Predicting Suicide Attempts in Women with Eating Disorders

Rates of suicide attempts among people with eating disorders range from 13% to 31% (Acta Psychiatrica Scand 1999; 106:381). Among patients with anorexia nervosa, recent studies have pinpointed some warning signs, such as being older, having a longer duration of illness, weighing less, and having substance abuse problems. Among patients with bulimia nervosa, warning signs include having more co-morbid psychiatric symptoms and reporting a history of sexual abuse.

Dr. Debra Franko and colleagues at Massachusetts General Hospital, Northeastern General Hospital, and Harvard University conducted a prospective longitudinal study of women with either DSM-IV anorexia nervosa (136) or bulimia nervosa (110). The women were interviewed and assessed for suicidal attempts and suicidal intent every 6 to 12 months for nearly 9 years (Psychological Medicine 2004;34:843).

After the women were screened by telephone, a trained research assistant interviewed them, and follow-up interviews using the LIFE-EAT were conducted every 6 to 12 months. Suicidal behavior was assessed by asking all subjects to report the number of suicidal gestures or attempts made since the previous interview; if either gesture or attempt was reported, suicidal intent was characterized on a 1-6 rating scale (1 indicating no intent to die, and 6 was extreme intent with careful planning and expectation of death). Psychosocial functioning was assessed during the LIFE-EAT interviews with a series of questions that probed relationships, employment, household duties, recreation and overall social adjustment.

Before entering the study, 30% of anorexic subjects and 23% of bulimic subjects reported a history of suicidal attempts (this difference was not significant). Rates of suicide attempts differed across diagnostic groups, although in a very different direction from the original hypothesis because more patients with anorexia nervosa than patients with bulimia nervosa made suicide attempts.

Anorexia nervosa patients

Before the study began, the authors had hypothesized that more suicide attempts would be reported among patients with bulimia nervosa; however, more patients with anorexia nervosa than bulimia nervosa attempted suicide (30/106, or 22.1%, of anorexia nervosa subjects and 12/110, or 10.9%, of bulimia nervosa subjects). The authors questioned whether there had been a crossover from restricting type anorexia nervosa to binge-purge type anorexia nervosa (62% of all the patients with anorexia nervosa engaged in regular binge eating at least once a week for 8 consecutive weeks).

Neither individual bulimic symptoms (such as binge eating, vomiting, use of diet pills/laxatives/diuretics) nor the bulimic symptoms measured over the study significantly predicted suicidal behavior in subjects diagnosed with anorexia at intake. Instead, the study results revealed that when anorexia nervosa is combined with significant co-morbidity (such as depression or substance abuse), the risk of suicide attempts escalates, regardless of the presence of bulimic behaviors.

A history of suicide attempts was found to significantly predict time to a future attempt. Eating disorder symptoms were not a significant predictor for patients with anorexia nervosa.

Bulimia nervosa patients

Unlike the anorexic patients, eating disorders symptoms, particularly laxative use, were a significant predictor for bulimic patients. A history of substance abuse disorder prior to the start of the study significantly predicted the time to the first suicide attempt for BN subjects. Those who were more depressed, as measured bevy the Beck Depression Inventory, and who scored higher on all subscales of the SCI-90, were also at greater risk for suicide attempts. Within the bulimic group, the Ineffectiveness, Interpersonal Distrust, Interoceptive Awareness and Maturity Fears subscales were significantly higher in those who attempted suicide as compared to those who did not.

Results from the Eating Disorders Inventory suggested that, particularly in bulimic women, those who saw themselves as lacking in self-efficiency and who had more difficulty identifying internal states and were distrustful of others were more likely to have made a suicide attempt during the study.

The researchers also found that the amount of both individual and group therapy shortened the time to the first suicide attempt, which was consistent with their data that the more psychiatrically ill patients were more likely to attempt suicide. What could explain this counter-intuitive finding? The authors hypothesize that these patients had individual and group treatment in the hospital setting because of a suicide attempt. Or, it was possible that patients whose illness was persistent and severe were in treatment longer than those with a less chronic course of illness.