Patients with eating disorders commonly exhibit comorbid psychiatric disorders, including anxiety, depression and OCD. The presence of comorbid disorders has been shown to exacerbate the severity and chronicity of the disorder, and unfavourably affect treatment outcome. Moreover, comorbid disorders may necessitate specialized treatment plans that take into account all the co-occuring disorders. Recovery from an eating disorder is hard enough, but when it is complicated by depression and severe anxiety, it can be a lot harder.
Nonetheless, commonly co-occuring psychiatric disorders may also provide researchers and clinicians clues about the etiology of eating disorders, the underlying neuronal processes as well as possible pharmacological interventions.
Researchers have been identifying disorders that commonly co-occur with eating disorders and studying the differences in co-morbidity between disorders. I picked one to write about today, it is a study by Blinder and colleagues that came out in 2007. It is by no means the best, but also not the worst, and of course it has several limitations, which I will mention. But it is a place to start.
This is a retrospective study, meaning that the authors went back through previous medical records, in this case of patients admitted to Remuda Ranch Treatment Center from January 1, 1995 to December 31, 2000. The chief benefit of this approach is the large sample size (2,436 patients in total). The assessments at the treatment center were very thorough and detailed, furthermore a random sample was corroborated independently to ensure diagnostic reliability.
But, there are several downsides from the decision to collect data from Remuda:
- One, it is a faith-based treatment center with “an active religious commitment”, so it biased toward religious patients and families. This of course biases the sample population toward a more ethnically and socio-economically homogeneous group: more than 95% of the sample was white and as much as a third paid out of their own pocket (!!).
- Two, Remuda is a residential treatment center, which in and of itself attracts people that are willing to take the time to go there and do treatment, have the money or health insurance to pay for it, either recognize they are in need of help or are forced by guardians/parents. A treatment-seeking sample (that can afford it). In other words, it likely biases the sample to a much more severe and chronic (less treatment responsive) population (Remuda might not have been the first, second or even third treatment option) than in an outpatient or community population. Given that increased co-morbidity is associated with a protracted outcome, it is no surprise that the frequency of co-morbid disorders reported in this study seem very large.
- No males!
Table 1 summarizes the patient population.
Interestingly, note that patients in the AN-R group are younger and have a shorter duration of the illness compared to the other groups. This is not surprising given the studies on diagnostic crossover and the common fact that restricting, for many, turns to restricting and bingeing/purging (and then, the difference between AN-BP (ANB) and bulimia is just weight).
(The authors controlled for these and other factors in the statistical analyses.)
In the end, the analyses included 27 Axis I diagnoses occurring in 10 or more patients. If diagnoses were received by less than 10 people, they were groups together with other disorders, whenever possible. Table 2 summarizes the percentages of patients with Comorbid Axis I diagnoses based on the ED group.
- all groups had between 96-98% co-occurrence of Axis I disorder
- depressive disorders and major depression co-occur with similar frequency among all groups (40-50%)
- bipolar disorder was the highest in BN and lowest in AN-R (5% vs 2%), but this is not significant.
- total % for any anxiety disorder was similar among all the groups but OCD was most prevalent in AN-R and AN-BP (29 and 28%, respectively) and less common in BN (16%); posttraumatic stress disorder was less common in AN-R (10%) than in the other groups (23-25%)
- alcohol-abuse/dependence was significantly more common in BN than other groups (26% versus 3% in AN-R, and 14% in AN-BP and EDNOS)
- in general, the total for having any substance disorders were: 34% in BN, 20% in AN-BP and EDNOS and 5% in AN-R
- ADD was slightly more common in BN than in other disorders
- schizophrenia was three times more prevalent in AN-R (two times in AN-BP) than in other EDs
- around 0.5% had trichotillomania
Table 3 summarized the odds of an Axis I diagnoses occurring together with an anorexia, bulimia or ED-NOS. This table sums up the data in Table 2 in a nice way to illustrate the salient findings of the study. EDNOS is a group that contains a mix of subthreshold AN and BN disorders, which likely cancel each other out in terms of the odds.
It is interesting, too, that some co-morbid disorders, like substance-abuse, are more prevalent in binge-eating/purging patients, regardless of AN-BP or BN diagnoses (again, just a difference in weight, really). Whereas others, like OCD for example, was twice as likely anorexia, regardless of the sub-type, than in bulimia.
Besides the aforementioned limitations, it would be nice to compare this data to the non-ED population at large and compare the rates of these disorders not just between different EDs but also between non-ED and ED populations. Even more necessary, I think, would be the comparison between outpatient and community based samples, in addition to just treatment-seeking inpatient groups (albeit that’s likely harder, given the quality of the assessments required for this study.)
In any case, this paper definitely reveals some interesting trends and hopefully other/future studies address some of the limitations in this study, by including a more diverse and heterogeneous population group (ethnic background, socio-economic status, treatment-seeking/non-treatment seeking (how do you study this group in large enough numbers?, males!) and so on.