by Joye C. AnestisBy now it's pretty much common knowledge that eating disorders are devastating. For instance, we know that anorexia nervosa (AN) has the highest mortality of any psychiatric disorder (10%; APA, 2004). Surprisingly, most of these deaths are by suicide, not medical complications associated with starvation (e.g.,Crisp, Callendar, Halek, & Hsu, 1992; Fedorowicz et al., 2007). The mortality rate for bulimia nervosa (BN) is lower (approxiamtely 0.3%; Keel & Mitchell, 1997), although suicide attempts are highly elevated with this diagnosis (1/3 of women with BN will attempt suicide at least once; e.g., Corcos et al., 2002; Franko et al., 2004). Less is known about the mortality rate for eating disorder not otherwise specified (EDNOS), the most common eating disorder diagnosis. This catch-all category consists of almost any pathological eating behaviors that do not fit the criteria of AN or BN (e.g., binge eating disorder, purging disorder, night eating syndrome). Scott Crow and colleagues recently examined the mortality rate in all eating disorders, and specifically in EDNOS, to cover this gap in our knowledge (the article, from the December 2009 issue of American Journal of Psychiatry can be accessed here).
Subjects in this study were individuals who presented to the University of Minnesota Outpatient Eating Disorders Clinic between 1979 and 1997 (n = 1,885). Each client at this clinic completed the Eating Disorder Questionnaire and reported their Body Mass Index (BMI). These 2 pieces of information were used to generate diagnoses of a current ED (and this reliance on self-report to make diagnoses is a significant limitation to the study that should be considered). To obtain mortality information, the researchers examined the National Death Index, a computerized database of state vital statistics data compiled by the National Center for Health Statistics. Updated yearly, it provides vital status information for the entire United States. Crow and colleagues were able to search this database and classify causes of death into 4 groups: suicide, substance-use related, traumatic, and medical.
The entire sample was overwhelmingly female (95.1%) and white (94.9%). BN was the most common diagnosis (48.1%), followed by EDNOS (42.5%) and then AN (9.4%). The mortality rate for BN was 3.9%, which is higher than is generally found, while the mortality rate in AN was lower than expected at 4.0%. But the main focus of the study was on EDNOS - where they found a mortality rate higher than most would have expected at 5.2%. Along with reported percentages of death, the authors also calculated standardized mortality ratios, which they describe as "the ratio of the observed rate of deaths in a population divided by the expected rate, based on the sample's demographic characteristics" (pp. 1343-44). The standardized mortality ratios for all deaths were 1.70 for AN, 1.57 for BN, and 1.81 for EDNOS. They also reported separate standardized mortality ratios for deaths by suicide. They were 4.86 for AN, 6.51 for BN, and 3.91 for EDNOS. They noted that the elevated suicide risk for AN was not statistically significant, but the elevated risks for BN and EDNOS were significant.
The most interesting and novel finding reported here is the significant number of deaths by suicide in the EDNOS sample - a rate similar to that found in individuals with AN. It is (I hope) common knowledge within the mental health community that AN is an incredibly lethal illness (see this previous post for a brief discussion about why AN is so lethal, above and beyond the effects of starvation). The authors note that many perceive an EDNOS diagnosis to be less severe than other eating pathology, and these findings emphasize the fallacy in this perception. Suicide risk should always be assessed and monitored in EDNOS patients, just as in all patients with mental health problems. One downfall of this study is that they did not report subgroups of individuals within the EDNOS category. EDNOS is an incredibly heterogeneous group of behaviors, including individuals who meet criteria for AN except that they have regular menstrual cycles or their weight is in the "normal" range, individuals who meet criteria for BN except that they binge and purge less frequently than the DSM requires, folks with purging disorder, folks who chew and spit out their food without swallowing, and individuals with binge eating disorder. I think it would have been very interesting to see if certain types of behaviors (e.g., purging disorder) were carrying much of the suicide risk. Another notable limitation, and one which could have influenced the elevated mortality risk, is that obese individuals with binge eating disorder were referred to a different clinic. This could have led to a skewed EDNOS sample.
This study also found a lower mortality rate for AN than previous studies. The authors note that this could be the result of methodological differences between their study and others. The current study used a current diagnosis of AN, whereas previous studies have used lifetime diagnoses. In fact, when the authors reexamined their data using lifetime diagnoses, the standardized mortality ratios for all death and death by suicide both increased and became significant. Furthermore, these were outpatient clients, and perhaps more serious cases of AN would be more likely to seek treatment at an inpatient facility.
In sum, Crow et al.(2009) present evidence that suicide risk and overall mortality in EDNOS should be examined more closely. I bet, with the upcoming release of DSM-V and the revamping of the eating disorder section, more research into this heterogeneous category will be conducted.