The corresponding 2002 review of psychotherapy concludes—somewhat more cautiously—that “there is a small body of evidence for the efficacy of cognitive-behavior therapy in bulimia nervosa and similar syndromes, but the quality of trials is very variable and sample sizes are often small.”
In bulimia nervosa and other psychiatric disorders, comparing psychotherapy with drug therapy is hazardous because several factors bias the comparison in favor of psychotherapy. These factors include an expectational effect, a responsibility effect, and differential generalizability of study results.
Expectational effect. Patients in clinical trials are aware that they are receiving psychotherapy and, presumably, that study investigators hope to demonstrate its efficacy. This might account for much of psychotherapy’s apparent effect, as even placebos can produce 30% to 50% improvement in bulimia.14
Responsibility effect. If a patient fails to improve in a drug study, she will conclude that the drug has failed. But if she fails to improve in a psychotherapy study, she may conclude that she has failed. Because psychological treatments generally require patients to work in therapy, the patient may feel partially responsible for the outcome. Thus, to avoid cognitive dissonance, she may consciously or unconsciously exaggerate her improvement, both in her own mind and when reporting to treaters.
Differential generalizability. Psychological study protocols, such as administering several months of a behavioral treatment, usually mimic clinical practice fairly well. This is not the case with drug study protocols.
No responsible clinician would inflexibly administer a single dosage of a single drug for a fixed period to every bulimic patient and then declare failure for all nonresponders, as is done in study protocols. In practice, the clinician can offer nonresponders augmentation strategies and additional drug trials. Thus, calculations of bulimia response rates in drug studies substantially understate response to drug therapy in clinical practice.
Table 2
How effective are medications in treating bulimia nervosa?
| Medication | Evidence for efficacy | Remarks |
|---|---|---|
| Antidepressants | ||
| Selective serotonin reuptake inhibitors | +++ | Fluoxetine is only SSRI studied in controlled trials |
| Tricyclics | +++ | Generally more side effects than SSRIs |
| Monoamine oxidase inhibitors | ++ | High rates of remission, but dietary restrictions |
| Trazodone | ++ | Only one controlled trial |
| Venlafaxine, mirtazapine, nefazodone | ? | No controlled trials, but probably effective |
| Bupropion | (++) | Not recommended; caused seizures in bulimic patients |
| Anticonvulsants | ||
| Topiramate | ++ | Only one controlled trial, but substantial effect size |
| Phenytoin | + | Little efficacy in only controlled study |
| Carbamazepine | + | May be useful in bulimia with comorbid bipolar disorder |
| Valproate | + | May be useful in bulimia with comorbid bipolar disorder |
| Other agents | ||
| Liothyronine | + | Augmentation agent in patients with incomplete antidepressant response |
| Lithium | + | Ineffective in only controlled trial; possible augmentation strategy |
| Naltrexone | 0 | Ineffective in two controlled trials |
| Ondansetron | + | One controlled trial |
| 0 No apparent efficacy | ||
| + Occasional effect; limited evidence | ||
| ++ Clear effect; good evidence from controlled trial(s) | ||
| +++ Strongly documented effect; evidence from multiple controlled trials. | ||
| ( ) Negative effect | ||
One also might note that psychological study findings have not “sold” well in the clinical practice marketplace. For example, in a recent survey of more than 220 bulimic women treated with psychotherapy, only 6.9% said they received a full course of cognitive behavioral therapy (CBT)14 —despite two decades of evidence of its efficacy. By contrast, untested, inefficacious, and possibly harmful psychotherapies for bulimia—including recovered-memory therapy—appear to be thriving.
Recommendation. Interpret with caution any head-to-head comparisons of psychological versus drug therapies—especially when clinical practice recommendations are made. Certain psychological therapies provided by specifically-trained individuals likely do help patients with bulimia nervosa. However, biases inherent to the studies may inflate psychological therapies’ efficacy when compared with that of drug therapy.
Therefore, for a psychiatrist who does not specialize in eating disorders to offer exclusively psychological therapy to a bulimic patient—while withholding or postponing drug therapy—may now be a questionable practice.
Choosing drug therapies
Although consensus is lacking on an optimal treatment trial sequence for bulimia nervosa, we suggest a rational approach based on the evidence and our experience (Algorithm).
First-line antidepressants. A selective serotonin reuptake inhibitor (SSRI) trial is usually the first choice (Table 2), and some data suggest that higher-than-usual dosages may be required. For example, in a large multicenter trial of fluoxetine in bulimia nervosa, 60 mg/d was considerably more effective than 20 mg/d for reducing binge eating behavior and vomiting frequency.15
Based on our observations, however, we believe that noncompliance or irregular compliance may account for this difference in response. Bulimic patients’ impulsive and obsessional behavior may keep them from taking their medications as prescribed. The higher fluoxetine dosage may therefore have been more effective simply because it ensured adequate plasma levels, even when patients missed or forgot multiple doses.
Augmenting agents. A first antidepressant trial rarely leads to complete remission of bulimic symptoms. This is not a serious concern, however, because many other options are available.
Liothyronine. Partial responders to SSRIs often become complete responders when we add a 10-day trial of liothyronine (T3), 25 μg/d. If this fails, we may try augmenting with lithium carbonate, although bulimic patients are often afraid of weight gain or lithium’s other side effects.