Patients lost to follow-up include those who could not attend; missing patients include those with missing measurements.
Immediately after treatment, the difference between groups was statistically significant (P = .04); however, the favorable outcomes of the inpatient treatment group were not sustained during follow-up. Error bars indicate SE.
eTable. Exclusion criteria
eAppendix. Description of interventions
Customize your JAMA Network experience by selecting one or more topics from the list below.
van der Baan-Slootweg O, Benninga MA, Beelen A, et al. Inpatient Treatment of Children and Adolescents With Severe Obesity in the Netherlands: A Randomized Clinical Trial. JAMA Pediatr. 2014;168(9):807–814. doi:10.1001/jamapediatrics.2014.521
Copyright 2014 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Severe childhood obesity has become a major health problem, and effective, evidence-based interventions are needed. The relative effectiveness of inpatient compared with ambulatory treatment remains unknown.
To determine whether an inpatient treatment program is more effective than an ambulatory treatment program at achieving a sustained weight loss in children and adolescents with severe obesity.
Design, Setting, and Participants
We conducted a randomized clinical trial with a 2-year follow-up at a tertiary referral center for pediatric obesity in the Netherlands. We recruited 90 children and adolescents aged 8 to 18 years with severe obesity (body mass index [BMI] z score, ≥3.0 or >2.3 with obesity-related health problems).
Patients were randomly assigned to an inpatient (6 months of hospitalization on working days) or an ambulatory (12 days of hospital visits at increasing intervals during a 6-month period) treatment program. Both treatment programs involved an intensive, family-based, lifestyle intervention, including exercise, nutritional education, and behavior modification for the patients and their caregiver(s).
Main Outcomes and Measures
Change in BMI z score. Secondary outcomes included fasting insulin, fasting plasma glucose, 2-hour plasma glucose, and lipid levels, insulin sensitivity, liver function test results, waist circumference, blood pressure, body composition, and aerobic fitness (peak oxygen consumption, V̇o2). Outcomes were analyzed by intention to treat.
Immediately after treatment, reductions in the BMI z score were significantly larger for the inpatient than the ambulatory groups (mean [SE] difference, −0.26 [0.12; 95% CI, −0.59 to −0.01]; P = .04). Change from baseline for the BMI z score in the inpatient group was −18.0% (P = .001) immediately after treatment, −8.5% (P = .008) at 18 months, and −6.3% (P = .38) at 30 months; in the ambulatory group, changes from baseline were −10.5% (P = .001), −6.2% (P = .39), and −1.5% (P > .99), respectively. The favorable outcomes of the inpatient group could not be sustained at 12 and 24 months after treatment. In addition, significant differences in favor of the inpatient group immediately after treatment were found for levels of fasting insulin (−6.37 IU/L; P = .02), total cholesterol (−19.51 mg/dL; P = .01), low-density lipoprotein cholesterol (−13.48 mg/dL; P = .03), and triglycerides (−25.39 mg/dL; P = .01), and insulin sensitivity (−1.37; P = .02), fat mass (−3.31%; P = .03), and peak V̇o2 (378.2 mL/min; P = .01).
Conclusions and Relevance
In severely obese children and adolescents, inpatient treatment was superior to ambulatory treatment immediately after treatment, but effects were not sustained at long-term follow-up. These findings stress the need to further study maintenance strategies for sustainable weight loss.
trialregister.nl Identifier: NTR1172
At present, obesity is the most common medical disorder of childhood in the developed world, and severe obesity among children and adolescents is increasing, with no good therapeutic options.1,2 Pediatric obesity is an independent risk factor for increased morbidity and mortality throughout life.3-6 Strong evidence shows that pediatric obesity has a wide range of adverse short- and long-term consequences.7 These children are at the highest risk for medical and psychological sequelae, justifying active intervention.8,9 Treatment of obesity can improve health, physical fitness, moods, and psychosocial functioning.10,11 Several studies have shown that severe pediatric obesity (body mass index [BMI], ≥99th percentile) is less amenable to conventional ambulatory treatment than moderate obesity (BMI, ≥95th to <99th percentile).12,13 Effective medications for children are lacking, and the long-term safety and effectiveness of bariatric surgery in adolescents remains to be determined.
The relative effectiveness of different interventions remains unclear, and reports of long-term effectiveness are scarce.14 A recent systematic review15 of immersion treatments (weight loss camps and residential programs) showed a mean greater reduction of 191% in overweight after treatment and 130% at follow-up compared with a meta-analysis of outpatient treatments. However, the economic and practical challenges of inpatient treatment are daunting, especially because of the total number of severely obese children and adolescents. Therefore, cost-effectiveness of inpatient vs ambulatory care is an important issue to study.
The present study investigated the effectiveness of a 6-month inpatient treatment program compared with an intensive ambulatory program of the same duration. The goal of each program was to achieve a sustained weight loss (to 2 years) in severely obese children and adolescents who had previously failed to achieve sustained weight loss in other ambulatory programs aimed at moderate obesity.
This study was a randomized clinical trial that compared the effects of an inpatient obesity treatment program with those of an ambulatory treatment program with the same components, only varying the modality of treatment. The treatment period was 6 months, with a 2-year follow-up. We conducted the study from November 1, 2004, through December 31, 2009, in a single, pediatric, tertiary referral center in the Netherlands that specializes in the treatment of severely obese children. The study protocol was approved by the Medical Ethics Committee of the Academic Medical Center of the University of Amsterdam. Assent and written informed consent were obtained from the parents or the legal representatives of participants and from adolescents (aged 12-18 years). Parents and/or caregivers gave additional written consent to participate in their children’s treatment and educational regimens.
The study population consisted of children and adolescents with severe obesity who were referred to our tertiary care center after ambulatory, multidisciplinary treatment elsewhere had failed. Inclusion criteria were 8 to 18 years of age and a BMI z score of at least 2.3, corresponding to the 98.9th percentile, according to the growth curves based on the fourth Dutch National Growth Study of 1997 (calculated via http://groeiweb.pgdata.nl/calculator.asp), with obesity-related comorbidity (eg, obstructive sleep apnea syndrome, elevated insulin levels, type 2 diabetes mellitus, liver function disorders, hypertension, dyslipidemia, joint problems) or a BMI z score of at least 3.0 (corresponding to the 99.9th percentile). Exclusion criteria consisted of severe psychiatric disorder, intellectual disability, obesity caused by endocrine disorders (eTable in the Supplement), use of medication that could cause significant weight gain or weight loss, and/or participation in a concomitant weight management program.
Eligible and consenting patients were randomly assigned at a ratio of 1:1 to the inpatient or the ambulatory treatment program (Figure 1). The randomization procedure was stratified for sex, age (8-12 vs 13-18 years), and BMI z score (for girls, ≥3.41 or <3.41; for boys, ≥3.61or <3.61). Stratification of the BMI z scores was based on the mean pretreatment BMI z scores in our center in the years before study onset, making up 8 strata in total. Pairs of children within each stratum were simultaneously randomized. Allocation concealment was achieved by sealed, opaque envelopes. Study personnel performing the outcome assessments were masked to treatment assignments.
The intervention groups are described in detail in the eAppendix in the Supplement. Patients randomized to the inpatient treatment program were hospitalized for 26 weeks. They followed a program during weekdays and returned home for the weekends with homework assignments. The program consisted of an exercise schedule 4 days per week (30 to 60 minutes each day, with a mean duration of 45 minutes for each exercise session) and nutrition/behavior modification once per week (60 minutes for each session). Patients and caregivers received comparable information about nutrition and behavior, but the 1-hour lessons were held separately, at 3 times during the treatment period.
A child psychologist facilitated the behavior modification element of the weight management program in individual and group sessions. Topics included self-regulation, self-awareness, goal setting, stimulus control, coping skills training, cognitive behavioral strategies, contingency management, and positive reinforcement for meeting (self-imposed) goals. Behavior modification classes for caregivers included the same topic, with additional topics that reflected the challenges verbalized by the parents (eg, preparing healthy meals, implementing physical activity in the family’s routine, how to help the child when the caregiver fails to change their own habits). Aside from group sessions, individual meetings with a dietician, a psychologist, or a social worker were organized as needed.
An exercise therapist led the exercise component of the weight management program. The standardized training sessions with high-intensity aerobic exercise (indoor, outdoor, and swimming activities) occurred in groups (n ≤ 10). Besides these organized activities, patients were encouraged to participate in outside activities daily, with other patients or on their own, to change their sedentary behavior.
The nutritional educational component of the program used a nondiet approach, focusing on improving the quality of the dietary intake and on trying to establish controlled yet flexible eating behaviors. A stable and predictive pattern of eating was promoted, and children were encouraged to try unfamiliar foods. By creating more awareness of their feelings of hunger and satisfaction, physical regulation of food intake was stimulated. The nutritional education of the parents included the above-mentioned components with more in-depth knowledge of nutrition, such as understanding food labels.
Patients randomized to the ambulatory treatment program and their caregivers attended the program for 12 visits at increasing time intervals for a 6-month period. After weighing, the children exercised for an hour (swimming and gymnastics). Children and parents were also encouraged to exercise at home on 3 additional days per week and to reduce sedentary behaviors. After physical exercise, the children attended an educational program for 1 hour and a nutritional educational session for half an hour.
In parallel sessions, caregivers were given detailed instructions on nutrition and nutritional behavior. The classes emphasized the importance of the parents’ role in inducing changes in health behaviors. The exercise counseling was given by exercise therapists who delivered exercise classes to the patients. The content of the nutrition classes, behavioral modification classes, and homework assignments for children and parents/caregivers was identical to that given during the inpatient treatment program.
The primary outcome of the study was the BMI z score, standardized by the use of age- and sex-normative data from the Dutch National Growth Study of 1997.16 Weight was assessed using a digital scale (Seca) to 1 decimal point in light clothing and bare feet. Height was measured to the nearest 0.1 cm using a stadiometer. Secondary outcomes included (1) waist circumference, measured in the horizontal plane with the smallest circumference between the costal arch and the iliac crest; (2) fat mass, determined in the morning after an overnight fast by bioelectrical impedance analysis (bioelectrical impedance analyzer model BIA 101-S; RJL Systems) while in the supine position; (3) blood pressure, measured automatically with a sphygmomanometer (model E 60; Heine) 3 times after at least five 5-minute rest periods (the mean of the final 2 measurements were taken); (4) biomarkers of cardiovascular disease risk (ie, plasma glucose, insulin, total cholesterol, low- and high-density lipoprotein cholesterol, triglyceride, aspartate transaminase, and alanine transaminase levels); and (5) aerobic fitness.
Blood samples were obtained after an overnight fast. Plasma glucose levels were measured with a chemistry analyzer and plasma insulin levels were measured by radioimmunoassay. Plasma lipid levels were measured with an autoanalyzer. The homeostasis model assessment of insulin resistance (HOMA-IR) was used to determine insulin sensitivity and was calculated using the following formula17:[Fasting Blood Glucose Level (in micromoles per liter) × Fasting Plasma Insulin Level (in microunits per milliliter)]/22.5. Aerobic fitness (peak oxygen consumption [V̇o2 peak] relative to body mass in milliliters per kilograms per minute) was determined from maximal exercise tests, which were performed on a cycle ergometer (Lode Excalibur). All participants were familiarized with the test and the equipment used, and standardized verbal encouragement was given throughout the test to stimulate maximal performance. Measurements were taken at 0, 6, 9, 12, 18, and 30 months for the BMI z score; at 0, 6, 18, and 30 months for waist circumference, bioimpedance, blood pressure, and fitness (V̇o2 peak); and at 0 and 6 months for blood tests, including those for oral glucose tolerance, lipid levels, and liver function (aspartate transaminase and alanine transaminase levels).
The sample size was based on an expected difference in BMI z score of 0.25 between the treatment groups, an expected SD within each group of 0.42, and 4 repeated measurements (with an estimated correlation between repeated measures of 0.75). With a 2-sided α set at .05 to obtain 80% power, we required 37 participants per group. With an expected dropout of 15% to 20%, we enrolled 90 participants. We compared baseline characteristics of the 2 treatment groups and of dropouts vs completers using 2-tailed t tests for independent samples (after checking for normal distributions of continuous variables), and we used the Pearson χ2 test in cases of categorical variables. Linear mixed-effects models for repeated measures were used to analyze between-group differences in outcome over time. In addition, we compared the proportion of participants who achieved a clinically meaningful weight loss of 0.5 BMI z or more (which is within the range where positive effects on cardiometabolic risk markers can be found in children).18,19 Data were analyzed using the intention-to-treat principle. Statistical significance was set at P ≤ .05.
Participant flow is presented in Figure 1. Participants were recruited from November 1, 2004, through May 31, 2007, and follow-up measurements were completed by December 31, 2009. Characteristics of all patients are shown in Table 1.
After randomization, 9 patients (4 from the inpatient treatment group and 5 from the ambulatory treatment group) withdrew consent because of dissatisfaction with the randomization result. We found no significant differences in age or BMI z score between patients who withdrew and those who started treatment (mean [SD] age, 14.2 [1.3] vs 13.9 [2.3] years, respectively; mean [SD] BMI z score, 3.24 [0.37] and 3.36 [0.40], respectively). Treatment was discontinued by 10 patients (7 from the inpatient treatment group and 3 from the ambulatory treatment group) for various reasons (homesickness [n = 3], family problems [n = 2], motivational problems [n = 3], and behavioral problems [n = 2]). We found no significant differences in age, sex, or BMI z scores at baseline between completers and noncompleters (data not shown).
The BMI z scores are shown in Figure 2 and Table 2. Linear mixed modeling showed an overall significant effect between the interventions in favor of the inpatient treatment group (P < .01). Except at baseline, mean BMI z scores of the inpatient treatment group remained below those of the ambulatory treatment group throughout the entire study period (Figure 2). The BMI z score at the end of the 26 weeks of treatment differed significantly between the groups (P = .04). However, this difference was not sustained at 3, 6, 12, or 24 months of follow-up. The BMI z scores gradually returned to near baseline values during the 2-year follow-up, but the mean BMI z scores remained less than the starting levels. In the inpatient group, BMI z scores of these participants were still less than the starting levels at 30 months after the start of treatment. During the 26 weeks of treatment, 22 individuals in the inpatient group (54%) and 9 individuals in the ambulatory group (23%) (P = .008) achieved a clinically meaningful reduction in their BMI z score (ie, ≥0.5). During follow-up (at 3, 6, 12, and 24 months), the proportions of individuals with meaningful reductions in their BMI z score did not differ between the treatment groups (17 of 35 [49%], 12 of 27 [44%], 11 of 33 [33%], and 8 of 31 [26%] for the inpatient treatment group and 10 of 22 [45%], 8 of 23 [35%], 7 of 27 [26%], and 5 of 28 [18%] for the ambulatory treatment group). Change from baseline for the BMI z score was −18.0% (P = .001) immediately after treatment, −8.5% (P = .008) at 18 months, and −6.3% (P = .38) at 30 months in the inpatient treatment group and −10.5% (P = .001), −6.2% (P = .39), and −1.5% (P > .99), respectively, in the ambulatory treatment group.
Analyses of secondary outcomes immediately after treatment showed significant between-group differences in fat mass and V̇o2 peak (Table 3) and levels of fasting insulin, total cholesterol, low-density lipoprotein cholesterol, and triglyceride and HOMA-IR (Table 4), all of which favored the inpatient treatment group. Adverse effects were found in 3 girls (2 in the inpatient treatment group and 1 in the ambulatory treatment group) who developed silent gallstones.
Our trial shows that inpatient treatment of severely obese children and adolescents is more successful in reducing weight and the percentage of body fat and improving peak V̇o2 immediately after treatment compared with an ambulatory treatment program, with a trend in favor of inpatient treatment to 1 year after the start of treatment. At the 12- and 24-month follow-ups, we found no significant difference. Studies evaluating the effectiveness of treatment programs in young children and adolescents are rare.20 Therefore, important strengths of our study are the randomized design and the population of children and adolescents with the most severe degree of obesity (BMI z score, >98.9th percentile, with 74 [82%] having a score >99.9th percentile). The generalizability of the study remains uncertain, because of the single-center setting. Another limitation of our study is the lack of an economic evaluation, which may differ substantially between the treatment programs. Given the higher costs of inpatient treatment and the observation that effects could not be sustained during the 2-year follow-up, cost-effectiveness is unlikely until a sustained treatment effect can be reached. If interventions are effective, they are potentially cost-effective during the lifetime of a child.21 Cost-effectiveness of obesity treatments, however, remains a complex issue, and cost savings and health benefits may not appear until the sixth or seventh decade of life.21
The inpatient treatment group achieved a clinically relevant reduction in BMI z scores (≥0.5) to 12 months after the start of treatment, whereas the ambulatory treatment group did not. This finding is important because a reduction in BMI z score of at least 0.5 is expected to have a positive effect on cardiovascular risk markers in children.18,19 This result may be attributed to the daily professional counseling of the children, the supervision and imposed daily structure, and all other aspects promoting a healthy lifestyle. These aspects may have led to an increased compliance with the program components during treatment and in the 6 months thereafter.
In accordance with earlier data, we also found a positive effect immediately after treatment in favor of the inpatient treatment program for several cardiovascular risk factors, body composition (fat mass), and aerobic fitness levels. In contrast, fitness levels decreased during the treatment period in the ambulatory treatment group. The regular and more frequent supervised exercise in the inpatient treatment group likely contributed to this differential effect between the groups.
We were surprised by the relative effectiveness of the ambulatory treatment in this therapy-resistant group because therapy responses in outpatient settings are often unsuccessful in severely obese youth.10,11,22 Our considerable experience with inpatient treatment of severely obese children might have contributed to this success.
Treatment outcomes can differ by age. Danielsson et al13 showed only a reduction in BMI SD scores in young children but not in severely obese adolescents (14-16 years of age). In the present study, we stratified for age (8-12 and 13-18 years). Subgroup analysis based on age group was not performed because of the small numbers of young children in each treatment group.
Long-term follow-up studies (>1 year) of interventions for severely obese youth are scarce.21 Studies in overweight and obese children and adolescents22,23 are disappointing in that achieved weight reductions are often not maintained. In severe obesity, the long-term results are even more discouraging; some individuals may be highly successful, while others continue to gain weight, despite treatment.24 In line with our results, 2 earlier studies in severely obese children and adolescents aged 8 to 12 years25 and 6 to 16 years13 demonstrated no sustained weight loss effect at 6 and 12 months after treatment follow-up and after 1, 2, or 3 years, respectively . In fact, only 2% of the participants in the latter study experienced significant weight loss (≥0.5 at the 3-year end point).13 The Loozit trial,26 however, showed promising results in adolescents, with significant but modest reduction in BMI z scores to 24 months. Identifying significant predictors of weight loss outcomes is crucial to enable the development of personalized treatment.
Achieving behavioral change is a complex process involving several steps. Adverse environmental factors, often combined with complex psychosocial issues in families, overwhelm behavioral and educational techniques designed to reduce energy intake and augment physical activity. The relative intellectual and psychological immaturity of children and their susceptibility to peer pressure present additional challenges to effecting behavior modifications. Possibly a long-term care model, similar to those that have been successful in the long-term management of other diseases, such as chronic obstructive pulmonary disease and diabetes mellitus,27,28 would be more effective for these children. This type of model consists of the sustained application of self-management programs, which are aimed at (structural) behavioral changes, including an action plan in the event of a relapse, that are individually tailored, taking into account the patient’s and family’s views. Whether often-found psychosocial issues (eg, family problems, poverty, substance abuse and addiction, and poor parenting) should be addressed more intensively in addition to a lifestyle program in severe obesity in childhood needs further investigation. Possibly, interventions into these areas should even precede obesity lifestyle programs.20
This randomized clinical trial of 2 treatment programs for severely obese children and adolescents has shown statistically significantly better outcomes in the inpatient group immediately after treatment, but effects could not be sustained in the long term. More research is needed to identify the determinants of interindividual variability in intervention success to more appropriately tailor treatment programs to achieve a maximal sustained weight loss. Future research might also address the question whether these treatment programs may have stronger effects in children, who are less resistant to treatment.
Accepted for Publication: March 2, 2014.
Corresponding Author: Olga van der Baan-Slootweg, MD, PhD, Pediatric Department, Merem Childhood Obesity Centre Heideheuvel, Soestdijkerstraatweg 129, 1213 VX Hilversum, the Netherlands (firstname.lastname@example.org).
Published Online: July 14, 2014. doi:10.1001/jamapediatrics.2014.521.
Author Contributions: Drs van der Baan-Slootweg and Beelen had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: van der Baan-Slootweg, Benninga, Tijssen, van Aalderen.
Acquisition, analysis, or interpretation of data: van der Baan-Slootweg, Beelen, van der Palen, Tamminga-Smeulders, Tijssen, van Aalderen.
Drafting of the manuscript: van der Baan-Slootweg, Benninga, Beelen, van Aalderen.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Beelen, van der Palen, Tijssen.
Administrative, technical or material support: Tamminga-Smeulders.
Study supervision: van der Baan-Slootweg, Benninga, Beelen, Tijssen, van Aalderen.
Conflict of Interest Disclosures: Dr van der Baan-Slootweg and Ms Tamminga-Smeulders are employed as pediatrician and research nurse, respectively, in Merem Childhood Obesity Centre Heideheuvel. No other disclosures were reported.
Additional Contributions: We thank all professionals of the Merem Childhood Obesity Centre Heideheuvel for their assistance in the treatments.
Create a personal account or sign in to: