A Tailored Family-Based Obesity Intervention: A Randomized Trial
OBJECTIVE: To determine whether a 2-year family-based intervention using frequent contact and limited expert involvement was effective in reducing excessive weight compared with usual care.
METHODS: Two hundred and six overweight and obese (BMI ≥85th percentile) children aged 4 to 8 years were randomized to usual care (UC) or tailored package (TP) sessions at university research rooms. UC families received personalized feedback and generalized advice regarding healthy lifestyles at baseline and 6 months. TP families attended a single multidisciplinary session to develop specific goals suitable for each family, then met with a mentor each month for 12 months, and every third month for another 12 months to discuss progress and provide support. Outcome measurements (anthropometry, questionnaires, dietary intake, accelerometry) were obtained at 0, 12, and 24 months.
RESULTS: BMI at 24 months was significantly lower in TP compared with UC children (difference, 95% confidence interval: –0.34, –0.65 to –0.02), as was BMI z score (–0.12, –0.20 to –0.04) and waist circumference (–1.5, –2.5 to –0.5 cm). TP children consumed more fruit and vegetables (P = .038) and fewer noncore foods (P = .020) than UC children, and fewer noncore foods were available in the home (P = .002). TP children were also more physically active (P = .035). No differences in parental feeding practices, parenting, quality of life, child sleep, or behavior were observed.
CONCLUSIONS: Frequent, low-dose support was effective for reducing excessive weight in predominantly mild to moderately overweight children over a 2-year period. Such initiatives could feasibly be incorporated into primary care.
- MInT —
- Motivational Interviewing and Treatment study
- TP —
- tailored package
- UC —
- usual care
- WHtR —
- waist to height ratio
What’s Known on This Subject:
Although treatment programs for childhood obesity can demonstrate success, long-term outcomes have seldom been evaluated. The benefit of intervention when overweight is identified in a screening assessment and parental recognition of the problem is minimal is understudied.
What This Study Adds:
A low-dose (sessions every 1–3 months), but long-term (2 years), family-based intervention was effective at reducing BMI compared with usual care in children recruited via a weight screening initiative in which many parents had been unaware their child was overweight.
The prevalence of excess weight in children from all demographic groups1 highlights the need for strategies that assist the whole population to make effective behavioral changes. Although much remains unknown regarding the most effective and sustainable solutions,2 a multidisciplinary approach that engages whole families and is tailored to their needs is recommended.3 Although individual studies report mixed success,2 meta-analyses indicate that BMI is ∼1 unit lower in intervention children compared with control children.4–6 However, virtually all interventions are only 6 to 12 months in duration,4–6 with few exceptions.7,8 Whether long-term changes to weight and the behavioral practices underpinning them are sustained has rarely been examined,9,10 but this is important given that shorter studies report greater BMI differences than longer studies.4
Typically, obesity interventions in children include treatment-seeking families or recruit via advertisement or referrals.11–15 The widespread inability of both parents16 and health professionals17 to correctly identify children as overweight limits intervention to the very overweight. This seems sensible given that obese children have greater health risks and are more likely to be obese adults than mildly overweight children.18 However, the mildly overweight represent a large portion of the population,1 and it seems opportune to intervene when smaller, more achievable changes might make a difference over time.19
Screening for overweight provides the opportunity to proactively discuss with parents the weight status of their child.20 Although health professionals are generally reluctant to do so,21 parents are receptive to the information if delivered well,22 and recruitment into intervention can be high.23 However, few studies have recruited families into treatment via weight screening. Families with young overweight children in the LEAP (Live, Eat and Play) trials attended 4 consultations with their general practitioner over 12 weeks. Although no discernible effect on BMI was observed 12 to 15 months postbaseline,8,24 it is feasible that the intervention dose was insufficient, given families only attended a total of 67% to 70% of these 4 sessions.8,24
Frequent, low-dose contact can be a cost-effective mechanism for sustaining weight loss over 2 years in adults.25 However, virtually all obesity treatment interventions in children and families tend to run as intensive, usually weekly, sessions over a short time frame (typically just a few months).2 The aim of this study was to determine whether less intensive contact (monthly) over a longer intervention period (2 years), but tailored to the needs of each family, was an effective intervention in young overweight children compared with usual care (UC).
The Motivational Interviewing and Treatment (MInT) study was a 2-phase randomized controlled trial involving screening children for overweight (phase 1) to recruit into a subsequent intervention (phase 2). Phase 1 investigated whether motivational interviewing was more effective than UC at encouraging parents to participate in the intervention. In phase 2, all overweight children identified in phase 1 (n = 271) could participate in a 2-year intervention comparing UC with a frequent but low-intensity intervention tailored to the needs of their family.23,26,27 Families were invited to participate in phase 1 and, if eligible, could agree to phase 2. Although this manuscript reports the findings from phase 2, sufficient detail regarding phase 1 is provided for context.
Phase 1: Screening
All families with children 4 to 8 years of age enrolled at 9 general practices or attending secondary care clinics at 1 hospital were invited to participate (n = 3704). Of these, 198 (5.3%) families were excluded (mostly medical conditions affecting growth), 1166 (31.4%) declined, and 1023 (27.6%) were rated as noncontactable after 5 attempts. Of the 1317 families that indicated interest, 1093 attended the screening appointment at our university research rooms. Ethical approval was obtained from the Lower Regional South Ethics Committee (LRS/09/09/039), and parents gave informed consent. A separate information sheet and consent form was used for children. Anthropometry and blood pressure measurements on children were obtained as described below while parents completed a comprehensive online questionnaire (Table 1). Parents then received feedback about their child’s weight status using either best practice care (n = 540) or motivational interviewing (n = 553, see Dawson et al23 for additional information). Randomization to feedback condition occurred using random block lengths (Stata 12.0, StataCorp, College Station, TX) after stratifying for practice, with sealed opaque envelopes. Participants were blinded to feedback randomization condition.
Phase 2: Intervention
Parents of all children identified as overweight or obese (BMI ≥85th percentile)28 at screening (n = 271) were invited to attend a follow-up appointment 2 weeks after the screening visit27 (Fig 1). Of the 260 families with overweight children, 196 (75%) agreed to participate in the intervention (206 of 271 children, 76%). If >1 child per family was recruited into the intervention (n = 8 families with 2 children and 1 family with 3 children), all children within that family were randomized to the same intervention condition. Additional outcome measurements (described later) were obtained in these families at a baseline appointment held approximately 1 week after the follow-up appointment. Children were then randomized to either UC (n = 102) or tailored package (TP; n = 104) conditions. Randomization to intervention condition occurred using random block lengths (Stata 12.0, StataCorp) after stratifying for feedback condition. Participants were not blinded to intervention condition because the 2 conditions differed in the amount of contact. Families attended their first intervention appointment ∼2 weeks after the baseline appointment.
UC families (predominantly mothers only) met with a trained researcher at baseline and 6 months. The first appointment lasted 30 to 45 minutes, and parents received individualized feedback about their child’s diet and activity habits, based on the comprehensive data collected at the screening and baseline appointments. The child’s results were compared with guidelines (eg, <2 hours of screen time each day, participate in at least 1 hour of physical activity) or published data (eg, recommended scores for the dietary questionnaire).29 Generalized advice using publicly available resources suitable for children of this age was then provided about healthy eating, physical activity, and sleep. Parents could ask questions and discuss any aspect in more detail. A second appointment at 6 months reviewed progress, provided support and answered queries. No new information/resources were provided, and these sessions lasted 15 to 30 minutes. Estimated total intervention contact time per family over the 2-year intervention was 45 to 75 minutes.
The TP condition consisted of a single multidisciplinary consultant session (usually both parents, mentor, dietitian, exercise specialist, and clinical psychologist all together) followed by regular, brief contact (predominantly mothers only) with a MInT mentor (1 nutritionist, 1 exercise trainer) over the 2-year intervention. The intervention was family-based rather than solely targeting the overweight child. An extensive report was generated before the consultant session using the baseline data and compared with guidelines or published data. All specialists used this report to identify potential targets for change, specific to each family, to guide prioritization. These could be specific (dietary goals) or general (approaches to parenting). The consultant sessions were 1 to 2 hours long in total, with the family taking the lead in identifying targets for change. Each family then met with their MInT mentor, monthly in year 1 and every 3 months in year 2, typically alternating between face-to-face sessions (30–40 minutes) at the university or in their home and phone calls (5–10 minutes). In these sessions, individual goals were negotiated and relevant resources, based on well-established behavioral strategies, were discussed (Table 2). Resources were provided as required rather than providing all families with every resource. Frequency of contact was purposely reduced in year 2 to assess how families managed by themselves with more limited contact. These sessions provided personalized support to each family, while allowing continued monitoring and adjustment to target behaviors over time. Estimated total intervention contact time over the 2-year intervention was 6 to 7 hours per family. The consultants did not meet with the families again, but both mentors were supervised fortnightly by the clinical psychologist. These sessions involved self-review by the mentors, and assistance with intervention plans and problem solving for families.
Outcome assessments were undertaken at baseline (including screening), 12 and 24 months by trained assistants blinded to intervention allocation. Duplicate measures of height (Tanita portable stadiometer), weight (Tanita BC-418), and waist circumference (level of the umbilicus) were obtained after standard techniques. BMI was derived and z scores calculated.28 Waist to height ratio (WHtR) was calculated.30 Estimates of percentage fat mass were obtained by bioelectrical impedance (Tanita BC-418), which provide a good estimate of change in percentage fat at this age.31
Parents completed comprehensive questionnaires at each time point (Table 1). Questionnaires on demographics, motivation, and parental feeding and discipline practices were completed before notification of the child’s weight status. Questionnaires on level of household chaos, quality of life, child behavior, dietary intake, and home food availability were completed at the baseline appointment. Children wore an accelerometer (ActiGraph GT3×, Pensacola, FL) fitted over the right hip for 7 days and 8 nights to measure physical activity and sleep. The ActiGraphs were initialized using 15-second epochs in the uniaxial mode and processed with the normal frequency filter. A valid day was at least 8 hours of wear time, and participants were excluded from the analysis if fewer than 3 valid days of wear were obtained. Moderate-vigorous physical activity was estimated.32 Data were cleaned and scored using an automated script developed in MATLAB (MathWorks, Natick, MA).
Our study has 90% power (P < .05) to detect differences of 0.15 BMI z score units with 73 children in each intervention group (assuming SD of 0.5 and repeated-measures correlation of 0.8).33 Recruiting 202 children allowed for 20% dropout and a design effect of 1.15 to account for sampling of general practices rather than individuals.
The data were analyzed according to modified intention to treat; that is, participants were included if they had at least 1 of the measures obtained at 12 or 24 months. Mixed models with a random effect for participant nested within the practice from which they were recruited were used. To estimate differences at 24 months, the models included terms for treatment and time and their interactions and also adjusted for sex, age, baseline observations and the presence of siblings in the study. Analysis of covariance was used for measures collected on only 2 occasions. A fractional multinomial logit model that uses quasi-maximum likelihood was used to estimate differences in sleep, sedentary, light, and moderate/vigorous activity times obtained from accelerometery.34 Differences between participants and nonparticipants were estimated using logistic regression adjusted for clustering by practice. Because the proportion of missing data for BMI (primary outcome) was small (4.3%), we did not impute the missing data.
Table 3 shows the characteristics of the study population. Children averaged 6.5 years of age, and 45% were male. The majority (75%) was European and came from 2-parent homes (78%), and 38% of mothers were university educated. Almost 70% of mothers were overweight, with an average BMI >29 kg/m2. Children whose families agreed to participate in the intervention (n = 206) did not differ from those who refused (n = 65) in terms of sex (P = .995), age (P = .616), ethnicity (P = .073), or maternal education (P = .052), age (P = .705), or BMI (P = .763), number in the household (P = .091), family structure (P = .203), or child BMI z score (P = .104) at baseline. Retention was high, with 181 children (88%) remaining at 24 months. Children who dropped out of the study (n = 25) did not differ from those remaining in terms of sex (P = .441), age (P = .069), ethnicity (P = .144), maternal education (P = .872), maternal age (P = .179), maternal BMI (P = .677), number in the household (P = .785), or child BMI z score (P = .375) at baseline. However, they were more likely to be from single-parent households (P = .005).
Table 4 presents the differences in body composition at 24 months, adjusted for confounders including feedback condition. Children in the TP had a significantly lower BMI (difference, 95% confidence interval: –0.34, –0.65 to –0.03) and BMI z score (−0.12, –0.20 to –0.04) than children participating in the UC condition. Similar effects were observed for waist circumference (–1.5 cm, –2.5 to –0.5) and WHtR (–0.01, –0.02 to –0.00).
Small differences were observed in a limited number of behavioral variables (Table 5) at 24 months. Parents of TP children reported higher fruit and vegetable intake scores (1.0, 0.0 to 2.1) and lower noncore food intake scores (–0.3, –0.5 to –0.0) with no difference in the intake of sweetened beverages (P = .121). TP families also reported that less noncore foods were present in the home (P = .002). TP children were also more physically active as indicated by a higher mean counts per minute (60, 4–115) from the accelerometer measures, whereas no differences were observed for time in moderate-vigorous physical activity (P = .141), sedentary time (P = .684), or sleep duration (P = .317). No differences in parental feeding practices, various indicators of the home environment, types of motivation, or quality of life were apparent according to intervention group.
Our data demonstrate that providing families of young overweight children with regular low-intensity support can make small but significant differences to body weight and lifestyle behaviors over 2 years. Children in the TP condition had smaller gains in BMI, were more physically active, had improved diets, and reported that fewer noncore foods were available in the home than children in UC.
Although an intervention difference in BMI z score of only 0.12 seems small for obesity treatment, it fits with recent meta-analyses reporting overall differences of 0.06 to 0.09 in young children.4,5 Two other meta-analyses have reported larger intervention differences of 0.31 and 0.47.4,6 However, Ho et al4 showed that study duration was important with much larger effects in shorter studies than those lasting >6 months (0.31 vs 0.09 BMI z score units, respectively). Janicke et al6 included studies in adolescents and showed that age was a significant moderator, with greater effects in older children.6 Direct comparisons are also complicated given that each meta-analysis was examining related but subtly different types of interventions, from educational approaches5 to comprehensive lifestyle interventions.6 Alternatively, the relatively small intervention effect we observed may have arisen because our children were recruited through screening rather than families seeking treatment. This is reflected by our mean BMI z scores, which are considerably lower than is typically observed in childhood obesity interventions2 but more comparable to trials who recruited after screening.8,24 Children who are mildly overweight, and thus only identified via screening initiatives, may be an important group in the population whose needs are not met by current referral programs. They may also not be met by community-based obesity prevention initiatives, given that such approaches are successful at reducing excess weight in normal weight but not overweight children.33
Our use of UC as the comparator rather than a wait-list control might have also contributed to the intervention differences observed, given that both groups received an intervention. However, Janicke et al6 recently demonstrated that type of control group (standard care vs wait-list control) was not a significant moderator variable in obesity treatment trials. We chose to provide our own “UC” rather than encouraging participants to visit their general practitioner, so that we could ensure participants received the same intervention. Our UC approach, which provided personalized feedback on weight and behaviors with generalized advice regarding behavior change in 2 visits lasting a total of 60 minutes, demonstrated reductions in BMI z score at 12 months that were maintained at 24 months, despite the intervention ending 18 months earlier. The addition of a brief intervention to screening initiatives may therefore be sufficient to prevent further weight gain in young children.
Relatively few behavioral differences existed that might explain the body weight differences. Children in the TP condition reported higher intakes of fruit and vegetables (quarter of an SD) and lower intakes of noncore foods (one-third of an SD). They were also more physically active to a similar degree, small differences that could conceivably make a difference in body weight over time. However, despite parenting being a focus of our TP intervention, no significant differences were observed in terms of parental feeding practices, ineffective parenting, household chaos, or child behavior. This may have occurred because the intervention offered to each family differed according to need such that managing parenting and child behavior was only a focus with some families. However, others35 have suggested that general parenting and demandingness/responsiveness should not be major targets of obesity treatment initiatives because of the small effect sizes observed.
A major strength of our study is the high retention, with 88% of children at study end. We also met virtually all study quality criteria, including blinding of outcome assessors to treatment, allocation concealment, and appropriate statistical analyses.4 The 1 exception was our nonblinding of participants to intervention group, which is generally acknowledged as not possible in lifestyle interventions.4 We had repeated measures of physical activity and sleep from multiday accelerometry, and we also attempted to measure the home food environment. Attendance at intervention sessions was high: all but 2 usual-care participants attended baseline and 90% attended the 6-month session. Families in the TP condition should have received ∼14 sessions; the median attended was 11. However, our study also has some limitations. First, although our study was powered to detect significant differences in our main outcome variable, it may have been insufficiently powered for some of the other secondary outcomes. Second, our interventions were not matched in terms of frequency, timing or duration. Finally, we chose to use a short questionnaire to assess dietary intake,29 which was a pragmatic choice, but interpretation is difficult given the outcomes are scores rather than amounts of food.
Offering parents of young children personalized, brief monthly support with limited expert involvement was an effective strategy for reducing excessive weight in predominantly mild to moderately overweight children over 2 years. However, the positive outcomes in our UC group would suggest that screening for overweight in young children, followed by a minimal 2-session intervention might have the capacity to make small but sustained changes to relative weight that appear to be maintained over 2 years. This alternative approach is considerably less involved and could feasibly be incorporated into the primary care environment with relatively little effort.
- Accepted April 30, 2015.
- Address correspondence to Rachael W. Taylor, PhD, Department of Medicine, University of Otago, PO Box 56, Dunedin 9054, New Zealand. E-mail:
Dr R.W. Taylor was the principal investigator of the Motivational Interviewing and Treatment study and had overall responsibility for it; she conceptualized and designed the study and drafted the initial manuscript; Ms Cox and Drs Brown and Knight provided the clinical psychologist supervision of the mentors; they also contributed to study design, reviewed and revised the manuscript, and approved the final manuscript as submitted. Dr Dawson developed phase 1 of the project in conjunction with Drs Brown and Knight, and reviewed and revised the manuscript; Dr Meredith-Jones was the exercise specialist and reviewed and revised the manuscript; Dr Haszard was a mentor and reviewed and revised the manuscript; Dr Williams designed and undertook all statistical analyses and reviewed and revised the manuscript; Dr B.J. Taylor contributed to study design, provided pediatric consultant services when required, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Supported by the Health Research Council of New Zealand. Dr Dawson was in receipt of a Freemasons New Zealand Fellowship at the time the data were collected. Dr R.W. Taylor is funded by the KPS Research Fellowship. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
This trial has been registered with the Australian New Zealand Clinical Trials Registry (ACTRN12609000749202).
- Spear BA,
- Barlow SE,
- Ervin C,
- et al
- ↵Ho M, Garnett SP, Baur L, Burrows T, Stewart L, Neve M, et al. Effectiveness of lifestyle interventions in child obesity: systematic review with meta-analysis. Pediatrics. 2012;130(6). Available at: www.pediatrics.org/cgi/content/full/e1647
- ↵Janicke DM, Steele RG, Gayes LA, Lim CS, Clifford LM, Schneider EM, et al. Systematic review and meta-analysis of comprehensive behavioral family lifestyle interventions addressing pediatric obesity. J Pediatr Psychol. 2014;39(8):809–825
- Jiang JX,
- Xia XL,
- Greiner T,
- Lian GL,
- Rosenqvist U
- Kalarchian MA,
- Levine MD,
- Arslanian SA,
- et al
- Golley RK,
- Magarey AM,
- Baur LA,
- Steinbeck KS,
- Daniels LA
- Hill JO
- Westwood M,
- Fayter D,
- Hartley S,
- et al
- Klein JD,
- Sesselberg TS,
- Johnson MS,
- et al
- Wake M,
- Baur LA,
- Gerner B,
- et al
- Dale KS,
- McAuley KA,
- Taylor RW,
- et al
- Taylor RW,
- McAuley KA,
- Barbezat W,
- Strong A,
- Williams SM,
- Mann JI
- Pinquart M
- Levesque CS,
- Williams GC,
- Elliot D,
- Pickering MA,
- Bodenhamer B,
- Finley PJ
- White P,
- Gunston J,
- Salmon C,
- Atkinson J,
- Crampton P
- Musher-Eizenman D,
- Holub S
- Arnold DS,
- O’Leary SG,
- Wolff LS,
- Acker MM
- Matheny AP,
- Wachs TD,
- Ludwig JL,
- Phillips K
- Copyright © 2015 by the American Academy of Pediatrics