Stark 2009.

Study characteristics
Methods	RCT. Compared behavioral intervention plus nutrition education to nutrition education alone to improve caloric intake and weight in children with CF. Hypothesis: Children receiving the behavioral plus nutrition education intervention would have a significantly greater increase from pre‐treatment to post‐treatment and through 24‐month follow up on primary and secondary outcomes.
Participants	Population of interest, N = 177 (met eligibility). Number randomised, n = 79. Number of participants received the intervention = 67. group 'nutrition education', n = 34. group 'behavioral intervention plus nutrition education', n = 33. There were 6 drop outs in both arms prior to treatment, leaving 67 participants for analysis. Recruited from 5 CF centres located in the Eastern, Midwestern, and Southern USA. Children from 4‐12 years with a diagnosis of CF by sweat test, pancreatic insufficiency; and weight for age and height ≤ 40th percentile.
Interventions	Behavioral intervention and nutrition education in group setting. 1. Nutrition education (n = 34), 2. Behavioral intervention for change around nutrition an energy (Be‐In‐CHARGE!; n = 33) (available online at www.oup.com/us/pediatricpsych). 7 sessions (each 90 minutes): pre‐treatment (session 1), 2 weeks later 5 weekly groups sessions (sessions 2 to 6), 2 weeks later post‐treatment (session 7; follow up).
Outcomes	Primary Outcomes Change in caloric intake and weight pre‐ to post‐treatment. Secondary outcomes %EER (percentage of the estimated energy requirements; calculated by subtracting EER for an active child of same age and gender from individual subject's calorie intake X 100), BMIZ (BMI z scores), weight, %EER, BMIZ (assessed at baseline and post‐treatment) and additionally height, HAZ, and FEV1 were examined at 24 months following treatment. Caloric intake was assessed at baseline, post‐treatment, 3, 6, 12, 18 and 24, months follow up.
Notes	In one report effects of the intervention on 'Family Interactions at Mealtime' were reported, but data was not sufficient. For detailed information on this specific outcome the leading author has been contacted and we are waiting information. Treatment fidelity: the authors reported that 'treatment fidelity was assessed by raters coding 4 videotapes from each of the 7 sessions for each intervention.' (Stark 2009, p.917).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Participants were 'randomised to the treatment arms by coin flip by research assistant and postdoctoral fellow together' (Stark 2009, p.916).
Allocation concealment (selection bias)	Low risk	Assignment could not been foreseen by participants and investigators enrolling participants because of coin flipping by research assistant and postdoctoral fellow together.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	The authors of the study state that 'families were never explicitly told which treatment they had been assigned' (Stark et al 2009, p.916). But, 'as with any behavioral intervention, it is not possible to keep subjects unaware of the treatment they are receiving or therapists the treatment they are providing' (Stark 2009, p.921). No details are provided about blinding of outcome assessors.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Of the 79 enrolled children 40 were assigned to the nutrition education group (NE) and 39 to the behaviour plus nutrition education group. There have been 6 drop outs in both arms prior to treatment. Data of 67 children was available for analysis post‐treatment (NE n = 33 and behavioural plus nutrition education intervention n = 34). 24 month follow‐up data of 28 children in the behaviour plus nutrition education intervention group and of 31 children in the NE group was available for analysis. The authors provided a flow diagram of participants randomised to both study arms and assessed at each point in time from baseline to 24‐month follow up (seeStark 2009, p.916 Figure 1).
Selective reporting (reporting bias)	Low risk	The study protocol is available and all of the study’s pre‐specified outcomes have been reported.