a b s t r a c t

Background: Although the number of older adults who engage in nonsuicidal self-injury (NSSI) is not insignifi- cant, research on older adults’ NSSI is scant. The current study examined the prevalence and characteristics of NSSI compared to suicide attempt (SA) in adults older than 50 years who were seen at Emergency Departments (EDs) and their ED visit outcomes.

Methods: Data came from the 2012 Nationwide Emergency Department Sample. We used binary logistic regres- sion analysis to examine demographic and clinical characteristics of NSSI versus SA among 67,069 visits with a diagnosis of either SA or NSSI, and multinomial logistic regression analysis to examine associations between NSSI versus SA and ED outcomes.

Results: Of self-inflicted intentional injuries, 76.89% were SA and 23.11% were NSSI. Visits for NSSI were associated with lower levels of psychiatric disorders and Alcohol use disorders than SA and were more likely than SA visits to occur among older age groups (65-74 and 75+), females, and those with multiple injuries and drug use disor- ders. NSSI visits were also associated with greater risks of hospital admission (relative risk ratio [RRR] = 1.45, 95% CI = 1.36-1.54) and death (RRR = 18.64, 95% CI = 14.19-24.49), as opposed to treat-and-release, but lower risks of facility transfer/discharge with home health care (RRR = 0.77, 95% CI = 0.72-0.83).

Conclusions: The findings of higher hospitalization and Death rates among those with NSSI than SA show how le- thal intentional self-destructive behaviors in late life can be even if they are not classified as suicide attempts. The need for Mental health and substance abuse treatment is discussed.

(C) 2016

1. Introduction

Nonsuicidal self-injury (NSSI) is defined as the deliberate, self- inflicted destruction of body tissue (eg, cutting or burning oneself) without suicidal intent and for purposes not socially sanctioned [1]. In- dividuals engage in NSSI often as a means of alleviating negative emo- tions, self-derogation/self-punishment, escape from a situation/ responsibility, and/or attention seeking [2-5]. In the fourth edition of the Diagnostic and Statistical Manual of Mental disorders (DSM-IV), NSSI was considered a symptom of borderline personality disorder (BPD) [6]. However, in the DSM-5, NSSI is included as a disorder that is distinct from both BPD and suicidal behavior disorder, including sui- cide attempt (SA), but both NSSI and suicidal behavior disorder are noted as conditions needing further study [7]. Previous research pro- vides empirical evidence that NSSI is not unique to BPD and may present with or without any Psychiatric comorbidities [8].

In addition to the absence of suicidal intent, NSSI may differ from SA with respect to its risk factors. The National Comorbidity Survey showed that compared to the 2.7% of those aged 15 to 54 years who had at least

* Corresponding author. Tel.: +1 512 232 9590.

E-mail address: [email protected] (N.G. Choi).

one lifetime suicide attempt, the 1.9% who had at least one lifetime NSSI (“suicide gesture”) were more likely to be female, and they had a signif- icantly lower lifetime prevalence of DSM diagnoses of major depressive episode, drug abuse and dependence, conduct disorder, antisocial per- sonality disorder, and simple phobia and a lower prevalence of multiple incidents of sexual molestation and physical assault [9]. A study of psy- chiatric patients in two Canadian emergency departments (EDs) be- tween 2009 and 2011 also found that compared to the 14% who had attempted suicide, the 4.3% with NSSI had lower odds of depression, psychosis, acute life stressors, and higher rates of social support, but the 2 groups did not differ in gender, age, history of child abuse, and Substance use disorders [10].

Other studies found that NSSI and SA often co-occur frequently, driv- en by overlapping genetic factors, personality traits, and psychiatric co- morbidities in both adolescent and adult samples, although SA is associated with greater levels of psychiatric and psychosocial impair- ments [11-14]. In addition, one study found that psychiatric inpatients’ NSSI history (presence and frequency) was more strongly associated with a history of SA than were their depressive symptoms, hopeless- ness, and BPD symptoms, and as strongly associated with SA history as current levels of suicidal ideation [15]. Among patients with a history of SA, those with an NSSI history also reported significantly greater

http://dx.doi.org/10.1016/j.ajem.2016.02.058

0735-6757/(C) 2016

lethal intent for their most severe SA, and patients’ number of prior NSSI episodes was positively correlated with the level of lethal intent associ- ated with their most severe SA [15]. Other studies also support that NSSI is prognostic of suicide, especially among those who repeatedly engage in it and progressively use more lethal methods, though most suicides took place without intervening episodes of self-harm [16-20].

NSSI behaviors tend to begin during early adolescence, peak during adolescence and the early twenties, and then decrease with age [8]. Given the high prevalence of NSSI during adolescence and young adult- hood, it is not surprising that most research on NSSI has focused on this age group, with a dearth of research on adults [3]. Although fewer older than younger adults engage in NSSI, the number of older adults who do so is not insignificant. For example, New Jersey hospital discharge data in 2003 found that 11.8% of patients (N = 3600) with a primary diagno- sis of NSSI were aged 55 + (compared to 50.4% who were aged 15-34 and 37.9% who were aged 35-54); 9.7% of NSSI patients in the 55 +

age group (compared to 11.9% in the 15-34 and 16.8% in the 35-54 age groups) had subsequent self-harm episode(s); and 1.7% of those aged 55 + (compared to 1.1% and 1.5% in the 15-34 and 35-54 age groups) died of suicide by 2007 [18]. There is no other record of research on NSSI among older adults in the United States; however, a study of homecare clients aged 60+ in Ontario, Canada, found that hospital re- cords of intentional self-harm (including selected accidental poisoning) were present in 9.3 cases per 1000 [21]. The study identified younger age (60-74 years), any psychiatric diagnosis, alcohol use and depen- dence, psychotropic medication use, depressive symptoms, and moder- ate and severe cognitive impairment as risk factors [21], but it did not differentiate those with suicide intent from those without.

Previous studies found that common forms of NSSI among younger age groups include cutting, burning, scratching, banging, hitting, biting, and interfering with wound healing, with drug use and binge drinking contributing to such behaviors [22-24]. However, the aforementioned hospital discharge data in New Jersey in 2003, which included people of all age groups, showed that drug poisoning was the most common (76%) method (followed by cutting and piercing [13%] and other methods [11%]), and it continued to be the most popular method in sub- sequent follow-ups [18]. ED data between 2001 and 2004 also found that the rates of self-harm poisonings in NSSI for women were 101 per 100,000 (95% confidence interval [CI] 81-123) and 66 per 100,000

(95% CI 51-81) for men [25].

In 2013, the highest suicide rate was among people aged 45-64 years and the second highest suicide rate was among those aged 85+ years. Suicide was the fifth leading cause of death among the 45-54 age group, the eighth among the 55 to 64 years age group, and the seven- teenth among the 65+ age group [26]. The Centers for Disease Control and Prevention also reports that 494,169 people were treated in EDs for nonfatal self-inflicted injuries in 2013; however, the data do not distin- guish between SA and NSSI [26]. Most people with NSSI do not seek medical care [3]. However, ED visitors with NSSI or SA are likely to have Serious injuries requiring medical and psychiatric attention, and thus comprise a subset with potentially lethal consequences. Given the high suicide rates among middle-aged and older adults and the po- tential of NSSI to escalate to suicidal behavior, it is important to under- stand potential similarities and differences between those with NSSI and SA with respect to their prevalence, correlates, and choice of methods. Using a nationally representative emergency sample data base, the current study examined (1) the prevalence and methods of SA and NSSI in adults older than 50 years who were seen at EDs; (2) de- mographic and clinical characteristics of those with SA and NSSI; and

(3) ED visit outcomes in NSSI versus SA. Based on previous study find- ings, the study hypotheses were: Adjusting for demographic and clinical confounders, ED visits for NSSI, compared to SA, will be more likely to be by women and less likely to include a diagnosis of mental disorders and substance use disorders (H1); and they will be less likely to result in hospital admission, other facility transfer, discharge against medical ad- vice, and death (H2).

1. Methods
   1. Data and sample

Data came from the 2012 Nationwide Emergency Department Sam- ple (NEDS) sponsored by the Agency for Healthcare Research and Qual- ity. This de-identified, publicly available dataset is part of the Healthcare Cost and Utilization Project and is the largest all-payer ED database. In 2012, NEDS contained information on 31 million ED visits at 950 hospi- tals in 30 states and approximated a 20% stratified sample of all hospital-based EDs in the United States [27]. Stratification was based on geographic region, trauma center designation, hospitals’ urban or rural location, teaching hospitals, and hospital ownership/control (pub- lic, for-profit, and not-for-profit). The 31 million ED events (visits) contained in the 2012 NEDS are weighted to represent the estimated 134 million ED events nationwide in that year [27].

NEDS data elements include patient demographics (age and gen- der); patient location (in counties by population size); patient zip code area income (in national quartiles); diagnostic and procedure codes from the International Classification of Diseases, 9th Revision, Clini- cal Modification (ICD-9-CM) as well as the clusters/categories of diagno- ses in the Clinical Classifications Software (CCS) system [28]; chronic condition indicator (ICD-9-CM diagnoses that last 12 months or longer and place limitations on self-care/independent living/social interactions and/or result in the need for ongoing intervention with medical prod- ucts/services/special equipment); E Codes (for external causes of injury and poisoning: self-inflicted, intentional; unintentional; and assault- related); Total charges; and ED dispositions/outcomes. In this study, we first examined the 26,142,903 ED visits by patients aged 13 + years for injury record and type and then the 2,201,766 ED visits by pa- tients older than 50 years (representing 9,543,396 weighted events) with any type of injury. This report focuses on 67,069 visits by patients older than 50 years (representing 287,364 weighted events) with a di- agnosis of either SA or NSSI.

Measures

SA and NSSI: These were identified from the NEDS E Codes. For SA, we cross-referenced the CCS diagnosis classifications of suicide with the E codes (ie, self-inflicted intentional injury for suicidal purpose). NSSI was defined as self-inflicted intentional injury that was not diag- nosed as suicide. The E Codes were also used to identify unintentional and assault-related injuries and methods of SA and NSSI: falls, cutting/ piercing, fire/burning, firearm, machinery, motor vehicle traffic (MVT), pedal cycling, pedestrian (not MVT), natural/environmental nature, overexertion, poisoning, being struck by/against, suffocation, adverse effects of medical care, adverse effect of medical drugs, and other.

ED outcomes were (1) treat-and-release; (2) admission to the same hospital or transfer to a short-term hospital as an inpatient (and did not die); (3) death either in the ED or in the hospital; (4) transfer to a skilled nursing facility, Intermediate care facility, or other such facility (transfer to another facility); or discharge with initiation of home healthcare ser- vices (HHC); and (5) discharge against medical advice (AMA). Due to their small sample sizes, ED outcomes of transfer to another facility and HHC were combined in multivariate analyses, and “other/discharge with an unknown destination” was excluded from analyses.

Patient demographics and health status included (1) age in years (the 2012 NEDS data set provides chronological age up to 90 years and codes those who are older than 90 years), (2) age group (50-64, 65-74, 75+ years); (3) gender; (4) income by patient zip code area (lower 50% vs missing or upper 50% [reference category]); (5) number of chronic medical conditions; and (6) multi-injury indicator (1 = more than one injury; 0 = one injury only).

Mental health and substance use disorders (MHSUDs) were identified from the single-level CCS diagnosis classifications including 12 mental disorders (suicide is one of them) and two substance use disorders

(Alcohol-related disorders [AUDs] and drug-related disorders [DUDs]) as long as they were listed as the primary to the fifteenth diagnosis. We further collapsed the 11 mental disorders (excluding suicide or SA) into five: (1) anxiety disorders; (2) Mood disorders; (3) delirium/ dementia, amnestic, and other Cognitive disorders (collectively referred to as cognitive disorders hereafter); (4) other mental health disorders (adjustment disorders attention deficit, conduct, and disruptive behav- ior disorders; developmental disorders; impulse control disorders; per- sonality disorders; schizophrenia and other psychosis); and (5) other miscellaneous disorders.

Data analysis

Analyses were conducted with Stata/MP 14’s svy function to account for NEDS’s multi-stage, stratified sampling design. Stata’s subpop com- mand was used for all subsample analyses (ie, only those pertaining to SA or NSSI-related visits by individuals older than 50 years) to ensure that variance estimates incorporate the full sampling design. All esti- mates presented are weighted to discharges in the universe except for sample sizes (ie, number of visits). SEs for all study variables show sta- ble estimates. First, descriptive statistics for injury status and cause by age group (13+ years and then 50+ years) were used to examine po- tential age group differences. Second, ?² and t tests were used to com- pare demographic and clinical characteristics and ED outcomes of SA visits and NSSI visits among 3 age groups: 50-64, 65-74, and 75 +. Third, binary logistic regression analysis was used to examine demo- graphic and clinical characteristics of NSSI versus SA. Fourth, multino- mial logistic regression analyses were used to examine associations of NSSI versus SA with ED outcomes (the dependent variable, with treat- and-release as the base outcome) among the 50+ age group, with de- mographic and clinical characteristics as confounders. Variance inflation factor diagnostics (using a cut-off of 2.50) [29] indicated that multicollinearity among the predictors was not a concern.

1. Results
   1. Injury record and cause by age group

Table 1 shows that of all ED visits by patients aged 13+ years, 76.49% had no recorded injury, 0.83% had a diagnosis of suicide (or SA), 0.34% had NSSI, 21.17% had unintentional injury, 1.08% had injury by assault, and 0.09% had unspecified injury. As expected, the highest rate (34.76%) of injury visits was found in the 13-19 age group. With regard to specific causes of in- jury, suicide or SA was highest in the 13-19 age group (1.33%), followed by the 35-49 age group (1.13%), 20-34 age group (0.91%), 50-64 age group

(0.83%), 65-74 age group (0.24%), and the 75+ age group (0.11%). NSSI was also highest in the 13-19 age group (0.78%), followed by the 20-34 age group (0.43%), 35-49 age group (0.39%), 50-64 age group (0.25%), 65- 74 age group (0.08%), and the 75+ age group (0.04%).

Self-injury cause and methods among visits by the 50+ age group

Table 2 shows that of all injury-related visits by the 50 + -year olds, SA constituted 2.32% (ranging from 3.89% in the 50-64 age group

to 0.50% in the 75 + age group), and NSSI was 0.70% (ranging from 1.14% in the 50-64 age group to 0.17% in the 75 + age group). Data also show that of the ED visits with self-inflicted intentional injuries, 76.89% were classified as SA, and 23.11% were as NSSI.

Method of injury was missing for the majority of SA visits; however, analysis of available data found that the most common methods record- ed for visits by members of the 50-65, 65-74, and 75+ age groups, re- spectively, were poisonings (4.67%, 4.72%, and 3.30%), falls (1.46%,

2.95%, and 4.43%), and cutting (1.20%, 1.04%, and 1.04%). For methods of NSSI (which had only a small proportion with missing data), the most common methods recorded for visits by members of the 50-65, 65-74, and 75 age groups, respectively, were poisoning (76.80%,

76.29%, 59.12%), cutting (13.80%, 11.30%, 15.69%), firearms (1.71%,

3.64%, 7.26%), suffocation (1.82%, 1.75%, 3.36%), and falls (1.67%, 2.52%,

3.05%).

Demographic and clinical characteristics and ED Outcomes of SA and NSSI by age group: bivariate analysis results

Table 3 shows that in the 50-64 age group, visits for NSSI were by pa- tients who were slightly older than for SA visits, but there was no differ- ence in age among visits for NSSI and SA in the 65-74 and 75 + age groups. In the 50-64 and 65-74 age groups, there were more visits by women for NSSI than for SA. In all three age groups, fewer chronic con- ditions and lower rates of mood disorders and other mental disorders were reported in NSSI than SA visits. In the 50-64 age group, anxiety dis- orders were reported in more NSSI than SA visits; in the 65-74 and 75+ age groups, rates of cognitive disorders were lower for NSSI than SA visits. In the 50-64 and 65 + age groups, rates of AUDs were also lower for NSSI than SA visits, but NSSI visits had higher rates of DUDs re- ported for all three age groups. For all age groups, the rate of multiple in- juries (ie, more than one) among NSSI visits (32%-38%) was significantly higher than among SA visits (2%-3%).

ED outcomes show that NSSI visits resulted in lower rates of treat- and-release and facility transfer/HHC and higher rates of hospital ad- mission and death. For example, the death rate was 10% among NSSI visits in the 75 + age group, compared to 0.75% among the SA visits. Also notable is that in the 75+ age group, the AMA discharge rate was 10% for NSSI visits compared to 0.17% for SA visits, while in the two younger groups, the rates were under 1% for both NSSI and SA visits.

Demographic and clinical correlates of NSSI versus SA: binary logistic regression results

Table 4 shows that compared to the visits by the 50-64 age group, those by the 65-74 and 75 + age groups had slightly higher odds of being for NSSI than SA. Female gender, multi-injury, anxiety disorder, and DUDs were also associated with higher odds of being NSSI than SA, while mood disorders, cognitive disorders, other mental disorders, and AUDs were associated with lower odds of being NSSI than SA visits. By far, having multiple injuries was most strongly associated with NSSI than SA (OR = 19.69, 95% CI = 18.39-21.09). These findings mostly support H1.

Table 1

Injury record and cause by age group among ED visits in 2012

%	All (13+ y)		13-19 y		20-34 y		35-49 y		50-64 y		65-74 y		75 + y
	100%		9.16%		28.60%		22.43%		19.35%		8.59%		11.87%
No recorded injury	76.49		65.24		76.41		77.15		78.52		80.60		77.81
Suicide attempt (SA; diagnosed)	0.83		1.33		0.91		1.13		0.83		0.24		0.11
Nonsuicidal self-injury (NSSI)	0.34		0.78		0.43		0.39		0.25		0.08		0.04
Unintentional injury	21.17		30.62		20.34		20.10		19.71		18.88		21.97
Injury by assault	1.08		1.91		1.80		1.14		0.61		0.14		0.06
Unspecified	0.09		0.12		0.12		0.10		0.08		0.03		0.02

Total N = 26,142,903 (representing weighted population of 112,896,756).

Table 2

Injury cause among ED visits by patients 50+ years with injury record

	Cause of all recorded injuries1
%	All	50-64 y		65-74 y		75 + y
	100%	49.16%		19.68%		31.16%
Suicide attempt (SA; diagnosed)	2.32 (2.29-2.34)	3.89 (3.85-3.92)		1.26 (1.23-1.30)		0.50 (0.48-0.50)
Nonsuicidal self-injury (NSSI)	0.70 (0.68-0.71)	1.14 (1.12-1.16)		0.42 (0.40-0.44)		0.17 (0.16-0.18)
Unintentional injury	95.13 (95.10-95.16)	91.77 (91.71-91.82)		97.41 (97.36-97.45)		99.01 (98.98-99.03)
Injury by assault	1.62 (1.60-1.64)	2.84 (2.81-2.87)		0.73(0.71-0.76)		0.26 (0.25-0.27)
Unspecified	0.24 (0.23-0.24)	0.37 (0.36-0.38)		0.18 (0.17-0.19)		0.07 (0.06-0.07)
	SA or NSSI only2
%	All	50-64 y		65-74 y		75 + y
	%	100%		82.10%		10.99%
SA	76.89	77.28		75.04		75.19
NSSI	23.11	22.72		24.96		24.81

(): 95% CI.

Age group differences in each row are significant at P b .0001.

¹ Total N = 2,201,766 (representing weighted population of 9,545,396); N = 51,537 for SA; N = 15,532 for NSSI; N = 2,093,454 for unintentional injury; N = 35,777 for assault, and N = 5466 for unspecified cause.

² Total N = 67,069 (representing weighted population of 287,364); N = 54,998 for the 50-64 years; N = 7413 for the 65-74 years; and N = 4658 for the 75+ years.

Associations of NSSI versus SA with ED Outcomes: multinomial logistic regression results

Table 5 shows that compared to SA visits, NSSI visits were associated with greater risks of hospital admission (relative risk ratio [RRR] = 1.45, 95% CI = 1.36-1.54) and death (RRR = 18.64, 95% CI = 14.19-24.49), as

opposed to treat-and-release, but lower risks of facility transfer/HHC (RRR = 0.77, 95% CI = 0.72-0.83), adjusting for demographic and clin- ical confounders. The findings fail to support H2. Of the confounders, visits by those in older age groups resulted in increased risk for hospital admission and death. Multiple injuries also greatly increased the risk of hospital admission (RRR = 2.46, 95% CI = 2.25-2.70), death (RRR = 3.02, 95% CI = 2.43-3.74), and facility transfer/HHC (RRR = 1.28, 95% CI = 1.15-1.43). Female gender, number of chronic conditions, and

Table 3 Demographic and clinical characteristics of ED visits and visit outcomes for SA versus NSSI among those older than 50 years

50-64 y 65-74 y 75 + y

N SA NSSI SA NSSI SA NSSI 42,496 12,502 5548 1865 3493 1165

Age¹ (M,SE) 55.02 55.21 68.56 68.50^? 81.82 82.07^?

mood, cognitive, and other mental disorders also increased the risk of hospital admission, while living in a low-income area and AUDs de- creased the risk. Mood disorders also increased but AUDs decreased the risk of facility transfer/HHC. Female gender, mood disorders, and other mental disorders decreased the risk of discharge AMA, but living in a low-income area increased the risk. Female gender, living in a low-income area, anxiety, mood, and other mental disorders, AUDs, and DUDs decreased the risk of death, while the number of chronic con- ditions increased the risk.

1. Discussion

Using a nationally representative ED visit data, this study examined two types of intentional self-inflicted injury, NSSI and SA, in adults older than 50 years and helps fill the research gap on NSSI in this age group. As expected, NSSI and SA are smaller proportions of the injury-related ED events in this age group compared to younger age groups. Among visits by patients older than 50 years, those by older individuals (aged 65-74 and 75 +) had overall lower rates of intentional self-inflicted injuries than in the younger (aged 50-64) group. The findings show that among visits in the 50+ age group, a little less than one-fourth of inten- tional self-inflicted injuries were NSSI, while the rest were SA. As in pre- vious studies [11-14], this study found that compared to SA visits, NSSI

	(0.02)	(0.04)	(0.04)	(0.07)	(0.08)	(0.15)	visits were associated with lower levels of psychiatric disorders (mood,
Female (%)	41.40	56.36	49.31	57.99	54.01	50.58?
low income area2 resident (%)	59.89	57.29	53.12	53.13?	46.32	47.83?
No. of chronic conditions	4.22	4.07	4.96	4.57	5.43	4.97	Table 4
(M,SE) Anxiety disorder (%)	(0.01) 19.82	(0.03) 23.39	(0.04) 21.72	(0.07) 21.56?	(0.05) 17.09	(0.10) 15.46?	Associations of demographic and clinical characteristics with NSSI versus SA-related visits by patients older than 50 years: odds ratios (OR) and 95% CI
Mood disorder (%)	73.22	66.03	74.72	66.18	70.17	54.79
Cognitive disorder (%)	1.17	1.09?	9.53	4.83	34.76	18.47	NSSI vs SA
Other mental disorder (%)	25.10	16.75	21.68	12.99	17.73	11.23
Alcohol use disorder (AUD, %)	33.60	29.04	19.97	15.01	6.20	5.98?		OR (95% CI)
Drug use disorder (DUD, %)	22.48	29.74	9.03	21.97	3.47	14.80	Age group
AUD and/or DUD (%)	46.41	38.28	25.97	25.22	9.29	13.82	(50-64 y)
More than one injury (%)	2.71	37.60	3.26	34.55	3.11	32.20	65-74 y	1.13 (1.05-1.21)
ED disposition/outcome3 (%)							75 + y	1.19 (1.08-1.31)
Treat and release	25.19	20.05	18.89	15.12	16.25	12.10	Female	1.61 (1.54-1.67)
Hospital admission	50.18	61.28	59.53	65.69	65.62	67.48	Low-income area resident	0.96 (0.91-1.01)?
Facility transfer/home	23.98	15.91	20.96	13.96	17.20	10.30	No. of chronic conditions	1.01 (1.00-1.02)?
health care							Multi-injury	19.69 (18.39-21.09)
Discharge AMA	0.57	0.44	0.37	0.22	0.17	10.18	Anxiety disorders	1.22 (1.16-1.29)
Died at ED or hospital	0.07	2.13	0.24	5.01	0.75	9.94	Mood disorders	0.53 (0.50-0.55)

¹ The ages of visitors aged 90 years or older are recorded as 90 years.

² Defined as zip code area income <=50% of the national median.

³ 140 visits by the 50-64 age group, 9 visits by the 65-74 age group, and 8 visits by the 75+ age group were excluded as their ED outcomes were unknown.

? Nonsignificant difference (P N .05) between SA and NSSI; all other paired compari-

sons in each age group are significant at P b .0001.

N = 67,067; F (12, 31,090,927) = 691.49, P b .001.

Cognitive disorders	0.58 (0.50-0.66)
Other mental disorders	0.54 (0.51-0.57)
Alcohol use disorders	0.74 (0.70-0.78)
Drug use disorders	1.39 (1.32-1.47)

^?P >= .095; P b .001 for all other ORs.

Table 5

Effects of NSSI versus SA on ED outcomes among patients older than 50 years: RRR and 95% CI

	Treat and release vs
	Hosp admission	Fac transfer/HHC		Discharge AMA		Died
	RRR (95% CI)	RRR (95% CI)		RRR (95% CI)		RRR (95% CI)
NSSI (vs SA)	1.45 (1.36-1.54)	0.77 (0.72-0.83)		0.91 (0.67-1.24)?		18.64 (14.19-24.49)
Age group
(50-64 y) 65-74 y	1.23 (1.14-1.33)	1.09 (1.00-1.19)?		0.84 (0.55-1.29)?		2.16 (1.67-2.80)
75 + y	1.27 (1.15-1.41)	0.97 (0.86-1.09)?		0.53 (0.24-1.17)?		3.36 (2.56-4.42)
Female	1.10 (1.05-1.15)	0.99 (0.94-1.04)?		0.66 (0.51-0.85)?		0.39 (0.32-0.49)
Low-income area resident	0.83 (0.79-0.87)	0.86 (0.82-0.90)		1.34 (1.04-1.72)?		0.66 (0.54-0.80)
No. of chronic conditions	1.58 (1.56-1.60)	0.99 (0.97-1.00)?		1.05 (0.98-1.13)?		1.49 (1.43-1.55)
Multi-injury	2.46 (2.25-2.70)	1.28 (1.15-1.43)		0.84 (0.50-1.43)?		3.02 (2.43-3.74)
Anxiety disorders	0.79 (0.74-0.84)	0.95 (0.89-1.02)?		0.85 (0.61-1.20)?		0.24 (0.16-0.80)
Mood disorders	1.40 (1.33-1.48)	1.24 (1.17-1.31)		0.51 (0.39-0.67)		0.22 (0.18-0.28)
Cognitive disorders	1.34 (1.15-1.56)	1.09 (0.90-1.30)?		0.43 (0.11-1.76)?		0.61 (0.37-1.01)?
Other mental disorders	1.29 (1.21-1.37)	0.96 (0.89-1.03)?		0.69 (0.49-0.98)?		0.17 (0.11-0.27)
Alcohol use disorders	0.72 (0.68-0.75)	0.62 (0.58-0.66)		0.74 (0.57-0.97)?		0.15 (0.11-0.22)
Drug use disorders	1.11 (1.05-1.18)?	0.96 (0.89-1.03)?		0.99 (0.73-1.34)?		0.32 (0.23-0.43)

N = 66,910; F (11, 31,090,781) = 239.43, P b .0001.

^?P >= .05; ^?P b .05; P b .001 for all other RRRs.

cognitive, and other mental disorders and AUDs). However, NSSI visits were more likely than SA visits to be associated with older ages (65- 74 and 75 +), female gender, multiple injuries, anxiety disorders, and DUDs. Multiple injuries and DUDs were especially prevalent in NSSI visits among the 50+ age group, and these important clinical character- istics appear to differentiate visits by older NSSI patients from visits by older SA patients. The greater likelihood of multiple injuries and DUDs in NSSI visits than in SA visits is not surprising as drug poisoning, the most common method of NSSI, is likely to have contributed to both in- ternal and external injuries, especially in older age groups.

Given significantly higher rates of multiple injuries and DUDs among NSSI visits, greater risks for hospital admission in these visits than in SA visits are also not surprising. NSSI patients may have been admitted to the hospital for drug overdose and drug-related injuries and other health conditions as well as for detoxification and DUD treatment. Previous stud- ies have found that older ED trauma patients who screened positive for cocaine or marijuana had significantly longer lengths of stay with in- creased ICU admission compared to those who screened negative, while alcohol-Intoxicated patients, compared to non-intoxicated patients, had significantly lower in-hospital mortality, decreased ICU admission, de- creased Intubation rate, and shorter hospital length of stay [30].

Our findings also show that although the intent of NSSI is not suicide, NSSI visits more often resulted in fatal consequences compared to visits among older adults who were brought to the ED with intentional self- inflicted injuries classified as suicide attempts. However, our study may underestimate fatalities from SA. It is well known that suicide com- pletion rates are higher among older than younger adults [31], and those suicides that are completed before identification and transport are not brought to the ED and therefore not included in NEDS. The find- ings of NSSI’s high lethality among the 50 + age group are noteworthy given the scant research on older adults who present at EDs with NSSI. Further research is needed to examine NSSI causes/methods in older adults that are likely to lead to high fatality.

Interestingly, our findings show that both AUDs and DUDs were as- sociated with decreased risk of death among older adults with inten- tional self-inflicted injuries. Other studies of community-dwelling older adults found higher mortality rates (eg, from HIV- and liver- related death) among older adults with substance use disorders com- pared to older peers without substance use disorders and younger adults with these disorders [32,33]. Since the current study focused on those with intentional self-inflicted injuries, direct comparisons with these previous study may not be warranted. Differences in study popu- lations may be responsible for these discrepancies between our findings and previous research findings.

Our study has several limitations due to NEDS data constraints. First, possible reporting bias in ED physicians’ assignment of an E code and di- agnosis of SA versus no such diagnosis to individual patients, for which we could not account, may have resulted in discrepancies or inaccura- cies in the classification and coding of SA and NSSI at different EDs and among physicians at the same ED. Many physicians may be reluc- tant to code SA and/or NSSI in older adults for a variety of reasons: family- and insurance-related pressures, medicolegal issues, and time constraints. Patients may also have overlapping signs and symptoms of SA and NSSI that may be difficult to distinguish from each other. The resources available for mental health care also vary greatly across EDs; over half of EDs lack a mental health care professional available to evaluate adults who deliberately harm themselves [34]. A study of self-harm assessment practices in eight US EDs found that only about 26% (11% excluding outliers) of patients age 18+ years were assessed for self-harm thoughts or behaviors and that patients age 65 + years were significantly less likely to be assessed than younger patients [35]. Even for those patients with a history of NSSI as a chronic psychiatric condition, more acute or severe clinical (medical) problems may divert attention away from these mental health concerns. It is also possible that some ED physicians may have become complacent about reporting NSSI in patients who chronically injure themselves [36].

Second, because the methods of injury were missing in the majority of SA visits, we could not compare methods used in NSSI and SA. Third, for DUDs, the data set does not allow for differentiating illicit drug mis- use from prescription drug misuse, which would have provided more information on the correlates of NSSI and SA. Fourth, because observa- tion units are visits, not individuals, return visits by the same individuals could not be examined. Examinations of the frequency of ED visits by patients with NSSI versus SA may also be useful in identifying differ- ences or similarities between these two groups.

1. Conclusions

Overall, our study found that older adults who engaged in NSSI differ in important ways from their age peers who made a suicide attempt. Previous research findings about younger age groups showed that inju- ries tended to be more severe or lethal for suicide attempters than those with NSSI. However, our study findings show that the reverse may be true for older age groups, at least among those older adults who were seen at EDs for intentional self-inflicted injuries. Higher hospitalization and death rates among older ED patients with NSSI than among suicide attempters show how lethal intentional self-destructive behaviors in late life can be, even in the absence of suicidal intent (though it should

be noted that completed suicides are not brought to the ED). These find- ings suggest the care with which older adults who present with NSSI must be treated given the potential for death from their injuries follow- ing presentation at the ED.

The findings provide the following clinical and research implica- tions. First, ED physicians should take care to identify both NSSI and SA and refer older-adult survivors of NSSI as well as SA to community- based mental health and/or substance abuse treatment programs. Sec- ond, mental health and substance abuse treatment providers should work with primary care physicians to monitor the status of older adults who have engaged in NSSI and SA to prevent them from engaging in fur- ther self-injurious, destructive behaviors. Third, healthcare and mental health providers should pay close attention to prevent NSSI and SA among older substance abusers. Fourth, given scant research on older adults with NSSI, more effort is needed to examine differences and sim- ilarities between NSSI and SA in late life with respect to risk and protec- tive factors, methods of injury, and treatment options. Research is also needed to examine classification accuracy of NSSI versus SA at EDs.

Author Contributions

NGC, DMD, and BYC jointly conceptualized, designed, reviewed the lit- erature, and obtained the data set (with data use agreement) for the study. NGC did statistical analyses and drafted the manuscript. DMD and BYC edited the draft and further contributed to writing. CNM provided statisti- cal consulting and contributed to the final editing of the manuscript.

Conflict Of interest

None for any author.

Funding source

None.

Acknowledgments

The data set for this study was made available by the United States Agency for Healthcare Research and Quality.

References

1. International Society for the Study of Self-Injury. Definition of non-suicidal self- injury. http://www.itriples.org/isss-aboutself-i.html; 2007. [Accessed 28 May 2014].
2. Klonsky ED, Muehlenkamp JJ. Self-injury: a research review for the practitioner. J Clin Psychol 2007;63(11):1045-56.
3. Klonsky ED. Non-suicidal self-injury in the United States adults: prevalence, sociodemographics, topography and functions. Psychol Med 2011;41(9):1981-6. http://dx.doi.org/10.1017/S0033291710002497.
4. Nock MK. Why do people hurt themselves? New insights into the nature and func- tions of self-injury. Curr Dir Psychol Sci 2009;18(2):78-83.
5. Nock MK. Self-injury. Annu Rev Clin Psychol 2010;6:339-63. http://dx.doi.org/10. 1146/annurev.clinpsy.121208.131258.
6. American Psychiatric Association. Diagnostic and statistical manual of mental disor- ders. 4th ed., text rev. Washington, DC: Author; 2000.
7. American Psychiatric Association. Diagnostic and statistical manual of mental disor- ders. 5th ed. Arlington, VA: American Psychiatric Publishing; 2013.
8. Zetterqvist M. The DSM-5 diagnosis of nonsuicidal self-injury disorder: a review of the empirical literature. Child Adolesc Psychiatry Ment Health 2015;9:31. http:// dx.doi.org/10.1186/s13034-015-0062-7.
9. Nock MK, Kessler RC. Prevalence of and risk factors for suicide attempts versus sui- cide gestures: analysis of the National Comorbidity Survey. J Abnorm Psychol 2006; 115(3):616-23.
10. Chartrand H, Bhaskaran J, Sareen J, Katz LY, Bolton JM. Correlates of nonsuicidal self- injury and suicide attempts among tertiary care, emergency department patients. Can J Psychiatry 2015;60(6):276-83.
11. Claes L, Muehlenkamp J, Vandereycken W, Hamelinck L, Martens H, Claes S. Compar- ison of non-suicidal self-injurious behavior and suicide attempts in patients admit- ted to a psychiatric crisis unit. Pers Indiv Differ 2010;48(1):83-7.
12. Fliege H, Kocalevent R-D, Walter OB, Beck S, Gratz KL, Gutierrez PM, et al. Three as- sessment tools for deliberate self-harm and suicide behavior: evaluation and psy- chopathological correlates. J Psychosom Res 2006;61(1):113-21 [PubMed: 16813853].
13. Klonsky ED, May AM, Glenn CR. The relationship between nonsuicidal self-injury and attempted suicide: converging evidence from four samples. J Abnorm Psychol 2013;122(1):231-7. http://dx.doi.org/10.1037/a0030278.
14. Maciejewski DF, Creemers H, Lynskey MT, Madden PAF, Heath AC, Statham DJ, et al. Different outcomes, same etiology? shared genetic and environmental influences on non-suicidal self injury and suicidal ideation. JAMA Psychiatry 2014;71(6):699-705. http://dx.doi.org/10.1001/jamapsychiatry.2014.89.
15. Andover MS, Gibb BE. Non-suicidal self-injury, attempted suicide, and suicidal intent among psychiatric inpatients. Psychiatry Res 2010;178(1):101-5. http://dx.doi.org/ 10.1016/j.psychres.2010.03.019.
16. Anestis MD, Khazem LR, Law KC. How many times and how many ways: the impact of number of nonsuicidal self-injury methods on the relationship between nonsuicidal self-injury frequency and suicidal behavior. Suicide Life Threat Behav 2015;45(2):164-77. http://dx.doi.org/10.1111/sltb.12120.
17. Fliege H, Lee JR, Grimm A, Klapp BF. Risk factors and correlates of deliberate self- harm behavior: a systematic review. J Psychosom Res 2009;66(6):477-93. http:// dx.doi.org/10.1016/j.jpsychores.2008.10.013.
18. Miller M, Hempstead K, Nguyen T, Barber C, Rosenberg-Wohl S, Azrael D. Method choice in nonfatal self-harm as a predictor of subsequent episodes of self-harm and suicide: implications for clinical practice. Am J Public Health 2013;103(6): e61-8. http://dx.doi.org/10.2105/AJPH.2013.301326.
19. Skegg K. Self-harm. Lancet 2005;366(9495):1471-83.
20. Bryan CJ, Rudd MD, Wertenberger E, Young-McCaughon S, Peterson A. Nonsuicidal self-injury as a prospective predictor of suicide attempts in a clinical sample of mil- itary personnel. Compr Psychiatry 2015;59:1-7. http://dx.doi.org/10.1016/j. comppsych.2014.07.009.
21. Neufeld E, Hirdes JP, Perlman CM, Rabinowitz T. Risk and protective factors associat- ed with intentional self-harm among older community-residing home care clients in Ontario, Canada. Int J Geriatr Psychiatry 2015;30(10):1032-40. http://dx.doi.org/10. 1002/gps.4259.
22. Klonsky ED, Olino TM. Identifying clinically Distinct subgroups of self-injurers among young adults: a Latent class analysis. J Consult Clin Psychol 2008;76(1): 22-7. http://dx.doi.org/10.1037/0022-006X.76.1.22.
23. Martin G, Swannell SV, Hazell PL, Harrison JE, Taylor AW. Self-injury in Australia: a community survey. Med J Aust 2010;193(9):506-10.
24. Whitlock J, Eckenrode J, Silverman D. Self-injurious behaviors in a college popula- tion. Pediatrics 2006;117(6):1939-48.
25. Prosser JM, Perrone J, Pines JM. The epidemiology of intentional non-fatal self-harm poisoning in the United States: 2001-2004. J Med Toxicol 2007;3(1):20-4.
26. CDC. Suicide facts at a glance 2015. http://www.cdc.gov/violenceprevention/pdf/sui-

cide-datasheet-a.pdf; 2015.

1. U.S. Agency for Healthcare Research and quality healthcare Cost and Utilization Project (HCUP). Introduction to the HCUP Nationwide Emergency Department Sam- ple (NEDS) 2012. HCUP Central Distributor; 2014 [http://www.hcup-us.ahrq.gov].
2. Elixhauser A, Steiner C, Palmer L, Clinical Classifications Software (CCS). U.S. Agency for Healthcare Research and Quality Healthcare Cost and Utilization Project (HCUP); 2014[https://www.hcup-us.ahrq.gov/toolssoftware/ccs/CCSUsersGuide.pdf].
3. Allison P. When can you safely ignore multicollinearity? http://statisticalhorizons. com/multicollinearity; 2012.
4. Lank PM, Crandall ML. Outcomes for older trauma patients in the emergency depart- ment screening positive for alcohol, cocaine, or marijuana use. Am J Drug Alcohol Abuse 2014;40(2):118-24. http://dx.doi.org/10.3109/00952990.2014.880450.
5. Conwell Y, Duberstein PR, Cox C, Herrmann J, Forbes N, Caine ED. Age differences in behaviors leading to completed suicide. Am J Geriatr Psychiatry 1998;6(2):122-6.
6. Larney S, Bohnert AS, Ganoczy D, Ilgen MA, Hickman M, Blow FC, et al. Mortality among older adults with Opioid use disorders in the Veteran’s Health Administra- tion, 2000-2011. Drug Alcohol Depend 2015;147:32-7. http://dx.doi.org/10.1016/j. drugalcdep.2014.12.019.
7. West NA, Severtson SG, Green JL, Dart RC. Trends in abuse and misuse of prescrip- tion opioids among older adults. Drug Alcohol Depend 2015;149:117-21. http:// dx.doi.org/10.1016/j.drugalcdep.2015.01.027.
8. Olfson M, Marcus S, Bridge J. Emergency treatment of deliberate self-harm. Arch Gen

Psychiatry 2012;69(1):80-8. http://dx.doi.org/10.1001/archgenpsychiatry.2011.

108.

1. Caterino JM, Sullivan AF, Betz ME, Espinola JA, Miller I, Camargo Jr CA. Boudreaux ED; Emergency Department safety assessment and follow-up evaluation (ED- SAFE) investigators. evaluating current patterns of assessment for self-harm in emergency departments: a multicenter study. Acad Emerg Med 2013;20(8): 807-15. http://dx.doi.org/10.1111/acem.12188.
2. Commons Treloar AJ, Lewis AJ. Professional attitudes towards deliberate self-harm in patients with borderline personality disorder. Aust N Z J Psychiatry 2008;42(7): 578-84.