The safety and effectiveness of different methods of ear wax removal: a systematic review and economic evaluation

Article history

The research reported in this issue of the journal was commissioned by the HTA programme as project number 06/77/04. The contractual start date was in April 2008. The draft report began editorial review in April 2009 and was accepted for publication in October 2009. As the funder, by devising a commissioning brief, the HTA programme specified the research question and study design. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the referees for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

None

Copyright statement

© 2010 Queen’s Printer and Controller of HMSO. This journal is a member of and subscribes to the principles of the Committee on Publication Ethics (COPE) (http://www.publicationethics.org/). This journal may be freely reproduced for the purposes of private research and study and may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NETSCC, Health Technology Assessment, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2010 Queen’s Printer and Controller of HMSO

Aim

The project will evaluate the clinical effectiveness and cost-effectiveness of the different methods for the removal earwax in adults and children. It will review systematically the evidence assessing the clinical effectiveness and cost-effectiveness of interventions that are currently available for softening and/or removing earwax, including the use of drops, irrigation, mechanical removal and other methods. The project will systematically search for, appraise and summarise evidence on the safety of the different methods, identifying any significant harms or adverse events (AEs). If the systematic review of cost-effectiveness shows that there are no appropriate good-quality economic evaluations, a new economic model relevant to the UK setting will be developed. Also, it will identify any research needs and use value-of-information approaches to help in prioritising them.

Description of the health problem

Earwax (cerumen) is a normal secretion in the external ear canal, produced by small glands in the ear. The purpose of earwax is generally thought to be to protect the ear by trapping any particles in the ear canal and preventing them from entering the deeper part of the ear. Particles may include dirt, dead skin or other fragments. Normally, earwax moves particles to the outer ear at a rate that prevents any significant build-up. When this process fails, there can be an excessive build-up of wax, which can block or occlude the auditory canal. 1,2 Although this can be a relatively minor problem, it can result in several other related problems, including hearing loss, discomfort, balance disorders, tinnitus and even infection. 3,4 It is often these symptomatic conditions that are the key concern for the person suffering from excessive earwax.

Although people with an excessive build-up of earwax can experience any of these conditions, hearing loss and the associated discomfort are probably the most frequent that occur. 5 The effects on hearing can be significant. Severity of hearing loss is measured by how well an individual can hear the frequencies or intensities of sound [measured by decibel (dB)] most often associated with speech. A person who can hear within the normal range can perceive sounds at a threshold intensity as low as 20 dB. Blockage of the ear from wax may elevate the hearing threshold to 40–45 dB. 6,7 In those presenting with age-related hearing loss, some will have earwax that, if removed, can reduce the hearing loss by around 10 dB. 8 The sense of blockage of the ear, the effects of hearing loss and the other comorbidities can cause discomfort and irritation to the person.

The occurrence of these related comorbidities provide evidence of the extent and severity of the build-up of earwax. Often people will present prior to the development of severe symptoms. In such instances, assessment of their condition is usually made based on the degree of occlusion of the auditory canal or tympanic membrane (TM), the extent of impaction of earwax, and the nature and characteristics of the earwax itself. These are often thought to be proportionally related to the severity of comorbidities. 7 Assessment of these symptoms is usually undertaken through direct visualisation of the ear canal with an otoscope. 9 There do not appear to be any documented standard criteria on which to base such judgements, meaning that they are open to variation in their measurement, reporting and interpretation. Such differences make judging the extent and severity of the earwax uncertain.

Current service provision and description of interventions

Despite the problems associated with the accumulation of excessive earwax and the demands placed upon primary care within the UK National Health Service (NHS), it appears that there are no nationally agreed guidelines for its diagnosis and treatment. As a consequence, it is likely that services will vary with many based primarily on local custom and practice rather than a strong clinical evidence base. Some locally and internationally based guidelines have been developed28 and these, along with opinions from clinicians, including practice nurses, provide a basis for understanding the clinical pathway that may be followed by people who are suffering from the problems associated with excessive earwax. Problems that may lead a person to seek help include a feeling of a blockage, discomfort, hearing loss in one or both ears, tinnitus and/or dizziness.

In the UK, people requiring the removal of earwax have traditionally attended primary care practices for confirmation of the diagnosis and treatment. The method of treatment should take account of the severity of the condition, the possibility of any contraindications (e.g. perforations) or comorbidities (e.g. tinnitus), the skills of the practitioner and the setting for treatment. The majority of practitioners currently advise the use of some form of drops or softeners as a first stage. The British National Formulary (BNF) lists several preparations, including almond oil, olive oil, sodium bicarbonate drops, Cerumol^®, Exterol^®, Molcer^®, Otex^® and Waxsol^®. Other softeners may also be used. A summary of different preparations is presented in Table 1, grouping them into water-based, oil-based or non-water-non-oil-based products, using the classification adopted by Hand and Harvey. 29 Not all of these preparations are currently available in the UK. Brand names will be used throughout the report unless those are unclear, when the generic name will be reported.

The intention of these remedies or drops is to either soften the wax prior to removal at the clinic or to help remove the wax on its own. Their specific action varies. For example, in vitro studies suggest that preparations including urea or glycerine increase water penetration of the earwax, while preparations including peroxide break up the earwax through the release of gas/bubbles, therefore aiding mechanical removal. 30 Limited data are available to guide people’s choice of drops in particular clinical situations. AEs differ for each intervention but are generally thought to be mild. It is usually suggested that people allow between 3 and 7 days for these remedies or drops to take effect, although for some it may take longer (i.e. 14 days of drops) or several cycles.

If the wax has not dissolved or dissipated using drops or remedies, people are recommended to attend the primary care practice to have the wax removed by mechanical removal, through either irrigation or curettage. 11 In irrigation, a pressurised flow of water is used to remove the earwax. Although flushing wax with metal piston syringes (e.g. Reiner-Alexander ear syringe) was common practice in primary care practices, these have largely been replaced with electronic irrigators, such as the oral jet irrigator, nebuliser or Propulse ear irrigator. 31,32 The use of metal piston syringes is no longer recommended. 33 Irrigation is contraindicated in people with perforated ear drums, history of ear surgery or chronic ear conditions. Reported harms of irrigation are pain, infection and injury to the ear, including TM perforation and tinnitus. 34,35 Curettage, which allows the removal of earwax under direct vision, using various implements, such as cerumen spoons, hooks, loops and probes, is rarely undertaken in primary care practices. Although it has the advantage of not using water to remove the earwax, and so perhaps lessening the risk of infection, it can be a difficult procedure, requiring specialist skill and time. 36

Increasingly the role of the general practitioner (GP) has been taken over by the practice nurse (or, for the house-bound, the district nurse), who can confirm the problem by examination of the ear, recommending/prescribing the use of the drops or remedies and then removing any wax by irrigation. 31 On most occasions irrigation will successfully remove earwax on the first attempt; however, for a limited proportion of people it may prove more difficult and these people may need to attend on several occasions. Rarely, when it proves impossible for the primary care practice to remove the earwax, the person will be referred to a hospital-based specialist.

With the availability of several different softeners and proprietary drops, some people decide to self treat. As with their use in primary care practices, drops provide the possibility of treating the wax to allow it to dissipate or dissolve without further treatment. While such preparations offer the opportunity for self-treatment, caution needs to be exercised when used in combination with cotton wool swabs, as inadvertent damage to the inner ear canal or TM can occur. People may make several attempts to clear the wax with drops before consulting their primary care practice. Having self-treated with drops, people subsequently consulting the primary care practice for treatment may be able to have their earwax removed without further delay. Although not currently recommended, and not widely available within the UK, some forms of syringes for self-treatment are available. Soft bulb irrigators can be used by people after drops to flush their own ears. These can be purchased from suppliers through the internet and over the counter (OTC) in some European countries and in the USA. Other syringes for self-treatment are available, including plastic piston syringes (e.g. The Real McCoy and Master Blaster).

A minority of people who are unable to have their ears cleared through self-care or at the primary care practice, or have particular clinical conditions (e.g. cholesteatoma), or contraindications to standard treatment (e.g. pre-existing perforations of the TM) may be referred to specialist care. Ear, nose and throat (ENT) or otolaryngology departments use techniques such as microsuction or curettage through direct microscopic or endoscopic vision to clear earwax. 33,37,38 These methods are used in combination with suction or the use of a Jobson-Horne probe and a St Bartholomew’s wax hook, or crocodile forceps. 38 This requires specialist equipment and considerable skill. 33,39 In very rare instances where clearance of earwax is prevented due to a narrowing of the external auditory canal, for example, surgery may be required. 11

In addition to the above technologies, complementary therapies can also be used (e.g. ear candling).

Methods for reviewing effectiveness

The a priori methods for systematically reviewing the evidence of clinical effectiveness and cost-effectiveness are described in the research protocol (Appendix 1), which was sent to experts for comment. Although helpful comments were received relating to the general content of the research protocol, there was none that identified specific problems with the methods of the review. The methods outlined in the protocol are briefly summarised below. Methods for the economic evaluation are outlined in Chapter 5 (see Methods of the economic valuation).

Quantity and quality of research available

A total of 202 records of publications were identified through literature searching. Of these, 158 were excluded on title and abstract. Full reports for the remaining 45 were requested for more in-depth screening. Of these, 19 were excluded (see list of excluded studies in Appendix 6). An inclusion flow chart can be seen in Appendix 2.

Twenty-six published studies met the inclusion criteria – of these, 22 were RCTs and four were CCTs. These trials fell into three categories: studies undertaken in primary care (14 studies41–54), studies undertaken in secondary care (eight studies4,54–60), and studies of self-care or those undertaken in other care settings (four studies61–64). Where the setting was in emergency care this has been grouped under primary care in the present review because it was assumed that this was the first point of contact with a health professional. In some other cases the setting was not made explicit within the publication; where this was the case an assumed grouping was used based on the author(s) affiliation (e.g. if they worked in an outpatient clinic this was assumed to be a secondary care setting). Additionally, studies are also grouped by participant group (children, mixed adults and children, and adult groups). In some cases the study did not indicate who the target population was and these were classed under the mixed group as population ‘unknown’. In some of these studies a mean age of the study participants was given but no range or measure of variance was reported and so these were also classed under the mixed population group. The subsequent discussion of these studies (below) will follow these conventions (Table 2).

There is very little consistency among the included studies, which makes it difficult to fully summarise the results and in many studies some basic data were not available. Across all of these studies there are variations in the characteristics of the participants recruited, in terms of age and, to some extent, gender, and in terms of the extent of the earwax problem. In many studies there is very limited discussion of baseline characteristics, which makes it difficult to establish the representativeness of the respective populations. Many of the included studies (eight studies4,42,43,46,51,55,57,61) were presented as either a short paper format (less than two sides), an abstract format or a conference proceeding, and, where this was the case, the available data were further limited.

Study sample sizes were typically small, varying from 36 participants45 to 237 participants,63 and only a few studies reported undertaking a sample size calculation (7 out of 24 studies45,47,49,52,53,56,63). In some studies the overall population sample size was not reported, rather the numbers of ears were presented. 42,43,60 Seventeen studies were two-arm comparisons,41–45,48–54,56,58,59,63,64 five were three-arm studies,46,47,55,57,62 and there was one study each with four,4 five61 and six60 intervention arms.

There were also a wide range of interventions used between the included studies and differences in the length of follow-up used. For ease of understanding the review that follows is also divided into studies of immediate follow-up and delayed follow-up (the exact length of follow-up will be discussed for each study individually) and into studies where the intention was to use a softening agent alone (even when subsequent irrigation was used, if this was after initial outcome assessment) and those where the intention was always to use a softening agent and an irrigation as the intervention. Owing to the wide variation in interventions and comparisons used in the included studies the review has not been split by the different comparisons. However, each comparison is reported in sequence, in line with the conventions discussed above (setting, population, follow-up, softening agent with or without irrigation).

The outcome measures used also varied across the included studies. Often information concerning the definitions of the outcome measures used was limited and in many cases it was not possible to assess how valid, objective or consistently applied these different measures were. In the instances where this was clear the review has made note of this; for all other studies it should be assumed that care is required in the interpretation of the outcomes. For the purpose of this review, the range of outcomes have been categorised as far as possible into those measuring the degree of occlusion; those measuring the ease of wax removal; those measuring participant satisfaction; those measuring recurrence of earwax; and AEs. Measures of the type and consistency of earwax removed were tabulated but not reported in the narrative synthesis. In many cases the baseline values were not presented for a particular outcome, which makes the interpretation of the evidence of the effectiveness of the interventions more difficult. Where pretreatment and post-treatment measurements of an outcome were reported this has been discussed (e.g. a study might report the varying degrees of occlusion before and after intervention). However, in many cases it is not clear how the final outcome is related to the finding at baseline (how many in the ‘no change category’ post intervention started in the fully occluded category pretreatment, how many were in the partially occluded and so on). Many of the studies also did not report measures of variance around the estimates presented.

A number of studies also did not report results of any statistical significance testing, and, of those that did, a number did not report the statistical analytical approach taken. As noted above, in some cases the allocation to interventions was undertaken on the basis of the number of ears rather than the number of participants. Also, in some studies where allocation was based on the number of participants, the analysis was undertaken on the number of ears. Where this is the case the present review has identified these studies. Care is recommended when interpreting the outcomes of these studies, as it is unclear whether valid statistical analyses (where reported) will have been undertaken (see Cochrane Handbookfor Systematic Reviewsof Interventions for discussion of appropriate analyses of these types of data65). Finally, there was a wide range in the publication dates of the included studies (1950–2007), and eleven studies were undertaken more than 20 years ago,41,43,48,50,51,54,57,59–61,64 which may affect the generalisability of the studies to current practice (and account for some of the other issues already raised above). It is likely in current practice that an irrigator rather than a metal syringe will be used for the removal of earwax; however, syringes were used for many of the included studies, although this may not always be clear because ‘syringing’ and ‘irrigation’ may sometimes be used interchangeably. The present report refers to ‘syringing’ only where this was the term reported by the included study. In the case of self-treatment, the present report uses the term ‘irrigation’ when a soft bulb was used, as these are not syringes, even if the study authors have referred to this as a self-syringe.

Studies in primary care settings

Characteristics of the primary care studies are shown in Table 3. Seven RCTs41,42,46–49,51 followed participants up immediately after interventions, and seven RCTs43–45,50,52–54 had a delay between intervention and follow-up.

Two RCTs46,47 compared docusate sodium (DS) versus triethanolamine polypeptide (TP) versus saline in children, while one RCT compared DS versus TP in a mixed population of adults and children. 49 Another trial51 compared TP versus carbamide peroxide in an unspecified population. All of these had immediate follow-up after treatment.

Of two RCTs44,50 comparing Cerumol, Dummer and colleagues44 compared Cerumol with Audax in an adult population. This RCT had a delayed follow-up, averaging 4 days between the first and second visit (range 3–7 days). The second study by Jaffe and Grimshaw50 compared Cerumol with Otocerol in a mixed population of adults and children. This study had a delayed follow-up, but the length was unspecified, with participants requested to revisit after three instillations. Cerumol was further compared with Waxsol in an RCT48 with a mixed population of adults and children. This RCT had an immediate follow-up after treatment. Another comparison, of Audax with Earex,45 was an RCT in a population of adults aged 16 years and above, and had a delayed follow-up of 5 days. A CCT by Fahmy and Whitefield54 compared Exterol with Cerumol in an unspecified population and had a delayed follow-up of 1 week. The trial consisted of three studies, but only study three was in a primary care setting (for study one and study two see Studies in secondary care settings, below).

One RCT52 compared aqueous sodium bicarbonate with aqueous acetic acid in a mixed population of children and adults. This trial had a delayed follow-up of 14 days. An RCT by the General Practitioner Research Group,41 as well as a CCT by Burgess,43 compared dioctyl-medo ear drops with oil in adults. These studies had an immediate follow-up and delayed follow-up of between 2 and 7 days after the intervention, respectively.

Eekhof and colleagues53 compared the instillation of water at body temperature with a group self-administering oil. The population for this RCT was unspecified, but had a mean age of 51 years for all participants. Follow-up was immediate after the water treatment, but delayed by 3 days for the oil arm of the study. There was only one RCT42 comparing either ‘wet’ (with prior instillation of warm tap water) syringing or ‘dry’ syringing (no prior instillation of water) in an adult population. Follow-up was immediately after treatment.

The methodology and quality of reporting of included studies was generally poor (Table 4). A number of studies pre-date RCT reporting guidelines, but this may also reflect set word limits for some of the publications. Only three RCTs47,49,52 were assessed as adequate for their randomisation procedure, with one assessed as partially meeting this criteria. 42 For several studies the method was judged inadequate,46,51,53 but for the majority41,44,45,48,50 it was not possible to judge due to a distinct lack of information. The same three studies47,49,52 that were judged adequate for their randomisation procedures were also judged as adequate in their concealment allocation. For the remaining RCTs, concealment of allocation was either inadequate42,53 or unknown,41,44–46,48,50,51 making it impossible to rule out selection bias.

None of the studies were judged as adequate in their method for blinding caregivers or participants. While the difficulties in blinding for some of the interventions could have been addressed somewhat by blinding the assessor of the outcomes, none of the studies did so adequately. Measurement bias can therefore not be ruled out for the majority of included studies. With only 4 out of 14 studies reporting baseline characteristics,42,47,49,50 it is also not possible to judge the similarities between the treatment groups at baseline for the majority of the studies. Four studies were judged adequate for the description of missing values,45,47,49,52 but none of the studies reported an adequate intention-to-treat (ITT) analysis. Caution should be exercised in the interpretation of the results of these studies, as overall these appear to be at a high risk of bias, which may affect the direction of any effects shown.

Likewise, the CCTs (Table 5) appear at risk to bias and their results have to be interpreted with care. The two included studies43,54 were of low quality, with unreported baseline characteristics in both and blinding of outcome assessors adequate in only one study. 43 Neither study was judged to be adequate for eligibility criteria, primary outcome results, ITT analysis, missing values or representativeness of the populations.

Studies in secondary care settings

Eight studies undertaken in secondary care settings were identified (Table 7): seven were RCTs and one was a CCT. Three RCTs55,57,59 followed participants up immediately after interventions. One57 was a three-arm trial in children, of hydrogen peroxide, mineral oil or TP; one other55 was a three-arm trial in adults, of chlorobutanol, sodium carbonate and saline. The third study59 compared TP with olive oil; the participant group was not described in this study. Three included studies had a longer delay in their follow-up. Keane and colleagues’4 RCT, which appears to be in elderly adults, was a four-arm trial of Cerumol, sodium bicarbonate, sterile water and no treatment (control). The interventions were administered for 5 days, after which the participants were reassessed. Fraser60 assessed sodium bicarbonate, olive oil, Cerumol, Waxsol, TP and dioctyl to a within-participant control (one ear test, one ear control) in elderly participants over a period of 3 days follow-up. The third study, a CCT by Fahmy and Whitefield,54 included three separate studies, one of which was in participants from primary care and was discussed previously (Chapter 3, Studies in secondary care settings). Of the other two studies, one compared Exterol with glycerol and the other Exterol with Cerumol. The target population for these studies was not clear in the publication, but the ear drops were given over a one-week period before follow-up. In addition to the studies comparing the use of different softening agents, two other studies were included that either compared two de-waxing techniques56 or the application of a skin oil with no treatment. 58 In the first of these RCTs adults in an otolaryngology clinic requiring removal of earwax were randomised to endoscopic otoendoscopic de-waxing or microscopic de-waxing. In the second RCT, participants had their earwax removed and were then randomised to either Ceridal lipolotion skin oil or no treatment to assess the recurrence of cerumen. This study was undertaken in adults and children with learning difficulties and follow-up was for 12 months.

The quality of reporting and methodology of the included studies was generally poor (see Tables 8 and 9). Of the seven RCTs, none of the studies was assessed as having an adequate generation of the randomisation sequence. In one RCT the generation of the randomisation schedule was judged to be partially met,56 in three this was judged to be inadequate,57,58,60 and in the remainder it was not possible to judge due to a lack of information. 4,55,59 Only one RCT adequately met the criteria for concealment of allocation. 59 Therefore, most of these studies may be subject to selection bias, with the allocation sequence open to possible manipulation. Double-blinding (participant and caregiver) was assessed as being adequate only in one study. 59 While double-blinding may have been difficult for some interventions, blinding of the outcome assessors could have been undertaken; however, there were no studies that were assessed as adequately blinding the outcome assessors. This factor is particularly important when subjective outcome measures are used and therefore this may mean the studies are subject to measurement bias. Many studies did not present baseline characteristics of the participants within their intervention arms so it is unclear how similar the groups were at baseline. The description of withdrawals were assessed as being adequate in four studies,4,56–58 but only one of these studies also described an adequate ITT analysis. 56 Overall these studies are at a high risk of bias, which may affect the direction of any effects shown, and care should be taken in the interpretation of the results.

The CCT54 was similarly judged to be only of low quality, the baseline characteristics between the study arms were unknown, the blinding of outcome assessors was unclear, and the report does not refer to withdrawals and dropouts or any ITT analysis. The results of this study should also be interpreted cautiously due to the high risk of bias.

Studies of self-care

Two RCTs63,64 investigated the use of self-irrigation after participants had been assessed for earwax in primary care. One63 compared self-irrigation to nurse irrigation in adult participants who had been given sodium bicarbonate to use for at least 2 days. The follow-up was 1–2 weeks later for the comparison of interventions. A later follow-up (within 6 weeks) was undertaken to assess the need for further treatment and for the assessment of AEs and adherence to treatment. The other study64 randomised participants to either TP overnight and self-irrigation in the morning, or just self-irrigation in the morning. Both studies used soft bulb irrigators and followed up participants subsequent to the irrigation attempts. Table 11 shows the characteristics of the two self-care studies.

The quality of the included RCT of self-irrigation63 (Table 12) compared with nurse irrigation was assessed as having reasonable quality. The method of generating the randomisation sequence and the concealment of allocation were rated as being adequate. These factors limit the risk from selection bias. Blinding of the participants and care-providers was not possible due to the nature of the intervention; however, the outcome assessors could have been blinded to the treatment allocation and this was therefore judged as inadequate. Withdrawals were reported but there appeared to be additional numbers missing in the analysis of many of the outcomes and there was no discussion of the application of an ITT principle to the analysis of data. The RCT of self-irrigation with or without a previous softener64 was assessed to be of low methodological quality. Details of the randomisation schedule and allocation were not adequately described and it is also unclear whether the outcome assessors were blinded to treatment allocation.

Studies in other care settings

One RCT62 was undertaken in a corporate research clinic in adult volunteers with excessive or impacted cerumen on examination (Table 15). The study compared TP, Murine and placebo with saline, and follow-up was immediate. One CCT61 was undertaken in a military research unit in a population of adult entrants to the Royal Air Force who had impacted cerumen (Table 15). The study compared sodium bicarbonate, Cerumol, hydrogen peroxide, olive oil and ‘no treatment’ control. Follow-up was also immediate. Both studies also followed the softening agents with an irrigation procedure before assessing the outcomes.

The quality of reporting and methodology of the included studies was generally poor (Tables 16 and 17). In the RCT the randomisation schedule and the concealment of allocation were unknown, which may lead to an increased risk of selection bias. Blinding of outcome assessors and the participant were judged to be adequate, with blinding of the caregiver judged as partially met. The number of withdrawals was discussed, but there was no discussion on an analysis based on the ITT principle. The CCT was judged to meet the criteria only partially for blinding of outcome assessors, but the reporting of outcomes and any withdrawals values was inadequate. These studies are at a high risk of bias, which may affect the direction of any effects shown. Caution is therefore required when interpreting the results.

Summary of results of the systematic review of clinical effectiveness

Research in progress

We identified a total of five ongoing trials from our searches (Table 21). Schroeder and colleagues aimed to evaluate olive oil as a softening agent, comparing the usual care of self-administered olive oil for 5 days prior to irrigation versus 3 weeks. The primary outcome of the trial was ear canal clearance and the trial was set to finish in 2006.

Gillett and colleagues aimed to investigate the differences in discomfort and/or complications between microsuction and the use of instruments. The trial was set to finish in July 2008, although no subsequent publication has been identified at the time of writing.

The third study by Caballero and colleagues was set in a hospital clinic in Spain and aimed to compare Otocerum^® ear drops, Taponoto^® ear drops and placebo. This is reported to be a phase IV trial, non-randomised, double-blind and parallel-assignment trial. The primary outcome was the proportion with a completely visualised TM, with either the treatments alone or with irrigation if needed. The study was expected to finish in October 2008 and no publication has so far been identified.

Two other controlled trials were identified from searches of the National Research Register. A study by MacKeith and colleagues was sited at the Bristol Royal Infirmary and aimed to evaluate wax softening agents used immediately prior to de-waxing by microsuction. The outcome was the impact of softening agents on the pain or discomfort of the procedure. The trial was anticipated to end in June 2007 and the trial status is shown as completed.

The second study, by Hand and the Suffolk and Norfolk Research and Development Consortium (SAND), aimed to compare preparations for the treatment of symptomatic earwax. No further details are provided and the trial was set to end in January 2007.

No publication of the trial data for any of the five studies has been found so far.

Searching

A targeted search of the literature was undertaken to identify additional studies reporting AEs associated with the removal of earwax, to assess the safety of earwax removal and to inform the economic modelling of the problem. The search was not restricted by study design. The following section complements the clinical effectiveness section, reporting overall results from both the RCTs identified, and any relevant observational studies. No systematic critique of the studies was undertaken; however, key methodological limitations of the studies are discussed in the relevant sections below. In addition, only AEs that the study authors attributed to the interventions were included. The studies were assessed on an ITT basis.

Adverse events from drops alone

Four RCTs4,44,50,52 were identified that included AEs from using drops alone with no irrigation. These RCTs are all critically assessed and data extracted in the clinical effectiveness section above (see Chapter 3). Two of these RCTs, by Carr and Smith52 and Keane and colleagues,4 did not find any AEs. The other two RCTs are shown in Table 22. One observational study by Midani and colleagues73 was identified and data is also summarised in Table 22. This was an open-label study of Sofenz cerumenolytic solution in 109 participants with excessive and impacted cerumen. A total of 58 AEs were reported, but only 16 were reported to be directly related to the treatment.

There were major differences in all of the studies in terms of population, drops used, care setting and methodologies used. Therefore, no formal meta-analysis of this data was attempted. However, an estimated probability for individually defined AEs from relevant studies was generated for illustration. An overall probability of an AE occurring was also calculated from the studies that included AEs as outcomes. Caution is required therefore in the interpretation of this data. The ranges of AEs reported for each condition and overall proportion for all AEs from the studies and the one observational study are reported in Table 22.

The overall number of AEs was 43 out of a total study population of 407 participants from the RCTs and observational study. The probability of minor AEs occurring was 10.57%. It is worth noting, however, that the probability of AEs may be variable depending on the type of drops used. The Cerumol arm of the Jaffe and colleagues RCT,74 for example, accounts for 13 AEs, whereas the arm using Otocerol accounts for a further eight AEs out of the total of 43. However, due to the heterogeneity of the studies it is not possible to draw any reliable conclusions about which drops are likely to cause more adverse outcomes than others. Based on data available it would seem that AEs from drops are relatively rare and tend to be of a mild nature.

Adverse events for drops and irrigation

An assumption was made that the reporting of AEs from drops and irrigation would be split into major AEs (those events that would be more likely to require additional treatment or resource use), and minor AEs (those that were likely to be transient and would not require additional treatment).

Major adverse events from drops and irrigation

Sixteen RCTs that included AEs as outcomes were identified. Three of these RCTs did not explicitly state what AEs had occurred and so it was not possible to assess whether the AEs were major or minor in nature, and so they are discussed separately, below. One other study, by Burgess43 was excluded because only the number of ears syringed were reported and it was not possible to tell how many participants were involved in the trial. Therefore, 12 RCTs were included.

In addition, two observational studies were identified. The first study was a retrospective analysis of 2400 patient records from a tinnitus clinic in the USA, who had experienced severe tinnitus. Eleven participants reported that their tinnitus started as a result of cerumen removal. However, it was not clear how many of the 2400 participants with severe tinnitus had actually received ear irrigation, and there was no clear association between the two events, and so the study was excluded. 75 The other study76 was a large prospective observational study (952 ears in 622 subjects) of people attending an ENT clinic in Nigeria, who were using drops to soften earwax and then undergoing irrigation of the ear with a Propulse II irrigation system. The only adverse outcomes reported were one participant with vertigo and another with a TM perforation. No minor AEs were reported. The study was undertaken in a different care setting to the UK, and the majority of participants were children, therefore this may limit the generalisability of the studies results. However, the equipment used and the ‘wet’ phenotype of cerumen is similar to both countries and populations and, as such, the study was included.

As described, the individual RCTs and the one observational study results were used to produce a probability for individually defined AEs. An overall probability of major AEs from drops and irrigation was also generated from all studies that included AEs as outcomes. Caution is recommended in the interpretation of these data. The RCTs were also heterogeneous, with reports from a variety of settings and using several different types of drops and irrigation techniques as active treatments and controls. Furthermore, the external validity of the Nigerian observational study to a UK context requires caution in its interpretation. The ranges of AEs reported for all major AE from the studies are reported in Table 23.

There were nine occurrences of AEs in a total study population of 1515 participants. The overall probability of major adverse outcome was therefore 0.59%. Six of the nine AEs came from the Fraser and colleagues study60 (all otitis externa) involving 124 participants. If this study was excluded then the incidence of AEs would drop to 0.22%. This would be close to the Sharp and colleagues’6 estimation a 0.1% incidence of major complications based on their otolaryngological unit in Scotland. No incidence of tinnitus, otitis media and hearing loss were reported in any of the included studies. It would seem that major AEs occur relatively infrequently, but, given the heterogeneous nature of the studies included, this result should be interpreted with caution.

Minor adverse events from drops and irrigation

The same 12 RCTs that were identified for the major AEs section above were all included in this section. No observational studies were identified reporting rates of minor AEs from the literature search. As mentioned above an estimate of the probabilities of minor AEs from drops and ear irrigation were generated. These need to be treated as illustrative only. The ranges of AEs reported for each condition and overall proportion for all AEs from the studies are reported in Table 24.

There were 128 occurrences of AEs from a total study population of 893. Overall, the incidence of minor AEs was 14.33% from all the included studies. The Coppin and colleagues’ study reported 108 out of the 128 incidences of AEs. 63 The higher incidence of AEs in this study, especially for discomfort and dizziness suggest a more sensitive inclusion of AEs than the other studies. 63 One of the arms in the trial was a self-syringing group and this may also explain the higher number of AEs in the study. If this study was excluded then the probability of minor AEs would fall to 3.05% for the remaining included studies.

Three further RCTs41,48,61 reported an overall rate of AEs but did not define the adverse outcomes that had occurred therefore it was not possible to determine whether they were major or minor AEs. One of these studies, by Hinchcliffe and colleagues, reported only the number of ears treated and as it was not possible to determine how many participants there were in this trial it was excluded. There was a total 257 participants in the two remaining RCTs. 41,48 There were an estimated 21 AEs (the studies reported AEs only as a percentage of each arm with or without AE). This gave an overall probability of 8.17% for the occurrence of AEs. Another study76 in Nigeria using drops to soften earwax, and then irrigation of the ear, reported that there were no minor AEs.

The incidence of minor AEs is relatively common compared to the major AEs, but it is unclear how long many of these AEs last. It is likely, however, by their very nature, that they will be relatively transient and have little or no impact on resource use.

Conclusion

The evidence from the included RCTs and the two observational studies suggest that using both drops and ear irrigation are safe techniques for the removal earwax, with the number of SAEs being low. However, some studies may have been more sensitive at including AE data than others, and in many the length of follow-up was not suitable for capturing AEs. These factors and the methodological issues noted above (and in Chapter 3) suggest that care is required before interpreting earwax removal techniques as safe procedures. The probabilities of more SAEs obtained from this simple aggregation technique were not thought to be reliable enough to put into the economic model. The minor incidence rates were felt unlikely to have a major impact on either health-related quality of life (HRQoL) or resource use and so were also not used.

Systematic review

A systematic review of the literature was undertaken to identify economic evaluations considering the treatment of earwax. The methods for the systematic review are described in Chapter 2. The details of the inclusion and exclusion criteria are shown in Appendix 1 and the search strategies are shown in Appendix 2.

No cost-effectiveness studies were identified that met the inclusion criteria for the review.

Southampton Health Technology Assessments Centre (SHTAC) economic analysis

A modelled economic evaluation was undertaken to estimate the comparative cost-effectiveness of three alternative treatment strategies for earwax removal in adults, without any known contraindications to the use of softeners or irrigation.

Methods of the economic evaluation

Model structure

This section describes in detail the clinical pathways associated with each active treatment alternative introduced in Chapter 5 (SHTAC economic analysis) above. Figure 1 presents the structure of the model.

In the base-case scenario it was assumed that patients in each of the active treatments follow a GP’s advice to apply softeners for 1 week. The use of softeners has two advantages: first, the softeners assist any subsequent irrigation by softening earwax and, second, they facilitate a spontaneous earwax clearance in some patients, after which no irrigation is required. Use of softeners is considered to be the first line of treatment followed by irrigation as the second line of treatment. Finally, if the earwax is still not removed after three rounds of applying softeners and syringing, patients may be referred to secondary care for removal of earwax by a specialist otolaryngologist (OTL). In the first active treatment alternative (current or standard practice), patients presenting at primary care for earwax treatment are initially assessed by a GP (assumption used in the base-case analysis) or a nurse (a scenario analysis); they are advised to use olive oil or sodium bicarbonate for a week and if spontaneous earwax clearance does not occur then return to the practice for irrigation. The irrigation is administered by a practice nurse using an electric irrigator.

FIGURE 1.

The structure of the economic model of earwax treatment alternatives; OLT, otolaryngologist, a specialist doctor providing treatment at secondary care; SAE, serious adverse event.

Alternatively, in the second active treatment option, patients may be advised to apply softeners for a week, and if spontaneous earwax clearance has not occurred they should proceed with self-irrigation using a bulb irrigator, which is available from chemists. Patients for whom self-irrigation does not result in earwax removal return to primary care for a second irrigation. Unlike the first irrigation, the second and third irrigations are administered professionally by a practice nurse using an electric irrigator. The second (and the third, if required) irrigation attempts do not require prior assessment by a GP.

The base-case scenario includes both active treatment alternatives, as well as a conventional ‘no treatment’ comparator.

There are no data to estimate the proportion of patients given each of the treatment options; therefore for the purpose of an economic evaluation the treatment strategies listed above represent the decision choices available for GPs and their patients.

Assumptions used in the model

The economic evaluation does not differentiate between patients who present with one or both ears occluded. Although this is a simplifying assumption, there were no reliable data that would allow differentiation of the outcomes with respect to unilateral or bilateral presentation of earwax.

While there is a non-zero probability of experiencing spontaneous earwax clearance associated with the first application of softeners, it is not the case for severely impacted and hardened earwax. In such instances, spontaneous earwax clearance assisted by softeners is unlikely to occur. Severely occluding and/or hardened earwax may not even be successfully removed at the first irrigation attempt at primary care and will require a second, and sometimes third, round of using softeners followed by irrigation in primary care. In such instances it is assumed that the use of drops merely facilitates the next professionally administered irrigation attempt, and is not associated with the probability of achieving a spontaneous earwax clearance.

If the first irrigation attempt was unsuccessful, the probability of a successful earwax removal at the second and the third irrigation attempt at primary care would not depend on whether the first irrigation was administered by a practice nurse or the patients themselves.

Most patients achieve the desired outcome (earwax removal) without a SAE associated with irrigation. It is assumed that the use of softeners alone as well as the ‘do nothing’ option are not associated with a SAE. In the base-case scenario the probability of a SAE is assumed to be the same regardless of the mode of irrigation (self-irrigation or administered by a practice nurse) and whether irrigation is undertaken for the first, second or the third time.

The published clinical evidence used to populate the model4,63 does not report an incidence of SAEs. However, the occurrence of a SAE was viewed as a clinically important factor, which can also affect the results of an economic evaluation. By including the likelihood of a SAE in the model we have effectively reduced the benefits that occur following successful removal of earwax. 4,63

As there are no published estimates of the incidence of SAEs associated with irrigation, an expert OTL’s opinion was obtained. The probability estimate of having a SAE as a result of irrigation was based on the observed rate of presentation to secondary care for treatment of a TM puncture/rupture or a serious ear infection. The corresponding assumption used in the model is that all the SAE patients are referred and treated at secondary care. However, it is possible that some patients with a TM puncture/rupture or a serious ear infection are monitored and treated by their GPs. If this is the case the estimate used in the model is an underestimate.

The patients who had a SAE associated with irrigation (typically a TM perforation or an infection) experience pain and discomfort. However, following the experts’ advice it was assumed that after 1 week the pain subsides and in 2 weeks most patients experience a spontaneous healing of a TM or resolution of a serious ear infection.

As discussed in Chapter 3, irrigation is a fairly safe procedure, with only a small proportion of patients experiencing minor complications such as superficial erythema and dizziness. Minor complications are assumed to be short-lasting and not significant enough to be associated with a measurable decrease in HRQoL or an additional use of health-care resources. Therefore, the minor complications are not included in the modelled economic evaluation. There is paucity of evidence on the incidence of more serious and possibly lasting complications such as tinnitus. These are not included in the modelled economic evaluation.

Apart from the temporary loss of hearing in the affected ear(s), there are a few other symptoms such as aural fullness, vertigo, itching and tinnitus that affect HRQoL of patients being treated for earwax problems. However, there are no published estimates of utility values associated with symptoms other than loss of hearing. Therefore, the estimated loss of utility for the model is limited to the loss of hearing that was applied to all patients until the earwax is cleared. In addition, utility decrements are applied to patients who experienced a SAE that involves both loss of hearing and pain.

Model type and characteristics

We developed an economic model of treatment alternatives for a single event of earwax removal. A deterministic decision tree (Figure 1) analysis was conducted, focusing on binary outcomes representing earwax that was removed or not removed, with results expressed as incremental cost per patient with earwax safely removed. Analysis also incorporated loss of utility associated with temporary loss of hearing due to occlusion, and for pain and loss of hearing due to a SAE, with results presented as incremental cost per QALY. The base-case analysis was conducted in a population aged 35–44 years, with the corresponding age-adjusted utility value of 0.91 (Kind and colleagues77).

Results of the deterministic decision-analytical model in terms of QALYs were extrapolated to a lifetime horizon (45 years) using an excel spreadsheet. The major assumption in these calculations was that, according to clinical experts, the recurrence rate of earwax is 0.35, which is close to one episode occurring every 3 years or 15 single presentations over the assumed time horizon. The age-specific utility values were used for the estimated effects (QALYs)77 as the cohort, aged 35 years at the baseline, was progressing through the different age categories. Both costs and outcomes were discounted using a 3.5% discounting rate. The calculations also take the UK age-adjusted annual mortality rate into account. Although the assumption of no treatment over this period of time is unlikely to be realistic, the no treatment alternative is included in the analysis for the purpose of maintaining consistency across all stages of an economic analysis. Table 25 presents characteristics of the decision-tree model and the lifetime extrapolation.

TABLE 25 - Characteristics of the modelled economic evaluation

QALY(s), quality-adjusted life-years; SAE, serious adverse event.

Seven weeks is a minimally acceptable time horizon to allow time for up to three presentations at primary care for irrigation, which, including the first week of using softeners, takes 4 weeks, and may be followed by 2 weeks of monitoring a SAE and at least 1 week to include the outcomes of a hospital admission of the patients for whom a SAE was not resolved.

The base-case calculations were repeated four times using the age-adjusted utility values77 to obtain age-adjusted estimates of QALYs for each of the treatment alternatives.

Type of the model	Treatment alternatives/pathways	Outcome	Time horizon	Discounting
Deterministic decision tree	No treatment Softeners followed by self-irrigation Softeners followed by irrigation administered by a practice nurse	Binary outcome ‘earwax removed’ (yes/no)	7 weeksa	Not applicable
Deterministic decision tree	No treatment Softeners followed by self-irrigation Softeners followed by irrigation administered by a practice nurse	QALY	7 weeksa	Not applicable
excel model with interactive inputs from the deterministic decision treeb	No treatment Softeners followed by self-irrigation Softeners followed by irrigation administered by a practice nurse	QALY	Lifetime (until the baseline cohort of 35-year-olds reaches the age of 80)	3.5% applied to both costs and outcomes

Assessment of uncertainty (sensitivity analysis)

A deterministic sensitivity analysis was used to address particular areas of uncertainty in the model. We investigated the uncertainties around the probability estimates that were expected, a priori, to have a disproportionate impact on the study results.

Scenario analysis was used to address the uncertainty associated with some aspects of the chosen structure of the model.

Parameter uncertainty is addressed using probabilistic sensitivity analysis (PSA). Probability distributions are assigned to the point estimates used in the base-case analysis. The point estimates for treatment effects are reported in Tables 26 and 27.

TABLE 26 - Estimates of clinical effectiveness used in the base-case analysis

SAE, serious adverse event; TM, tympanic membrane.

The range of estimates used in the sensitivity analysis was calculated by varying the point estimate by 30%.

The outcomes were originally reported with respect to number of impacted ears and were subsequently converted into the outcomes with respect to patients.

In total, 14% of patients in softeners and nurse irrigation arm were lost to follow-up in the Coppin and colleagues63 study. It is reasonable to assume that patients did not present for the professionally administered irrigation because the earwax was resolved after application of softeners.

Treatment alternatives including ‘no treatment’ option	Probability of earwax removed (range used in the sensitivity analysis)a	Source	Comments
No treatment (spontaneous earwax removal)	0.05 (0.035–0.065)	bKeane and colleagues, 19954	The outcomes were assessed on an ordinal scale with three categories: ‘impacted’, ‘moderately clear’ and ‘completely clear’. The model outcome ‘earwax removed’ corresponds to the ‘completely clear’ (of wax) outcome in Keane and colleagues4
Use of softeners preceding irrigation at primary care	0.20 (0.14–0.26)	bKeane and colleagues, 19954
Softeners followed by self-irrigation	0.48 (0.34–0.62)	Coppin and colleagues, 200763	The outcomes were initially assessed on an ordinal 4-point scale assessing the degree of obstruction of the membrane (see Chapter 3, Research in progress). Coppin and colleagues63 reported a binary outcome of the degree of wax clearance. The model outcome ‘earwax removed’ corresponds to the combined scores of 0 (no or minimum wax with TM fully visible) and 1 (minor amount of wax with TM essentially visible)
Softeners followed by irrigation administered by nurse at primary care	0.62c (0.43–0.81)	Coppin and colleagues, 200763
Second irrigation – at primary care	0.805 (0.69–0.91)	Linear extrapolation
Third irrigation – at primary care	0.99 (0.95–1.0)	Expert advice	Remaining patients assumed to undergo de-waxing in secondary care
Successful de-waxing at secondary care without SAE	0.97 (0.8–1.0)	Pothier and colleagues, 200656

TABLE 27 - Base-case analysis probabilities associated with a SAE

SAE(s), serious adverse event(s); TM, tympanic membrane.

The following assumptions were made in the calculation of the estimated rate of SAEs: the incidence of symptomatic earwax is 8%7 and 80% of the population with earwax require irrigation.

Event/treatment	Probability of event/treatment (range used in sensitivity analysis)	Source	Comments
SAE associated with irrigation administered by nurse at primary care	0.0006 (0.0003–0.0009)	Expert advice	The base-case estimate is based on 11 SAEs observed in secondary care, servicing the population of 300,000a The upper limit in the range is based on 18 SAEs observed annually in secondary care, servicing a population of 280,000a
SAE associated with self-irrigation	0.0006 (0.0003–0.0009)	Expert advice	Assumed to be the same as the probability of SAE in primary care
Hospital admission for either myringoplasty or treatment of a serious infection	0.05 (0.035–0.065)	Expert advice
Probability of partial permanent hearing loss if myringoplasty is only partially successful	0.25 (0.18–0.33)	Expert advice
Proportion of patients with TM perforation in the total number of patients with SAE	0.333 (0.23–0.43)	Expert advice	There are two types of SAE: a TM perforation and a serious ear infection. These SAEs differ with respect to treatment pathways and costs

The purpose of this analysis was to test the robustness of the cost-effectiveness results to variation in structural assumptions and parameter inputs.

Probabilistic sensitivity analysis

Table 39 reports the mean costs and outcomes from the PSA of results of a single presentation model. The PSA generated cost and QALY estimates for each active treatment alternative that were similar to those for the base-case analysis (see Table 33 for base-case analysis). Variables included in the PSA, distributions and parameters of distributions used can be seen in Appendix 7.

TABLE 39 - Probabilistic sensitivity analysis (single presentation)

SD, standard deviation; QALY, quality-adjusted life-year.

Treatment alternative	Mean cost (£)	SD	Mean effects (QALY)	SD
No treatment	22.62	2.3	0.121727	0.000014
Softeners followed by self-irrigation	37.20	2.6	0.122327	0.000011
Softeners followed by irrigation at primary care	42.30	3.6	0.122342	0.000009

Figure 2 shows the cost-effectiveness acceptability curves for all alternative treatments. The chart indicates the probability that a given treatment option is optimal compared with the alternatives. This suggests that a ‘no treatment’ option is a cost-effective option at lower threshold levels of willingness-to-pay for health outcomes (QALYs). At the threshold above £30,000 per QALY, the treatment involving self-irrigation becomes more likely an optimal option among the three alternative treatments. As the threshold is increased, softeners followed by irrigation at primary care is increasingly likely to be the optimal treatment alternative. At the threshold of about £330,000 per QALY both active treatment alternatives are equally likely to be optimal.

FIGURE 2.

Cost-effectiveness acceptability curves comparing no treatment, softeners followed by irrigation and softeners followed by self-irrigation.

Statement of principal findings

Strengths and limitations of the assessment

The assessment has certain strengths:

• It was independent of any vested interest.

• The review brings together the evidence for the clinical effectiveness and cost-effectiveness of different methods of earwax removal using consistent methods of critical appraisal, presentation and transparency. In addition a de novo economic model has been developed, following recognised guidelines.

• The evidence synthesis was guided by the principles for undertaking systematic reviews and economic evaluations. Prior to undertaking the assessment, the methods were set out in a research protocol (Appendix 1), and this was commented on by an advisory group. The protocol defined the research question, inclusion criteria, quality criteria, data extraction process and methods used to undertake the different stages of the assessment.

• An advisory group has informed the review from its initiation, through the development of the research protocol and completion of the report.

• Systematic searches were undertaken to identify data for the economic model, and main results were summarised and presented.

• The quality of the clinical effectiveness studies was assessed using criteria recommended by the NHS CRD.

In contrast, the assessment was affected by certain limitations:

• The studies identified by the systematic review of clinical effectiveness were published over a considerable period, from 1950 to 2007. Inevitably the nature of clinical trials in terms of their methods and their reporting in publications has changed significantly during this period. Of the trials included in this systematic review, those early published studies tended to provide very limited details of their methods and results. As a consequence, it was difficult to assess the quality of the studies and interpret their results appropriately. In addition, the technologies have continued to develop, with new softeners, methods of irrigation and changes in the nature of the service used to deliver the interventions. Again, this limits the comparability of the evidence.

• Limited details are given in many of the studies of the participants involved in the studies, the dose and frequency of softeners used and the length of follow-up. All affect the generalisability of the findings.

• The effectiveness of the different interventions for removing earwax was assessed through a range of different outcomes. While many of the outcomes assessed the extent and ease of clearance, the specific nature of the different outcomes varied considerably. The extent of clearance was assessed by outcomes such as amount of TM visualised, proportion achieving complete visualisation of the TM, degree of wax removal, hearing, wax clearance, wax occlusion, degree of impaction, amount of wax removed, recurrence of cerumen impaction and further treatment required. Ease of removal of earwax was assessed through outcomes such as volume of water for syringing and syringefuls used, mean number of syringing attempts, mean time to syringe, ease of wax removal, ease of syringing, frequency of syringing, and time taken to de-wax. In many of the studies there was no clear description or definition of these outcome measures. Many are open to subjective measurement and interpretation. As a consequence, it was difficult to directly compare and synthesise the outcomes from the different studies. No studies assessed benefits of the different interventions on quality of life. Although a limited proportion of studies did examine different measures of patient satisfaction, these were limited to whether patients were satisfied with the intervention and if they found them effective and tolerable. Again, these outcomes were not clearly defined or described and open to subjective interpretation.

• Some studies used a different unit of allocation to that used for the analysis (e.g. participants compared with ears). Although there are methods for handling the analysis of such data, it was unclear whether these were followed by the primary studies.

• Although recurrence of earwax is thought to be a common problem for sufferers, it was not considered in the studies included in the systematic review. Recurrence may be a consequence of differences in the anatomical structure of different people’s ears, a result of damage that has been caused to the mechanism that allows earwax to be removed or to the use of medical devices (e.g. hearing aids). Inevitably people within this group will use a disproportionate amount of services and research should focus on identifying methods for diagnosing, treating and, if possible, preventing further problems. The economic evaluation assumed a recurrence rate of one event every 3 years. Inevitably this was a simplification and some people will incur a higher recurrence rate.

• Follow-up with authors to clarify details of their methods and results was not routinely undertaken. Given that the majority of studies that lacked such information were those published nearly 60 years ago, it was considered of limited value to do so as authors would be difficult to trace and further details of studies are unlikely to be available. As technologies have developed and practice has changed, these early studies may have been of more limited relevance to current practice. Where there were uncertainties in key aspects in more recent publications, contact was attempted and was beneficial in some cases.

• Synthesis of the studies included in the systematic review of clinical effectiveness was through narrative review. Due to the limitations of the literature, meta-analysis was not possible.

• The limitations evident in the systematic review of clinical effectiveness and cost-effectiveness and in the evidence on the epidemiology and aetiology of earwax have impacted upon the development of the de novo economic evaluation. With studies either lacking detail of study characteristics or being heterogeneous in nature, the evidence base was limited. As a consequence, assumptions were used to develop the structure of the clinical pathway and to populate the different elements of the model (see Chapter 5, Assumptions used in the model). This has rendered the economic evaluation as exploratory, from which definitive recommendations for changing clinical practice should not be drawn.

• The economic evaluation was developed from a NHS perspective, with patients having to attend primary care for assessment prior to decisions about subsequent treatment. Inevitably, some people will not attend primary care for assessment and/or treatment and will either suffer from the problem untreated or decide to self treat. These people are not included within the evaluation. The rationale for excluding these people was that the intention of providing evidence to decision-makers as to possible future guidance to the NHS, and, as such, the research had to focus on aspects of current provision that could be affected by policy. This led to the exclusion of the evaluation of a service where advice from the NHS on self-syringing is provided without attendance at primary care practice (e.g. NHS Direct or through advice from pharmacists) or the effects of personal costs to the patient. Also, limited evidence was available as to the safety and effectiveness of self diagnosis and treatment.

• With no definitive guidelines identified for the management of symptomatic earwax in the UK, the clinical pathway used to structure the economic evaluation was developed using assumptions based on available evidence and advice from clinical experts. As a consequence, it was uncertain if the clinical pathway used in the model was representative of current practice. For example, it was unclear who undertakes the first assessment within primary care, whether a practice nurse or a GP. Opinion appears to vary and it is likely that different primary care practices adopt different approaches. Also, the duration of the use of softeners was uncertain. Expert advice indicates that people use softeners for a week prior to self-based or primary-care-practice-based irrigation. Others have suggested that the effectiveness of softeners may not depend on the duration of their use49 with spontaneous clearance achieved in a short period (e.g. 20 minutes). Although alternatives have been examined through sensitivity and scenario analyses where possible, in some instances a pragmatic approach has been taken.

• Similar concerns affect the parameter inputs used to populate the economic evaluation. No published evidence was found on the incidence of SAE associated with either professional irrigation or self-irrigation. As a consequence the base-case estimate was based on expert advice from an OTL. This may be an underestimate of the probability of SAE as it is based on the assumption that all patients are referred to secondary care. It is possible that some patients with SAE are first treated in primary care, and only those whose infection or TM perforation does not heal within the expected time are referred to secondary care. It was also assumed that the probability of SAE in both active treatment alternatives is the same regardless of the mode of irrigation (self-irrigation or administered by a primary care practice nurse) and whether irrigation is undertaken for the first, second or the third time, which may not correspond to actual clinical practice.

• There were no published estimates of the loss of HRQoL associated with symptomatic earwax. The utility decrement of 0.006 (standard error = 0.0001) was obtained from Sullivan and Ghushchyan. 78 This was based on SF-12 quality-of-life estimates in the general US population, with the loss of hearing as defined by the ICD-9-CM diagnosis code 389. The original estimates were then converted into EQ-5D values. The estimates are not specific to patients presenting with earwax, although earwax is associated with temporary loss of hearing, as well as other symptoms. The utility decrement does not reflect minor complications, such as itching and tinnitus, and other minor symptoms, such as aural fullness and vertigo that affect HRQoL of patients with earwax problem. The direction of bias in the results of an economic evaluation associated with the use of utility decrement of 0.006 based on the ICD-9-CM diagnosis code 389 is unknown. An alternative utility estimate of 0.0679 obtained in the population eligible for a hearing aid was used in the scenario analysis. The results indicated that the ICER estimates are sensitive to the choice of value of the utility gain.

• The economic evaluation was conducted for a general population of adults aged 35–44 years without any known contraindications to irrigation, reflecting the participants in the primary studies of efficacy used to develop the economic evaluation. Although limited epidemiological data were available, it was felt that this group may not represent those in which the condition is most prevalent. As such, the results may not necessarily apply to other population groups, such as the elderly, children and people with disabilities. This is particularly important when considering the use of new technologies, such as self-irrigation, which require care in their use and adherence to instruction. It is likely that self-irrigation will only be relevant to a subgroup of the general population and therefore it should not necessarily be considered as an option for all groups.

Comparison with previous reviews

The findings of our evidence synthesis are generally in line with those of other systematic reviews in the area. We identified four previous reviews,1,29,37,74 each with slightly different methodologies and therefore different number of included studies. One review74 included nine RCTs, as well as seven in vitro studies and centred on commercially available products obtainable to facilitate ear syringing. All included RCTs were quality assessed and date of last searching was 2002. Excluding the findings of the in vitro studies, the review found that no one product was superior to another.

A Cochrane Collaboration review updated in 2009, which included nine RCTs, centred on the effectiveness of ear drops only. 88 All included RCTs were quality assessed. Although most studies were judged not to be comparable and of poor quality, two46,47 of the nine included trials were meta-analysed. While the overall findings of the review were inconclusive, results from the meta analysis suggest TP is statistically superior to saline in preventing the need for syringing. A third review29 included 18 RCTs and concentrated on the effectiveness of topical preparations for the treatment of earwax. Trials were quality assessed and the date of last searching was January 2004. Preparations in this review were categorised into water-, oil- and non-water-based. Although the review undertook a meta-analysis, it also found that no one preparation was superior to another in either clearing earwax or facilitating syringing.

A more recent review37 included trials and systematic reviews,1,29 and considered treatments for both ear syringing and manual removal. The review carried out a grade evaluation of interventions for earwax on four of the included RCTs. 4,42,56,62 Date of last searching was June 2007. Similarly to our conclusion, the overall conclusion of the review was that there was not enough evidence showing softeners alone to be effective or that one type of softener is superior to another.

Our review differed from these previous reviews, in that we assessed studies by setting, intention to use softening agent alone or as part of the irrigation procedure, followed by population and subgrouped into immediate or delayed follow-up. We assessed all methods of treatment, including self-syringing, all available preparation comparisons and each study was assessed for methodological quality. We did not feel that quantitative pooling of the data was appropriate. In the Hand and Harvey study,29 data was combined. However, with such diverse outcomes and participant groups, we question whether this is appropriate. Furthermore, it is also questionable whether pooled data based on such poor quality studies provides reliable evidence about the effectiveness of these products.

Contributions of authors

AJ Clegg: Developed the original research grant application, acted as principal investigator and project managed the research, developed the research protocol, assisted in the development of the search strategy, assessed studies for inclusion, extracted data from and quality assessed included studies, synthesised evidence, and drafted and edited the final report.

E Loveman: Developed the research protocol, led the systematic review of clinical effectiveness, assisted in the development of the search strategy, assessed studies for inclusion, extracted data from and quality assessed included studies, synthesised evidence, and drafted and edited the final report.

E Gospodarevskaya: Developed the research protocol, assisted in the development of the search strategy, assessed studies for inclusion and extracted data for the economic evaluation, led the development of the economic evaluation and drafted the final report.

P Harris: Developed the research protocol, assisted in the development of the search strategy, assessed studies for inclusion, extracted data from and quality assessed included studies, synthesised evidence and drafted the final report.

A Bird: Developed the research protocol, assisted in the development of the search strategy, assessed studies for inclusion and extracted data from included studies, synthesised evidence, developed the economic evaluation and drafted the final report.

J Bryant: Developed the original research grant application, developed the research protocol, and drafted and edited the final report.

DA Scott: Developed the original research grant application, acted as guarantor for the economic evaluation, developed the research protocol, developed the economic evaluation and drafted the final report.

P Davidson: Developed the original research grant application, developed the research protocol and assisted in the drafting of the final report.

P Little: Developed the original research grant application, developed the research protocol and assisted in the drafting of the final report.

R Coppin: Developed the original research grant application, developed the research protocol and assisted in the drafting of the final report.

We would like to thank members of our advisory group panel who provided expert advice and comments on the protocol and/or a draft of this report:

Professor George Browning, Emeritus Professor of Otorhinolaryngology, MRC Institute of Hearing Research, University of Glasgow, Glasgow Royal Infirmary, Glasgow, UK.

Deborah Williams, Bromley Primary Care Trust, Kent, UK.

Professor Mark E Lutman, Institute of Sound and Vibration Research, University of Southampton, Southampton, UK.

Steve Palmer, Senior Research Fellow, Deputy Director of the Team for Economic Evaluation and Health Technology Assessment and Head of NICE Programme, Centre for Health Economics, University of York, York, UK.

Desmond A Nunez, Director, Department of Otolaryngology Honorary Senior Lecturer, North Bristol NHS Trust and Bristol University, Bristol, UK.

We are also grateful to Karen Welch, Information Specialist (SHTAC, University of Southampton), for conducting the searches; Elizabeth Hodson, Information Assistant (Wessex Institute, University of Southampton), for retrieving references; Mr Jonathan Blanshard, Consultant ENT Surgeon (Basingstoke and North Hampshire NHS Foundation Trust, Basingstoke), Mr Paul Spraggs, Consultant ENT Surgeon (Basingstoke and North Hampshire NHS Foundation Trust, Basingstoke), Dorothy Wicke, Research Nurse (Oakley and Overton Partnership, Basingstoke), Joanne Nesbitt, Practice Nurse (Denton Park Medical Group, Newcastle PCT, Newcastle) and Julia Johns, Practice Nurse (Padnell Road Surgery, Hampshire PCT) for data for the economic model; and, Dr Jill Colquitt, Senior Research Fellow (SHTAC, University of Southampton), for reviewing a draft of this report.

Disclaimers

The views expressed in this publication are those of the authors and not necessarily those of the HTA programme or the Department of Health.