To the Editor: Chronic suppurative otitis media (CSOM) is a long-term problem which often has serious effects on hearing, speech and learning. Couzos and colleagues have shown that local treatment with ciprofloxacin (CIP) eardrops clears episodes of purulent discharge more efficiently than the commonly used framycetin, gramicidin and dexamethasone (FGD) eardrops.1 This hardly ranks as a “cure” as suggested by the authors.

The natural history of CSOM is known2-4 (see Box). The disease commonly starts in infancy with painless perforation of the ear drum and purulent discharge. The perforation is usually central and often large. It remains for several years. During this time there are episodes of painless discharge of foul-smelling pus associated with a blocked ear canal and poor hearing. Between these episodes, the perforation remains, but the ear is usually dry and hearing can be normal or at least adequate. By mid-childhood the perforation often closes spontaneously. In some children this leaves a scarred retracted eardrum, but, in others, fluid collects behind the now intact drum, and this chronic serous otitis media decreases hearing. Eventually, this chronic serous otitis media clears, leaving a scarred, retracted eardrum. Hearing then improves and is often functionally normal. Cholesteatoma and other medical complications are uncommon, but the loss of hearing during childhood has severe social and educational effects on the child.

Attempts to hasten closure of the perforation and limit the episodes of purulent discharge have had incomplete success.5,6 Any treatment that hastens recovery is welcome, but we should ask:

• Does local CIP treatment retain its effectiveness in repeated episodes of purulent discharge? (The most common organisms are Pseudomonas spp which rapidly develop resistance to antibiotics);

• Does CIP treatment alter the interval between purulent episodes compared with other treatments?; and

• Is there any evidence that local CIP treatment alters the natural history of the disease or lessens the hearing loss? (it is probably too early to detect this).

CSOM is a disease of poverty and overcrowding, but the mechanism leading from social disadvantage to ear disease is not clear. In Cherbourg Aboriginal Community where I work, social and living conditions have improved and CSOM is now much less common than reported by Stuart and co-workers more than 25 years ago.2 I hope that treatment with local CIP eardrops will maintain its promise as a significant improvement in the management of this disease in children who have already acquired it, while we work towards eliminating the disease in the long term.

To the Editor: Although the study by Couzos et al1 is important, I have serious concerns about (i) the lack of “intention-to-treat” analysis and (ii) the implication that the use of ototopical aminoglycosides is unethical and could lead to litigation.

The CONSORT statement indicates that analysis by intention-to-treat (ITT) is a key measure of methodological quality in reporting of randomised trials.2,3 Of 147 children randomly allocated to the two treatments in this trial, the primary outcome (resolution of otorrhoea) is only reported for 111 children (75.5%), and the secondary outcomes of healed tympanic membrane perforation in 64 (43.5%), and of improved hearing in 49 (33.3%). In addition, although the difference in cure rates is indeed 24.6% if drop-outs are ignored, the authors report incorrect 95% confidence intervals (15.8%–33.4%, instead of 7.3%–41.8%).

Analysis by intention-to-treat means that the 36 children lost to follow-up are assumed to still have ear discharge, so 33 of 75 in the ciprofloxacin group and 43 of 72 in the combined framycetin, gramicidin and dexamethasone group would not have “clinical cure”. This reduces the absolute difference to 15.7% (95% CI, 0–32%; P = 0.07), which is no longer a significant difference.

It is also unlikely that the clinical cure rates of 50%–70% in the study can be replicated in the real world of remote Top End Aboriginal communities with a high prevalence of chronic suppurative otitis media (CSOM). Cleaning ears twice a day with gentle syringing with 0.5% povidone-iodine, followed by ototopical ciprofloxacin for 10–14 days is not a feasible intervention among the competing priorities in most Aboriginal communities in the Northern Territory. The really important clinical outcomes for CSOM are the two secondary outcomes of healed perforations and improved hearing, but these outcomes were not significantly improved by ciprofloxacin compared with combined framycetin, gramicidin and dexamethasone.

Couzos and colleagues conclude that the use of aminoglycosides in CSOM is unethical and could lead to litigation on the basis of potential ototoxicity. Were this true, it would make this trial unethical. This misuse of ethics and law in the medical literature must be denounced. Nearly all drugs have potential side effects which have to be measured against their potential benefits. The potential ototoxicity of ototopical aminoglycosides (and other antibiotics) is not an absolute contraindication to their use, and has not been documented in any of the randomised controlled trials that have measured hearing before and after use of topical aminoglycosides.4

Couzos and colleagues misinterpret the strength of evidence for changing to ototopical ciprofloxacin. Although it is probably true that ciprofloxacin is superior to combined framycetin, gramicidin and dexamethasone in eradicating Pseudomonas aeruginosa, and for the short-term resolution of ear discharge, it is still unclear whether this transient benefit will be maintained in the long term, or whether resistant organisms will emerge in the middle ear, mitigating any benefit from the new drug. Haphazard use of ototopical ciprofloxacin in Top End Aboriginal communities is likely to result in resistance (indeed there is already anecdotal and unpublished evidence of this), so I believe that its use should currently be restricted to studies and programs whose outcomes are healed perforations and improved hearing (rather than only transient resolution of ear discharge).

In reply: Dugdale correctly refers to the natural history of chronic suppurative otitis media (CSOM), and the spectrum of endpoints. The primary endpoint chosen in our trial after ototopical treatment was a dry ear1 (commonly referred to as a “clinical cure” by other trials2), an outcome to be expected after short-term follow-up. Surgical closure of the tympanic membrane (TM) has also been defined as “curing” CSOM. Attaining a dry ear is essential to TM healing, and is therefore a functionally important outcome.

Brewster seems to confuse intention-to-treat (ITT) analysis with sensitivity analysis. Patients for whom there is no data after random allocation to study groups fall into the “missing” category, and may be included as “failures” in a sensitivity analysis. This may reveal additional information if missing data occurred differentially (ie, in some way associated with the treatments). However, in an ITT analysis, patients (with recorded data) are analysed in the group to which they were randomly allocated, irrespective of the actual treatment they received.3 Our analysis followed this principle and included all children, irrespective of the completeness of their treatment regimen.1 A sensitivity analysis revealed no new information, which was not surprising as the missing data were lost for (ascertained) reasons exclusively unrelated to the treatment, and double-blinding excluded any differential follow-up efforts by the healthcare workers. Consequently, the missing values occurred randomly and our analysis not only follows the ITT principle but is also unbiased (ie, valid with respect to “missing” patients).

We are surprised that our treatment regimen is not feasible in the Northern Territory given that it has been used for years in remote Aboriginal communities in Western Australia. We also confirmed the effectiveness of twice-daily use of ototopical medications, which is simpler than current four-times daily schedules.

In 1996, the World Health Organization recommended that topical aminoglycosides (AG) not be used for CSOM because of ototoxicity.4 Such use is also contraindicated by manufacturers.1 Given the availability of a safer alternative, healthcare professionals face potential medicolegal challenges if they choose ototopical AG to treat CSOM.

The relative superiority of ototopical fluoroquinolones (FQ) over AGs in effecting a dry ear is likely to persist with repeated treatments, as the risk of bacterial resistance generated in CSOM pathogens appears to be very small,5 and is far outweighed by the risks of resistance found with oral or parenteral FQs. In our trial, bacterial resistance to ciprofloxacin in ear isolates was not demonstrated in the short term. Systemic absorption of FQ through ototopical use is also negligible.5 In Japan, ototopical ofloxacin has been used as treatment for CSOM since 1992. Based on repeated nationwide surveys from 1995, increased FQ resistance attributed to ototopical use has not been seen in chronic otitis media isolates (Professor K Suzuki, Department of Otolaryngology, The Second Hospital, Fujita Health University School of Medicine, personal communication).6

Whether antibiotics can affect the interval between purulent episodes of CSOM is predicated on host and environmental factors, as well as the duration of effective drug therapy.

A multifaceted approach to the problem of CSOM, requiring the political will to improve the living conditions of Aboriginal families, access to appropriate primary healthcare, ototopical FQs, and surgery will see a reduction in the rate of this disabling disease.