Henderson 1994.

Methods	6‐week, parallel design trial with 4 arms. Randomised using stratified randomised block design.
Participants	12 children and young adults diagnosed with CF and pancreatic insufficiency (mean (SD) age 12.2 (5.4) years) on genotype or sweat test and able to complete the spirometric tests. Also pancreatic insufficient and plasma vitamin A and E levels within the normal range. 13 gender and age‐matched people, without CF (mean (SD) age 13.4 (6.3) years), 7 male, 6 female.
Interventions	8 x 1 g capsules fish oil (4 capsules twice daily) (3.19 g EPA and 2.21 g DHA) compared with olive oil placebo capsules flavoured to obtain a slight fish taste, over 6 weeks.
Outcomes	Outcomes included in this review: number of people experiencing adverse events; number of deaths; changes in haematological indices; changes in plasma and erythrocyte levels of EPA and DHA and EPA/AA ratio.
Notes	Significantly lower plasma n‐6 fatty acids (linoleic acid and arachidonic acid) noted at baseline in participants with CF compared with the group who did not have CF.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomised using a stratified randomised block design.
Allocation concealment (selection bias)	Unclear risk	Not discussed in paper.
Blinding (performance bias and detection bias) All outcomes	Low risk	Described as double blind ‐ while the capsules were not described as identical, it was stated that the placebo olive oil capsules were flavoured to obtain a slight fish taste which we agreed would be sufficient to blind participants.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Withdrawals from the study were discussed with explanations (20 out of 25 randomised participants completed the study). Study included all participants in the data analysis, which was performed according to the intention‐to‐treat principle.
Selective reporting (reporting bias)	Unclear risk	Protocol not available for comparison, so unable to definitely eliminate selective reporting.
Other bias	Low risk	No additional bias identified.