Interventions for the prevention of postoperative ear discharge after insertion of ventilation tubes (grommets) in children

Abstract

Background

Grommets are frequently inserted in children's ears for acute otitis media and otitis media with effusion. A common complication is postoperative ear discharge (otorrhoea). A wide range of treatments are used to prevent the discharge, but there is no consensus on whether or not intervention is necessary nor which is the most effective intervention.

Objectives

To assess the effectiveness of prophylactic interventions, both topical and systemic, in reducing the incidence of otorrhoea following the surgical insertion of grommets in children.

Search methods

We searched the Cochrane Ear, Nose and Throat Disorders Group Trials Register; the Cochrane Central Register of Controlled Trials (CENTRAL); PubMed; EMBASE; CINAHL; Web of Science; BIOSIS Previews; Cambridge Scientific Abstracts; ICTRP and additional sources for published and unpublished trials. The date of the search was 3 July 2012.

Selection criteria

We included randomised controlled trials (RCTs) that compared the efficacy of prophylactic interventions against placebo/control and/or with other prophylactic interventions for postoperative otorrhoea in children.

Data collection and analysis

Two review authors independently assessed study eligibility and risk of bias, and extracted data. The outcome data were dichotomous for all the included trials. We calculated individual and pooled risk ratios (RR) using the Mantel‐Haenszel fixed‐effect method. We also calculated the numbers needed to treat to benefit (NNTB).

Main results

We found 15 eligible RCTs (2476 children, aged from four months to 17 years). We graded seven RCTs as being at a low risk of bias (n = 926 children) and for an eighth RCT we also graded two of the arms as being at a low risk of bias. We graded the other seven trials as being at a high risk of bias.

For a single application at surgery, there was evidence from two low risk of bias trials that at two weeks postoperatively the risk of otorrhoea was reduced by multiple saline washouts (from 30% to 16%; RR 0.52, 95% confidence interval (CI) 0.27 to 1.00; NNTB 7; one RCT; 140 children) and antibiotic/steroid ear drops (from 9% to 1%; RR 0.13, 95% CI 0.03 to 0.57; NNTB 13; one RCT; 322 ears). A meta‐analysis of two low risk of bias trials (222 ears) failed to find an effect of a single application of antibiotic/steroid ear drops at four to six weeks postoperatively.

For a prolonged application of an intervention, there was evidence from four low risk of bias trials that the risk of otorrhoea was reduced two weeks postoperatively by antibiotic ear drops (from 15% to 8%; RR 0.54, 95% CI 0.30 to 0.97; NNTB 15; one RCT; 372 children), antibiotic/steroid ear drops (from 39% to 5%; RR 0.13, 95% CI 0.05 to 0.31; NNTB 3; one RCT; 200 children), aminoglycoside/steroid ear drops (from 15% to 5%; RR 0.37, 95% CI 0.18 to 0.74; NNTB 11; one RCT; 356 children) or oral antibacterial agents/steroids (from 39% to 5%; RR 0.13, 95% CI 0.03 to 0.51; NNTB 3; one RCT; 77 children).

Only one trial assessed the secondary outcome of ototoxicity, but no effect was found. There were no trials that assessed quality of life.

Authors' conclusions

Our review found that each of the following were effective at reducing the rate of otorrhoea up to two weeks following surgery: (1) multiple saline washouts at surgery, (2) a single application of topical antibiotic/steroid drops at surgery, (3) a prolonged application of topical drops (namely antibiotic ear drops, antibiotic/steroid eardrops or aminoglycoside/steroid ear drops) and (4) a prolonged application of oral antibacterial agents/steroids. However, the rate of otorrhoea between RCTs varied greatly and the higher the rates of otorrhoea within a RCT, the smaller the NNTB for therapy.

We conclude that if a surgeon has a high rate of postoperative otorrhoea in children then either saline irrigation or antibiotic ear drops at the time of surgery would significantly reduce that rate. If topical drops are chosen, it is suggested that to reduce the cost and potential for ototoxic damage this be a single application at the time of surgery and not prolonged thereafter.

Plain language summary

Interventions for the prevention of postoperative ear discharge after insertion of ventilation tubes (grommets) in children

The insertion of a grommet is a common surgical procedure performed in children. Ear discharge (otorrhoea) is often complained of after the procedure. A wide range of treatments are used to prevent the discharge, but the evidence for their effectiveness is inconclusive: indeed, it is not known whether treatment is even necessary.

This review found 15 randomised controlled trials which assessed the effectiveness of the interventions used to prevent ear discharge after surgery. The findings in seven low risk of bias trials (and part of an eighth) showed a reduction in the rate of ear discharge up to two weeks after the operation using either saline washouts or a single application of antibiotics at the time of the operation. A prolonged application of antibiotics was also effective. The effects were largest in studies with a high rate of ear discharge.

We conclude that the use of an intervention to prevent postoperative ear discharge should be restricted to those at a high risk of this discharge, but the choice of which treatment may be left to the surgeon.

Background

Description of the condition

The two main indications for the insertion of grommets (also known as ventilation tubes or tympanostomy tubes) are the restoration of hearing in children with otitis media with effusion (glue ear) and the prevention of recurrent acute otitis media. The operation is one of the most common paediatric surgical procedures: each year in the UK over 30,000 patients have ventilation tubes inserted (Sood 2007). However, the procedure can often lead to sequelae, of which postoperative ear discharge (otorrhoea) is common. The prevalence of otorrhoea ranges from 3% to 74% (Baker 1988; Gates 1986; Giles 2007; Kay 2001; Per‐Lee 1981; Scott 1992). The otorrhoea can be accompanied by foul odour, otalgia and fever. It can also block the grommet, making it ineffective in ventilating the middle ear or in improving the hearing. The cause of the otorrhoea remains uncertain: it may indicate the presence of persistent or recurrent middle ear infection, or it may be a response to the grommet as a foreign body (Hubbard 1985). 

Why it is important to do this review

There have been two previous reviews assessing interventions for the reduction of the incidence of post‐tympanostomy otorrhoea (Garcia 1994; Hochman 2006). Garcia et al reviewed five randomised trials of topical antibiotics and topical antibiotic‐steroid combinations in children that were published up to 1992. They found that four of the five reported a reduction in the incidence of otorrhoea, but in only one was the reduction statistically significant. Three trials were sufficiently similar to be meta‐analysed, giving an odds ratio of 0.12 (95% confidence interval (CI) 0.04 to 0.37). Garcia et al recommended the use of the interventions, but due to a potential risk of ototoxicity they suggested it should be limited to cases at a higher risk of otorrhoea. The issue of ototoxicity is important, as transtympanic aminoglycosides are known to be ototoxic in humans as evidenced by their use for chemical labyrinthectomy (Pullens 2011), although the main supporting evidence for ototoxicity from ear drops is from case reports of patients with chronic otitis media where the tympanic membrane is perforated (Phillips 2007). Hochman 2006 analysed nine trials of both children and adults, published up to 2004. All nine trials reported an odds ratio that was less than 1, though the reduction in postoperative otorrhoea was statistically significant in only three trials. The overall odds ratio was 0.52 (95% CI 0.39 to 0.69).

The Scottish Intercollegiate Guidelines Network (SIGN 2008) included otorrhoea from grommet insertion as part of a wider review of antibiotic prophylaxis in surgery. A single dose of topical antibiotic was recommended as "prophylaxis reduces short term morbidity, reduces hospital costs and may decrease overall consumption of antibiotics" (p. 13). They reported the odds ratio for otorrhoea as 0.46 and the number needed to treat to benefit as 13. The conclusions were based on three RCTs (Kocaturk 2005; Nawasreh 2004; Zipfel 1999), though the choice of which out of those potentially available was not stated.

None of these previous reviews assessed the quality of the trials included. An up‐to‐date review of the literature focusing on a wide range of interventions and an in‐depth quality assessment was warranted.

There have been many Cochrane reviews looking at grommets or otitis media ‐ the most recent is van Zon 2012. One of particular interest is Vaile 2006, which reviewed non‐surgical management of discharge from ears with grommets in place, although this looked at the management of post‐ventilation tube otorrhoea rather than prophylaxis.

Objectives

To assess the effectiveness of prophylactic interventions, both topical and systemic, in reducing the incidence of otorrhoea following the surgical insertion of grommets in children.

Methods

Criteria for considering studies for this review

Types of studies

Any randomised controlled trials (RCTs) that assessed the prevention of postoperative otorrhoea by prophylactic interventions initiated at the time of insertion of grommets for otitis media. The grommets could be inserted either unilaterally or bilaterally; the otitis media could be either otitis media with effusion or recurrent acute otitis media.

Types of participants

Children between the ages of three months and 17 years who had a grommet inserted for otitis media with effusion and/or recurrent acute otitis media.

Types of interventions

Any trials of antibiotics or steroids (namely systemic antibiotics, topical antibiotics, topical steroids and topical antibiotic‐steroid combinations), or any trials of other methods (e.g. irrigation and decongestant instillation). The experimental designs had to compare treatments versus either placebo, a no treatment group or other prophylactic treatments.

Types of outcome measures

Primary outcomes

• Otoscopic observation of postoperative otorrhoea from one or both ears

Secondary outcomes

• Any parent or patient reports of ear discharge

• Bacterial culture, both pre‐ and post‐insertion

• Tympanometry assessment for patency of the grommet

• Any reports of adverse effects (e.g. toxicity of drops, hearing loss and balance, irritant or allergic reaction)

• Increased burden on family or dependents

• Any loss of time due to repeat consultations

• Cost‐effectiveness

We also looked for predictor factors for otorrhoea and potential benefit.

Search methods for identification of studies

We conducted systematic searches for randomised controlled trials. There were no language, publication year or publication status restrictions. The date of the search was 3 July 2012.

Electronic searches

We searched the following databases from their inception for published, unpublished and ongoing trials: the Cochrane Ear, Nose and Throat Disorders Group Trials Register; the Cochrane Central Register of Controlled Trials (CENTRAL, The Cochrane Library 2012, Issue 6); PubMed; EMBASE; CINAHL; AMED; LILACS; KoreaMed; IndMed; PakMediNet; CAB Abstracts; Web of Science; BIOSIS Previews; CNKI; ISRCTN; ClinicalTrials.gov; ICTRP; Google Scholar and Google.

We modelled subject strategies for databases on the search strategy designed for CENTRAL. Where appropriate, we combined subject strategies with adaptations of the highly sensitive search strategy designed by The Cochrane Collaboration for identifying randomised controlled trials and controlled clinical trials (as described in the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0, Box 6.4.b. Handbook 2011). Search strategies for the major databases including CENTRAL are provided in Appendix 1.

Searching other resources

We scanned the reference lists of identified publications for additional trials and contacted trial authors where necessary. In addition, we searched PubMed, TRIPdatabase, The Cochrane Library and Google to retrieve existing systematic reviews relevant to this systematic review, so that we could scan their reference lists for additional trials. We searched for conference abstracts using the Cochrane Ear, Nose and Throat Disorders Group Trials Register.

Data collection and analysis

Selection of studies

Two authors (IS and SLS) independently examined the titles and abstracts found by the searches, and then retrieved the full text of those found relevant. Any differences in opinion over the inclusion of the studies were resolved by discussion with the other authors (GGB or MAA).

Data extraction and management

Two authors (IS and SLS) independently extracted the data from each trial using a standardised data extraction form. If any essential data were missing from a study then at least one author (IS or SLS) contacted the authors to request the additional information, with reference to the CONSORT guidelines and checklist (see Altman 2001; Moher 2001).

We documented for each trial the following information:

• Methods: randomisation method; unit of randomisation (by child or by ear); blinding method; assessment method.

• Participants: number of children entered and analysed; age and sex; inclusion and exclusion criteria.

• Type of interventions: details of the intervention used; the start of application; duration of treatment.

• Outcomes: number of participants with otorrhoea in the treatment and control groups; time of postoperative assessment; whether otorrhoea was observed in one or both ears; any reports of secondary outcomes.

Assessment of risk of bias in included studies

We assessed the risk of bias of the trials using the methods outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Handbook 2011). This was performed by two authors (IS and SLS), with any disagreements resolved by discussion with the other authors (GGB and MAA). We assessed the risk of bias using six quality domains: sequence generation, allocation concealment, blinding of participants, blinding of assessors, completeness of outcome data and any other sources of bias. We used the 'Risk of bias' tool in RevMan 5 (RevMan 2012) to assess the risk of bias in the included trials and to assign a judgement to each.

Data synthesis

We performed the data analysis on an intention‐to‐treat basis. We analysed the data as risk ratios (RR) giving the risk of developing otorrhoea for each comparison: a RR of less than 1 indicates that there was an improvement from treatment, a RR of 1 indicates no change and a RR greater than 1 indicates a worsening. We used a fixed‐effect model using the Mantel‐Haenszel method, as it was a simple, two‐level model of otorrhoea present versus otorrhoea not present. We also calculated the number needed to treat to benefit (NNTB).

We were able to conduct six meta‐analyses. When combining the data for meta‐analysis we checked for any heterogeneity using the Chi² test, setting a significance level of 0.1. We also used the I² statistic to estimate the total variation across studies. Where relevant we conducted separate subgroup analyses for those trials considered to be at high or low risk of bias. We did not find enough trials to be able to conduct subgroup analyses for age, sex, indication of ventilation tube insertion or type of effusion at time of surgery.

Results

Description of studies

Results of the search

The searches were carried out in April 2010, November 2010, August 2011 and July 2012. The search history is shown in Figure 1.

1.

Study flow diagram.

Included studies

We found 15 trials to be eligible for this review. See Characteristics of included studies for details.

Method of randomisation

Seven trials were randomised by ear (Arya 2004; Nawasreh 2004; Salam 1992; Shinkwin 1996; Welling 1995; Zacharia 1993; Zipfel 1999). Eight trials were randomised by child (Cannon 1997; Daly 1995; Epstein 1992; Giles 2007; Kocaturk 2005; Morpeth 2001; Poetker 2006; Scott 1992).

Participants

All trials recruited children who were prospectively undergoing ventilation tube insertion for otitis media with effusion or for recurrent acute otitis media. Five trials also included children undergoing adenoidectomy and tonsillectomy in their trials (Daly 1995; Nawasreh 2004; Scott 1992; Zacharia 1993; Zipfel 1999). One trial stated that they had included patients with co‐morbid factors such as Down's syndrome or cleft palate (Arya 2004).

The age of the children tested ranged from four months to 17 years. Six trials reported the mean age of their participants (between 2.1 and 7.6 years), three trials reported the median age (between one and five years), and the remaining six trials only reported the range of ages.

Across the studies a wide range of exclusion criteria were used, including cranio‐facial abnormalities, sensorineural hearing loss, general medical conditions, previous surgery (e.g. adenoidectomy, tonsillectomy, middle ear), concurrent oral or topical medication, known allergies to antibiotics, and unreliable attendance or care givers. 

Setting and location

All trials were located in private or public hospitals. The locations were as follows:

• USA (Cannon 1997; Daly 1995; Epstein 1992; Giles 2007; Morpeth 2001; Poetker 2006; Scott 1992; Welling 1995; Zipfel 1999);

• UK (Arya 2004; Shinkwin 1996; Salam 1992; Zacharia 1993);

• Turkey (Kocaturk 2005);

• Jordan (Nawasreh 2004).

Interventions

The interventions used in the trials were:

Irrigation

• Isotonic saline irrigation (Kocaturk 2005)

• Povidone‐iodine topical antiseptic (Betadine)/saline irrigation (Cannon 1997)

Irrigation/antiseptic cleaning of the ear canal combined with topical antibiotics

• Povidone‐iodine topical antiseptic/saline irrigation and gentamicin drops (Cannon 1997)

Topical antibiotics

• Gentamicin drops (Cannon 1997; Scott 1992)

• Ciprofloxacin drops (Morpeth 2001; Nawasreh 2004; Zipfel 1999)

• Ofloxacin drops (Kocaturk 2005; Poetker 2006)

Oral antibiotics

• Sulbactam/ampicillin combination (Kocaturk 2005)

Oral antibacterial agents/steroids

• Oral prednisone followed by oral trimethoprim/sulfamethoxazole suspension (Daly 1995)

Topical antibiotic/steroid combinations

• Betamethasone/neomycin drops (Salam 1992)

• Ciprofloxacin/dexamethasone drops (Giles 2007)

• Dexamethasone/framycetin/gentamicin, instilled (Arya 2004)

• Gentamicin/hydrocortisone, instilled (Shinkwin 1996; Zacharia 1993)

• Neomycin/hydrocortisone drops (Poetker 2006)

• Neomycin/polymyxin‐B/hydrocortisone, instilled (Welling 1995)

• Neomycin/polymyxin‐B/hydrocortisone (Morpeth 2001)

• Sulfacetamide/prednisolone acetate (Epstein 1992)

Application of intervention

The interventions were administered either topically to the ear or systematically (orally) as:

• a single application at the time of surgery;

• a prolonged application for a period after surgery;

• a combination of both.

A range of experimental designs was used:

• one intervention versus placebo (Daly 1995);

• one intervention versus no treatment (Arya 2004; Epstein 1992; Giles 2007; Salam 1992; Shinkwin 1996; Welling 1995; Zacharia 1993; Zipfel 1999);

• two or more interventions versus each other but without a control group (Morpeth 2001);

• two or more interventions versus each other and versus no treatment (Cannon 1997; Kocaturk 2005; Poetker 2006);

• a single application of an intervention versus a prolonged application of the same intervention and versus no treatment (Nawasreh 2004; Scott 1992).

Outcomes

The primary outcome was postoperative otorrhoea from one or both ears. Two trials reported the incidence of purulent otorrhoea (Cannon 1997; Kocaturk 2005) whereas the other trials reported the incidence of otorrhoea without further specification. Note that in Scott 1992 four patients who were found to have purulent otorrhoea or purulent adenitis intra‐operatively were withdrawn. In the 'by child' studies the outcomes were reported either from one or both ears (Daly 1995; Giles 2007; Kocaturk 2005), from the left ear only (Epstein 1992), by the number of ventilation tubes (Poetker 2006) or it was not specified (Cannon 1997; Scott 1992). 

The secondary outcome of any adverse affect on hearing was reported by Giles 2007 and the secondary outcome of cost‐effectiveness was reported by Kocaturk 2005. No other trials reported secondary outcomes.

Excluded studies

We excluded 17 studies after reading their full texts ‐ see Characteristics of excluded studies for the reasons for their exclusion.

Risk of bias in included studies

The 'Risk of bias' assessment for each individual trial is reported in the Characteristics of included studies table. The risks are illustrated in Figure 2 and Figure 3.

2.

Individual ratings of the 15 trials for each category of risk of bias.

3.

Distribution of classifications for each risk of bias category. All 15 trials are included.

The grading of quality was primarily based on whether the assessor was blind when performing otoscopy. We graded seven trials as being at a low risk of bias (Arya 2004; Daly 1995; Giles 2007; Morpeth 2001; Poetker 2006; Shinkwin 1996; Welling 1995). In one four‐armed trial, we graded two of the arms as being at a low risk of bias because the observer was blind to the interoperative management but not to the postoperative therapy (Kocaturk 2005). We graded the other two arms of this trial as being at a high risk of bias. We also graded all the remaining trials as being at a high risk of bias (Cannon 1997; Epstein 1992; Nawasreh 2004; Salam 1992; Scott 1992; Zacharia 1993; Zipfel 1999). 

We graded the generation of the allocation sequence as adequate in Arya 2004, Daly 1995, Giles 2007, Poetker 2006, Shinkwin 1996, Welling 1995 and Zipfel 1999. In two trials we graded the allocation sequence as being inadequate, as the patients were randomised by a deck of cards (Salam 1992) or by alternation (Cannon 1997). The allocation method was unclear in the remaining trials, usually because the only description was "patients were randomised" or similar, and attempts to contact the authors for more information were unsuccessful.

We graded the blinding of the assessor as adequate in Arya 2004, Daly 1995, Giles 2007, Kocaturk 2005, Morpeth 2001, Poetker 2006, Shinkwin 1996 and Welling 1995. In the remainder the method was either unclear (Epstein 1992; Nawasreh 2004; Salam 1992; Scott 1992; Zacharia 1993) or the assessor was not blinded to the treatment group (Cannon 1997; Zipfel 1999).

We graded the blinding of the participant to the intervention they received as adequate in Arya 2004, Daly 1995, Morpeth 2001, Shinkwin 1996, Welling 1995 and Zipfel 1999. In one trial the method was unclear (Zacharia 1993). In the remainder the children were not blinded because they knew that one ear was having treatment and the other was not (Cannon 1997; Epstein 1992; Giles 2007; Kocaturk 2005; Nawasreh 2004; Poetker 2006; Salam 1992; Scott 1992). 

In 13 trials the outcome data were either complete or adequately addressed, but in two incomplete data were not adequately addressed (Daly 1995; Zacharia 1993).

Many trials had specific exclusion criteria which may have been an additional source of bias (see above). We regarded none of these as being substantial enough to affect the overall grades.

Effects of interventions

For assessments made at up to two weeks, the overall prevalence of otorrhoea was 25% in the no treatment/placebo groups of the low risk of bias trials (79 cases out of 309; Daly 1995; Giles 2007; Poetker 2006). It was 13% in the high risk of bias trials (117 cases out of 884; Cannon 1997; Epstein 1992; Kocaturk 2005; Nawasreh 2004; Salam 1992; Scott 1992; Zacharia 1993; Zipfel 1999).

To structure the presentation of the results below they are reported by whether the trials: (1) compared control versus treatment or compared treatment versus treatment, (2) which treatment was tested, (3) whether the treatment was single application or prolonged application, (4) whether the design was by child or by ear, and (5) whether the trial was graded as low risk of bias or high risk of bias. Note that the majority of studies assessed outcomes at up to two weeks. The possible combinations of intervention type, randomisation type and intervention duration are summarised in Table 1. Entries are only included if we found a trial that considered a particular combination, and only trials that compared an intervention against either a placebo or a no treatment group are included (thus Morpeth 2001 is excluded). 

1. Summary of those trials that compared an intervention versus no treatment/placebo.

Type of intervention	Type of randomisation	Outcomes measured at up to 2 weeks		Outcomes measured at other time points
		Single application	Prolonged application	Single application	Prolonged application
#1. Irrigation of middle ear	By ear	‐	‐	‐	‐
...	By child	High risk of bias Kocaturk 2005	‐	‐	‐
#2. Antiseptic preparation	By ear	‐	‐	‐	‐
...	By child	‐	‐	High risk of bias Cannon 1997 (1 week)	‐
#3. Topical antibiotics	By ear	High risk of bias Nawasreh 2004 Zipfel 1999 Scott 1992	High risk of bias Nawasreh 2004	‐	‐
...	By child	‐	Low risk of bias Poetker 2006 High risk of bias Cannon 1997 Kocaturk 2005 Scott 1992	‐	‐
#4. Topical antibiotic/steroids	By ear	Low risk of bias Shinkwin 1996 High risk of bias Zacharia 1993	High risk of bias Salam 1992	Low risk of bias Arya 2004 (6 weeks) Welling 1995 (4 weeks)	‐
...	By child	‐	Low risk of bias Giles 2007 Poetker 2006 High risk of bias Epstein 1992	‐	‐
#6. Oral antibiotics	By ear	‐	‐	‐	‐
...	By child	‐	High risk of bias Kocaturk 2005	‐	‐
#7. Oral antibacterial agents/steroids	By ear	‐	‐	‐	‐
...	By child	‐	Low risk of bias Daly 1995	‐	Low risk of bias Daly 1995 (3 weeks)
#8. Antiseptic/saline preparation and topical antibiotics	By ear	‐	‐	‐	‐
...	By child	‐	‐	‐	High risk of bias Cannon 1997 (1 week)

The table summarises which combinations of intervention type (rows), randomisation type (rows) and intervention duration (columns) we found data for. Blank cells indicate that no relevant trial was found. Only trials that compared an intervention against either a placebo or a no treatment group are included (thus Morpeth 2001 is excluded). The index numbers refer to the subsections of Effects of interventions where the analyses are reported. Low risk of bias trials are highlighted in bold. 

For most of the combinations of interventions we found no more than one or two trials. There were sufficient trials for two meta‐analyses of the low risk of bias trials (see Analysis 4.3 and Analysis 13.1). The I² statistic for both was 0%, indicating minimal heterogeneity. We also conducted four meta‐analyses of the high risk of bias trials (Analysis 3.1, Analysis 3.3.2, Analysis 4.2.3 and Analysis 5.1). The I² statistic for two of these was 0% but for two others was 68% and 74%. Although the latter two are above the usual criterion for heterogeneity (Higgins 2002; Higgins 2003), given that the data were from high risk of bias trials we felt that the meta‐analyses were reasonable.

4.3. Analysis.

Comparison 4 Control versus topical antibiotics/steroids, Outcome 3 Single application/assessment at up to 4 to 6 weeks/by ear (low risk of bias).

13.1. Analysis.

Comparison 13 Topical antibiotics versus topical antibiotics/steroids, Outcome 1 Prolonged application/assessment at up to 2 to 3 weeks/by ear (high risk of bias).

3.1. Analysis.

Comparison 3 Control versus topical antibiotics, Outcome 1 Single application/assessment at up to 2 weeks/by ear (high risk of bias).

3.3. Analysis.

Comparison 3 Control versus topical antibiotics, Outcome 3 Prolonged application/assessment at up to 2 weeks/by child.

4.2. Analysis.

Comparison 4 Control versus topical antibiotics/steroids, Outcome 2 Prolonged application/assessment at up to 2 weeks/by child or ear.

5.1. Analysis.

Comparison 5 Control versus topical antibiotics or topical antibiotics/steroids, Outcome 1 Single application/assessed at up to 2 weeks/by child or ear (high risk of bias).

Primary outcome: postoperative otorrhoea

A. Tests of a treatment against control

1. Control versus irrigation of the middle ear

A 'by child' trial that was of overall high risk of bias included two relevant arms of low risk of bias (Kocaturk 2005). These were repeated middle‐ear irrigation of 1 ml of saline until the middle ears were cleared of 'sticky fluid' versus no irrigation after tube insertion. The prevalence of otorrhoea at two weeks was significantly reduced from 30% in control children to 16% in children whose ears were irrigated (risk ratio (RR) 0.52, 95% confidence interval (CI) 0.27 to 1.00, number needed to treat to benefit (NNTB) = 7) (Analysis 1.1).

1.1. Analysis.

Comparison 1 Control versus irrigation of the middle ear, Outcome 1 Single application/assessment at up to 2 weeks/by child (low risk of bias).

2. Control versus antiseptic/saline irrigation of the middle ear

A high risk of bias 'by child' trial compared triple irrigation of the ear canal with 50% solution of povidone‐iodine topical antiseptic (Betadine) and saline for one minute before the tube was inserted versus no canal irrigation (Cannon 1997). The risk ratio at one week was not significant.

3. Control versus topical antibiotics

Three high risk of bias 'by ear' trials compared a single application of topical antibiotics versus control (Nawasreh 2004; Scott 1992; Zipfel 1999). A meta‐analysis demonstrated that the prevalence of otorrhoea at two weeks was significantly reduced from 13% in 268 ears with no treatment to 6% in 267 ears with ciprofloxacin or gentamicin (RR 0.47, 95% CI 0.27 to 0.83, NNTB 14.9) (see Analysis 3.1). One of those trials also compared a prolonged application of ciprofloxacin versus control (Nawasreh 2004). The risk ratio at two weeks was not significant (Analysis 3.2).

3.2. Analysis.

Comparison 3 Control versus topical antibiotics, Outcome 2 Prolonged application/assessment at up to 2 weeks/by ear (high risk of bias).

A low risk of bias 'by child' trial compared a prolonged application of ofloxacin ear drops versus control (Poetker 2006). The prevalence of otorrhoea at two weeks was significantly reduced from 15% in 174 control ears to 8% in 198 ears given ofloxacin (RR 0.54, 95% CI 0.30 to 0.97, NNTB 14.6) (Analysis 3.3.1). 

Three high risk of bias 'by child' trials compared a prolonged application of topical antibiotics versus control (Cannon 1997; Kocaturk 2005; Scott 1992). A meta‐analysis demonstrated that the prevalence of otorrhoea at two weeks was significantly reduced from 21% in 137 control children to 12% in 138 children given ear drops (gentamicin ophthalmic drops (Garamycin), ofloxacin ear drops or gentamicin otic drops) (RR 0.55, 95% CI 0.31 to 0.96, NNTB 10.4) (Analysis 3.3.2).

4. Control versus topical antibiotics/steroids

A low risk of bias, 'by ear' trial compared administration of four drops of gentamicin/hydrocortisone (Gentisone HC) at the time of surgery in one ear versus four drops of saline in the other ear (Shinkwin 1996). The prevalence of otorrhoea at two weeks was significantly reduced from 9% in the 161 ears with saline drops to 1% in the 161 ears with hydrocortisone/gentamicin drops (RR 0.13, 95% CI 0.03 to 0.57, NNTB 12.5) (Analysis 4.1.1).

4.1. Analysis.

Comparison 4 Control versus topical antibiotics/steroids, Outcome 1 Single application/assessment at up to 2 weeks/by ear.

A high risk of bias 'by ear' trial compared a single application of gentamicin/hydrocortisone versus control (Zacharia 1993). The prevalence of otorrhoea at two weeks was significantly reduced from 13% in 122 ears with no treatment to 2% in 122 ears with gentamicin/hydrocortisone drops (RR 0.13, 95% CI 0.03 to 0.53, NNTB 8.7) (Analysis 4.1.2).

Two low risk of bias trials compared prolonged applications of topical antibiotic/steroids versus control (Giles 2007; Poetker 2006). It was not possible to perform a meta‐analysis as Giles 2007 reported the outcomes by child at either one or both ears, but Poetker 2006 reported the outcomes as the total number of ears. The results were therefore analysed separately. In Giles 2007, the prevalence of otorrhoea at two weeks was significantly reduced from 39% in 99 control children to 5% in 101 children given ciprofloxacin/dexamethasone (RR 0.13, 95% CI 0.05 to 0.31, NNTB 2.9) (Analysis 4.2.1). In Poetker 2006 the prevalence of otorrhoea at two weeks was significantly reduced from 15% in 174 control ears to 5% in 182 ears given neomycin otic drops (RR 0.37, 95% CI 0.18 to 0.74, NNTB 10.6) (Analysis 4.2.2).

Two high risk of bias 'by child' trials compared a prolonged application of topical antibiotic‐steroid versus control (Epstein 1992; Salam 1992). A meta‐analysis demonstrated that the prevalence of otorrhoea at two weeks was significantly reduced from 11% in 374 control children to 7% in 380 children given ear drops (sulfacetamide/prednisolone or betamethasone/neomycin (Betnesol‐N) drops) (RR 0.61, 95% CI 0.38 to 0.97, NNTB 22.8) (Analysis 4.2.3).

Two low risk of bias 'by ear' trials compared a single application of a topical antibiotic/steroid combination versus control: one used hydrocortisone/neomycin/polymyxin B (Cortisporin) and measured outcomes at up to four weeks (Welling 1995), the other used dexamethasone/framycetin/gramicidin (Sofradex) and measured outcomes at six weeks (Arya 2004). A meta‐analysis of the six‐week data showed no significant difference in the prevalence of otorrhoea with dexamethasone/framycetin/gramicidin or hydrocortisone/neomycin/polymyxin B in comparison to control (Analysis 4.3; Figure 4).

4.

Forest plot of Analysis 4.3: Control versus single application of topical antibiotics/steroids, assessed at up to 4 to 6 weeks (by ear/low risk of bias RCTs). See section A.4 of Effects of interventions for more information.

5. Control versus topical antibiotics or topical antibiotics/steroids

We combined parts of the two previous analyses into one, as four high risk of bias trials compared a single application of either a topical antibiotic or topical antibiotic/steroid combination versus control (Nawasreh 2004; Scott 1992; Zacharia 1993; Zipfel 1999). A meta‐analysis demonstrated that the prevalence of otorrhoea at two weeks was significantly reduced from 13% in 390 control ears to 5% in 389 treatment ears receiving topical antibiotics with or without steroids (RR 0.36, 95% CI 0.22 to 0.61, NNTB 13) (Analysis 5.1).

6. Control versus oral antibiotics

A high risk of bias 'by child' trial compared a prolonged application of an oral antibiotic versus control (Kocaturk 2005). The prevalence of otorrhoea at two weeks was significantly reduced from 30% in 70 control children to 14% in 70 children given sulbactam ampicillin (RR 0.48, 95% CI 0.24 to 0.94, NNTB 6.3) (Analysis 6.1).

6.1. Analysis.

Comparison 6 Control versus oral antibiotics, Outcome 1 Prolonged application/assessment at up to 2 weeks/by child (high risk of bias).

7. Control versus antibacterial agents/steroids

A low risk of bias 'by child' trial compared a prolonged application of an oral combination versus placebo (Daly 1995). The prevalence of otorrhoea at two weeks was significantly reduced from 39% in 36 children given placebo to 5% in 41 children given prednisone and trimethoprim‐sulfamethoxazole (RR 0.13, 95% CI 0.03 to 0.51, NNTB 2.9) (Analysis 7.1). The same trial also reported outcomes at three weeks. The prevalence of otorrhoea showed a significant decrease from 54% given placebo to 17% given treatment (RR 0.31, 95% CI 0.14 to 0.68. NNTB 2.7) (Analysis 7.2).

7.1. Analysis.

Comparison 7 Control versus oral antibacterial agents/steroids, Outcome 1 Prolonged application/assessment at up to 2 weeks/by child (low risk of bias).

7.2. Analysis.

Comparison 7 Control versus oral antibacterial agents/steroids, Outcome 2 Prolonged application/assessment at 3 weeks/by child (low risk of bias).

8. Control versus combination of topical antibiotics and antiseptic/saline preparation

A high risk of bias 'by child' trial compared triple irrigation of the ear canal with a 50% solution of povidone‐iodine topical antiseptic and saline for one minute before the tube was inserted versus no irrigation (Cannon 1997). The risk ratio at one week was not significant (Analysis 8.1).

8.1. Analysis.

Comparison 8 Control versus combination of topical antibiotics and antiseptic/saline preparation, Outcome 1 Prolonged application/assessment at up to 1 week/by child (high risk of bias).

B. Tests of one treatment against another

9. Irrigation of the middle ear versus topical antibiotics

A high risk of bias 'by child' trial compared repeated middle ear irrigation of 1 ml of saline until the middle ears were cleared of 'sticky fluid' versus a prolonged application of ofloxacin (Kocaturk 2005). The risk ratio at two weeks was not significant (Analysis 9.1).

9.1. Analysis.

Comparison 9 Irrigation of middle ear versus topical antibiotics, Outcome 1 Prolonged application/assessment at up to 2 weeks/by child (high risk of bias).

10. Irrigation of the middle ear versus oral antibiotics

A high risk of bias trial compared repeated middle ear irrigation of 1 ml of saline until the middle ears were cleared of 'sticky fluid' versus a prolonged application of oral sulbactam/ampicillin (Kocaturk 2005). The risk ratio at two weeks was not significant (Analysis 10.1).

10.1. Analysis.

Comparison 10 Irrigation of middle ear versus oral antibiotics, Outcome 1 Prolonged application/assessment up to 2 weeks/by child (high risk of bias).

11. Antiseptic/saline preparation versus topical antibiotics

A high risk of bias 'by child' trial compared triple irrigation of the ear canal with a 50% solution of povidone‐iodine topical antiseptic and saline for one minute before the tube was inserted versus a prolonged application of gentamicin ophthalmic drops (Cannon 1997). The risk ratio at one week was not significant (Analysis 11.1).

11.1. Analysis.

Comparison 11 Antiseptic/saline preparation versus topical antibiotics, Outcome 1 Prolonged application/assessment at up to 1 week/by child.

12. Antiseptic/saline preparation versus combination of topical antibiotics and antiseptic/saline preparation

A high risk of bias 'by child' trial compared triple irrigation of the ear canal with a 50% solution of povidone‐iodine topical antiseptic and saline for one minute before the tube was inserted with a prolonged application of gentamicin versus a prolonged application of gentamicin (Cannon 1997). The risk ratio at one week was not significant (Analysis 12.1).

12.1. Analysis.

Comparison 12 Antiseptic/saline preparation versus combination of topical antibiotics and antiseptic/saline preparation, Outcome 1 Single and prolonged application/assessment up to 1 week/by child (high risk of bias).

13. Topical antibiotics versus topical antibiotics/steroids

Two low risk of bias trials compared a prolonged application of topical antibiotic/steroids versus a prolonged application topical antibiotics (Morpeth 2001; Poetker 2006). Morpeth 2001 randomised by ear whereas Poetker 2006 randomised by child but reported outcomes by ear. A meta‐analysis showed that the risk ratio at two to three weeks was not significant (Analysis 13.1; Figure 5).

5.

Forest plot of Analysis 13.1: Prolonged application of topical antibiotics versus topical antibiotics/steroids, assessed at up to 2 to 3 weeks (by ear/low risk of bias RCTs). See section A.13 of Effects of interventions for more information.

14. Topical antibiotics versus combination of topical antibiotics and antiseptic/saline preparation

A high risk of bias 'by child' trial assessed the additive effect of triple irrigation of the ear canal with a 50% solution of povidone‐iodine topical antiseptic and saline for one minute before the tube was inserted with a prolonged application of gentamicin versus a prolonged application of gentamicin (Cannon 1997). The risk ratio at one week was not significant (Analysis 14.1).

14.1. Analysis.

Comparison 14 Topical antibiotics versus combination of topical antibiotics and antiseptic preparation, Outcome 1 Single and prolonged application/assessment at up to 1 week/by child (high risk of bias).

15. Topical antibiotics versus oral antibiotics

A high risk of bias 'by child' trial compared a prolonged application of oral sulbactam/ampicillin versus a prolonged application of ofloxacin (Kocaturk 2005). The risk ratio at two weeks was not significant (Analysis 15.1).

15.1. Analysis.

Comparison 15 Topical antibiotics versus oral antibiotics, Outcome 1 Prolonged application/assessment at up to 2 weeks/by child (high risk of bias).

C. Tests of single versus prolonged applications of treatment

16. Single versus prolonged application (topical antibiotics)

A high risk of bias 'by ear' trial compared a single application of gentamicin versus a prolonged application of gentamicin (Scott 1992). The risk ratio at two weeks was not significant (Analysis 16.1).

16.1. Analysis.

Comparison 16 Single versus prolonged application of topical antibiotics, Outcome 1 Gentamicin/assessment at up to 2 weeks/by child (high risk of bias).

A second high risk of bias 'by ear' trial compared a single application of ciprofloxacin versus a prolonged application of ciprofloxacin (Nawasreh 2004). The risk ratio at two weeks was not significant (Analysis 16.2).

16.2. Analysis.

Comparison 16 Single versus prolonged application of topical antibiotics, Outcome 2 Ciprofloxacin/assessment at up to 2 weeks/by ear (high risk of bias).

Secondary outcomes

One low risk of bias trial (Giles 2007) reported adverse effects (e.g. fever) in 20 cases out of a total of 200 children (eight in the control group and 12 in the intervention group), but they were unrelated to the treatment of a prolonged application of topical antibiotic‐steroids. A second low risk of bias trial considered ototoxicity and reported hearing before and after treatment (Welling 1995). The treatment was middle ear instillation of 0.5 ml of a topical aminoglycoside antibiotic/steroid combination. No significant change in hearing thresholds was found. 

Some other studies reported ear discharge (Arya 2004; Giles 2007), bacterial culture (Cannon 1997; Epstein 1992) and patency of ventilation tube (Daly 1995; Epstein 1992; Poetker 2006; Zacharia 1993; Zipfel 1999), but none gave sufficient details to be reported.

Only one trial evaluated cost‐effectiveness (Kocaturk 2005). The study was conducted in Turkey with the costs reported in US dollars. The costs per patient were USD 0.15 for saline + syringe, USD 1.51 for topical antibiotics and USD 9 for oral antibiotics. Given their measurements of the percentage of successful patients (respectively 84%, 91%, 86%) the cost per successfully treated patient was USD 0.18, USD 1.65 and USD 10.50 respectively. Current costs are likely to differ since this study was conducted between 2000 and 2004.

No trials reported the secondary outcomes of societal burden or industry funding.

Predictor factors for otorrhoea and potential benefit

None of the trials analysed if the outcomes were related to whether the indication for surgery was acute otitis media or otitis media with effusion. There was, however, a clear link between the values of the NNTB and the initial rates of otorrhoea (see Table 2): the higher the initial rate, the lower the NNTB.

2. Number needed to treat to benefit (NNTB).

Study	Time of assessment	% Otorrhoea in control group	% Otorrhoea in experimental group	Risk ratio	NNTB	Analysis
Daly 1995	3 weeks	54%	17%	0.31	2.7	7.2
Giles 2007	Up to 2 weeks	39%	5%	0.13	2.9	4.2.1
Daly 1995	Up to 2 weeks	39%	5%	0.13	2.9	7.1
Kocaturk 2005	Up to 2 weeks	30%	16%	0.52	7.0	1.1
Poetker 2006	Up to 2 weeks	15%	5%	0.37	10.6	4.2.2
Shinkwin 1996	Up to 2 weeks	9%	1%	0.13	12.5	4.1.1
Poetker 2006	Up to 2 weeks	15%	8%	0.54	14.6	3.3.1

The table summarises the rates of otorrhoea in the control group and the values of NNTB for those low risk of bias trials that gave a significant effect of treatment. The studies are sorted by NNTB. The rates in the experimental group, the risk ratio and the index number of the relevant analysis are reported for reference.

Discussion

Summary of main results

Primary outcomes

We found 15 randomised controlled trials (RCTs) that considered interventions for preventing post‐tympanostomy otorrhoea in children. Seven low risk of bias RCTs were found and an eighth RCT had two low risk of bias arms, although was otherwise at high risk of bias. The primary outcome was otoscopic observation of postoperative otorrhoea from one or both ears. The grading of quality was primarily based on whether the assessor was blind when performing otoscopy. No trial documented child or parent reports of otorrhoea. Due to the methodological heterogeneity of the RCTs it was not possible to conduct a meta‐analysis of all the RCTs, though some smaller meta‐analyses were possible. The primary methodological differences were in the choice of intervention, the duration of the application of the intervention and the unit of randomisation.

For a single application at the time of surgery, we found that the risk of otorrhoea at two weeks postoperatively was reduced by:

• multiple saline washouts (from 30% to 16%; risk ratio (RR) 0.52, 95% confidence interval (CI) 0.27 to 1.00) in 140 children (Analysis 1.1);

• antibiotic/steroid ear drops (from 9% to 1%; RR 0.13, 95% CI 0.03 to 0.57) in 322 ears (Analysis 4.1.1).

A meta‐analysis of two trials failed to find an effect of antibiotic/steroid ear drops at four to six weeks postoperatively (222 ears; Analysis 4.3).

For prolonged applications we found that the risk of otorrhoea was reduced at two weeks postoperatively by:

• antibiotic ear drops (from 15% to 8%; RR 0.54, 95% CI 0.30 to 0.97) in 372 children (Analysis 3.3.1);

• antibiotic/steroid ear drops (from 39% to 5%; RR 0.13, 95% CI 0.05 to 0.31) in 200 children (Analysis 4.2.1);

• aminoglycoside/steroid ear drops (from 15% to 5%; RR 0.37, 95% CI 0.18 to 0.74) in 356 children (Analysis 4.2.2);

• oral antibacterial agent with oral steroids (from 39% to 5%; RR 0.13, 95% CI 0.03 to 0.51) in 77 children (Analysis 7.1).

Variations in incidence of otorrhoea

In the low risk of bias trials, the number needed to treat to benefit (NNTB) to eliminate otorrhoea in one child ranged from 3 to 14. The range is because of the difference in the otorrhoea rates in the control groups, varying from 9% to 54%. The most likely reason for this effect is a difference in the proportion of children whose indication was recurrent acute otitis media.

Secondary outcomes

Ototoxicity

One trial investigated ototoxicity but found no effect on the hearing thresholds with the use of aminoglycoside ear drops. This result must be treated with caution, however, as it was based on a small sample of 100 ears in 50 children: it would require much larger studies of this type to reliably exclude a small risk of ototoxicity. We argue that it is up to the surgeon to balance the potential benefit to the child of a single intervention of saline irrigation or topical antibiotic/steroid drops against the potential damaging effects of saline irrigation or ototoxicity of the ear drops.

Cost‐effectiveness

Only one trial reported cost‐effectiveness data (Kocaturk 2005), with costs per patient of USD 0.15 for saline + syringe, USD 1.51 for topical antibiotics and USD 9 for oral antibiotics. It is likely that these costs are now out of date, as the study was conducted in Turkey between 2000 and 2004. It is also not certain that the benefits to the children are sufficient to justify the costs involved. In modern health services cost is a crucial factor and it is often necessary to demonstrate cost‐effectiveness for an intervention. Given the paucity of data here we argue that new data need to be collected.

Comparison to previous reviews

Two earlier reviews have been reported (Garcia 1994; Hochman 2006). Garcia 1994 recommended the use of a prolonged application of topical antimicrobial prophylaxis following ventilation tube insertion, based on a meta‐analysis of three studies randomised by child. Hochman 2006 reported a meta‐analysis of nine heterogeneous studies and found a significant benefit of antimicrobial prophylaxis, though the recommendations were guarded.

Otorrhoea from grommet insertion was also included as part of a wider review by SIGN 2008. They recommended that a single application of antibiotic prophylaxis is used at the time of surgery in adults and children.

Nevertheless, each earlier review is less than ideal. None of them conducted an in‐depth quality appraisal, and the conclusions drawn were based on fewer trials than the present review. Garcia 1994 identified five trials, two of which we excluded from the present review because they were not true randomised controlled trials. A further two trials included both children and adults and the outcomes for children could not be separated out. The remaining trial was graded in our review as a being at a high risk of bias. Hochman 2006 included nine trials, two of which were not randomised controlled trials. Three trials included children and adults whose outcomes could not be separated for our review; the remaining four trials were graded here as being at high risk of bias. SIGN 2008 only considered three trials selected on an unclear basis, each of which we graded as being at a high risk of bias. Note that Garcia 1994 and Hochman 2006 only included trials where a prolonged application of prophylaxis was used.

Our review is the first to have considered a wide range of prophylaxis and whether it was given as a single or prolonged application. We also considered the type of children that are more likely to benefit from using prophylaxis as opposed to recommending that prophylaxis is used in all patients (as suggested by SIGN 2008). We also reported: (1) the risk ratios for the interventions with 95% confidence intervals, instead of odds ratios, so facilitating interpretation, and (2) the numbers needed to treat to benefit (NNTB) which is an important consideration in the implications for practice.

Our review supports the three earlier reviews regarding the efficacy of antibiotics prophylaxis but with better quality evidence analysed in multiple analyses. What we have added is quality evidence of the beneficial effect of saline middle ear irrigation at the time of insertion of the ventilation tube. We recommend prophylaxis at the time of surgery in those at high risk of otorrhoea, most likely infants. A single application of topical antibiotics with or without steroids and multiple middle ear saline washouts were found to be effective with a NNTB of 3 to 7. However, given the potential risks of ototoxicity with aminoglycosides, or of middle ear trauma due to inexpert irrigation, the recommendation is that the choice between these treatments is left to the surgeon, balancing the likely adverse effects along with their cost.

Authors' conclusions

Implications for practice.

The results indicate that various interventions are effective in reducing otorrhoea up to two weeks postoperatively:

• multiple saline washouts of the middle ear at the time of surgery;

• a single application at surgery of topical antibiotics with steroids;

• prolonged application of topical antibiotics (with or without steroids) or oral antibacterial agents/steroids.

Beyond the period of two weeks the evidence was limited, but the efficacy of these interventions was reduced and in one trial significantly increased. All but one of the five topical antibiotic trials used aminoglycosides, so more research is needed to clarify the risk of ototoxicity from topical therapy. However, if topical drops are chosen, it is suggested that to reduce the cost and potential for ototoxic damage this be a single application at the time of surgery and not prolonged thereafter.

Implications for research.

There is a need for studies on both persistent otitis media with effusion and recurrent acute otitis media that:

• assess the effects from postoperative otorrhoea on the children's and parents' quality of life;

• assess the rates of postoperative otorrhoea in relation to the indications for surgery.

The results of these studies are necessary to gauge the potential need for future, focused RCTs of individual treatments.

Appendices

Appendix 1. Search strategies

CENTRAL	PubMed	EMBASE (Ovid)	CINAHL (EBSCO)
#1 MeSH descriptor MIDDLE EAR VENTILATION explode all trees #2 Grommet* #3 middle NEXT ear NEAR (ventilation OR tube*) #4 (ventilation OR tympanostomy OR tympanic) NEAR tube* #5 tubulation OR tubulate #6 #1 OR #2 OR #3 OR #4 OR #5 #7 MeSH descriptor CEREBROSPINAL FLUID OTORRHEA explode all trees #8 OTOR* OR LEAK* OR DISCHARG* #9 LIQUORRH* OR LIQUORH* OR OTOLIQUORRH* OR OTOLIQUORH* #10 #7 OR #8 OR #9 #11 #6 AND #10 #12 MeSH descriptor Premedication explode all trees #13 MeSH descriptor Preoperative Care explode all trees #14 MeSH descriptor Perioperative Care single term #15 premedicat* OR prophyl* OR prevent* OR preoperative*:ti OR pre‐operative*:ti OR perioperative*:ti OR peri‐operative*:ti #16 #12 OR #13 OR #14 OR 15 #17 #11 AND #16 #18 MeSH descriptor Cerebrospinal Fluid Otorrhea explode all trees with qualifier: PC #19 #18 AND #6 #20 #17 OR #19	#1 “MIDDLE EAR VENTILATION” [Mesh] OR Grommet* [tiab] OR (middle [tiab] AND ear [tiab] AND (ventilation [tiab] OR tube* [tiab])) OR ((ventilation [tiab] OR tympanostomy [tiab] OR tympanic [tiab]) AND tube* [tiab]) OR tubulation [tiab] OR tubulate [tiab] #2 “CEREBROSPINAL FLUID OTORRHEA” [Mesh] OR OTOR* [tiab] OR LEAK* [tiab] OR DISCHARG* [tiab] OR LIQUORRH* [tiab] OR LIQUORH* [tiab] OR OTOLIQUORRH* [tiab] OR OTOLIQUORH* [tiab] #3 #1 AND #2 #4 "Middle Ear Ventilation/adverse effects"[Mesh] #5 #3 OR #4 #6 “Premedication” [Mesh] OR “Preoperative Care” [Mesh] OR “Perioperative Care” [Mesh] OR premedicat* [tiab] OR prophyl* [tiab] OR prevent* [tiab] OR preoperative* [ti] OR pre‐operative* [ti] OR perioperative* [ti] OR peri‐operative* [ti] #7 #5 AND #6 #8 "Cerebrospinal Fluid Otorrhea/prevention and control"[Mesh] #9 #7 OR #8	1 tympanostomy tube/ 2 (Grommet* or (middle and ear and (ventilation or tube*)) or ((ventilation or tympanostomy or tympanic) and tube*) or tubulation or tubulate).tw. 3 1 or 2 4 otorrhea/ or cerebrospinal fluid otorrhea/ or tympanostomy tube otorrhea/ 5 (OTOR* or LEAK* or DISCHARG* or LIQUORRH* or LIQUORH* or OTOLIQUORRH* or OTOLIQUORH*).tw. 6 4 or 5 7 3 and 6 8 exp prophylaxis/ or prevention/ 9 preoperative care/ 10 (premedicat* or prophyl* or prevent*).tw. 11 (preoperative* or pre‐operative* or perioperative* or peri‐operative*).ti. 12 8 or 9 or 10 or 11 13 7 and 12	S1 (MH "Middle Ear Ventilation") S2 TX (Grommet* or (middle and ear and (ventilation or tube*)) or ((ventilation or tympanostomy or tympanic) and tube*) or tubulation or tubulate) S3 S1 or S2 S4 (MH "Cerebrospinal Fluid Otorrhea") S5 TX (OTOR* or LEAK* or DISCHARG* or LIQUORRH* or LIQUORH* or OTOLIQUORRH* or OTOLIQUORH*) S6 S4 or S5 S7 (S4 or S5) and (S3 and S6) S8 (MH "Premedication") S9 TX (premedicat* or prophyl* or prevent*) S10 TI (preoperative* or pre‐operative* or perioperative* or peri‐operative*) S11 S8 or S9 or S10 S12 S3 and S6 and S11
Web of Science	BIOSIS Previews (Web of Knowledge)	CAB Abstracts (Ovid)	ISRCTN (mRCT)
#1 TS=(Grommet* or (middle and ear and (ventilation or tube*)) or ((ventilation or tympanostomy or tympanic) and tube*) or tubulation or tubulate) #2 TS=(OTOR* or LEAK* or DISCHARG* or LIQUORRH* or LIQUORH* or OTOLIQUORRH* or OTOLIQUORH*) #3 #2 AND #1 #4 TS=(premedicat* or prophyl* or prevent*) #5 TI=(preoperative* or pre‐operative* or perioperative* or peri‐operative*) #6 #5 OR #4 #7 #6 AND #3	#1 TS=(Grommet* or (middle and ear and (ventilation or tube*)) or ((ventilation or tympanostomy or tympanic) and tube*) or tubulation or tubulate) #2 TS=(OTOR* or LEAK* or DISCHARG* or LIQUORRH* or LIQUORH* or OTOLIQUORRH* or OTOLIQUORH*) #3 #2 AND #1 #4 TS=(premedicat* or prophyl* or prevent*) #5 TI=(preoperative* or pre‐operative* or perioperative* or peri‐operative*) #6 #5 OR #4 #7 #6 AND #3	1 (Grommet* or (middle and ear and (ventilation or tube*)) or ((ventilation or tympanostomy or tympanic) and tube*) or tubulation or tubulate).tw. 2 (OTOR* or LEAK* or DISCHARG* or LIQUORRH* or LIQUORH* or OTOLIQUORRH* or OTOLIQUORH*).tw. 3 1 AND 2 4 exp prophylaxis/ or prevention/ 5 preoperative care/ 6 (premedicat* or prophyl* or prevent*).tw. 7 (preoperative* or pre‐operative* or perioperative* or peri‐operative*).ti. 8 4 OR 5 OR 6 OR 7 9 3 AND 8	Grommet% OR (ear AND tube%) OR (ear AND ventilation)

Data and analyses

Comparison 1. Control versus irrigation of the middle ear.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Single application/assessment at up to 2 weeks/by child (low risk of bias)	1	140	Risk Ratio (M‐H, Fixed, 95% CI)	0.52 [0.27, 1.00]

Comparison 2. Control versus antiseptic irrigation of middle ear.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Single application/assessment at up to 1 week/by child (high risk of bias)	1	100	Risk Ratio (M‐H, Fixed, 95% CI)	1.25 [0.36, 4.38]

2.1. Analysis.

Comparison 2 Control versus antiseptic irrigation of middle ear, Outcome 1 Single application/assessment at up to 1 week/by child (high risk of bias).

Comparison 3. Control versus topical antibiotics.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Single application/assessment at up to 2 weeks/by ear (high risk of bias)	3	535	Risk Ratio (M‐H, Fixed, 95% CI)	0.47 [0.27, 0.83]
2 Prolonged application/assessment at up to 2 weeks/by ear (high risk of bias)	1	196	Risk Ratio (M‐H, Fixed, 95% CI)	0.49 [0.22, 1.09]
3 Prolonged application/assessment at up to 2 weeks/by child	4		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.1 Low risk of bias trial (reported by ear)	1	372	Risk Ratio (M‐H, Fixed, 95% CI)	0.54 [0.30, 0.97]
3.2 High risk of bias trials	3	275	Risk Ratio (M‐H, Fixed, 95% CI)	0.55 [0.31, 0.96]

Comparison 4. Control versus topical antibiotics/steroids.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Single application/assessment at up to 2 weeks/by ear	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1 Low risk of bias trial	1	322	Risk Ratio (M‐H, Fixed, 95% CI)	0.13 [0.03, 0.57]
1.2 High risk of bias trial	1	244	Risk Ratio (M‐H, Fixed, 95% CI)	0.13 [0.03, 0.53]
2 Prolonged application/assessment at up to 2 weeks/by child or ear	4		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
2.1 Low risk of bias trial	1	200	Risk Ratio (M‐H, Fixed, 95% CI)	0.13 [0.05, 0.31]
2.2 Low risk of bias trial (reported by ear)	1	356	Risk Ratio (M‐H, Fixed, 95% CI)	0.37 [0.18, 0.74]
2.3 High risk of bias trials	2	754	Risk Ratio (M‐H, Fixed, 95% CI)	0.61 [0.38, 0.97]
3 Single application/assessment at up to 4 to 6 weeks/by ear (low risk of bias)	2	222	Risk Ratio (M‐H, Fixed, 95% CI)	0.67 [0.28, 1.57]

Comparison 5. Control versus topical antibiotics or topical antibiotics/steroids.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Single application/assessed at up to 2 weeks/by child or ear (high risk of bias)	4	779	Risk Ratio (M‐H, Fixed, 95% CI)	0.36 [0.22, 0.61]

Comparison 6. Control versus oral antibiotics.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Prolonged application/assessment at up to 2 weeks/by child (high risk of bias)	1	140	Risk Ratio (M‐H, Fixed, 95% CI)	0.48 [0.24, 0.94]

Comparison 7. Control versus oral antibacterial agents/steroids.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Prolonged application/assessment at up to 2 weeks/by child (low risk of bias)	1	77	Risk Ratio (M‐H, Fixed, 95% CI)	0.13 [0.03, 0.51]
2 Prolonged application/assessment at 3 weeks/by child (low risk of bias)	1	77	Risk Ratio (M‐H, Fixed, 95% CI)	0.31 [0.14, 0.68]

Comparison 8. Control versus combination of topical antibiotics and antiseptic/saline preparation.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Prolonged application/assessment at up to 1 week/by child (high risk of bias)	1	100	Risk Ratio (M‐H, Fixed, 95% CI)	1.25 [0.36, 4.38]

Comparison 9. Irrigation of middle ear versus topical antibiotics.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Prolonged application/assessment at up to 2 weeks/by child (high risk of bias)	1	140	Risk Ratio (M‐H, Fixed, 95% CI)	1.83 [0.72, 4.68]

Comparison 10. Irrigation of middle ear versus oral antibiotics.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Prolonged application/assessment up to 2 weeks/by child (high risk of bias)	1	140	Risk Ratio (M‐H, Fixed, 95% CI)	1.1 [0.50, 2.42]

Comparison 11. Antiseptic/saline preparation versus topical antibiotics.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Prolonged application/assessment at up to 1 week/by child	1	100	Risk Ratio (M‐H, Fixed, 95% CI)	0.71 [0.24, 2.10]

Comparison 12. Antiseptic/saline preparation versus combination of topical antibiotics and antiseptic/saline preparation.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Single and prolonged application/assessment up to 1 week/by child (high risk of bias)	1	100	Risk Ratio (M‐H, Fixed, 95% CI)	1.0 [0.31, 3.24]

Comparison 13. Topical antibiotics versus topical antibiotics/steroids.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Prolonged application/assessment at up to 2 to 3 weeks/by ear (high risk of bias)	2	580	Risk Ratio (M‐H, Fixed, 95% CI)	0.73 [0.46, 1.16]

Comparison 14. Topical antibiotics versus combination of topical antibiotics and antiseptic preparation.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Single and prolonged application/assessment at up to 1 week/by child (high risk of bias)	1	100	Risk Ratio (M‐H, Fixed, 95% CI)	0.71 [0.24, 2.10]

Comparison 15. Topical antibiotics versus oral antibiotics.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Prolonged application/assessment at up to 2 weeks/by child (high risk of bias)	1	140	Risk Ratio (M‐H, Fixed, 95% CI)	1.67 [0.64, 4.34]

Comparison 16. Single versus prolonged application of topical antibiotics.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Gentamicin/assessment at up to 2 weeks/by child (high risk of bias)	1	195	Risk Ratio (M‐H, Fixed, 95% CI)	1.03 [0.40, 2.64]
2 Ciprofloxacin/assessment at up to 2 weeks/by ear (high risk of bias)	1	35	Risk Ratio (M‐H, Fixed, 95% CI)	0.71 [0.13, 3.72]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Arya 2004.

Methods	Randomisation type: by ear Randomisation method: sealed envelopes Blinding: double‐blind, patient and surgeon Design: test versus no treatment
Participants	Entered: 75 children Analysed: 122 ears Age range (entered): not stated Age range (analysed): median 5 years, IQR 4 to 7 years Sex (entered): not stated Sex (analysed): 30 boys, 31 girls Inclusion criteria: persistent middle ear effusions
Interventions	1. Sofradex (dexamethasone 0.05%, framycetin 0.5% and gramicidin 0.005%) instilled following grommet insertion 2. No treatment (in other ear)
Outcomes	Primary outcome: otorrhoea Assessed at 6 weeks postoperatively Assessment made by surgeon Assessment technique: ventilation tube patency accessed via tympanometry, clinical examination Secondary outcomes: parent or patient reports of ear discharge (no data reported)
Notes	Exclusion criteria: children with known allergy to Sofradex
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote (from correspondent): "A box was placed in theatre filled with shuffled identical envelopes. Each envelope contained 2 pieces of paper. One contained instructions to either place or not place the drops plus a section to record patient / disease information. This was then placed in a separate file and was not looked at until the end of the study. The second piece of paper was identical in both envelope types and was placed in the notes to record the findings at follow up. The patient was therefore selected at random and the examiner was blinded at follow up. Analysis was only commenced at the end of the study."
Allocation concealment (selection bias)	Low risk	Quote (from report): "Identical sealed envelopes indicating which ear was to be selected for Sofradex treatment were available in theatre and opened in consecutive order by a member of the theatre staff who were instructed the surgeon as to which ear to treat." (pg. 599)
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote (from report): "Reasons for incompleteness were patients who were lost to follow‐up (12 patients) and two grommets that had extruded before 6 weeks follow‐up." (pg. 599)
Blinding of participants?	Low risk	Intervention was received intra‐operatively. Cotton wool was placed in both ears thus blinding children as to which ear received treatment.
Blinding of assessor?	Low risk	Quote (from report): "The treated side was double blinded to both patient and surgeon at follow‐up." (pg. 599)
Free of other bias?	Low risk	Down’s syndrome and cleft palate children were included in this study (number of children not specified)

Cannon 1997.

Methods	Randomisation type: by child Randomisation method: simple alternation (1:1) Blinding: unblinded Design: test versus intervention versus intervention versus no treatment
Participants	Entered: 200 children Analysed: 200 children Age range (entered): less than 12 years old (overall age 25 months) Age range (analysed): unclear Sex (entered): 59% girls, 41% boys Sex (analysed): 59% girls, 41% boys Inclusion criteria: OME
Interventions	1. Triple irrigation of the ear canal with 50% povidone‐iodine topical antiseptic (Betadine)/saline (intra‐operatively) 2. No ear canal preparation, gentamicin ophthalmic drops (Garamycin) (postoperatively) 3. Triple irrigation of the ear canal with 50% povidone‐iodine topical antiseptic/saline followed by gentamicin ophthalmic drops postoperatively (4 days) 4. No treatment
Outcomes	Primary outcome: purulent otorrhoea (not stated whether at one or both ears) Assessed at 1 week postoperatively Assessment made by: not stated Assessment technique: not stated Secondary outcomes not reported
Notes	Exclusion criteria not stated but it was reported that no patients had cleft palate or neurosensory hearing loss
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote (from correspondent): "...patients randomised by simple alternation on a 1:1 basis."
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit a judgement of 'yes' or 'no'
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Blinding of participants?	High risk	Not possible
Blinding of assessor?	High risk	Quote (from correspondent): "Reviewer not blinded"
Free of other bias?	High risk	It was reported that 97.5% patients were taking antibiotics for otitis media with effusion at the time of surgery. The outcome (otorrhoea) was not stated as to whether it was at one or both ears.

Daly 1995.

Methods	Randomisation type: by child Randomisation method: permuted block randomisation Blinding: double‐blind Design: test versus placebo
Participants	Entered: 171 children Analysed: 77 children Age range (entered): 6 months to 8 years Age range (analysed): 6 months to 8 years Sex (entered): not stated Sex (analysed): 64% male, 46% female Inclusion criteria: ventilation tube insertion for MEE with/without adenoidectomy (one unilateral tube inserted, all other participants had bilateral tubes inserted)
Interventions	1. Prednisone (1 mg/kg/day for 7 days divided into 2 doses, followed by 1 mg/kg/day once every other day for the next 7 days) and TMP/SMX suspension (8 mg of trimethoprim and 40 mg of sulfamethoxazole/kg/day for 14 days in 2 doses) 2. Placebo group received preparations that were similar in taste and appearance to the active drugs
Outcomes	Primary outcome: otorrhoea (from one or both ears) Assessed at 14 days, 3 weeks and 3, 6, 9 and 12 months (though data only reported at 14 days and 3 weeks) Assessment made by otolaryngologist and research nurse Assessment technique: pneumatic otoscopy and tympanostomy Secondary outcomes: tympanometry assessment for patency of the ventilation tube (noted in the report but no outcome data reported)
Notes	Exclusion criteria: previous adenoidectomy or tonsillectomy, sensorineural hearing loss greater than 15 dB at 1 or more frequencies between 500 Hz and 4000 Hz; chronic illnesses; inability to return to follow‐ups; varicella exposure in the preceding 3 weeks if they had no history of varicella; systematic corticosteroid treatment within the previous 3 months; allergies to sulphonamide, penicillin or corticosteroid; anaemia, leukopenia or thrombocytopenia
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote (from report): "The trial was double blind: children were assigned to active drug or placebo groups from two strata with the use of permuted block randomisation with blocks of four consecutive patients." (pg. 1069)
Allocation concealment (selection bias)	Low risk	Quote (from report): "Those assigned to placebos received preparations that were similar in taste and appearance to the active drugs." (pg. 1069)
Incomplete outcome data (attrition bias) All outcomes	High risk	171 children agreed to participate. 80 participated in the clinical trial. Only 77 children were recorded in the analysis.
Blinding of participants?	Low risk	See above
Blinding of assessor?	Low risk	See above
Free of other bias?	Unclear risk	Some children received systematic antibiotics before surgery. Children with craniofacial abnormalities were excluded from the trial.

Epstein 1992.

Methods	Randomisation type: by child Randomisation method: unclear Blinding: unclear Design: test versus no treatment
Participants	Entered: 478 children Analysed: 430 children Age range (entered): 6 months to 14 years Age range (analysed): 4 months to 14 years (M = 3.1 years) Sex (entered): not stated Sex (analysed): 262 boys, 168 girls Inclusion criteria: insertion of ventilation tube for chronic OME and recurrent AOM
Interventions	1. Sulfacetamide/prednisolone acetate otic drops (sulfacetamide sodium and prednisolone acetate) intra‐operatively and continued postoperatively (3 drops, for 3 days, 3 times) 2. No treatment
Outcomes	Primary outcomes: otorrhoea at left ear and tube patency (purulent otorrhoea not stated, purulent otorrhoea not ruled out) Assessed at 1 week postoperatively Assessment made by: surgeon Assessment technique: otoscopy and immittance audiometry Secondary outcomes: tympanometry assessment for patency of the ventilation tube (noted in the report but no outcome data reported)
Notes	Exclusion criteria: cleft palate and Down's syndrome
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote (from report): "All subjects were randomly assigned to one of two groups ‐ experimental or control." (pg. 759)
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit a judgement of 'yes' or 'no'
Incomplete outcome data (attrition bias) All outcomes	Low risk	48 missing children (only children who attended the 1‐week postoperative follow‐up were included in the analysis)
Blinding of participants?	High risk	Not possible
Blinding of assessor?	Unclear risk	Insufficient information to permit a judgement of 'yes' or 'no'
Free of other bias?	High risk	Outcomes reported at the left ear only. Children with craniofacial abnormalities were excluded from the trial

Giles 2007.

Methods	Randomisation type: by child Randomisation method: generated randomisation table Blinding: single‐blind (physician/assessor blinded) Design: test versus no treatment
Participants	Entered: 207 children Analysed: 200 children Age range (entered): not stated Age range (analysed): 4.5 months to 4  years (median = 1 year) Sex (entered): not stated Sex (analysed): 127 boys, 73 girls Insertion criteria: insertion of ventilation tube for chronic OME, recurrent AOM and atelectatic OM (all bilateral tube insertion)
Interventions	1. Ciprofloxacin (0.3%)/dexamethasone (0.1%) otic suspension (Ciprodex) for 5 days, twice daily (in each ear) 2. No treatment
Outcomes	Primary outcome: otorrhoea and tube patency (one or both ears) Assessed at 2 weeks postoperatively or earlier if needed Assessed by investigator Assessment technique not stated Secondary outcomes: any reports of adverse effects: 20 adverse events were reported during the study, including 12 in the ciprofloxacin/dexamethasone group and 8 in the control group. All adverse events were deemed unrelated to treatment by the study investigator (see pg. 753 of report).
Notes	Exclusion criteria: Down's syndrome or any craniofacial abnormalities; chronic illness of known immunodeficiency; middle ear pathology in surgical ear(s) other than otitis media; elongated shaft tubes inhibiting visualisation of tube patency; allergies to quinolones or any inactive ingredients of Ciprodex; suspected fungal or mycobacterial infection; or non‐compliance and lost to follow‐up
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote (from report): "...generated randomization table..." (pg. 748)
Allocation concealment (selection bias)	Low risk	Low risk due to the design of the study
Incomplete outcome data (attrition bias) All outcomes	Low risk	7 missing children ‐ lost to follow‐up (1 ciprofloxacin/dexamethasone patient/4 control children) and noncompliance (1 ciprofloxacin/dexamethasone patient/1 control patient)
Blinding of participants?	High risk	Not possible
Blinding of assessor?	Low risk	Quote (from report): "The evaluating study investigator remained blinded to the treatment during the post‐op period and did not perform phone interviews with the children/caregivers." (pg. 749)
Free of other bias?	Unclear risk	Children with craniofacial abnormalities were excluded from the trial

Kocaturk 2005.

Methods	Randomisation type: by child Randomisation method: unclear Blinding: single‐blind (surgeon/assessor) Design: test versus test versus test versus no treatment
Participants	Entered: not stated Analysed: 280 Age range (entered): not stated Age range (analysed): 3 years to 11 years (M = 5.9 years) Sex (entered): not stated Sex (analysed): 160 boys, 120 girls Inclusion criteria: insertion of ventilation tubes for AOM; serous otitis media (bilateral tube insertion)
Interventions	1. Isotonic saline irrigation (1 ml) of middle ear intra‐operatively 2. Oral sulbactam ampicillin combination (25 mg/kg) postoperatively for 5 days starting at the day of the operation 3. Ofloxacin ear drops postoperatively (twice a day for 5 days) 4. No treatment
Outcomes	Primary outcome: purulent otorrhoea (from one or both ears) Assessed at follow‐up 2 weeks postoperatively Assessed by physician Assessment technique not stated Secondary outcomes: bacterial culture pre‐ and post‐insertion (some details reported but no data); cost‐effectiveness data (costs per patient, reported in US dollars)
Notes	Excluded criteria: children with adenoidectomy and/or tonsillectomy performed in the same session; cleft palate complications of otitis media; allergies to penicillin or quinolone group antibiotics; sinusitis
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote (from report): "Patients were selected randomly for each treatment group." (pg 124) ‐ insufficient information to permit a judgement of 'yes' or 'no'
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit a judgement of 'yes' or 'no'
Incomplete outcome data (attrition bias) All outcomes	Low risk	No withdrawals or drop‐outs reported
Blinding of participants?	High risk	Not possible
Blinding of assessor?	Low risk	Quote (from report): "The physician doing the postoperative examination did not know the type of treatment the patient received." (pg. 125)
Free of other bias?	Unclear risk	Children with craniofacial abnormalities were excluded from the trial

Morpeth 2001.

Methods	Randomisation type: by child Randomisation method: unclear Blinding: double‐blind (surgeon and participant) Design: test versus test
Participants	Entered: 100 children Analysed: 200 ears Age range (entered): 7 months to 11 years Age range (analysed): 7 months to 11 years Sex (entered): 57 boys, 43 girls Sex (analysed): 57 boys, 43 girls Inclusion criteria: insertion of ventilation tube for RAOM and COME
Interventions	1. Hydrocortisone/neomycin/polymyxin B (Cortisporin) otic suspension with equal volume of 4% lidocaine intra‐operatively and then hydrocortisone/neomycin/polymyxin B otic suspension without lidocaine continued postoperatively for 3 days (3 times daily) 2. Ciprofloxacin with equal volume of 4% lidocaine placed into the ear intra‐operatively and then ciprofloxacin without lidocaine continued postoperatively for 3 days (3 times daily) Note: if otorrhoea persisted after 3 days parents were instructed to continue drops for 3 days beyond end of the drainage
Outcomes	Primary outcome: otorrhoea (not stated from one or both ears) Assessed at 3 weeks follow‐up and earlier if needed Assessed by physician Assessment technique: otomicroscopy Secondary outcomes not reported
Notes	Exclusion criteria: children with cleft palate, craniofacial dysmorphism, immunodeficiency and mucociliary clearance disorders, and undergoing concurrent procedures
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote (from report): "... patients were randomised into either group A or B for a total of 50 patients in each group." (pg. 100). Insufficient information to permit judgement of 'yes' or 'no'
Allocation concealment (selection bias)	Low risk	The drug containers were labelled 'A' or 'B' when used by the attending surgeon, and used by the parents when continuing the medication postoperatively
Incomplete outcome data (attrition bias) All outcomes	Low risk	No missing outcome data
Blinding of participants?	Low risk	See above
Blinding of assessor?	Low risk	Quote (from correspondent): "... as assessor I was blinded to treatment."
Free of other bias?	Unclear risk	Children with craniofacial abnormalities were excluded from the trial

Nawasreh 2004.

Methods	Randomisation type: by ear Randomisation method: unclear Blinding: not stated Design: test versus test versus no treatment
Participants	Entered: unclear Analysed: 150 children Age range (entered): not stated Age range (analysed): 3 to 14 years Sex (entered): not stated Sex (analysed): not stated Inclusion criteria: insertion of ventilation tubes for chronic serous OM, recurrent AOM with/without concomitant adenoidectomy, tonsillectomy or both
Interventions	1. Ciprofloxacin (0.3%), 3/4 drops in the external ear immediately after surgery 2. Ciprofloxacin (0.3%), intra‐operative 3/4 drops in the external ear immediately after surgery and continued postoperatively for 5 days (3 times a day) 3. No treatment
Outcomes	Primary outcome: otorrhoea Assessed at 2 weeks postoperatively or earlier if needed Assessed by: not stated Assessment technique: not stated Secondary outcomes not reported
Notes	Exclusion criteria: diagnoses of adhesive otitis media
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote (from report): "Patients were randomised into 3 groups." (pg. 39). No further information given, therefore insufficient information to permit judgement of 'yes' or 'no'
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit judgement of 'yes' or 'no'
Incomplete outcome data (attrition bias) All outcomes	Low risk	150 children randomised, 292 tubes placed (assumed a mix of bilateral and unilateral tube insertions). All included in the outcome data.
Blinding of participants?	High risk	Not possible
Blinding of assessor?	Unclear risk	Insufficient information to permit a judgement of 'yes' or 'no'
Free of other bias?	Low risk	The trial appears to be free of other sources of bias

Poetker 2006.

Methods	Randomisation type: by child (but outcomes reported by ear) Randomisation: random number selection Blinding: single‐blind (assessor) Design: test versus test versus no treatment
Participants	Children entered: 306 Children analysed: 277 Age range (entered): not stated Age range (analysed): 6 months to 12.16 years (M = 22.5, 26.0 and 22.6 months in each intervention group) Sex (entered): not stated Sex (analysed): 151 males, 126 females Inclusion criteria: insertion of ventilation tube for otitis media (bilateral tube insertion)
Interventions	1. Ofloxacin otic drops (children with fluid demonstration in 1 or both middle ear spaces at the time of tube placement were instructed to use their drops postoperatively for 10 days. If no fluid was present, children were treated postoperatively for 3 days) 2. Neomycin sulfate/polymyxin B sulfate/hydrocortisone otic drops (children with fluid demonstration in 1 or both middle ear spaces at the time of TT placement were instructed to use their drops for 10 days postoperatively. If no fluid was present, children were treated for 3 days postoperatively). 3. No treatment
Outcomes	Primary outcome: otorrhoea and tube patency Assessed at 2 weeks follow‐up (majority seen between 10 and 14 days) Assessed by attending surgeon Assessment technique: otoscopy and tympanometry Secondary outcome: tympanometry assessment for patency of the ventilation tube (noted in the report but no outcome data reported)
Notes	Exclusion criteria: children with syndromic conditions, known craniofacial abnormalities, other chronic illness or other severe medical conditions, sensorineural hearing loss; cancellation or postponement of surgery; failure to follow up
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote (from report): "Random allocation was achieved using random number selection by the research coordinator and group allocation was kept concealed until the end of the trial." (pg. 1295)
Allocation concealment (selection bias)	Low risk	See above
Incomplete outcome data (attrition bias) All outcomes	Low risk	306 patient randomised, 277 analysed in their assigned group. All excluded children (n = 29) were accounted for in the report.
Blinding of participants?	High risk	Not possible
Blinding of assessor?	Low risk	Quote (from report): "All assessments made at the follow‐up appointment regarding TTO and tube patency were made with the observer blinded to the patients treatment group." (pg. 1295)
Free of other bias?	Unclear risk	Children with craniofacial abnormalities were excluded from the trial

Salam 1992.

Methods	Randomisation type: by ear Randomisation method: nurse chose blindly from a set of cards Blinding method: unclear Design: test versus no treatment
Participants	Entered: 176 children Analysed: 162 children (324 ears) Age range (entered): not stated Age range (analysed): 3 to 12 years (M = 4.6 years) Sex (entered): not stated Sex (analysed): 67 boys, 95 girls Inclusion criteria: insertion of ventilation tubes for OME
Interventions	1. Betamethasone/neomycin (Betnesol‐N) drops in 1 ear intra‐operatively and continued for 3 days postoperatively 2. No treatment in other ear
Outcomes	Primary outcome: otorrhoea (blood/mucopus) and tube patency (not assessed via tympanometry) Assessed at 2 weeks postoperatively Assessed by: not stated Assessment technique: (1) unclear (2) ventilation tube patency Secondary outcomes not reported
Notes	Exclusion criteria not stated
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote (from report): "The side treated with drops was allocated by one of the nurses choosing blindly a card from a pack of cards in two colours and the other ear left as a control." (pg. 188). Possibly open to selection bias.
Allocation concealment (selection bias)	Unclear risk	Not enough information to permit a judgement of 'yes' or 'no'
Incomplete outcome data (attrition bias) All outcomes	Low risk	14 children were excluded from the study because of failure to follow‐up
Blinding of participants?	High risk	Not possible due to postoperative treatment group
Blinding of assessor?	Unclear risk	Insufficient information to permit a judgement of 'yes' or 'no'
Free of other bias?	Unclear risk	Exclusion criteria not reported

Scott 1992.

Methods	Randomisation type: by child Randomisation method: unclear   Blinding: not stated Design: test versus test versus no treatment
Participants	Entered: 53 children Analysed: 52 children Age range (entered): 9 months to 11 years Age range (analysed): unclear (mean 3.5 years) Sex (entered): not stated Sex (analysed): 33 male and 19 females Inclusion criteria: COME and RAOM with/without concurrent adenoidectomy
Interventions	1. Gentamicin otic drops following tube placement 2. Gentamicin otic drops intra‐operatively and for 2 days postoperatively (4 drops in each ear, 3 times a day) 3. No treatment
Outcomes	Primary outcome: otorrhoea (unclear from one or both ears) Assessed at 2 weeks (physician contacted if otorrhoea had developed earlier) Assessed by otolaryngology resident Assessment technique: otoscopy, tube patency, audiometry and tympanostomy Secondary outcomes: tympanometry assessment for patency of the ventilation tube (noted in the report but no outcome data reported)
Notes	Exclusion criteria: children with purulent ears
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote (from report): "Subjects were randomized at the time of surgery into one of three groups." (pg. 34). No further information reported.
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit a judgement of 'yes' or 'no'
Incomplete outcome data (attrition bias) All outcomes	Low risk	1 patient data lost and accounted for in the report
Blinding of participants?	High risk	Not possible
Blinding of assessor?	Unclear risk	Insufficient information to permit a judgement of 'yes' or 'no'
Free of other bias?	High risk	The trial included a small sample size. Randomisation did not meet with equal number of children with preoperative diagnosis in each group, and it was unclear if the outcomes were reported from one or both ears per child.

Shinkwin 1996.

Methods	Randomisation type: by ear Randomisation method: using sealed envelopes Blinding: unclear Design: test versus placebo
Participants	Entered: 193 children (386 ears) Analysed: 161 children (322 ears) Age range (entered): not stated Age range (analysed): 1 to 11.5 years (mean = 6 years) Sex (entered): unclear Sex (analysed): 83 boys (52%) and 78 girls (48%) Inclusion criteria: symptomatic persistent middle ear effusions and RAOM
Interventions	1. Gentamicin/hydrocortisone (Gentisone HC) drops (4 drops) into the ear intra‐operatively 2. Normal saline drops (4 drops) intra‐operatively in other ear (taken as no treatment)
Outcomes	Primary outcome: otorrhoea Assessed at 2 weeks postoperatively (or earlier if needed) Assessment made by: not stated Assessment technique: otoscopy Secondary outcomes not reported
Notes	Exclusion criteria: children with pre‐existing medical conditions that predisposed them to otorrhoea
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote (from report): "The side determined randomly and blindly by an independent observer using sealed envelopes." (pg. 531)
Allocation concealment (selection bias)	Unclear risk	Insufficient information to permit a judgement
Incomplete outcome data (attrition bias) All outcomes	Low risk	32 children lost to follow‐up
Blinding of participants?	Low risk	Corresponding author confirmed that the participants were blinded to the treatment groups
Blinding of assessor?	Low risk	Corresponding author confirmed that the assessor was blinded to which ear had been treated
Free of other bias?	Unclear risk	Children with pre‐existing medical conditions that predisposed them to otorrhoea were excluded

Welling 1995.

Methods	Randomisation type: by ear Randomisation method: unclear Blinding: not stated Design: test versus no treatment
Participants	Entered: 60 children Analysed: 50 children Age range (entered): 4 to 17 years Age range (analysed): not stated (M = 7.6 years) Sex (entered): not stated Sex (analysed): 22 girls, 28 boys Inclusion criteria: RAOM or chronic serous otitis media
Interventions	1. Hydrocortisone/neomycin/polymyxin B (Cortisporin) otic suspension (0. 5 ml) middle ear instillation intra‐operatively 2. No treatment
Outcomes	Primary outcome: otorrhoea and hearing level Assessed at 4 weeks postoperatively Assessment made by: not stated Assessment technique: not stated Secondary outcomes not reported
Notes	Exclusion criteria: children with a sensitivity to Cortisporin; children unable to complete study; loss to follow‐up
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote (from correspondent) when contacted regarding sequence generation: "...random number generator"
Allocation concealment (selection bias)	Low risk	Quote (from correspondent) when contacted regarding allocation concealment: "...it was."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Accounted for 8 patients lost to follow‐up; 2 children need further treatment
Blinding of participants?	Low risk	Quote (from correspondent) when contacted regarding the blinding of the parents/children to the treatment groups: "They were blinded to the side."
Blinding of assessor?	Low risk	Quote (from correspondent): "The audiologist who did the hearing test was blinded." Due to the design of the study the assessor of the outcome (otorrhoea) would have been blinded.
Free of other bias?	Low risk	Noted that prior to the start of this trial, 3 children were receiving oral antibiotics; these patients were left in the study. The age range of the participants was from 4 to 17 years.

Zacharia 1993.

Methods	Randomisation type: by ear Randomisation method: sequence generated by odd or even date of birth Blinding: not stated Design: test versus no treatment
Participants	Entered: 125 children   Analysed: 122 children Age range (entered): not stated Age range (analysed): 4 months to 13 years (median = 4.8 years) Sex (entered): not stated Sex (analysed): 76 males and 46 females Inclusion criteria: OME or RAOM with and without adenoidectomy or adeno‐tonsillectomy
Interventions	1. Gentamicin and hydrocortisone combination (5 drops into the external ear followed by tragal massage, intra‐operatively) 2. No treatment
Outcomes	Primary outcomes: otorrhoea and tube patency Assessed at 2 weeks postoperatively Assessment made by: one of the authors Assessment technique: not stated Secondary outcomes not reported
Notes	Exclusion criteria not stated
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quote (from report): "Patients were randomized so that those with an odd final digit of the date of birth e.g. (1983) had drops inserted into the right ear and those with an even digit into the left." (pg. 239)
Allocation concealment (selection bias)	Unclear risk	Not enough information to permit a judgement of 'yes' or 'no'
Incomplete outcome data (attrition bias) All outcomes	High risk	122 children completed the trial out of 125 children
Blinding of participants?	Unclear risk	Not enough information to permit a judgement of 'yes' or 'no'
Blinding of assessor?	Unclear risk	Not enough information to permit a judgement of 'yes' or 'no'
Free of other bias?	Unclear risk	Exclusion criteria not stated

Zipfel 1999.

Methods	Randomisation type: by ear Randomisation method: computer‐generated list Blinding: single‐blind (participants) Design: test versus no treatment
Participants	Entered: 183 children Analysed: 154 children (308 ears) Age range (entered): 6 months to 14 years Age range (analysed): 6 months to 14 years Sex (entered): not stated Sex (analysed): 101 boys, 53 girls Inclusion criteria: insertion of ventilation tubes for chronic OME recurrent AOM and atelectatic OM with/without concomitant adenoidectomy or adeno‐tonsillectomy and tonsillectomy
Interventions	1. Ciprofloxacin, placed in the middle and external ear after surgery (3 drops pumped 4 times to ensure good penetration) 2. No treatment in other ear
Outcomes	Primary outcome: otorrhoea (from both ears) Assessed at 24 hours to 2 weeks postoperatively Assessed by: not stated Assessment technique: otologic examination Secondary outcome not reported
Notes	Exclusion criteria: children with unilateral disease, craniofacial anomalies and any other chronic illness of known immune deficiencies
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote (from report): "Randomisation was accomplished using a computer generated list of random assignments." (pg. 417)
Allocation concealment (selection bias)	Low risk	Allocation adequately concealed due to blinding of participants
Incomplete outcome data (attrition bias) All outcomes	Low risk	All missing data accounted for and missing outcomes balanced across intervention groups
Blinding of participants?	Low risk	Quote (from report): "Patients and their parents, but not the investigators, were blinded to the randomisation." (pg. 417)
Blinding of assessor?	High risk	See above
Free of other bias?	Unclear risk	Children with craniofacial abnormalities were excluded from the trial

AOM: acute otitis media 
 COME: chronic otitis media with effusion 
 IQR: interquartile range 
 MEE: middle ear effusion 
 OM: otitis media 
 OME: otitis media with effusion 
 RAOM: recurrent acute otitis media 
 TMP/SMX: trimethoprim/sulfamethoxazole 
 TT: tympanostomy tube 
 TTO: tympanostomy tube otorrhoea

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Ah‐Tye 2001	ALLOCATION Not a randomised controlled trial
Baker 1988	ALLOCATION Randomised controlled trial PARTICIPANTS Included children and adults undergoing VT insertion INTERVENTION Gentamicin versus control OUTCOMES Otorrhoea: unable to separate children and adult outcome data
Balkany 1983	ALLOCATION Randomised controlled trial PARTICIPANTS Included children and adults undergoing VT insertion INTERVENTION Topical antibiotic/steroids versus oral ampicillin versus no treatment control OUTCOMES Otorrhoea: unable to separate adults and children outcome data
Balkany 1986	ALLOCATION Randomised controlled trial PARTICIPANTS Included children and adults undergoing VT insertion INTERVENTIONS Irrigation of the middle ear OUTCOMES Otorrhoea: unable to separate children and adults outcome data
Coates 1990	ALLOCATION Not randomised
Force 1995	ALLOCATION Not randomised
Giebink 1992	ALLOCATION Not randomised
Golz 1998	ALLOCATION Retrospective study
Granath 2008	ALLOCATION Randomised controlled trial PARTICIPANTS Included children enrolled with tubes in situ
Gross 2000	ALLOCATION Randomisation failure ‐ part studies were allocated systematically
Harlock 1992	ALLOCATION Although the order as to which ear was operated on was randomised, the treated ear was chosen systematically (right ear)
Heslop 2010	ALLOCATION Randomised controlled trial, double‐blind PARTICIPANTS Included children with tubes in situ
Hester 1995	ALLOCATION Randomised controlled trial, double‐blind PARTICIPANTS Included children and adults undergoing VT insertion INTERVENTIONS Polymyxin B/neomycin/hydrocortisone drops OUTCOMES Otorrhoea: unable to separate children and adult outcome data
Pearson 1996	ALLOCATION Randomised controlled trial, double‐blind PARTICIPANTS Included children and adults undergoing VT insertion INTERVENTIONS Sofradex (dexamethasone/framycetin/gramicidin) drops versus control OUTCOME Only functioning of the ventilation tube data were reported
Roland 2003	ALLOCATION Randomised controlled trial PARTICIPANTS Included children with tubes in situ
Ruohola 2003	ALLOCATION Randomised controlled trial PARTICIPANTS Included children with tubes in situ
Spraggs 1995	ALLOCATION Randomised controlled trial PARTICIPANTS Included children with tubes in situ

VT: ventilation tube

Differences between protocol and review

• The text of the Background has been redrafted.

• To improve the clarity and flow of the text the order of intervention types has been changed.

• In the Criteria for considering studies for this review we have increased the maximum age of included children from 16 years to 17 years.

• To the Types of outcome measures we have added the time at which the assessment was made.

• The text of the Primary outcomes has been simplified from "The objective visualisation (using otoscopy) of purulent otorrhoea from one or both ears postoperatively" to "the otoscopic observation of postoperative otorrhoea from one or both ears".

• We have added cost‐effectiveness as a secondary outcome, as one trial reported it.

• We computed risk ratios instead of odds ratios. We have also computed the number needed to treat (NNTB).

• In the Effects of interventions section we have added a new subsection named Predictor factors for otorrhoea and potential benefit.

Contributions of authors

Akeroyd MA: drafting the protocol, interpretation of data, drafting and editing the final review.

Browning GG: design concept, drafting the protocol, interpretation of data, editing the final review.

Syed MI: design concept, bibliographic searches, data collection for the review, drafting the protocol, extracting data from studies, writing to authors, 'Risk of bias' assessment, analysis of data, interpretation of data, drafting the final review.

Suller SL: bibliographic searches, data collection for the review, drafting the protocol, extracting data from studies, 'Risk of bias' assessment, analysis of data, interpretation of data, drafting the final review.

Sources of support

Internal sources

• Medical Research Council, London, UK.

  Intramural funding to MRC Institute of Hearing Research (grant number: U135097131)

• Chief Scientist Office, Edinburgh, UK.

  Intramural funding to MRC Institute of Hearing Research

External sources

• No sources of support supplied

Declarations of interest

No conflicts of interest are known.

New