Effectiveness of ototopical antibiotics for chronic suppurative otitis media in Aboriginal children: a community-based, multicentre, double-blind randomised controlled trial

Children aged less than 15 years with at least 2 weeks of otorrhoea and TM perforation were eligible for inclusion.1 Exclusion criteria were: current febrile illness, current antibiotic use or use in the preceding 2 weeks, allergy to ototopical medications and specific allergy to fluoroquinolones, need for renal dialysis, recent ear surgery or an in-situ grommet or tympanostomy tube, mastoid surgery in the preceding 12 months, congenital ear or hearing problems, obstructed middle ear (eg, polyp), pregnancy, and being unlikely to be resident in the study region over the follow-up period.

Informed consent was obtained from each child’s parents.

A statistical program20 was used to produce balanced random sequences for each site to assign the two ototopical medications to a list of client identification numbers. Participants were then assigned a client number according to the sequence, which was concealed from them and from investigators for the duration of the trial and analysis. None of the investigators, health team nor the children knew which agents had been allocated (double-blind).

Both topical antibiotics had the same pH (4.5), colour, viscosity and similar storage requirements, but they had slightly different odours. To achieve blinding, third parties transferred the agents into brown glass bottles (each containing 8 mL) with droppers, checked them for sterility, labelled them with client numbers, generic product safety information, and site number, packaged them according to the randomisation sequence and distributed them to clinic sites.

Children with bilateral CSOM received treatment for both ears, although only one ear (randomly chosen) was monitored for the study.

A clinical cure was defined as a complete absence of discharge in the middle ear and canal determined by otoscopy. During the treatment period, each child was assessed daily, with half the ototopical treatments applied by health workers and the other half by parents/guardians. If clinical cure was achieved by Day 10, treatment was stopped and the child was reassessed at Day 14. Guardians were instructed in applying eardrops (supine into external meatus), including the use of tragal pressure (pressing several times on the flap of skin in front of the ear canal) which aids middle-ear penetration.21

Ears were cleaned twice each day before the eardrops were administered. Cleaning comprised gentle syringing of the discharge in the external ear canal with 0.5% povidone-iodine solution until the TM perforation was clear of purulent material. Children were not permitted to swim while receiving ototopical medications.

All recruitment, treatment, and clinical assessment was conducted by trained Aboriginal Health Workers (AHWs) at each participating ACCHS. Most workers had previously completed the Commonwealth-sponsored training program by Australian Hearing.22 Acquisition of skills in otoscopy, video otoscopy/photography capture, and audiometry during training were audited by Australian Hearing. All sites had calibrated screening audiometers (Welch Allyn, Skaneateles Falls, NY), soundproof rooms, and otoscopy and video otoscopy equipment.

Baseline measurements included: age, sex, height, weight, living conditions, family size, type of housing, prior history and treatment of otorrhoea, and risk factors for CSOM. The clinical assessment at baseline and at Days 10 and 14 included otoscopy for otorrhoea, grading of the discharge, sample of the discharge for bacteriological examinations, size estimates of the TM perforation, pure-tone audiometry (air conduction) for hearing thresholds at frequencies of 500 Hz, 1000 Hz, 2000 Hz and 4000 Hz (after ear cleaning and drying with tissue paper spears). Pure-tone audiometry was conducted in sound-proof rooms in children aged over 3 years.

An ear swab was obtained by inserting a sterile swab deep in the ear canal before commencing ear cleaning. This method was selected because pathogens isolated from the middle ear correlate well with those from the external canal in patients with CSOM-related otorrhoea.23 All microbiological processing of swabs was blinded and was conducted at the same site in Perth to assure uniform standards. Queensland sites used an interstate courier transport system, while Western Australian sites used existing storage and transport policies (insulated packing with ice bricks), to ensure swabs would reach the laboratory within 72 hours of sampling.

Ear swabs were cultured on cysteine lactose electrolyte deficient agar, blood agar, and horse haematin agar (all incubated at 35°C for 48 hours in 5% CO2), as well as blood agar containing colistin and nalidixic acid and Sabourauds agar (incubated at 35°C and 30°C, respectively, for 48 hours in air). All isolates of conventional pathogens were reported individually. Other organisms not usually colonising the healthy external ear canal were reported only when seen as a heavy predominant growth. Susceptibility testing of significant isolates was performed according to NCCLS guidelines.

The size of TM perforation was assessed by visual grading as a proportion of the TM surface area. A sample of digital images of tympanic membranes (using video otoscopy) were copied to CD-ROM and sent to two otolaryngologists for validation of perforation size as reported by AHWs.

Adherence to treatment was assessed by the number of AHW checks per child and the proportion of scheduled drops applied by the AHW and guardian.

We hypothesised that the use of ciprofloxacin in children would lead to a 20 percentage point increase in resolved CSOM after 9–10 days of twice-daily treatment than would the use of FGD. From cure percentages in previous randomised trials of ototopical aminoglycosides, we calculated that 100 children were needed in each treatment arm to detect an improvement in resolution of CSOM from 50% to 70% with a power of 80% at an α level of 5%.24,25 To allow for a 30% loss to follow-up, 300 children were needed (30–60 per recruitment site).

Univariate statistics: Categorical variables were described as percentages. As the distributions of most of the numerical variables proved to be skewed, medians were used as measures of central tendency, and 25% and 75% percentiles as measures of dispersion for descriptions of numerical information. The hearing threshold level for each child was averaged over four frequencies. Differences in average hearing threshold and grade of perforation were calculated for each child to compare pretreatment and posttreatment outcomes.

Bivariate statistics: Comparisons between treatment groups were performed using exact versions of χ2-type tests for nominal and exact-trend tests for ordinal information. Numerical variables were compared using exact versions of the Mann–Whitney tests. For all statistical tests, exact P values are stated; values below 0.05 were regarded as significant. Confidence limits (95%) for proportions were calculated using exact binomial probabilities.

Multivariable statistics: Stepwise logistic regression modelling for dichotomous outcomes was applied to check for potential confounders of the bivariately observed difference in rates of successful cures between the two treatment regimens. No “intention to treat” analysis was conducted as no information on children lost to follow-up was available. No “interim” analysis was performed.

Overall, cure of CSOM was observed in 71 of the 111 children (64%). Cure was significantly correlated with the type of treatment (P = 0.009). In the FGD group, 29 of the 56 children (51.8%) were cured compared with 42 of the 55 (76.4%) in the CIP group. This reflects an absolute difference of 24.6% (95% CI, 15.8%–33.4%). No significant difference (before and after treatment) was found in the size of TM perforation or the level of hearing impairment, nor was there a difference in these variables according to the ototopical medication used (Box 2).

We found no association between clinical cure and any of the assessed potential confounders (age, sex, history of exposure to cigarette smoke inside dwellings, crowding in households, other children with CSOM in the household, duration and grade of discharge at baseline, history of prior ear infection, history of swimming with head underwater in the preceding 2 weeks, and number of visits by AHWs).

Children who were not cured after a course of ototopical medication (31/83 with baseline audiometry) were more likely to be those with poor hearing at baseline (average hearing threshold, 36.2 dB compared with 23.7 dB, P = 0.01). No relationship between clinical cure and the initial size of the TM perforation was shown.