Autoinflation for treatment of glue ear

Editor—A review of nasal balloons is welcome, given the high prevalence of glue ear (otitis media with effusion) and the devices’ simplicity.¹ The authors conclude from three studies that the balloon may be of benefit. The two published studies to which they refer merit closer attention.²,³ The studies suggest short term benefit, but it is not sustained. No significant differences could be found in either study between treated and control groups in tympanometric resolution at two or three months’ follow up. Compliance was poor in over half the children and got worse with time. Any benefit is lost after treatment is stopped.

Reidpath et al have included the two-week outcomes in one trial² and the 12-week outcomes in the two others (Blanshard et al³ and the unpublished data), even though benefit decreases significantly with time, biasing the results in favour of the treatment. The greater prevalence of type B tympanometry in one untreated group³ is an additional source of bias, as is the lack of blinding and adequate randomisation in any study. The result is the conclusion that the treatment is effective, with an odds ratio of 3.5. Meta-analysis of poor research gives poor answers.

Reducing all the information in a paper to a single odds ratio can be a powerful tool but also hides much of the available information. Meta-analysis is not the best way to analyse three small studies of variable methodology. The conclusion is the inevitable refrain that “more/bigger/better studies are needed.” In his commentary on the paper Haggard mentions the need to prioritise research endeavour to maximise payback.¹ Surely it is wrong to recommend that large, expensive, time consuming studies be done on an issue for which the answer is already available. The nasal balloon produces no sustained benefit in glue ear, and after three months’ watchful waiting no large difference can be expected in the number of children requiring surgery. This is the real issue.

Haggard mentions that nasal balloons may have a place as a temporising measure. Is he suggesting that although the device produces no worthwhile benefit, we can use it as a placebo while nature takes its course? This would be deception of the worst kind, since the balloon is not available on prescription in the United Kingdom and must be paid for by parents. Glue ear is associated with social deprivation, and we must not ask our poorest patients to pay for a worthless device simply to keep them amused.