Included studies

Four randomised controlled trials (RCTs) were included for this review (Mair 2016; Park 2016; Subtil 2019; Wang 2022).

The included studies are summarised in Table 2.

One study compared intraoperative ciprofloxacin ear drops alone, and intraoperative and postoperative ciprofloxacin ear drops to no topical ciprofloxacin (Wang 2022). Two studies compared intraoperative intratympanic ciprofloxacin injection to placebo/sham (Mair 2016, Park 2016). One study compared water precautions to no precautions (Subtil 2019).

One study excluded participants with craniofacial anomalies (Subtil 2019) and 3 studies did not report data on whether any participants had craniofacial anomalies (Mair 2016, Park 2016, Wang 2022). None of the studies reported data on whether any participants had mucociliary conditions.

All studies were conducted in children aged 17 years or under, and included children with a mean age in years of 2.75 (Mair 2016), 2.45 (Park 2016), and 4.4 (Subtil 2019) years. One study (Wang 2022) did not report the mean age, but the median participant age in years was 4.02, with a range of 0.72 to 13.1 years. None of the studies reported the number of previous episodes of otorrhoea or previous episodes of acute otitis media (AOM) participants.

See the literature search strategy in appendix B and study selection flow chart in appendix C.

Excluded studies

Studies not included in this review are listed, and reasons for their exclusion are provided in appendix J.

Intraoperative ciprofloxacin ear drops compared to no topical ciprofloxacin

Intraoperative ciprofloxacin ear drops had an important benefit in terms of reducing otorrhoea compared with no topical ciprofloxacin. In addition, there was a possible important benefit of intraoperative ciprofloxacin ear drops for reducing tube blockage (90% CI: 0.30 to 0.96).

Intraoperative plus postoperative ciprofloxacin ear drops compared to no topical ciprofloxacin

Intraoperative plus postoperative ciprofloxacin ear drops had an important benefit for reducing tube blockage compared with no topical ciprofloxacin; however, there was no important difference between interventions for otorrhoea.

Intraoperative intratympanic ciprofloxacin injection compared to placebo/sham

There was evidence that a 6mg intraoperative intratympanic ciprofloxacin injection had an important benefit in terms of reducing otorrhoea compared with placebo/sham, and very low quality evidence of a possible important benefit for a 12mg intraoperative intratympanic ciprofloxacin injection when compared with placebo/sham (90% CI: 0.09 to 0.96). However, there was no evidence of an important difference at a lower dosage of 4mg when compared to placebo, though this evidence was low quality (90% CI: 0.11 to 1.37). When evidence from intraoperative intratympanic ciprofloxacin injection at 4 mg, 6 mg, and 12 mg was pooled, intraoperative intratympanic ciprofloxacin injection overall had an important benefit for reducing otorrhoea when compared with placebo/sham (moderate quality). This was also the general pattern within the subgroups although differences often failed to reach clinical importance. One study (Mair 2016) included two groups of no active treatment (placebo and sham), and we arbitrarily assigned each of these groups as the comparison group against intraoperative intratympanic ciprofloxacin injection 4 mg and 12 mg groups. Sensitivity analyses where we swapped the comparison groups over showed that this did not affect the overall or subgroup results for otorrhoea. The pooled effect of intraoperative intratympanic ciprofloxacin injection showed no important difference for adverse effects of the intervention, and this was also the general pattern within the 6 mg and 12 mg subgroups whereas intraoperative intratympanic ciprofloxacin injection of 4 mg had an important benefit for reducing adverse effects of intervention compared with sham. However, adverse effects data from Mair 2016 included otorrhoea as an adverse event, and therefore instances of otorrhoea are double counted within the effect of ciprofloxacin injections at 4 mg and 12 mg for adverse effects outcomes. It is unclear how removing instances of otorrhoea from these data would affect adverse effects outcomes, because otorrhoea was not reported at 29 days follow-up. Although the pooled result of the sensitivity analyses also agreed with the main analysis for adverse events, the results within the subgroups differed in the sensitivity analyses relative to the original analyses, showing no important difference in the 4 mg group and important benefit in the 12 mg group. See appendix L for further details of the sensitivity analysis, included risk ratios and absolute effects.

There was no important difference for tube blockage and tube extrusion.

Water precautions compared to no water precautions

A comparison between water precautions and no precautions showed no important difference and no evidence of important difference (90% CI 1.00 to 2.06) for quality of life and otorrhoea, respectively. The outcomes of acceptability and ‘surgical intervention to remove ventilation tubes’ were not reported by any studies.

See appendix F for full GRADE tables.

Included studies

A systematic review of the economic literature was conducted but no economic studies were identified which were applicable to this review question.

Economic model

No economic modelling was undertaken for this review because the committee agreed that other topics were higher priorities for economic evaluation given the low cost of the interventions under consideration in this review.

The outcomes that matter most

Otorrhoea is a common complication after grommet surgery, which may both recur and lead to poor quality of life in children with otitis media with effusion. Therefore, otorrhoea was prioritised as a critical outcome. To reduce the risk of otorrhoea following grommet surgery, different types of preventive measures/interventions can be used during (for example, antibiotic ear drops with or without corticosteroids, systemic antibiotics, and systemic corticosteroids) and after surgery (for example, antibiotic ear drops with or without corticosteroids, water precautions and oral antibiotics). However, these preventive measures may have adverse effects in children (for example, local discomfort, rash, headache, gastrointestinal discomfort, and antimicrobial resistance), and the overall ability and willingness of children to use effective preventive measures is important. Therefore, adverse effects and acceptability of preventive measures and interventions were prioritised as critical outcomes.

Tube blockage, tube extrusion and requirement of surgical intervention to remove grommets were selected as important outcomes as they are relatively common in children who undergo grommet placement surgery, and tube blockage and tube extrusion may be related to recurrence of otitis media with effusion. In addition, quality of life was selected as an important outcome as this is a global measure that takes into account both beneficial and adverse effects of the interventions.

The quality of the evidence

The quality of the evidence was assessed using GRADE methodology. The evidence for otorrhoea and adverse effects of the interventions was moderate to very low quality due to risk of bias arising from the intervention allocation, missing outcome data and measurement of outcome, seriously imprecise findings, and the inclusion of indirect populations and interventions. The evidence for tube blockage was low to very low quality because of risk of bias from the intervention allocation, seriously imprecise findings, and the inclusion of indirect populations and interventions. The evidence for tube extrusion and quality of life was low to very low quality due to risk of bias arising from the intervention allocation, missing outcome data and measurement of outcome, seriously imprecise findings, and the inclusion of indirect outcomes.

No evidence was found that reported on the outcomes of acceptability or surgical intervention to remove grommets.

Benefits and harms

There was moderate quality evidence that intraoperative ciprofloxacin drops had an important benefit in terms of presence of otorrhoea at 6 weeks follow-up compared to no drops, and that intraoperative intratympanic ciprofloxacin injection had an important benefit in terms of presence of otorrhoea at 15-29 days follow-up compared to placebo or sham. There was also low quality evidence that intraoperative ciprofloxacin drops had a possible important benefit in terms of tube blockage at 6 weeks follow-up compared to no drops, but there was uncertainty around the estimate and very low quality evidence from another trial showed no evidence of an important difference in terms of tube blockage for an intraoperative intratympanic ciprofloxacin injection compared to sham at 29 days follow-up. The committee agreed there were further inconsistencies in the evidence which raised concerns, such as the fact that Wang 2022 showed no evidence of an important difference in terms of presence of otorrhoea at 6 weeks follow-up when intraoperative and postoperative ciprofloxacin drops were used. The committee agreed it was unusual that the evidence showed a worse result with the addition of postoperative drops than the application of intraoperative drops alone, and there was uncertainty in the importance of the outcome. Additionally, the committee discussed the indirectness of the population in Wang 2022 and agreed the fact that 34% of the population had acute otitis media (AOM) and not OME was of serious concern because antibiotics would be expected to be more effective for AOM as it is an acute infection, while OME is sterile.

Having considered the reliability and strength of the evidence, the committee agreed it should be interpreted with caution and supplemented with their own expertise. They discussed their knowledge of current practice and agreed there is variation regarding the application of intraoperative ear drops; they tend to only be applied intraoperatively if the ear looks infected when the grommets are put in, or for anticoagulative properties if there is bleeding after insertion of the grommets, to prevent blood clots in the lumen. The committee also discussed whether otorrhoea was a significant issue for people with OME, and whether the resolution of otorrhoea justified the potential for increasing antibiotic resistance. Lay members of the committee observed that some clinicians’ experience with incidences of otorrhoea might not reflect the real prevalence of otorrhoea following grommet insertion, because they would usually see patients at 6 weeks follow-up, by which point the otorrhoea might have resolved. They noted that otorrhoea was very painful for the patient and could cause difficulties for families, and therefore attempts to prevent otorrhoea should be made. The committee also agreed that a single dose of ciprofloxacin drops is unlikely to influence antibiotic resistance, and therefore the risk is minimal while the benefits could include prevention of otorrhoea and tube blockage, and increased quality of life in children with OME and their families.

In Mair 2016 and Park 2016, the intraoperative ciprofloxacin was applied using intratympanic injection, which is not standard practice. The committee discussed their expectation that injecting the ciprofloxacin would allow it to be applied more precisely and penetrate the middle ear better but agreed that applying ear drops using the standard method had a lower risk of complications such as dislodging the grommets. Additionally, although there was no important difference between intraoperative intratympanic ciprofloxacin injection and placebo/sham in terms of adverse effects, the committee discussed the fact that this outcome double-counted otorrhoea as an adverse event, and therefore the potential harms of intraoperative intratympanic ciprofloxacin injection when otorrhoea is not included were not known. As a result, the committee recommended that a single application of ciprofloxacin ear drops intraoperatively should be considered to prevent otorrhoea and tube blockage based on the available evidence and their knowledge and experience. However, a stronger recommendation could not be made due to the lack of high quality, reliable evidence. The committee also discussed dosage of ciprofloxacin drops. The committee felt that ciprofloxacin 0.2% ear drops in a 0.25 ml single ampoule dose may be the most appropriate option. However, they were aware that in some areas, ciprofloxacin 0.3% eye drops may also be used in the ear as a single dose of 3-5 drops (equivalent dose as per above). As there was variation in practice regarding the dosage, the committee felt that they could not add details on dosage to the recommendation. There was no evidence of an important difference for water precautions such as earplugs and headbands after grommet insertion in terms of presence of otorrhoea at 6 months follow-up compared to no water precautions, although this evidence was of very low quality and there was uncertainty in the importance of the outcome. The study did not analyse the effectiveness of different precautions, so it was unclear whether some interventions were more effective than others (such as earplugs versus headbands). In addition to the low quality of the evidence, the committee agreed it was difficult to ascertain from the information presented in the study whether and to what extent people in the control group might have avoided water independent of their use of ear plugs or headbands, and therefore whether this would have affected results. Considering the uncertainty of the evidence and the committee’s agreement that these interventions also had the potential to dissuade children from swimming at all, no recommendations were made about the use of earplugs and headbands while swimming or bathing. The committee discussed what should happen if a patient has a history of otorrhoea or if otorrhoea repeatedly occurs after surgery and agreed at this point the patient would be considered to be susceptible to ear infections and should be treated according to the recommendations in Evidence Review L.

The committee agreed that in practice, more involved interventions such as those investigated in Subtil 2019 do not tend to be recommended after grommet insertion. Instead, standard practice is to advise patients to keep the ear dry (that is, avoid swimming and take care when bathing and washing hair) to prevent the risk of water permeating the lumen while the wound is still healing around the grommet, although there is currently variation regarding the number of weeks this is advised for. The committee agreed to make a strong recommendation despite the lack of high quality evidence because water avoidance is sensible to prevent infection at the site of the tube but agreed that 6 to 8 weeks of water avoidance until first follow-up after surgery was excessive as the risk would lower once the wound was healed, which would usually happen long before follow-up. The committee agreed that 2 weeks would usually be enough time for the wound to heal. They agreed that, considering the lack of evidence about this, a shorter timeframe would be more practical as it would have a lower risk of impacting both the child’s development (for example, the committee were concerned that water precautions could interfere with children learning to swim) and the quality of life of children and their families.

The committee also considered whether further water precautions after the initial 2 weeks post-surgery should be recommended. They discussed the fact that soapy water (due to soap reducing the surface tension of the water) and diving both increase the potential for permeation of water into the lumen, – and therefore should be avoided for the whole duration that the grommet is in place. However, the available evidence did not provide information regarding these considerations. The committee agreed, based on their experience, that the risk of permeation in these situations after the wound has healed is still very low, and, therefore, that further recommendations were unnecessary.

There was low quality evidence that intraoperative and postoperative ciprofloxacin drops had an important benefit in terms of tube blockage at 6 weeks follow-up compared to no drops, however the committee agreed the evidence for this outcome was unreliable considering intraoperative application only had better outcomes than intraoperative and postoperative application in terms of presence of otorrhoea. The committee agreed that repeat applications of antibiotics would increase the risk for antibiotic resistance and therefore the evidence of effectiveness was not strong enough to recommend postoperative use of ciprofloxacin drops. There was also no evidence available regarding the use of oral antibiotics postoperatively, or for antiseptic or saline washouts, systemic antibiotics, or systemic corticosteroids intraoperatively, and no recommendations were made about these.

There was limited evidence on water precautions for preventing otorrhoea following grommet insertion for children with OME and no evidence on the effectiveness of water avoidance, or on the comparative effectiveness of different water precautions. The committee agreed it would be useful to know which water precautions were the most effective, including the optimal length of time to use water precautions in order to prevent otorrhoea. This would enable the least restrictive recommendations to be made in the future, to promote children’s development while also ensuring they are not at risk of developing otorrhoea. Therefore, the committee made a research recommendation.

Cost effectiveness and resource use

This review question was not prioritised for economic analysis and therefore the committee made a qualitative assessment of the likely cost-effectiveness of their recommendations. The committee noted that advice about water precautions would be given in conjunction with other patient information and would have negligible costs. Therefore, the committee concluded that such advice would be cost-effective in minimising the risk of water permeating into the lumen whilst the wound following grommet insertion is still healing. The committee also considered that such advice would be in line with current practice.

The committee considered that the clinical evidence and their own expertise and experience provided some evidence for the likely cost-effectiveness of a single application of intraoperative non-ototoxic antibiotic-containing topical ear drops (such as ciprofloxacin) during grommet insertion to prevent otorrhoea and tube blockage. They reasoned that only a small beneficial effect of treatment would be required for it to be considered cost-effective given the low cost of ear drops. The committee thought it unlikely that a single dose of ciprofloxacin would promote antibiotic resistance and concluded that any risks of intervention were likely to be outweighed by the benefits in the prevention of otorrhoea, tube blockage and improved health related quality of life. Whilst the committee believed that current practice is varied, they recognised that their recommendations could represent a change in practice for some units. However, given the low cost of ear drops and the potential for some savings from reduced rates of otorrhoea the committee did not believe that their recommendation would represent a significant resource impact to the NHS.

Otorrhoea outcomes

Swapping the placebo and sham groups in the comparison did not affect the overall or subgroup results for otorrhoea: the pooled effect of intraoperative intratympanic ciprofloxacin injection showed no important difference for adverse effects of the intervention, and this was also the general pattern within the 6 mg and 12 mg subgroups whereas intraoperative intratympanic ciprofloxacin injection of 4 mg had an important benefit for reducing adverse effects of intervention compared with sham. The risk ratios and absolute risk within the sensitivity analysis for the outcomes were as follows:

• Sensitivity analysis swapping the sham no intervention comparison group with the placebo group for the 4 mg comparison and the placebo group with the sham group for the 12 mg comparison showed the following results: RR 0.42 (0.29 to 0.62); Absolute risk: 145 fewer per 1000 (from 95 fewer to 178 fewer)
• Sensitivity analysis swapping the sham no intervention comparison group with the placebo group showed the following results: RR 0.26 (0.06 to 1.09); Absolute risk: 269 fewer per 1000 (from 342 fewer to 33 more)
• Sensitivity analysis swapping the placebo group with the sham no intervention comparison group showed the following results: RR 0.42 (0.09 to 1.92); Absolute risk: 145 fewer per 1000 (from 227 fewer to 230 more)

Adverse event outcomes

Although the pooled result of the sensitivity analyses also agreed with the main analysis for adverse events, the results within the subgroups differed in the sensitivity analyses relative to the original analyses, showing no important difference in the 4 mg group and important benefit in the 12 mg group.

• Sensitivity analysis swapping the sham no intervention comparison group with the placebo group for the 4 mg comparison and the placebo group with the sham group for the 12 mg comparison showed the following results: RR 0.85 (0.59 to 1.25); Absolute risk: 87 fewer per 1000 (from 238 fewer to 145 more)
• Sensitivity analysis swapping the sham no intervention comparison group with the placebo group showed the following results: RR 1.13 (0.68 to 1.89); Absolute risk: 71 more per 1000 (from 175 fewer to 485 more)
• Sensitivity analysis swapping the placebo group with the sham no intervention comparison group showed the following results: RR 0.53 (0.32 to 0.86); Absolute risk: 423 fewer per 1000 (from 126 fewer to 612 fewer)