Study design—From 1996 to 1999 we performed a double blind, randomised study with two arms on 130 children aged between 6 months and 6 years. Randomisation was undertaken by a technician with no access to information on the patients or doctors. Three ear, nose, and throat specialists at Lundby Hospital, Gothenburg, Sweden, were engaged in the study.
Patients—Children eligible for inclusion in our study were those who had had recurrent otitis media and who had been either referred by their general practitioner or a paediatrician to the open care unit of the ear, nose, and throat department at Lundby Hospital or were directly seeking medical advice for ear pain. The children had had at least two episodes of acute otitis media during the past six months or five episodes during the past year. At the next occurrence of ear pain the children were examined, and those with a red or pale, bulging, thickened tympanic membrane were included in the study. We excluded those with penicillin allergy, serious underlying disease, immunological deficiency, a valvular heart defect, major lesions in the mouth or nose, a grommet in the ear, or chronic otitis media.
Informed written consent was obtained from one of the parents of each eligible child. The study protocol was approved by the Medical Products Agency.
Antibiotic treatment—Those children with no recurrences during the past month but who had acute otitis media were given phenoxymethylpenicillin (Kåvepenin, AstraZeneca, Sweden) 25 mg/kg bodyweight. Those children with a recurrence within the past month were given amoxicillin clavulanic acid (Spektramox, AstraZeneca) 20 mg/kg bodyweight. Both antibiotics were given twice daily for 10 days.
—The streptococcal spray was made up by isolating α haemolytic streptococci from the opening of the eustachian tube of the healthy children and selecting five strains (of about 800 tested) for their superior ability to inhibit the growth of S pneumoniae, H influenzae, M catarrhalis
, and S pyogenes
(group A streptococci), using a method described earlier.10
The streptococci represented two strains of S sanguis
, two strains of S mitis
, and one strain of S oralis
in equal proportions. They were freeze dried in skimmed milk, reconstituted in 0.9% sodium chloride immediately before use, and kept cold during the treatment period. The mixture corresponded to a suspension of 5×108
colony forming units per millilitre. The viable counts in the bottle at the end of treatment still exceeded 5×106
colony forming units per millilitre. Placebo comprised skimmed milk powder, with the same texture and colour as the spray. Parents were informed both verbally and in writing on how to give the spray, and this was demonstrated at the first visit. The bottle was given to the doctor at follow up visits to ensure that adequate amounts of spray had been given. At least five days of spray treatment (more than 50% of the suspension) had to be given for the patient to be evaluated for efficacy.
Participant flow and follow up analysis
At the first visit (day 1) the child's medical history and background data were recorded and a clinical examination was performed, including otomicroscopy (this is superior to otoscopy and allows a detailed inspection of the tympanic membrane). A nasopharyngeal swab was taken for bacteriological analysis and a 10 day course of antibiotics prescribed. Information about the study was given both in writing (signed by a parent) and verbally.
At the second visit (days 8-10) the children were examined by otomicroscopy and excluded from the study if signs or symptoms of infection were still present. The parent was given a fresh bottle of spray with instructions to give three puffs into each nostril twice a day for 10 days, starting 2-12 hours after the last dose of antibiotic. At the end of spray treatment the children returned for a third visit (days 25-30), and the children or parents, or both, were interviewed about compliance. At the fourth visit (days 55-60) a further 10 day course of spray treatment was started. The last valid visit was between days 88 and 92. At the third and fifth visits the amount of unused spray in the bottle was recorded. A clinical examination was performed at all visits, the status of the tympanic membranes checked by otomicroscopy, and the clinical response classified as cured (normal tympanic membrane), improved (major decrease in signs and symptoms—only applicable at the second visit), secretory otitis media (signs of middle ear fluid, but no signs of infection), or recurrence (new otitis media).
The children or parents, or both, were advised to seek medical help at any time during the study if there were signs and symptoms of new otitis media. If the children needed help from a doctor other than the three specialists, they took a form with them on which the doctor described the status of the ears and the treatment given. Whether a recurrence had taken place was decided on the basis of the formulary and medical records. Adverse events that were spontaneously reported and information obtained by non-leading questions were recorded at follow up visits. The local ethics committee in Gothenburg approved the study.
We used Fisher's exact test and logistic regression, both bivariate and multivariate. We regarded a sample size of 130 patients, 65 in each group, as sufficient for an analysis of the clinical efficacy and safety of treatment. Earlier studies have shown that at least 50% of patients acquire new otitis media during the three months after an episode. We allowed for a drop out rate of 15%.