There are a small group of studies that examine the outcomes in AN that also provide some information on the effects of early intervention. In a study conducted by Eisler et al. [34], it was found that patients with AN who were less than 18 years of age and had been ill for less than 3 years (mean was 1.2 years) had better outcomes following family therapy treatment [34]. However, AN patients who had the illness for more than 3 years (mean was 5.9 years) had poorer outcomes. There was a clear association between the length of the illness and outcome; patients who had a poor outcome at the end of treatment had been ill for a significantly longer period than those who had a good outcome both at 1- and 5-year follow-ups [35, 36].

In a review conducted by Kreipe et al. (1996) examining the outcomes of three studies where the mean duration of illness prior to the onset of treatment averaged less than 12 months, excellent outcomes were reported. The authors also report that the duration of illness identified in patients treated in the adolescent medicine paediatric programmes is one-half to one-third that of patients treated in the child and adolescent psychiatry programmes, and treatment location (with more complex or chronic patients being seen by psychiatry) may be a significant factor relating to prognosis [37]. Moreover, Hall et al. conducted a follow-up study of 50 patients that they had assessed who were completely healthy at a minimum of 4 years since onset and found that only 20% of the sample were free from any physical or mental health issue. Once again, length of illness was found to be a factor related to outcome, with shorter length of illness predicting a better outcome [38]. One of the problems with this literature is that it is difficult to parse out the DUI as opposed to the duration of illness prior to the treatment that is being described. For example, comparing patients treated by adolescent medicine as compared to psychiatry. The results seem to be confounded by the greater likelihood of patients treated by psychiatry having had a more complex course. So, although they may be new patients to the psychiatry programme, they may have had previous treatment. Borrowing from researchers in psychosis, reporting on duration untreated in AN (DUAN) may improve the analysis and interpretation of the data. This of course would require a definition for the treatment.

Steinhausen [39, 40] addressed age of onset in two review papers. In the first, he compared adolescent subjects with AN to adult subjects with AN. There was a somewhat better global outcome for the adolescent patients in terms of recovery, improvement and chronicity, but there was no consensus on the prognostic value of age at onset. Only in some studies was the age associated with a better outcome [39]. In a more recent study with an expanded literature review [40], outcome in adolescent onset AN more clearly had a lower mortality rate and more favourable outcome, as measured by rates in recovery, improvement and chronicity.

Most studies of adolescents suggest that there are better outcomes compared to adults. Comparing studies of adults and adolescents is confounded by the earlier age of onset of adolescents, along with a tendency towards shorter duration. As a result, some authors believe that the shorter duration of illness is more important than the early onset [41]. Le Grange and Loeb (2007) also propose that adolescents fare better than adults because of the reduced severity and chronicity of their illness. It remains unclear if it is the shorter duration of illness, or early age of onset, or both are the variable(s) of interest.

In children and adolescents, there are three primary elements to treatment of AN: outpatient-based family therapy, inpatient treatment and psychopharmacotherapy. The National Institute for Health and Clinical Excellence (NICE) guidelines from the United Kingdom suggest that family interventions should be the treatment of choice for adolescents with AN [8, 32]. Family therapy is more effective than individual therapy in producing higher rates of remission and lower rates of relapse and hospital admissions [35, 36, 42, 43].

The best-studied family therapy intervention was originally referred to as the Maudsley method and more recently as family-based treatment or FBT [44]. This is a manualised treatment consisting of three distinct phases. The first phase focuses on helping parents take charge and manage AN-related behaviours; the second phase slowly transitions control of these behaviours back to the adolescent and the third phase focuses on the impact of AN on adolescence. Sessions last around an hour and involve the entire family. The treatment includes approximately 15 sessions delivered over 9 months, initially occurring weekly and less often as time goes on [44].

A number of other models of family therapy have been described in the literature and appear to be effective, however, the studies that exist are limited [45, 46]. One model that has spread throughout Europe and is beginning to make inroads in North America is multifamily therapy (MFT) [36]. This therapy is an intensive group based therapy where six families participate in 4 days of group therapy as families. There are 4–12 one-day follow-up days that are spread out over the course of the next 6–12 months and occur about every 1–2 months [36]. Intensive MFT in the treatment of patients with AN results in dropout rates that are extremely low, roughly 2–3%, and inpatient admission rates reduced by 30%. In addition, length of stay for inpatients has decreased by 25%, while readmissions have been cut in half [45]. In 2001, Scholz and Asen showed (in their preliminary results) that the use of MFT in adolescents with EDs was acceptable to families and produced significant positive changes in symptoms and recovery rates [45]. Currently, there is limited research examining the effectiveness of MFT in relation to symptomatic improvement.

Inpatient care may be required in the course of treatment and generally occurs when a patient is considered to be medically unstable or too ill to respond to outpatient treatment. This regularly does occur early in the course of illness or in the early stages of treatment. In some cases, this can be the first presentation to treatment and does not predict the preceding length of illness. Some adolescents become quite medically unstable very quickly and this can occur soon after onset of illness. Length of stay varies from a few days to weeks or months [47–49]. Treatment is comprised of medical stabilisation and refeeding with the aim of increasing the patient’s body weight towards a healthy weight. Treatment also includes other therapeutic interventions, most commonly family, individual, and group therapy [47–49]. Outcomes vary by study, with some predicting a good outcome in the majority of patients at 1-year follow-up [48], while more recent studies suggest as many as 60% of patients have a poor outcome at 1-year follow-up [47] and 8–32% have a poor outcome after 8 years of follow-up [50]. Outcome appears to be predicted by discharge weight, with higher weights predicting a better outcome [47, 48, 50].

Data on the effectiveness of psychotropic medication remain limited. There are no adequate well-designed studies that examine the effect of antidepressants in this population. Case reports and case series suggest that atypical antipsychotics, particularly olanzapine, may be helpful in this population [51–54]. More recent retrospective cohorts [55] and blinded and controlled studies [56] cast doubt on this hypothesis. While there are no reports of unexpected side effects [57], there appears to be no added benefits. However, the sample sizes of these studies are quite small (on average 20 patients) and may not have the statistical power to detect differences [55, 56].

In contrast to the progress made with adolescents in recent years, the treatment of AN in adults continues to present a significant challenge. To date, there are no effective evidence-based treatments for adults with AN. The relative rarity of this condition, and the very high dropout rates from treatment studies are the primary barriers to the study of treatment for AN in adults. Completed studies are rare; one recent study examining the utility of fluoxetine on relapse prevention [58] showed no benefit from fluoxetine, but also had a 50% dropout rate, thus making the results difficult to interpret. There is a limited literature examining the correlates of successful inpatient treatment in AN [59, 60], and one small double-blind study suggesting a modest effect of adding olanzapine to inpatient treatment [61]. However, beyond these few papers, a review of the literature documents no evidence of the effectiveness of any specific treatment, psychotherapeutic or pharmacologic, for AN. There is no significant data examining the effect of early intervention in AN in adults, and there remains controversy about how to establish and date onset of illness in adults.

There are three sets of treatment guidelines for AN in adults: the APA guidelines [33], the NICE guidelines from Great Britain [32] and RANZCP (Australasian) Clinical Practice Guidelines (CPGs) [62]. These provide recommendations for clinical treatment based on the best available evidence, which is slim. Both sets of guidelines emphasise the need to make a careful treatment plan, with clear goals, and matching the site of treatment to those goals. Outpatient treatment, usually consisting of cognitive-behavioural treatment (CBT) and nutritional advice, is unlikely to provide a benefit except for those patients with milder forms of the illness, although the NICE guidelines recommend an initial course of outpatient management. It is expected that patients with more than mild illness will require more intensive treatment, either in a residential or a partial hospitalisation setting.

Most intensive programmes have similar combinations of elements—behavioural change elements, including weight gain and normalised eating, and various types of psychological treatments. There is no clear evidence as to the optimal ‘basket’ of such additional treatments. Treatments that do not include weight gain and nutritional rehabilitation are not recommended. A rate of weight gain of 1.0 kg/week is typical for intensive treatment. There is no evidence that slower rates of weight gain are beneficial in intensive treatment settings.

The literature examining outcomes in patients with BN is scarce and has significant methodological shortcomings. In a systematic review conducted by Reas et al. [67] examining the prognostic value of duration of illness in BN and early intervention in BN, only five studies met inclusion criteria. Of the five studies that met methodological criteria for inclusion in the systematic review, only one of the studies found a significant negative association between the duration of illness and outcome. The other four studies in this systematic review [67] revealed no differences in duration of illness between those who had recovered at follow-up and those who did not. Steinhausen and Weber’s (2009) review of 79 studies examining outcomes in BN revealed that duration of illness had been assessed in most of the studies, but found no effect on the course of illness. Most of the studies failed to find a significant association between age at onset and outcome. There is no consistent evidence to support that illness duration and age of onset results in better outcomes in BN. Once again, the lack of evidence to support early intervention may be due to the paucity of adequate research or possibly, once again, the failure to differentiate between duration of untreated BN (DUBN) from duration of BN.

The evidence-based literature in the treatment of adolescents with BN is limited to a handful of studies. This is not surprising, given the later mean age of onset for BN. As in AN, family therapy, specifically FBT, appears to be an effective form of treatment for this population, with better outcomes at 6-month follow-up [68]. Family therapy has a similar outcome at 1-year follow-up to CBT-guided self-care [69]. A case series examining CBT in adolescents with bulimia reported that while some changes need to be made to account for developmental differences, with good adherence to treatment, an abstinence rate of 56% for binge eating and purging could be achieved [70].

The psychopharmacologic literature in BN in adolescence is very limited, with only 1 open trial examining 10 adolescents treated with fluoxetine [71, 72]. A confound in this literature is that older adolescents are typically included in adult medication trials, and their response is not described separately from the adults in the studies. Kotler et al. [71] examined 8 weeks of acute therapy with fluoxetine in 10 adolescents (average age 16.2 ± 1.2 years) with BN. The primary outcome measures were frequencies of binge eating and purging. Average weekly binges decreased significantly from 4.1 ± 3.8 to 0 (p < 0.01) and average weekly purges decreased significantly from 6.4 ± 5.2 to 0.4 ± 0.9 (p < 0.005).

In contrast to AN, there are effective evidence treatments for BN in adults. This is may be in part due to the higher incidence of BN, which allows for easier recruitment into studies. There is evidence for the efficacy of self-help and psychoeducational treatments as a first step in the course of outpatient treatment of BN [73] regardless of the length of illness. CBT for BN has been extensively studied and has excellent efficacy [74]. Other forms of psychotherapy, such as interpersonal therapy (IPT), or dialectical behaviour therapy (DBT) [75] have also been studied and found to be as efficacious as CBT. However, response to IPT may be slower than CBT [74]. There is no evidence for the efficacy of psychodynamic psychotherapy as a treatment for BN and there is limited evidence supporting the efficacy of nutritional therapy alone in the treatment of BN.