Clarifying the relationship between physical injuries and risk for suicide attempt in a Swedish national sample

The Interpersonal‐Psychological Theory of Suicide proposes that capability for suicide is acquired through exposure to painful and provocative events (PPEs). Although there is robust evidence for a positive association between aggregate measures of PPEs and risk for suicidal behavior, little is known about the contributions of physical injuries. The present study investigated the relationship between injuries and risk of subsequent suicide attempt (SA).


| INTRODUCTION
Suicide attempt (SA) is a significant public health concern.The annual costs attributable to non-fatal SA have been estimated at $5.2 billion in the United States, 1 and SA before age 20 is associated with an 18-and 11-year reduction in life expectancy among men and women, respectively. 2Clarifying the etiology of SA is critical to improve risk assessment and inform prevention and intervention efforts.
The Interpersonal-Psychological Theory of Suicide (IPT) 3,4 proposes that capability for suicide is acquired through physically painful and/or fear-inducing experiences, collectively referred to as painful and provocative events (PPEs).Through repeated exposure to PPEs, an individual may habituate to painful or fearful aspects of self-harm, increasing their likelihood of making an attempt when experiencing thoughts of suicide. 3,47][8][9] While prior research has considered the effects of some individual PPEs, such as non-suicidal self-injury [10][11][12][13] and physical and sexual abuse, [14][15][16] less is known about the relationship between physical injuries and suicidal behavior.Several studies have shown that individuals with a traumatic brain injury (TBI) [17][18][19] or spinal cord injury 20,21 are at higher risk for death by suicide than individuals in the general population.In addition, a recent case-crossover study

Significant outcomes
• Physical injuries are associated with increased risk for later suicide attempt, particularly in the first year following an injury.• The relationship between injuries and risk for suicide attempt is complex: The strength of associations differs based on injury type, sex, age, and genetic liability for suicide attempt.• Psychiatric and substance use disorders, genetic factors, and familial environmental factors contribute to, but do not fully account for, the relationship between injuries and suicide attempt, supporting a modest causal effect of injuries on suicide attempt risk.

Limitations
• Use of medical registries limits the identification of physical injuries and suicide attempt to those that require medical attention.• It is possible that some self-harm events were misclassified as unintentional physical injuries, which may inflate observed associations between injuries and suicide attempt.• Co-relative models only account for factors shared by members of a relative pair and do not adjust for unmeasured individual-level confounders.
investigated falls, transport-related injuries, TBIs, and injuries from personal assault as proximal triggers for self-harm, showing that the week following an injury was associated with a sixfold to eightfold increased risk of self-harm when compared to earlier weekly control periods. 22owever, a number of questions remain about the relationship between injuries and risk for SA.With the exception of TBI and spinal cord injury, injuries have not been investigated as distal risk factors for suicidal behavior.It is also important to examine potential sex differences in the relationship between injuries and SA, given that the prevalence of unintentional injuries is substantially higher in males than females, 23,24 and females are at higher risk for SA than males. 25Moreover, additional work is needed to evaluate the degree to which associations between injuries and risk for SA are causal or confounded.The IPT hypothesizes that PPE exposure is a causal risk factor for suicidal behavior, 3,4 but it is plausible that the relationship between injuries and SA may be partly driven by potential confounders, such as psychiatric and substance use disorders [26][27][28][29][30][31] or family-level factors (e.g., neighborhood characteristics, genetic factors).Finally, prior research has not investigated interactions between injuries and genetic liability for SA.Jointly considering the roles of genetic factors and injury exposure may offer additional insight into who is most at risk for suicidal behavior, thereby informing targeted prevention efforts.
To further clarify the relationship between injuries and risk for SA, we used nationwide Swedish registry data to test the longitudinal associations between 10 injury types-eye injury, fracture, dislocation/sprain/ strain, injury to the nerves and spinal cord, injury to blood vessels, intracranial injury, crushing injury, internal injury, traumatic amputation, and other or unspecified injuries-and risk for SA.We aimed to answer four research questions:

| Injuries
Injuries were determined from the Hospital Discharge Register, the Outpatient Care Register, and regional Primary Health Care research datasets.Table 1 shows the ICD codes used to identify 10 injury types: eye injury, fracture, dislocation/sprain/strain, injury to nerves and spinal cord, injury to blood vessels, intracranial injury, crushing injury, internal injury, traumatic amputation, and other or unspecified injuries.Injuries that occurred on the same day as a suicide attempt were excluded.For individuals who sustained multiple injuries within the same category (e.g., multiple fractures), only the first injury within that category was considered.

| Covariates
Mean years of parental education (standardized by sex and year of birth) and birth year were included as covariates.Analyses were stratified by sex.In sensitivity analyses, psychiatric and substance use disorder diagnoses, including alcohol use disorder (AUD), drug use disorder (DUD), affective disorders, psychotic disorders, and anxiety disorders were included as additional covariates.See the Supporting Information for details.

| Family genetic risk scores (FGRS SA )
FGRS for SA were calculated based on registrations for non-fatal SA among 1st-5th degree relatives.Scores accounted for the individual's genetic relatedness to each relative (e.g., 0.5 for parents and full siblings, 0.125 for first cousins), the relative's age at first registration with the outcome (when applicable), years of cohabitation, and differences in the number of relatives for whom information is available.Additional information about the calculation of FGRS can be found in Kendler et al. 33 2.3 | Statistical methods

| Univariable and multivariable models
To investigate the relationship between each injury type and risk for SA, we first performed a series of Cox models with each injury included in a separate model.To test for non-proportionality, we included an extra effect of the injury that started 1 year after injury exposure.If this term was significant, it suggested that the effect of injury on SA was different within the first year following the injury versus the remainder of the follow-up time.In the case of a significant term, we present separate hazard ratios (HRs) for the time period within 1 year of the injury and the time period more than 1 year after the injury.If the term was not significant, we report one HR for the association between injury and SA.Next, we conducted a multivariable analysis with all injury types entered in the same model as time-varying covariates.

| Co-relative models
We performed co-relative analyses to further investigate the nature of the relationship between each injury type and SA.The co-relative model, an extension of the co-twin control design, 34,35 evaluates whether differences between relatives in their injury exposure predict differences in SA status, controlling for genetic and environmental factors shared by members of the relative pair.For these analyses, we identified all cousins, half-siblings, and full siblings born within 10 years of one another from the Swedish Multi-Generation Register.In addition, we identified monozygotic (MZ) twins from the Swedish Twin Registry.Additional details can be found in the Supporting Information.

| Tests for gene-environment interaction
For injury types with a putative causal effect on SA, inferred by a HR significantly different from 1 in corelative models with MZ twins, we tested the additive interaction effect of injury type and FGRS SA on risk for SA by calculating the relative excess risk due to interaction (RERI). 36To do this, we included FGRS SA in the Cox regression models alongside an interaction term between FGRS SA and the time-varying injury variable.

| Secondary analyses
We conducted univariable models, multivariable models, and tests for gene-environment interaction further stratified by age group (young people = 10-24 years, adults = 25+ years, as defined by the World Health Organization 37 ) to explore potential age-based differences in the pattern of associations.

| Sensitivity analyses
To address the possibility of confounding by psychiatric and substance use disorders, we repeated the analyses described above with registrations for AUD, DUD, affective disorders, psychotic disorders, and anxiety disorders included as additional time-varying covariates.Analyses were performed using SAS 9.4 (©2002-2012 SAS Institute Inc., Cary, NC, USA).

| Descriptive statistics
Table 2 shows frequencies for injuries, SA, and covariates separately by sex.Tetrachoric correlations among injury types are provided in Table S1.

| Univariable models
We used Cox models to evaluate the relationship between injuries and risk for SA, with each injury type entered into a separate model.For all injury types, except injury to blood vessels among females, the association with SA was non-proportional.Therefore, we present two HRs: one for the period within 1 year of the injury and one for the period more than 1 year after the injury.
As shown in Figure 1 and Table S2, all 10 injury types were associated with elevated risk for SA.Effect sizes were larger within 1 year of the injury (HRs = 1.8-7.0).Nonetheless, all injury types exhibited a sustained relationship with SA more than 1 year after injury exposure (HRs = 1.4-2.6),except for traumatic amputation in females.
When psychiatric and substance use disorders were included as covariates, the associations between injuries and risk for SA were attenuated but remained statistically significant, with two exceptions: Eye injury was no longer related to risk for SA more than 1 year after injury exposure in females, and the association between injury to blood vessels and SA in females was non-significant (Figure 1, Table S3).

| Multivariable models
Next, we performed a multivariable analysis with all injury types entered in the same model.Effect sizes were somewhat attenuated after accounting for exposure to other injuries, but HRs remained significantly above 1 in all but one case (Figure 2, Table S4).
Including psychiatric and substance use disorders as covariates further reduced HRs for the association of each injury type with risk for SA (Figure 2, Table S5).However, the first year following injury exposure was still related to higher risk for SA across injury types.Associations between most, but not all, injury types and SA persisted more than 1 year after injury exposure.

| Co-relative models
Complete results are provided in Tables S6 and S7 for females and males, respectively.We focus on results from the predicted model in the main text (see the Supporting Information for details on competing models).Parameter estimates are presented in Figures 3 and 4 for females and males, respectively.The results varied by injury type but were generally consistent across sex.We observed three patterns of associations.First, for eye injury, HRs did not substantially decline across relative pairs of increasing genetic relatedness, which is consistent with a potential causal effect of eye injury on risk for SA.Second, for fracture, dislocation/sprain/strain, intracranial injury, and other and unspecified injuries, HRs declined across relative pairs of increasing genetic relatedness but remained significantly greater than 1 in MZ twins, consistent with partial confounding by familial factors and a residual causal pathway.Finally, for the remaining injury types, HRs were not significantly different from 1 in MZ twins.This is consistent with confounding by familial factors but may also reflect low statistical power, as these injuries were each experienced by <1% of the sample.
Results from co-relative models adjusted for psychiatric and substance use disorders are presented in Figure 3 and Table S8 for females and in Figure 4 and Table S9 for males.Findings did not change substantively, except that eye injury was not significantly associated with risk for SA in male MZ twins.

| Tests for gene-environment interaction
We estimated additive interaction effects between FGRS SA and injuries with a putative causal effect on SA (i.e., eye injury, fracture, dislocation/sprain/strain, intracranial injury, and other and unspecified injuries).Results are presented in Table 3.
In females, there was evidence to support a positive additive interaction with FGRS SA for fracture, dislocation/sprain/strain, intracranial injury, and other and unspecified injuries (RERI = 0.16-0.31).In males, we observed significant positive additive interactions across injury types (RERI = 0.13-0.33).The largest interaction effect was with intracranial injury across sexes.We repeated these tests for gene-environment interaction with further correction for psychiatric and substance use disorder registrations.As observed in the primary model, there was a positive additive interaction between FGRS SA and intracranial injury in females.All other interactions Results from univariable models of the relationship between injuries and risk for suicide attempt.Results are presented as hazard ratios with 95% confidence intervals.Confidence intervals were wide in some cases, and error bars were permitted to exceed the plot margins.Statistically significant parameter estimates are shown as solid circles, and non-significant parameter estimates are shown as empty circles.Models were stratified by sex.For all injuries, except injury to blood vessels among females, the association with suicide attempt was non-proportional.Hence, we present two hazard ratios: one for the period within 1 year of injury exposure and one for the period more than 1 year after injury exposure.For injury to blood vessels among females, we only present one hazard ratio (labeled "Overall" in the Figure ).In the primary models, birth year and parental education were included as covariates.The adjusted models further corrected for alcohol use disorder, drug use disorder, affective disorder, psychotic disorder, and anxiety disorder registrations.blood vessels, injury to blood vessels; CI, confidence interval; crushing, crushing injury; dislocation, dislocation/sprain/strain; internal, internal injury; intracranial, intracranial injury; nerves/spinal cord, injury to nerves and spinal cord; other injuries, other and unspecified injuries; traumatic amp, traumatic amputation.
became non-significant or changed direction after controlling for comorbid psychopathology (Table 3).

| Secondary analyses
Finally, we conducted univariable models, multivariable models, and tests for gene-environment interaction further stratified by age group (10-24 vs. 25+ years).Findings from the univariable models are presented in Figure 5 and Table S10.Overall, the relationship between injuries and risk for SA was similar across age groups.Several injury types (e.g., fracture, injury to nerves and spinal cord, intracranial injury) appeared to be more strongly associated with risk for SA in individuals aged 25+ years than in individuals aged 10-24 years, though 95% confidence intervals (CIs) overlapped in most cases.Sensitivity analyses adjusted for psychiatric and substance use disorders yielded a similar pattern of results (Figure 5, Table S11).
In a multivariable analysis stratified by sex and age group, HRs were attenuated but remained statistically significant in most cases (Figure 6, Table S12).HRs were further reduced after adjusting for psychiatric and substance use disorder registrations, and several injury types were no longer related to risk for SA more than 1 year after injury exposure in the 25+ age group (Figure 6, Table S13).
Finally, we conducted tests for gene-environment interaction within each age group for injury types with a putative causal effect in the primary co-relative analyses.In females, there was evidence to support a positive additive interaction with FGRS for fracture, dislocation/sprain/ strain, intracranial injury, and other and unspecified injuries across age groups (RERI = 0.10-0.35).In males, we observed significant positive additive interactions across injury types and age groups (RERI = 0.08-0.44),except for eye injury in the 25+ age group (Table 3).
In sensitivity analyses adjusted for psychiatric and substance use disorders, additive interaction effects remained positive and statistically significant for fracture, dislocation/sprain/strain, and other and unspecified injuries among females in the 10-24 age group (RERI = 0.14-0.28),and for all injury types among males in the 10-24 age group (RERI = 0.11-0.31).Therefore, young people with higher FGRS for SA were generally more sensitive to the effects of injury exposure than young people with lower FGRS.In contrast, negative and statistically significant F I G U R E 2 Results from multivariable models of the relationship between injuries and risk for suicide attempt.Results are presented as hazard ratios with 95% confidence intervals.Confidence intervals were wide in some cases, and error bars were permitted to exceed the plot margins.Statistically significant parameter estimates are shown as solid circles, and non-significant parameter estimates are shown as empty circles.Models were stratified by sex.For all injuries, except injury to blood vessels among females, the association with suicide attempt was non-proportional.Hence, we present two hazard ratios: one for the period within 1 year of injury exposure and one for the period more than 1 year after injury exposure.For injury to blood vessels among females, we only present one hazard ratio (labeled "Overall" in the Figure ).In the primary model, birth year and parental education were included as covariates.The adjusted model further corrected for alcohol use disorder, drug use disorder, affective disorder, psychotic disorder, and anxiety disorder registrations.blood vessels, injury to blood vessels; CI, confidence interval; crushing, crushing injury; dislocation, dislocation/sprain/strain; internal, internal injury; intracranial, intracranial injury; nerves/spinal cord, injury to nerves and spinal cord; other injuries, other and unspecified injuries; traumatic amp, traumatic amputation.interaction effects were observed across sex and injury types in the 25+ age group (RERI = À0.07 to À0.15).The only exception was a non-significant interaction with intracranial injury in females.These findings suggest that, in adults, the synergistic effects of FGRS SA and injury exposure were less than expected based on their main effects (Table 3).

| DISCUSSION
Leading theories of suicidal behavior highlight PPEs as a key risk factor for SA: Exposure to PPEs is thought to increase pain tolerance and lower fear of death, increasing the likelihood that an individual will attempt suicide when experiencing suicidal ideation. 3,4However, few studies have focused on physical injuries as distal risk factors for SA.In the current study, we investigated the relationship between 10 injury types and risk for SA in a Swedish birth cohort of more than 2 million individuals.We focus on five key takeaways from this study.First, at the population level, all 10 injury types were associated with elevated risk for SA: Individuals who had experienced a physical injury were between 1.2 and 7.0 times more likely to later attempt suicide when compared to individuals who had not experienced an injury.Associations were stronger within the first year of injury exposure, which is consistent with prior evidence to support injury as a proximal trigger for self-harm 22 and suggests that the period immediately following an injury may be the most important to target in prevention and intervention efforts.However, nearly all HRs remained significantly above 1 more than 1 year after injury exposure.
F I G U R E 3 Results from co-relative analyses of the relationship between injuries and suicide attempt in female individuals.Hazard ratios and 95% confidence intervals are presented for the predicted estimates within each relative type.Confidence intervals were quite wide in some cases, and error bars were permitted to exceed the plot margins.Complete results are available in the Supporting Information.For all injuries, except injury to blood vessels, the association with suicide attempt was non-proportional.Hence, we present two hazard ratios: one for the period within 1 year of injury exposure and one for the period more than 1 year after injury exposure.For injury to blood vessels, we only report one hazard ratio (presented on the plots labeled "More than 1 year" for ease of presentation).In addition, several parameters could not be estimated due to a limited number of informative pairs; these parameters are shown as gray dots at y = 0.In the primary models, birth year and parental education were included as covariates.The adjusted models further corrected for alcohol use disorder, drug use disorder, affective disorder, psychotic disorder, and anxiety disorder registrations.blood vessels, injury to blood vessels; CI, confidence interval; crushing, crushing injury; dislocation, dislocation/sprain/strain; internal, internal injury; intracranial, intracranial injury; MZ, monozygotic; nerves/spinal cord, injury to nerves and spinal cord; other injuries, other and unspecified injuries; traumatic amp, traumatic amputation.
Second, the relationship between injuries and SA is complex: The strength of associations varied based on injury type, sex, and age.For instance, some injury types (e.g., crushing injury) were more strongly associated with risk for SA in females than males, whereas other injuries (e.g., fracture, dislocation/sprain/strain, other and unspecified injuries) showed a similar pattern of associations across sex.
Third, the majority of associations between injuries and SA persisted after accounting for exposure to other injuries, as well as for psychiatric and substance use disorder registrations-potentially key confounders.These findings suggest that comorbid psychopathology contributes to, but does not fully account for, the relationship between injuries and SA.
Fourth, co-relative analyses demonstrated that the nature of the relationship between injury exposure and risk for SA varies based on injury type.Consistent with the IPT's hypothesis that exposure to PPEs causally increases SA risk, 3,4 there was evidence to support modest causal effects of eye injury, fracture, dislocation/ sprain/strain, intracranial injury, and other and unspecified injuries on risk for SA.The remaining injury types were each experienced by <1% of the sample; as a result, HRs were imprecisely estimated.Nonetheless, several interesting findings emerged.For example, HRs for the association between internal injury and risk for SA were similar in magnitude among first cousins, half-siblings, full siblings, and MZ twins, suggesting that there could be a causal effect of internal injury on SA that we did not have sufficient statistical power to detect.Conversely, the association between crushing injury and risk for SA steadily declined across relative pairs of increasing genetic relatedness, which is consistent with confounding by familial factors.
Fifth, there was mixed evidence for interaction effects of genetic liability for SA and injury exposure on SA risk.In analyses adjusted for birth year and parental education, Results from co-relative analyses of the relationship between injuries and suicide attempt in male individuals.Hazard ratios and 95% confidence intervals are presented for the predicted estimates within each relative type.Confidence intervals were quite wide in some cases, and error bars were permitted to exceed the plot margins.Complete results are available in the Supporting Information.For all injuries, the association with suicide attempt was non-proportional.Hence, we present two hazard ratios: one for the period within 1 year of injury exposure and one for the period more than 1 year after injury exposure.Several parameters could not be estimated due to a limited number of informative pairs; these parameters are shown as gray dots at y = 0.In the primary models, birth year and parental education were included as covariates.The adjusted models further corrected for alcohol use disorder, drug use disorder, affective disorder, psychotic disorder, and anxiety disorder registrations.blood vessels, injury to blood vessels; crushing, crushing injury; dislocation, dislocation/sprain/ strain; HR, hazard ratio; internal, internal injury; intracranial, intracranial injury; MZ, monozygotic; nerves/spinal cord, injury to nerves and spinal cord; other injuries, other and unspecified injuries; traumatic amp, traumatic amputation.
T A B L E 3 Interaction effects of injuries and family genetic risk scores on risk for suicide attempt.

Note:
Results are presented as the relative excess risk due to interaction (RERI) with 95% confidence intervals.Statistically significant parameter estimates are shown in bold font.In the primary models, birth year and parental education were included as covariates.The adjusted models further corrected for alcohol use disorder, drug use disorder, affective disorder, psychotic disorder, and anxiety disorder registrations.
we observed a number of positive additive interaction effects between FGRS and injuries, such that individuals with higher FGRS for SA were more sensitive to the effects of injury exposure than individuals with lower FGRS.After further correcting for psychopathology, interaction effects remained in the positive direction among 10-to 24-yearolds.In contrast, we observed negative interaction effects in individuals aged 25 years or older.These findings suggest that the relationship between genetic liability and injury exposure may be impacted by psychiatric illnesses, many of which have onsets during adulthood.However, it is also plausible that these contrasting directions of effect are artifactual, 38,39 and replication in another sample would improve confidence in this pattern of results.
This study is subject to several limitations.First, our use of medical registries limits the identification of injuries and SA to those that require medical attention, which may underestimate the occurrence of injuries and suicidal behavior.ICD codes are known to have poor sensitivity in the classification of suicide-related outcomes, 40 and changes in the coding of self-harm from ICD-9 to ICD-10 produced changes in the rates of intentional self-harm and events of undetermined intent identified through electronic health records. 41These limitations of registry-based data are offset by the availability of longitudinal data, absence of reliance on self-reported behavior, and absence of recall bias.
Second, we defined non-fatal SA based on ICD-10 X codes but did not have access to ICD-10 code T14.91 F I G U R E 5 Results from univariable models of the relationship between injuries and risk for suicide attempt, stratified by sex and age group.Results are presented as hazard ratios with 95% confidence intervals.Confidence intervals were quite wide in some cases, and error bars were permitted to exceed the plot margins.Complete results are available in the Supporting Information.Statistically significant parameter estimates are shown as solid circles, and non-significant parameter estimates are shown as empty circles.For most injury types, the association with suicide attempt was non-proportional.Hence, we present two hazard ratios: one for the period within 1 year of injury exposure and one for the period more than 1 year after injury exposure.For injury types that showed a proportional association with suicide attempt, we only present one hazard ratio (labeled "Overall" in the Figure).In the primary models, birth year and parental education were included as covariates.The adjusted models further corrected for alcohol use disorder, drug use disorder, affective disorder, psychotic disorder, and anxiety disorder registrations.blood vessels, injury to blood vessels; CI, confidence interval; crushing, crushing injury; dislocation, dislocation/sprain/strain; internal, internal injury; intracranial, intracranial injury; nerves/spinal cord, injury to nerves and spinal cord; other injuries, other and unspecified injuries; traumatic amp, traumatic amputation.
(assigned when an event is known to be a SA but there is insufficient information regarding the method or type of injury).As a result, some attempts may have been missed.Third, it is possible that some self-harm events were misclassified as unintentional physical injuries, which may inflate associations between injuries and SA.This misclassification bias is likely to impact some injury types (e.g., injury to blood vessels) more than others (e.g., traumatic amputation).Fourth, the present study focused only on the occurrence of physical injury and did not consider the context in which each injury occurred.Further research is needed to investigate whether the magnitude of the association between physical injury and risk for future SA varies based on the nature of the injury-causing event (e.g., whether the injury occurred as the result of an accident vs. nonaccidental trauma).Future studies might also consider injury severity and sequelae, such as residual pain and job loss, as potential mediators and moderators of the relationship between injuries and SA risk.
Fifth, although the co-relative design controls for genetic and environmental factors shared by members of the relative pair, this design does not control for unmeasured individual-level confounders (e.g., physical health, trauma exposure), which precludes a strong causal interpretation of results.Sixth, because information on familial relations is often unavailable for immigrants, the present analyses were restricted to individuals with two F I G U R E 6 Results from multivariable models of the relationship between injuries and risk for suicide attempt, stratified by sex and age group.Results are presented as hazard ratios with 95% confidence intervals.Confidence intervals were quite wide in some cases, and error bars were permitted to exceed the plot margins.Complete results are available in the Supporting Information.Statistically significant parameter estimates are shown as solid circles, and non-significant parameter estimates are shown as empty circles.For most injury types, the association with suicide attempt was non-proportional.Hence, we present two hazard ratios: one for the period within 1 year of injury exposure and one for the period more than 1 year after injury exposure.For injury types that showed a proportional association with suicide attempt, we only present one hazard ratio (labeled "Overall" in the Figure ).In the primary model, birth year and parental education were included as covariates.The adjusted model further corrected for alcohol use disorder, drug use disorder, affective disorder, psychotic disorder, and anxiety disorder registrations.blood vessels, injury to blood vessels; CI, confidence interval; crushing, crushing injury; dislocation, dislocation/sprain/strain; internal, internal injury; intracranial, intracranial injury; nerves/spinal cord, injury to nerves and spinal cord; other injuries, other and unspecified injuries; traumatic amp, traumatic amputation.
Swedish-born parents.As a result, it is unclear whether these results would generalize to individuals who immigrated to Sweden or have a non-Swedish parent (historically marginalized groups in Sweden).Seventh, we used a phenotype-based measure, FGRS, to index aggregate genetic liability for SA.Prior work suggests that phenotype-and genotype-based genetic instruments are both valid estimators of genetic liability but account for largely independent variance in psychiatric outcomes. 42herefore, investigating interactions between injury exposure and genotype-based measures of genetic liability for SA remains an important area for future work.
To conclude, the current study adds to the literature by investigating the relationship between an understudied PPE-physical injuries-and risk for SA in a nationally representative cohort.We found that 10 injury types were positively associated with risk for SA, and the strength of associations differed based on injury type, time elapsed since injury exposure, sex, age, and genetic liability for SA.Psychiatric and substance use disorders and familial confounding factors contributed to the relationship between injuries and SA, but often did not fully account for it, supporting a potential causal effect of injuries on SA risk.These findings have important implications for prevention and intervention efforts, suggesting that the development of effective injury prevention strategies may have a downstream impact on suicide risk.However, additional work is needed to identify the mechanisms through which injury exposure affects risk for SA.The IPT 3,4 highlights lowered fear of death and increased pain tolerance as potential mediators for further research.
Frequencies for injuries, suicidal behavior, and covariates presented separately by sex.
T A B L E 2Note: Injuries that occurred after a suicide attempt are not included.