Introduction

Eating disorders (EDs) have the potential to cause a number of acute as well as long-term complications. The prevalence of EDs has increased in Western countries among adolescent girls in the high-risk group of 15–19 years [1] and many adolescents go untreated [2]. The incidence of Anorexia Nervosa (AN) is highest in the adolescent period [3], which is an important time for growth and independence, and the illness can be protracted, causing a significant disruption to schooling and family life.

Family-based therapy (FBT) for AN has shown significantly faster weight gain early in treatment, fewer days in the hospital, and lower treatment costs per patient in remission at the end of treatment in the US/Canadian-based populations [4,5,6]. FBT is emerging as the first-line therapy for paediatric patients with AN [7]. However, the success of FBT relies on the population to which the families belong, their cultural backgrounds, and eating behaviours. Despite the success of FBT in western populations, there is scant literature on whether FBT is similarly effective in Asian populations. Although one qualitative study [8], looking at the acceptance of family therapy in 24 patients treated in Hong Kong, found that families generally perceived it as helpful, to date, there have been no longitudinal studies investigating the effectiveness of FBT in Asia. Hence, in this study, we aimed to assess the effectiveness of FBT in a Singaporean cohort of adolescent patients with AN by comparing the outcomes of patients treated with FBT with those receiving individualized Adolescent Focussed Therapy (AFT). We hypothesized that FBT would similarly improve remission rates and reduce the time taken to achieve remission in our patients with AN.

Methods

Study design and setting

This is a retrospective, cohort study over a 7-year period from January 2011 to December 2017, of patients with AN treated within a large tertiary hospital in Singapore. The Khoo Teck Puat National University Children’s Medical Institute has an EDs program which manages paediatric patients (≤ 18 years old) admitted to the paediatric wards of the National University Hospital (NUH), Singapore, or referred to our clinics from elsewhere. The ED team comprises of paediatricians trained in adolescent medicine, adolescent nurses, psychiatrists, psychologists, dietitians, medical social workers, and art therapists. To be included in this study, patients should have completed at least 1 year of treatment.

Based on encouraging results from studies on FBT in Western populations, we started to offer FBT to all our patients with AN since 2015, hence, shifting our focus from individualized AFT to manualised FBT.

Previously, patients underwent individualized AFT (non-FBT group), where the methods varied for each patient, but was primarily based on the principles of cognitive, behavioural, expressive, dynamic, and developmental therapeutic techniques. Family involvement was supportive in nature, but this was not protocol based and the primary focus was on reintegration of the patient back into the family once improvements were seen. Patients were admitted if they were medically unstable (usually for bradycardia) and stayed in the hospital until they reached 85% of their ideal body weight. Since 2015, patients are encouraged to enrol in manualised FBT. All the patients undergoing FBT are discharged as soon as medical stabilisation is achieved, even if they are below their ideal body weight as we could review them frequently and regularly as outpatients. Manualised FBT is employed with a formal program that assigns specific goals to each session and involves regular and frequent outpatient treatment sessions with the whole family, which typically is at least 20 sessions over a 6–12 month period. The focus was on parent empowerment, rapid weight gain, and inducing the early behavioural changes.

All patients were seen by the entire team including psychiatrist, adolescent medicine specialist, dietician, and therapists.

Participants

The study subjects were all patients diagnosed with AN and followed up by our ED management team. The clinical diagnosis of AN was made based on the consensus among the team members which included two medical consultants—an adolescent paediatrician and a child psychiatrist experienced in managing AN. The diagnosis was in accordance with the guidelines of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-4) prior to June 2013 and Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), thereafter. To be included in this study, patients should have completed at least 1 year of treatment. Accordingly, the follow-up period was 7 years for the earliest patients and a minimum of 1 year of follow-up for all patients.

Variables and data sources/measurement

The outcome variables were time to remission and number of patients in remission at 1 and 2 year follow-up. The required data were extracted from hospital case notes. The operational definition of remission was the achievement of a minimum 95% of expected Ideal Body Weight for sex, age, and height and no longer meeting the DSM-5 criteria for AN, with consistent improvement in behaviours (≥ 3 months) as assessed by the therapist. Time to remission was defined as the time from the onset of treatment to the time that the remission criteria were met. Patients were followed up regularly until remission was achieved. The minimum frequency of follow-up visits was monthly until achieving remission, and remission criteria were assessed at every follow-up visit.

Demographic information such as age at diagnosis, weight, BMI, time to diagnosis, sex, race, parent’s marital status, and comorbidities were extracted from the patient case notes.

Statistical analysis was done in SPSS (IBM SPSS Statistics Version 25). To determine the difference in the time to remission between the groups, we used survival analysis with a log-rank test for comparison of median time to remission. Kaplan–Meier curves were used to determine remission-free survival rates at 1 and 2 years from the commencement of treatment. We used cox regression to obtain unadjusted and adjusted hazard ratios for achieving remission within the study period. We adjusted for gender, age, BMI at diagnosis, ethnicity, and the presence of psychiatric comorbidity.

Ethics

This study was approved by the National Healthcare Group Domain Specific Review Board. Being a retrospective study with anonymised data, formal consent was not required.

Results

During the 7-year period of interest, we completed at least 1 year of treatment for 119 patients with AN. Of these, 77 patients were treated with individualized AFT (non-FBT group) and 42 with manualised FBT.

Patient characteristics

The patient characteristics in the two groups are similar (Table 1) except that in the FBT group patients had a higher mean BMI at diagnosis. The prevalence of psychiatric comorbidities was similar in both groups (54 and 50% of patients in the non-FBT and FBT groups, respectively). Among the patients with psychiatric comorbidities, 78% of those in the non-FBT group and 81% in the FBT group received psychiatric medications, which were primarily fluoxetine, mirtazapine, escitalopram, or olanzapine. Cox regression was used to statistically adjust for differences between the groups.

Table 1 Baseline non-FBT and FBT patient characteristics
Full size table

Patient outcomes—remission rates

Patients in the non-FBT group had significantly longer time to remission compared with patients in the FBT group (p < 0.001, HR = 3.191, 95% CI 1.863–5.468). The median time to remission was 16.0 months (95% CI 11.4–20.6 months) for patients in the non-FBT group, and 5.0 months (95% CI 3.4–6.6 months) for patients in the FBT group. The difference in the median time to remission was 11.0 months (p < 0.001, 95% CI 7.9–14.1 months).

The FBT group achieved higher rates of remission at both 1 and 2 years. The remission rates in the FBT group were 69% and 90% at 1 and 2 years, respectively. For the non-FBT group, remission rates were 30% and 57% at 1 and 2 years, respectively.

Cox regression was performed to adjust for the following variables; gender, age, BMI at diagnosis, ethnicity, and the presence of other concomitant psychiatric conditions. The difference in remission-free survival between the two groups remained significant (p value = 0.003, HR = 2.523, 95% CI 1.372–4.639). All of the other tested variables were not statistically significant.

The improved results with FBT were achieved despite a strategy that reduced the mean length of hospital stay by 30 days (− 30 days 95% CI − 56 to − 3 days, p = 0.03). In the patients requiring admission, those in the FBT group had a mean duration of stay of 27 days compared to 57 days in the non-FBT group. There was also less variability in the duration of hospital stay in the FBT group. The median length of stay in the FBT group was 23 days (IQR 16–27 days) compared to 33 days (IQR 20–60 days) in the non-FBT group.

Our loss to follow-up was 7% (n = 8) for the whole cohort; seven patients in the non-FBT group and one patient in the FBT group. The results assume that the patients who were lost to follow up were non-remitters. A sensitivity analysis was done, where we assumed that patients lost to follow up remitted at 12 months or 24 months, but the time to remission, as well as the proportion in remission at 1 year and 2 years remained significant between the groups.

Discussion

AN causes much morbidity and is disruptive to patients and their families, so strategies to induce faster recovery are sought after. FBT has become a well-established therapy in Western populations. We note that such efficacy studies have not been done in the Asian context.

The utilization of FBT, with regular and frequent outpatient consultations, has helped in inducing much faster and higher remission rates in our cohort compared to individualized Adolescent Focused Therapy. We have seen the initial promising results with an absolute reduction in the median time to remission by 11 months and an absolute increase in remission rates by 39% at the end of 1 year of treatment, and 33% at 2 years from the commencement of treatment. These results were achieved despite a strategy that has reduced the average duration of inpatient hospital stay by 1 month. The length of hospital stay for inpatients was also much less variable for patients who were on FBT.

This study is the first to assess the effectiveness of FBT in an Asian context. Key strengths of the study are a relatively large sample size for AN, given its rarity, though modest compared to studies in other fields; and a high follow-up rate. We had a multidisciplinary team comprising of adolescent medicine physicians as well as psychiatrists to give accurate diagnoses, and therapists trained in the facilitation of individual and family-based therapies. Our findings not only add to the growing body of literature supporting the efficacy of FBT, but also verify its effectiveness in an Asian population.

As previously reported, it is difficult to compare remission rates between studies because of varied definitions used [9]. Some have recommended assessment tools such as the Eating Disorder Examination (EDE) [9], but they do note that extensive assessment tools and surveys may not be practical in a clinical setting. Since this tool has been mostly used in older patients and we do not have norms for our population, we were not able to apply them in our setting. Our assessment of remission was made based on weight restoration, as well as no longer fulfilling the DSM-5 criteria for AN over a period of at least 3 months. This assessment was done by the same team for all cases, thus ruling out the variability of assessments within our patient cohort.

Notwithstanding the difficulties in comparing studies, we have tried to compare our results with the randomised-controlled trial of Lock et al. [5], looking at AFT versus FBT. They found improvement in full and partial remission rates in the FBT group. Partial remission [defined as achieving a weight of more than 85% of expected ideal body weight (IBW) for age, height, and sex] was seen in 89.1% of patients treated with FBT compared to 66.7% of patients treated with AFT at end of 1 year of treatment [p = 0.02]. With a more stringent definition of full remission, defined as body weight more than 95% of IBW and normalisation of global EDE scores, there was 41.8% of FBT patients and 22.6% of AFT patients in full remission at the end of treatment [p = 0.06] which was not statistically significant, but was shown to be significantly superior at 1 year follow-up (mean difference 26.2%, p = 0.02). Our definition of remission is less strict than their definition of full remission but more stringent than their definition of partial remission. As expected, our remission rates in both FBT and non-FBT groups at 1 year of treatment fall in between full and partial remission rates at the end of treatment for the respective groups in their study.

Limitations of this study include the retrospective nature of the study. Randomised-controlled trials would be superior, but, given that studies have been previously done to prove efficacy, our aim was to verify the previously noted results within the Asian context and as such observational studies can be useful for this purpose. There may be differences between the families who participated in FBT compared to those who did not. Good family support would be a beneficial factor in recovery despite the method of treatment; however, the main difference between these groups was only the time of recruitment. Most of our patients treated prior to 2015 received individualized AFT, while those treated after 2015 received manualised FBT. Our results were adjusted for potential confounders including gender, age, BMI at diagnosis, the presence of psychiatric comorbidity, and ethnicity. The increase in remission rate seen from 1 to 2 years indicates that the benefits of FBT may be more pronounced and recognized over the longer term. Further research to study sustained remission rates over a longer period of time will be useful.

Our results are quite similar to those from studies on western population, which show both increased remission rates and shorter time to remission [4,5,6]. It is, therefore, pertinent to note that FBT for the treatment of adolescents with AN is also effective in the Asian context. The advantages of FBT as an effective treatment for AN lie in the fact that it induces faster remission and higher remission rates without the need for lengthy hospital stays. This is especially important in adolescence, which is a critical stage of development.