CONTEXT: Although practice guidelines suggest that primary care providers working with children and adolescents incorporate BMI surveillance and counseling into routine practice, the evidence base for this practice is unclear.
OBJECTIVE: To determine the effect of brief, primary care interventions for pediatric weight management on BMI.
DATA SOURCES: Medline, CENTRAL, Embase, PsycInfo, and CINAHL were searched for relevant publications from January 1976 to March 2016 and cross-referenced with published studies.
STUDY SELECTION: Eligible studies were randomized controlled trials and quasi-experimental studies that compared the effect of office-based primary care weight management interventions to any control intervention on percent BMI or BMI z scores in children aged 2 to 18 years.
DATA EXTRACTION: Two reviewers independently screened sources, extracted data on participant, intervention, and study characteristics, z-BMI/percent BMI, harms, and study quality using the Cochrane and Newcastle-Ottawa risk of bias tools.
RESULTS: A random effects model was used to pool the effect size across eligible 10 randomized controlled trials and 2 quasi-experimental studies. Compared with usual care or control treatment, brief interventions feasible for primary care were associated with a significant but small reduction in BMI z score (–0.04, [95% confidence interval, –0.08 to –0.01]; P = .02) and a nonsignificant effect on body satisfaction (standardized mean difference 0.00, [95% confidence interval, –0.21 to 0.22]; P = .98).
LIMITATIONS: Studies had methodological limitations, follow-up was brief, and adverse effects were not commonly measured.
CONCLUSIONS: BMI surveillance and counseling has a marginal effect on BMI, highlighting the need for revised practice guidelines and the development of novel approaches for providers to address this problem.
- CI —
- confidence interval
- RCT —
- randomized controlled trial
- SMD —
- standardized mean difference
Concerns about the rising prevalence of pediatric obesity, as well as the associated comorbidities and long-term medical consequences, have led to well-publicized public health initiatives to reduce obesity in youth. In this effort, primary care practitioners have been charged with the task of identifying and intervening when at-risk young patients present for a routine appointment.1–3 Although these recommendations seem reasonable, there are some concerning gaps in the science. In a 2005 comprehensive literature review, there were no studies that addressed the key question of whether screening/intervention at the level of primary care for overweight in children and adolescents improves behavior, health outcomes, or weight.2 Moreover, a 2010 systematic review of obesity interventions feasible for implementation in a pediatric primary care setting found consistently poor quality studies with the majority showing little to no change in BMI or other physiologic measures (eg, lipid levels, glucose tolerance, blood pressure, physical fitness measures).4
According to the American Academy of Pediatrics, the recommendation regarding screening and behavioral counseling for children at risk for obesity is largely extrapolated from primary care-based prevention in other areas, such as physician conversations about smoking cessation or breastfeeding.4 Not only are these behaviors distinct from weight management, the latter also differs in that evidence suggests that physicians’ conversations about weight loss may have unintended consequences including increasing weight-related stigma, dieting behaviors, consequent binge-eating and weight gain, as well as risk for eating disorders.5–8 In addition, data have suggested that perceived weight stigma, including being weighed and given feedback about gaining weight, contributes to adults with higher BMIs avoiding or delaying necessary and routine medical appointments.9 Similarly, a recent study has found that overweight and obese children are more likely to receive routine medical care in an emergency department, as opposed to a primary care setting.10 Although no causal mechanism was reported, the study suggests the potential for overweight/obese pediatric patients to underutilize preventative health care services in a manner comparable to their adult counterparts.10
Understanding the balance of potential benefits and harms is particularly important in light of research suggesting the financial impact of these interventions to both the family and society is considerable.11–14 If these interventions have a marginal benefit, resources may be better used developing novel programs or directed toward interventions that have a larger impact on children’s health and well-being. Consequently, high-quality empirical data regarding both the benefits and possible harms of screening and behavioral counseling for pediatric obesity prevention conducted within the primary care setting are needed.
The objective of this study was to summarize the available observational and interventional evidence in a systematic review and meta-analysis to determine the effect of typical primary care, office-based, weight management interventions (eg, motivational interviewing, lifestyle modification education) compared with any control intervention (eg, usual care, no intervention, BMI feedback only, active control treatment) on BMI in children and adolescents aged 2 to 18 years. Although several systematic reviews have summarized primary care interventions for pediatric weight management,2,15 these reviews have included studies with substantial threats to external validity, including aspects of intervention design and delivery that are not feasible for implementation in primary care (eg, home visits, 18 session protocols, specialized obesity treatment, behavioral specialist–delivered treatment). Because the goal of this study was to understand how physicians’ conversations about children’s weight, as well as guidance and interventions typically offered in primary care influence children’s BMI, we limited our focus to brief interventions appropriate and feasible for the average primary care setting rather than interventions representative of specialty weight management services. A second goal of this study was to examine potential adverse effects of these practices.
Using an unpublished review protocol (see Supplemental Information) that followed PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, this review set out to determine the effect of primary care–level interventions for weight management on z-BMI or BMI percentile in children and adolescents (ages 2–18).
Patients, carers, and laypeople were not systematically involved in the development of the research question, study design, or outcome measures and were not involved in the implementation of the study. However, the idea behind this study originated from adolescents’ experiences with routine primary care BMI monitoring and healthy habits coaching during primary care visits. There are no plans to disseminate the results of the study to patients or caregivers but results will be used to inform primary care practices related to obesity prevention.
Eligible studies were randomized controlled trials (RCTs), quasi-experimental trials, nonrandomized trials, and prospective cohort studies published in any language that compared the effect of office-based primary care weight management interventions (eg, lifestyle modification education, BMI feedback and lifestyle counseling, motivational interviewing/solution-focused therapy) to any control intervention (eg, usual care, no intervention, BMI feedback only, active control treatment) on BMI in children and adolescents aged 2 to 18 years. Studies that were considered eligible included those where the majority of the intervention was delivered by staff members routinely involved in primary care or by research assistants supervised by disciplines routinely involved in primary care (eg, physicians, physicians-in-training, nurse practitioners, physicians assistants, registered nurses, bachelor’s degree health educators) as opposed to specialists or staff members not regularly involved in primary care (eg, psychologists, physical therapists, dieticians). Eligible studies assessed percent BMI or BMI z scores before and at the end of and/or after treatment.
We excluded studies that included patients presenting for targeted weight management services at non-primary care/specialty clinics (eg weight management clinics, endocrinology, bariatric surgery, psychology) and studies of interventions representative of specialty weight management services (eg, family-based behavioral treatment of obesity, intensive behavioral treatment of obesity), or with interventionists that had specialty training in behavioral interventions (eg, psychologists, psychology graduate students) or interventions that were beyond the scope of practice for typical primary care staff members. Dieticians could be involved as long as a primary care staff member was the primary interventionist.
Information Sources and Searches
With input from a methodologist (M.H.M.) with expertise in conducting systematic reviews, a reference librarian (P.J.E.) designed and conducted the electronic search strategy. This systematic search included electronic databases (Medline, CENTRAL, Embase, PsycInfo, and CINAHL) from January 1, 1976, to March 25, 2016. We used a combination of text words and indexed terms related to “primary health care,” BMI,” “child” or “adolescent,” and “intervention” (see Supplemental Information). To identify additional candidate studies, we reviewed the reference section of each of the eligible primary studies and of narrative and systematic reviews.
Working independently and in duplicate, reviewers (J.L. and L.A.S.) screened all abstracts and titles. Reviewers obtained all potentially eligible studies in full text. Acceptable chance adjusted agreement (κ = 0.70) was observed between the 2 reviewers who determined the eligibility of full text reports. Reviewers resolved disagreements by consensus or arbitration (A.K.).
Data Collection Process
Using a pilot-tested computerized extraction form, J.L. and L.A.S., working in duplicate, abstracted data describing the patient population and treatments studied. In the case of disagreements, the same 2 researchers met to review and resolve discrepancies for final data extraction. Authors were contacted to obtain missing data and to verify the data as abstracted.
Data were abstracted on participant age, gender, ethnicity, and BMI. We extracted information on type of intervention, number and frequency of in-person sessions, number of interim phone calls, and duration of intervention, type of treatment provider, target of intervention (parent only versus parent and child), type of control group (active control versus usual care/wait-list), and duration of follow-up.
The outcomes extracted included percent BMI and BMI z scores (z-BMI) including end of study and/or change from baseline values. Outcomes extracted were those reported at the longest point of complete follow-up.
Pairs of reviewers worked independently to determine the reported risk of bias of eligible RCTs using the Cochrane Collaboration Risk of Bias Tool16 with acceptable interrater agreement.
BMI z scores or converted BMI percentiles to BMI z scores were extracted from the included studies. When BMI data were available without z scores, authors were contacted to obtain these data. The summary measure was the weighted mean difference in change in BMI from baseline to follow-up between children and adolescents who were exposed to primary care weight management interventions and those in control conditions. BMI z scores were then pooled by using the DerSimonian-Laird random effect model, and pooled effects and their 95% confidence intervals17 were estimated.18 Inconsistency was assessed by using the I2 statistic, which describes the proportion of the observed overall between-study variability not due to chance.17 I2 <25% reflects small inconsistency, and I2 > 50% reflects large inconsistency. All statistical analyses were conducted by using Stata 13.1 (StataCorp, College Station, TX).
To explain heterogeneity across studies, a priori hypotheses for subgroup analyses included participant age (≤6 vs >6 years), target of intervention (parent only, child only, parent and child), treatment provider (physician/physician-in-training/nurse practitioner/physician assistant vs registered nurse/health educator), number of in-person sessions (≤4 visits vs >4 visits), interim telephone contact (none vs telephone sessions), duration of intervention (≤6 vs >6 months), control intervention (active control vs usual care/wait-list), follow-up (no follow-up vs follow-up), and study quality including type of study, blinding of outcome assessors and extent of loss to follow-up. We conducted tests for treatment-subgroup interactions,19 considering a significant interaction when P < .05.
After screening 800 abstracts, 27 full text articles were identified. A review of reference sections identified another 54 studies and 27 full text articles. A review of full text articles initially identified 15 article reporting on 10 RCTs20–28 and 4 quasi-experimental studies.29–32 Three studies were excluded because of missing outcome data that the authors were unable to provide.31–33 The final analysis included 13 articles reporting on 10 RCTs12,20–28,34 and 2 quasi-experimental studies.29,30 All of the RCTs had missing methodological quality indicators. We contacted all of the corresponding authors (10 authors replied with additional details). Study selection flow is depicted in Fig 1.
All but 1 of the studies included children who were in the overweight to mildly obese weight range. Two studies recruited both children and adolescents (ages 4–18 and 7–16 years), yet the mean age of these participants was under 12. The majority of studies recruited preadolescent children with 5 studies including participants with a mean age of under 6 years. The majority of the interventions studied offered ≥4 in-person meetings, and most included between-session phone calls from intervention staff members. All but 2 of the interventions used motivational interviewing/solution-focused therapy approaches, and all but 1 of the interventions delivered information regarding nutrition education. In terms of intervention delivery, 9 of treatments were delivered by primary care providers (physicians, physicians-in-training, nurse practitioners, or physician assistants), and 3 were delivered by RN staff members, bachelor’s degree health educators, and/ or research assistants who functioned as a health educator. Only 2 studies had a follow-up period of ≥1 year; all others had follow-up periods of <12 months, including 5 studies that had no follow-up period at all, and only evaluated outcomes at end of treatment. For a description of study and interventions characteristics, see Table 1.
Risk of Bias
Overall, the included RCTs had minimal reporting of methodological features that protect against bias (Table 2). Although 60% of the trials clearly blinded data collectors and assessors, 1 of the studies blinded providers and 3 blinded participants of the interventions. The median loss to follow-up was 14.15%. These numbers should be interpreted cautiously, because half of the RCT studies had no follow-up period posttreatment. Only 3 studies measured potential harms of the intervention (body image dissatisfaction, quality of life, perceived appearance).
The quasi-experimental studies were representative of exposed individuals in the community (Table 3). Both of the studies drew their comparison sample from the same community as the exposed cohort, yet neither of the studies controlled for baseline characteristics. Both of the studies had >30% of the participants lost to follow-up. Neither of the studies examined potential harms of the intervention.
Figure 2 summarizes the results of meta-analysis of the effects of interventions feasible for primary care on BMI over 14 comparisons. Compared with no treatment, usual care, or active control treatments, brief, office-based, primary care–level interventions for pediatric obesity were associated with a significant effect on z-BMI of –0.04, (95% confidence interval [CI],–0.08 to –0.01), P < .02; with no inconsistency across studies (I2 = 0%). This compares to an average effect size of family-based behavioral treatments for pediatric obesity of –0.37, (95% CI,–0.05 to –0.73).35
Compared with no treatment, usual care, or active control treatments, office-based, primary care interventions for pediatric obesity were associated with a nonsignificant effect on body satisfaction (standardized mean difference [SMD] 0.00, [95% CI,–0.21 to 0.22]); P = .98, I2 = 64.1%, child-reported quality of life (SMD 0.06, [95% CI,–0.12 to 0.24]), P = .53, I2 = 0.0%, parent-reported quality of life (SMD 0.13, [95% CI,–0.05 to 0.31]), P = .15, I2 = 0.0%, and physical appearance/self-worth (SMD 0.71 [95% CI, –0.17 to 1.58]), P = .55, I2 = 93.5% (Fig 3). These results suggest that these interventions are not associated with harm, at least with regard to these measures.
Results of subgroup analyses found no significant interactions caused by participant age (≤6 vs >6 years; P = .44), target of intervention (parent veruss child versus parent and child; P = .10, treatment provider (primary care provider versus nurse/health educator; P = .82), interim telephone calls (yes versus no; P = .66), number of sessions (≤4 visits vs >4 visits; P = .65), duration of intervention (≤6 vs >6 months; P = .46), duration of follow-up (posttreatment versus follow-up; P = .70), type of study (RCT versus quasi-experimental; P = .15), control intervention (active control versus usual care/wait-list; P = .07), and study quality including blinding of outcome assessors (blind versus not blind; P = .30), and extent of loss to follow-up (<30% vs ≥30%; P = .10).
This systematic review and meta-analysis found a marginal effect for primary care–based early interventions for pediatric obesity with regard to BMI reduction. To put the finding in context, for a 10-year-old girl with a BMI at the 90th percentile, the effect is equivalent to a difference between the intervention and control groups of 1 kg over a 0- to 3-year follow-up period. Moreover, the change in z-BMI found in this study of –0.04 compares with an average effect found in studies of family-based behavioral weight management treatments of –0.37.35 Because a BMI z score reduction of 0.5 to 0.6 is needed to be sure of clear fat reduction and associated health benefit,36 the approach examined in the reviewed studies is considered to be generally ineffective.
Subgroup analyses, although underpowered, found no differences based on participant age; whether the child or parent participated in the intervention; the intensity, duration, or type of intervention; the type of provider delivering the intervention; whether telephone calls were included; or any other aspect of intervention or study design. The included studies were of variable quality, sample sizes were small, and follow-up was relatively brief (ranging from posttreatment to 3 years). Of concern, only 2 studies followed their sample for a year or more posttreatment, and several studies (5 of 10) did not measure outcomes posttreatment. This is notable because data suggest that primary care–based weight loss programs for adults, although effective for short periods of time, have low rates of long-term participant compliance and do not result in sustained benefits after 2 years.37 Had there been a meaningful effect size, the lack of longer follow-up would have been an important limitation. However, given the overall lack of efficacy and tendency found in obesity research for any BMI effects to typically wash out over time, additional follow-up would be unlikely to threaten our conclusions.
Unfortunately, less than one-third of included studies measured adverse effects. In light of this concern, along with data that demonstrate that seemingly innocuous public health campaigns regarding healthy habits can be perceived to contain inherent weight stigma by young people,38,39 the lack of measurement of potential harms across the majority of studies is a considerable oversight. Because only 2 of the studies recruited adolescents in their sample and the mean age of the participants was <12 years, it is unclear how the results would generalize to an adolescent population and whether adverse effects would be more likely in this group.
A potential harm not measured in this study is that financial resources used in implementing these interventions may be directed away from other, possibly more beneficial health care interventions. In light of the substantial financial cost of these interventions to the family and to the society,12–14 the lack of a meaningful effect of these primary care efforts in reducing a child’s BMI trajectory suggests that resources may be better devoted to other public health agendas and to the development and testing of novel approaches to address this problem in primary care.
Although all studies included BMI data and several measured other parameters, data could not be included for other behavior change variables (eg, physical activity, dietary choices, kilocalorie intake) due to lack of consistency between assessment measures. Of concern, despite this study’s focus on interventions feasible for primary care, most of the included studies contained threats to external validity because they evaluated interventions considerably more elaborate than what practice guidelines suggest. All but 2 of the interventions used motivational interviewing and solution-focused techniques. Several of the interventions implemented computerized decision support tools and systems, physician training, tertiary physician/specialist consultations, frequent follow-up appointments, educational materials, and regular telephone calls. As such, caution is suggested in generalizing these results to standard physician conversations about weight management with children and their parents.
Although the primary literature has limitations, this study has several notable strengths including a focused review question, a comprehensive and systematic literature search, assessment of the methodological quality of included studies that focused on randomized trials and observational studies, and successful author contact. Although there have been previous reviews of the literature, to our knowledge, this is the first meta-analysis of primary care–based interventions for pediatric obesity. The findings of this study were similar to previous systematic reviews that have found little evidence regarding the effectiveness of primary care-based pediatric obesity programs.2,4,40 In adults, a systematic review of primary care-based behavioral treatments for obesity found clinically meaningful weight outcomes for intensive behavioral counseling, yet insufficient evidence to support the feasibility of integrating targeted and intensive counseling into standard primary care practice.41 In light of these results, standard practice guidelines regarding BMI surveillance and counseling should be revised. Moreover, novel approaches feasible for primary care to address pediatric obesity should be developed and tested, as opposed to continuing to pursue programs that do not appear to have sizeable impacts on the pediatric population.
This review suggests that primary care interventions that incorporate a systematic approach to addressing pediatric overweight and obesity (eg, patient-centered communication, patient education, regular visits and phone calls) have only a marginal effect on reducing pediatric overweight and obesity in the short term. Furthermore, the clinical significance of this finding remains questionable, and there continue to be several important knowledge gaps in primary care prevention and weight management interventions. It appears that a paradigm shift might be indicated, in which novel programs are designed and tested, potentially taking into account the evidence about elements of effective behavioral weight loss programs in other settings and about more meaningful markers of health compared with BMI. Large methodologically rigorous RCTs on new approaches implemented with children and adolescents with overweight and obesity are needed to provide evidence as to what interventions might be effective and sustainable in treating this population. Furthermore, it is imperative that researchers examining interventions for pediatric obesity in primary care collect data on potential adverse effects of interventions, including increased dieting behaviors, low self-esteem, perception of weight bias and stigma, and eating-disordered cognitions and behaviors, particularly in adolescents. It is equally important that financial cost be evaluated, with regard to both the individual and society as a whole. For individuals who are at risk or already affected by the serious medical complications and functional impairments related to pediatric obesity, available data appear to support referral to a more intensive behavioral weight management program run by trained specialists who can deliver feedback and counseling about behavior change over an extended period of time, during multiple regular visits.1,2,42
- Accepted June 9, 2016.
- Address for correspondence to Leslie Sim, PhD, Mayo Clinic, 200 First St SW, Rochester, MN 55905. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2016-2497.
- Krebs NF,
- Jacobson MS; American Academy of Pediatrics Committee on Nutrition
- Whitlock EP,
- O’Conner EA,
- Williams SB,
- Beil TL,
- Lutz KW
- Schvey NA,
- Puhl RM,
- Brownell KD
- Wake M,
- Baur LA,
- Gerner B, et al
- Wells GA,
- Shea B,
- O’Connell D, et al
- Altman DG,
- Bland JM
- Wake M,
- Lycett K,
- Clifford SA, et al
- Resnicow K,
- McMaster F,
- Bocian A, et al
- Hunt LP,
- Ford A,
- Sabin MA,
- Crowne EC,
- Shield JPH
- Holzapfel C,
- Cresswell L,
- Ahern AL, et al.
- Garner P,
- Panpanich R,
- Logan S
- Wald ER,
- Moyer SC,
- Eickhoff J,
- Ewing LJ
- Copyright © 2016 by the American Academy of Pediatrics