Kheirandish‐Gozal 2008.

Methods	Cross‐over RCT
Participants	62 children, age 6 to 11 years, with habitual snoring and mild OSA on PSG
Interventions	Budesonide 32 μg per nostril once daily for 6 weeks; comparison against placebo for 6 weeks; washout period of 2 weeks
Outcomes	AHI, total arousal index, respiratory arousal index, nadir SpO₂, sleep efficiency, latency to sleep onset, latency to REM sleep, % REM sleep, end‐tidal CO₂ tension, adenoidal/nasopharyngeal ratio; measured after 6 (phase 1) and 12 weeks (phase 2).
Notes	Enrolment period: April 2004 to May 2006 Funding sources: Investigator‐initiated grant from AstraZeneca Ltd to Dr. Leila Kheirandish‐Gozal Declaration of interest: Dr. Leila Kheirandish‐Gozal is the recipient of investigator‐initiated grants from AstraZeneca Ltd and Merck & Co. for studies in paediatric sleep apnoea; Dr. David Gozal is on the National Speaker Bureau of Merck & Co. and has received honoraria for lectures.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"using a double‐blind, randomisation, crossover procedure generated using random computer numeral allocation"
Allocation concealment (selection bias)	Unclear risk	Research coordinator was involved in both allocation and patient management "children who fulfilled the study inclusion criteria on the basis of the PSG‐1 findings were recruited by 1 of the authors and assigned by a clinical research coordinator to 1 of 2 treatment groups [...] and managed by the same clinical research coordinator who was otherwise not involved in any aspect of the study"
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Unclear whether placebo was indistinguishable from the intervention and whether data analyst was blinded. "randomized, double‐blind, crossover design" "All of the studies were initially scored by a certified technician and were then blindly reviewed by 2 physicians who were experienced in paediatric polysomnography" "The adenoidal/nasopharyngeal ratio was then measured according to the method of Fujioka et al by 1 of the investigators (Dr Gozal), who was blinded to the treatment group and PSG findings of the children."
Incomplete outcome data (attrition bias) All outcomes	High risk	"Of the 62 participants, 43 completed all phases of the protocol, and 19 children completed the first arm of the protocol and subsequently decided to withdraw. Among the latter, 14 children were in the placebo treatment group and 5 were in the budesonide treatment group. Thus, the relative risk for withdrawal from the study was 2.80 for children who started on placebo compared with those who started with intranasal budesonide (95% confidence interval: 1.15– 6.83; P=0.02 after Mantel‐Haenszel correction)."
Selective reporting (reporting bias)	Low risk	All pre‐specified outcomes reported; no evidence of selective reporting of outcomes
Other bias	High risk	1. Study groups were imbalanced at baseline with the placebo‐first group having a mean AHI 2 standard errors lower than in the intervention group 2. The authors presented and compared outcomes from children receiving budesonide in either arm with those who received placebo in the first arm, thereby omitting data from 25 children that received placebo in the second arm 3. Children who received budesonide in the first arm and placebo in the second arm had comparable sleep outcomes after the treatment phase versus the placebo phase, possibly indicating a carry‐over effect