Attrition and the Management of Pediatric Obesity: An Integrative Review

Abstract

Background:

A key challenge in managing pediatric obesity is the high degree of program attrition, which can reduce therapeutic benefits and contribute to inefficient health services delivery. Our aim was to document and characterize predictors of, and reasons for, attrition in pediatric obesity management.

Methods:

We searched literature published until January 2014 in five databases (CINAHL, EMBASE, MEDLINE, PsycINFO, and Scopus). Articles were included if they were English, included participants 0–18 years of age, focused on pediatric obesity management, incorporated lifestyle and behavioral changes without pharmacotherapy, provided attrition data, and reported information about predictors of, and/or reasons for, attrition from family-based interventions provided in research or clinical settings. Twenty-three articles (n=20 quantitative; n=2 qualitative; n=1 mixed methods) met our inclusion criteria. Clarity of study aims, objectives, methods, and data analysis were appraised using Bowling's checklist.

Results:

Attrition varied according to definition (minimum to maximum, 4–83%; median, 37%). There were few consistent predictors of attrition between studies, although dropout was higher among US-based families receiving public health insurance. Older children were also more likely to discontinue care, but sex and baseline weight status did not predict attrition. The most commonly reported reasons for attrition were logistical barriers and programs not meeting families' needs.

Conclusions:

Developing and evaluating strategies designed to minimize the risk of attrition, especially among families who receive public health insurance and older boys and girls, are needed to optimize the effectiveness of pediatric obesity management.

Introduction

Pediatric obesity is an urgent health issue given its high prevalence and associated comorbidities.^1–4 Given that obesity early in life often persists into adulthood,^5,6 there is a clear need for effective interventions to enable weight management and improve the health and well-being of boys and girls with obesity.⁷ To date, most pediatric obesity management research has examined the efficacy⁸ and effectiveness⁹ of lifestyle and behavioral interventions. Findings from these reports revealed that interventions can help children and youth to achieve clinically meaningful improvements in their weight status and cardiometabolic health. However, intervention-related benefits are often compromised by program attrition, a common phenomenon that impacts up to 75% of participants and their families who start a pediatric weight management intervention.¹⁰

Intervention attendance and completion are positively linked with success in obesity management for adults^11–13 and likely for children as well. Given that children and their families who attend and participate in interventions for managing obesity are likely to experience positive long-term effects,¹⁴ it is reasonable to assume that children with obesity who discontinue care prematurely are less likely to achieve health improvements. Further, attrition can create feelings of failure for both clinicians and families, which can lead families to believe that treatment is ineffective and reduce the likelihood that families will reinitiate treatment in the future. It is not uncommon for attrition to be preceded by a series of no-show or cancelled appointments, and these events can decrease clinic efficiency and clinician productivity delay access to care for those waiting for treatment, and increase healthcare costs.^15,16

Attrition is not unique to pediatric obesity; high levels (e.g., 30–80%) have also been reported in other areas of pediatric healthcare, including children's outpatient mental health services.¹⁷ Mental health researchers have shown that tailored strategies (e.g., motivation enhancement therapy) can reduce the likelihood that families discontinue care prematurely.¹⁸ This experience can inform the development of obesity-specific interventions designed to minimize attrition, ensuring that families derive maximal benefits from available services. To guide intervention development and future research in pediatric obesity, critically reviewing our current knowledge of attrition is a key, formative step. The aim of this review was to characterize predictors of, and reasons for, attrition in the management of pediatric obesity.

Methods

Design

In designing our review, we chose an integrative approach.¹⁹ In contrast to other reviews (e.g., systematic review), an integrative review is advantageous because it is the one method that allows researchers to combine diverse methodologies (e.g., qualitative and quantitative as well as experimental and nonexperimental), which can be useful to inform evidence-based health services. We expected to identify a range of methodological approaches and study designs through our literature search, so an integrated approach was well suited to our research. As others have noted,^20,21 integrative reviews are preferred when studying conceptual issues and complex phenomena, which we believe applies to our research in program attrition.

Search Methods

An electronic literature search strategy was created with the assistance of a research librarian and conducted using search terms to locate studies that were indexed in the following bibliographical databases: CINAHL (January 1982–January 2014), EMBASE (January 1990–January 2014), MEDLINE (January 1946–January 2014), PsycINFO (January 1887–January 2014), and Scopus (January 1966–January 2014). Sample search strategies used in EMBASE and MEDLINE are presented in Table 1. Our initial search was conducted in December 2012 and was updated in February 2014.

Table 1.

Sample EMBASE and MEDLINE Search Strategies

Concept 1		Concept 2		Concept 3
(weight or obesity or life-style^* or lifestyle^) adj2 (program^ or clinic or clinics or center or centers or centre or centres)	AND	retention.mp. or attrition.mp. or exp Patient Dropouts/ or “drop out^”.mp. or dropout^.mp. or success^.mp. or exp Patient Compliance/ or compliance^.mp. or non-return^.mp. or nonreturn^.mp. or non complian^.mp. or noncomplian^.mp or (“non adher^” or nonadher^).mp. or (prematur^* adj end^).mp. or predict^.mp. or motivat^.mp. or enroll^.mp. or (discontin^* or stop or disengag^* or dis-engag^).mp. or “treatment refus^”.mp.	AND	pediatric.mp. or exp Pediatrics/ or child.mp. or exp Child, Preschool/ or exp Child/ or minor.mp. or exp Minors/ or exp Adolescent/ or adolescen^.mp. or teen^.mp. or youth^.mp. or children.mp. or (limit set # to (“all infant (birth to 23 months”)) or “all child (0 to 18 years)”) or ((pediatric^ or paediatric^* or peadiatric^* or child^* or teen^* or adolesc^* or youth or puberty). jn,jw,nj,nw.)

Inclusion Criteria

Articles were included in this review if they (1) were published in English, (2) included 0–18 year olds, (3) had a primary focus on pediatric weight management, (4) incorporated lifestyle and behavioral changes without the use of pharmacotherapy, (5) provided data on attrition (families that discontinued an obesity management intervention before its completion or stopped attending appointments within a clinic), and (6) reported predictors of attrition (variables that predicted dropout) and/or reasons for attrition (descriptions explaining why children/families dropped out) from a family-based intervention that was delivered in a research or clinical setting.

Exclusion Criteria

Studies were excluded if they (1) focused on the primary prevention of obesity, (2) included school- or community-based obesity interventions, (3) provided attrition data (e.g., percent dropout) in the absence of predictors or reasons for dropout, (4) included data from studies with a primary focus on adult weight management, (5) reported composite data that were not exclusive to attrition (e.g., data provided for families who sporadically attended appointments and for families who withdrew from care), and (6) provided attrition data at time points that did not include intervention conclusion.

Search Outcome

Our initial search strategy (Fig. 1) identified 1071 potentially relevant sources of evidence, of which 193 were removed as duplicates. The remaining 878 article titles and abstracts were reviewed by one author (J.D.) to determine relevance. The full text of these manuscripts was reviewed because initial reviews revealed that attrition data were not universally reported in the abstract. A full review of the remaining 154 articles was conducted independently by two authors (J.D. and N.M.I.N.). Reviewers met to discuss their decisions to achieve consensus as to whether articles met inclusion criteria; however, consensus was not achieved for four articles (4 of 154; 2.6%). In these instances, a third author (G.D.C.B.) reviewed the articles and inclusion/exclusion criteria to make a final decision. Our updated search helped us to identify two additional manuscripts that met inclusion criteria. In total, 23 original articles were included in our review (Fig. 2).

Figure 1.

Flow diagram of original integrative review article search.

Figure 2.

Initial and updated literature search processes.

Quality Appraisal, Data Extraction, and Synthesis

Bowling's quality checklist^22,23 was used to appraise the articles (see Table 2), which allowed us to assess and compare study objectives, design, methods, analysis, results, discussion, and clinical implications. For our review, studies were deemed to be of relatively high, moderate, or low quality if they met 12 of 12 (100%), ≥9 of 12 (≥75%), or <9 of 12 (<75%), respectively, of the quality criteria.

Table 2.

Critical Appraisal of Scientific Literature Using Bowling's Quality Assessment Tool

	Barlow and Ohleymeyer⁴⁴	Braet et al.¹⁶	Cote et al.²⁷	De Niet et al.³⁸	Dolinsky et al.³³	Halvorson and Skelton⁴²	Hampl et al.⁴³	Hampl et al.²⁶	Israel et al.⁴⁵	Israel et al.⁴¹	Jelalian et al.²⁸	Jelalian et al.³⁴	Kitscha et al.²⁵	Savoye et al.³⁵	Skelton et al.¹⁰	Skelton et al.³⁰	Sothern et al.³⁶	Tershakovec and Kuppler³⁹	Van den Akker et al.³²	Vignolo et al.³⁷	Walker et al.³¹	Williams et al.⁴⁰	Zeller et al.²⁹
Clearly stated aims and objectives	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔
Study design adequately described	✔	✔	✔	✔	✔	✔	✔	✔		✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔
Appropriate research methods	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔
Appropriate use of tools (reliability and validity)		✔	✔	✔	✔		✔				✔	✔	^*	✔			✔	✔	✔		✔	✔	✔
Adequate description of sample, inclusion and exclusion criteria, response rates	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔		✔	✔	✔	✔	✔	✔
Statistical power discussed					✔		✔				✔	✔	^*	✔	✔						✔
Ethical considerations discussed	✔	✔			✔	✔	✔				✔	✔	✔	✔		✔	✔	✔		✔	✔	✔	✔
Appropriate analyses	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔
Results clear and adequately reported	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔
Discussion of results reported in light of study question and relevant literature	✔	✔	✔	✔	✔	✔	✔	✔		✔	✔	✔	✔	✔	✔	✔		✔	✔	✔	✔	✔	✔
Limitations of research and design discussed	✔	✔	✔	✔	✔	✔	✔	✔			✔	✔		✔	✔	✔		✔	✔	✔	✔	✔
Implications of research discussed	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔	✔

Asterisk (^*) denotes “not applicable.”

All 23 articles were reviewed in detail. Potential predictors of, and reasons for, attrition were summarized to create a data matrix.²⁴ We categorized study findings based on predictors and reasons because we speculated that this dichotomy might help us to differentiate the more quantitative/objective (predictor) data from the qualitative/subjective (reasons) data. Articles were assigned randomly to one of two authors (J.D. and N.M.I.N.) for independent data extraction and organization. To facilitate analysis and interpretation, predictors of attrition were grouped to create categories based on (1) child factors, (2) parent factors, (3) family factors, and (4) health services factors (see Tables 3 and 4 for list of extracted predictors). Reasons for attrition were grouped into five categories according to: (1) logistical barriers, (2) families' needs, wants, or expectations, (3) costs, (4) motivation/readiness, and (5) health services factors. Statistical approaches were classified based on: (1) descriptive plus multivariate, (2) multivariate only, and (3) descriptive only. The manner with which data analyses were conducted (a priori or post hoc) allowed us to differentiate those studies that planned to examine attrition from the onset (a priori) from those that included analyses after the fact (post hoc).

Table 3.

Overview of Results From Articles in Which Reasons and/or Predictors of Attrition Were Investigated With A Priori Analyses

Study (country)	n	Age (years)	Attrition (%)	Definition of attrition	Reasons for attrition	Variables significantly associated with attrition	Variables not significantly associated with attrition
Barlow and Ohleymeyer⁴⁴ (US)	157	1–18	61	Attended ≤2 visits	Logistical barriers, families' needs, cost, motivation health services factors
Braet et al.¹⁶ (Belgium)	72	4–16	47	Attended ≤6/8 sessions	Families' needs, health services factors	Child: Older age,^ainternalizing behavior^aParent: Low motivation	Child: Sex, higher BMI, motivationParent: Age, baseline weight statusFamily: Lower SES,^b structure, bigger size^bHealth services: Expectations of group treatment,^b treatment history
Cote et al.²⁷ (US)	120	5–17	55 (at 3 months)	Did not complete intensive phase	Logistical barriers, families' needs, cost, motivation, health services factors	Family: Public insurance^bHealth services: Lower caregiver-rated quality of care^a	Child: Age,^a sex,^a ethnicity,^a baseline weight status,^a health status^bFamily: SES^a
de Niet et al.³⁸ (Netherlands)	248	8–14	44	Children dropped out anytime during treatment period (0–12 months)		Child: Higher baseline BMI SDS,^b participated in fewer activities,^b did not eat breakfast regularly^bParent: Nonwhite mother^bFamily: Lack of static adaptability structure^b	Child: Age,^b sex,^b internalizing behavior,^b externalizing behavior,^b social problems,^b school problems^bParent: Psychosocial factorsFamily: Number of siblings, SES
Halvorson and Skelton⁴² (US)	194	2–18	57 (at 4 months)	Did not complete intensive phase		Health services: More missed appointments
Hampl et al.²⁶ (US)	24	N/A	32 for follow-up visits	Did not attend scheduled visit and did not reschedule	Logistical barriers, families' needs, cost
Hampl et al.⁴³ (US)	147	Parents of 2–18	N/A	Did not attend appointments within 6 months of the first clinic appointment or attended <25% of program sessions	Logistical barriers, families' expectations, motivation, cost
Israel et al.⁴¹ (US)	111	8–13	18	Completed initial assessment, but not full program		Parent: Degree of marital satisfaction^c Family: Lower SES,^c single-parent household^c	Parent: Baseline weight statusFamily: Sibling weight status
Israel et al.⁴⁵ (US)	54	8–13	26	Withdrew from program or missed 2 consecutive sessions and did not attend following session		Health services: Any adherence, but extremely high baseline adherence to treatment protocols
Jelalian et al.³⁴ (US)	76	13–16	18	Failure to complete end-of-treatment evaluation (16 weeks)		Parent: Higher baseline BMI^a	Child: Sex, higher BMI,^b ethnic minority status,^b psychosocial/behavioral/ lifestyle factorsFamily: SES
Kitscha et al.²⁵ (Canada)	152	2–17	33	Attended ≤2 clinic appointments	Logistical barriers, families' needs, motivation, health services factors
Skelton et al.³⁰ (US)	133	2–18	32 (at 4 months)	Self-identified as dropping out of program, or missed or cancelled visit and did not reschedule	Logistical barriers, families' needs, cost motivation, health services factors	Child: Older age, lower BMI, below-average school performance	Child: Ethnicity, sexFamily: Home environmentHealth services: travel distance, number of appointments before dropout
Tershakovec and Kuppler³⁹ (US)	518	5–17	49	Attended initial and intervention session only		Child: African American ethnicityFamily: Private insurance^c	Child: Sex, baseline weight status
Walker et al.³¹ (US)	1080	0–17	38	Did not return for follow-up visit after initial visit		Child: Male, <6 years^bFamily: Public insurance^bHealth services: Living in tertiary service area,^b enrolment during summer^b	Child: Baseline BMI z-score ≥2.5,^b ethnicity^b
Williams et al.⁴⁰ (US)	204	4–7	32 (at 6 months)	Attended <2 of total scheduled sessions		Child: Ethnic minorityFamily: Lower SES, single parent household, disengagement^b	Child: Sex, baseline weight status, psychosocial/ behavioral/lifestyle factors Family: Chaos,^b rigidity^b
Zeller et al.²⁹ (US)	212	6–17	55 (at 4 months)	Completed ≥1 visit(s), but withdrew before completion of intensive phase		Child: Older age,^a African American ethnicity,^a little BMI change, self-report depressive symptoms,^a lower self-conceptFamily: Public insurance^a	Child: Sex, higher BMI/BMI z-score,^b HDL, LDL, blood pressure, internalizing behavior, externalizing behavior,^b personal adjustment, clinical maladjustment, greater school maladjustment,^b adaptive behaviorParent: Psychosocial factorsFamily: Home environment^bHealth services: Travel distance

By descriptive and multivariate analysis.

By multivariate analysis only.

Approached significance.

SES, socioeconomic status; HDL, high-density lipoprotein cholesterol; LDL, low-density lipoprotein cholesterol; N/A, not applicable.

Table 4.

Overview of Results From Articles in Which Predictors of Attrition Were Investigated With Post-Hoc Analyses

Study	n	Age (years)	Attrition (%)	Definition of attrition	Variables significantly associated with attrition	Variables not significantly associated with attrition
Dolinsky et al.³³ (US)	983	2–20	83	Dropout (not returning for any follow-up visits within 1 year of the first visit); Noncompletion (not attending at least 6 visits within 1 year after entry)	Child: Ethnicity (nonwhite, non-Hispanic), normal cholesterol levels	Child: Age, sex, BMI, abnormal glucose metabolism, elevated blood pressureFamily: Insurance, family history of diabetes
Jelalian et al.²⁸ (US)	76	13–16	26	Did not complete study	Child: Higher baseline BMI	Child: AgeFamily: SES
Savoye et al.³⁵ (US)	209	8–16	56	Did not complete study	Child: Higher baseline diastolic blood pressure	Child: Age, sex, ethnicity, baseline weight status, health status
Skelton et al.¹⁰ (US)	248	2–18	73	Did not complete program during study period		Child: Age, sex, ethnicity, baseline weight statusFamily: Insurance, home environment
Sothern et al.³⁶ (US)	73	7–17	34	Did not complete study	Child: Male sex	Child: Age, ethnicity, baseline weight status, longitudinal weight status
Van den Akker et al.³² (Netherlands)	73	8–15	4 (at 3 months) 33 (at 12 months)	Did not return at end of intensive phase or did not attend follow-up visits	Child: Older age, higher baseline BMI-SDS, less successful in BMI reduction during treatment
Vignolo et al.³⁷ (Italy)	31	6–12	36	Did not complete study		Child: Age, sex, baseline weight status, psychosocial/behavioral/lifestyle factors Parent: Baseline weight status

SES, socioeconomic status; SDS, standard deviation score.

Results

Most studies included in our review were quantitative (20 of 23; 87%), although there were two qualitative studies^25,26 and one mixed-methods report.²⁷ Attrition varied from 4% to 83% (mean, 41%; median, 37%). Sixteen articles used a priori analyses (Table 3) and seven used post-hoc analyses (Table 4) to analyze predictors of, and reasons for, attrition from pediatric weight management. A total of 16 articles summarized predictors of attrition, four summarized reasons for attrition, and three highlighted both predictors of and reasons for attrition (Fig. 2). Of the 23 articles included in the review, 22 were clinic based and one was research based.²⁸ Most studies (18 of 23; 78%) were completed in the United States, and there was no appreciable difference in the degree of attrition reported in US- (mean, 44%) versus non-US-based (mean, 41%) studies. The variety of ways in which attrition was defined prevented us from testing whether the predictors of attrition varied according to earlier or later dropout; some definitions were specific (e.g., children did not return for follow-up after initial appointment), whereas others were general (e.g., children dropped out during treatment period). However, we observed that most studies designed to study attrition a priori tended to take a broader view to the potential predictors by including variables representing multiple levels of influence (e.g., child, parent, family, and health services), whereas those studies that completed post-hoc analyses included (primarily) variables at the child level. Generally, greater sample sizes were used in studies analyzing data a priori, generating more-consistent results compared to studies that used post-hoc analyses. The quality appraisal of studies revealed a mix of relatively high (n=6), moderate (n=14), and low (n=3) quality studies. The most common deficiency that lowered ratings was inadequately described research methods. Irrespective of their quality rating, data from all reports were included in this review. Data heterogeneity precluded a meta-analysis, so analysis and interpretation were completed using a narrative approach.

Predictors of Attrition

Child factors

Four of the seven studies that examined children's age showed a positive correlation with dropout using a priori analyses.^18,29–31 For example, three studies indicated that older age was predictive of attrition.^16,29,30 Apart from one study,³² studies reporting post-hoc analyses indicated that age was not predictive of attrition.^{10,31–33,55,56}

Nine studies^{6,27,29–31,34,38–40} examined the role of children's sex as a predictor of attrition using a priori analyses, and only one³¹ found that sex predicted attrition. Similarly, four of five reports^{10,33,35–37} that included post-hoc analyses indicated that children's sex was not associated with dropout.

Three of six studies that applied multivariate analyses a priori revealed that children's ethnicity was not predictive of attrition,^23,30,34 consistent with a study that used only descriptive analyses.²⁸ The three remaining studies that found ethnicity to predict dropout^29,39,40 showed that Caucasians were more likely to continue care. There were four studies that examined ethnicity through post-hoc analyses,^10,33,35,36 of which only one³³ detected a significant association.

Seven of nine studies^{16,27,29,30,34,38–41} that assessed children's baseline weight status as a predictor a priori found no association with attrition. Similarly, of the seven studies that applied post-hoc analyses,^{10,28,32,33,35–37} only two articles^28,32 found that baseline weight status was associated with attrition. Children's general health status was examined in two studies,^27,35 but was not predictive of attrition. Three studies assessed the association between metabolic risk factors (e.g., blood pressure and cholesterol levels) and program attrition, although findings were inconsistent. Two articles (one that included a priori analyses²⁹ and another that included post-hoc analyses³³) did not detect a relationship between blood pressure and attrition; however, in two articles using post-hoc analyses, one indicated that a higher baseline diastolic blood pressure was associated with dropout,⁵⁵ whereas another found that normal cholesterol levels were associated with dropout.³³

Six studies examined the link between child psychosocial/behavioral/lifestyle variables and attrition.^{16,29,30,34,37,39} Braet and colleagues¹⁶ found that dropout was predicted by internalizing behaviors (e.g., anxiety) at baseline using descriptive and multivariate analyses. Conversely, two studies revealed that children's internalizing behaviors, externalizing behaviors (e.g., aggressive and antisocial/delinquent behavior), social problems, and school problems were not predictive of attrition.^29,38 One study examined child psychosocial/behavioral/lifestyle factors post hoc, but identified no associations with dropout.³⁷

Parent factors

Three studies examined parents' baseline weight status a priori; two^16,41 found that it did not predict dropout, whereas Jelalian and colleagues³⁴ reported that a high parent BMI at baseline predicted attrition. One study³⁷ that included post-hoc analyses reported no link between parental weight status and dropout. Among the parental psychosocial constructs that were examined, low motivation for treatment,¹⁶ and not degree of marital satisfaction⁴¹ predicted attrition.

Family factors

Four of six studies^16,27,34,38 revealed that socioeconomic status (SES) was not predictive of attrition. However, two reports that included descriptive analyses showed that families that dropped out had a lower SES than those who completed the intervention.^40,41 Another report using post-hoc analyses showed that SES did not predict dropout.²⁸

Four studies^27,29,31,39 examined insurance type a priori and found that families holding public insurance (e.g., Medicaid) were more likely to discontinue care, compared to families with private insurance. Using both logistic regression and descriptive statistics, Zeller and colleagues²⁹ revealed that families who discontinued care were more likely to be Medicaid recipients. Other investigators^27,31,39 reported similar findings. Two studies^10,33 that used post-hoc comparisons found no association between insurance type and attrition.

Four of the six studies that investigated the family environment a priori found that it was not predictive of attrition.^16,29,30,40 Conversely, two studies^40,41 that applied descriptive analyses indicated that children from single-parent households were more likely to discontinue care. One study¹⁰ applied post-hoc analyses and found no association between single- or dual-parent households and attrition.

Health services factors

Six studies^{16,27,29,31,41,42} examined different health services factors a priori and reported inconsistent findings. Using multivariate analyses, one report³¹ found that summertime intervention enrollment and a greater travel distance between families' homes and clinic location predicted attrition, and Hampl and colleagues⁴³ had similar results for travel distance; however, others^29,30 applied descriptive analyses to show that travel distance did not impact dropout. Different reports showed that missed appointments did⁴² and did not³⁰ predict attrition. Finally, while low caregiver-rated quality of care predicted attrition,²⁷ expectations from group-based treatment and treatment history¹⁶ were not related to attrition.

Reasons for Attrition

Logistical barriers

Kitscha and colleagues²⁵ completed a telephone survey to examine reasons for dropout from a dietitian-led pediatric weight management program. From this sample, 79% (n=11 of 14) of caregivers identified scheduling, parking, and clinic location as reasons for attrition. Barlow and Ohlemeyer⁴⁴ found that 21% (n=9) of caregivers reported scheduling conflicts as a barrier to care; they also found that 28% (n=12) of parents were concerned that children missed too much school to attend appointments and that 23% (n=10) perceived the clinic to be too far from their home. Skelton and colleagues³⁰ showed that transportation difficulties (37%; n=8), work commitments (41%; n=9), and concerns about children missing too much school (41%; n=9) led to parents discontinuing care. Similar reasons for dropout were reported by others.^26,43

Families' needs, wants, and expectations

Seven studies^{16,25–27,30,43,44} queried families' needs, wants, and expectations regarding weight management services. For instance, 37% (n=16) of participants in one study said the intervention was not what they were looking for⁴⁴; in other studies, 52% (n=12) of children³⁰ and 12% (n=8)²⁷ of families reported that their expectations (e.g., treatment focus and program length) were not met or were mismatched with the program. Whereas some parents (33%; n=22) reported their children's desire to terminate care was the main reason for dropping out,²⁷ few parents (7%; n=1) felt they no longer required professional support.²⁵ Finally, some clinic administrators (n=7; 36%) perceived that families derived little benefit from care,²⁶ which led to family attrition.

Cost

Five studies^{26,27,30,43,44} reported that the cost of clinical visits and insurance coverage were important reasons for dropout. In two studies, 33% (n=22)²⁷ and 21% (n=9)⁴⁴ of parents reported that they had challenges securing insurance coverage for health services, which impacted their decision to terminate care. Among hospital administrators,²⁴ 23% (n=6) believed that the cost of clinic visits negatively influenced family participation.

Motivation/readiness

Caregivers surveyed by Barlow and Ohlemeyer⁴⁴ reported that they withdrew from care because their children (16%; n=7) and families (5%; n=2) were not ready to make the healthy lifestyle changes (e.g., dietary and physical activity) that were recommended by the program. Others reported that 63% (n=15) of parents believed that their children were not ready to make changes.³⁰ Additionally, 7% (n=1) of caregivers described low motivation as a barrier to continuing care.²⁵ Hampl and colleagues⁴³ reported that 39% (n=58) and 37% (n=54) of parents cited that family motivation and mismatched expectations were reasons for nonreturn, respectively.

Health services factors

Caregivers reported that the clinic environment (14%; n=2) and intervention content (7%; n=1) led to family dropout.²⁵ Barlow and colleagues⁴⁴ found that caregivers perceived clinic visits to be either too infrequent (12%; n=5) or too frequent (7%; n=3), which led to attrition. Others^16,27,30 reported that problems with booking and attending appointment were key reasons for attrition. Hampl and colleagues⁴³ reported that problems with scheduling appointment were cited by more than half their sample, and this influenced their decision to drop out.

Discussion

Our integrative review of attrition and the management of pediatric obesity revealed several relevant findings. First, the overall quality of the studies included in our review was high. Second, there was substantial heterogeneity between studies in the prevalence and definition of attrition. Third, despite inconsistent findings across studies regarding variables that predicted attrition, two relatively consistent findings were that (1) US-based families receiving public health insurance and (2) older children were at increased risk of discontinuing care. Finally, families reported a number of practical (e.g., scheduling difficulties) and personal (e.g., unmet expectations) reasons for discontinuing care, insights that can inform clinicians' interactions with families and future interventions designed to minimize the risk of attrition.

In the United States, the prevalence of obesity is high among individuals receiving Medicaid (public insurance)⁴⁶ and, as we showed, are at higher risk of attrition, compared to their peers with private insurance. One plausible contributing factor is the lack of government reimbursement for the cost of obesity-related treatment. Users of Medicaid also tend to have poorer health status than individuals with private insurance coverage,⁴⁷ perhaps because of obesity-related comorbidities, which may be of greater concern to parents than weight status per se. Given that users of Medicaid tend to be have lower family incomes,⁴⁷ issues such as a lack of transportation, child care, and job security⁴⁰ can make it difficult to commit to the lifestyle changes recommended by most clinical programs. Therefore, determining the family-specific characteristics that contribute to the discontinuation of care among recipients of Medicaid, which is likely a proxy measure of other, more proximal issues, can inform care that is tailored to minimize the risk of dropout. Given that only one quarter of the studies in our review were completed outside of the United States, additional evidence from countries with different models for funding health services is needed to establish external validity. The similarity in attrition prevalence between US and non-US studies suggests that more personal or family factors (e.g., social support and interpersonal relationships) may be more important in determining dropout than program funding and type of family insurance coverage.

The tendency for older children to discontinue care more consistently than their younger peers is not surprising. In general, interventions for younger children with obesity tend to involve parents to a great extent, many of which apply a parents as agents of change approach that enlist parents as primary recipients of care.⁴⁸ Interventions designed for older boys and girls typically include parents in supportive, secondary roles, which align with the increasing autonomy and cognitive/social development of children as they mature. These realities emphasize the need for obesity screening, preventive, and management services to be available and accessible to families earlier in children's lives as well as customizing interventions for older children to meet their unique developmental needs.

Several studies revealed that logistical barriers or conflicting activities led to families discontinuing care.^{25–27,30,43,44} Attending regular clinic appointments can result in children missing school and parents taking time off from work, factors that can have a negative impact on learning and family income, respectively. These issues suggest that intervention options that are less burdensome for families (e.g., evening/weekend appointments, Internet-based interventions, or telemedicine for distance support) may better meet families' needs in clinical care.^48–50 Although logistical barriers were reported by many families, these barriers are not exclusive to families who discontinue care. Families who remain engaged in treatment also experience barriers that impact their treatment,¹⁶ but parents' perceptions of the quality of their care, the value they place on the care received, their ability to prioritize competing demands, and their resilience may help them to remain in care while others do not.

Many families discontinued care because their needs, wants, or expectations were not met. A number of the reviewed studies highlighted families' mismatched expectations with the treatment program (e.g., Hampl and colleagues⁴³) and dissatisfaction with treatment outcomes (e.g., Barlow and Ohlemeyer⁴⁴). Conversely, other studies demonstrated families' premature termination of the program resulting from parents' and children's perceived satisfaction with their health outcomes (e.g., Braet and colleagues¹⁶ and Kitscha and colleagues²⁸). Because families are often limited in the types and styles of services they can choose from, understanding the kind of care families need by incorporating screening tools (e.g., educational needs assessment to identify lifestyle knowledge and needs²⁵) can help match families to appropriate services.⁴⁴ Though there is limited research in this area,⁵¹ one recent study of parents whose children discontinued care for weight management revealed that making changes to logistical factors (e.g., clinic hours and location) would have helped them to remain in the program.⁵²

Children's desire to discontinue care was a prominent reason for discontinuing care.^27,53 However, most research has focused on parents' perceptions of pediatric care.⁵⁴ Only one study¹⁶ revealed parent motivation as a contributor to attrition, suggesting that attrition should not be viewed exclusively as patient nonadherence. Clinicians often view lack of motivation as a barrier in managing pediatric obesity,¹⁵ so identifying families at baseline who have lower motivation may be useful in determining when to use additional motivational strategies (e.g., motivational interviewing)⁵⁶ to enhance engagement and retention. This may be an area in which researchers and clinicians working in obesity management can learn from their colleagues in related fields. For instance, mental health professionals specializing in eating disorders (e.g., anorexia nervosa) have a successful history of using motivational assessments to identify readiness and willingness to change unhealthy habits⁵⁶ and providing tailored therapies that match motivation levels in order to optimize treatment effects.⁵⁷ The use of similar clinical measurements and models of care in pediatric weight management represents a logical and timely extension.

It is important to acknowledge the limitations of our review. First, we were limited in completing direct study-to-study comparisons because study designs, interventions, and definitions of attrition varied. Though our narrative analysis provided context, the between-study variability precluded a formal meta-analysis. Second, attrition was conceptualized as a single category,²⁷ regardless of when it occurred. There may be important qualitative differences between families that terminate care early on in treatment, compared to those who drop out at a later phase.^27,38 Establishing a consistent definition of attrition that (1) accounts for attrition at early versus late phases of treatment and (2) differentiates between attrition from structured weight management interventions (with defined start and end points) versus clinical pediatric programs that provide open-ended care can better delineate this phenomenon, which will help to develop strategies to enhance child and family engagement and reduce the risk of attrition.

Conclusion and Future Directions

Attrition from pediatric obesity management remains a challenging issue in delivering health services for families. Our review indicated that public insurance status was predictive of dropout; families who received public insurance (e.g., Medicaid) were more likely to discontinue pediatric weight management care. Children's sex and baseline weight status were not predictive of attrition. Additionally, the most commonly reported reasons for attrition were logistical barriers (e.g., parents taking time off from work or children missing school) and families' needs, wants, or expectations not being met (e.g., dissatisfaction with treatment outcomes). Based on these results, to minimize the risk of attrition, tailoring interventions to families receiving public insurance and minimizing perceptions of common barriers faced by families (e.g., scheduling) are warranted. Further, because child and family characteristics (e.g., demographics) were not primary contributors to patient dropout, we believe that additional research investigating child and parent reasons for their discontinuation of care is warranted. In order to meet the needs of families, the perspectives of both children and parents can be used to develop effective strategies that optimize the likelihood of program completion and treatment success.²⁶ Our review highlighted that research in this field is nascent, so future research to examine other familial factors, beyond demographic and resource-related variables, that may be related to program attrition (e.g., family functioning, social support, and intra- and interpersonal stress) may offer important, practical insights into potentially modifiable factors that clinicians and families can work to address either in advance of, or in concert with, health services to manage pediatric obesity.

Footnotes

Acknowledgments

J.D. was supported by graduate scholarships from the University of Alberta and the Canadian Institutes of Health Research (CIHR) Frederick Banting and Charles Best Canada Graduate Scholarship. L.Z. was supported by the Stollery Children's Hospital Foundation Chair in Autism Research and an Alberta Innovates–Health Solutions Health Scholar award. When this research was conducted, G.D.C.B. was supported by a Population Health Investigator award from Alberta Innovates–Health Solutions and a New Investigator award from the CIHR. The authors thank Ms. Sandy Campbell, BA, MLS, AALIA (CP), medical librarian at the University of Alberta, for assistance in developing and refining the search strategy used in this study.

Author Disclosure Statement

No competing financial interests exist.

References

Fagot-Campagna

, Pettitt

, Engelgau

, et al. Type 2 diabetes among North American children and adolescents: An epidemiologic review and a public health perspective. J Pediatr, 2000; 136:664–672.

Freedman

, Dietz

, Srinivasan

, et al. The relation of overweight to cardiovascular risk factors among children and adolescents: The Bogalusa Heart Study. Pediatrics, 1999; 103:1175–1182.

Kang

, Lee

, Weng

, et al. Body weight status and obstructive sleep apnea in children. Int J Obes (Lond), 2012; 36:920–924.

Must

, Strauss

. Risks and consequences of childhood and adolescent obesity. Int J Obes Relat Metab Disord, 1999; 23(2 Suppl):S2–S11.

Serdula

, Ivery

, Coates

, et al. Do obese children become obese adults? A review of the literature. Prev Med, 1993; 22:167–177.

Power

, Lake

, Cole

. Measurement and long-term health risks of child and adolescent fatness. Int J Obes Relat Metab Disord, 1997; 21:507–526.

Wang

, Lobstein

. Worldwide trends in childhood overweight and obesity. Int J Pediatr Obes, 2006; 1:11–25.

Oude Luttikhuis

, Baur

, Jansen

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

, Garnett

, Baur

, et al. Effectiveness of lifestyle interventions in child obesity: Systematic review with meta-analysis. Pediatrics, 2012; 130:e1647–e1671.

10.

Skelton

, DeMattia

, Flores

. A pediatric weight management program for high-risk populations: A preliminary analysis. Obesity, 2008; 16:1698–1701.

11.

Wadden

, Letizia

. Predictors of attrition and weight loss in patients treated by moderate and severe caloric restriction. In: Wadden

, VanItallie

(eds), Treatment of the Seriously Obese Patient. Guilford: New York, 1992, pp. 383–410.

12.

Fontaine

, Cheskin

. Self-efficacy, attendance, and weight loss in obesity treatment. Addict Behav, 1997; 22:567–570.

13.

Honas

, Early

, Frederickson

, et al. Predictors of attrition in a large clinic-based weight-loss program. Obesity, 2003; 11:888–894.

14.

Epstein

, Valoski

, Wing

, et al. Ten-year outcomes of behavioral family-based treatment for childhood obesity. Health Psychol, 1994; 13:373–383.

15.

Skelton

, Irby

, Beech

, et al. Attrition and family participation in obesity treatment programs: Clinicians' perceptions. Acad Pediatr, 2012; 12:420–428.

16.

Braet

, Jeannin

, Mels

, et al. Ending prematurely a weight loss programme: The impact of child and family characteristics. Clin Psychol Psychother, 2010; 17:406–417.

17.

Warnick

, Bearss

, Weersing

, et al. Shifting the treatment model: Impact on engagement in outpatient therapy. Admin Pol Mental Health, 2014; 41:93–103.

18.

Nock

, Kazdin

. Randomized controlled trial of a brief intervention for increasing participation in parent management training. J Consult Clin Psychol, 2005; 73:872–879.

19.

Broome

. Integrative literature reviews for the development of concepts. In: Rodgers

, Knafl

(eds), Development in Nursing: Foundations, Techniques, and Applications, 2nd ed. Saunders: Philadelphia, PA, 1999, pp. 231–250.

20.

Bowman

. A research synthesis overview. Nurs Sci Quart, 2007; 20:171–176.

21.

Whittemore

, Knafl

. The integrative review: Updated methodology. J Adv Nurs, 2005; 52:546–553.

22.

Desborough

, Forrest

, Parker

. Nurse-led primary healthcare walk-in centres: An integrative literature review. J Adv Nurs, 2012; 68:248–263.

23.

Bowling

. Research Methods in Health: Investigating Health and Health Services, 2nd ed. Open University Press: Berkshire, UK, 2002.

24.

Garrard

. Health Sciences Literature Review Made Easy: The Matrix Method, 3rd ed. Jones & Bartlett: Sudbury, UK, 2010.

25.

Kitscha

, Brunet

, Farmer

, et al. Reasons for non-return to a pediatric weight management program. Can J Diet Pract Res, 2009; 70:89–94.

26.

Hampl

, Paves

, Laubscher

, et al. Patient engagement and attrition in pediatric obesity clinics and programs: results and recommendations. Pediatrics, 2011; 128(2 Suppl):S59–S64.

27.

Cote

, Byczkowski

, Kotagal

, et al. Service quality and attrition: An examination of a pediatric obesity program. Int J Qual Health Care, 2004; 16:165–173.

28.

Jelalian

, Mehlenbeck

, Lloyd-Richardson

, et al. ‘Adventure therapy’ combined with cognitive-behavioral treatment for overweight adolescents. Int J Obes, 2006; 30:31–39.

29.

Zeller

, Kirk

, Claytor

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

30.

Skelton

, Goff

, Ip

, et al. Attrition in a multidisciplinary pediatric weight management clinic. Child Obes, 2011; 7:185–193.

31.

Walker

, Smolkin

, O'Leary

, et al. Predictors of retention and BMI loss or stabilization in obese youth enrolled in a weight loss intervention. Obes Res Clin Pract, 2012; 6:e330–e339.

32.

Van Den Akker

ELT

, Puiman

, Groen

, et al. A cognitive behavioral therapy program for overweight children. J Pediatr, 2007; 151:280–283.

33.

Dolinsky

, Armstrong

, Østbyestbye

. Predictors of attrition from a clinical pediatric obesity treatment program. Clin Pediatr, 2012; 51:1168–1174.

34.

Jelalian

, Hart

, Mehlenbeck

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

35.

Savoye

, Nowicka

, Shaw

, et al. Long-term results of an obesity program in an ethnically diverse pediatric population. Pediatrics, 2011; 127:402–410.

36.

Sothern

, Hunter

, Suskind

, Brown

, Udall

Jr , Blecker

. Motivating the obese child to move: The role of structured exercise in pediatric weight management. Southern Med J, 1999; 92:577–584.

37.

Vignolo

, Rossi

, Bardazza

, et al. Five-year follow-up of a cognitive-behavioural lifestyle multidisciplinary programme for childhood obesity outpatient treatment. Eur J Clin Nutr, 2008; 62:1047–1057.

38.

de Niet

, Timman

, Jongejan

, et al. Predictors of participant dropout at various stages of a pediatric lifestyle program. Pediatrics, 2011; 127:e164–e170.

39.

Tershakovec

, Kuppler

. Ethnicity, insurance type, and follow-up in a pediatric weight management program. Obesity, 2003; 11:17–20.

40.

Williams

, Coday

, Somes

, et al. Risk factors for poor attendance in a family-based pediatric obesity intervention program for young children. J Dev Behav Pediatr, 2010; 31:705–712.

41.

Israel

, Silverman

, Solotar

. An investigation of family influences on initial weight status, attrition, and treatment outcome in a childhood obesity program. Behav Ther, 1986; 17:131–143.

42.

Halvorson

, Skelton

. Appointment attendance in a pediatric weight management clinic. Clin Pediatr., 2012; 51:888–891.

43.

Hampl

, Demeule

, Eneli

, et al. Parent perspectives on attrition from tertiary care pediatric weight management programs. Clin Pediatr, 2013; 52:513–519.

44.

Barlow

, Ohlemeyer

. Parent reasons for nonreturn to a pediatric weight management program. Clin Pediatr, 2006; 45:355–360.

45.

Israel

, Silverman

, Solotar

. Baseline adherence as a predictor of dropout in a children's weight-reduction program. J Consult Clin Psychol, 1987; 55:791–793.

46.

Lee

, Sheer

JLO

, Lopez

, et al. Coverage of obesity treatment: A state-by-state analysis of Medicaid and state insurance laws. Public Health Rep, 2010; 125:596–604.

47.

Health Care Financing A. A Profile of Medicaid: Chartbook 2000. US Department of Health and Human Services: Washington, DC, 2000.

48.

Faith

, Van Horn

, Appel

, et al. Evaluating parents and adult caregivers as “agents of change” for treating obese children: Evidence for parent behavior change strategies and research gaps: A scientific statement from the American Heart Association. Circulation, 2012; 125:1186–1207.

49.

Ambler

, Hagedorn

DWJ

, Ball

GDC

. Referrals for pediatric weight management: The importance of proximity. BMC Health Serv Res, 2010; 10:302.

50.

Haugen

, Tran

, Wyatt

, et al. Using telehealth to increase participation in weight maintenance programs. Obesity, 2007; 15:3067–3077.

51.

Ball

GDC

, Perez Garcia

, Chanoine

, et al. Should I stay or should I go? Understanding families' decisions regarding initiating, continuing, and terminating health services for managing pediatric obesity: The protocol for a multi-center, qualitative study. BMC Health Serv Res, 2012; 12:486.

52.

Sallinen Gaffka

, Frank

, Hampl

, et al. Parents and pediatric weight management attrition: Experiences and recommendations. Child Obes, 2013; 9:1–9.

53.

Skelton

, Beech

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

54.

Litt

, Cuskey

. Satisfaction with health care. A predictor of adolescents' appointment keeping. J Adolesc Health Care, 1984; 5(3):196–200.

55.

Barlow

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

56.

Geller

, Brown

, Srikameswaran

, et al. The psychometric properties of the readiness and motivation questionnaire: A symptom-specific measure of readiness for change in the eating disorders. Psychol Assess, 2013; 25:759–768.

57.

Geller

, Brown

, Srikameswaran

. The efficacy of a brief motivational intervention for individuals with eating disorders: A randomized control trial. Int J Eat Disord, 2011; 44:497–505.