The effectiveness and cost-effectiveness of cochlear implants for severe to profound deafness in children and adults: a systematic review and economic model

Incidence and/or prevalence

Children

A 15-year study by Fortnum and colleagues7 that examined birth cohorts of those born in the UK between 1980 and 1995 found nearly 3600 (21%) cases of children with permanent severe hearing loss (71–95 dB) and 4262 (25%) cases with permanent profound hearing loss (> 95 dB). The aetiology of severe hearing loss was 22% more likely than other levels of deafness to have perinatal causes (p < 0.001). Those with profound deafness were more likely to have a genetic (42%; p < 0.001), postnatal (20%; p < 0.001) or prenatal (12%; p < 0.001) aetiology. Fortnum and colleagues also looked at the subset of children with cochlear implants. Here, significantly more of these children had a postnatal aetiology (47.7%; p < 0.001) than those profoundly deaf children not implanted.

Adults

The most common cause of eventual severe to profound deafness in the elderly is presbycusis. 1 This is progressive hearing loss due to the failure of hair-cell receptors in the inner ear, in which the highest frequencies are affected first. Hearing loss may also be due to noise exposure, ototoxic drugs, metabolic disorders, infections or genetic causes. 8 Communication problems from deafness may lead to social isolation and depression. 9–11

Children

In children, hearing loss may have significant consequences for linguistic, cognitive, emotional and social development. 12 Many deaf children live in relative isolation in their early years and their first contact with other deaf children may be when they attend school. 13

Early life may be dominated by trying to adapt to their impairment. This may involve learning to lip-read and/or using cued speech or sign language, either at mainstream or special schools. 14 The inability to communicate wants and needs may alienate children from family members. 12

At school, deaf children may also exhibit more behavioural problems than their hearing peers. Greater problems are evident in those with bilateral severe to profound deafness. 13,15 Congenitally deaf children fare poorly academically. 13,15 In the longer term children with uncorrected hearing loss are at an increased risk of becoming underemployed. 16,17

Measurement of quality of life in young children (i.e. < 5 years) is often by proxy through parents (or teachers). 18,19 In total, 90% of deaf children have two hearing parents and 95% have at least one. 13,20 There are no standardised measures to assess quality of life specific to deaf children, deaf adolescents or their parents. 15 However, two generic profile measures have been used to assess quality of life in deaf children, the Child Health Questionnaire (CHQ)15,21 and the Munich Quality of Life Questionnaire for Children (KINDLr). 19,22 Both have been used to assess quality of life in children with severe to profound deafness, including those who are prelingually deafened, either using an acoustic hearing aid or with a cochlear implant. Prelingual deafness refers to deafness occurring before a child has developed speech, with an age of 3 years often taken as a proxy for this. Postlingual deafness refers to deafness occurring after this time.

An Australian cross-sectional study15 used the 28-item short version of the parent proxy report CHQ to compare quality of life in children aged 7–8 with significant congenital hearing loss (with mild to profound hearing impairment, including cochlear implant users) with their hearing peers. The CHQ has 12 subscales – physical functioning, role (social–physical), role (social–emotional), bodily pain, behaviour, mental health, self-esteem, general health, parent impact (emotional), parental impact (time) and family activities – and produces two summary scores (physical and psychosocial). The CHQ has only been provisionally validated. 23 Children with congenital hearing loss scored significantly worse in six domains [role (social–physical), behaviour, mental health, parent impact (emotional), parental impact (time) and family activities]. The psychosocial summary score (out of 100) was also significantly lower in children with congenital hearing loss (49.2, SD 9.6) than in children with normal hearing (53.1, SD 8.2). Ceiling scores of 100 were reported on four subscales in both groups. 15 The study did not control for differences in parental level of tertiary education or co-morbidity.

An Austrian study22 used the KINDLr to assess the quality of life of children (aged 8–16 years) with cochlear implants. It has six domains (physical health, general health, family functioning, self-esteem, social functioning and school functioning). Total self-reported scores for boys (67.5, SD 9.6) and girls (63.1, SD 8.6) were significantly below those of their hearing peers (76.8, SD 8.6 and 76.7, SD 8.7 respectively).

Adults

Studies indicate that deafness may adversely affect the quality of life of adults24–29 and that of their family members. 31,32 Mulrow and colleagues33 reported that 82% of the elderly deaf stated that deafness had an adverse effect on their quality of life and 24% felt depressed.

Commonly reported difficulties identified by postlingually deaf adults include feelings of isolation, loss of confidence and tinnitus. 10 In social settings, in particular those with background noise, communicating with others can be very challenging. 17 In a study of 47 severely to profoundly postlingually deafened adults in Wales,34 nearly two-thirds identified feelings of isolation, loss of confidence and loss of social life as causing them difficulties. Such difficulties may influence the viability of personal networks and, therefore, the sense of self. 16 These effects can lead to reduced feelings of well-being. 17 The difficulties caused by hearing loss may result in withdrawal from social activities, reducing intellectual and cultural stimulation and cognitive functioning. 12,17

In an Italian study of 1191 non-institutionalised elderly,35 those with hearing impairment had twice the risk [odds ratio (OR) 2.1, 95% CI 1.36–3.25] of poor functioning in daily living activities compared with non-impaired elderly, with over 20% of the elderly deaf having a level of functioning classified as poor by the Instrumental Activities of Daily Living (IADL) scale. A similar relationship between hearing loss and self-sufficiency was seen among middle-aged adults (51–61 years) living in the community. 36

A US cross-sectional study31 of 178 adults (17–84 years) with profound postlingual deafness showed that 13% showed clinically elevated levels of depressed mood (T-score ≥ 70) and 16% had feelings of significant social isolation on the Minnesota Multiphasic Personality Inventory (MMPI). Their levels of anxiety in social contexts, measured using the Social Avoidance and Distress (SAD) scale, were also greater than those of people diagnosed with simple phobias. 31 A follow-on study involving 95 of these participants also showed that they had lower levels of social participation than 44 age-matched hearing control subjects. 31 Candidates experienced lower levels of pleasant social events (16%) and non-social events (19%) than control subjects (23% and 27% respectively; p < 0.05). 31

Dalton and colleagues27 used the Short-Form 36 (SF-36) to measure the impact of hearing impairment on quality of life in 2688 adults. The SF-36 measures physical functioning, role limitation because of health problems, social functioning, role–emotional, general health, bodily pain, mental health and vitality on a scale from 0 to 100. They found that severity of hearing loss was significantly associated with worse quality of life. Those with moderate to severe hearing loss (> 40 dB) had the lowest scores. Scores were 1.9–5.9 points lower than in those without hearing loss across six of the eight domains. The greatest differences were in the domains of role (physical) (5.9), physical functioning (5.2), vitality (4.2) and role (emotional) (3.9). There was no association with general health (2.1) or bodily pain (1.9), although scores did decline with hearing loss. 27 There was also a statistically significant difference in the two adjusted summary component scores in physical and mental health between people with no hearing loss (40.3, SE 1.87 and 50.2, SE 1.59 respectively) and those with moderate to severe hearing loss (38.8, SE 1.89 and 49.0, SE 16.1 respectively). 27 The impact of increasing deafness on quality of life has been shown in other studies. 37 It is not clear to what extent these relationships are causally related to the hearing impairment rather than to other disabilities or diseases associated with ageing or the aetiology of deafness (e.g. premature birth).

In a Dutch study38 of 46 people waiting for cochlear implants, SF-36 scores in those with profound postlingual deafness, mean age 51 years (SD 16), were between 60.2 (SD 41.5) [role (physical) domain] and 79.2 (SD 24.8) (physical functioning domain). A Norwegian study39 of 27 postlingually deaf, cochlear implant candidates compared pre- and postimplantation scores. They found that postimplantation participants had similar physical functioning scores (80.8) as preimplantation participants but higher role (physical) scores (71.0, SD 40.0). The vitality domain had the lowest scores (58.8, SD 21.8).

Relevant national guidelines

• National action plan for audiology – Improving Access to Audiology Services in England. 48 This framework document sets out how health reform levers can be brought to bear to improve quality, efficiency and access to audiology services. It also describes national work intended to support this for adults and children.

• MHAS – Modernisation of Hearing Aid Services (Adults)

• MCHAS – Modernisation of Hearing Aid Services (Children) (2001)

• NHS Newborn Hearing Screening Programme – seeks to identify deafness within 26 days of birth.

Significance to the NHS

Although deafness per se is not an illness, it does impact on NHS resources through the need for procedures of diagnosis and assessment and the possible provision of acoustic hearing aids and cochlear implants with the associated follow-up and support required.

Cochlear implants have been available in England and Wales since the late 1980s. Currently there are 14 tertiary implant centres in England and three in Wales. Treatment is provided by multidisciplinary teams of clinical scientists in audiology, audiologists, surgeons, speech and language therapists, hearing therapists and administrators. Within paediatric services, teams also include teachers of the deaf. Some units use or have access to clinical and/or educational psychology, link nurses and paediatricians. 49

The recently published best practice guidance Improving Access to Audiology Services in England48 seeks to improve the responsiveness of audiological services to cut waiting times to a maximum of 18 weeks.

NHS Newborn Hearing Screening Programme

Nationally all newborns are screened for hearing problems within 26 days of birth with positive cases referred to NHS audiology departments. If confirmed deaf the baby should be provided with a hearing aid within 2 months. Referrals to other services are usually coordinated by the audiology department. These include:

• paediatric services, to assess for possible co-morbidities

• ear, nose and throat services, to consider surgery, including possible referral to a tertiary centre for cochlear implant assessment

• educational services

• social services.

Variation in services for hearing loss

There is geographical variation in the way that hearing impairment is managed in different parts of the UK. In general, models of adult services tend to follow the MHAS. Differences occur in the types of digital hearing aids fitted, the diagnostic facilities available and the access to hearing therapists. The professionals providing services may also vary. The larger departments and teaching hospitals tend to employ clinical scientists (audiology) whereas local district general hospitals are more likely to employ audiologists.

Paediatric services also vary; there are different models of newborn hearing screening, with some being maternity-unit based and others community based. In some areas second-tier paediatric services are delivered in the community, usually by community paediatricians with support from audiologists. In other areas these services have been integrated into the main audiology department and are clinical scientist/audiologist led. Hearing aid services for paediatrics will usually follow the MCHAS model.

The referral process for cochlear implants may also vary, but referrals will go to the major centres (14 in England and three in Wales). Similar protocols are used for adults and children.

There may be slight variation nationally in the initial screening and diagnostic services described above. Although follow-up care may vary more, the following description may be considered reasonably typical (expert advisory group, 2006, personal communication).

For children, following diagnosis and fitting with an initial hearing aid 2 months later, services generally conform to the following pattern:

• visit to the audiology department every 2 weeks for new ear moulds for 6 months

• visit from educational services every week

• formal diagnosis of level of hearing loss at 3 months

• potential cochlear implant use considered very early on, i.e. usually within the first year

• audiological assessment at 6 months, then every 6 months until 2 years old or until hearing aid use is stable and consistent

• once stable audiological checks every 6 months until 5 years old, then annually until adult services take over at 18 when there are 4-yearly reassessments.

Adults make up the vast majority of people seen in the NHS for hearing problems. The NHS provides over 2,600,000 adult hearing aid services per year; 600,000 of these are assessments of hearing, 500,000 are hearing aid fittings, 500,000 are follow-up appointments and more than 1,000,000 are for ‘repairs’ of devices. 50 Services are coordinated by audiology departments. Adults normally have a 4-yearly review, although this varies across the UK (Dr Jonathan Parsons, Mid, East Devon and Exeter Area Primary Care Trust, 2006, personal communication).

Aim of cochlear implants

The aim of cochlear implants is to improve quality of life by enabling people with hearing loss to hear and interpret sounds, thus improving their ability to understand others, communicate effectively and move safely in their environment.

Insertion procedure

The procedure for cochlear implant surgery takes between 2 and 3 hours under general anaesthetic. It involves the insertion of the electrode array into the cochlea through a tunnel that has been drilled above the external ear canal, bypassing the mastoid cavity. 52

Bilateral implantation in the UK

Throughout the UK there had been 115 bilateral implantations by the year ending April 2006; of these, 33 were simultaneous implantations (both ears implanted in the same operation) and 82 were sequential implantations (ears implanted in different operations). In the year ending March 2007 there were an additional 32 child and 11 adult bilateral implantations. There had also been 34 bilateral reimplantations to this date either because of contraindication of more surgery to the first ear or because residual function of the first device was considered likely to contribute to the benefit from a second implant. 49

Estimated future demand

The BAA, BCIG and ENT UK joint submission to NICE in March 200749 estimated (based on the assumption that 25% of severely deaf people with > 85 dB HL may benefit from an implant) that in 2005 there were potentially 625 children and 1620 adults per year who could benefit from implantation.

Intervention

This assessment considers one or two multichannel cochlear implants using whole-speech processing coding strategies that attempt to transmit as much sound signal information as possible, for example ACE, SPEAK and CIS, rather than earlier feature extraction strategies. In cases in which the coding strategy was not disclosed in the research paper, attempts were made to contact authors for this information. When there was no response it was assumed that studies which collected data after 1995 used whole-speech processing and that those before did not.

This distinction between coding strategies was made following expert advice that whole-speech processing strategies are considered more effective and that older coding strategies are no longer being implanted by the NHS (Professor Quentin Summerfield, University of York, January 2007, personal communication). The devices currently supplied to the NHS and those in the included studies are shown in Table 4. There are currently 11 cochlear implant devices sold on contract to the NHS. Only two of these were used in the studies included in this report. Fourteen others were used in the studies but are no longer supplied under contract to the NHS.

Comparator

One cochlear implant was compared with non-auditory support, acoustic hearing aids and two cochlear implants. Two cochlear implants were compared with one cochlear implant plus a contralateral acoustic hearing aid.

Population

The population was children aged from 12 months to 18 years and adults.

Outcomes

These included:

• sensitivity to sound

• speech perception

• speech production

• psychological outcomes

• educational outcomes

• adverse events

• health-related quality of life.

Relevance to the UK NHS of the technology

Studies were included if they were in health-care settings that were considered to be sufficiently similar to the UK to be relevant to this assessment (e.g. Europe, North America and Australasia).

Systematic reviews and randomised controlled trials

All systematic reviews and RCTs were included, including those with waiting list controls. Systematic reviews ideally only consider well-conducted RCTs; however, in this instance the evidence base is methodologically highly variable across the policy questions of interest. The inclusion criteria for studies of clinical effectiveness were as follows.

Controlled studies

Other types of controlled studies (i.e. non-RCTs, cross-sectional studies and pre/post studies with people acting as their own controls) were included. These designs, including within-subject designs, were considered acceptable because levels of sensitivity to sound outcome at preimplantation were near or at zero and because hearing loss was unlikely to improve over time. Thus, benefits seen over time can be attributed to the intervention. However, with speech outcomes for children it could be expected that there would be a natural improvement over time. Prospective cohort designs, in which other people acted as control subjects, were included when baseline levels of hearing loss between the two groups were similar.

The inclusion of prospective cohort studies in a systematic review requires caution. The absence of randomisation introduces the possibility of bias in the selection of participants so that the group receiving the intervention may have different characteristics from the control group. These dissimilarities may cause confounding. Further bias may occur in measurement, for example ceiling effects from the benefit of a unilateral cochlear implant may obscure the benefit of an additional implant.

A number of the included studies were prospective case series; although these had the advantage of allowing participants to be their own controls, the validity of the results obtained is uncertain as familiarity with test materials, and therefore procedural learning, may affect results. 58 In observational studies confounding is a greater issue than lack of statistical power. A review59 evaluating non-randomised intervention studies has concluded that:

Results from non-randomised studies sometimes, but not always, differ from results of randomised studies of the same intervention. Non-randomised studies may still give seriously misleading results when treated and control groups appear similar in key prognostic factors.

Internal validity

Consideration of internal validity addressed the selection of appropriate study groups, the identification of sources of possible confounders and their effects on analyses, whether the study was prospective, the blinding of assessors and data analysts, the validity and reliability of outcome measures, the reporting of attrition and the appropriateness of data analysis.

External validity

External validity was judged according to the ability of a reader to consider the applicability of findings to a patient group in practice. Study findings can only be generalisable if they (1) describe a cohort that is representative of the affected population at large or (2) present sufficient detail in their outcome data to allow the reader to extrapolate findings to a patient group with different characteristics. Studies that appeared representative of the UK population with regard to these factors were judged to be externally valid.

Definitions of study design used in this report

• Waiting list RCTs – These are RCTs in which participants are randomly allocated to have the intervention immediately or to go onto a waiting list and have the intervention in the future. Outcomes from both groups are then compared at baseline and at the same time points from baseline. The weakness of this design is that confounding variables may affect the control group in the time before they receive the intervention.

• Pre/post studies – This design consists of measuring and comparing outcomes before and after the intervention, with participants usually acting as their own controls. The main weaknesses of this design are its inability to account for maturation effects and selection bias.

• Cross-sectional studies – These measure differences in outcomes between intervention and control groups at one point in time. Usually the intervention and control groups are two different groups of people. However, in the case of cochlear implants they may be the same, as the external component of an implant can be removed and outcomes measured without the device. The main weaknesses of this design are that it cannot report changes over time and if different groups are measured then selection bias may occur.

• Prospective cohort studies – In this design the intervention group is compared with control subjects who have been selected to have similar characteristics. The weakness of this design is the lack of randomisation, which would control for selection bias and potential confounders.

Summary tables for each comparison are shown at the beginning of the relevant section.

Only seven studies reported both sensitivity to sound and functional measures of severity of deafness. Moreover, there were insufficient studies (with the same comparators) to reveal any apparent relationships between the preimplantation sensitivity to sound hearing level and size of functional outcome.

Of the studies reporting both types of measures of deafness (sound sensitivity and functional ability) only one62 used health utility outcomes; this study classified implant recipients according to preimplantation speech perception using standard sentence tests and when using optimally fitted hearing aids. This can be viewed as a classification according to level of ‘functional hearing’, and was predictive of levels of utility gain with implantation. Given that, in the current UK NHS, ability to benefit from cochlear implantation is primarily judged on the basis of level of functional hearing ability, it is unfortunate that the vast majority of the evaluative research on this technology only reports the audiologically measured severity of deafness of implantation candidates.

Sensitivity to sound

Only one study, conducted by Manrique and colleagues (n = 182),122,127 measured sensitivity to sound in this comparison. They found a significant improvement in pure-tone audiometry (PTA) scores (p < 0.05) at 12 months post activation compared with preimplantation (preimplantation = 115.8, SD 3.25; 12 months post implantation = 34.3, SD 8.25). This indicates that a fundamental change in the children’s ability to detect sound had occurred.

Speech perception

In total, 666 children were measured for speech perception ability across seven studies using 22 different instruments.

Staller and colleagues (n = 78)124 used four measures of speech perception [ESP battery, GASP, Lexical Neighbourhood Test (LNT) and the Hearing In Noise Test for Children (HINT-C)]. The results showed a range of improvements for children (age 3–17 years) over 6 months, from a 35% difference with the LNT (word recognition) to a 50% difference with the HINT-C (sentence recognition).

Further evidence for the benefits of cochlear implants came from the MED-EL report (for the FDA) (n = 82). 125 This measured speech perception 6 months post activation with six instruments [ESP battery, GASP, Communicative Skills Checklist for all ages (18 months to < 18 years) and the Multisyllabic Lexical Neighbourhood Test (MLNT), LNT and Bamford–Kowal–Bench (BKB) test for older children (≥ 5 to < 18 years)]. The scores for younger children ranged from a 50% difference on the ESP spondee identification test (two long syllables) to a 70% difference on the ESP battery (pattern perception test). Older children’s scores ranged from a 53% difference with the BKB test (simple sentences) to a 79% difference with the ESP spondee identification test and the GASP. However, not all children were entered for all tests.

An earlier study by Illg and colleagues (n = 167)70 reported on seven measures in children from 12 months to 15 years over a 2-year follow-up period [Test for Auditory Perception and Speech (TAPS), GASP (word and sentence recognition), Mr Potato Head (following instructions to assemble the toy), pattern perception, one- and two-syllable tests and a minimal pairs test]. All results showed a trend favouring cochlear implants, with scores for younger children (< 7 years) ranging from an 8% (SD 8.5) difference with the GASP to a 59% difference with a pattern perception test. Older children’s (7–15 years) scores ranged from a 15% (SD 14.5) difference with the Mr Potato Head assembly task and the minimal pairs test (words that differ by one feature) to a 39% difference with a pattern perception test.

Kessler and colleagues’ much smaller study (n = 49)126 measured speech perception with the Phonetically Balanced Kindergarten Word List (PB-K), ESP, GASP and Mr Potato Head over 6 months for children aged 7 years or over. They found that all outcome measures showed a benefit from cochlear implants, with a range of improvement in scores from a 32% difference with the PB-K to a 54% difference with the ESP.

Additionally, the MED-EL, Staller and Kessler studies reported parental ratings of listening behaviours (e.g. responding to a door bell) using the MAIS as the mean percentage point improvement over 6 months. Although significance was not reported, all found increased scores (MED-EL = 38%, Staller = 16%, Kessler = 20%).

All of these four studies70,124–126 found a positive association between early age at implantation and improvements in speech understanding. However, only one study reported significance levels.

Harrison and colleagues (n = 82)123 used four speech perception tests [Test for Auditory Comprehension (TAC), GASP, PB-K word test and PB-K phoneme test] with children aged from 2 to 13 years. They found a positive trend associated with earlier implantation with mean differences from preimplantation ranging from 6.36% with the TAC to 84.25% with the GASP. A similar association between age at implantation and positive outcome was found by Nikolopoulos and colleagues (n = 82)120 who examined participants’ understanding of English grammar and found a link between earlier age at implantation and greater understanding of the construction of grammar. The proportion of those with understanding comparable to normal-hearing peers rose from 2% at preimplantation to a remarkable 67% after 5 years when measured with the Test for the Reception of Grammar (TROG).

In an earlier study Nikolopoulos and colleagues121 found significant negative correlations with age at implantation at 3 (–0.38) and 4 (–0.58) years from baseline on the connected discourse tracking (CDT) measure of auditory performance (p < 0.01), and at 4 years (–0.44) with the Iowa Matrix Sentence Test (IMST), a closed-set sentence test (p < 0.05), thus further indicating that increased benefit from cochlear implants was associated with earlier implantation. Interestingly, Nikolopoulos and colleagues also showed significantly greater improvements for younger-implanted children than older-implanted children in Categories of Auditory Performance (CAP; a measure of performance on a range of auditory tasks performed in a quiet situation); scores were recorded at between 24 and 48 months of implant use (correlation coefficients with age at implantation: 24 months = –0.32, p = 0.006; 36 months = –0.48, p = 0.0007; 48 months = –0.58, p = 0.002). This indicates that in quiet situations there was again benefit from earlier implantation.

Speech production

Results for speech production were similarly positive. Only one study, that by Nikolopoulos and colleagues (n = 126),121 examined the effects of age at implantation on speech production. Using the Speech Intelligibility Rating (SIR) scale they found that at 4 years post activation there was a significant correlation (–0.49) between earlier implantation and better speech production (p < 0.01).

Sensitivity to sound

In the study by van den Borne and colleagues65 a total of 43 children had auditory outcomes measured. The ability to detect everyday sounds was measured on a scale from 1 to 4; both groups were measured before implant and at 6-month intervals, up to 24 months post implant. Both groups were measured with acoustic hearing aids before implant after which the cochlear implant group were measured with implants alone. The score in the cochlear implant group improved by 3.5 points and that in the acoustic hearing aid group by 1.9 points during this time.

Speech perception

Across all studies a total of 209 children had their ability to understand speech measured; two studies reported significance levels.

Mildner and colleagues (n = 49)128 used a cross-sectional study design to compare children with cochlear implants or acoustic hearing aids. They found a mean percentage gain in understanding visually and orally presented words for the cochlear implant group, with an overall difference in word scores of 22.4%, (p < 0.01) (cochlear implant group = 82.8%, acoustic hearing aid group = 60.4%).

An earlier pre/post implantation study by Osberger and colleagues (n = 58)130 measured speech perception using five tests (ESP, GASP, Mr Potato Head, common phrases test, PB-K phonemes and words). Improvements were seen on all measures over 18 months, ranging from a mean score difference between times of 19.9 on the common phrases test to 56.5 on the ESP. All measures showed a significant difference in favour of cochlear implants (p < 0.0001).

A much larger study by Svirsky and colleagues (n = 297)131 compared the difference between actual PB-K words scores for implanted children and predicted PB-K scores for children using acoustic hearing aids. However, they reported insufficient information to calculate the difference in scores for the acoustic hearing aid group. The cochlear implant group mean scores improved by 6.3% over 18 months for those aged less than 6 years and by 6.5% over 12 months for those aged between 6 and 12 years.

A small study by Osberger and colleagues (n = 30)131 measured speech perception using three instruments (ESP, GASP words and sentences, PB-K phonemes and words). Measures were taken before implantation with acoustic hearing aids and 6 months post implantation with cochlear implants. The results showed improvements on all measures over 6 months for the cochlear implant group. The difference in scores between the groups ranged from a mean percentage score difference of 33.3% on PB-K phonemes to 49.6% on PB-K words; however, statistical significance was not reported. These participants may be a subset of those of Osberger and colleagues. 130

van den Borne and colleagues (n = 43)65 also reported on speech perception, this time using the Scales of Early Communication Skills for Hearing Impaired Children (SECSHIC), in a prospective cohort study. Their results showed a small relative improvement in verbal receptive skills over baseline for cochlear implant users compared with those using acoustic hearing aids of 0.1 over 24 months. However, the actual scores at 24 months were better for acoustic hearing aid users (cochlear implant group = 50, acoustic hearing aid group = 54), although it should be noted that the baseline scores were lower for the cochlear implant group (cochlear implant group = 43, acoustic hearing aid group = 47.5). As both groups made gains from their baseline scores (cochlear implant group = +7.0, acoustic hearing aid group = +6.9) it would appear that maturation effects contributed to improvements in receptive language.

Speech production

Only Tomblin and colleagues (n = 58)129 reported speech production measures, using the Index of Productive Syntax (IPSyn) to analyse transcripts of children retelling stories in a prospective cohort study. Their results showed a mean difference in 5-year total scores of 19.6 in favour of cochlear implants. However, these results may be susceptible to bias as the cochlear implant group had the advantage of repeated exposure to the test whereas the acoustic hearing aid group had only one exposure. Regression analysis showed that, when age was included, length of use of cochlear implants was the main factor in IPSyn scores.

The visual summary of results in Table 14a–c shows the overall positive impact of cochlear implants compared with hearing aids for the profoundly deaf children who participated in these studies.

Sensitivity to sound

Only one small study, that by Litovsky and colleagues (n = 13),134 reported sensitivity to sound, using the MAA, which assess the narrowest angle at which a person can detect a change in sound direction. This is used to determine whether there is an advantage of bilateral implantation in improving the ability to tell the direction that a sound has come from. However, of the 13 participants recruited, only the nine who found the task easiest were measured, somewhat undermining these results. Litovsky and colleagues found that these children were able to discriminate sound location better with two implants than with one (mean score first side 27.7%, second side 29.7%, bilateral 16.2%; p < 0.001).

Speech perception

Two studies with a total of 48 participants measured speech perception using six tests.

The cross-sectional study by Peters and colleagues (n = 30)133 measured speech perception in quiet conditions with the MLNT, LNT and HINT-C and in noise with the Children’s Realistic Intelligibility and Speech Perception (CRISP) test. The participants were grouped by age (group 1: 3–5 years, group 2: 5.1–8 years, group 3: 8.1–13 years). Children recruited were not a representative sample of candidates for cochlear implantation, as only those who already had open-set speech perception abilities with their first implant were eligible.

None of the results in quiet conditions reached significance. However, all of the results showed a trend in favour of the use of bilateral implants. The difference in scores ranged from 5% with the HINT-C in the oldest group to 13% with the LNT for the youngest group. They found that children who received their second implant when they were younger than 8 years old did better at word recognition than those who received their second implant when they were older (10.6% mean improvement < 8 years, 5.5% mean improvement 8–13 years). The CRISP test directs sound from the front and at both ears individually to test the ability to identify picture and sound combinations. All sound directions showed a significant bilateral advantage with the greatest advantage when noise was directed at the ear that was implanted first (mean improvement of 13.2%; p < 0.0001).

Similar results were found in an earlier small cross-sectional study by Kuhn-Inacker and colleagues (n = 18). 79 They measured speech perception with the Gottinger test in quiet conditions and with a discrimination in noise test. Both tests showed a trend in favour of bilateral implantation. Mean scores in quiet conditions were 70% and 71% for each ear independently and 87% bilaterally. When tested in noise the unilateral mean score was 60% and the bilateral mean score was 80%. However, this study was very poorly reported with no description of the selection criteria and other key quality indicators.

Table 16a and b provides a visual summary of the results.

Sensitivity to sound

A total of 39 children had sensitivity to sound measured by determining their ability to detect the direction that sounds came from using the MAA.

Litovsky and colleagues (n = 19)134 found that bilaterally implanted children were significantly better than children with one implant plus a hearing aid at detecting sound direction measured using the MAA (which indicates the smallest change in position of a sound that can be detected, with lower scores being better) (bilateral = 28.0°, unilateral + acoustic hearing aid = 44.4°; p < 0.05). In another study (n = 20)135 with the same outcome measure a similar result was obtained (bilateral = 20.0°, unilateral + acoustic hearing aid = 27.0°; p < 0.05).

Speech perception

Peters and colleagues133 measured the ability of 50 children to understand speech using three different instruments [MLNT words (age 3–5 years), LNT words (age 5.1–13 years) and HINT-C sentences (age 8.1–13years)]. All results showed a trend in favour of bilateral implantation; for some groups this reached significance (MLNT, 3–5 years: bilateral = 92.3, unilateral + acoustic hearing aid = 67.3, p = 0.003; LNT, 5.1–13 years, bilateral = 86.0, unilateral + acoustic hearing aid = 69.4, p = 0.004).

Speech production

Litovsky and colleagues (n = 20)135 measured speech production using the CRISP test in quiet and in noise. Both conditions showed a significant benefit for bilateral implantation (quiet: bilateral = 20.00, unilateral + acoustic hearing aid = 24.00, p < 0.0001; noise: bilateral = 11.00, unilateral + acoustic hearing aid = 17.50, p < 0.005).

Table 18a and b provides a visual summary of the overall benefit from bilateral implantation reported by these studies.

Educational placement

The data in Table 23 indicate that, taken together:

• children with cochlear implants are more likely to be in mainstream education, including a special unit within the school (75–95%), than in a school for the deaf (5–21%).

• children with cochlear implants are less likely to be in schools for the deaf (5–21%) than profoundly deaf children without cochlear implants (29–46%).

Effect of implanting before or after starting school on educational placement

Archbold and colleagues143 looked at the effect of implantation before and after children had started school on educational placement. They found that 53% of children who were implanted before starting school were in mainstream schools compared with 6% of those who were implanted after (Table 24 and Figure 2).

FIGURE 2.

The effect of implanting before and after starting school on educational placement.

Educational attainment

Speech perception

The total number of participants in this comparison was 984, with the four studies using nine instruments; however, there is some overlap between those taking part in the UK Cochlear Implant Study Group (UKCISG)62,144 study and those in the Mawman and colleagues study145 (number unknown); therefore the actual total number will be less.

The UKCISG study (n = 316)62,144 measured speech perception with the BKB and a normalised index of audiovisual gain (AVGN) preimplantation and 9 months later. Participants were divided into ‘traditional candidates’ [TC: mean hearing level = 117.1 dB (95% CI 115.7–118.5)] or ‘marginal hearing aid users’ [MHU: mean hearing level = 108.7 dB (95% CI 106.8–110.5)] on the basis of their score on speech intelligibility tests taken before implantation. Both groups were profoundly sensorineurally deaf. The MHU results are also reported in the next comparison (one cochlear implant versus acoustic hearing aid) as their preimplantation measures were with acoustic hearing aids. They are recorded here to show the comparative effects of level of hearing loss.

The mean scores for both outcome measures improved at 9 months compared with preimplantation, with the TC group showing significantly more improvement than the MHU group [BKB: TC = 53.0 (95% CI 48–58), MHU = 44.0 (95% CI 37–51), p < 0.05; AVGN: TC = 68.0 (95% CI 63–71), MHU = 31.0 (95% CI 26–37), p < 0.001].

Further evidence for the benefits of cochlear implants came in the same year from Mawman and colleagues (n = 214),145 who looked retrospectively at patient records from one UK cochlear implant centre. Speech perception results were measured with BKB sentences and Arthur Boothroyd (AB) monosyllable words preimplantation and 18 months later. They found non-significant trends in favour of cochlear implants for both measures [BKB mean difference = 64.0 (SD 24.0); AB mean difference = 50.0 (SD 17.3)].

An earlier study by Parkinson and colleagues (n = 216)146 used a pre-/postimplantation design and evaluated speech perception in quiet conditions using City University New York (CUNY) sentences and words and HINT sentences, and in noise using CUNY sentences. They found significant positive benefits for cochlear implants at 3 months post implant (p < 0.001 for all measures). Mean change (SD) scores from pre- to post implantation ranged from 34.5 (22.6) for CUNY words in quiet to 67.0 (31.5) for CUNY sentences in quiet.

Kessler and colleagues (n = 238)126 found similar benefits from cochlear implants when they measured outcomes from preimplantation to 12 months post implantation on a range of instruments [Minimal Auditory Capabilities (MAC) vowels and consonants, CUNY sentences, Central Institute for the Deaf (CID) sentences, Northwestern University Auditory Test number six (NU-6), words and everyday sentences listened to over the telephone]. Positive trends were found on all measures; these ranged from a median 36% improvement in score with NU-6 words to a 73% improvement with everyday telephone sentences.

Quality of life

Only the UKCISG study62 measured quality of life, using three instruments [Health Utilities Index 3 (HUI-3), Glasgow Health Status Inventory (GHSI) and the Glasgow Benefit Inventory (GBI)]. This study compared the preimplant scores of 64 people with their 9-month post implant scores. All measures showed a trend towards improvement in quality of life. In particular, the HUI-3 showed that traditional candidates had a significantly better health-related quality of life than marginal hearing aid users after 9 months [mean changes: TC = 0.22 (95% CI 0.19–0.24); MHU = 0.15 (95% CI 0.11–0.19), p < 0.01].

Association with age at implantation

The UKCISG study62 also considered the association between speech perception and quality of life and age at implantation. Participants were divided into six age groups (years): < 30, 30–39, 40–49, 50–59, 60–69, ≥ 70.

There were no strong links between speech perception and quality of life and age at implantation. Pearson correlations revealed that there was a non-significant decline in speech intelligibility as age at implantation increased when measured with the BKB. However, there was a significant increase in benefit with age in audiovisual gain at 9 months post operation [r = 0.164 shown by the AVGN (p < 0.01)]. The HUI-3 and GHSI quality of life measures declined with age at implantation, significantly with the latter measure (r = –0.114, p < 0.05). The GBI did not vary significantly with age.

Association with duration of deafness

The UKCISG study62 showed a stronger effect on speech perception and quality of life with duration of deafness, with greater effectiveness being associated with implantation in the ear with a shorter duration of profound deafness. On all measures effectiveness declined with duration of deafness (r = –0.203, p < 0.01), with a significant difference being found between the group with the shortest duration of deafness and those with more than 30 years of deafness.

Table 27a and b provides a visual summary of these outcomes showing the pattern of results.

Sensitivity to sound

Ching and colleagues (n = 21)148 conducted the only study that measured sensitivity to sound in people with severe to profound deafness. They used a cross-sectional design and measured the ability of participants to detect the direction of sound in quiet laboratory conditions. They found a minimal benefit for cochlear implants by measuring the average root mean squared errors [cochlear implant = 4.5 (95% CI 4.1–4.9), acoustic hearing aid = 4.6 (95% CI 4.3–4.9)].

Speech perception

All studies measured speech perception, using six instruments; the total number of participants was 121.

The UKCISG study (n = 84)62 used a prospective cohort design to measure speech perception with the BKB and AVGN before implantation and 9 months later in people who were profoundly deaf. Marginal hearing aid users were classified on the basis of their score on speech intelligibility tests taken before implantation [mean hearing level = 108.7 dB (95% CI 106.8–110.5)].

Results showed an improvement in scores on both outcome measures at 9 months [BKB: MHU = 44.0 (95% CI 37–51); AVGN: MHU = 31.0 (95% CI 26–37)]. This is the same study as reported in the previous comparison.

Ching and colleagues (n = 21)148 used a cross-sectional design in a small study to measure the same people with cochlear implants or acoustic hearing aids. Before each condition was measured participants used only that type of device in the preceding week. Ching and colleagues used BKB sentences in noise to measure speech perception and found a significant benefit for cochlear implant users (mean scores, cochlear implant = 39, acoustic hearing aid = 2, p < 0.001).

A few years earlier the MED-EL study (n = 63)125 measured speech perception in quiet conditions with HINT and CUNY sentences and in noise with HINT sentences and Consonant Nucleus Consonant (CNC) words. They compared participants’ scores preimplantation with hearing aids to scores with cochlear implants 6 months later. They conducted two sets of subgroup analyses: (1) pre- and postlingually deaf and (2) duration of deafness in postlingually deaf people (more or less than 25 years old). The mean difference (pre/post) for postlingually deaf people in quiet was 62%, with people with 25 years or less of hearing loss showing greater benefit from cochlear implants than those with more than 25 years of hearing loss (≤ 25 years = 71%, > 25 years = 53%). Prelingually deaf participants had a mean benefit in quiet of 20%. In noisy conditions postlingually deaf people with 25 years or less of deafness again did better than those with more than 25 years of deafness (mean scores: ≤ 25 years = 40% and > 25 years = 29% with CNC words).

Hamzavi and colleagues (n = 37)147 used number and monosyllable tests to measure speech perception preimplantation and 12 months later in participants who were severe to profoundly deaf, with cochlear implants or acoustic hearing aids, in a small prospective cohort study. They also measured changes between 12 and 36 months post implant in quiet and noise with the Hochmaier, Schultz and Moser (HSM) sentence test. They found that people with cochlear implants had a mean improvement in pre/post implant scores of 90% whereas over the same time acoustic hearing aid users’ mean scores improved by 37%. The monosyllable word test showed a mean improvement of 43% for cochlear implant users and 19% for acoustic hearing aid users. Over 2 years the HSM scores in quiet improved by 16% for cochlear implant users and 0% for acoustic hearing aid users. In noise, acoustic hearing aid users again showed no improvement over a range of decibels; however, cochlear implant users showed improvement over all levels, ranging from 3.5% at a signal to noise ratio (SNR) of 0 dB to 19.5% at a SNR of 10 dB.

Speech production

The MED-EL study (n = 63)125 measured speech production with CID sentences via a telephone. As with the speech perception results, the MED-EL study found an advantage for those who had been deaf for less than 25 years. Cochlear implant recipients were more able to correctly repeat back uncommon sentences (≤ 25 years = 68% and > 25 years = 42%).

Functional performance

Ching and colleagues (n = 21)148 measured functional performance in real-life situations by giving participants an ad hoc questionnaire after they had used each condition unaided by the other for a week. The questions considered the use of the devices, their performance in quiet and noisy conditions and awareness of environmental sounds. Participants with cochlear implants had significantly higher overall scores than those with acoustic hearing aids [cochlear implant = 59% (95% CI 52–65), acoustic hearing aid = 40% (95% CI 36–44), p < 0.001], indicating greater satisfaction with the functional performance of cochlear implants.

Quality of life

The UKCISG study (n = 84)62 measured quality of life with the HUI-3, GHSI and GBI. This study compared preimplant scores with scores 9 months post implant. All measures showed a trend towards improvement in quality of life [mean scores (95% CI): HUI-3 = 0.15 (0.11–0.19); GHSI = 0.19 (0.16–0.22); GBI = 42.0 (37–47)].

Participants in the MED-EL study (n = 63)125 were given an ad hoc quality of life questionnaire after 6 months with a cochlear implant. Overall, 84% of postlingually and 83% of prelingually deaf people were quite or very positive about the impact of cochlear implants on their quality of life.

Adverse events

Adverse events were measured by the UKCISG53 and the MED-EL study. 125 The UKCISG found that, out of 311 participants, there were 37 adverse events in 27 (9%) participants. Twelve of these events required readmission but did not lead to revision surgery and 25 events did lead to revision surgery. Eleven people had wound infections treatable by antibiotics. Six people had wound revisions, one of which went onto permanent explantation; one person had the device explanted and the other ear implanted; six people needed the device electrodes replacing; two needed the electrodes repositioning; and one needed wound revision. Three became non-users (1%); one was explanted because of complications; one had vertigo; and one had poor non-specified outcomes.

The MED-EL results were taken from all 106 adults implanted in the USA with a COMBI 40+. A total of 22 adverse events occurring in 20 (19%) people were reported. Seven of these were medical and 15 were device related. Only one of these required revision surgery (0.9%).

A visual summary of the results for this comparison is shown in Table 29a–d.

Sensitivity to sound

A total of 44 people in two studies had sensitivity to sound measured in laboratory conditions.

The RCT of Summerfield and colleagues (n = 24)149 measured self-reported spatial hearing, qualities of hearing and hearing for speech (Speech Hearing, Spatial Hearing and Qualities of Hearing questionnaires, SSQ) in adults who either had sequentially received a second cochlear implant or were waiting for one. The scores are an average score for the domain in question with a range of 0–10. They found that there was a significant benefit for spatial hearing at 3 and 9 months post implantation compared with preimplantation [mean difference (SD) scores: 3 months = 1.46 (0.83–2.09), p < 0.01; 9 months = 0.71 (0.08–1.33), p < 0.01]. When the groups’ bilateral results were pooled a stronger effect was seen [3 months = 1.56 (0.95–2.17), p < 0.001; 9 months = 2.00 (1.47–2.53), p < 0.001]. Pooling of the group results showed significant binaural gains for quality of hearing and hearing for speech [quality of hearing: 3 months = 0.9 (0.5–1.3), p < 0.05; 9 months = 0.7 (0.2–1.2), p < 0.05; hearing for speech: 3 months = 6.00 (0.00–12.00), p < 0.01; 9 months = 9.00 (3.00–15.00), p < 0.01].

Verschuur and colleagues (n = 20)151 investigated the ability to detect the direction of sound with either unilateral or sequential bilateral implants. They found that bilaterally aided participants made significantly fewer errors in sound direction detection, however speakers were positioned (mean absolute angular error scores: unilateral = 67°, bilateral = 24°, p < 0.001).

Speech perception

Three studies measured speech perception in a total of 103 participants using seven outcome measures.

Litovsky and colleagues (n = 37)152 used three outcome measures (CNC words and HINT sentences in quiet conditions and BKB sentences in noise) to measure speech perception in simultaneously implanted adults. They found significant binaural gains on all instruments (CNC: left ear 40%, right ear 36%, bilaterally 54%, p < 0.0001; HINT: left ear 66%, right ear 67%, bilaterally 76%, p < 0.0001).

In particular, bilaterally implanted participants were able to use the head shadow effect when in noise. This occurs when speech and noise come from different directions producing a difference in the SNR because of the presence of the head. The mean (SD) head shadow effects were 4.95 dB (3.6) for noise right and 6.34 dB (3.8) for noise left, i.e. a slightly greater effect for noise left. When speech reception thresholds were compared for bilateral implants and either ear unilaterally there was a significant gain for bilateral versus unilateral implants (data not reported, p < 0.0001).

An earlier study similarly evidenced the benefits of bilateral implantation in noise. The RCT of Ramsden and colleagues (n = 29)150 measured speech perception with the CNC and CUNY in quiet and noise in sequentially implanted adults. They found a significant binaural benefit over the first ear alone for speech and noise from the front (12.6 ± 5.4%, p < 0.001) and when noise was ipsilateral to the first ear (21 ± 6%, p < 0.001). No bilateral advantage over the first ear was found in quiet.

Improved speech perception through accessing the head shadow effect was found by Laszig and colleagues (n = 37). 153 They used three tests in this pre/post study with its own controls [the Freiburger monosyllabic word test (FMWT) words and HSM sentences in quiet, and HSM and Oldenburg sentence test (OLSA) sentences in noise]. They found a significant binaural benefit in quiet conditions compared with the poorer unilateral ear alone (mean score: unilateral = 49%, bilateral = 58%, p = 0.00009). In noisy conditions they found a significant head shadow effect, with bilateral advantage greater when the better ear was closest to the speech source than when the poorer ear was closest (poorer ear closest to noise –10 dB and better ear closest to noise –11.4 dB, p < 0.00001).

Quality of life

Quality of life was measured for 54 participants in two studies with five different instruments.

The RCT of Summerfield and colleagues (n = 24)149 measured quality of life with five instruments [GHSI, HUI-3, overall quality of life visual analogue scale (VAS), EuroQol 5 dimensions (EQ-5D) and a tinnitus questionnaire]. At 9 months post implantation they found that scores on the GHSI showed a positively significant result in favour of bilateral implantation [GHSI = 4.00 (95% CI 1.00–0.08), p < 0.05]. Other measures showed neutral or negative mean differences between unilateral and bilateral conditions at 9 months [HUI-3: –0.01 (95% CI –0.1 to 0.08), not significant; VAS: –0.06 (95% CI 0.12–0.00), not significant; EQ-5D: –4.5 (95% CI –12.0 to 3.0), p < 0.05]. These results were coincidental with worsening tinnitus that followed the second implantation (seven out of 16 people who reported tinnitus before the second implant said that tinnitus worsened after the second implant). The reduction in quality of life because of tinnitus reached significance at 3 months (mean score on the tinnitus questionnaire: 12 (95% CI 1.0–23), p < 0.05). Summerfield and colleagues examined these outcomes with multivariate analyses, which showed that the positive gains that came from improved hearing were offset by worsening tinnitus.

Litovsky and colleagues (n = 37)152 used the Abbreviated Profile of Hearing Aid Benefit (APHAB) to measure quality of life. On four of the subscales they found significant gains for bilateral implantation; these ranged from mean scores of 4.4% (p < 0.0001) for reverberant conditions and background noise to 5.7% (p < 0.0001) for communication. Table 31a–c provides a visual summary of these results.

Abandoned initial procedure

Abandoned procedures represent the irreversible failure of an implant operation.

The study by Ray and colleagues158 reported on complications experienced in 844 consecutive implants in a mixture of adults and children, the majority of which were performed after 1994. Only one (0.12%) operation was ‘abandoned’. Similarly, Bhatia and colleagues159 report that in 300 consecutive paediatric implantations no major postsurgical complications occurred perioperatively and only one (0.33%) operation was abandoned.

In an Australian model-based analysis of the cost-effectiveness of cochlear implants for both adults and children, published in 1999, Carter and Hailey160 made provision for 5% of operations to result in some form of complication and assumed a 99% clear-up rate.

Major complications

A major complication is defined as one that leads to revision surgery under general anaesthetic. These may include flap breakdown, cholesteatoma, ear drum perforation, facial nerve damage, persistent infection, meningitis, extrusion of the electrode array or device failure. Revision surgery may also be required to reposition a suboptimally placed electrode array. The surgery to implant the device may mean the loss of residual hearing and it is not possible to predict which patients may suffer such loss. 4

In considering the safety and effectiveness of the COMBI 40+ implant system MED-EL125 collected adverse event data on 106 adults and 82 children. The cumulative implant experiences were 713 months and 533 months respectively. In adults there was one major complication requiring revision surgery. In children there were three major postsurgical complications, two involving resuturing and one explantation. The corresponding complication rates are therefore 1.7 events per 100 patient-years for adults and 6.8 events per 100 patient-years for children.

Two other papers allow the approximation of cumulative implant experience. Proops and colleagues161 reported on complications occurring in 100 adults who received devices as part of the UK-based implant programme. Implants were fitted between December 1990 and May 1996 and the number of operations per year reported. Assuming that all operations occurred in the middle of each year gives an approximate follow-up period of 2638 patient-months. Over that time there were four events reported. The crude rate is therefore 1.8 events per 100 patient-years.

Dutt and colleagues162 report the same type of data for a different group of 100 adults from the same implant programme. The study period was 1999–2001, over which period 122 operations were carried out. This follow-up period is relatively short and assuming that the same number of operations was carried out in each month gives an approximate follow-up period of 2257 patient-months. Three events classified as major postsurgical complications were reported, giving a crude rate of 1.6 events per 100 patient-years.

Fayad and colleagues163 studied the clinical outcomes of children following revision surgery. In total, 28 of the 496 children required some form of revision surgery, leading to an ‘overall revision rate’ of 5.6%. However, without knowing how many children had implants in each year it is impossible to calculate the cumulative implant experience.

Minor complications

Minor complications may resolve with conservative treatment and may include wound infections, flap oedema, haematoma, facial nerve stimulation, tinnitus and temporary vertigo.

In evaluating the safety and effectiveness of the COMBI 40+ implant system125 MED-EL noted that there were 21 ‘minor’ events in adults and 16 such events in children. These correspond to event rates of 35.3 minor complications per 100 patient-years for adults and 34.7 events per 100 patient-years for children.

Meningitis

Before 2003 an increased risk of meningitis associated with cochlear implantation was reported. 164 Summerfield and colleagues165 ascertained that, of 1851 children implanted in the UK before October 2002, none had contracted meningitis and there were no significant differences compared with the general population. Of 1779 adults, five had contracted meningitis, of whom three died. 165 This incidence was significantly higher than that in the general population. For the total UK cochlear implant cohort the incidence rate per 100,000 population was 29 cases (95% CI 9–68), compared with 1.31 per 100,000 population in the general population. 165

Since 2002 the Medicines and Healthcare Products Regulatory Agency (MHRA) in the UK has advised that patients should be routinely vaccinated against pneumococcal meningitis before surgery for cochlear implantation. 166 An international consensus on meningitis and cochlear implants167 reported that since these and other measures have been in place the incidence of meningitis has fallen to ‘its previously low, acceptable level, and may even have fallen below it’. Nevertheless, the risk of meningitis is discussed with prospective implantees or their carers and is therefore included in sensitivity analyses in the PenTAG cost-effectiveness model but not in the base case.

The increased incidence of meningitis among patients with cochlear implants in 2002 was shown to be significantly associated with a particular type of electrode array, which was subsequently withdrawn from use. It may be that the withdrawal of this array, plus careful attention to preoperative vaccination and postoperative intervention with antibiotics in case of middle ear infection, has reduced the incidence of postoperative meningitis worldwide.

A similar investigation to that of Summerfield and colleagues165 was undertaken in the USA by Reefhuis and colleagues168 of paediatric cochlear implant users. The person-years of exposure in this cohort was much larger than that analysed by Summerfield and colleagues (9652 person-years compared with 2478). Reefhuis and colleagues noted 10 cases of meningitis in this cohort, giving an incidence rate of 104 episodes per 100,000 patient-years.

Non-use of devices

Non-use of devices refers to the choice of recipients of cochlear implants to no longer use them for a variety of reasons.

Summerfield and Marshall169 published a paper reporting the incidence of elective non-use among the first cohort of adult patients to receive implants in the UK (n = 313); they found that cumulative elective non-use was stable at 6.3% (95% CI 3.6–9.1%) between 4 and 7 years post implantation but rose to 11.0% (95% CI 1.75–20.3%) at 7.5 years post implantation. Risk factors for non-use were low auditory performance (odds ratio = 8.2, 95% CI 2.1–31.9), low self-reported benefit (odds ratio = 19.6, 95% CI 4.6–84.4) and experiencing a major complication (odds ratio = 3.2, 95% CI 1.0–10.6).

More recently, Bhatt and colleagues170 conducted a retrospective case review of 214 adults who received implants between June 1988 and June 2002. They found that 29 (13.6%) had at some time not used their device for more than 4 consecutive weeks. The cumulative follow-up period was 1126 patient-years. Over that period two people (0.93%) elected for non-use, three (1.40%) became non-users because of co-morbid illnesses and one (0.47%) became a non-user because of audiological complications. Two people (1%) became non-users because of the deterioration of hearing.

Archbold and colleagues171 looked at long-term cochlear implant use in children (n = 138). They found that over 7 years 83% of children wore their implants full-time, 12% most of the time, 2% some of the time and 3% not at all. When the children were classified according to age at implantation they found a significant effect. Those who were full-time users had a median age at implantation of 4.4 years, whereas those who were not full-time users had a median age at implantation of 5.5 years (p = 0.0009). All of the children who were total implant non-users had been implanted over the age of 5 years.

These results are similar to those of Ray and colleagues172 who retrospectively looked at 172 children and 251 adults implanted in the Birmingham programme between 1990 and 2000. They found that five (2.9%) of the children (mean age 11 years) and three (1.2%) of the adults (mean age 42 years) chose not to use their cochlear implants. For children the main reason for non-use was peer pressure and for adults reasons included depression, tinnitus, concomitant neurological problems and non-auditory stimulation.

Non-reimplantation of a cochlear implant during a revision procedure

A cochlear implant may need to be removed for a variety of reasons, for example infection or device failure. Normally, once the problem has been dealt with the ear is reimplanted; however, this is not always the case. Available information concerning reported instances of cochlear implants being permanently explanted is summarised in Table 35.

Failure and replacement of cochlear implant internal components

Maurer and colleagues176 conducted a review of studies looking at the reliability of cochlear implants. However, only two (Von Wallenberg and Brinch177 and Ajayi and colleagues178) of the nine studies reviewed reported the cumulative reliability of devices (Table 36).

Maurer and colleagues176 then looked at the reliability of 192 devices implanted over 11 years (1990–2001) and found an overall cumulative survival of device rate of 91.7% over 11 years.

Conboy and colleagues179 followed 363 devices for 13 years (1989–2002) and found a similar cumulative survival (90.0%) to that of Maurer and colleagues. 176 Data on the number of implants for each year as well as cumulative device survival are reported. Overall, 94.3% of devices survived to 7 years and 90.0% to 13 years. Again, neither confidence intervals nor standard errors were presented for each time point. This makes it difficult to assess whether the differences between devices reported in the two studies are significantly different. This finding is less favourable than that from Lehnhardt and colleagues,180 who reported a cumulative reliability for 16,427 devices over 12 years of 94.9%.

In a review of cochlear implant failures and revisions in 2005, Lassig and colleagues175 found that, in 900 cochlear implant patients from one centre, 27 (3%) underwent revision surgery because of the failure of the internal device.

The Cochlear Europe submission to NICE presented information about the numbers of devices given to children and the cumulative survival for several different devices. The information was reportedly correct as of 30 June 2006, with each graph representing a type of Nucleus^® device based on the receiver/stimulator portion. These data are reproduced in Figure 3 alongside the annual data presented in Conboy and colleagues. 179

FIGURE 3.

Paediatric cumulative survival plots for a range of cochlear implants as reported by Cochlear Europe and Conboy and Gibbin. 179

The Cochlear Europe submission also contains cumulative survival curves for various devices as used in adults. These are reproduced in Figure 4.

FIGURE 4.

Cumulative survival plots for a range of cochlear implants given to adults as reported by Cochlear Europe.

Search strategy

Appendix 1 describes the range of sources searched and the search strategy. The search was limited to English language papers only. Databases were searched from their inception to the most recent date available.

Study selection criteria

The inclusion and exclusion criteria for the systematic review of economic evaluations were identical to those for the systematic review of clinical effectiveness, except that:

• decision model-based analyses or analyses of patient-level cost and effectiveness data alongside observational studies were included

• only full cost-effectiveness analyses, cost–utility analyses, cost–benefit analyses and cost–consequence analyses were included (economic evaluations that report only average cost-effectiveness ratios were included only if the incremental ratios could easily be calculated from the published data)

• stand-alone cost analyses based in the UK NHS were also sought.

Using these inclusion/exclusion criteria, initial study selection was made on the basis of titles and abstracts from the search results by one reviewer (ZL), with unblinded checking by a second reviewer (RA).

Data extraction strategy

Data were extracted by one researcher (ZL) and checked by another (RA) into two summary tables, one to describe elements of the study design of each economic evaluation and the other to describe the main results (see Appendix 6).

For each study the following information was recorded in the study design table: author and year, whether model or trial based, type of model (when relevant), design type (e.g. cost-effectiveness analysis, cost–utility analysis or cost analysis), service setting/country, study population, comparators, research question(s), perspective, time horizon and discounting, main costs included, main outcomes included and sensitivity analyses conducted (see Appendix 7).

In the main results table, incremental costs and benefits as well as the incremental cost-effectiveness ratio (ICER) were recorded for each reported pairwise comparison. Occurrences of either dominance or extended dominance were also noted.

Study quality assessment

The methodological quality of any full UK-based economic evaluations was assessed using the international consensus-developed criteria reported by Evers and colleagues. 181 This formed the basis of a fuller narrative appraisal of these studies. Because of the relatively large number of full economic evaluations discovered we did not conduct a full assessment of the quality of studies from outside the UK.

UK-based full economic evaluation studies

Four of the eight full economic evaluations analysed from a UK perspective involved postlingually deafened adults,53,189–191 one involved children with prelingual deafness190 and the remaining three either failed to report whether the children were deafened pre- or post lingually or contained a mixture of pre- and postlingually profoundly deafened children. 192–194 All of the UK-based full economic evaluations were published between 1995 and 2006. Four of the nine studies are at least a decade old.

No studies were identified in which prelingually deafened adults were analysed from a UK perspective. The clinical and service settings, comparators and basic designs of the eight studies are summarised in Tables 37 and 38.

All of these studies were cost–utility analyses. Five were based on clinical effectiveness results from UK-based cochlear implant programmes. 53,189–191,194 Although the settings in the other three studies190,192,193 were not explicitly reported in the papers it is apparent from related papers that they were also based on NHS treatment settings.

All of the UK-based full economic evaluations used average remaining life expectancy as the time horizon. All of the studies applied discounting to both costs and benefits. None of the included studies was funded by manufacturers of cochlear implants.

UK-based economic evaluations of cochlear implantation in adults

All four studies in adults presented cost–utility analyses and used a decision model to produce cost and utility estimates. All four were based on cochlear implant programmes conducted in the UK NHS (see Table 37). Summary information on the results is shown in Table 39.

All of the four studies examined costs and effects of cochlear implantation as a treatment and reported the cost–utility of cochlear implants relative to either non-implanted or preimplanted adults. None of the studies, however, reported both costs and effects of the comparator.

The earliest UK-based analysis of multichannel cochlear implantation in adults was that conducted by the MRC Institute for Hearing Research, assessing the cost-effectiveness of the technology as used from 1990 to 1994. 191 Using a decision model, over their remaining lifetimes of 26 years, the base-case cost–utility of cochlear implantation was estimated as £11,440 per QALY (with costs and benefits discounted at 6% per year). Although this was higher than cost-effectiveness estimates from US-based studies, this partly reflects the high discount rate used. However, they also speculated that the cost-effectiveness of the technology would improve over time because of longer duration of implant use (with people implanted sooner after a diagnosis of profound deafness, and people also living longer); expected further increases in utility gain because of improvements in electrode and speech processor technology; and efficiency gains in the organisation and provision of services. This was also the first study to publish comprehensive sensitivity analyses.

The UK-based cost-effectiveness analyses published before 2003 serve to highlight the widespread use of health-related quality of life as the only sensible outcome measure for use in cost-effectiveness analyses of the technology; the lack of large well-designed studies of the quality of life impact (and utility gains) associated with cochlear implantation; the critical importance of study perspective (and particularly the potential inclusion of educational cost savings in assessment of paediatric cochlear implantation); the importance of including both device (‘hardware’) costs and postimplantation tuning, rehabilitation and maintenance costs in determining cost-effectiveness; and the paucity of economic studies evaluating bilateral implantation.

Many of these parameters were not adequately empirically assessed until the series of linked studies by the UK Cochlear Implant Study Group (in adults) and Barton and colleagues (in children) published from 2003 to 2006. 22,53,55,62,138,192,195,196 These used unit cost data from the majority of the UK cochlear implant centres, combined with resource use and outcome data from much larger numbers of implant recipients than any of the earlier studies. A fuller description of the costing methods used in these studies is provided later in this chapter (see Resource use estimation, Costs for cochlear implant model).

The only two currently published economic studies of bilateral cochlear implantation were both based in the UK and assessed the technology in adults from an NHS perspective. 149,189 The more recent (2006) study,149 based on a small RCT, was not classed as a full economic evaluation as the exploratory cost–utility analysis forms just a small part of the discussion. The earlier of the two studies189 elicited health-state values from 70 normal-hearing volunteers who were clinical professionals from cochlear implant centres or academics with experience of profoundly deaf people and cochlear implantation. The other study149 obtained HUI-3 scores alongside an RCT of 24 adults, 12 of whom were randomised to receive a second implant immediately, the other 12 subjects having their second implant after a 12-month wait. The unadjusted between-trial arm results of this trial for the HUI-3 outcome at 9 months after the second implant are only reported in diagram form. They appear to show a modest but non-significant increase in utility [approximately +0.11, 95% CI –0.11 to +0.29, whereas the whole group result (i.e. before versus after difference) is approximately –0.01, 95% CI –0.1 to +0.08]. On the VAS and EQ-5D there are small negative differences in quality of life at 9 months after the second implant (with the whole group and between-trial arm analyses).

However, this is a small trial and because these negative impacts on quality of life were largely explained by a few trial participants who experienced worsening tinnitus after their second implant (and because research on unilateral implantation suggests that there is usually an overall positive impact on the prevalence and intensity of tinnitus) the authors decided to adjust for the impact of tinnitus using a regression model. Using this model (Table 1 in the paper), and therefore assuming an overall neutral impact of tinnitus, gives an estimated utility gain from the second cochlear implant of +0.03 (95% CI –0.045 to +0.104). It is this figure that is used in the discussion section to generate an estimate of cost-effectiveness, and this is also the initial estimate used in our model-based analysis of bilateral implantation.

This HUI-3 measured utility gain from the RCT assumes a neutral impact of change in tinnitus, which was not the case in the raw results of this small study (the gain of 0.03 is based on a regression analysis, i.e. after adjusting for the impacts of tinnitus). Also, as second implant recipients in this trial had been unilateral implant users for between 1 and 6 years, these results may not reflect the actual gains of simultaneous bilateral implantation nor those of second implant recipients who have been unilateral implant users for more than 6 years. The cost-effectiveness estimates of bilateral compared with unilateral cochlear implantation from these two studies are £68,916 and £66,600 per QALY, assuming an overall neutral impact of tinnitus on quality of life. However, the authors are duly cautious about these estimates given the weaknesses in the data on utility gains.

The most recent study was by the UKCISG53 and examined both the expected lifetime costs incurred by the UK NHS of providing and maintaining a cochlear implant and the gains in health benefits (measured using HUI-3) associated with implant use. A total of 311 profoundly deaf adults were classified as belonging to one of four subgroups. These subgroups were chosen to represent a progressive relaxation of implant candidacy criteria relating to severity of deafness. Preimplantation HUI-3 utility values were elicited for each of the groups.

The economic evaluation published in 2002 by Summerfield and colleagues189 reported estimates of incremental costs, benefits and cost–utility ratios for both unilateral implantation compared with either non-technological support or hearing aids and bilateral implantation (simultaneous and sequential) compared with unilateral implantation. Regardless of comparator, health-related quality of life was measured using the HUI-2 tool.

UK-based economic evaluations of cochlear implants in children

The remaining five included studies were in groups of children. Although the primary perspective was that of the NHS all also performed some analyses which included education cost savings. One included the cost of support services at home196 and another study costs to the family. 195

Three of the five studies involved either a mixture of pre- and postlingually deafened children or failed to specify their age of onset of deafness, with the other study reporting that only prelingually deafened children were included. 190

The results of the five paediatric evaluations are shown in Table 40.

The earliest UK-based study in children was by Hutton and colleagues193 and was explicitly only a preliminary analysis, primarily of health system costs and potential cost savings when education costs are included. Their cost-effectiveness estimates were only tentative (e.g. using the speculative assumption that cochlear implantation would increase the utility of deaf people from 0.6 to 0.7, and assuming zero costs for the no implantation alternative). Two later studies of paediatric cochlear implantation, by O’Neill and colleagues,186,194 were both based on the Nottingham Paediatric Cochlear Implant Programme; they included the cost savings associated with different postimplantation schooling and educational support needs (the second paper mainly highlighted how regional variations in these costs can critically alter the cost-effectiveness of the technology). With the inclusion of educational savings, and relying on a mean utility gain of 0.23 (extrapolated from adult studies), they estimated that unilateral paediatric cochlear implantation achieved an ICER of £2532 per QALY gained (in 1998 UK pounds). Another early study190 of unilateral cochlear implantation similarly used cost data from the Nottingham Programme and also relied on the assumed average utility gain of 0.23 (from adults); their projected lifetime cost-effectiveness estimates were £15,600 per QALY without educational cost savings, and from £10,000 to £12,100 per QALY when educational cost savings and/or special equipment for daily living were included (all in 1996 UK pounds).

The most recent study, published in 2006 by Barton and colleagues,192 used regression analysis of a large sample of individual patient data – including 403 implant recipients – to examine the gain in health utility associated with the implant. They used a version of the HUI-3 instrument with slightly adapted and simplified wording for the UK context. Utility was modelled as a linear function of preoperative average hearing level, age at implantation, the time period over which gains are accumulated and level of deafness (profound or severe). They combined these estimates of utility gain with comprehensive NHS costs (from Barton and colleagues55) and other cost estimates to produce a range of incremental cost–utility estimates for children at two different ages (3 and 6 years old), for three different levels of preimplantation hearing loss and according to three analytical perspectives (NHS, NHS plus education sector, and ‘societal perspective’).

Finally, two other minor sources of data on paediatric cochlear implantation were identified. One was a brief study by Summerfield and colleagues190 in which paediatric cochlear implantation was discussed alongside a main study on implantation in adults, and the other was a section in the study by Summerfield and Marshall191 in which the costs of paediatric implantation were discussed.

Summary of reviews and other studies

One systematic review of economic evaluations198 and one meta-analysis of a number of cost–utility and cost–benefit analyses185 were also included. Quite unusually, this latter study involved the pooling of cost–utility ratios across a number of studies.

There was also a fairly recent and very comprehensive costing study by Barton and colleagues55 that reported an audit and survey of resource use in all 12 UK cochlear implant centres in 1998/9. This study provided the per patient costs for their economic evaluation and also for our cost–utility analysis later in this assessment report. Another related study195 showed that much of the variation in costs between implant centres could be explained by differences in the volume of implantation activity in that centre. Finally, a study by Sach and colleagues,199 using time-adjusted individual patient outcome and cost data from the Nottingham Paediatric Cochlear Implant Programme from 1989 to 1996, showed that the per patient cost of the programme was reducing over time during this period. This was thought to be partly explained by learning effects and economies of scale.

Age at implantation

A mean implant age of 50 years was used for all postlingually deafened adult cohorts. Prelingually deafened children are assumed to be implanted at 1 year of age. A non-reference case analysis of children implanted at age 8 is also conducted (although separate utility gain estimates for postlingually deafened children are not available).

Simulation

For each comparator, single birth cohorts of either adults or children were modelled independently and results used to produce a deterministic ICER (i.e. using best point estimates for each input parameter). A cycle length of 6 months was used to suitably capture the complexity of the process and to maintain flexibility in the model. The impact of running the model using different time horizons was assessed in sensitivity analysis.

Policy comparisons

The primary research questions investigated in this report are listed in Table 43.

Even though, strictly, ‘no cochlear implantation’ should be compared with three main policy comparators – unilateral cochlear implantation and both types of bilateral implantation (simultaneous and sequential) – we have chosen to break down the decision problem into simpler pairwise comparisons. This is for two main reasons. First, in terms of both costs and effectiveness, unilateral cochlear implantation is inherently intermediate between having no cochlear implant and having two. Second, the current dominant de facto clinical practice in the NHS for the patient groups in our reference case analyses is unilateral implantation. It therefore makes sense to examine the cost and QALY implications of changing policy from this current standard clinical practice.

In addition, we have chosen not to present a head-to-head formal comparison of simultaneous versus sequential bilateral implantation. This is primarily because the difference in QALY gains between these two strategies is entirely due to the difference in age at implantation (and hence life expectancy) when the second implant is put in, rather than to any known difference in the effectiveness between the two strategies of bilateral cochlear implantation. To be consistent with NICE’s principles for the use of social value judgements about age in the development of NICE guidance (Principle 6202) – and in the absence of reliable clinical evidence that outcomes such as speech perception and quality of life are different between children implanted by each method – then any modelled difference in QALY gain between sequential and simultaneous bilateral implantation should be ignored as purely resulting from the difference in age (and life expectancy) at implantation.

Furthermore, although there is a difference in the surgical procedure costs of simultaneous versus sequential bilateral implantation, these are small compared with the initial device hardware and other maintenance costs involved.

Gender distribution

Hospital Event Statistics (HES) have reported cochlear implantation as a distinct category [Healthcare Resource Group (HRG) C60 ‘Cochlear implants’] since 2003/4.

Table 45 shows the number of finished consultant episodes for males and females during the period 2003–6 as well as the average over this period. The data corresponds to HRG v3.5 category C60 and represents the totals for all English strategic health authorities.

Using the 0–14 years category for children, on the basis of the 3-year averages, the male–female ratio is approximately 52:48. A similar calculation can be performed using the age 15+ category as a proxy for adults. The resulting male–female ratio is approximately 41:59.

Starting ages

Severity of baseline hearing impairment

In all base-case analyses individuals are assumed to be profoundly deaf (average hearing level in better ear of > 95 dB). The cost-effectiveness of cochlear implantation in severely deaf individuals is not presented as there are no published utility gain values for this subgroup of cochlear implant recipients, and no studies that would allow them to be reliably estimated.

Benefit from acoustic hearing aids

It was not possible to generate separate parameter values for implant users who would otherwise have been users of either acoustic hearing aids or non-acoustic support because of the lack of research reporting whether hearing aids were used.

We have therefore chosen to match the underlying baseline populations used in each of the cohorts to the real-world situation faced by audiologists. Therefore, we have assumed that, in the absence of cochlear implants, 50% of profoundly deaf individuals will gain benefit from hearing aids (2007, personal communication with expert advisory group).

Unsuccessful candidacy

No published information on the current UK situation was found; however, clinical opinion suggests that around 20% of paediatric and 30% of adult referrals do not go on to receive an implant (Professor Quentin Summerfield, University of York, 2007, personal communication). These values have been used in all model cohorts.

Background mortality

The most up-to-date life tables (2003–5) produced by the UK government actuarial department203 were consulted for gender-specific annual values for inhabitants of England and Wales. Equivalent cycle values were generated and combined using the proportions above to derive the required parameter values.

Rare adverse events

By definition mathematical models are simplifications of reality and, as such, decisions about what not to incorporate into the model have to be made. In representing the patient pathway experienced by cochlear implant users the following events were deemed ‘rare’ and were therefore not included in the base-case analysis: abandoned initial implant procedures, surgical death and the likelihood of contracting meningitis.

For people scheduled to have sequential bilateral implantation, a proportion of candidates may forego the second operation and remain as users of either one cochlear implant or, if the first operation failed, no implants. This event has also been classified as rare and not included in the base-case analysis.

The effect of introducing these variables is explored in sensitivity analyses.

Number of acoustic hearing aids used

All potential implantees (adults or children) are normally provided with two acoustic hearing aids before implantation as part of the assessment process. We have assumed that, in the absence of cochlear implants, of individuals who showed signs of benefiting from these devices, 50% will remain using two acoustic hearing aids and the remaining 50% will use only one acoustic hearing aid.

Furthermore, the assumption has been made that 70% of adults and 80% of children who undergo unilateral implantation use a contralateral acoustic hearing aid (2007, personal communication with members of the expert advisory group]. Clearly, no one who receives bilateral implants continues to use an acoustic hearing aid.

Expected lifetime of an acoustic hearing aid

In an assessment of best practice standards for adult audiology published in 2002204 the Royal National Institute for the Deaf (RNID) stated that the ‘full patient journey’ (assessment, fitting and follow-up) should reoccur every 3 years with an upgrade in technology. Currently, independent sector contracts use a patient journey of 5 years (Jonathan Parsons, Consultant Clinical Scientist, Clinical Director of Audiology, East, Mid Devon and Exeter Area, 2007, personal communication). The assumption has been made that these contracts rather than the guidelines produced by the RNID reflect current practice within the NHS. Therefore, the lifetime of a conventional acoustic hearing aid has been assumed to be 5 years. All individuals in the model who are still alive and users of acoustic hearing aids are given new devices every 5 years.

Failure and replacement of cochlear implant external components

In the absence of any published data, information from the Advanced Bionics submission was used to generate a parameter estimate. This submission reported that around 12% of Auria processors required replacing over a 1-year period. This value was also used in their economic evaluation. Assuming a constant hazard, the expected lifetime of the external components is approximately 7.8 years. The values used in the model are summarised in Table 46.

Failure and replacement of cochlear implant internal components

Analysis of internal device reliability is usually presented in the form of cumulative survival graphs. 176,180,205 Such graphs show the proportions of devices fitted that survive to a particular time point.

Conboy and Gibbin179 present results relating to the reliability of 377 paediatric implantations carried out as part of a UK programme between 1989 and 2002. Data on the number of implants for each year as well as cumulative device survival are reported. Overall, 94.3% of devices survived to 7 years and 90.0% to 13 years.

The Cochlear Europe submission presented information about the number of devices given to children and the cumulative survival for several different devices. The information was correct as of 30 June 2006 and each graph in the submission represents a different type of Nucleus^® device based on the receiver/stimulator.

Information was pooled from both of these sources to generate a survival curve that is representative of all cochlear implant devices currently in use. The resulting curve is shown in Figure 7. The time-dependent probability for internal device failure was calculated from the cumulative survival values using standard formulae. 206 The construction of this curve is explained in detail in Appendix 10.

FIGURE 7.

Cumulative survival for internal components in cochlear implants worn by children.

The same process can be undertaken to generate a combined survival curve for internal failure for cochlear implants worn by adults. Figure 8 shows the resulting survival curve.

FIGURE 8.

Cumulative survival for internal components in cochlear implants worn by adults.

Major complications

In the PenTAG model a major complication refers to any adverse event that is not related to device failure and which results in some form of reoperation. As discussed in Chapter 2 only four studies were identified that contained enough information to derive estimates of the likelihood of an event during a model cycle. The relevant information is summarised in Table 47.

On the basis of information presented in the study by the UKCISG (Figure 3, p. 317) the consequences of our definition mean that the vast majority of major complications are wound related. These require wound revision, electrode replacement and, more rarely, device explantation followed by implantation of the other ear. On the basis of this we have assumed that it was inappropriate to use the same event probability during each cycle and instead have used one probability for major complications during the first year post surgery and another for all years thereafter.

Of the studies in Table 47 two53,125 have an average per individual follow-up period of less than 1 year and the remainder report the number of events that occur within the first year. Therefore, the weighted average of the complication rates for these studies can be used to derive a cycle probability for the first year post implantation. Although Dutt and colleagues162 and Proops and colleagues161 both follow groups of patients for longer than 1 year, not enough information is presented to calculate the probability of major complications after the first year. Therefore, we have simply assumed that the long-term probability is a tenth of the year 1 probability. The values used in the model are summarised in Table 48.

Permanent elective non-use

To incorporate information on elective non-use (reported in Chapter 2, Non-use of devices) we have assumed a trial period during which individuals use their devices before deciding whether to continue or not. We have therefore made the assumption that all non-use occurs after 2 years (i.e. at the start of the third year). Thereafter, all remaining users are assumed to use their implants fully (Table 49).

Device explantation without reimplantation

Because of the small number of studies that report these results for adults and children, the small numbers of events and the wide range of values, the assumption that the probability of permanent removal is the same for adults and children has been made. The value derived from data presented in Table 35 has therefore been applied throughout the model.

Sequential bilateral implantation

Conditional on successful assessment, individuals in the cohort used to model sequential bilateral implantation have two operations scheduled. In a UK-based multicentre study by Verschuur and colleagues151 of 20 individuals who underwent sequential implantation, the mean delay between the two operations was 35 months. The interval used in the model is therefore 3 years.

A summary of the PenTAG model parameters, values and sources is presented in Table 50.

Candidacy/assessment for implantation

Candidacy or assessment costs are all NHS costs incurred between referral to a cochlear implant centre and the day of the implant operation. We used the converted and inflated costs reported in the published studies, as in Tables 3 and 51.

First implantation (unilateral cochlear implant)

The mean NHS cost for ‘implantation of intracochlear prosthesis’ or ‘implantation of extracochlear prosthesis’ has the HRG code of C60. The National Schedule of Reference Costs (NSRC) 2005/6 cost of this inpatient episode is £18,005. However, the NHS Supply Chain agency has also provided us with detailed data on the prices currently paid by the NHS under an NHS purchasing contract for cochlear implants. Depending on the exact cochlear implant model and manufacturer, these prices (for ‘applicable national price bands’) vary from £12,250 to £15,550 for single implant systems. Within this contract (for one manufacturer’s products) there is also a single price for two full implant systems for bilateral implants of £18,375.

There is therefore a choice between using the NSRC cost for the HRG code for cochlear implants and using separate estimates for the costs of the devices and the costs of preoperative, operative and perioperative procedures and care. To retain more flexibility we have decided to use current device costs as provided by the NHS Supply Chain and the converted and inflated costs from the two UK costing studies described above.

All but one of the cochlear implant systems are for use in either children or adults. One of the DIGISONIC products from Neurelec is intended only for use in children under the age of 3 years.

Bilateral cochlear implantation

In the reference case analysis we assume that bilateral implantation requires two complete (unilateral) cochlear implant systems and therefore the device costs are twice the device costs of a unilateral implant. However, it is current practice for all four of the manufacturers that sell cochlear implant systems to the UK NHS to offer price discounts when two systems are being implanted in the same person (information supplied by manufacturers and also suggested in the joint submission to NICE from BAA/BCIG/ENT UK). Nevertheless, the continued presence and size of these discounts in the future is impossible to guarantee and so we have decided initially to assess the technology on the basis of those prices that are contractually agreed with the NHS (via the NHS Supply Chain).

Therefore, for both simultaneous and sequential bilateral implantation, the device costs are twice those for a single implant system (£14,611 × 2 = £29,222). The cost of the implantation procedure and hospital stay is assumed to be incurred twice for sequential cochlear implantation (£5628 in adults, £6960 in children, derived from the two previous UK costing studies53,55). However, although for simultaneous bilateral implantation only one surgical procedure and hospital stay is required, we assume that these costs are 50% higher than for unilateral implantation (£4221 in adults, £5220 in children), mainly because of the additional time in surgery.

With regard to preimplantation assessment costs we assume that for either simultaneous or sequential bilateral implantation these costs are incurred only once and at the same level as for unilateral implantation. However, the cost of tuning and rehabilitation (in the first year after implantation) is assumed to be incurred after each implantation operation and is therefore incurred twice for sequential implantees. The long-term costs of routine maintenance (4+ years post implantation) are assumed to be the same whether people have one or two cochlear implants, although the risks of device failures and major complications are doubled in those using two implants (see below).

Device tuning and other early postimplantation costs

In the first year after a successful operation to implant a cochlear implant the recipient requires various specialist appointments during which the devices themselves are adjusted and the person is further assessed and ‘trained’ to maximise their capacity to benefit from the implants, for example in terms of speech perception and other goals.

We have used the costs of tuning and other care in the first year post implantation from the two recent UK-based studies,53,192 as cited at the beginning of this section. After inflation and conversion to 2005/6 UK pounds these NHS care costs in the first year after implantation are £9148 in children and £5000 in adults.

Routine maintenance costs

The routine costs of device maintenance used are those derived from the two previous UK costing studies53,55 (from year 4 onwards post implantation: £1364 per year for children and £599 per year for adults). The only exception is that, in generating cost–utility ratios for all paediatric subgroups, the model assumes that children will at some point incur the lower annual costs of device maintenance and hearing support which adults experience. In our model, from the age of 16 years, children incur the annual adult cost (£599) for the remainder of their lives as cochlear implant users rather than the estimated annual cost for children (£1364).

Device failure – internal

In the model internal device failures were attributed the mean NHS cost of a replacement implant device (electrode) (£14,498) plus the operation costs to implant it (£2814 in adults, £3480 in children).

The internal component of a cochlear implant is under warranty for free repairs and/or replacements (information supplied to NICE by manufacturers) and therefore separate costs need to be used for the periods of time inside and outside the warranty.

During the first 10 years after initial implantation all devices are assumed to be within warranty and therefore upon failure individuals only incur the costs associated with implantation. Thereafter, during each model cycle a proportion of internal failures are assumed to be in warranty and the remainder not (and hence incurring the full cost of replacement). The proportions used were derived using the relevant event probabilities in adults and children.

Device failure – external

Similarly, the external component of a cochlear implant is also under warranty for free repairs and/or replacements, with the warranty period being 3 rather than 10 years (information supplied to NICE by manufacturers).

During the initial warranty period we have assumed that all replacements incur no cost; thereafter, a proportion incur the full NHS cost of a replacement speech processor (£4114) and the remainder do not. These proportions were again calculated on the basis of the relevant event probabilities for adults and children.

Major complications

Major complications are defined for our modelling purposes as those requiring a reoperation at the implantation site but not associated with a device failure. Most complications are wound related; more rarely they result in operations to reposition the electrode or receiver/stimulator. We estimated the cost of these on the basis of data on wound-related complications in adults from the UKCISG study53 (specifically, data presented in Figure 3 of that paper).

For 21 out of the 311 patients in this study a profile is provided of complications that required treatment (e.g. a course of antibiotics); these included wound revision, electrode repositioning, electrode replacement (functioning electrode but wound-affected) and in some circumstances cochlear implant removal and a new cochlear implant in the other ear. We calculated a weighted average of these reported costs (inflated to 2005/6 prices and converted from euros to UK pounds), except using current reimplantation and device costs (as described in the previous section).

The resultant costs of treating major postsurgical complications were £7777 in adults and £7935 in children for unilateral implantees and £6117 in adults and £6212 in children for bilateral implantees (for bilateral implantees complications are, on average, slightly cheaper to treat because implantation in the other ear is not an option).

Speech processor upgrades

These are assumed to take place every 10 years and attract the same cost as a replacement external processor due to device failure (£4114).

Digital hearing aids

In the model, digital hearing aids may be used either in conjunction with cochlear implants or by deaf people in the absence of cochlear implantation. The cost (2007 prices) to the NHS of a moderate-power digital hearing aid varies from £68 to £118, and the cost of a high-power digital hearing aid from £105 to £152. As there are a vast number of products, many with different prices, we have made the reference case assumption that on average they cost £100 each and are replaced every 5 years. (We have not taken into account the cost of hearing aid batteries supplied by the NHS because they are relatively inexpensive.)

Adults

The best study that estimates the utility associated with being a severely or profoundly deaf adult is that by the UKCISG,53 which elicited values from 311 postlingually deafened adults who completed the HUI-3 instrument both before and after cochlear implantation (Table 55). Alternative possible sources that were less suitable were studies by Summerfield and colleagues189 [smaller sample (n = 202), and HUI-2 instrument], Francis and colleagues208 [smaller sample (n = 47) and older sample, retrospective assessments, algorithm for utility derivation not stated], Wyatt and colleagues209 [smaller sample (n = 32), US sample], Lee and colleagues210 [small sample (n = 11), Korean implantees, retrospective assessments, HUI version not stated] and Krabbe and colleagues211 [smaller sample (n = 45), Netherlands sample, retrospective assessments, HUI-2 used].

In the UKCISG study53 the age at time of implant ranged from 18 to 82 years in the whole group. Although their AHL in the better-hearing ear ranged from 85 dB HL to 140 dB HL, nearly all were profoundly deaf; the mean preoperative AHL in their better ear ranged from 119 dB HL (95% CI 117.7–121.3 dB HL) in ‘non-benefitting traditional candidates’ to 107.4 dB HL (95% CI 104.3–110.6 dB HL) in ‘scoring marginal hearing aid users’ (Table 4, p319, in UKCISG 2004 REF CEA I paper61).

Children

The best study that estimates the utility associated with being a deaf child is that by Barton and colleagues,192 which elicited proxy values from the parents of a representative sample of hearing-impaired British children using an adapted version of the HUI-3 instrument (Table 56). We could not find any studies that had tried to directly elicit utility values from deaf children. Alternative possible sources were less suitable, either because they used estimated values for the hypothetical pure state of ‘being deaf’ (on the HUI instrument) or because they were based on much smaller samples of children in the USA.

Adults

The best study that estimates the utility associated with unilateral cochlear implantation in deaf adults is that by the UKCISG,53 which elicited values from 311 postlingually deafened adults who completed the HUI-3 instrument both before and after cochlear implantation. Alternative possible sources that were less suitable were rejected for the same reasons as already listed in the previous section.

Table 57 shows the mean utility at 9 months post implantation and the resultant mean change in utility compared with preimplantation. All improvements in utility were statistically significant at the 95% confidence level. On average, traditional candidates were older (mean age 52.5 years) than the marginal hearing aid users (mean age 46.3 years) at the time of implantation.

A recently published study by Damen and colleagues212 is the first to have evaluated long-term changes in quality of life following cochlear implantation. In a group of 37 implant recipients followed for 6 years, and using a number of different quality of life measures, including the SF-36 and HUI-3, they showed that the health-related quality of life of adult implant recipients appears to decrease slightly over time, although this may reflect ageing rather than any supposed diminishing benefits of cochlear implant use.

Children

The study by Barton and colleagues192 of the cost–utility of paediatric cochlear implantation in the UK provides the most relevant utility estimates for this analysis. The parents of 403 profoundly deaf children with unilateral cochlear implants completed a modified HUI-3 instrument according to their perception of their children’s health-related quality of life, together with the parents of 549 profoundly deaf children and 464 severely deaf children without cochlear implants. The responses in relation to the implanted children yielded a mean post implant utility of 0.575 (95% CI 0.553–0.598). This utility weight was intermediate between the raw utility weights of 0.616 for children with severe deafness (AHL 71–95 dB HL) and 0.497 for those with an AHL between 96 and 105 dB HL. The mean preoperative AHL of children with implants was 115 dB and approximately 93% had an AHL of between 100 dB and 130 dB (i.e. they were nearly all profoundly deaf before implantation).

However, linear regression analysis of these data including child-specific data on age, age at onset of hearing impairment, severity of preoperative hearing impairment and years since implantation reveals considerable variations in the estimated utility gain from implantation. The main results of this regression analysis are shown in Table 58. The study shows that the estimated utility change due to cochlear implantation, even in profoundly deaf children (AHL ≥ 105 dB HL), varies considerably from a non-significant (p > 0.05) increase of 0.066 to a significant increase of 0.232, depending on preoperative AHL, number of years of use and age at implantation. However, amongst profoundly deaf children who have been implant users for more than 4 years the estimated utility gain is at least 0.183.

No regression analysis was conducted involving cochlear implant variables stratified by age at onset of hearing impairment and so the possible different utility gains for postlingually deafened children were not specifically estimated. The results shown should be regarded as relating to children who became deaf prelingually [of those implanted at < 5 years of age less than 2% became deaf while aged 4 years and only 10% of those implanted at age 5 years or over became deaf after the age of 3 years (using data from Table 4 in Stacey and colleagues21)].

Adults

Only two published studies have assessed the utility of bilateral cochlear implantation. Summerfield and colleagues189 used the time trade-off method to elicit values from 70 normal-hearing volunteers, all of whom had familiarity with deaf adults and cochlear implantation in a professional capacity (either clinicians working in the UK adult cochlear implant programme or staff at the MRC Institute of Hearing Research). A more recent study, also by Summerfield and colleagues,149 randomised 24 adult unilateral cochlear implant users to receive a second cochlear implant either immediately (12 users) or 12 months later (12 patients). At 9 months after bilateral implantation there was only a small and non-significant difference in HUI-3 estimated utility between the bilateral and unilateral groups, of +0.030 (95% CI –0.045 to +0.104). This very modest utility increment, although based on a small sample of actual implant users, is remarkably similar to the utility increment estimated from the time trade-off exercise with normal-hearing volunteers in the earlier study (+0.031). However, it should be noted that the (HUI-3) utility gain of +0.03 estimated from the RCT assumes a neutral impact from changes in tinnitus; we believe this is justified because a larger body of evidence about the impact of unilateral implantation on tinnitus experience implies reductions in tinnitus are more likely. Also, changes in utility at 3 months and 9 months on the EQ-5D and VAS were neutral or negative.

Although the population in the RCT had been unilateral implant users for between 1 and 6 years, we have assumed that the utility gain estimate from this study more closely applies to simultaneous bilateral implanted adults and those adults sequentially implanted in relatively close succession (3 years in our base case). Unfortunately, neither of the two studies that report utility estimates for bilateral implantation in adults report empirical data from deaf adults who received their second implant more than 5 years after the first implant and so there are no utility gain estimates for this group of potential bilateral implant recipients.

Children

We could not find any published studies evaluating the impact of bilateral cochlear implantation on the quality of life of deaf children; therefore, we assume the same value as for adults, +0.03.