Diagnostic aspects related to the transition from DSM-IV to DSM-V

The core symptomatology of anorexia nervosa (AN) consists of an intertwining of primary behavioral features and cognitions with mental and physical symptoms due to starvation, which both differ depending on age and stage of the disorder [1, 2]. We need a broad classification for AN, which (a) is able to account for this complexity and other relevant aspects such as cross-cultural differences [3], (b) is based on empirically based, easy-to-apply and replicable diagnostic criteria for clinicians and researchers and (c) has a high sensitivity and specificity [2].

Lack of a standard or references for the weight criterion

The weight criterion “minimally normal weight for age and height” of the fourth edition of diagnostic and statistical manual of mental disorders (DSM-IV) has repeatedly been altered by the DSM-5 Working Group on eating disorders; in the last update (April 2012) it has been changed to “significantly low body weight in the context of age, sex, developmental trajectory, and physical health”; the term “body weight less than 85 % of that expected” has been eliminated. Significantly low weight is defined “as a weight that is less than minimally normal, or, for children and adolescents, less than that minimally expected” (Table 1).

Table 1 Comparison between DSM-IV and proposed DSM-5 diagnostic criteria for anorexia nervosa after revision in April 2012 (see http://www.dsm5.org/ProposedRevision/Pages/proposedrevision.aspx?rid=24)
Full size table

Underweight per se is not obligatorily associated with pathology or an increased mortality rate [2]. The necessary distinction between healthy and harmful underweight (with clinically apparent symptoms of starvation) can now be inferred from the DSM-5 reference to “significantly low weight in context of […] physical health”, that we strongly welcome. We would nevertheless argue that the term physical health is rather general; the diagnostic assessment would undoubtedly benefit from a reference to those AN-related symptoms that are not compatible with physical health. Such symptoms could be categorized as starvation-related somatic findings as proposed previously [1, 2].

In accordance with the WHO classification of different weight categories, we have proposed the use of the body mass index (BMI) threshold of 18.5 kg/m² in adults, which at age 18 roughly corresponds to the 10th BMI age centile in US and European populations entailing that this centile can be used to define underweight in minors. The lack of a standard method to validly measure energy intake and expenditure [4] in our opinion precludes the pseudoscientific reference to the very first and thus central current DSM-5 A-criterion term “restriction of energy intake relative to requirements”. A clinician can merely infer that this restriction is operative in a particular patient [2]. A delineation of behavioral symptoms that can entail a negative energy balance or more generally a too low caloric intake would provide the clinician with a readily observable framework for judging this criterion. Inclusion of symptoms indicative of a disordered eating behavior would make reference to the fact that AN is classified as an eating disorder.

Removal of stigmatizing terms

Therapeutic management, family support and treatment outcome potentially depend on the psychiatrist’s, patient’s, relative’s and even societal conceptualization and understanding of a particular disorder. Obviously, the diagnostic criteria are of utmost importance for such conceptualizations. Based on previous criticism the terms “refusal to maintain body weight at or above a minimally normal weight for age and height” in criterion A and “denial of the seriousness of the current low body weight” in criterion C were replaced in DSM-5 (see Table 1), because both words imply an empirically unsubstantiated deliberate willful action of the patient and additionally convey a paternalistic and pejorative attitude [1, 5].

Lack of developmental and cultural considerations in the B and C criteria

According to DSM-5, the clinician is dependent on the patient’s endorsement of weight phobia and body image disturbance or their inference; however, many underweight eating disordered patients—in particular at younger ages—do not report an intense fear of gaining weight or body image disturbance [5, 6]. The DSM-IV diagnostic criteria are not sensitive to the timing of neurocognitive maturation in children and adolescents with AN; the cognitive capacity for complex abstract reasoning usually develops during adolescence [6]. Furthermore, the endorsement of the symptom “fear of weight gain” depends on illness stage and culture; for example, the proportion of Chinese AN patients reporting weight phobia increases with exposure to Western cultural values [7].

It is fortunate that the DSM-5 criterion B is alternatively fulfilled if the additional term “persistent behavior that interferes with weight gain” is met. With this alternative, we have a readily detectable clinical symptom at hand which is independent of weight phobia. On a theoretical basis, it would suffice to merely pay reference to this persistent behavior to avoid weight gain, because weight phobia can be subsumed within this term.

Except for the novel term “persistent lack of recognition of the seriousness of the current low body weight”, the DSM-IV criterion C was not changed (see Table 1). However, there is no definition or instruction for the clinician to judge the “seriousness of the current low body weight”. In light of the fact that even clinicians will not always agree to the seriousness of the current low body weight of an individual patient, it appears all the more problematic to additionally require the persistent lack of recognition by the patient.

Reference to symptoms of starvation

The DSM-IV D criterion (amenorrhea) is not applicable to males, prepubescent and postmenarcheal females and women who take contraceptives. Amenorrhea may also not be reliably reported by patients [6, 8]. The DSM-5 Eating Disorder Working Group has thus proposed to eliminate the D criterion. Hence, in the proposed DSM-5 classification there is no single somatic (e.g. loss of fat mass, low pulse rate, low blood pressure, amenorrhea) or psychological symptom (e.g. irritability, preoccupation with food, depressed mood), which pertains to the starvation-related neuroendocrine dysfunction characteristic of AN [2].

Hyperactivity, which figures prominently in several patients with AN, is not included as a possible symptom of the disorder. It is potentially indirectly and to our opinion poorly referred to in the DSM-5 A criterion, which refers to “restriction of energy intake relative to requirements” [2]. In addition, in comparison to other psychiatric disorders unique opportunity to include endocrinological markers (i.e. hypoleptinemia) with high specificity for the diagnosis of AN in adolescence and early adulthood [9] was not considered.

Usefulness of the restricting and binge eating/purging subtypes

The DSM-IV AN subtypes are not stable over time [10]; modifications have been proposed to allow for the crossing-over between the subtypes [1, 2, 6]. Therefore, the DSM-5 Working Group has recommended that the subtyping be specified for the last 3 months (instead of the former reference to “during the current episode”, see Table 1); 3 months is the timeframe used for Bulimia Nervosa (BN) and the novel diagnosis of binge eating disorder. Nevertheless, because of the initial presentation of young patients with mainly the restricting type of AN [11] the subtypes are of limited value only for further delineating the symptomatology in children and adolescents [12].

Reduction of the high percentage of eating disordered patients who receive a diagnosis of EDNOS

The main current classification problem resides in the high percentage of patients with a clinically relevant eating disorder associated with functional impairment, who do not fulfill all criteria for AN or BN and who are thus subsumed under the diagnostic label of eating disorders not otherwise specified (EDNOS) [5, 13]. Thus, in several recent clinical and epidemiological studies well over 50 % of adolescents and adults presenting for eating disorder treatment are diagnosed with EDNOS [13]. The high prevalence of EDNOS is additionally problematic because of the clinical severity (e.g. higher prevalence of comorbid psychopathology in adolescent patients with EDNOS then in patients with AN) and heterogeneity (e.g. pathology, demographics, comorbidity). Furthermore, specific treatment strategies for EDNOS have only infrequently been recommended; the evidence for such recommendations is weak [13]. To reduce EDNOS as a residual category, several research groups support modifying the current DSM-IV criteria for AN and BN to increase the validity of the classification system, which has implications for clinical research and practise as well as for public policy [2, 13].

In college students, the application of the proposed DSM-5 revision (update April 2010) of the diagnostic criteria for AN, BN and the addition of binge eating disorder as a separate diagnostic category led to the desired reduction of the rate of EDNOS [14]. Birgegard et al. [15] confirmed this reduction in a large naturalistic database (children, adolescents and adults at specialized eating disorder clinics); however, the problem of a still high percentage of EDNOS remained, especially in younger patients.

Conclusions

In our opinion, the changes in the latest DSM-5 revision represent a clear improvement for the diagnosis of AN. We strongly welcome that underweight is associated with age, sex, developmental trajectory and physical health in the weight criterion and that an easy to detect “persistent behavior that interferes with weight gain” has been included in criterion B. Nevertheless, we would support further changes in light of the fact that DSM V will apply for the upcoming 15–20 years. In our opinion, major issues that could still be addressed include: (a) omission of the pseudoscientific and difficult to verify term “restriction of energy intake relative to requirements”, (b) reference to starvation-related (mental and physical) symptoms would provide a clinician with the means to judge if energy intake is indeed restricted relative to requirements and would no longer require reference to the very general novel term ‘physical health’, (c) a standard or reference(s) for the weight criterion, (d) better delineation of behaviors and cognitions readily observable in AN patients thus also substantiating the classification as an eating disorder, (e) avoidance of the term seriousness of the current low body weight as long as no explanation is provided as to how to judge “seriousness”. We deem this discussion very worthwhile to achieve higher diagnostic sensitivity, specificity and reliability.

Guideline based treatment with a focus on recommendations for inpatient versus outpatient treatment

The methodological rigor of the NICE (National Institute for Health and Clinical Excellence; [16]) guidelines for the treatment of eating disorders has been contrasted with that of the current Practice Guideline for Eating Disorders (PGED) of the American Psychiatric Association. Unlike NICE, the PGED does not detail criteria for evaluating supporting research. Instead of making clear recommendations, PGED is uncritically inclusive and emphasizes subjective judgment of individual clinicians [17]. The NICE guidelines, which were methodologically processed in five basic steps [16], balance recommending specific treatments against the importance of clinical judgment when guideline recommendations are not readily applicable [17]. According to NICE, “Grade A” evidence requires at least one randomized controlled trial as part of a body of literature of overall good quality and consistency addressing the specific recommendation (evidence level I) without extrapolation. “Grade B” is defined by well-conducted clinical studies but no randomized clinical trials on the topic of recommendation (evidence levels II or III) or extrapolated from level I evidence. Finally, “Grade C” is based on expert committee reports or opinions and/or clinical experiences of respected authorities (evidence level IV) or extrapolated from level I or II evidence. Despite the fact that guidelines have been published in other languages by national societies (e.g. German Guidelines for Eating Disorders) we focus on the NICE guidelines to illustrate guideline-based treatment. We particularly address one of the three key NICE priorities, according to which AN patients should be treated on an outpatient basis.

NICE recommendations for eating disorders

The currently very limited evidence for treatment strategies for AN is based on few studies only. Over the past 30 years, 23 controlled trials have investigated psychological treatments [18]. The infrequency of the disorder, the presence of medical complications that sometimes require inpatient management, the prolonged course and the extended period of treatment necessary for full symptom remission entail a deficit of controlled treatment research in AN. The problems of sample recruitment, attrition prevention and secure participation in follow-up assessments are amongst other things caused by the patients’ ambivalent attitudes about recovery [19]. In accordance with our limited knowledge of the optimal treatment for AN, 50 of the 51 recommendations in the NICE guideline were defined as “Grade C”. The strongest recommendation “Children and adolescents should be offered family interventions that directly address the eating disorder” was graded “B”.

The 17 recommendations for BN comprise one A and seven and nine B and C gradings, respectively. The strongest recommendation (“Grade A”) in BN refers to cognitive behavioral therapy (CBT), because of evidence of its superiority to other psychological and drug treatments. In contrast to AN (no evidence), the use of anti-depressants in BN was graded B.

Out of the ten recommendations for ‘atypical eating disorders’, two were defined as “Grade A”: (1) the recommendation for CBT in patients with binge eating disorder and (2) the recommendation that “patients should be informed that all psychological treatments for binge eating disorder have a limited effect on body weight”. Out of the other eight recommendations five received “Grade B” and three “Grade C”.

The following three “key priorities” are highlighted within the NICE guidelines for AN [16]:

  1. 1.

    “Most people with AN should be managed on an outpatient basis with psychological treatment provided by a service that is competent in giving that treatment and assessing the physical risk of people with eating disorders” (Grade C).

  2. 2.

    “People with AN requiring inpatient treatment should be admitted to a setting that can provide the skilled implementation of refeeding with careful physical monitoring (particulary in the first few days of refeeding) in combination with psychosocial interventions” (Grade C).

  3. 3.

    “Family interventions that directly address the eating disorder should be offered to children and adolescents with AN (Grade B).”

Outpatient versus inpatient treatment in anorexia nervosa

During the course of AN, clinicians have to repeatedly decide which setting is optimal for the patient. The majority of referred AN patients are treated as inpatients at least once during the course of their eating disorder; e.g. this applies to 50 % of the patients in the United States [19]. In Germany, 10–20 % of all AN patients are treated as inpatients each year based on the evaluation of benefit data from statutory health insurance, statutory pension insurance and a review of literature of the epidemiology of AN [20]. In both the United States and Australia, patients are generally admitted for short-term medical stabilization on a medical or pediatric ward [16]. In Germany, inpatient treatment is frequently pursued until target weight is achieved, entailing hospitalization durations frequently exceeding 3 months.

NICE recommendations for treatment setting in AN

The NICE guidelines provide the following Grade C recommendations [16]:

  1. 1.

    “Most people with AN should be treated on an outpatient basis.”

  2. 2.

    “Inpatient treatment or day patient treatment should be considered for people with AN, whose disorder has not improved with appropriate outpatient treatment, or for whom there is a significant risk of suicide or severe self-harm.”

  3. 3.

    “Inpatient treatment should be considered for people with AN, whose disorder is associated with high or moderate physical risk.”

Critical reflection of the NICE recommendation for priority of outpatient treatment

The typical duration of AN in an individual patient extends for many months [21]; obviously, even if a patient is admitted as an inpatient during the course of the disorder, they will be outside of a hospital for a substantial proportion of the total duration of the disorder. The NICE recommendation “most people with AN should be treated on an outpatient basis” is phrased somewhat ambiguously because it is unclear if of all people with AN the majority should be treated as outpatients during the total course of their eating disorder or alternatively if at any given time point most patients with AN should be treated as outpatients. Additionally, the lack of research evidence has led NICE to conclude that no specific treatments can be recommended, but on the other hand NICE recommended (“Grade C”) that most patients with AN should be managed as outpatients. The NICE guidelines do not provide a definition for “a high or moderate physical risk” or for an “adequate outpatient treatment” (see above). In contrast to other guidelines, NICE did not define any specific BMI threshold entailing an outpatient or inpatient treatment. The APA guidelines recommend an inpatient treatment if a BMI is ≤16 kg/m², whereas in the UK the BMI threshold is lower (BMI <13 kg/m²; [16]). According to the German S3 Guideline on the diagnosis and treatment of eating disorders [22] the following criteria should entail an inpatient treatment: a rapid and continuous weight loss (>20 % over 6 months), a BMI less than 15 kg/m² (or <3rd age matched BMI percentile in children and adolescents).

The NICE recommendations for inpatient/outpatient treatment mainly rely on a single RCT [23, 24] and three reviews [25–27] that specifically addressed the treatment setting (one of these compared only day hospitalization programs [27]). The single RCT [23, 24] encompassed 90 consecutively referred female patients. Inclusion criteria were a DSM-III-R diagnosis of AN and an eating disorder’s duration <10 years. The exclusion criterion was based on an address located more than 40 miles from the treatment center. Psychotropic drugs were not prescribed to any of the patients. At the end of a diagnostic assessment interview, they were randomized to four treatment arms:

  1. 1.

    Inpatient treatment lasted “several” months (not specified) and involved weight restoration to the mean matched-population weight (MMPW). This treatment included weekly individual therapy, family therapy, group therapy, dietary counseling and occupational therapy, including psychodrama and projective art techniques and was followed by 12 sessions of outpatient treatment.

  2. 2.

    Outpatient individual and family psychotherapy plus separate dietary counseling: Each patient received 12 outpatient sessions which included to a variable extent individual and family therapy according to the needs of the respective patient. Initially, sessions were at 1 or 2 week intervals. They were gradually spaced. For treatment completers treatment covered approximately 10 months.

  3. 3.

    Outpatient group psychotherapy for patients and parents separately included ten outpatient meetings with the patients and ten group meetings for parents separately at monthly intervals. Additionally, dietary counseling and advice to promote a slow steady weight gain were offered on four occasions within the outpatient treatment packages 2 + 3.

  4. 4.

    If patients were allocated to the option “no further treatment”, they were referred back to their general practitioners.

The patients were not permitted to reject a treatment arm in favor of another and there was no change of treatment arm as a result of lack of progress. The randomization resulted in no significant differences across the four groups for age, weight, height, MMPW, duration of AN, presence of vomiting or purging, or Morgan and Russell mean global score.

The mean age of the sample was 22 years. The mean length of illness was 39 months (range 4–107 months) and the mean deviation below MMPW % in the inpatient group was 28 versus 26.4 (non-significant difference) in the two outpatient groups. The inpatient group consisted of 30 patients, 12 of whom dropped out during the year. Two and three patients dropped out of the two outpatient groups (20 patients each). Outcome measures after 1 and 2 years were deaths, weight, MMPW, and the Morgan and Russell mean global score.

Both after 1 and 2 year follow-up, significant improvements in % MMPW and Morgan and Russell mean global scores were found for each arm; however, no significant differences were observed between the inpatient and outpatient groups. Weight changes were significantly better for the three treatment groups than for the assessment only group.

The following limitations of this RCT should be pointed out: (1) The assessment only treatment arm was confounded by patients, who sought treatment elsewhere. (2) The apparently good outcome in the outpatient group included two patients who developed binge eating and became obese. (3) The study lacked power (20–30 patients/treatment arm). (4) Neither the patients nor the assessors were blind to treatment allocation.

After publication of the NICE guidelines, one additional RCT that compared inpatient and outpatient treatment was published between 2004 and 2012 [28]. This RCT focused on children and adolescents aged 12–18 years with a DSM-IV diagnosis of AN. The results showed no superiority of one treatment setting over the other and were thus similar to the earlier RCT of Gowers and coworkers [23, 24]. The mean cost for specialized outpatient treatment was lower than that for inpatient care. The major limitation of this RCT was the poor adherence to randomization.

Cost considerations

Obviously, economic considerations play an important role in the light of poor evidence for one or the other treatment setting. AN is associated with a significant cost burden, because physical comorbidities can develop including osteoporosis, cardiovascular problems and renal failure. Psychiatric comorbidities are anxiety disorders, obsessive compulsive disorders, mood disorders and personality disorders; a risk of suicidality must also be considered. Mortality is elevated as well [29]. The data of Striegel-Moore et al. [30] show that, per patient, treatment for AN (and BN) was comparable in costs per year to treatment of schizophrenia. However, cost comparisons between studies are hard to accomplish due to different methods of cost estimation. Comprehensive evaluations of direct and indirect costs are still lacking [31].

Conclusion

The literature reviewed in the NICE guidelines do not strongly substantiate the Grade C level recommendation of outpatient treatment in AN. Outpatient treatment regimens reviewed in the NICE Guidelines differed markedly both in the frequency of treatment and the therapeutic methods. Thus, the comparability is limited. In our opinion, future research is required to differentiate between patients who profit from inpatient treatment versus those for whom outpatient treatment suffices. If inpatient treatment of a particular subgroup is proven more effective than outpatient treatment, the required length of hospitalization, the mode of treatment and treatment goals need to be determined. The characteristics of this subgroup need to be delineated. In the light of potentially low participation, low adherence and high drop-out rates we are aware of the difficulties such research efforts pose. Nevertheless, we need to advance the currently limited knowledge as to appropriate treatment strategies; multi-center studies are inevitable to achieve this goal.