Weight outcomes for adolescents with atypical anorexia nervosa in family-based treatment

Abstract

Background

Although over one-third of adolescents presenting with restrictive eating disorders have a history of being overweight, there is no evidence-based treatment for atypical anorexia nervosa (AAN). Family-Based Treatment (FBT) is a feasible treatment and is routinely applied to treat atypical anorexia nervosa in adolescents; however, identifying a treatment target weight within FBT for these patients is a challenge.

Objective

This study aimed to 1) increase understanding regarding recommendations for weight gain versus weight stabilization in FBT for adolescents with AAN and 2) examine treatment outcomes in FBT for adolescents with AAN.

Method

Using a retrospective design, we reviewed the files of 41 patients with AAN who were referred for FBT at a pediatric eating disorder program located within a tertiary care health centre.

Results

We found variability in recommendations for weight gain, with 56% of the sample recommended to gain weight and 44% recommended to stabilize weight. Baseline BMI for age appeared to be a key factor in establishing recommendations for weight gain. AAN patients in our sample gained a significant amount of weight across treatment, with those recommended to gain weight showing more weight gain during treatment. Forty-nine percent of the sample completed FBT; those patients displayed a mean of 10kg of weight gain during treatment.

Conclusions

Findings suggest that many patients gained weight during the course of FBT for AAN. Further study on weight changes during FBT for adolescents with AAN and increased diagnostic consistency for AAN will be important for this field.

Introduction

A large proportion (37%; 1) of adolescents presenting with restrictive eating disorders have a history of being overweight. Many of these previously overweight adolescents may be captured by the relatively new atypical anorexia nervosa (AAN) diagnostic profile, which was introduced in the Diagnostic and Statistical Manual of Mental Disorders – Fifth edition (DSM-5) in 2013 and defined as meeting “all of the criteria for anorexia nervosa … except that despite significant weight loss, the individual’s weight is within or above the normal range” (2). AAN is equally or more common than AN but is often underrepresented in clinical settings (3). Many factors may contribute to this, including bias against identifying restrictive eating problems and associated medical complications in normal- or over-weight youth, AAN being a relatively new diagnosis that was absent from previous editions of the DSM, and AAN being only one of several specifiers within the broader Otherwise Specified Feeding and Eating Disorders (OSFED) diagnostic category, rather than its own diagnostic category. Even when AAN is suspected, assessing what constitutes “normal weight” and “significant weight loss” often poses difficulty for diagnosing clinicians, and leads to inconsistency in diagnosing AN and AAN (e.g., 3, 4).

When studying AAN, many researchers have used more objective numerical criteria to attempt to clarify definitions for “normal weight” and “significant weight loss” (for a review, see 3). Sawyer and colleagues (5), for instance, applied cut-offs to define “normal weight” as ≥90% median body mass index for age (mBMI), and “significant weight loss” as ≥10% body weight in the past year. Using these criteria, they found that adolescents with AAN have similar medical complications to adolescents with AN, and are admitted to hospital at similar rates to adolescents with AN. Interestingly, they also showed that adolescents with AAN displayed more severe psychological eating disorder symptoms compared to those with AN. Many researchers have found similar results, even when using slightly different numerical criteria (e.g., 3, 6) – further highlighting that adolescents with AAN have equal or greater eating disorder symptomology and equal medical risks to those with AN, despite being within a normal weight range.

There is currently no evidence-based treatment for AAN. Family-based treatment (FBT; 7) has the strongest evidence for treatment of adolescent AN (8, 9). Weight gain is a key focus of FBT, and it has been less clear how much weight gain is required to recover from AAN. Hughes and colleagues (10) have shown FBT to be a feasible treatment for adolescents with AAN, with 37.5% of patients achieving full remission, and another 25% achieving partial remission. These remission rates are similar to those observed in studies of AN (8). Of note, Hughes and colleagues (10) found that adolescents with AAN, especially those who were hospitalized prior to starting treatment, displayed little overall change in weight over the course of FBT; however, patients displayed reductions in eating disorder symptoms and behaviours across treatment. There were no explicit guidelines about setting target weights for adolescents with AAN in this study, but the authors noted that the approach tended toward weight stabilization or modest weight gain (10), which is quite different from the focus on weight gain within FBT for AN (7).

In a recent qualitative study (11), a sample of FBT-trained clinicians were unanimous in identifying FBT as an appropriate “first-line” intervention for AAN. Practitioners reported markedly different approaches to weight gain, with some aiming for weight restoration and others weight stabilization. Indeed, practitioners indicated that identifying the goal weight for these patients was the most salient challenge when delivering FBT for adolescents with AAN. These reflections highlight the need for further study of the efficacy of FBT for AAN, and consideration of guidelines for applying this model to AAN.

One factor influencing treatment target weights that has been highlighted in the literature is weight suppression (i.e., the difference between the highest previous weight at the current height and the current weight). Recent research indicates that psychological symptoms (12) and medical complications (6, 13) among patients with AN and AAN are better predicted by amount of total weight loss than by baseline assessment weight. As such, weight suppression has been identified as a key indicator of illness severity and amount of weight gain needed for health restoration (e.g., 14). In line with this recommendation, weight suppression has been shown to be positively associated with both rate of gain and total weight gain for older adolescent and adult females engaged in eating disorder treatment (15, 16). Weight suppression may be a particularly important consideration amongst patients with AAN, as research with previously over-weight adolescents has shown that these individuals need to gain a similar amount of weight as their non-overweight peers and restore lost weight (therefore reaching a higher absolute weight than same-aged peers) prior to demonstrating signs of physical recovery, such as return of menses (17).

The utility of weight suppression is more limited, however, when setting treatment targets for children and younger adolescents who are still growing. Some youth may meet criteria for AN or AAN due to failing to make expected developmental gains, rather than due to weight loss, and other youth in this age range experience weight loss compounded with failure to make expected gains. As a result, when setting treatment target weights for youth it is especially important to use an individualized approach that takes a wide variety of factors into consideration (18, 19). When considering previous weight history, examining discrepancies between baseline and historical growth relative to percentile curves (when available) may be more informative than absolute values of weight suppression. Data on historical growth relative to percentile curve is often limited (e.g., 6, 16), however, making setting treatment target weights for children and adolescents particularly difficult.

Given the specific difficulty of establishing treatment target weights within FBT for children and adolescents with AAN, we designed the current study to 1) investigate recommendations for weight gain versus weight stabilization in FBT for adolescents with AAN and 2) examine treatment outcomes in FBT for adolescents with AAN in a real-world hospital setting. We utilized a retrospective file review to accomplish these goals.

Methods

Setting

This study took place at a child and adolescent eating disorder program located within a tertiary care health centre. Patients could be referred to the program by themselves, their families, their family physician, or a health care provider within the health centre. To aid in the intake and triage process, patients were encouraged to participate in a medical evaluation with their family physician or a walk-in clinic prior to the multidisciplinary assessment (e.g., to identify youth that may need to be directed to the Emergency Department for urgent care/inpatient hospitalization, or to be assessed by the outpatient clinic on a ‘priority’ basis). Multidisciplinary assessments within the inpatient psychiatric unit and outpatient eating disorder clinics included evaluation by a psychiatrist or psychologist, dietitian, and nurse, as well as detailed review of referral information and any available prior psychiatric and medical records (e.g., historical growth charts). Prior to mid-March 2020 all assessments were conducted in-person, typically within a half day period. After the onset of the COVID-19 pandemic in mid-March 2020, there was a brief period during which some portions of the multidisciplinary assessment (e.g., clinical interview) were conducted via videoconference, but the team quickly returned to conducting assessments in person and attempted to maintain previous standards of practice as much as possible. Pandemic protocols at times prevented more than one caregiver from being permitted to accompany the youth to the assessment, but additional caregivers were included whenever possible (e.g., having caregivers join via videoconference/phone, requesting management approval for in-person attendance). If diagnosed with an eating disorder after completing a multidisciplinary diagnostic assessment with either the outpatient team or the inpatient unit, patients would be eligible for outpatient FBT. Treatment would commence if/when patients were sufficiently stable medically to be managed on an outpatient basis.

Manualized outpatient FBT was delivered by specialist outpatient mental health clinicians who had either obtained certification in the model or were under active weekly supervision with certified FBT therapists and/or a certified FBT trainer. To maintain fidelity to the model post-certification, all certified FBT clinicians in the program participated in a weekly consultation group. The FBT clinician acted as the head of the treatment team and the youth’s family physician or nurse practitioner was responsible for medical monitoring throughout treatment. Other members of the treatment team typically included a clinic-employed nurse who was responsible for consulting with family physicians/nurse practitioners and FBT therapists about eating disorder-specific medical recommendations and questions, and a clinic dietitian who was responsible for establishing and updating treatment target weights. The nursing and dietetics roles were largely consultative, with direct clinical contact only occurring when specifically requested (e.g., urgent vitals checks by nursing). If any other clinician was involved in the patient’s care (e.g., psychiatrist), they would also be included in the treatment team. Because ongoing medical monitoring was not conducted within the clinic, obtaining consistent collaboration and documentation from family physicians (including updated height measurements) could be a challenge. In cases where youth did not have a family physician, medical monitoring was typically conducted at a walk-in clinic (in consultation with the eating disorder clinic nurse) but was in some instances conducted by nursing or psychiatry staff within the eating disorder clinic. FBT was exclusively conducted in-person prior to mid-March 2020, but after the onset of the COVID-19 pandemic, FBT was often conducted via videoconference (due to a variety of factors, including gathering limits, masking mandates, family preference, and space limitations associated with cleaning protocols).

Participants

Participants included patients who were assessed at this child and adolescent eating disorder program between March 2015 and July 2020. This retrospective chart review procedure received REB approval from the health centre (#1025788) and the need to obtain consent was waived due to the retrospective design / lack of ongoing relationship at the time of the file review, minimal perceived risk to participants, and risk of introducing bias into the results if a segment of the patient population could not be reached for consent. We screened 144 patients who had been referred for FBT during this time period by applying an established diagnostic algorithm (5) retrospectively to patients’ multidisciplinary outpatient assessment report and/or multidisciplinary inpatient unit discharge report. Adolescents with AAN were defined as those who had lost at least 10% of their body weight, and were ≥ 90% of median body mass index (mBMI) for their age and sex (20). These cut-offs were selected based on Society for Adolescent Health and Medicine clinical guidelines for classifying malnutrition in eating disorders (21). Adolescents also needed to meet criteria for intense fear of gaining weight/compensatory behaviours and disturbance in perception of body weight/shape. The first author completed the screening for all 144 patients, while the second author completed screening for a random subset (10%) of patients. Inter-rater reliability was high (14/15; 93.3%), with the sole difference being due to a discrepancy in how target weights were reported in assessment reports from 2015 to 2017. Once this was flagged, the first author re-screened all patients assessed from 2015–2017, and inter-rater reliability increased to 100% for the same subset. We identified 41 patients who met criteria for atypical AN and had completed treatment by April 2021.

For those 41 patients, a full file review was conducted based on electronic records which included the initial multidisciplinary outpatient assessment and/or inpatient unit discharge report, treatment progress notes, dietitian consultation throughout treatment, and discharge summary. The first author completed the full file review, with input from the registered nurse and dietitian for clarification.

Measures

Recommended weight gain

Treatment target weights were set by the clinic dietitians at the start of FBT. When setting treatment target weights, clinic dietitians used an individualized approach, pulling information from age and sex-based norms; historical growth curve data from family physician/medical records (when available); family report of pre-morbid weight, height, and eating history during assessment; medical records/results; and family history. Treatment target weights were used to calculate recommended weight gain as well as to classify participants into “weight gain recommended” or “weight stabilization recommended” groups. Weight stabilization was defined as less than 1kg weight difference between baseline weight and treatment target weight. Weight gain was defined as more than 1kg difference between baseline weight and treatment target weight.

Treatment completion

We classified participants as “treatment complete” or “treatment not-complete” based on information provided by the treating clinician in their final FBT session note or discharge summary. Reasons for classification as treatment not completed included: family declined to continue FBT, referral to another eating disorder treatment without having completed FBT, non-attendance/no contact, or family moved out of service area.

Weight changes

Median BMI (mBMI) was defined as 50^th percentile BMI by age (in 6 month increments) and gender (20). Baseline BMI was calculated as baseline weight (kg)/baseline height (m²). Discharge BMI was calculated as discharge weight (kg)/discharge height (m²). (Note that discharge height was defined as the last updated height available within the patient chart.) Baseline percentage of mBMI (% mBMI) was calculated as baseline BMI/median BMI for age and gender x 100. Discharge percentage of mBMI (% mBMI) was calculated as discharge BMI/median BMI for age and gender x 100. Weight change in kilograms (kg) was calculated as discharge weight (kg) – baseline weight (kg); Weight change in % mBMI was calculated as discharge % mBMI – baseline % mBMI. We also examined discharge weight (kg) as a percentage of baseline weight (kg), treatment target weight (kg), and premorbid highest weight (kg).

Covariates

We included a number of covariates at baseline, including length of illness, psychiatric comorbidity, age, gender, pre-morbid highest weight (kg), and weight suppression. Weight suppression in kilograms was calculated as pre-morbid highest weight (kg) – baseline weight (kg). Percentage of weight suppression was calculated as weight suppression (kg)/pre-morbid highest weight (kg) x 100.

Statistical analysis

Descriptive statistics of participants at assessments were completed. Recommended weight gain groups and treatment completion groups were compared on baseline covariate measures using independent t-tests. We used paired samples t-tests to compare baseline weight to discharge weight for the entire sample, by recommended weight gain categories, and by treatment completion categories. We used independent t-tests to compare treatment completion groups on discharge weight metrics. Finally, we used individual-level data to create graphs that included main weight outcomes.

Results

Of the 144 patients who were referred for FBT during the study timeframe, 41 (28.5%) met criteria for AAN based on the study criteria. Of note, at clinical assessment, only 12 (29.3%) of the 41 were actually given a diagnosis of AAN. Twenty-three (56.1%) participants were diagnosed with AN, while 6 (14.6%) participants received a diagnosis of OSFED or unspecified ED.

Table 1 presents the participant characteristics of the sample (n=41). Participants ranged in age from 8 to 18 years, with a mean age of 14.7 years. Most were female (85.4%). Participants completed a mean of 14.85 sessions (SD=9.91) of FBT.

Table 1.

Participant characteristics (n=41)

	% (n)	Mean (SD)	Min, Max
Age at presentation		14.68 (1.95)	8, 18
Female	85.4 (35)
Baseline weight
Weight (kg)		55.0 (9.6)	28.5, 72.7
% mBMI		102.7 (8.8)	90.0, 125.3
Premorbid weight
Highest weight (kg)		68.9 (16.44)	29.8, 124.3
Weight suppression
Weight loss (kg)		14.2 (9.7)	3.2, 52.2
% weight loss		19.8 (8.2)	10.0, 42.0
Psychiatric comorbidity	36.6 (15)
Length of illness (months)		11.2 (11.5)	2, 60
Inpatient admission prior	14.6 (4)

We found that 18 participants (43.9%) had treatment target weights less than 1kg more than their baseline weight, and were thus recommended to stabilize or maintain their weight (“weight stabilization”) during treatment, while 23 participants (56.1%) had treatment target weights more than 1kg above their baseline weight, and were thus recommended to gain weight (“weight gain”) during treatment. Table 2 presents baseline characteristics by weight change category. We found that those in the stabilization group had higher baseline % mBMI compared to those in the weight gain group (109% vs. 98%). There were no significant differences between the group in terms of baseline weight or premorbid highest weight in kilograms. Of note, premorbid weight was not available in % mBMI due to limitations in these data (exact age and height not available). Similarly, the groups did not differ on metrics of weight suppression, length of illness, age, comorbidity, or inpatient admission. We did find that a higher proportion of those in the weight gain group were female (96%) compared to the stabilization group (72%).

Table 2.

Baseline characteristics by weight change category and whether completed FBT

	Weight stabilization recommended (n=18) Mean (SD)	Weight gain recommended (n=23) Mean (SD)	t (p)	FBT not completed (n=21) Mean (SD)	FBT completed (n=20) Mean (SD)	t (p)
Age at presentation	14.1 (2.2)	15.2 (1.6)	1.87 (0.1)	15.0 (2.2)	14.3 (1.6)	1.2 (0.3)
Length of illness (months)	11.7 (14.1)	10.8 (9.3)	0.23 (0.8)	11.0 (9.5)	11.4 (13.7)	0.11 (0.9)
Baseline weight
kg	58.3 (12.1)	53.0 (6.4)	1.9 (0.1)	57.0 (10.7)	52.8 (8.0)	1.42 (0.2)
% mBMI	109.2 (7.8)	97.7 (5.4)	5.6 (<0.001)	104.2 (9.3)	101.0 (8.1)	1.2 (0.3)
Premorbid weight
Highest (kg)	73.7 (21.5)	64.9 (9.6)	1.6 (0.1)	71.1 (19.8)	66.4 (11.7)	0.9 (0.4)
Weight suppression
kg	16.5 (13.2)	12.4 (5.4)	1.4 (0.2)	15.3 (12.5)	13.0 (5.6)	0.8 (0.5)
%	20.4 (9.8)	19.3 (6.9)	0.5, (0.7)	19.6 (9.3)	20.0 (7.1)	0.2 (0.9)
Recommended weight gain
kg	0.1 (0.8)	8.2 (3.7)	9.4 (<0.001)	3.9 (5.6)	5.3 (4.3)	0.9 (0.4)
%mBMI	0.004 (1.4)	15.3 (7.2)	8.4 (<0.001)	8.0 (10.5)	10.0 (8.5)	0.6 (0.5)
	% (n)	% (n)	χ2 (p)	% (n)	% (n)	χ2 (p)
Female	72.2 (13)	95.6 (22)	4.4 (0.04)	90.5 (19)	80.0 (16)	0.9 (0.3)
Psychiatric comorbidity	44.4 (8)	30.4 (7)	0.9 (0.4)	38.1 (8)	35.0 (7)	0.04 (0.8)
Inpatient admission prior	16.7 (3)	13.0 (3)	0.1 (0.8)	19.0 (4)	10.0 (2)	0.7 (0.4)

Bold data represents statistically significant at p<.05

We found that 21 participants (51.2%) did not complete treatment (due to declining to continue treatment, being referred to another treatment modality, not attending appointments, or moving out of catchment area) and were categorized as “treatment non-completers.” Treatment non-completers attended on average 9.7 (SD=7.86) sessions. We found that 20 participants (48.7%) were categorized as having completed FBT and were categorized as “treatment completers.” Treatment completers attended on average 20.3 (SD=9.05) sessions. Table 2 also presents baseline characteristics by treatment completion group. Note that these two groups did not differ significantly on age, gender, mental health comorbidity, length of illness, highest weight, weight suppression, baseline weight, or recommended weight gain.

Table 3 presents weight change data from baseline to discharge. Results for the full sample showed a significant increase in mean weight across treatment. Results indicated that both the stabilization group and the weight gain group displayed a significant increase in mean weight across treatment. However, only the weight gain group showed a significant increase in mean % mBMI during treatment. Similarly, while both treatment completers and non-completers displayed a significant increase in mean weight across treatment, only the treatment completion group were found to have a significant increase in mean % mBMI during treatment. A post-hoc analysis revealed that weight change in % mBMI was significantly higher for the weight gain group (10.56% mBMI) compared to the stabilization group (4.43% mBMI), t=−2.28, p=.03. Similarly, weight change in % mBMI was significantly higher for the treatment completion group (13.0%) compared to the treatment non-completion group (2.98%), t=−4.45, p<.001).

Table 3.

Mean weight from baseline to discharge, by weight change category and whether completed FBT

	Baseline Mean (SD)	Discharge Mean (SD)	t (p)
Total sample (n=41)
kg	55.0 (9.6)	60.3 (10.7)	6.4 (<0.001)
% mBMI	102.7 (8.8)	110.5 (11.1)	5.8 (<0.001)
Weight stabilization (n=18)
kg	58.3 (12.1)	62.3 (14.1)	2.7 (0.02)
% mBMI	109.2 (7.8)	113.6 (11.6)	2.0 (0.06)
Weight gain (n=23)
kg	52.3 (6.1)	58.8 (7.2)	7.2 (<0.001)
% mBMI	97.5 (5.5)	108.1 (10.3)	7.0 (<0.001)
FBT not completed (n=21)
kg	57.0 (10.7)	59.1 (11.6)	2.2 (0.03)
% mBMI	104.2 (9.3)	107.2 (9.4)	1.97 (0.06)
FBT completed (n=20)
kg	52.8 (8.0)	61.7 (9.8)	9.5 (<0.001)
% mBMI	101.0 (8.1)	114.0 (11.9)	7.8 (<0.001)

Bold data represents statistically significant at p<.05

We compared treatment completion versus non-completion groups on a number of weight-related variables at discharge; table 4 presents the findings from these analyses. We found the completion group had a discharge weight that was higher as a percentage of baseline weight, discharge weight that was higher as a percentage of treatment target weight, and discharge weight that was higher as a percentage of premorbid highest weight as compared to the non-completion group.

Table 4.

Discharge weight variables by treatment completion status

Discharge weight	Treatment not completed (n=21) Mean (SD)	Treatment completed (n=20) Mean (SD)	t (p)
kg	59.1 (11.6)	61.7 (9.8)	0.8 (0.5)
% mBMI	107.2 (9.4)	114.0 (11.9)	2.0 (0.5)
% baseline weight	103.7 (6.8)	116.9 (8.4)	5.6 (<0.001)
% target weight	96.6 (9.9)	106.1 (9.8)	2.9 (0.003)
% highest weight	84.9 (9.3)	93.7 (7.1)	3.4 (<0.001)

Bold data represents statistically significant at p<.05

Given our relatively small sample size and variability within groups, we also presented our findings at the individual level to allow for a more refined understanding of the results. Figure 1 displays individual-level bar graphs that indicate weight change from baseline to discharge. Figure 2 displays individual-level weight gain data (presented as percentage of baseline weight at discharge) that is separated by completion status.

Figure 1.

Individual weight change across treatment, ranked by baseline weight (%mBMI)

Note: light grey bars indicate weight gain; dark grey bars indicate weight loss. S = weight stabilization group; G = weight gain group.

Figure 2.

Individual level discharge weight as a percentage of baseline weight; divided by treatment completion category

Discussion

This study used a retrospective file review to investigate recommendations for weight gain versus weight stabilization in FBT for adolescents with AAN and to examine treatment outcomes in FBT for adolescents with AAN in an outpatient hospital setting, where care was provided outside of a clinical trial.

In line with previous qualitative research that indicated a lack of consensus among FBT practitioners about weight gain versus weight stabilization in adolescent patients with AAN (11), the current study found a divide within the sample in terms of recommendations for weight gain versus weight stabilization. An individualized approach to setting treatment target weights is utilized in this setting, thus we anticipated that premorbid weight and weight suppression would be key predictors of recommended weight gain. However, we found that weight gain and weight stabilization groups did not significantly differ on either of these variables, although there was a trend to indicate that those in the stabilization group had higher premorbid weight. One limitation is that premorbid highest weight was only available in kilograms, and was not able to be adjusted for height and age due to limited data. This is a common limitation observed in this clinic, with limited growth curves available from primary care physicians and highest weight is often self- or parent-reported.

In fact, when we examined the weight stabilization and weight gain groups across a number of covariates, we observed that groups differed only in terms of their baseline weight (in % mBMI). Of note, those in the stabilization group were actually above the median BMI for age (109%) while those in the weight gain group were slightly below the median BMI for age (98%). It may be that those patients at a lower weight were more medically compromised, or it is possible that there exists reticence for practitioners to suggest weight gain for those who are already slightly above average weight. Another complicating factor is the diagnostic inconsistency within AAN. Similar to previous work in this field (4), we discovered inconsistency between clinical diagnosis and the diagnosis assigned based on the standardized research criteria. Inconsistencies in diagnostic classification appeared to be due to lack of clarity on how to define “normal” weight criteria, and “significant” weight loss. We found that many of the patients in our study who actually met criteria for AAN were assigned a diagnosis of AN at assessment. This inconsistency likely contributed to differences in recommendations regarding weight gain in treatment.

In terms of treatment outcomes, we observed significant weight gain across treatment in this sample of adolescents with AAN. This is in contrast with results from a previous study of FBT for adolescents with AAN, which found little overall weight change across the course of treatment (10). We found that weight change (in % mBMI) was higher for the group that was recommended to gain weight compared to the group that was recommended to stabilize weight. It is not clear from the current study whether the increased weight gain we observed in the weight gain group is due to the recommendation itself (i.e., treatment team advising further weight gain in this group) or due to another variable, such as lower weight at baseline. Lower weight at baseline may contribute to increased willingness to gain weight from caregivers and patients, or may contribute to patients more easily gaining weight once eating disorder symptoms were interrupted. It is important to note that FBT for AAN may be complicated by resistance from parents about their child achieving a weight that is above average for height and age. Parents may cite fear of bullying, health risks of being overweight, and concerns about their child relapsing due to body image distress as barriers to reaching a weight that is above average. Parents’ conscious or unconscious biases about being overweight may further contribute to their fears.

Due to limitations in the information provided at discharge in this setting, our retrospective review was not able to rigorously apply remission criteria. Thus, we elected to use “treatment completion” as a marker for treatment success, recognizing its limitations. Rather than a session number cut-off, we examined the reasons for ending treatment and classified participants accordingly. Based on this definition, we found that just under half of the sample completed treatment. Those who completed treatment did not differ on baseline characteristics but did sustain additional weight gain during treatment compared to those who did not complete treatment. Those who completed treatment gained on average 17% of their baseline body weight over the course of treatment. Of note, they seemed to surpass treatment target weight slightly, but did not reach their premorbid highest weight by the end of treatment. Taken together, our findings suggest that weight gain may be an important component of treatment of AAN in adolescents.

This study has several methodological limitations, including a small sample size, retrospective design, and measurement of outcomes only at the end of treatment. In particular, the small sample size limits the conclusions that can be drawn from the data and the generalizability of the findings. We elected to present findings by both recommended weight gain and treatment completion status, but note that multiple statistical comparisons increase the risk of Type II error. The current study also lacks inclusion of symptom-based questionnaires and medical variables that would have allowed for a more thorough examination of treatment outcomes in this patient population. Hughes and colleagues (10) showed an improvement in eating disorder symptoms in their study, in spite of limited weight gain. Both of these studies were limited by a lack of control group to test hypotheses. Future studies would benefit from a more controlled examination of different components of FBT in AAN across treatment to understand the active ingredients necessary for recovery from this disorder.

It is also noteworthy that some of the current sample received services after the onset of the COVID-19 pandemic, which resulted in some alterations to typical practice standards. Most notably, FBT sessions were delivered via video-conference and session weight were taken on home scales. We observed that it was increasingly difficult for youth to obtain in-person medical monitoring from family physicians/nurse practitioners after the onset of the pandemic, which may have contributed to delays in obtaining updated height data needed for re-assessing treatment target weights. Given that our inclusion period for study participants ended only 4 months into the onset of the pandemic, it is unlikely that those presenting for assessment during the period of the pandemic differed significantly from those assessed pre-pandemic, as their illness onset and referral to our clinic would have occurred prior to the onset of the pandemic.

This study also has some important strengths, especially the application of FBT for AAN in a real-world setting. As such, this study is able to highlight some of the practical limitations of treating adolescents with AAN. The first pertains to diagnostic inconsistency for AAN. It is possible that some of the confusion in the field about whether weight gain is recommended for this population relates to the diagnostic confusion. For instance, if a patient is mistakenly diagnosed with AAN and has not sustained significant weight loss or their weight loss is part of a larger pattern of weight gain and weight loss due to bulimia nervosa or more general over-eating, then this youth may not need to gain weight as part of their treatment. Because clear and consistent diagnosis is so important for treatment recommendations, we recommend increased standardization in differential diagnosis in clinical settings. Diagnostic inconsistency is common with AAN (3) and may be partly due to ambiguity in DSM-5 criteria. As such, applying numerical guidelines to define terms like “normal weight” (e.g., ≥90% of mBMI) and “significant” weight loss (e.g., ≥10% of body weight in past year, or ≥7% in past 3 months) may be particularly helpful. Given that there is considerable variability in the numerical guidelines used to operationalize these terms in the research literature (3), consensus may not emerge in the field without further guidance in a future edition of the DSM. Nonetheless, it may be helpful for clinicians working within the same setting to come to their own consensus about how they will operationalize these terms, with the goal of increasing consistency in practice and provision of care in their clinic. In our clinic, we created a diagnostic algorithm based on both DSM 5 criteria and numerical guidelines from selected research studies, and are trialing using this during our multidisciplinary assessments.

Another clinical limitation involves limited availability of growth history to set treatment target weights. We have found that youth in our service area often have incomplete or no growth curve data, as in our region youth are not typically followed by pediatricians and it is not routine practice for family physicians/nurse practitioners to collect this form of developmental data. Moreover, approximately half of the residents in our province do not have access to a family physician/nurse practitioner and therefore rely exclusively on walk-in and emergency services. Although less reliable than objective measurements, there is some evidence that self- and parent-reported data can be relatively accurate (23, 23), and it is standard practice to incorporate this data into assessments and research protocols (e.g., 6, 16). It may be helpful to gather these self-report data in a more systematic and comprehensive way during assessments, such as asking parents for as many weight and height estimates that they can provide throughout childhood. If given advance notice and educated about the importance of these data, families may also be able to collect more objective information from different sources that they have at home (e.g., height measurements taken on the wall; records from medical visits). More comprehensive historical self-report data are likely to contribute to improvements in estimates of treatment target weights.

Finally, we identified barriers to applying guidelines for updating treatment target weights across treatment (i.e., every 3 months for youth who are still growing and every 6 months for older youth among whom further growth is not expected (18, 19). A key barrier in our clinic was inconsistent re-assessment of height by primary care physicians during treatment, and delays in communication of this information to team dietitians. This is especially relevant for patients who were engaged in FBT for close to one year, as a certain amount of weight gain would be expected for adolescents over that time period, just to remain on their developmental trajectory. For younger patients who are still growing in height, the amount of normative weight gain would be even higher. In our clinic, we have sought to establish reminders for dietitians to request updated height information from primary care physicians, as well as working with FBT therapists and families to acquire this information. We have also established a system for tracking treatment target weights across treatment, ensuring that these are truly “moving targets” reflective of the patient’s developmental stage.

In conclusion, this study adds to previous qualitative studies (11) on the difficulty of establishing recommendations about weight gain in FBT for adolescents with AAN, and extends previous studies by explicitly looking at weight change outcomes by weight change recommendations. We found significant weight gain during the course of FBT for AAN, especially in those who were recommended to gain weight, and in those who completed treatment. Although limited by our small sample size and methodological limitations, findings from our retrospective chart review provide some preliminary evidence that weight gain may be an important part of treatment in adolescents with AAN, as it is in the treatment of adolescents with AN. The current study underlines some of the challenges in treating adolescents with AAN, and highlights opportunities for further research in this area as well as opportunities for improving clinical care for these youth.