Estimated Cost-effectiveness of Active Middle-Ear Implantation in Hearing-Impaired Patients With Severe External Otitis

Objective  To determine the cost-effectiveness of middle-ear implantations in hearing-impaired patients with severe external otitis in the Netherlands.

Design  Cost-effectiveness analysis, using single-subject repeated measures of quality of life and total cost determinations.

Setting  Hospital based.

Patients  Moderately to severely sensorineurally hearing-impaired patients (n = 21) with severe chronic external otitis, eligible to receive a middle-ear implant.

Main Outcome Measure  Cost per quality-adjusted life-year (QALY), based on scores of the Medical Outcomes Study Short-Form Health Survey (SF-36) generic quality of life questionnaire. Only direct costs were included in cost calculation of middle-ear implantation.

Results  Mean health utility gain was 0.046 (0.012-0.079) (P = .01) measured at the mental component of the SF-36. With a mean profitable time of 19.4 years and an overall cost of €14 354, minimal cost-effectiveness of middle-ear implantation was €16 085/QALY.

Conclusion  Based on the cost per QALY, middle-ear implantation proved to be a cost-effective and justified health care intervention in the Netherlands.

In patients with moderate to severe sensorineural hearing loss, conventional hearing aid fitting may be troublesome owing to chronic external otitis. Although some patients benefit from vented or silver- or gold-coated ear molds, others continue to experience pain or itching when the ear canal is occluded, which leads to nonuse of the hearing aid. In these patients, the electronic middle-ear implant may form a solution, because the device does not block the ear canal. At present, 2 types of middle-ear implants are commercially available: the Vibrant Soundbridge (Med-El, Innsbruck, Austria)^1-3 and the Otologics MET device (Otologics LLC, Boulder, Colo).^4-6

Compared with conventional hearing aids, the application of the middle-ear implant involves surgery and higher financial cost, which has led to health care policy questions regarding treatment effectiveness and cost-benefit analyses. Cost-effectiveness of a treatment can be determined by combining the qualitative and quantitative health benefit with the cost of the treatment.

To assess the health benefit of a medical intervention, generic health-related quality of life (HR-QOL) instruments are often used. These questionnaires are not disease specific and can therefore be applied to areas not directly affected by a specific disorder. Improvement in quality of life (QOL), also called health utility gain, is expressed on a linear scale from 0.00 (death) to 1.00 (perfect health).⁷ Unfortunately, several studies have shown only small changes in generic HR-QOL questionnaire scores after hearing aid fitting.⁸ An example is the general Medical Outcomes Study Short-Form Health Survey (SF-36) questionnaire that assesses various aspects from physical as well as mental health.^9,10 Although various studies assessing the benefit of hearing aid fitting showed marginally better mental health status, no improvements were noted, and sometimes negative changes were seen in some aspects of physical health, for example physical functioning and bodily pain.^10-12 This is not in accordance with the results of disability and handicap-specific questionnaires that have shown significant improvements in hearing aid benefit in hearing-impaired subjects.¹³ It has been concluded that a major problem with generic HR-QOL questionnaires is their insensitivity to problems concerning audition and communication.^8,10,13,14 Barton et al¹⁴ reported low benefit scores but marked differences between the outcomes of 3 different HR-QOL instruments, which indicates that cost-effectiveness depends on the questionnaire used. The BAHA system (bone-anchored hearing aid), a semi-implantable bone-conduction device, was not found to have any significant effect on QOL when generic instruments were used, whereas disability-specific questionnaires showed significant improvements.¹² In contrast, significant changes in HR-QOL questionnaires were reported in postlingually deafened adults after cochlear implantation.^11,15,16

The relative insensitivity of existing HR-QOL questionnaires to audition and communication problems has led to the development of several new instruments to assess health benefit after hearing aid fitting. The Glasgow Benefit Inventory (GBI)¹⁷ is an HR-QOL questionnaire that was specially developed to measure outcomes of otorhinolaryngologic interventions. It examines how the health status of the patient has been affected by the intervention, including psychological, social, and physical well-being.¹⁷ The GBI has been used successfully to evaluate the Vibrant Soundbridge,³ the BAHA system,^18,19 and recently, cochlear implants in a large population of recipients.¹⁶ The Nijmegen Cochlear Implant Questionnaire (NCIQ)²⁰ is an HR-QOL questionnaire that was specially developed to assess health status after cochlear implantation. It comprises questions in communicative, psychological, and social domains.^11,20 Although the NCIQ has not yet been used widely, it has proven to be reliable and sensitive to clinical changes.²⁰

Up to now, none of the HR-QOL questionnaires has received a general recommendation. Therefore, it remains advisable to use several instruments in parallel to determine postintervention health status.

After the health utility gain has been determined, the number of quality-adjusted life-years (QALYs) can be calculated. A QALY is designed to aggregate the total health improvement in a group of individuals, while capturing improvements in QOL (health utility) and quantity of life.⁷ It is calculated by multiplying the life expectancy after the intervention (years) by the health utility gain due to that intervention.¹⁵ However, the assumption that a subject will benefit from the intervention until death is unrealistic in case of middle-ear implant application. In any subject, hearing can be expected to deteriorate with age, which makes every hearing aid less effective and eventually ineffective at some stage in life. Therefore, life expectancy after the intervention should be replaced by profitable years, ie, the number of years of effective middle-ear implant use.

In the United States, England, and Canada, health interventions with a cost-utility ratio of less than about €20.632/QALY are considered to be acceptably cost-effective.¹⁵ (The assumed US exchange rate, as of October 10, 2005, for determination of this figure was $1.00 = €0.825292.) This is comparable with the ratios of €18 150/QALY and €20 000/QALY calculated in the Netherlands.^21,22 One exception is the British Cochlear Implant Study Group²³ who use an upper limit of acceptability of €50 000/QALY.

Several cost-effectiveness studies have been performed on hearing devices, especially on cochlear implants. Cochlear implantation was shown to be a cost-effective treatment, with cost-utility ratios that ranged from €10 553 and €12 107/QALY^15,24 to €22 283/QALY.²³ At a mean cost of €13 933 and €15 807/QALY, conventional hearing aid fitting was also considered to be cost-effective in elderly patients.^22,25 However, to our knowledge, no research has been done on the cost-effectiveness of middle-ear implantations. The aim of the present study was to determine the cost-effectiveness of middle-ear implantations in patients with sensorineural hearing loss. It was expected that this treatment would be cost-effective in terms of cost per QALY.

The study had a prospective, single-subject, repeated-measures design in which each patient served as his or her own control subject in the QOL reports. Early in the selection phase, prior to being selected for middle-ear implantation, patients filled out 2 HR-QOL questionnaires, the SF-36⁹ and the NCIQ.²⁰ The same 2 questionnaires were also filled out 6 and 12 months after the middle-ear implant audio processor had been fitted. The GBI questionnaire¹⁷ was added to the protocol later and was filled in by the patients at 6 to 24 months of follow-up. As it concerns a retrospective questionnaire, this change in protocol was not seen as a problem. To minimize enthusiasm bias in the patients who received implants, all the questionnaires were sent by mail, and a relatively long evaluation period was chosen. The QALYs were calculated based on the health utility gain reflected in the questionnaires.

To assess whether the audio processor fittings were adequate while dealing with the nonlinear amplification of the audio processors, it was decided to measure amplification for normal speech levels (speech gain) and to compare this gain with target gain values obtained with the well-validated National Acoustic Laboratories (NAL) nonlinear rule,²⁶ which prescribes desired gain as a function of input level based on hearing thresholds alone. Speech gain was defined as the shift between the unaided and aided speech audiograms (speech recognition intensity graph) at the 65-dB sound pressure level input.² Mean NAL target gain values at frequencies of 0.5, 1.0, and 2.0 kHz for an input level of 65 dB sound pressure level were used to compare with measured speech gain.

Differences between baseline and implant-aided scores were analyzed using a paired t test; 2-sided comparisons were considered statistically significant at P<.05. Data are expressed as mean ± SD, while hearing differences between post–middle-ear implantation and baseline are expressed as means (95% confidence intervals [CIs]). SPSS statistical software, version 12.0, was used to make the calculations (SPSS Inc, Chicago, Ill).

The study population comprised 21 middle-ear implant users who had been operated on between January 2000 and May 2004 at the Department of Otorhinolaryngology, University Medical Center St Radboud, Nijmegen, the Netherlands. In 13 patients, the Vibrant Soundbridge had been implanted, fitted with the 404 audio processor, and 8 patients received the Otologics MET device, fitted with the standard button audio processor. There were 9 men and 12 women, mean ± SD age 52.4 ± 13.9 years (age range, 18-79 years) at the time of implantation. All the patients had symmetrical sensorineural hearing loss (54.4 ± 11.4 dB; range, 32-77 dB) and chronic, therapy-resistant external otitis. At some stage before implantation, all the subjects had been using conventional hearing aids for at least 3 months, fitted by an audiologist. Special ear molds were applied, mostly vented, made of hypoallergenic or silver-coated material. Several patients decided to stop using their hearing aids partly or even completely due to the chronic otitis.

Data on 4 more patients were excluded because their characteristics changed during follow-up. One patient manifested a degenerative disease and died; another patient had hereditary diabetes that resulted in sudden symmetrical hearing deterioration and a final hearing threshold poorer than that allowed for middle-ear implantation; a third patient developed an air-bone gap of 30 dB after implantation, and the implant was removed because of ineffective repeated surgery; the fourth patient had serious and unresolved complaints about environmental sounds that have only recently been understood. Thus, the original study group comprised 25 patients, and 4 of these have been excluded.

The SF-36 is a multidimensional outcome instrument to measure QOL. It has been designed to prospectively monitor patient outcomes in medical and clinical research settings.⁹ The questionnaire assesses 8 different health concepts: physical functioning, role limitations due to physical and emotional problems, vitality, mental health, social functioning, bodily pain, and general health perceptions. Because hearing device fitting may not have any direct effect on physical health but may affect mental health, 2 overall scores should be determined.¹⁰ The physical component summary mainly contains physical functioning, role limitations due to physical problems, bodily pain, and general health perceptions. The mental component summary is mainly composed of role limitations due to emotional problems, vitality, mental health, and social functioning. The physical and mental component summary scores are expressed as a value between 0 and 1.

The NCIQ is a prospective questionnaire used to measure QOL. It was specially developed to evaluate how cochlear implantation affects health status.²⁰ The general physical domain consists of 3 subdomains (basic and advanced sound perception and speech production) and mainly focuses on communication (referred to as NCIQ communication). The psychological domain (NCIQ psychological) contains mainly self-esteem questions, while the social domain (NCIQ social) addresses activity limitations and social interactions. Because our subjects had acquired hearing loss, 6 questions on deaf speech and sign language were deleted (questions 8, 15, 33, 39, 56, and 57; see questionnaire in Hinderink et al²⁰). Scores can range from 0 to 100 based on a 5-point scale that indicates the degree of ability in different situations.

The GBI is a retrospective standardized questionnaire that examines the impact of an otologic treatment, such as middle-ear implantation, on the health status of the patient.¹⁷ Scores can range from –100 to +100 on the basis of 5-point scales that measure from severe deterioration to great improvement in health status.

The QALYs were calculated from the health utility gain scores obtained with the QOL questionnaires. The hearing of every subject will deteriorate over time, and any middle-ear implant will eventually become ineffective. Snik et al⁶ estimated the maximum hearing loss for proper middle-ear implant application: 70 dB hearing loss for the Vibrant Soundbridge and 80-dB hearing for the Otologics MET (average at 0.5, 1.0, and 2.0 kHz). Individual profitable usage periods in years can be calculated by subtracting the average hearing loss from the maximum hearing loss and dividing this by an assumed deterioration of 1 dB per year (average at 0.5, 1.0, and 2.0 kHz).²⁷ The formula would be expressed as follows: profitable years Vibrant Soundbridge = 70 dB – average hearing loss in decibels (0.5, 1.0, and 2.0 kHz)/1 dB per year and profitable years Otologics MET = 80 dB – average hearing loss in decibels (0.5, 1.0, and 2.0 kHz)/1 dB per year.

A QALY was calculated by multiplying the number of profitable years with a middle-ear implant by the health utility gain from that specific middle-ear implant.

Treatment cost was calculated according to the methods described by Severens et al²⁸ and only included the direct cost of middle-ear implantation, ie, during the phases of selection, implantation, rehabilitation, and aftercare. It was assumed that the cost of long-term aftercare for today's semi-implantable devices is comparable to that for conventional hearing aid fittings. The cost of personnel was made up of 3 parts: the gross salary of the employee, a 21% raise for the social liabilities paid by the employer, and a 37% raise by the hospital for the facilities and equipment. Reimplantations were not taken into account because these were assumed to be the result of “growing pains” of the new device and expected to become insignificant in number as experience increases. On the basis of the total cost of middle-ear implantation and the value of 1 QALY, we calculated the cost per QALY.

The mean ± SD difference between the measured gain and target gain estimated by the NAL rule was 2.4 ± 6.4 dB (range, –11.5 to 7.6 dB); thus, measured gain was a little higher than NAL target gain. This suggests that the audio processor fitting can be considered as adequate.

All 21 patients completed the SF-36 and NCIQ preoperatively and postoperatively, but only 17 patients (81%) returned the retrospective GBI questionnaire. Table 1 lists the overall results of the SF-36 and the NCIQ at baseline and after middle-ear implantation. Because GBI scores implicate effectiveness by definition, only health utility gain is shown for this questionnaire.

Significant differences were found in the physical and mental components of the SF-36 between post–middle-ear implantation and baseline (P = .05 and P = .01, respectively, for the mean individual difference scores). The mental component summary had higher scores after middle-ear implantation. Social functioning was the only underlying health concept with significantly higher scores after middle-ear implantation (P = .01). Surprisingly, the physical component summary was significantly poorer after middle-ear implantation. Bodily pain was the only subdomain that had just significantly lower scores compared with baseline (P = .05). The communication-related physical, psychological, and social domains of the NCIQ were significantly higher after middle-ear implantation than at baseline (P = .002, P = .01, and P = .001, respectively). All of the underlying subdomains of the NICQ showed significantly higher scores after middle-ear implantation than at baseline (P≤.01 for all subdomains). The mean GBI score was highly significant (P<.001).

Because the SF-36 is the only questionnaire that is widely accepted as a health utility instrument, QALY and cost per QALY values are given only for this questionnaire, using the SF-36 mental component summary score as the health utility gain measure.

The mean ± SD estimated profitable period of middle-ear implant use was 19.4 ± 9.0 years. Multiplying this estimate by the health utility gain score obtained with the mental component of the SF-36 questionnaire produced a QALY value of 0.89.

The overall cost of middle-ear implantation was estimated to be €14 354, based on the costs listed in Table 2. Dividing this amount by the value of 1 QALY, cost per QALY for middle-ear implantation was €16 085 according to the SF-36.

The present prospective study addressed the cost-effectiveness of middle-ear implantation in hearing-impaired patients with severe external otitis and is the first to our knowledge to show the cost-effectiveness of middle-ear implantation, although the treatment has been applied for almost a decade. These results can be expected to encourage approval and justification of middle-ear implantation in the future.

Since no other QOL and cost-effectiveness data were available on middle-ear implantation, comparisons were made with studies on cochlear implants. It must be noted that the target populations for the 2 hearing aids are not equal. The middle-ear implant is used in subjects with sensorineural hearing loss, while the cochlear implant is offered to deaf subjects. When we compared the present SF-36 (mental component score) and NCIQ results obtained after middle-ear implantation with those after cochlear implantation in adults at the same research institute,^11,20 improvement with cochlear implantation was found to be from about 2 to 2.5 times higher than that with middle-ear implantation (Table 3) (the SF-36 physical summary score was not considered because no change in physical functioning was expected to occur as a result of middle-ear application). This implies that the middle-ear implant for hearing-impaired subjects is about 2 times less effective than a cochlear implant hearing device for deaf subjects when measured in terms of QOL. However, middle-ear implantation is also cheaper than cochlear implantation.

It was remarkable that a deterioration in the physical component of the SF-36 questionnaire was observed after middle-ear implantation, which was mainly caused by the increase in the subdomain bodily pain. Such a negative change in physical health status had been reported earlier in the cochlear implant study by Krabbe et al¹¹ (Table 3) as well as in the BAHA system and, surprisingly, with the conventional hearing aid.¹⁰ Worsening of physical health has a neutralizing effect on the overall SF-36 outcome. Increased bodily pain might be a consequence of the event of surgery or of the presence of any type of implant (cochlear implant, middle-ear implant, or BAHA system). However, the direct causes of the unfavorable scores with respect to pain are unknown and should be clarified to diminish this adverse effect.

Our GBI scores were comparable with those obtained in other studies. After cochlear implantation, GBI scores were reported to be 40¹⁷ and 43,¹⁶ while BAHA treatment resulted in GBI scores of 31¹⁸ and 33.¹⁹ Thus, our value of 34 for middle-ear implantation was comparable with that reached with the BAHA system and close to that reached after cochlear implantation. Sterkers et al³ reported a low GBI score of only 17 after Vibrant Soundbridge treatment in a multicenter study. It is speculated that the difference between this low result and the present results is caused by the baseline situation of the participants. For the study by Sterkers et al, patients were dissatisfied conventional hearing aid users who did not necessarily have external otitis.

The relatively higher health utility gain scores obtained with the GBI than with the other questionnaires might be explained by the design of the GBI. Compared with most questionnaires that are prospective and have to be filled out twice before utility scores can be calculated, the GBI is retrospective and therefore provides a direct measure of the effectiveness of an intervention. It will be less easily influenced by the patient's mood of the day.

To summarize the qualitative effect of middle-ear implantation, the impact on quality of life was smaller than that with cochlear implantation but about equal to that with the BAHA system.

The estimated overall cost of middle-ear implantation was based on calculations by Severens et al,²⁸ who studied the cost of cochlear implantation in the Netherlands. In 2004, the official tariff for cochlear implantation in adults amounted to €47 500, based on the data reported by Van den Broek.²⁹ In comparison with the estimated overall cost in this study, middle-ear implantation was 3.3 times cheaper than cochlear implantation. However, the differences in cost per QALY were less evident between the 2 types of implant because middle-ear implantation was about 2 times less effective than cochlear implantation, according to the questionnaire scores (Table 3).

Studies on the cost-effectiveness of cochlear implantation showed amounts ranging from €12 107/QALY^15,24 (data from 1999) to €22 283/QALY²³ (data from 2004), and cochlear implantation was found to be cost-effective. With a cost-utility ratio of €16 085/QALY, middle-ear implantation can also be considered a cost-effective intervention.

The cost per QALY might be criticized because we used the SF-36 mental health score, not the overall score. However, the comparison made with cochlear implantation showed that middle-ear implantation is at least as cost-effective as cochlear implantation.

Some methodologic remarks can be made about the study. The mean difference between the speech gain measurements and the NAL target gain was small. This implies that the audio processor fittings of the middle-ear implant can be considered adequate, which is a prerequisite for a health utility study. In addition, owing to the limited number of subjects per type of middle-ear implant (either the Vibrant Soundbridge or the Otologics MET), we decided to pool our data. A previous study on the gain and maximum output of these 2 devices showed some differences, especially in the maximum output.⁶ However, the overall gain and frequency response of the 2 devices were highly comparable, which also applies to the purchase price in the Netherlands. Wilcoxon rank sum tests did not show any significant differences between the changes in SF-36 and NCIQ results or between the GBI scores of the 2 groups. Therefore, in retrospect, pooling these data seems justified.

Regarding the choice of the HR-QOL questionnaires in the present study, most of the HR-QOL questionnaires are insensitive to audition and communication problems that are often present as a result of diminished hearing or hearing aid fitting. Therefore, we decided to use the additional NCIQ and GBI questionnaires, which include questions on these topics. A disadvantage of these questionnaires is that they were not designed to calculate the cost per QALY in that an overall utility score has not yet been defined for these questionnaires.

In conclusion, this study was the first to our knowledge to address cost-effectiveness of middle-ear implantation. With an estimated cost of €16 085/QALY according to the SF-36 QOL questionnaire, middle-ear implantation proved to be a cost-effective and justified health care intervention for the treatment of hearing-impaired patients with severe external otitis. Additional support came from a comparison with a previous study on QOL in cochlear implant users; middle-ear implantation was 2 to 2.5 times less effective but more than 3 times cheaper than cochlear implantation, resulting in a more favorable cost-effectiveness ratio.

Correspondence: Ad F. M. Snik, PhD, Department of Otorhinolaryngology, University Medical Center St Radboud Nijmegen, PO Box 9101, 6500 HB Nijmegen, the Netherlands (a.snik@kno.umcn.nl).

Submitted for Publication: January 10, 2006; final revision received April 23, 2006; accepted May 31, 2006.

Author Contributions: Dr Snik and Ms Duijnhoven had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Snik and Cremers. Acquisition of data: Duijnhoven. Analysis and interpretation of data: Snik, Duijnhoven, and Mylanus. Drafting of the manuscript: Snik, Duijnhoven, and Cremers. Critical revision of the manuscript for important intellectual content: Snik and Mylanus. Statistical analysis: Duijnhoven. Study supervision: Snik, Mylanus, and Cremers.

Financial Disclosure: None reported.

Acknowledgment: We thank Madeleine Wiltingh, Carine Hendriks, and Thomas Snik for data collection.