Effectiveness of a Hospital-Based Multidisciplinary Pediatric Weight Management Program: Two-Year Outcomes of PHIT Kids

Abstract

Objective:

For children with obesity, long-term sustainability of weight loss after treatment is difficult to achieve. This study examined 2-year anthropometric outcomes of a moderately intensive group behaviorally based weight management program.

Methods:

One hundred seventy-three children with obesity ages 8–18 years participated with their parent or adult caregiver in a 24-week multicomponent intervention, which was followed by monthly sessions for a total of 2 years. Children were considered treatment completers if they attended ≥50% of the 24 weekly sessions. A multilevel model (multiple assessment time points nested within participants) was used to test person-level change in BMI z-score (BMIz) for program completers between (1) pre- and post-treatment, (2) pretreatment and 24-month follow-up, (3) post-treatment and 12-month follow-up, and (4) post-treatment and 24-month follow-up.

Results:

One hundred twenty-four (72%) of the participants completed the 24-week intervention. Significant reductions in BMIz were observed over the course of treatment (β = −0.03; standard error [SE] = 0.004; t = −6.85; p < 0.001). Completers showed a significant reduction in BMIz between initiation of treatment and 2-year follow-up (n = 110 at 24 weeks; n = 38 at 24 months; β = −0.02; SE = 0.005; t = −4.12; p < 0.001). Children did not show any significant changes in BMIz between post-treatment and 24-month follow-up (β = −0.006; SE = 0.011; t = −0.61; p = 0.54), suggesting that treatment effects were maintained.

Conclusions:

Children maintained treatment gains achieved during a 24-week family-based behavioral weight management program at 2-year follow-up. Although these findings suggest that gains are sustainable, further research is needed to understand how these long-term changes impact child health.

Introduction

Current reports indicate that approximately 32% of children ages 2–19 are overweight or obese.¹ Youth who are obese are at higher risk than their healthy weight peers for a spectrum of illnesses historically encountered in adulthood.² In order to prevent and decrease the severity of these comorbidities, pediatric obesity treatment clinics and programs have emerged across the United States.

Numerous independent research studies have supported the effectiveness of comprehensive, multidisciplinary, family-based programs for treating childhood overweight and obesity.^3,4 Many of these programs are group based, offering a more cost-effective approach compared to individual therapy.⁵ Moreover, a multitude of programs incorporate the family system as part of the intervention, given that results demonstrate the effect of parental behavior on child outcome,⁶ especially in preadolescent populations.^7–9

To be categorized as comprehensive, programs are expected to include components of behavior modification that encourage food monitoring, physical activity (PA) goal setting, and altering the home environment to make it amenable to dietary changes.¹⁰ Family-based interventions frequently involve an adaptation of curricula to match children's developmental levels while covering similar nutritional and PA content. Parents learn strategies to change their dietary habits to model more nutritious options for their child,¹¹ given that parental modeling of healthy behaviors has a significant impact on children's performance in obesity interventions.^6,12,13 Additionally, parents are coached on behavioral management techniques to curb their unhealthy eating patterns and maintain lifestyle changes addressed in the program.

The 2010 systematic review of weight management interventions performed by the US Preventive Services Task Force (USPSTF)⁴ found that for children with obesity age 6 and older, comprehensive behavioral interventions of moderate-to-high intensity (defined as having a minimum of 26 contact hours) were the most effective in reducing BMI at 12 months after start of treatment. However, a limited number of studies have measured long-term outcomes from pediatric weight management programs beyond 12 months. Many investigators have found it difficult to reach patients for follow-up assessment 1 year or longer after the end of their treatment period. For example, Reinehr and colleagues¹⁴ noted that follow-up data were missing for 92% of patients at 2-year follow-up, and that this limitation makes it difficult for many centers to examine the long-term effectiveness of their interventions. Yet, the few available studies in which longer-term outcomes are included report findings that suggest sustained effectiveness of these interventions. For example, in 2010, Reinehr and colleagues published a 5-year follow-up study in which 12 months postintervention BMI z-score (BMIz) was a significant predictor of long-term BMIz reduction.¹⁵ In 2007, Epstein and colleagues reported favorable 24-month to 10-year outcomes of family-based behavioral treatment on BMIz.¹⁶ Despite these promising findings, the USPSTF reported on the need for more lifestyle interventions aimed at children and adolescents, targeting key health behaviors such as dietary changes, PA, and behavioral counseling. The USPSTF recognized that a limited number of lifestyle interventions have been published, and evidence was limited on the long-term sustainability of weight loss after completion of these interventions.⁴ Therefore, the present study aimed to determine the 24-month outcomes of a moderate-intensity stage 3 intervention for preadolescents and adolescents with obesity presenting to a Midwest tertiary care children's hospital. Study hypotheses included that (1) youth would demonstrate reductions in BMIz over the course of treatment and (2) reductions in BMIz would be maintained at 24-month follow-up.

Methods

Participants

Children with obesity between 8 and 18 years old participated in a 24-week group-based weight management program, Promoting Health In Teens and Kids (PHIT Kids), at a Midwest children's hospital. PHIT Kids group participants were referred from the PHIT Kids Weight Management Clinic, which provides multidisciplinary evaluation and treatment for children with obesity ages 2–18 years. Patients are referred to this clinic from their community primary care provider (PCP) or a hospital-based PCP or subspecialist. Inclusion criteria included a BMI ≥95th percentile for age and sex and availability of an English- or Spanish-speaking parent or caregiver to attend the weight management program with the child. Children with significant intellectual disabilities that prevented age-appropriate participation in group sessions were excluded. Specifically, clinic staff did not refer children requiring full-time paraprofessional assistance in school to the group sessions. These children receive care for their obesity in a separate clinic (Special Needs Weight Management Clinic) at the hospital.

Measures

Child anthropometric data were recorded at baseline, 6 weeks, 12 weeks, 24 weeks, 12 months, 18 months, and 24 months. Child height and weight were measured at PHIT Kids group program sessions, PHIT Kids Weight Management Clinic appointments, and/or other hospital clinic appointments. Height and weight was obtained from the child's medical record from the group program or clinic appointment occurring most closely to the follow-up date (±2 weeks from time points during active treatment and ±2 months during follow-up). Height (to the nearest 0.1 cm) and weight (to the nearest 0.1 kg) at PHIT Kids group program sessions were taken by trained intervention staff using a stadiometer (Harpenden Holtain Model 603; Holtain, Crymych, UK) and a portable SECA digital scale (Seca GmbH & Co. KG, Hamburg, Germany). Child height and weight were used to calculate BMI, BMI percentile, and BMIz using the CDC BMI calculator for the Statistical Analysis Software program (available at http://www.cdc.gov/nccdphp/dnpao/growthcharts/resources/sas.html).

Intervention

The PHIT Kids group program is a multidisciplinary, behaviorally based weight management program for children and adolescents (8–18 years of age) with obesity and their families. The PHIT Kids group program included 24 group-format weekly meetings and was delivered by health educators (with at least a bachelor's degree in dietetics, education, nursing, or related field) in English for children and English-speaking parents and in Spanish for Spanish-speaking parents. The program targeted changes to diet and PA through a curriculum developed by a multidisciplinary clinical and educational team. Nutrition education included topics such as family meals, shopping and eating on a budget, and fruits and vegetables. In addition to education about PA benefits and options, each meeting included at least 30 minutes for actual PA for families. Behavioral strategies taught included goal setting, self-monitoring, and contingency management, among others. The health educators were trained in the curriculum and its delivery and had weekly meetings with faculty to problem-solve issues related to the group program. Although the weekly topic was the same, parents and children received separate curricula, and children (ages 8–12) and adolescents (ages 13–18) were divided into two groups based on age to deliver age-appropriate content. Each group (tweens, teens, and parents) received guidance and feedback about their role as an individual and as a family member. Meal and snack demonstrations were utilized to reinforce key educational topics. Other details about the PHIT Kids group curriculum have been previously published.¹⁷

Subsequent to the group-format program, families were invited to participate in a maintenance program, which consisted of follow-up visits, as indicated, to the PHIT Kids Weight Management Clinic and monthly group sessions led by the health educators. The monthly maintenance group sessions included educational activities, active games, or group exercise and often a meal or snack demonstration. The educational activities reinforced the curriculum delivered during the evening group program, and the focus of maintenance sessions was to provide families with support and a dedicated space to practice behavior change.

Procedures

Participant demographic and anthropometric data were used in the current study. This information was collected as part of a larger battery of patient outcome measures (e.g., laboratory measures and measures of psychosocial functioning), but limited response rate on these other variables precluded their inclusion in the current analysis. Before measurement collection, the data collection procedures included in the PHIT Kids group program were approved by the Children's Mercy Hospital Institutional Review Board (IRB# 05 02-024E).

Statistical Analysis

Children considered completers of the weight management program (i.e., attended ≥50% of the 24 weekly sessions) were included in analyses. Before analyses, it was examined whether these children differed from noncompleters on demographic characteristics. Of the 173 children enrolled in the PHIT program, 124 (72%) were completers. Completers and noncompleters did significantly differ on insurance status (χ² = 5.08; p = 0.02). For noncompleters, 67% of families had Medicaid, no insurance, or received financial assistance from the hospital, in comparison to only 48% of families who completed treatment. However, the groups did not differ on child sex, ethnicity, age, or BMIz at baseline.

A multilevel model (MLM) was used to test person-level change in BMIz between (1) pre- and post-treatment, (2) pretreatment and 24-month follow-up, (3) post-treatment and 12-month follow-up, and (4) post-treatment and 24-month follow-up. MLM has advantages for longitudinal analyses because it does not require an equal number of observations for each person and allows for management of missing data without the need for data imputation.¹⁸ Thus, all participant cases were retained in our analyses despite attrition, which reduces the biases possible in long-term program outcomes analyses resulting from participant drop-out. The model specified was person-level repeated measures of BMIz (level 1) within individuals (level 2), using a random slope and intercept and a homoscedastic error structure. The possible interactions between (1) child age and BMIz change and (2) insurance status and change in BMIz were explored to identify whether these child characteristics related to differences in the trajectories of BMIz change.

Analyses were completed in R software (version 2.15.2; R Foundation for Statistical Computing, Vienna, Austria) using the “lmer” function and the full information maximum likelihood estimator (FIML). This estimator was used, given its advantages for calculating fixed effects in comparison to other estimators (e.g., restricted maximum likelihood), which are the primary variables of interest in this study. Additionally, FIML is appropriate for addressing missing data in longitudinal analyses when data are missing at random or completely at random.¹⁸ t-statistics for the main effect are provided by the model. Given concern about the ambiguity of degrees of freedom calculations for multiple-level models,¹⁸ the “lmer” function does not provide calculations for degrees of freedom; therefore, these are not reported. Instead, a normal approximation was used to provide p values for the inferential statistics, which is appropriate for large sample sizes (n > 100¹⁹).

Although MLM does allow for inclusion of participants with incomplete data and an appropriate estimator was used to address missing data, it is important for interpretation of conclusions to note the level of attrition and missing data across time points. Of the 124 children completing treatment, only 38 (31%) had follow-up data at the 2-year time point. Children who had 24-month follow-up data did not differ from those missing data at this time point on most demographics and characteristics examined (i.e., ethnicity, insurance, age, and baseline BMIz). However, females were more likely to be missing data at 24-month follow-up (χ² = 5.14; p = 0.02).

Results

Approximately two thirds of the 124 participants were female, with a mean age of 12.3 years. Participants represented a number of races/ethnicities (42.7% African American, 29.8% Caucasian, and 25.0% Latino). Sixty-four participants (51.6%) had private insurance, and 60 (48.4%) had Medicaid, used hospital financial assistance to pay for medical services, or had no insurance. Participants' average baseline BMIz was 2.38 (range, 1.67–3.33; standard deviation [SD] = 0.28), corresponding to an average BMI percentile of 98.9%. Approximately 60% of patients (n = 74) presented at baseline with a BMI percentile >99.0%. See Table 1 for complete participant demographics.

Table 1.

Participant Demographics

Demographic	Total sample (n = 124)
Sex (%)
Female	82 (66.13)
Male	42 (33.87)
Ethnicity (%)
African American	53 (42.74)
Caucasian	37 (29.84)
Latino	31 (25.00)
Other	3 (2.42)
Insurance (%)
Private insurance	64 (51.61)
Medicaid/financial Assistance/self-pay	60 (48.39)
Age	12.26 (2.12)
Baseline BMIz	2.38 (SD = 0.28)
Baseline BMI %tile	98.91 (0.82)

BMIz, BMI z-score; SD, standard deviation.

Child age (β < 0.001; standard error [SE], <0.001; t = 1.44; p = 0.15) and insurance status (β = 0.009; SE = 0.008; t = 1.09; p = 0.27) did not moderate change in BMIz from baseline to post-treatment. Therefore, the model examining change in BMIz was reanalyzed without these interaction terms. Significant reductions in BMIz were observed over the course of treatment (β = −0.03; SE = 0.004; t = −6.85; p < 0.001), and the average BMIz at the final treatment session (24 weeks) was 2.32 (SD = 0.32). The slope for change over the course of treatment (β = −0.03) indicates that, on average, children showed a BMIz reduction of 0.03 between each assessment time point. Regarding long-term follow-up, no significant age by change over time (β < 0.001; SE <0.001; t = 1.31; p = 0.19) or insurance status by change over time (β = −0.001; SE = 0.010; t = −0.16; p = 0.88) interactions were found for BMIz change from the initiation of treatment to 2-year follow up. When the model was run without these nonsignificant interaction terms, children showed a significant reduction in BMIz between initiation of treatment and 2-year follow-up (β = −0.02; SE = 0.005; t = −4.12; p < 0.001) with an average BMIz of 2.27 (SD = 0.44) at 2-year follow-up. The slope for BMIz change between start of treatment and 2-year follow-up (β = −0.02) demonstrates that children, on average, showed a BMIz reduction of .02 between each assessment time point. Of note, 89% of children's anthropometric follow-up data was obtained from either the PHIT Kids maintenance sessions or the PHIT Kids Weight Management Clinic.

Regarding change between post-treatment and long-term follow-up, child age (β < 0.001; SE <0.001; t = −0.25; p = 0.81) and insurance status (β = −0.019; SE = 0.021; t = −0.88; p = 0.38) did not moderate change in BMIz. Further, children did not show any significant changes in BMIz between post-treatment and 24 month follow-up (β = −0.006; SE = 0.011; t = −0.61; p = 0.54), suggesting that treatment effects were maintained. See Table 2 for average BMIz at each time point. Table 3 presents the number of participants who demonstrated reduction, stabilization, or increase in BMIz between pre- and post-treatment and between pretreatment and follow-up.

Table 2.

Average BMIz at Various Time Points

Baseline (n = 124)	2.38 (SD = 0.28)
12 weeks (n = 119)	2.33 (SD = 0.30)
24 weeks^a (n = 110)	2.32 (SD = 0.32)
12 months (n = 68)	2.31 (SD = 0.33)
18 months (n = 45)	2.29 (SD = 0.36)
24 months^b (n = 37)	2.27 (SD = 0.44)

Indicates end of the weekly group treatment program.

Indicates end of the follow-up period.

BMIz, BMI z-score; SD, standard deviation.

Table 3.

Participant-Level Change in BMIz

Baseline to post-treatment (n = 110)	n (%)
BMIz decrease	76 (69.09)	Average decrease = −0.10(SD = 0.08; range = −0.45 to −0.01)
BMIz stable	4 (3.64)
BMIz increase	30 (27.27)	Average increase = 0.04(SD = 0.03; range = 0.01–0.15)

Baseline to follow-up (n = 37)
BMIz decrease	23 (62.16)	Average decrease = −0.25(SD = 0.28; range = −1.18 to −0.01)
BMIz stable	1 (2.70)
BMIz increase	13 (35.14)	Average increase = 0.13(SD = 0.10; range = 0.04–0.31)

BMIz, BMI z-score; SD, standard deviation.

Discussion

Children maintained treatment gains (i.e., reduction in BMIz) achieved during a 24-week family-based behavioral weight management program at 2-year follow-up, suggesting that such programs have the potential to impact long-term changes in child health. These findings are consistent with past research^14,16,22 that has identified long-term maintenance of treatment effects in children with follow-up data available. The treatment program examined in the current study also supported long-term maintenance of treatment gains by providing participants the opportunity for follow-up visits in an outpatient weight management clinic and monthly group meetings. Interestingly, though these visits may have contributed to the maintenance of treatment effects, children did not show greater reductions in BMIz between post-treatment and follow-up, suggesting that the maintenance program did not lead to further improvement in body composition (weight status). However, the high rate of attrition and small portion of the sample with 2-year follow-up data may have limited the statistical power for identifying an effect for change in BMIz.

The current study demonstrated that the pediatric weight management group program PHIT Kids is effective for reducing child BMIz in a diverse patient population (70% ethnic minorities) with a high BMI percentile (60% of patients had a baseline BMI percentile >99.00). Past research has highlighted the difficulty in treating children with obesity in ethnically diverse populations and children with a higher BMI percentile,^20,21 particularly outside of a traditional randomized, controlled trial setting, signifying the importance of adapting intervention location and content to address cultural and other differences found in a real-world environment.

Almost three quarters of patients enrolled in PHIT Kids were considered treatment completers, which supports the acceptability of such a program in often underserved populations. However, individuals with Medicaid insurance were less likely to complete treatment than their privately insured counterparts, and females were less likely than males to have 2-year follow-up data. This may limit the generalizability of our findings. Future research should examine ways to tailor interventions to meet the needs of families more likely to withdraw and discover the causes of this attrition. Participant attrition over the course of treatment and follow-up may also have impacted the current study findings and the generalizability of results. First, only completers were included in the current analyses, because we sought to examine treatment effects for individuals receiving an adequate treatment dose. Whereas noncompleters and completers did not differ on most demographic variables, completers were more likely to have commercial insurance, which is consistent with past research that has demonstrated greater attrition in populations receiving Medicaid services.^22,23 Additionally, 2-year follow-up data were only available for around 31% of children included in the current analyses, or approximately 22% of children who sought treatment. Past research examining attrition in pediatric weight management programs has suggested that severely affected children (e.g., children with high BMI percentile) and vulnerable families (e.g., ethnic minorities) may be at increased risk for treatment dropout.²² Given the characteristics of our current sample, this may explain the high rates of attrition. Though these attrition rates are high, they are smaller than those in other studies examining long-term follow-up outcomes of pediatric weight management programs. For example, Reinehr and colleagues¹⁴ found that only 8% of children across 129 centers had follow-up data 2 years after participation in a weight management program. However, it is important to note that Reinehr and colleagues¹⁴ showed significant differences in long-term outcomes when conducting an intent-to-treat analysis using an earlier observation carried forward for missing date in comparison to an analysis using only actual follow-up measurements. Therefore, the findings of the current study should be interpreted with caution. Finally, the fact that the group program participants were referred from a single clinical source (PHIT Kids Weight Management Clinic) in the Midwest United States and that their anthropometric measures were, at times, obtained by nonresearch clinical staff (e.g., while at clinic visits) limits the generalizability of the results.

Although the current study demonstrated statistically significant reductions in BMIz over the course of treatment and maintenance of these effects at 2-year follow-up, future research should examine whether these longer-term improvements reach a level of clinical significance. Average BMIz for the sample declined to 2.27 at 2-year follow-up from 2.38 at baseline. Whereas past research has suggested that BMIz reductions of greater than or equal to 0.25 units over the course of 1 year are required in order to see clinically significant health improvements,²⁴ more recent literature has suggested that changes in BMIz alone are insufficient to depict the full range of potential improvements in children's health status after childhood obesity interventions, such as changes in PA or self-esteem.²⁵ Unfortunately, our sample size for these additional measures did not support definitive analysis of these constructs. Though it is certainly beneficial for children with obesity to reduce their BMIz and maintain that loss, there is mounting evidence to suggest that improvements in cardiovascular and metabolic profiles are observed, even when youth remain obese postintervention.²⁶ Future research should examine whether children participating in weight management programs show maintenance of clinically meaningful treatment effects at long-term follow-up.

Future research should also examine predictors of children maintaining treatment gains at follow-up. Though it may be the case that all children display improved outcomes at follow-up, it may also be the case that certain subgroups of children are more likely to maintain these improvements. Identification of children who may need additional support to achieve and maintain treatment gains after completion of weight management programs could inform development of future treatment and maintenance programs.¹⁶

Conclusions

In conclusion, ethnically diverse adolescents and preadolescents attending a family-based group intervention for obesity successfully maintained their initial BMIz loss over 2 years. Accessibility of such interventions for these populations is important and should be a goal of future clinical practice. Given that these populations may be at increased risk for program dropout, additional supports or tailoring of programs to these families' needs may be required to reach optimal outcomes. Long-term follow-up of patients participating in weight management programs can be extremely challenging, but is necessary to document durability of these interventions and support continuity of care. Staff of pediatric weight management clinics and programs should identify mechanisms within their own healthcare systems to increase capability for long-term follow-up, such as closer involvement with PCPs.

Footnotes

Acknowledgments

S.H. and M.D.G. contributed to study design, data analysis, data interpretation, literature search, and writing of the manuscript. K.B. and C.B. contributed to data collection, data analysis, data interpretation, literature search, and writing of the manuscript. C.O.S. contributed to data analysis, data interpretation, literature search, generation of figures, and writing of the manuscript. K.P.C. contributed to literature search and writing of the manuscript. All authors had final approval of the submitted version.

Author Disclosure Statement

No competing financial interests exist.

References

Ogden

, Carroll

, Kit

, et al. Prevalence of childhood and adult obesity in the United States, 2011–2012. JAMA, 2014; 311:806–814.

Krebs

, Himes

, Jacobson

, et al. Assessment of child and adolescent overweight and obesity. Pediatrics, 2007; 120(Suppl 4):S193–S228.

Oude Luttikhuis

, Baur

, Jansen

, et al. Interventions for treating obesity in children. Cochrane Database Syst Rev, 2009;(1):CD001872.

Whitlock

, O'Connor

, Williams

, et al. Effectiveness of weight management interventions in children: a targeted systematic review for the USPSTF. Pediatrics, 2010; 125:e396–e418.

Goldfield

, Epstein

, Kilanowski

, et al. Cost-effectiveness of group and mixed family-based treatment for childhood obesity. Int J Obes Relat Metab Disord, 2001; 25:1843–1849.

Kitzmann

, Dalton

III , Stanley

, et al. Lifestyle interventions for youth who are overweight: A meta-analytic review. Health Psychol, 2010; 29:91–101.

Golan

, Fainaru

, Weizman

. Role of behaviour modification in the treatment of childhood obesity with the parents as the exclusive agents of change. Int J Obes Relat Metab Disord, 1998; 22:1217–1224.

Golan

, Kaufman

, Shahar

. Childhood obesity treatment: Targeting parents exclusively v. parents and children. Br J Nutr, 2006; 95:1008–1015.

Munsch

, Roth

, Michael

, et al. Randomized controlled comparison of two cognitive behavioral therapies for obese children: Mother versus mother-child cognitive behavioral therapy. Psychother Psychosom, 2008; 77:235–246.

10.

Barlow

. Expert committee recommendations regarding the prevention, assessment, and treatment of child and adolescent overweight and obesity: Summary report. Pediatrics, 2007; 120(Suppl 4):S164–S192.

11.

Epstein

, Paluch

, Saelens

, et al. Changes in eating disorder symptoms with pediatric obesity treatment. J Pediatr, 2001; 139:58–65.

12.

Wilfley

, Tibbs

, Van Buren

, et al. Lifestyle interventions in the treatment of childhood overweight: A meta-analytic review of randomized controlled trials. Health Psychol, 2007; 26:521–532.

13.

Janicke

, Sallinen

, Perri

, et al. Comparison of program costs for parent-only and family-based interventions for pediatric obesity in medically underserved rural settings. J Rural Health, 2009; 25:326–330.

14.

Reinehr

, Widhalm

, l'Allemand

, et al.; APV-Wiss STudy Group and German Competence Net Obesity. Two-year follow-up in 21,784 overweight children and adolescents with lifestyle intervention. Obesity (Silver Spring), 2009; 17:1196–1199.

15.

Reinehr

, Kelber

, Lass

, et al. Body mass index patterns over 5 y in obese children motivated to participate in a 1-y lifestyle interevention: Age as a predictor of long-term success. Am J Clin Nutr, 2010; 91:1165–1171.

16.

Epstein

, Paluch

, Roemmich

, et al. Family-based obesity treatment, then and now: twenty-five years of pediatric obesity treatment. Health Psychol, 2007; 26:381–391.

17.

Dreyer Gillette

, Stough

, Best

, et al. Comparison of a condensed 12-week version and a 24-week version of a family-based pediatric weight management program. Child Obes, 2014; 10:375–382.

18.

Little

, Schnabel

, Baumert

. Modeling longitudinal and multilevel data: Practical issues, applied approaches, and specific examples. Erlbaum: Mahwah, NJ, 2000.

19.

Hox

. Multilevel Analysis: Techniques and Applications, 2nd ed. Routledge: New York, 2010.

20.

Eliakim

, Friedland

, Kowen

, et al. Parental obesity and higher pre-intervention BMI reduce the likelihood of a multidisciplinary childhood obesity program to succeed—A clinical observation. J Pediatr Endocrinol Metab, 2004; 17:1055–1061.

21.

Jelalian

, Hart

, Mehlenbeck

, et al. Predictors of attrition and weight loss in an adolescent weight control program. Obesity, 2008; 16:1318–1323.

22.

Zeller

, Kirk

, Claytor

, et al. Predictors of attrition from a pediatric weight management program. J Pediatr, 2004; 144:466–470.

23.

Skelton

, Beech

. Attrition in paediatric weight management: A review of the literature and new directions. Obes Rev, 2011; 12:e273–e281.

24.

Ford

, Hunt

, Cooper

, et al. What reduction in BMI SDS is required in obese adolescents to improve body composition and cardiometabolic health?. Arch Dis Child, 2010; 95:256–261.

25.

Kolotourou

, Radley

, Chadwick

, et al. Is BMI alone a sufficient outcome to evaluate interventions for childhood obesity?. Child Obes, 2013; 9:350–356.

26.

Kelly

, Barlow

, Rao

, et al. Severe obesity in children and adolescents: Identification, associated health risks, and treatment approaches. A scientific statement from the American Heart Association. Circulation, 2013; 128:1689–1712.