Cheng AK, Niparko JK. Cost-Utility of the Cochlear Implant in AdultsA Meta-analysis. Arch Otolaryngol Head Neck Surg. 1999;125(11):1214-1218. doi:10.1001/archotol.125.11.1214
Copyright 1999 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1999
To conduct a meta-analysis of the cost-utility of the cochlear implant in adults.
MEDLINE literature search, review of article bibliographies, and consultation with experts.
Studies that reported (1) data on adults (age ≥18 years) with bilateral, postlingual, profound deafness; (2) a health-utility gain from cochlear implantation on a scale from 0.00 (death) to 1.00 (perfect health); (3) a cost-utility ratio in terms of dollars per quality-adjusted life-year (QALY); and (4) at least 1 conventional statistical parameter (ie, SD, 95% confidence interval [CI], or P value).
From each study, we extracted the number of subjects, study design, health-utility instrument used, health-utility associated with profound deafness, health-utility gain from cochlear implantation, cost-utility of cochlear implantation, and reported statistical parameters.
Weighted averages were calculated using a statistical weight of 1 per variance. Pooling 9 reports (n=619), the health-utility of profoundly deaf adults without cochlear implants was 0.54 (95% CI, 0.52-0.56). Pooling 7 studies (n=511), the health-utility of profoundly deaf adults after cochlear implantation was 0.80 (95% CI, 0.78-0.82). This improvement of 0.26 in health-utility resulted in a cost-utility ratio of $12,787 per QALY.
Profound deafness in adults results in a substantial health-utility loss. Over half of that loss is restored after cochlear implantation, yielding a cost-utility ratio of $12,787 per QALY. This figure compares favorably with medical and surgical interventions that are commonly covered by third-party payers in the United States today.
APPROXIMATELY 600,000 to 1 million individuals in the United States are profoundly deaf and have hearing loss so debilitating that they obtain little to no benefit from conventional hearing aids.1 The multichannel cochlear implant often confers substantial benefit to these individuals; as of 1997, approximately 20,000 individuals had received cochlear implants.2,3
However, the cochlear implant is an expensive technology; the implant device itself costs from $14,027 to $37,572 (The Listening Center at Johns Hopkins University, Baltimore, Md, unpublished data, May 1999). In the United States, patients often endure a lengthy, challenging process in order to obtain third-party payment; in some cases, reimbursement is significantly below cost. In this era of health care reform, there is increasing scrutiny and pressure on the medical profession to provide both effective and cost-effective care. Evidence regarding effectiveness and cost-effectiveness now influences policy decisions on reimbursement and prioritization of health resources.4
Cochlear implants belong to a class of medical interventions that have no significant effect on survival but provide substantial benefit in terms of improved quality of life. To assess the value of all medical interventions on a uniform scale, whether the benefit accrued is additional years of life or improved quality of life, cost-utility analysis is useful. A special form of cost-effectiveness analysis, the cost-benefit ratio1 is expressed in terms of quality-adjusted life-years (QALYs); QALYs are defined as life-years weighted by a quality-of-life factor. This factor, when quantitatively expressed on a linear scale from 0.00 (death) to 1.00 (perfect health), is termed health-utility.5 For example, 20 years of life at a health-utility of 0.50 would be equivalent to 10 years of life at a health-utility of 1.00.
A number of cost-utility studies6- 17 have now been performed for the cochlear implant in adults. The majority of these studies have concluded that the cochlear implant compares favorably with other accepted health interventions, but the range of results was considerable: a health utility gain of 0.07 to 0.41 and a cost-utility ratio of $7405 to $31,711 per QALY. In the United States, England, and Canada, health interventions with a cost-utility ratio of less than about US $25,000 per QALY are considered to be acceptably cost-effective.18 A quantitative synthesis of all cochlear implant studies would thus be a timely endeavor. Furthermore, different studies have used disparate scaling methods. Only some of the analyses reported actual data from patients; others simply estimated how cochlear implant users would respond. The studies included few or no controls, and some cohorts of patients have been used in multiple analyses. To help resolve these discrepancies, we have performed a meta-analysis of the cost-effectiveness of cochlear implants in adults.
A comprehensive literature search of the MEDLINE database (January 1966 through May 1999) was performed by using the following key words: cochlear implantation or cochlear implant or cochlear implants and cost-benefit or cost-effectiveness or cost-utility or cost analysis or quality-adjusted life-years. The search yielded 17 journal articles, all in English, which were then reviewed. Other search techniques did not reveal any additional relevant articles. Of these 17 articles, 9 pertained to cost-utility of the cochlear implant. Additional studies were sought by cross-checking the reference section of each retrieved article; this search identified 7 more reports in book chapters, including 6 in 1 book.14 Abstracts from meetings relevant to cochlear implantation were reviewed, resulting in 1 more report. Consultation with experts produced 1 unpublished report. Excluding 4 duplicate reports, 14 unique studies on the cost-utility of the cochlear implant in adults were further analyzed.
The following inclusion criteria were used: (1) reporting of patient data on adults (age ≥18 years) with bilateral, postlingual, profound deafness; (2) a health-utility gain from cochlear implantation on a scale from 0.00 (death) to 1.00 (perfect health); (3) a cost-utility value in terms of dollars per QALY; and (4) at least 1 conventional statistical parameter (SD, 95% confidence interval [CI], or P value).
Only 7 of the above 14 studies report actual patient data.8,9,14- 17 The other studies performed "theoretical mapping" (ie, hypothetically estimating how deaf patients would answer a health-utility instrument). These studies were not included in the meta-analysis (Table 1).
Prospective studies are traditionally recognized as preferable to retrospective studies because they eliminate recall bias. However, 2 studies15,17 obtained virtually the same results using the Health Utilities Index–Mark III (HUI),19 whether collected prospectively or retrospectively. There may be a lack of recall bias for cochlear implant users, since they turn off their implants on a daily basis at bedtime and when bathing (and lose their use when the batteries run out). Thus, they continually reexperience what it is like to be profoundly deaf. We have used this information to justify the pooling of retrospective cochlear implant data with that acquired prospectively. Sensitivity analysis was also performed by stratifying prospective and retrospective studies (Table 2).
One prospective study reported data on only 7 patients.8 We elected to include this study of small sample size because it was prospective and used a different health utility measure (the Quality of Well-being Scale20) than the other studies. Because of the high variability associated with such a small sample size, sensitivity analysis was performed excluding this study (Table 2). If a cohort of patients was analyzed using 2 different instruments, the meta-analysis weighted each analysis by half (ie, statistical weight=0.5×[1/variance]). When the same data were published twice, only 1 report was included.
The abstraction process was not blinded to the journal, year of publication, or authors. The data abstracted were (1) sample size of patients; (2) sample size of controls, if any; (3) prospective or retrospective study design; (4) health-utility instrument used; (5) health utilities reported, including SD or 95% CI; and (6) cost-utility ratio (dollars per QALY) reported.
The meta-analysis calculated weighted averages for (1) the health-utility loss from profound deafness in adults, (2) the change in health-utility reported by adult cochlear implant users, and (3) the cost-utility ratio (dollars per QALY). A summary effect size for each was derived using the "fixed-effects model to a continuous measure" method.21,22 Sensitivity analyses were conducted to assess the robustness of the meta-analysis by varying the inclusion criteria (Table 2 and Table 3).
A comprehensive search of MEDLINE followed by cross-checking citations and consulting with experts yielded 14 unique studies, of which 7 presented actual patient data.8,9,14- 17 Within the 7 studies, there were 9 reports of cases or controls with a loss of health-utility from profound deafness. Controls were defined as adults with profound deafness who had not received a cochlear implant. They were on the waiting list to receive an implant, were rejected as an implant candidate for medical or insurance reasons, or did not wish to receive an implant. Pooling these 9 reports (n=619), the health-utility of profoundly deaf adults without cochlear implants was 0.54 (95% CI, 0.52-0.56) (Table 4).
Pooling the 7 studies (n=511), the health-utility of profoundly deaf adults after cochlear implantation was 0.80 (95% CI, 0.78-0.82) (Table 5). This improvement of 0.26 in health-utility resulted in a cost-utility ratio of $12,787 per QALY for cochlear implantation (Table 6). The 7 theoretical mapping studies (n=0) are summarized and compared in Table 1.
Sensitivity analyses were performed to test the robustness of the meta-analysis by varying the inclusion criteria (Table 2 and Table 3). This demonstrated that the baseline meta-analysis results did not change substantially as inclusion criteria were modified. Statistical pooling of the 2 prospective studies only8,17 yielded a cost-utility ratio of $19,999 per QALY.
This meta-analysis provides a structured quantitative literature review of all health-utility values associated with profound deafness and cochlear implantation in adults. Statistical pooling suggests that the cochlear implant in adults has a favorable cost-utility ratio of $12,787 per QALY.
One limitation of the pooled studies is an inadequate comparison of the health utilities of individuals who have received a cochlear implant and controls who have not. Only 1 study incorporated the use of longitudinal controls. Palmer et al17 evaluated 37 cases and 14 controls prospectively for 1 year. Whereas there was a 0.20 increase in health-utility (from 0.58 to 0.78) in the implanted group, the control group reported no change in the same baseline health-utility with time (from 0.58 to 0.58). Two cross-sectional studies14,15 asked "controls" to determine their health-utility loss from profound deafness but did not evaluate their health-utility values over time. Another potential problem is the possibility of recall bias in the retrospective studies, although this may be mitigated for cochlear implant users as they reexperience profound deafness on a daily basis when their implants are off.
Among studies, results differed depending on which instrument was used to assess health-utility. To some extent, this heterogeneity limits the meaningfulness of statistical pooling. The summary health-utility gain of 0.26 was approximately halfway between those values found in studies using the HUI and those using the Visual Analog Scale (VAS). Of the 7 studies, 4 used the VAS, 2 used the HUI, and 1 used the Quality of Well-being Scale. Other instruments were used in theoretical mapping studies and resulted in more modest health-utility gains. If patients were asked to use these instruments, less cost-effective ratios may have resulted.
The HUI incorporates into its scale a health-utility loss for "complete deafness" of –0.40 (ie, 1.00 – 0.40=0.60), a value derived by the standard gamble method from 532 individuals of the general population in Ontario.19 The VAS is a "feeling thermometer" rating scale in which patients simply rate their own quality of life on a scale from 0.00 (0%) to 1.00 (100%). It is noteworthy that 2 disparate health-utility methods, the HUI (–0.42) and the VAS (–0.47), derived similar utilities for profound deafness. However, the subsequent health-utility gain from cochlear implantation was less similar for the HUI (+0.20) and the VAS (+0.31). The VAS has been criticized for overestimating losses in health-utility from mild disease because the respondent is not forced to make a choice under conditions of uncertainty.23 Also, the linearity of the VAS has been questioned.23 However, the HUI can be faulted for attaching a value to profound deafness based on responses of people who have never experienced deafness; directly eliciting health utilities from individuals who have experienced both profound deafness and normal hearing may provide more meaningful results.
The Panel on Cost-effectiveness in Health and Medicine24 in 1996 was unable to select a criterion standard health-utility instrument. The current recommendation is to use 1 or more instruments in a study and compare results with other studies that used the same instrument(s). Interestingly, no study thus far has directly elicited health utilities from cochlear implant users using the 2 most commonly accepted health-utility methods (standard gamble and time trade-off). This would be a logical next step for future studies. Future studies should also be prospective in nature, evaluating an adequate number of cases and controls longitudinally. Statistical pooling of the 2 prospective studies only8,17 yielded a reasonable cost-utility ratio of $19,999 per QALY.
In summary, this meta-analysis pooled 7 cost-utility analyses of the cochlear implant in profoundly deaf adults. The pooled cost-utility ratio of $12,787 per QALY suggests that the cochlear implant in adults is cost-effective compared with health interventions commonly covered by third-party payers in the United States today, at least when using the VAS and HUI as health-utility instruments.
Accepted for publication July 2, 1999.
We thank Deborah McClellan, MD, Ann Bergin, MD, Norman Beauchamp, MD, Julie Brahmer, MD, Lorna Fitzpatrick, MD, Jeanette White, MD, Rodney Willoughby, MD, and Susan Zieman, MD, for their valuable critique.
Corresponding author: John K. Niparko, MD, The Listening Center, Department of Otolaryngology–Head & Neck Surgery, The Johns Hopkins University, Baltimore, MD 21203-6402 (e-mail: firstname.lastname@example.org).