Powers 2005.

Study characteristics
Methods	Parallel RCT. Comparing behavioral and nutrition treatment (BEH) with usual standard care (CTL). Cross‐over from CTL to BEH offered after stage 1 of study finished with 3‐ and 12‐month follow up. Hypotheses: 1. BEH would improve child's energy intake relative to CTL; 2. improvements would be maintained for 12 months. Computer‐generated block randomisation code with independent blind allocation.
Participants	Population of interest N = 21 with n = 14 meeting criteria. Four withdrew or lost to follow up. Thus, n = 10 participants included aged 18 ‐ 48 months with confirmed diagnosis of CF with pancreatic insufficiency; diagnosis for at least 3 months; and, on unrestricted fat diet. Exclusions for those with developmental delay; receiving supplemental enteral nutrition; an additional diagnosis or medication for condition known to affect growth; baseline diet record of 120% RDA or greater.
Interventions	1. BEH (N = 4) nutritional counselling ‐ targeted at 1 meal each week. 3 focus areas: a. increasing calorie and fat intake; b. dosage and timing of pancreatic enzymes; c. teaching parent management skills. Treatment over 8‐week period with baseline at week 1 and 6; intervention sessions during weeks 3 ‐ 6. 2. CTL (n = 6) scheduled clinic visits every 3 months with CF and dietitian consulted whenever diet and growth issues were identified.
Outcomes	Change in average energy intake % fat intake measured with 7‐day diet diaries 1 week before intervention and 1 week after intervention (8 weeks apart).
Notes	POI: change in average energy intake per day over a 7‐day period from pre‐ to post‐treatment. 71% recruitment rate. No significant difference between the groups at baseline for age, weight, height, sex, nutritional status ethnicity or marital status of caregiver. Attrition: 2 families did not attend regularly and 2 declined to participate.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The authors described the randomisation process in detail. 'The computer‐generated randomisation code was developed independently by the study biostatistician.' (Powers 2005, p. 1443)
Allocation concealment (selection bias)	Low risk	The method of concealment was described in detail by the authors. They reported that a decentralised blinded research staff member then was given a copy of the list. Once a family was enrolled [...], another research staff member called the decentralised research staff member and obtained the random assignment.' (Powers 2005, p. 1443) The authors noted additionally that all relevant study team members were unaware of the specific allocation concealment.
Blinding (performance bias and detection bias) All outcomes	High risk	Participants and personnel providing intervention could not be blinded to intervention as usual for psychological interventions. The primary outcome 'daily calorie intake' was assessed via self‐reported daily diet dairies provided by the parents. The measure, which is not objective, might be influenced by the knowledge of allocation to one of the conditions.
Incomplete outcome data (attrition bias) All outcomes	Low risk	The authors reported that ITT analysis has been conducted for the three‐month follow up. Two families failed to complete the dietary record at this time point. Scores for the two participants were imputed 'by calculating the mean change in: (1) energy intake per day; (2) percentage of RDA; and (3) percentage of fat intake from post‐treatment to 3‐month follow up for the 7 completers. These mean changes then were used to calculate estimated 3‐month follow up data for the 2 noncompleters' (Powers 2005, p. 1446). Furthermore, they reported that two other approaches for imputation revealed similar results.
Selective reporting (reporting bias)	Unclear risk	All of the study’s pre‐specified outcomes have been reported. It is unclear if additional outcomes were pre‐specified in the study protocol but not reported.