The following changes were made to the eligibility criteria and outcome measures from those stipulated in the protocol:

1. ELIGIBILITY CRITERIA
Types of studies: quasi‐randomised controlled trials were specified in the protocol but excluded from the review.

Types of participants:
Protocol: People of any age with a diagnosis of CSOM meeting the WHO definition
Review: any diagnosis of CSOM as defined by the trial authors.

Types of interventions ‐ the following were stipulated in the protocol:
Intervention: topical (aural) antibiotics (all and individual)
Comparator: no intervention; placebo; other topical antibiotics with and without steroids; systemic antibiotics (all and individual); combination of topical and systemic antibiotics; antiseptics.

However, we have divided these trials into the following three reviews:
* Topical treatment with antibiotics (THIS REVIEW)
comparisons: no treatment or aural toilet, topical antiseptics, various topical antibiotics, excluding steroids
* Systemic versus topical treatments for CSOM
comparisons: any systemic treatment against any topical treatment excluding steroids
* Systemic or topical steroids, as monotherapy or combination therapy
comparisons: no treatment or aural toilet, topical antiseptics, topical antibiotics, systemic antibiotics

2. OUTCOMES
Types of outcome measures ‐ the following primary outcomes have been changed:
Protocol:
* Resolution of CSOM at 2 to 4 weeks and after 4 weeks, according to the following findings:
a) No report of otorrhoea
b) Disappearance of discharge on otoscopy
c) Healing of the eardrum perforation on otoscopy
d) Time to resolution of CSOM according to any other definition of resolution made by the authors, including improvement in mucosal appearance

Review:
* Resolution of CSOM at 2 to 4 weeks, and after 4 weeks, according to the investigators' criteria.
We have also analysed results for treatment failure rather than success.

The others were analysed as separate, secondary outcomes (where reported by trialists):
* Healing of perforation at 2 to 4 weeks, and after 4 weeks;
* Time to resolution of CSOM as defined by the investigators.

The other protocol outcomes were unchanged, except that results before 2 weeks were also included and reported separately.

3. SEARCH STRATEGY
The protocol stated that we would obtain all relevant studies regardless of publication status. However, trials reported in conference proceedings or on posters have not yet been sought, but will be sought for inclusion in an update of this review. Additionally, the further potential sources that we will search for future updates of this review, were also specified in the protocol for this review.

4. METHODS ‐ REVIEWERS' CONTRIBUTIONS
The protocol stated that we would follow the statistical guidelines in the Cochrane ENT Group Guidelines for Reviewers (CochraneENTGuideline, updated November 2000), and that two reviewers (CM and JA) would independently:
* select trials (at least for electronic searches), review the titles and abstracts of articles identified by the search strategy, and assess full texts for eligibility;
* assess the methodological quality of all trials identified as eligible for inclusion (and report the level of agreement between the two reviewers);
* extract data of study characteristics, including methods, participants, interventions and outcomes, and record these on standard forms; and
* enter data onto Review Manager 4.2.

However, while CM and JA independently reviewed titles and abstracts of articles identified by the search strategy, only CM assessed the full texts for eligibility and methodological quality, extracted data, and entered data onto Review Manager 4.2. JA and CG provided a second opinion on trials CM had selected for inclusion; CG reviewed those where there was any ambiguity about the methods used, for the methodological quality and data extraction; and JA provided further information where this had been obtained from authors of trials included in the previous review 'Interventions for chronic suppurative otitis media' (Acuin 1998). The three authors resolved any disagreements through discussion.

5. MINIMISING CONFLICTS OF INTERESTS
The protocol also stated that where one of the authors has worked on a trial in the review, we will seek help from another person at the ENT editorial base, to minimise bias due to potential conflict of interests. However, CM and CG both worked on one trial, Macfadyen 2005; no third party input was enlisted.

6. ASSESSMENT OF METHODOLOGICAL QUALITY
The protocol stated that allocation concealment, and inclusion of randomised participants, would be assessed as inadequate if an allocation concealment approach was not reported, or it was not clear how many people were originally randomised into the trial, respectively. However, we have classed these as unclear.

We assessed the methodological quality of the trials using the following dimensions and criteria based on four methodological aspects.

a) Generation of allocation sequence

Adequate: if sequences are suitable to prevent selection bias and the method used is described.
Adequate methods include random numbers generated by computer, table of random numbers, drawing of lots or envelopes, tossing a coin, shuffling cards, throwing dice, or other methods of allocation that appear to be unbiased.

Unclear: stated but method not described.
The trial describes itself as being randomised but no further information is given.

Inadequate: if sequence could be related to prognosis.
Inadequate methods include case record number, date of birth, time, day, month or year of admission.

b) Allocation concealment

Adequate: if participants and investigators enrolling participants cannot foresee assignment. Adequate measures include a priori numbered or coded containers of identical appearance, central randomisation; sequentially numbered, opaque, sealed envelopes; or other descriptions that contained convincing elements of concealment.

Inadequate: trials in which the authors reported an approach that could not be considered adequate (e.g. methods of allocation such as alternation methods or use of case record numbers are not concealed).

Unclear: trials in which the authors either did not report an allocation concealment approach at all or allocation concealment was stated but method not described.

Baseline comparison of experimental groups will confirm whether treatment arm allocation appears to be unbiased.

c) Blinding

Double blind: the trial uses a placebo, or a double dummy technique such that neither the participant or care provider/assessor know which treatment is given.

Single blind: the participant or care provider/assessor is aware of the treatment given.

Open: all parties are aware of treatment.

d) Inclusion of all randomised participants

Adequate: More than 90% of people randomised in the trial were included in the analysis.

Inadequate: Less than 90% of those randomised in to the trial were included in the analysis.

Unclear: It is not clear how many people were originally randomised into the trial or analysed.

7. DATA ANALYSIS
The protocol stated that we would only include the results for CSOM participants where available, for studies that also enrolled people with otitis externa, draining surgical cavities or acute otitis media. However, we only did so if the authors reported stratifying randomisation by diagnosis. Where information was not reported regarding whether alternative diagnostic groups were stratified at randomisation, we included all participants (groups may be unbalanced, and decisions by trialists to report subgroups may have been linked to trend).