Vernacchio 2014.

Methods	Double‐blind, pragmatic practice‐based RCT
Participants	326 otitis‐prone children ages 6 to 71 months with history of at least 3 clinically diagnosed episodes of acute otitis media (with/without middle ear effusions) in the previous 12 months with at least 1 in the previous 6 months, in good general health (see criteria for ‘good health’ on page 290) and English or Spanish speaking All children were identified and referred by participating physicians from 3 paediatric practice‐based networks (the Slone Center Office‐based Research Network at Boston University, the Pediatric Physicians’ Organization at Children’s (Boston), and the North Carolina Child Health Research Network at the University of North Carolina)
Interventions	Active treatment: Xylitol oral solution (Xylarex; Arbor Pharmaceuticals, Atlanta, GA) consisting of a 66.7% aqueous xylitol solution with the addition of carboxymethylcellulose and potato starch as mucosal adherence agents and natural flavouring. The dose for the xylitol syrup was 7.5 mL 3 times daily for 12 weeks (i.e. 5 g xylitol per dose) Placebo medication: 30% sorbitol oral solution with the addition of carboxymethylcellulose and potato starch as mucosal adherence agents and natural flavouring at a dose of 2.25 g sorbitol 3 times per day
Outcomes	Primary outcome: comparison of time to first clinically diagnosed AOM episode in the two study groups, using proportional hazards model. Secondary outcomes: 1. Proportion of participants in each group with no AOM episodes and no antibiotic use during the study period 2. Reduction in the 3‐month cumulative incidence of AOM episodes, as measured by the risk difference and the hazard rate for AOM (in Poisson regression) 3. Reduction in antibiotic use per 90 days, as measured by the risk difference and the hazard rate (in Poisson regression) Frequency of occurrence of adverse events in either group, as measured by the risk difference
Notes	Analyses were based on the intention‐to‐treat principle. For the comparison of incidence of AOM episodes and antibiotic use per 90 days between the two groups, “rates were calculated individually (events divided by days of enrolment), assuming a zero rate for the 12 participants with no follow‐up.” Higher amount of xylitol per dose and per day over previous studies and the addition of mucosal adherence agents to the xylitol solution as compared to other studies
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	No direct quote was found. There was an acknowledgement to Qiaoli (Lily) Chen who generated randomisation assignments
Allocation concealment (selection bias)	Unclear risk	No information is provided
Blinding (performance bias and detection bias) All outcomes	Low risk	Blinded study syrups. Sorbitol is similar in appearance and taste to xylitol. “After randomisation, study materials, including the blinded study syrup, appropriate oral syringes, and a study calendar, were shipped to the subject’s home.” The principal investigator reviewed medical records in a blinded fashion. Quote: “At the end of the study, medical records from each subject’s primary care physician and from any other health care provider whom the parent identified as having treated the subject during the study period were obtained and reviewed by the principal investigator (LV) in a blinded fashion.”
Incomplete outcome data (attrition bias) All outcomes	High risk	62/160 allocated to xylitol (38.8%) declined participation, were lost to follow‐up or discontinued intervention versus 58/166 allocated to placebo (34.9%)
Selective reporting (reporting bias)	Low risk	As compared to protocol NCT01044030, 2 secondary outcomes were not reported, although they do not represent key outcomes: 1) To determine the effect of viscous‐adherent xylitol on nasopharyngeal and oropharyngeal colonisation with S pneumoniae and non‐typable H influenzae and 2) To compare the antimicrobial resistance patterns of isolates of S pneumoniae and non‐typable H influenzae cultured from the oropharynx and/or nasopharynx of subjects treated with viscous‐adherent xylitol compared to placebo
Other bias	Low risk	Potential bias for outcome assessment: The study did not have any protocol of assessment for healthcare providers. Quote: “For the primary outcome of clinical diagnoses of AOM, the medical record was considered the gold standard. For those cases in which the medical record was not available, the parent’s report of AOM diagnoses was used” The researchers assumed that inaccurate diagnoses would be equal in both groups. Quote: “AOM diagnoses were made by a wide range of clinicians and were not otherwise verified.” “… assuming that AOM diagnoses were equally inaccurate in both study groups, the overall effect would be to bias the study toward a null result.” The trial was performed under FDA New Drug application number 107246 and was registered at www.clinicaltrials.gov (NCT01044030) No financial incentives paid to parents/subjects or to enrolling practices, except a nominal reimbursement for staff time involved in referring potentially eligible parents.

AOM: acute otitis media
 ARI: acute respiratory infection
 n: number
 SD: standard deviation