards, contributing in terms of person-years to each category for the time each cohort member stayed in a specific category.
All suicides (N=503) and deaths of undetermined intent (N=149) occurring during the follow-up were considered as the outcome. Previous studies have shown that missing data in Sweden's Hospital Discharge Register are between 1% and 2% and that the external code was missing in around 3% of cases. However, there are no reasons to believe that missing information or misclassification in the Hospital Discharge Register and Cause of Death Register are more common for some social groups than others.24–26
Demographic and socioeconomic variables
We obtained information about socioeconomic position from the Swedish Population and Housing Census of 1995.27 Each parent's socioeconomic position was defined in accordance with a classification used by Statistics Sweden, which is primarily based on occupation, but takes educational level of occupation, type of production, and position at work into account.28 The categorisation is based on the ‘dominance’ principle developed by Erikson.29 This principle is built on the concept of a dominance relationship between different work categories and their influence on the ideology, attitudes and consumption patterns of the family, and the category most important for the child's life prospects. Accordingly, each participant was allocated to one of four groups according to the highest-ranking parent: high/intermediate non-manual employees; self-employed professionals and farmers; assistant non-manual employees and manual workers; and others (eg, students, people on sickness and disability pensions, long-term unemployed). Similarly, the highest level of education achieved in the household was used to classify the parental level of education, classified as: higher education (university); upper-secondary education (10–12 years); basic education (9 years or less). Information on household disposable income (divided in quintiles) and social welfare benefits (yes or no at period of time from 1993 through 1997) was obtained from Sweden's Total Enumeration Income Survey, maintained by Statistics Sweden and comprising information about taxes and incomes of all Swedish residents.
Subjects in the highest socioeconomic position (high/intermediate non-manual employees), with the highest level of education (≥13 years of education), in the highest quintile of disposable income, and not receiving social benefit were used as reference groups.
Parental country of birth was classified into three categories. Cohort members with both or with at least one Swedish-born parent were considered as the reference group. If one or both parents were born in a Nordic country other than Sweden, the subjects were assigned to an intermediate group, ‘Nordic countries’. A third category grouped together all those with both parents born outside the Nordic countries. If one of the parents was born in Sweden and the other one in a Nordic country other than Sweden, the subject was assigned to the reference group.
Statistical analysis
Deaths codified as either suicides or of undetermined intent were combined and referred to as suicidal cases, as in previous studies.20 30
We estimated the effect of each single variable on the rate of suicide through Poisson regression with 95% CIs. Because of differences in rates of suicide in different age groups, all analyses were adjusted for age at the time of death.
A multivariate model was tested with history of injury hospitalisation as the independent variable, with adjustment for demographic and socioeconomic variables. We also tested the model with interactions between previous hospitalisation and each of the demographic/socioeconomic factors. As none of the interaction terms proved to be significant, they were removed from the model.
The likelihood ratio test was used to obtain the overall p value for each variable in the multivariate analyses, with relative risks and 95% CI based on the Wald test.
Data were processed with SAS V.9.1.3 for Windows.
Results
The rate of suicide was found to increase steeply with the age of adolescents and young adults, with the highest rate of 17.7 cases per 100 000 person-years in the population aged 22–26 (table 1).
Cases (N) and rates (per 100 000 person-years) of suicide and death from undetermined intent by age and method of death (cohort subjects=1 616 342)
Hanging was the most common method of fatal intentional self-harm, accounting for 43% of cases, while poisoning accounted for 66% of deaths of undetermined intent; when considered together, hanging was still the most common method of death, with a mortality of 2.3 per 100 000 person-years in the population aged 7–26.
Table 2 shows details of sociodemographic factors and method of suicide. A strong association was found between age and method of death, with a disproportionate number of hangings among young adolescents (p=0.01). Suicide by poisoning and by firearm were over-represented in females and males, respectively (p<0.01). Young people from families with low income (p=0.02) and social benefit recipients (p<0.01) resorted to poisoning more often than expected.
Distribution of method of death by suicide by demographic and socioeconomic factors among subjects aged 7–26 (n=652)
Table 3 presents estimates of RR of suicide by demographic, socioeconomic factors and previous hospitalisation for injury, both from bivariate and multivariate models.
Estimates of relative risk of suicide by demographic, socioeconomic factors and previous hospitalisation among subjects aged 7–26 (cohort subjects=1 616 342)
Young females had a risk of suicide that was 2.5 times lower than males (95% CI 0.34 to 0.48; p<0.01). In relation to the parental demographic factors, children of parents born in a Nordic country other than Sweden have a risk of suicide that is 2.29 times that of children with Swedish parents (95% CI 1.64 to 3.21; overall p<0.01). Children with a teenage mother were also found to have a higher risk of youth suicide than the reference group (RR 1.82; 95% CI 1.32 to 2.52; p<0.01).
When we analysed socioeconomic factors, having received a social benefit showed an association with a risk of suicide that was stronger (RR 2.11; 95% CI 1.80 to 2.47; p<0.01) than other common indicators, such as socioeconomic position and income. Moreover, the level of education of parents did not influence the overall risk of suicide among their children (overall p=0.15).
Subjects with a history of hospitalisation for any kind of injury, either intentional or unintentional, had a much higher risk of dying from a later suicide than those with no injury history.
Being hospitalised for self-inflicted injury or injury of undetermined intent was associated with a risk of suicide 36 to 47 times, respectively, that of subjects never hospitalised in the period under study (95% CI 28.36 to 45.58 and 26.67 to 83.87 for intentional self-harm and event of undetermined intent, respectively; overall p<0.01). Furthermore, not only a previous hospitalisation for assault (RR 9.43; 95% CI 5.04 to 17.67), but also a previous hospitalisation for unintentional injury (RR 3.08; 95% CI 2.26 to 4.19), markedly increased the risk of suicide compared with the no-injury group.
When adjusted for other sociodemographic covariates (multivariate model), the socioeconomic position lost its significance in increasing the risk of suicide (overall p=0.10), and the relative risks for other demographic factors such as parental country of birth and teenage mother decreased. Having received social benefit also remained strongly associated with later youth suicide in the multivariate model including sociodemographic factors.
Hospitalisations for both intentional and unintentional injuries were highly associated with the risk of future youth suicide, regardless of sociodemographic factors in the family. Having as a reference category subjects without previous hospitalisation, the risk of suicide was 40 times higher among subjects previously hospitalised for self-inflicted injuries (95% CI 31.09 to 51.47), and even higher among those sustaining lesions of undetermined intent (RR 43.30; 95% CI 24.36 to 76.96). Although to a lower extent than hospitalisations for intentional injuries, being hospitalised for strictly unintentional injuries, such as falls or road traffic crashes, was found to be significantly associated with the risk of youth suicide (RR 2.55; 95% CI 1.85 to 3.51).
Discussion
The study shows that, in addition to a previous hospitalisation for deliberate self-harm, which is the strongest known risk factor for suicide, even hospitalisation for injury with other intent—unintentional, interpersonal violence/assault and undetermined—is strongly associated with a subsequent completed suicide among young people. These effects are not only high but also very robust, as they are not markedly affected by demographic and socioeconomic factors.
These findings probably apply to most young people in Sweden, as Swedish registers are known for good coverage and high quality.28 Nonetheless, some methodological drawbacks need to be underlined. The first is the non-coverage of subjects living in institutions, such as orphans. These young people, few in number, are at particular risk of disadvantaged health outcomes,31 and are probably not well represented by the family sociodemographic attributes considered here. The study could not include injuries that did not lead to hospitalisation, the distribution of which across family demographic and socioeconomic attributes is uncertain, although unlikely to be biased to any socioeconomic group in the Swedish context.17 Swedish studies also suggest that misclassification of injuries—that is, from intentional to undetermined or unintentional—is not likely to reveal a strong socioeconomic bias.17
Further, it has been suggested that, with increasing age, socioeconomic status of destination (often measured in terms of education) is a better measure of socioeconomic status than socioeconomic status of origin (ie, family related). In some studies, it has indeed proved to have a stronger association with ill-health.32 For the cohort followed here, it has not been possible to find a measure of ‘destination’ that would have satisfactorily suited all subjects. We trust that the measure chosen gives a reliable picture of the socioeconomic position of the vast majority of the subjects, and we also know that parental socioeconomic position affects health-related behaviours and lifestyles throughout young adulthood.33 Furthermore, the use of income quintiles rather than absolute income is a means of overcoming problems of comparability over different periods of time or changes in the tax system.34
Finally, we aggregated suicide cases and deaths from undetermined intent as one single outcome, which may have introduced some heterogeneity into the outcome. Not only is this practice common in previous youth suicide studies,20 30 but also we observed very similar patterns for the two outcomes in the univariate analyses conducted on this material (data not shown).
The high risk of suicide among young people with previous unintentional injuries is interesting. One possible interpretation is that unintentional injuries may trigger psychopathological disorders of various forms and eventually exacerbate latent self-destructive behaviour.15 16 35 This finds support in earlier studies on injury repetition and indicates that subjects with previous trauma are more prone to new injuries,10 36 in particular, but not exclusively, after injuries due to interpersonal violence. Even if the association between previous unintentional injury and subsequent deliberate self-harm is not causal, the occurrence of unintentional injury may serve as a marker for the presence of underlying shared risk factors.13
An alternative, artefactual explanation is that injuries classified as unintentional may actually conceal a self-directed harm. On a related note, it has been posited that even injuries classified as assaults may mask deliberate self-harm episodes that were explained to hospital staff as being inflicted by others.13
An additional finding of importance is that the effects on youth suicide of previous hospitalisation for injury do not substantially change after adjustment for family socioeconomic status. In this cohort, the various socioeconomic attributes impacted inversely on the risk of suicide. We believe this can be explained in part by the fact that poisoning is the most common method of death in undetermined intent,8 and is over-represented among young people from households of low socioeconomic status, in the lowest income quintile and receiving social benefit.
As expected, youth suicide was higher in disadvantaged groups.19 37 In the Swedish context, social benefit, rather than being strongly related to the financial situation, can be considered as a proxy of accumulation of psychosocial and health problems in the family.19 38 It may therefore engender problem behaviours that are detrimental to health and well-being, such as alcohol misuse and interpersonal violence, and also lead to depressive moods and suicide ideation.
Our finding that young people living in a family with at least one parent born in a Nordic country other than Sweden are at higher risk than young people with Swedish-born parents, and those with a parent born outside the Nordic countries are at lower risk, is in line with the findings of an earlier study among Swedish children aged 5–17 years.30 It also finds support in a study of children with a mother of southern European or non-European origin, which was a protective factor against injury mortality compared with children with Swedish mothers.20 While the likelihood of selection mechanisms coming into play ought to be underlined,39 the underlying mechanisms most certainly need closer investigation.
Conclusion
This study indicates that youth suicide is strongly associated with previous injuries of any intent. The association is robust and unaltered by family socioeconomic circumstances.
What is already known about this subject
People who have attempted suicide are at greater risk of dying by suicide.
Young people who belong to low socioeconomic groups have a higher risk of committing suicide than those belonging to higher socioeconomic groups.
What this study adds
Subjects who are hospitalised for injuries of any intent have a higher risk of committing youth suicide than those without history of injury.
Previous hospitalisation for assaults and unintentional injuries—such as falls or road traffic crashes—increases the risk of suicide among young people by 9 and 3 times, respectively, compared with young people without history of injury.
The association between previous injury and later youth suicide is not influenced by family socioeconomic status.
Acknowledgments
We thank Dr Ugo Fedeli, Dr Giovanni Milan and Dr Francesco Avossa for their statistical advice and helpful comments.
References
Footnotes
Competing interests None.
Ethics approval The study was undertaken with the approval of the regional ethics committee in Stockholm, Sweden.
Provenance and peer review Not commissioned; externally peer reviewed.