Bone-anchored hearing aids (BAHAs) for people who are bilaterally deaf: a systematic review and economic evaluation

Epidemiology of hearing loss

Although an understanding of the epidemiology of hearing loss is key to the assessment of the clinical effectiveness and cost-effectiveness of a technology and to the subsequent development of guidance on its provision and use, limited research has been undertaken. 15 Assessments of the epidemiology of hearing loss have tended to focus on retrospective cohort studies of its prevalence and have been limited in the type of hearing loss considered. They often use surrogate measures of prevalence such as use of health services, rather than population-based studies, resulting in the potential to underestimate needs. In addition, studies have tended to be affected by differences in the methods for assessing and diagnosing hearing loss (e.g. self-assessment) and variations in definitions and classification of hearing loss (including arbitrary nature of thresholds). Several studies have been undertaken within the UK and elsewhere and those most relevant to this evaluation are discussed in the following section. As there is little evidence focusing on CHL, the epidemiology of hearing loss in general is discussed to give some context to this evaluation.

Estimates of burden of hearing loss in England and Wales

Using the studies of the prevalence of hearing loss and population estimates for England and Wales,16,17,19,22,23 it is possible to provide a provisional estimate of the burden of bilateral hearing loss in England and Wales (Table 3). These estimates should be interpreted with caution owing to the differences in the nature of the studies and the classifications of hearing loss used. Estimates of the prevalence of bilateral hearing loss among children in England and Wales indicate that there could be between 900 and 1000 children in each annual birth cohort with a bilateral hearing loss of ≥ 40 dB HL. Although the majority of children would have a moderate hearing loss of 41–70 dB HL, around 400 would have either a severe (71–95 dB HL) or profound loss (≥ 95 dB HL). It was evident that the majority of impairments among children would be congenital in origin, accounting for hearing impairment in around 750–775 children per annual birth cohort in England and Wales. 16,19 Most of these congenital hearing impairments are thought to be permanent sensorineural (approximately 730 per annual birth cohort). 16 It was estimated that among adults in England and Wales there would be around 1.6 million people aged 17–80 years with a hearing loss of ≥ 45 dB, with around a quarter of these having a loss of ≥ 65 dB (Table 4). Around 60% (275,000) of those with a hearing loss ≥ 65 dB HL would be aged 60–80 years.

Impact of hearing loss

For those people with hearing loss who identify with the ‘Deaf community’ (people whose first or preferred language is British Sign Language), being deaf is seen as part of their total identity and not as a deficiency. 24 However, deafness and hearing loss can have a profound effect on individuals and have been associated with a range of negative consequences, including educational and employment disadvantages, social isolation and stigmatisation. 25,26 According to a report by the World Health Organization, hearing loss is the second leading cause globally of ‘years lived with disability’ and has a larger non-fatal burden than alcohol use disorders. 26 The impact of hearing loss is influenced by the severity of the loss and age at onset. Deafness present at birth or during early childhood (the pre-lingual period) has considerable effects on speech acquisition and cognitive and psychosocial development. 27 Deafness acquired post-lingually requires the individual to adopt new communication strategies and often an entirely different lifestyle,24 and can result in isolation and compromised quality of life (QoL). 27 Hearing loss affects not only individuals, but also the people around them such as family and co-workers. 28 These people have to put more effort into communication with the individual, for example speaking more slowly and with better articulation, turning their face to allow lip-reading and moving closer. 28 As a consequence, there is a risk that people will make less contact and the individual will become more isolated.

Early hearing loss delays the development of basic auditory skills, including auditory detection, discrimination, recognition, comprehension and attention, which negatively affects the child’s ability to learn and use an auditory–oral language system. 29 Difficulties with the rules of language, the meaning of words and the use of language in social contexts lead to comprehension, expressive communication and learning problems, and can result in reduced academic achievement. 29 In contrast, a number of studies have shown that children with hearing loss who are raised by parents with hearing loss often have psychosocial advantages over those who are born to hearing families, as they grow up in an environment where communication is naturally dependent on visual, not oral, cues. 24

A recent study used both parent-report and videotaped data from 116 severely and profoundly deaf and 69 hearing preschool-age children, and demonstrated that hearing-impaired children displayed more behaviour problems and greater difficulties with oral language, parent–child communication and sustained attention than hearing children. 30 High rates of behavioural and emotional problems and a high rate of social maladjustment according to general population norms were also found by a cross-sectional study of 84 children and adolescents (age 2–18 years) attending schools for the deaf. 31 According to parents’ descriptions, children were socially isolated and not participating in structured activities. Similar results were demonstrated by a study in Upper Austria to evaluate mental health and QoL in a representative sample of deaf pupils with a bilateral impairment of at least 40 dB, from both mainstream schools and a school for the hearing impaired. 32 Using the strengths and difficulties questionnaire,33 deaf children scored higher for conduct, emotional and peer problems than children from a normative sample, though differences were less marked for hyperactivity/inattention. Whereas parents of deaf children had a generally positive view of their child’s QoL, deaf children provided a more complex picture, stressing areas of dissatisfaction.

A non-systematic review of mild bilateral hearing loss described studies demonstrating that many children with even mild hearing loss do not perform at expected academic levels, especially in the areas of vocabulary, reading comprehension and language use, and that they expend more effort in listening to speech in quiet and in the presence of background noise than children with normal hearing. 34 The author suggests that children with even a relatively mild degree of bilateral hearing loss may exert more energy than their normal-hearing peers to listen in a classroom setting, leaving them with less energy or attention capacity for processing what they hear, taking notes and other activities required of school children. 34

There is evidence to suggest that the effects of hearing loss in adults differ according to age group. For example, a large Norwegian health-screening survey examined the association between hearing loss, measured by pure-tone audiometry, and self-report symptoms of mental health and well-being in a normal population sample of over 50,000 people aged between 20 and 101 years. 35 The survey found a moderate but clear effect of hearing loss on anxiety, depression, self-esteem and well-being among young and middle-aged people. The strongest effects were found for depression and self-esteem among young men; however, the effects were almost absent among elderly people.

These findings are supported by a recent cross-sectional study which used the internet to determine both hearing status and self-reported psychosocial health in 1511 adults aged 18–70 years. 36 Hearing status was assessed using the ‘National Hearing test’, an online speech-in-noise screening test (also available via telephone) that has been implemented in the UK and the Netherlands. The study found significant associations between hearing status and distress, somatisation, depression and loneliness, but not between hearing status and self-efficacy or anxiety. For every dB signal-to-noise ratio (SNR) reduction in hearing status, the odds for developing moderate or severe depression increased by 5%, and the odds for developing severe or very severe loneliness increased by 7%. The study also found that different age groups exhibit different associations between hearing status and psychosocial health; increased loneliness was an issue for the 18–29 years group and the 40- to 49-year-olds had the greatest number of significant associations (distress, somatisation, self-efficacy, depression and anxiety), but in the 60–70 years group none of the adjusted associations reached statistical significance. The authors suggest that the differences in age groups could be due to differences in the time of onset of hearing loss, in the use of health care, or in the way hearing loss is regarded; it may be considered as part of the normal ageing process by older adults,36 whereas younger people may suffer from being different in terms of not being fully able to function as expected for people of their age. 35 Nevertheless, hearing loss can still affect the lives of older adults, as demonstrated by a population-based longitudinal study of 2688 participants aged 53–97 years. 37 This study used pure-tone audiometry to assess hearing loss, and reported a significant association between severity of hearing loss and reduced QoL.

Variation in services

The number of BAHAs in use is unknown as there are no formal records, but it is thought that there are about 6000–7000 BAHAs in current use in the UK (David Proops, Birmingham Children’s NHS Hospital Trust, March 2010, personal communication). Services for BAHA users vary throughout the NHS and funding is not universally available. Primary care trusts (PCTs) differ in policy on the provision of BAHAs, the eligibility criteria for funding BAHAs and sometimes the number of aids funded. 40 For most children with bilateral hearing loss, PCTs will fund a unilateral BAHA as long as a range of criteria around the nature of the hearing loss, the indication and the social and psychological impact have been met. Few PCTs, however, will fund bilateral BAHAs. Furthermore, as stated previously, not all NHS hospitals have an audiology department or one that specialises in the fitting of BAHAs;2 therefore, people referred for BAHAs may have to travel a considerable distance for treatment. 41

Current service cost

Bone-anchored hearing aid funding is recovered on an individual cost-per-case basis via the PCT. 39 BAHAs are more costly than other hearings aids,42 with the sound processor having to be replaced every 3–5 years. 13 The NHS reference costs 2007/08,43 report that a day-case admission for one-stage insertion of fixture for a BAHA costs £1918. This does not include the cost of fixtures, surgical consumables and the BAHA sound processor, which are reimbursed separately, through a high-cost low-volume top-up payment. Prices from Cochlear UK suggest that the product cost for an implant, abutment and processor ranges from £2700 to £3800, this price being dependent on the type of processor used. The 2010 price list from Oticon Medical AB (William Demant Holding) gives a package deal (processor, implant and abutment) for the Ponto (Oticon Medical, Askim, Sweden) and Ponto Pro (Oticon Medical, Askim, Sweden) of £2654.64 and £2886.60, respectively. Prices for surgery, inpatient episode and internal device cost published by the Nottingham University Hospitals44 in 2007–8 were £2683 for adults and £1588 for children. Additional maintenance costs amounted to £3800 in the first year, reducing to £1250 annually, and these did not differ for adults or children. 44

People with BAHAs require lifelong rehabilitation. Every individual should be on a rolling maintenance programme, and therefore funding is required for ongoing maintenance and replacement. These costs, however, need to be considered in the light of the often considerable costs of hearing loss to the person, the NHS and wider society.

Criteria for treatment

Bone-anchored hearing aids are indicated for people with conductive or mixed hearing loss who can benefit from amplification of sound. BAHAs are also indicated for unilateral sensorineural deafness, also known as single-sided deafness, which is beyond the scope of this report. Otological indications for BAHAs include:45

• congenital malformation of the middle/external ear or microtia

• chronically draining ear or other infective state that does not allow use of an ACHA (e.g. external otitis, draining mastoid activity)

• patients with bilateral CHL due to ossicular disease (and not appropriate for surgical correction) or unable to be aided by conventional hearing aid devices.

Chronic suppurative otitis media and recurrent ear canal infections are the most common diagnoses for adults fitted with BAHAs, as these make it difficult to wear conventional ACHAs. 46 For children, the most common diagnoses are congenital ear malformations, with the BAHA often used instead of a conventional BCHA. 46 Bone thickness is critical for implant integration47 and is often insufficient in children under 4 years of age. 48 While it has been suggested that children as young as 3 years can be fitted with a BAHA,2 the devices are indicated for children aged ≥ 5 years. 49,50

Intervention

The BAHA consists of:

1. A permanent titanium implant (3–4 mm), which is surgically placed in the mastoid bone behind the ear, where it fuses with the living bone (osseointegration). The implant transfers sound vibrations to the functioning cochlea.

2. An abutment, which protrudes through the skin and connects the titanium implant to the sound processor, transferring sound vibrations.

3. A small sound processor, which picks up sound vibrations and transfers them to the abutment. The processor can be attached to the abutment and disconnected by the user. Some processors are at head level, although the more powerful are body worn.

Fitting a BAHA requires surgery and can involve either a one-stage or two-stage surgical procedure, with each stage taking around 1 hour. In the one-stage procedure, the implant and abutment are placed at the same time, whereas in the two-stage procedure, the abutment is fitted after a period of around 3 months in adults or 4–6 months in children to allow osseointegration (where bone fuses with the implant) to occur. 51 The advantage of one-stage surgery is that it requires only one surgical procedure, but it risks transmission of forces through the abutment to the fixture before osseointegration has occurred, resulting in a failure of osseointegration and loss of the fixture. The two-stage procedure is therefore most commonly used for young children, adults who may not be able to protect the abutment adequately (e.g. adults with learning difficulties) or adults with poor bone quality (e.g. irradiated bone following radiotherapy in cancer patients). The one-stage procedure is, however, being trialled in children in some centres and has been found to be safe for children as well as adults,52,53 and can be considered for the older child aged 14–16 years. Finally, the sound processor is connected to the abutment after a period of about 1 month. 54

Bone-anchored hearing aid surgery is generally uncomplicated. The most common potential side effects are soft tissue reactions (with poor hygiene being the most frequent reason for adverse skin reactions)55 and loss of fixture. 54 Failures in children tend to occur soon after implantation as, relative to the adult skull, the infant skull is lower in mineral and higher in water content. 49 Re-operation rates are more common in children than adults, for example a Health Technology Assessment review13 for the Ontario Ministry of Health and Long-Term Care found that re-operation rates for tissue reduction or repositioning were generally under 10% for adults but as high as 25% for children. Similarly, an association between younger age and increasing adverse outcomes, such as requiring revision surgery or experiencing fixture loss, was reported by a UK review of 71 children with BAHAs. 54

If trauma or failure of osseointegration occurs, a reserve or ‘sleeper’ implant may be fitted during the first procedure as a backup. This allows a new vibrating part to be fitted into the second implant as soon as a problem occurs with the first, without the need for a repeated first-stage procedure and subsequent 4- to 6-month wait for osseointegration to occur, during which time the individual would be without any hearing aid. It has been usual practice to fit the sleeper approximately 5 mm from the primary fixture; however, the sleeper is rarely needed and, as the bone is thinner, it is less likely to osseointegrate successfully. 57 For bilateral hearing loss it has been recommended that the sleeper should be placed on the contralateral side at the time of the primary surgery, where it is located in an optimum position and could be used if the decision is made to proceed to bilateral BAHA placement, reducing the number of procedures needed. 57

In the past the BAHA was fitted on just one side (unilaterally), which could be either the better hearing side if the two cochleae differ in acuity58 or the side preferred by the individual. The vibratory patterns of bone-conducted sound would suggest that one BAHA should be sufficient for good hearing amplification in bilateral hearing loss, as sound is transmitted to both the ipsilateral (same side) and contralateral (opposite side) cochleae. 59 However, it has been suggested that people with bilateral BAHAs benefit in terms of greater stimulation levels at the cochlea, better directional hearing and space perception, and better speech recognition in noise. 14,59–61 A potential advantage of this includes road safety, especially for children. A further benefit of bilateral BAHAs is that in the event of a problem with one side, for example an infected site or malfunctioning processor, the individual still has one functioning BAHA rather than being without any hearing aid while the problem is resolved. In a consensus statement from BAHA experts in 2005,42 bilateral application with thorough counselling was advocated in young children with severe congenital conductive hearing impairment.

However, the application of bilateral BAHAs is still debatable. Although the benefits of bilateral stimulation through air conduction (AC) are well established, the benefits with BC are less clear. One consequence of BC stimulation is crossover transmission, where the signal presented to one side of the head is transmitted to the contralateral cochlea. When bilateral stimulation occurs, the signals from each side are transmitted to both cochleae and thus interfere, potentially leading to the cancelling of the differences in signals arriving from the two ears and removing the benefits of binaural hearing. 62 The term binaural hearing ‘denotes our faculty for taking advantage from comparisons of the acoustic signals at the two ears’,63 implying the involvement of specialised brain processing that compares the neural correlates of the acoustic signals at the two ears. While empirical evidence suggests that some people with two BAHAs can use some available cues for localisation of sound, the processes remain unclear.

Participants

• Adults and children with bilateral deafness were included.

• Single-sided deafness was excluded.

• Studies reporting both bilateral and unilateral hearing loss were included only if the groups were reported separately or if the majority of participants had bilateral hearing loss.

Interventions

• Bone-anchored hearing aids, consisting of a surgically implanted titanium fixture. Devices in current use and devices no longer manufactured were included. BAHAs could be fitted unilaterally or bilaterally.

Comparisons

• Bone-anchored hearing aids versus:

  1. – conventional hearing aids (ACHA or BCHA)

  2. – unaided hearing

  3. – ear surgery (tympanoplasty, myringoplasty, ossiculoplasty, stapedectomy and stapedotomy).

• Unilateral versus bilateral BAHAs.

• Studies comparing different BAHA models were excluded.

Outcomes

• Hearing measures, aided hearing thresholds, speech recognition scores.

• Validated measures of QoL and patient satisfaction.

• Adverse events.

• Measures of cost-effectiveness [i.e. cost per quality-adjusted life-year (QALY), cost per life-year saved] and consequences in terms of health service resources.

Study design

For the systematic review of clinical effectiveness, studies were classified according to the criteria by Thomas and colleagues,73 with some adaptations to meet the requirements of this review. The following study designs were eligible:

1. randomised controlled trials (RCTs)

2. controlled clinical trials

3. prospective cohort analytic studies (two groups pre and post, i.e. assessments made before and after BAHA surgery in the intervention group and the control group)

4. prospective cohort one-group pre and post studies (no control group, assessments made before and after BAHA surgery)

5. cross-sectional ‘audiological comparison studies’ [no control group, assessments with intervention and comparator(s) made at one point in time, after BAHA surgery]

6. prospective case series (no comparator condition, outcomes reported with BAHA only).

Where evidence from different types of study design was identified for each of the above comparisons, only studies with the most rigorous designs were included. Where higher level evidence was limited to BAHA models no longer in current use, lower level evidence for models in current use (Divino, Intenso, Cordelle II, BP100, Ponto, Ponto Pro) was considered.

Studies published as abstracts or conference presentations were included only if sufficient details were presented to allow an appraisal of the methodology and the assessment of results to be undertaken.

Only full economic evaluations, those reporting both costs and outcomes, were eligible for inclusion in the systematic review of cost-effectiveness evidence. Conference abstracts were not eligible for inclusion in the cost-effectiveness section.

Characteristics of included studies

Quality assessment

All 12 included studies were rated overall as ‘weak’ for their methodological quality and quality of reporting (Table 7). 59,60,66,76–87 The studies were not described using recognised study types or descriptions, and therefore were termed ‘cohort pre and post studies’ (seven studies66,77–84) or cross-sectional ‘audiological comparison’ studies (five studies59,60,76,85–87) by reviewers. In the cohort pre and post studies, baseline measurements were undertaken with the individual’s previous aid or unaided or both, before BAHA surgery, and measurements with the BAHA were undertaken after a given period of use. In the audiological comparison studies, the measurements were undertaken with the intervention and comparator at the same point in time. With both of these designs, a potential source of bias is that, as with any hearing aid trial, participants are likely to prefer the second hearing aid tested. 89,90 As the study design method was not described by any of the included studies, all were assigned a rating of ‘weak’, although in the hierarchy of evidence, cohort pre and post studies are ranked higher than audiological comparison studies.

Six studies were rated as ‘moderate’ for selection bias,60,81–84,87 indicating that the selected individuals are at least somewhat likely to be representative of the target population, and at least 60% of those identified agreed to participate. The remaining six studies were rated as ‘weak’ for selection bias,59,66,76–78,86 as participants may not be representative of the target population if they are self-referred, or because selection method or the level of participation were not described.

Although some studies described some baseline characteristics of the participants, none of the included studies reported these as potential confounding variables or described how these may be causally related to the outcomes of interest. All 12 studies were therefore rated as ‘weak’ for the assessment of potential confounders.

It was assumed that outcome assessors were not blinded to the intervention as this was not reported by any of the studies. This can lead to bias in the care provided (performance bias) and how the outcomes are assessed (measurement or detection bias). Participants were blinded to the use of unilateral or bilateral BAHAs in one study;87 however, it was assumed that participants were not blinded to the research question in the remaining studies, leading to reporting bias. While it may be difficult to blind participants to the intervention, it is important to note that this bias is present when interpreting the results.

All 12 studies were rated as strong for data collection methods. The 11 studies that reported audiometric data all used methods known to be valid and reliable; however, some of these studies also reported self-reported subjective outcomes using methods that are not valid and reliable and are therefore not discussed in the systematic review (but can be viewed in Appendices 6–10). The study by Hol and colleagues82 used QoL measures that are known to be valid and reliable.

Two studies were rated ‘strong’ for withdrawals and dropouts,81,86 as they described both the numbers and reasons for withdrawals and dropouts, and had a follow-up rate of ≥ 80%. This information could not be deduced from the remaining studies, which may therefore be at risk from attrition bias.

Intervention integrity describes whether all participants received the intervention in the same way, for example the same frequency or duration of use. Only two studies78,82 reported the amount of use with the BAHA (such as number of hours per day), five studies reported that the BAHA was used ‘daily’ with no further description,60,79–81,84,86,87 and the remaining five studies did not report how often the BAHA was used during the study period. 59,66,76,77,83

Statistical analysis of the results was not undertaken by four studies. 60,77,83,86 The remaining eight studies reported statistical analyses, six of which were judged appropriate to the research question59,66,78,82,84,87 and in two studies this was not clear. 76,81 Only one of the studies reported on how missing data were dealt with in the analysis. 82

Audiological measures

Three included studies (reported in five publications)77–81 provided a comparison of audiological measures between BAHA and BCHA (Table 8). All three studies were cohort pre and post studies and none used BAHA models that are in current use. The high risk of bias in all three included studies should be considered when interpreting the results.

Self-reported measures

Béjar-Solar and colleagues77 reported QoL using a validated measure described as the Coop/Dartmouth test. 93 However, limited details were reported; the authors simply stated that ‘results uniformly showed the response “hardly could have done better” (options: hardly could have done better, pretty good, indifferent, pretty bad, hardly could have done worse)’. They also stated that physical and emotional condition was reported as very improved. Given that limited details were reported, these results should be interpreted with caution.

Hol and colleagues82 assessed QoL using a BCHA prior to BAHA surgery and again after 6 months’ experience with a BAHA. There were no statistically significant differences found by the SF-36 or EQ-5D (Table 9). The SF-36 showed increased emotional problems [indicated by a decrease in role limitations (emotional) score, p = 0.19], more pain experienced (p = 0.3) and improved mental health (p = not significant) with the BAHA, but the clinical effects were small (0.33, 0.24 and –0.36, respectively). Scores from the EQ-5D suggested that participants were slightly less mobile (p = 0.26) and experienced more pain/discomfort (p = 0.26) with the BAHA, but again the effect sizes were small (–0.30 and –0.28, respectively). However, statistically significant improvements in disability (p < 0.01) and handicap (p < 0.01) were found with the BAHA by the HHDI, and effect sizes indicated a large clinical impact (1.42 and 0.79, respectively).

Hol and colleagues82 also reported the number of otolaryngology visits over the preceding 6 months for draining ears, and found these to reduce from a mean of 5.40 (SD 4.19, range 0–20) visits with BCHA to a mean of 1.50 (SD 2.10, range 0–6) visits with the BAHA (statistical significance not reported).

Four studies reported the results of subjective questionnaires on patient preference,77,81,82 satisfaction,78 comfort79 and opinions on speech recognition in noise and quiet. 78–81 None of these questionnaires appeared to have been validated and are therefore not discussed here. The data from these studies can be viewed in the data extraction forms in Appendices 6 and 8; however, care should be taken when interpreting results because of the issues associated with non-validated questionnaires.

Audiological measures

Five included studies (in seven publications)76,78–81,83,84 provided a comparison of audiological measures between a BAHA and an ACHA (Table 10). Four studies were pre–post comparisons and the BAHAs used in these studies are no longer manufactured. One audiological comparison study assessed a BAHA model in current use (BAHA Intenso). 76 The high risk of bias in all five included studies should be considered when interpreting the results.

Self-reported measures

Hol and colleagues82 assessed QoL using an ACHA prior to BAHA surgery and again after 6 months’ experience with a BAHA. There were no statistically significant differences found by the SF-36 (Table 11), and although mental health improved slightly with the BAHA (p = not significant), the effect size was small (–0.28). The EQ-5D found a statistically significant increase in anxiety/depression (p < 0.01) with the BAHA, but again the clinical effect was small (–0.3). There were no other statistically significant differences in the EQ-5D. However, statistically significant improvements in disability (p < 0.01) and handicap (p < 0.01) were found with the BAHA by the HHDI, and effect sizes indicated a large clinical impact (0.79 and 0.86, respectively).

Hol and colleagues82 also reported the number of otolaryngology visits over the preceding 6 months for draining ears, and found these to reduce from a mean of 12.7 (SD 10.5, range 0–30) visits with ACHA to a mean of 3.3 (SD 4.8, range 0–25) visits with the BAHA (statistical significance not reported).

Five studies reported the results of subjective questionnaires on patient preference,81–84 satisfaction,78 comfort79,83 and opinions on speech recognition in noise and quiet. 78–81 None of these questionnaires appeared to have been validated and they are therefore not discussed here. The data from these studies can be viewed in the data extraction forms in Appendices 7 and 8, although care should be taken when interpreting results owing to the issues associated with non-validated questionnaires.

Audiological measures

Four cohort pre and post studies were identified that provide a comparison between BAHAs and unaided hearing (Table 12). 66,77,78,83 Three studies used BAHA models that are no longer manufactured. 77,78,83 The study by Kompis and colleagues66 included the BAHA Divino, which is in current use. As discussed in Characteristics of included studies, this study differs from the other included studies as the participants had prior experience with BAHAs when tested unaided at baseline. The high risk of bias in all four included studies should be considered when interpreting the results.

The studies by Béjar-Solar and colleagues,77 Burrell and colleagues83 and Cooper and colleagues78 also reported data comparing the BAHA with the hearing aids previously used, BCHAs,77 ACHAs83 or ACHAs and BCHAs. 78 In order to aid the narrative synthesis of studies for similar comparators, the BAHA data from these studies are repeated in Tables 8, 10 and 12. Care should therefore be taken to ensure that data for BAHAs are not ‘double-counted’ when interpreting the results. The entire data set for each study can be seen in the data extraction forms in Appendices 6–9.

Self-reported measures

No self-reported measures were described by the included studies comparing BAHAs with unaided hearing.

Audiological measures

Four audiological comparison studies59,60,85–87 compared unilateral and bilateral BAHAs (one study was reported in two publications60,85) (Table 13, Appendix 10). None of these studies assessed a BAHA model that is currently available. In each of the four studies, participants had several months’ experience with bilateral BAHAs before being tested with unilateral and bilateral BAHAs during the same session, which could lead to bias, as discussed in Quality assessment. The high risk of bias in all four included studies should be considered when interpreting the results.

Self-reported measures

Priwin and colleagues59 reported the validated MAIS and MUSS to assess hearing skills in ‘meaningful, real world situations’, and the IOI-HA to assess hearing aid outcomes. Scores appeared similar between unilateral and bilateral BAHA users for most items; however, given the very small sample sizes (n = 2–6), these results should be interpreted with caution (Table 14).

Included studies

Only 3 of the 12 included studies reported any data on adverse effects77,81,84 (Table 15). Béjar-Solar and colleagues77 reported that no irritation was present at most follow-up visits (71 of 82 visits); however, it is not clear how many of the 11 participants experienced irritation. No participants in this study experienced infection leading to loss of the implant or any major complications. Osseointegration could not be achieved in one participant following an impact to the mastoid area. Re-operations to remove or replace implants or to reduce the thickness of the subcutaneous layer around the implant were required in 15% of participants in the study by Snik and colleagues. 81 Two of 34 participants in the study by Mylanus and colleagues84 stopped using their BAHA owing to pain of unknown cause; this was after 3 months’ use in one participant and 2.5 years’ in the second participant.

Prospective case series

To supplement the limited data from the included studies, prospective case series reporting adverse events were identified from the list of potentially eligible studies (see Appendix 5). As they were not included in the systematic review, these studies did not undergo the same process of data extraction and quality assessment. Six prospective case series reporting adverse events were identified; five of these can be seen in Table 16. Further examination of the remaining study revealed no useful data for the purposes of this review so it is not discussed further. 99 The discussion below focuses on loss of implants and skin reactions as these were identified as being the most relevant to this report.

BAHA versus BCHA

Four cohort pre and post studies77–82 provided a comparison of BAHAs and BCHAs. Improvements in sound field PTA and warble-tone thresholds were found with a BAHA by all three studies77–81 that reported this outcome, but statistical analysis was reported by only one study (p < 0.01). 78 A statistically significant improvement in SRT in quiet and speech-to-noise ratio was found by one study,81 while another study found no statistically significant difference in speech discrimination score. 78 Statistical analysis was not reported for other results, which included a 23% improvement in 100% speech audiometry discrimination in noise,77 but little difference in speech recognition threshold. 79 Statistically significant improvements with large clinical effects were found with BAHAs compared with BCHAs for disability and handicap using the HHDI. 82

BAHA versus ACHA

Five cohort pre and post studies78–84 and one cross-sectional audiological comparison study76 provided a comparison of BAHAs and ACHAs. Results for sound field pure-tone or warble-tone thresholds were inconsistent among the five studies reporting audiological data. 76,78–81,83,84 Where statistical analysis was undertaken, one study83 found a statistically significant improvement in mean warble-tone thresholds (0.5–4.0 kHz) with a BAHA, while a different study76 comparing each frequency individually found a statistically significant improvement with a BAHA at 1 and 8 kHz, but not at 0.25, 0.5, 2 or 4 kHz. The remaining three studies79,80,83 did not compare data statistically; one study found that the ACHA was better at 0.25 kHz but there was improvement with BAHA at other frequencies,76 another study found that the BAHA was better at 0.25 and 0.50 kHz, but the ACHA was better at higher frequencies,79 while in another study data on average warble-tone thresholds (0.2–4.0 kHz) were described as ‘comparable’ between BAHAs and ACHAs. 83

The direction of the effect was also unclear for speech audiometry, with some studies finding improved outcomes with the ACHA and some with the BAHA. One study reported better outcomes with the ACHA for speech discrimination scores,83 and another for maximum phoneme score79 or speech recognition threshold,79 although statistical analysis was not conducted. A later publication by the same authors as the latter study found a statistically significant deterioration in SRT in quiet with BAHA (p < 0.05), but statistically significant improvement in speech-to-noise ratio (p < 0.05). 81 One study found no statistically significant difference in maximum phoneme score, but a statistically significant improvement in speech-to-noise ratio with BAHA. 84 Speech discrimination score was statistically significantly better with the BAHA in the congenital hearing loss group but not the chronic suppurative otitis media group in one study. 78 The final study reported an improvement in speech-in-noise with the BAHA described as ‘large and clinically significant’. 84

Statistically significant improvements with large clinical effects were found with BAHAs compared with ACHAs for disability and handicap using the HHDI. 82

BAHA versus unaided hearing

Four studies66,77,78,83 reported improvements in sound field thresholds and speech audiometry (where reported) with BAHA compared with unaided hearing, which were statistically significant where analysis was undertaken.

Unilateral versus bilateral BAHAs

An improvement in sound field average tone thresholds with bilateral BAHAs compared with unilateral BAHAs was found in adults87 and a small group (n = 3) of children59 with previous experience of BAHAs. Two studies found that speech recognition thresholds in quiet were statistically significantly lower with bilateral BAHAs,60,87 although one study found similar results between unilateral and bilateral BAHAs. 86 Bilateral BAHAs produced better results when noise was presented from baffle/best side, but not when noise was presented from the shadow side. 60,86,87 Three studies found that localisation of sound was improved with bilateral BAHAs. 59,60,87 Two studies60,87 reported the binaural masking level difference test and suggested that BAHAs give binaural hearing, although the validity of their methods is uncertain.

Adverse events

The included studies reported very limited data on adverse events. Prospective case series reported rates of loss of implants between 6.1% (9–25 months follow-up)102 and 19.4% (median 6 years’ follow-up). 100 The vast majority of participants in the prospective case series experienced no or minor skin reactions.

Published economic evaluations

A systematic literature search was undertaken to identify full economic evaluations and cost studies that included BAHAs and other potentially relevant hearings aids (BCHAs and ACHAs). The methods for the systematic review are described in Chapter 2. Details of the inclusion and exclusion criteria are shown in Chapter 2, Inclusion and exclusion criteria and the search strategies are documented in Appendix 2.

A total of 225 potentially relevant publications were identified by the searches. No relevant full economic evaluations involving BAHAs were found after screening titles and abstracts. The searches identified 29 economic evaluations in hearing aids. None of the identified economic evaluations (19 in cochlear implants and 10 in other hearing aids) was found to be directly applicable to the aim of this study.

Two studies107,108 that reported costs, but not outcomes, associated with BAHAs were reviewed for their relevance to estimating resource use and costing in our economic model. Catalano and colleagues107 assessed the costs (in terms of participants’ and physicians’ time, as well as fees for treatment) of outpatient and inpatient insertion of BAHAs in a retrospective study with 19 US participants. However, the costs in the study were not directly applicable to the development of our economic model because of the US perspective of the study. Moreover, day case surgery would be the current service standard for surgical implantation, preparatory to fitting of the BAHA processor, in the NHS. Therefore, this study was not used to inform the economic model.

Watson and colleagues108 in a retrospective analysis compared service use before and after BAHA insertion in 26 adults with suppurative otitis media exacerbated by behind-the-ear hearing aids. They identified a reduction in the number of treatments and visits after BAHA insertion compared with behind-the-ear hearing aids. As part of the analysis they outlined treatment protocols for people undergoing surgical implantation and post-surgical management (within the ENT clinic), as well as audiological management to fit and commission the BAHA processor in a UK district general hospital. This study was used, in conjunction with current service standards109 relevant to the NHS and expert opinion, to identify the management pathway for individuals considered eligible for a BAHA as a basis for costing the intervention in our economic model (see Resource use and cost data for more details).

Unpublished economic evaluations

An unpublished, UK-based, economic evaluation of BAHAs conducted for an MSc thesis was identified. 110 This study does not strictly meet the inclusion criteria for the review, as it appears that a minority of participants had bilateral hearing loss (33.3% were reported as having bilateral hearing loss, the remainder were unilateral/single-sided or not stated). However, given that this is the only economic evaluation of BAHAs compared with conventional hearing aids that was identified, it is briefly reviewed below.

The initial sampling frame for the study was all adults (greater than 16 years old) undergoing primary BAHA implantation by the Departments of Audiology and Otolaryngology at University Hospitals Birmingham NHS Foundation Trust (UHB) between April 2007 and June 2008. All patients were invited to participate in the study and were sent the Health Utilities Index (HUI) 15-item self-completion questionnaire, along with a set of questions concerning current hearing aid use and duration of hearing impairment. Of the 147 eligible participants, 89 returned the first questionnaire (61% response rate). Of those 89 who completed the baseline questionnaire, 70 completed a second questionnaire (at least 3 months after the fitting of their BAHA).

The mean age of participants responding to the initial questionnaire was 55 years, with 44% of respondents being male. The mean time from surgery to receiving their BAHA was 2.8 months and the mean time from fitting of their BAHA to receiving the second questionnaire was 6.1 months.

Costs for BAHA provision were based on charges by the provider to PCTs. These were £5689 for the first year, to cover surgical and audiological assessment, implantation surgery, post-surgical care and acquisition and fitting of the BAHA sound processor, and a cost for the annual maintenance contract. The contribution of each component to this total charge was not reported. For subsequent years, the continuing cost of providing BAHAs was based on the annual maintenance contract fee of £1004. The analysis appears not to take account of costs associated with adverse events or treatment failure in BAHA users. The comparator group for the analysis was modelled by assumption, using participants’ reported usage of hearing aids prior to receiving their BAHA [39/70 (56%) reported using one or more hearing aids prior to receiving their BAHA]. Costs of conventional hearing aids were based on provision of an NHS digital hearing aid (£260) replaced every 5 years.

Outcomes in the analysis were based on differences in utility (before and after BAHA provision), scored using the HUI3 and HUI2 algorithms, and participants’ age-sex-specific life expectancy, derived from UK life tables. The mean HUI3 utility at baseline was 0.57 (95% CI 0.52 to 0.62) for all 89 respondents to the initial questionnaire and 0.59 (95% CI 0.53 to 0.65) for the 70 participants who responded to both outcome questionnaires. The mean HUI3 utility score for participants post BAHA was 0.66 (95% CI 0.60 to 0.72). It is unclear whether individual utility values or the mean values were used to derive QALYs, as the study report refers both to using ‘pooled [utility] results’ and to ‘each subject[s]… utility scores’ in calculating QALYs.

Table 17 reports the costs, QALYs and incremental cost-effectiveness ratio (ICER) for BAHAs compared with participants’ previous hearing aid provision. Costs and outcomes are discounted at 3.5%.

A series of scenario analyses were reported, assuming that all participants in the modelled comparator group were using standard hearing aids (rather than the observed proportion, 56%) and adopting shorter time horizons. The study110 also reported a subgroup analysis, breaking the total population of participants down by the main indication for treatment. The ICER results were generally robust to variables in the scenario analyses, with the ICER reducing fractionally (to £17,224 per QALY gained) for increasing the proportion of the modelled comparator group using conventional hearing aids. Reducing the time horizon of the model led to higher ICERs (£18,820 at 20 years, £22,097 at 10 years and £28,928 at 5 years). In the analysis by indication for treatment, BAHAs were least cost-effective in participants with bilateral CHL (£32,331 compared with £7459 for unilateral CHL and £19,391 for single-sided deafness). However, the number of cases included for each comparison was low (13–31 cases).

Published quality-of-life studies

In addition to the searches for economic evaluations and cost studies, a systematic literature search was undertaken to identify QoL studies of BAHAs and other potentially relevant hearing aids (see Appendices 2 and 11 for details).

A total of 322 potentially relevant publications were identified by the searches. After screening titles and abstracts, only one relevant study (by Hol and colleagues,82 discussed in Chapter 3, Self-reported measures) was identified. This study reported statistically non-significant differences before and after BAHA implantation using generic measures (SF-36 and EQ-5D), but statistically significant differences using a condition-specific measure (HHDI). The difference between the condition-specific and the generic health-related QoL instruments was probably due to a lack of hearing dimension on the EQ-5D and SF-36 questionnaires. As has been noted elsewhere,112,113 the HUI3 appears to be a more suitable generic measure for QoL in a population with hearing difficulties rather than the EQ-5D or SF-36. Owing to a lack of sensitivity in the EQ-5D and SF-36 it was decided that the study by Hol and colleagues82 should not be used in the decision model.

Baseline cohort

The population in the base-case analysis are those with bilateral deafness who were already provided with BCHAs, but are considered for BAHA owing to convenience and improved wearability. For the purposes of the model, adults are considered to be aged ≥ 18 years. Children are those less than 18 years of age. The sex composition of the included cohorts reflects the general population and is applicable only in calculation of death rates.

Data sources used in the model

Clinical effectiveness data

Gain in hearing

Chapter 3 of this report, Assessment of clinical effectiveness: BAHA versus BCHA, presents the systematic review of outcomes for BAHAs compared with BCHA in terms of audiological outcomes. No quantitative summary of the reported outcomes could be produced and there was little consistency of reporting between included studies. Consideration was given to using one of these outcomes for inclusion in the economic model, but it is not clear from the included studies what these outcomes mean to those receiving BAHAs. We were not able to identify any robust methods to map from audiological outcome measures to QoL measures. As a result, the outcomes in the model are based on potential gains using a generic QoL scale that is sensitive to changes in hearing. Quality of life reports the approach adopted for outcome assessment (potential QoL gain) in the model.

Adverse events

Table 18 presents estimates of the annual risk of implant failure (loss of bone integration), using data reported and discussed in Chapter 3, Prospective case series. Data in columns 1–5 of Table 18 were extracted from the study by Bonding,100 which reported the number of patients followed up for up to 7 years and the count of implants lost each year. Annual risk of failure (column 6) has been estimated from the reported cumulative success proportions, using the declining exponential approximation to life expectancy (DEALE) method,115 which estimates a constant risk over time. The estimated annual risk of failure, beyond year 2, varies between 3.8% and 4.5%.

TABLE 18 - Loss of implants, estimated cumulative success and annual risk of failure

Annual failure risk at each time point is calculated based on the cumulative success up to that time point, e.g. annual failure risk at 3 years was calculated as 1/3 × –ln(0.885).

Annual failure risk (R) estimated using the DEALE method: R = 1/t × –ln(S), where ln(S) is natural log of cumulative survival at time t.

Time (years)	n	Implants lost	Success (% success per year)	Cumulative success (%)	Annual risk of failurea (%)
0	31	0	100.0	100.0
1	31	0	100.0	100.0	0.00
2	29	0	100.0	100.0	0.00
3	26	3	88.5	88.5	4.09
4	21	1	95.2	84.2	4.28
5	19	1	94.7	79.8	4.51
6	15	0	100.0	79.8	3.76
7	14	1	92.9	74.1	4.28
> 7	10	0	100.0	74.1

Figure 3 illustrates the fit of the predicted annual risk [based on the cumulative success proportion at 7 years (0.741)] to the data reported by Bonding,100 treating the cumulative success proportions as Kaplan–Meier survival estimates. This figure suggests that applying the constant risk may overestimate the failure rate in the first 2 years from implantation. For subsequent years there appears to be reasonable agreement between the Kaplan–Meier estimate and the line of fit using the DEALE method. The annual failure risk estimated at year 7 (4.28%) is applied in the base-case analysis.

FIGURE 3.

Goodness of fit of annual risk of titanium implant failure (derived using the DEALE method) and Kaplan–Meier estimates.

The clinical data discussed in Chapter 3, Prospective case series, typically report the proportion of BAHA users experiencing skin reactions, rather than an estimate of rates. The latter would be more appropriate given the differential follow-up periods shown for participants in all of the included studies. Table 19 reports our estimates of the rates of skin reactions by grade for people using BAHAs, based on the reported counts and cumulative person-years of observation in each study shown in Chapter 3, Prospective case series (95% CIs have been estimated using exact confidence limits for Poisson counts116). The estimated rates show considerable variation, with substantially higher rates of all skin reactions, except grade 3, reported by Portmann and colleagues104 and Mylanus and colleagues,102 than in other studies. It is difficult to interpret the significance of the results of these studies or suggest explanations for variation between studies, as little information is provided on characteristics of populations in the studies [for example, the proportion of paediatric cases or the proportion of subjects (adult or children) with learning disability]. The studies are comparatively small and are not all consistent in the categorisation of grade of skin reaction. We therefore adopted rates estimated from the study by Håkansson and colleagues101 in our model, as this is the largest study using standard definitions for grades of skin reaction. The sensitivity of results to this assumption is addressed in scenario analyses, in which estimates based on other included studies are applied.

TABLE 19 - Estimated rates of skin reaction, by grade, in included studies

One implant was removed owing to non-integration at 4 months. Rate per 100 PYAR = 2.05 (95% CI 0.05 to 11.41).

Source	Person-months at risk (n)	Count by grade			Rate per 100 person-years at risk (PYAR) (95% CI)
		Grades 1 and 2	Grade 3	Grade 4	Grades 1 and 2	Grade 3	Grade 4
Holgers et al. 1988105	1515 (64)	15	5	1	11.88 (6.65 to 19.60)	3.96 (1.29 to 9.24)	0.79 (0.02 to 4.41)
Håkansson et al. 1990101	5542 (164)	67	16	1	14.51 (11.24 to 18.42)	3.46 (1.98 to 5.63)	0.22 (0.01 to 1.21)
Jacobsson et al. 1992103	586 (15)	8	0	0	16.38 (7.07 to 32.28)	0.00 (0.00 to 7.55)	0.00a (0.00 to 7.55)
Mylanus et al. 1994102	476 (33)	18	0	1	45.34 (26.89 to 71.72)	0.00 (0.00 to 9.30)	2.52 (0.06 to 14.05)
Portmann et al. 1997104	1338 (41)	30	1	2	26.91 (18.15 to 38.41)	0.90 (0.02 to 5.00)	1.79 (0.22 to 6.48)

The impact of these adverse events is included in cost estimates used in the model (see Resource use and cost data for discussion of the costs of managing and treating adverse events). No direct estimate of the impact of adverse events on QoL was found in our literature searches and these are not included in the base-case analysis.

All-cause mortality

The most recent UK life tables were used in the model to estimate the percentage of the cohort dependent on age that dies in each cycle of the model. No increase in the mortality rate was assumed to be applicable to the baseline cohort.

Resource use and cost data

Both resource use and cost data are taken from an NHS perspective. The resource use associated with the implantation of BAHAs was identified using Watson and colleagues’s108 retrospective study of resource use and the Bone Anchored Hearing Aid Service Standards. 109 These sources were used to develop a resource use protocol (illustrated in Figure 4) which was discussed with clinical experts.

FIGURE 4.

Resource use protocol for patients considered eligible for a BAHA.

Four distinct phases of costs associated with BAHAs were identified. These cost phases were defined as being associated with:

• assessment of surgical and audiological eligibility

• surgery

• post-surgical management (up to 12 months following the initial surgical procedure and included fitting of external processor)

• long-term management.

All cost data and relevant sources are given and discussed in turn below. All unit costs and cost-effectiveness results are expressed in 2009 pound sterling.

Assessment of surgical and audiological eligibility costs

Potential BAHA users have an initial consultation with the specialist BAHA surgeon. This includes an examination of the middle and external ears and determination of the aetiology of hearing loss, as well as an assessment of the individual’s general medical status and suitability for BAHAs. The cost of consultation with the surgeon was taken from the NHS reference costs. 43 An ENT outpatient first attendance cost of £110.78 was used for adults and a paediatric ENT outpatient first attendance cost of £131.69 was used for children.

An audiological assessment is carried out by an audiologist, who assesses current middle and external ear status. AC and BC are tested using pure-tone audiometry. There may be an evaluation of the user’s current hearing aid provision, if applicable. The cost of the audiological assessment (£57.48) is taken from the NHS reference costs. 43

The final stage of the assessment phase involves a multiprofessional consultation (£147.36) to agree to the individual’s eligibility for a BAHA and to gain consent for surgery.

The breakdown of unit costs for the initial assessment of surgical and audiological eligibility for adults and children is given in Table 25. The total cost of the assessment of surgical and audiological eligibility cost was £315.63 for adults and £336.53 for children.

TABLE 25 - Unit and total cost of assessment of surgical and audiological eligibility for adults and paediatrics

Source: NHS reference costs 2007/08,43 inflated to 2008–9 values.

Consultant led: first attendance non-admitted face to face.

Audiological services: fitting of hearing aids and counselling (including tinnitus); using currency code for fitting of hearing aids and counselling: assessments (Code AS1A).

Consultant led: first attendance multiprofessional non-admitted face to face.

Item	Adult unit costs (£)	Paediatric unit costs (£)
Initial consultation with surgeona	110.78	131.69
Audiological assessmentb	57.48	57.48
Multiprofessional consultationc	147.36	147.36
Total cost	315.63	336.53

Surgery costs

The cost of one-stage surgery to implant the BAHA was taken from the NHS reference costs for a day-case BAHA operation with a cost of £2004.57 for adults. The paediatric two-stage method for implantation of a BAHA was assumed to cost twice the NHS reference cost for BAHA surgery and therefore had a cost of £4009.14. 43 Additional costs including the fixture and abutment costs and additional consumables from the surgery are given in Table 26. These had a total cost of £989.50. These costs are current list prices and were provided by the UHB. The breakdown of unit cost of surgery for adults and children is given in Table 26. The total cost for surgery in adults was estimated at £2994.07 in adults and £4998.64 in children.

TABLE 26 - Unit and total cost of surgery for adults and paediatrics

Sources:

Healthcare Resource Group Code CZ26Z bone-anchored hearing aids: day case. NHS reference costs 2007/08,43 inflated to 2008–9 values.

UHB.

Item	Adult unit costs (£)	Paediatric unit costs (£)
BAHA surgery procedurea	2004.57	4009.14
Guide drillb	15.50	15.50
Countersinkb	29.50	29.50
Fixtureb	310.00	310.00
Abutmentb	520.00	520.00
Cover screwb	62.00	62.00
Healing capb	23.50	23.50
Dermatome blade b	29.00	29.00
Total	2994.07	4998.64

Post-operative costs

Watson and colleagues108 identified three obligatory visits to ENT and audiology after surgery. This was verified by our experts. These visits consist of changing the dressing (removal of the mastoid bandages after 24 hours), removal of healing disc and stitches and an assessment to ensure osseointegration at around 3 months. The cost for these three visits was taken from the NHS reference costs. 43 There appears to be practice variation at this stage with either the surgeon or a specialist nurse undertaking the assessments. It was assumed that in the base case the surgeon undertook the assessment. This uncertainty was explored in a scenario analysis. Therefore, the cost of each consultation was taken from the NHS reference costs for an ENT outpatient follow-up attendance with a cost of £72.11 for adults and £90.93 for paediatric ENT43 (Table 27).

TABLE 27 - Unit and total cost of post-operative follow-up

Source: NHS reference costs 2007/08,43 inflated to 2008–9 values.

Consultant led: follow-up attendance non-admitted face to face using service codes for ENT (Code 120).

Consultant led: follow-up attendance non-admitted face to face using paediatric ENT (Code 215).

Audiological services: fitting of hearing aids and counselling (including tinnitus); using currency code: fitting of hearing aids and counselling: fitting (Code AS1FA).

Audiological services: fitting of hearing aids and counselling (including tinnitus); using currency code: fitting of hearing aids and counselling: follow-up (Code AS1FU).

Item	Unit costs (£)
	Adults	Paediatrics
ENT outpatient visit to change dressinga,b	72.11	90.93
ENT outpatient visit to remove healing disk and stitchesa,b	72.11	90.93
ENT outpatient visit to ensure osseointegrationa,b	72.11	90.93
Initial fitting of the BAHA by an audiologistc	64.80	64.80
4-week audiological follow-up of BAHA fittingd	50.17	50.17
3-month performance of processor by an audiologistd	50.17	50.17
Total costs	381.47	437.91

The first audiological attendance is to fit the BAHA device. This includes adjusting the device settings based on audiometric results and the user’s response, and recording the output of the device using electro-acoustic methods. Furthermore, consideration is given to fitting BAHA accessories, discussing the user’s expectation and providing information on the management of the device. The second audiological follow-up occurs at around 4 weeks and includes a site and abutment inspection when the tightness of the abutment is checked using a torque driver. A record of the output from the device using electro-acoustic methods is taken and this is compared with the measure taken at the time of fitting. The third audiological follow-up occurs at 3 months and includes a record of the output from the device to further assess performance and outcome measurement. An unaided and aided sound field audiometry test is undertaken and post-operative questionnaires are administered. The costs of the three audiological visits were taken from the NHS reference costs and were assumed to consist of a fitting cost of £64.80 and two follow-up visits of £50.17 each. 43 The breakdown of unit cost of post-operative follow-up for adults and children is given in Table 27. The total cost of post-operative follow-up for adults is £381.47 and £437.91 for children.

The cost of the BAHA processor is also included in this phase. The average cost of four of the BAHA processors currently used in the NHS and the cost of their maintenance plans are reported in Table 28. The average cost for both the processor and the maintenance plan was used in the model owing to uncertainty over which processors are currently used most in the NHS. A sensitivity analysis of the range of costs of the processor and first year maintenance plan was used to explore this uncertainty.

TABLE 28 - Unit and average cost of processors and maintenance plans for each BAHA

Source: UHB and Southampton University Hospitals Trust (SUHT).

BAHA model	Cost of processor (£)	Cost of maintenance plan (£)	Total cost (£)
Divino	1820.00	610.00	2430.00
Intenso	1980.00	665.00	2645.00
Cordelle	1970.00	670.00	2640.00
BP 100	2995.00	1000.00	3995.00
Average cost	2191.25	736.25	2927.50

The total cost of the post-operative follow-up and the processor (and maintenance plan) after surgery for adults is £3308.97 for adults and £3365.41 for children.

The cost per user at each stage of a successful implantation of a BAHA in the first year is given in Table 29. This includes the cost of audiological assessment, surgery costs and post-operative surgery costs with a total cost estimated at £6618.68 for adults and £8700.59 for children per successful implantation of a unilateral BAHA.

TABLE 29 - Total costs per patient for successful BAHA implantation in the first year

Stage of BAHA implantation	Total cost adults (£)	Total cost paediatrics (£)
Total initial audiological assessment cost (see Table 25)	315.63	336.53
Total surgery costs (see Table 26)	2994.07	4998.64
Total post-operative surgery costs, first year (see Tables 27 and 28)	3308.97	3365.41
Total cost for BAHA implantation	6618.68	8700.59

Long-term follow-up costs

Expert opinion and current service standards109 suggest that BAHA users will have periodic audiological follow-ups. In addition to the annual audiological follow-up, those who continue to use a BAHA will have the sound processor replaced every 3 years under the annual maintenance plan. At the annual audiological assessment a record of output from the device is measured using an electro-acoustic test and this is compared with the measure taken at time of fitting. Performance and outcome measurements are undertaken with unaided and aided sound field audiometry tests. Post-operative questionnaires may also be administered at this stage (Table 30).

TABLE 30 - Long-term costs of BAHA use

Sources:

NHS reference costs 2007/08,43 inflated to 2008–9 values. Audiological services: fitting of hearing aids and counselling (including tinnitus); using currency code: fitting of hearing aids and counselling: follow-up (Code AS1FU).

UHB and SUHT – see Table 28.

Item	Frequency	Unit costs adults (£)	Unit costs paediatrics (£)
Audiological assessment	Every year	50.17a	50.17a
BAHA maintenance plan	Every year	736.25b	736.25b
Total long-term costs every year		786.42	786.42

Adverse events costs

Two types of adverse events were identified from the review of the literature and confirmed by our clinical experts: skin irritation associated with the implantation of the BAHA and a loss of osseointegration of the fixture. The resource use associated with grade 1 or grade 2 skin irritation (using the Holgers and colleagues grading system105) was assumed to consist of an outpatient visit for dermatological care and antibiotics to treat the reaction. The costs of an outpatient dermatology visit was taken from the NHS reference costs, with a cost £118.10 for adults and £134.82 for children. 43 The antibiotic steroid cost of £5.40 was taken from the British National Formulary (BNF). 125 The total estimated cost of treating a grade 1 or grade 2 skin reaction is £123.50 for adults and £140.22 for children (Table 31). Grade 3 skin reactions were assumed to require surgical revision and were costed as intermediate skin procedures, provided as day cases. 43 Grade 4 skin reactions were assumed to require removal of the skin-penetrating implant and were costed using the reference cost applied for the initial day case surgery (see Table 26), excluding the costs of surgical consumables required for the initial implantation.

TABLE 31 - Cost of adverse events

Source: NHS reference costs 2007/08,43 inflated to 2008–9 values.

Dermatology: consultant led: first attendance non-admitted face to face (Code 330).

Intermediate skin procedure without complication and comorbidities (Healthcare Resource Group Code JC04C) as day case.

Assume same cost as initial surgery.

See Table 26.

See Tables 27 and 28. Excludes cost of BAHA sound processor.

Skin reaction		Cost, adults (£)	Cost, paediatrics (£)
Grades 1 and 2	Dermatologista	118.10	134.82
	Antibiotic steroids	5.40	5.40
	Total	123.50	140.22
Grade 3	Revision surgeryb	663.66	663.66
Grade 4	Day case surgery to remove implantc	2004.57	2004.57
Loss of bone integration
Re-operationd		2994.07	4998.64
Post-operative follow-up, year 1e		1117.72	1174.16
Total		4111.80	6172.80

If loss of bone integration occurs, then there is a cost of re-operation surgery to re-implant the fixture and an associated post-operative surgery cost. This was assumed to be the same as the initial surgical operation and post-operative follow-up, as calculated above (see Table 29). The breakdown of costs for skin irritation and loss of osseointegration are reported in Table 31. The overall total cost of a loss of osseointegration was £4111.80 for adults and £6172.80 for children.

Costs associated with comparator pathways

BCHA costs

People considered eligible for provision of BCHA were assumed to undergo an initial audiological assessment (including assessment of middle and external ear status as well as testing of AC and BC using pure-tone audiometry) similar to that for those considered suitable for BAHAs. People receiving a BCHA were assumed to have a single attendance to fit the BCHA with a follow-up visit at 3 months to assess the performance of the hearing aid. In subsequent years individuals were assumed to attend once a year for follow-up, with replacement of the hearing aid occurring every 5 years.

A range of costs was provided for the BCHA device itself, from a low cost of £117–183 for a body-worn aid (which is attached to the transducer via a cord) to £250 for a more cosmetically appealing option, incorporating a behind-the-ear hearing aid constructed onto the headband to drive the vibrator/transducer (using an NHS digital power aid). A higher cost of £300–350 was provided for proprietary behind-the-ear devices (Table 32). These costs were supplied by two NHS providers (see Table 32) and reflect current NHS purchasing arrangements for BCHA, rather than manufacturers’ list prices for any particular devices.

TABLE 32 - Unit and total costs for provision of BCHAs

Sources:

Audiological services: fitting of hearing aids and counselling (including tinnitus); using currency code for fitting of hearing aids and counselling: assessments (Code AS1A).

Audiological services: fitting of hearing aids and counselling (including tinnitus); using currency code: fitting of hearing aids and counselling: fitting (Code AS1FA).

SUHT.

UHB.

Audiological services: fitting of hearing aids and counselling (including tinnitus); using currency code: fitting of hearing aids and counselling: follow-up (Code AS1FU).

Item		Unit costs, adults (£)	Unit costs, paediatrics (£)
Initial audiological assessment costsa		57.48	57.48
Fitting costsb		64.80	64.80
Cost of BCHA device (£)	Low (body-worn)c	117–183	117–183
	Low (behind-ear worn)c	250	250
	Highd	300–350	300–350
Post-fitting, follow-up assessment (3 months) of device by an audiologiste		50.17	50.17
Total	Low (body-worn)	289–355	289–355
	Low (behind-ear worn)	422	422
	High	472–522	472–522

The total cost of providing a BCHA was estimated as between £289 and £522 for the first year (depending on the type of aid provided), with long-term costs of £50.17 (for an annual audiological assessment) and a replacement cost (for a new device, fitting and post-fitting assessment) of between £232 and £465 every 5 years.

Cost analysis

Exploratory cost-effectiveness analysis

An exploratory, augmented base-case analysis was developed to incorporate potential benefits from improved hearing resulting from BAHA provision and improvements in wearability, compared with BCHA. In this analysis we assumed that the QoL benefit from improved hearing could be proxied by levels of the hearing attribute in the HUI3 and that identical gains were potentially achievable through use of BAHA and BCHA. Differences between the utility associated with use of BAHA or BCHA were assumed to be realised through differences in the proportion of time that members of the modelled cohorts used each device, as outlined in Quality of life.

The results are reported in terms of total costs and total QALYs for each treatment strategy, incremental costs and benefits and the ICERs. Costs and outcomes are discounted at 3.5%.

The results of the augmented base-case analysis in paediatric cases are reported in Table 44. The average costs estimated for each cohort are identical to those reported in Table 38 (average costs, assuming BAHA treatment failures revert to BCHA). The QALY outcomes for BCHA and BAHA have been estimated on two different potential levels of hearing gain associated with the use of hearing aids (based on items in the HUI3 hearing domain). Under the first assumption (QALY1) the utility gain from aided hearing is 0.178, based on attaining level 5 on the HUI hearing domain (‘able, when using hearing aid, to hear conversation with one other person, but unable to hear what is said in a group conversation’; see Table 21 and Quality of life for more details), while under the second assumption (QALY2) the utility gain from aided hearing is 0.384, based on attaining level 3 on the HUI3 hearing domain (‘able to hear both conversation with one other person and what is said in a group conversation, when using hearing aid’; see Table 21 and Quality of life for more details). These assumptions apply equally to BCHA and BAHA. However, different assumptions apply to the proportion of the cohort using the relevant hearing aid (see Table 34), which give rise to different QALY estimates for the two modelled cohorts (BAHA vs BCHA).

For paediatric cases, the QALY gain associated with providing BAHA ranges from 0.14 to 0.30 (depending on the assumed level of utility associated with aided hearing), resulting in ICERs of £119,367 and £55,642, respectively.

The results of the augmented base-case analysis in adults, applying the same assumptions on potential utility gain from aided hearing and usage of hearing aids, are reported in Table 45. The estimated QALY gain from use of BAHA is similar to that for children, ranging from 0.14 to 0.29 depending on the assumed level of utility associated with aided hearing. However, the ICERs are lower (£100,029 and £46,628, respectively) given the lower incremental costs estimated for adults (resulting from the use of one-stage surgery in adults and lower costs for adult outpatient attendances).