A comparison of the frequency of familial suicide attempts across eating disorder diagnoses

Abstract

Objective

This study examined the prevalence of reported suicide attempts among family members of individuals with an eating disorder (ED).

Method

1870 individuals presenting for ED treatment reported whether their family members ever made a suicide attempt using the Eating Disorders Questionnaire.

Results

A lifetime suicide attempt by any family member was reported by 10.8% (n = 202) of the sample and ranged from 7.0% of those with eating disorder not otherwise specified to 16.1% of those with purging disorder. Controlling for age and gender, individuals with bulimia nervosa had a higher prevalence of any familial suicide attempt and mother suicide attempt than individuals with EDNOS; no other differences were observed across ED diagnoses. There were no differences in prevalence of reported suicide attempts made by fathers, brothers, sisters, uncles, or aunts by ED diagnosis.

Conclusions

Findings support a growing literature indicating a familial association between EDs and suicide risk.

Suicide is one of the primary causes of death among individuals with eating disorders (EDs). Suicide rates are higher in EDs than any other psychiatric illness (1, 2). The high mortality due to suicide for individuals with anorexia nervosa (AN) and bulimia nervosa (BN) is well documented (3–6). There is a growing body of research indicating increased prevalence of lifetime suicide attempts and increased mortality due to suicide across all types of EDs, including binge-eating disorder (BED), purging disorder (PD), and eating disorder not otherwise specified (EDNOS; 6, 7). However, potential contributing factors to the increased risk of suicide in individuals with EDs remain unclear.

Both EDs and suicide aggregate in families (8, 9). A recent twin study found a common genetic pathway for both suicidality (defined as any lifetime suicidal ideation) and a composite variable of “any eating disorder” (10). A separate twin study found shared genetic factors in the occurrence of AN and suicide attempts (11). Additionally, a recent family liability study demonstrated that individuals with a full sibling with AN or BN had increased risk of suicide attempts; an elevated but attenuated risk of suicide attempts was also found in more distant relatives (12). Thus, recent research has suggested a shared genetic liability for EDs and suicidality. However, these studies have either used a combined ED group or only investigated AN and BN. Notably, although suicide risk is elevated in EDs other than AN and BN including EDNOS (6), familial patterns of reported suicide attempts have not been examined.

Thus, the purpose of the current study was to determine the prevalence of familial suicide attempts in first and second degree family members of individuals with AN, BN, BED, EDNOS, and PD and additionally to examine prevalence of familial suicide attempts by ED diagnosis. This study aims to contribute to a growing literature establishing the familial association between EDs and suicide by comparing prevalence of familial suicide risk across different ED diagnoses.

Method

Participants and Procedures

Participants included 1870 patients (95.1% female; age M= 26.1, SD = 7.7) who presented for an ED clinical intake appointment between 1975 and 2004 at a Midwestern ED treatment center (13). Participation in the research portion was voluntary. The study had Institutional Review Board approval.

Measures

The Eating Disorder Questionnaire (EDQ) is a self-report questionnaire that includes a series of questions regarding different aspects of ED and related medical and psychiatric problems, including demographics, dieting and weight control behavior, exercise, psychiatric history, and family medical and psychiatric history (14). Diagnoses were derived from the EDQ based upon DSM-IV criteria, including AN, BN, BED (based on the proposed research criteria), PD (purging in the absence of objective binge eating), and EDNOS (subthreshold AN, BN, or BED). We chose to examine PD separately from EDNOS given recent research on increased suicide attempts and mortality in this population (6, 7). In the family medical and psychiatric history section, participants were instructed to indicate in the specified column “any of your blood relatives who has, or has had, the following conditions or problems” with “suicide attempt” listed as a condition. Response options were available for mother, father, brother, sister, uncle, and aunt.

Data Analysis

A composite “any familial suicide attempt” variable was created. A series of generalized linear models were run predicting presence of a suicide attempt made by any family member, mother, father, brother, sister, uncle, or aunt from ED diagnoses, controlling for participant age and gender (male coded as 1, female coded as 2). Age and gender were entered as covariates as each are associated with prevalence of suicide attempts. When the main effect of ED diagnoses were significant, post-hoc analyses were used to determine differences between specific diagnostic groups.

Results

A history of a suicide attempt by any family member was reported by 10.8% (n = 202) of the sample. Prevalence of familial suicide attempts across ED diagnoses are presented in Table 1. Results from generalized linear models predicting presence of a suicide attempt by ED diagnosis controlling for participant age and gender are presented in Table 2. The model predicting any familial suicide attempt was significant (p < 0.012) and ED diagnosis was significantly associated with familial suicide attempt (p < 0.014); pairwise comparisons indicated that the prevalence of familial suicide attempt was higher in BN than in EDNOS but no other pairwise comparisons were significant. The model predicting mother suicide attempt was significant (p < 0.007) and ED diagnosis was significantly associated with mother suicide attempt (p < 0.007); pairwise comparisons indicated that the prevalence of mother suicide attempt was higher in BN than in EDNOS but no other pairwise comparisons were significant. The models predicting presence of a father, brother, sister, uncle, or aunt suicide attempt from ED diagnosis were non-significant (ps > 0.05).

Table 1.

Prevalence of familial suicide attempts by eating disorder diagnosis.

	AN (n = 274)	BN (n = 836)	BED (n = 208)	EDNOS (n = 498)	PD (n = 54)
	N (%)	N (%)	N (%)	N (%)	N (%)
Any Family Member	26 (9.5)	101 (12.1)	27 (13.0)	35 (7.0)	9 (16.7)
Mother	10 (3.6)	42 (5.0)	9 (4.3)	8 (1.6)	6 (11.1)
Father	4 (1.5)	12 (1.4)	4 (1.9)	6 (1.2)	1 (1.9)
Sister	7 (2.6)	41 (4.9)	8 (3.8)	13 (2.6)	4 (7.4)
Brother	6 (2.2)	15 (1.8)	7 (3.4)	10 (5.0)	1 (1.9)
Aunt	1 (0.4)	2 (0.2)	0 (0.0)	1 (0.2)	0 (0.0)
Uncle	1 (0.4)	5 (0.6)	2 (1.0)	1 (0.2)	0 (0.0)

Note. Prevalence of familial suicide attempts were reported by the individual with the eating disorder. AN: anorexia nervosa; BN: bulimia nervosa; BED: binge eating disorder; EDNOS: eating disorder not otherwise specified; PD: purging disorder

Table 2.

Results of generalized linear models predicting familial suicide attempts from participant ED diagnosis.

		Χ2	p	Pairwise Comparisons
Any family member		16.37	0.012
ED diagnosis		12.48	0.014	BN > EDNOS
Participant age		3.22	0.07
Gender		0.03	0.86
Mother		17.71	0.007
ED diagnosis		14.04	0.007	BN > EDNOS
Participant age		0.18	0.67
Gender		1.05	0.31
Father		4.76	0.58
ED diagnosis		1.14	0.89
Participant age		1.89	0.17
Gender		2.29	0.13
Sister		3.46	0.036
ED diagnosis		8.38	0.08
Participant age		3.61	0.06
Gender		2.50	0.11
Brother		0.73	0.99
ED diagnosis		0.35	0.99
Participant age		0.24	0.62
Gender		0.16	0.69
Aunt		2.01	0.92
ED diagnosis		0.20	0.99
Participant age		0.20	0.65
Gender		0.00	0.99
Uncle		3.01	0.81
ED diagnosis		1.69	0.79
Participant age		0.00	0.98
Gender		0.00	0.99

Discussion

This study examined the prevalence of suicide attempts in family members of individuals with EDs and found that over 10% of the sample reported at least one family member having a lifetime suicide attempt. Prevalence of any familial suicide attempt ranged from 7.0% in EDNOS to 16.1% in PD. These prevalence estimates are comparable to prevalence of familial suicide attempts of individuals with major depressive disorder who died by suicide (10.8%) and non-suicidal individuals with major depressive disorder (6.5%), but appear higher than prevalence of familial suicide attempts of community controls (3.4%; 15).

We hypothesize that the prevalence estimates for specific family members are likely underestimates. Methodologically, it is important to note that there was no “not applicable” option for these questions, and participants were only able to mark whether the relative had made a suicide attempt. Thus, we were unable to remove individuals who did not have a sister, brother, aunt, or uncle and these would have been coded as a no lifetime suicide attempt, contributing to a conservative estimate of the prevalence of suicide attempts. Further, individuals who make suicide attempts often report feeling stigmatized (16) and thus may be less likely to disclose their suicide attempts to family members, particularly more distant relatives.

When controlling for participant age and gender, the only differences observed were higher prevalence of any familial suicide attempt and mother suicide attempt in individuals with BN compared to individuals with EDNOS. Given the small sample of individuals with PD, it is possible we were underpowered to detect significant differences in this group. It is interesting that the familial suicide attempts were highest in BN and appeared to also be elevated in PD, as both of these disorder are marked by impulsivity (17) which evidence indicates is a heritable trait (18) and is associated with suicide risk (19). Additionally, the highest prevalence of suicide attempts were observed in female relatives and females have higher prevalence of comorbid psychopathology (e.g., mood disorders, borderline personality disorder). Thus, the shared liability between ED and suicide attempts may due to shared liability for impulsivity or comorbid psychopathology. However, we were unable to test this hypothesis in the present study.

This study had several strengths, including the large sample size and the inclusion of the full spectrum of ED diagnostic categories. A few limitations should be noted. Suicide attempt data were based on a single item and were reported by the individual with the ED. Single item measures of suicide are problematic (20) and it is possible the individual with the ED under or over-reported familial suicide attempts or differentially reported based on ED diagnosis. We were unable to assess the reliability of this measure. Further, we were unable to determine whether individuals with ED had each type of relative, or whether they lived together, which might suggest a shared environmental effect rather than genetic underpinnings. Additionally, we did not have data on whether the individuals with the ED had a lifetime suicide attempt and the participants were treatment seeking; thus, results may not generalize to community samples. Finally, we did not have a control group and thus are unable to determine how these results compare to the prevalence of familial suicide attempts among healthy controls or other psychiatric diagnoses. These limitations notwithstanding, the present study enhances the literature by demonstrating that familial suicide attempts occur in individuals across the full ED diagnostic spectrum.