|
|
| |||||||||||||||||||||||||||||||
Background Cochlear implants restore some degree of hearing in patients with severe hearing impairment, but the efficacy of different implants has not been compared. We conducted a prospective trial to compare several cochlear implants.
Methods We studied 82 patients who were randomly assigned to receive one of three cochlear implants: the Ineraid multichannel implant (implant 1), the Nucleus multichannel implant (implant 2), and the 3M/Vienna single-channel implant (implant 3). All the patients had profound deafness, and none had derived benefit from hearing aids. The assigned device was successfully implanted in 80 patients. Twenty-four hearing tests were used to assess the patients' performance before implantation and 12 and 24 months after implantation. The tests were grouped into five categories according to their content, and a weighted composite index was developed to provide a single numerical indicator of the overall auditory response.
Results All the patients were able to hear with their implants. Nineteen of the 30 patients (63 percent) who received implant 2, 18 of the 30 patients (60 percent) who received implant 1, and 1 of the 20 patients (5 percent) who received implant 3 were able to distinguish some words and sentences. The scores for the composite index were similar in the patients who received implant 1 and those who received implant 2, and were higher in both these groups than in the patients who received implant 3 (P = 0.02). When 24 patients with implant 2 were given an improved speech processor, their composite index increased significantly within 3 months (P<0.001); their score at that time was also significantly higher (P = 0.04) than the score of the patients with implant 1 at 24 months. Age at implantation, lip-reading ability, and IQ were prognostic indicators of the patients' performance with a cochlear implant.
Conclusions Multichannel cochlear implants are superior to single-channel implants, especially for understanding speech. Changes in speech processing can improve patients' performance.
These approaches have led to the development of a multiplicity of implants, many of which have been abandoned. Implants reported to allow at least some recipients to understand some speech without lip reading are the 3M/Vienna single-channel implant, the Nucleus 22-channel implant, and the Smith and Nephew Richards Ineraid 4-channel implant1,2,3,4. Although these devices differ from one another in concept, design, number and shape of electrodes, manner of stimulation, and speech-coding methods (Table 1), all have produced similar results in small numbers of patients. One study comparing some of these implants has been reported,5 but it was not randomized or prospective.
|
Methods
Study Design
We enrolled only adults with postlingual deafness who had obtained no benefit from amplification devices (i.e., they had increases of less than 4 percent on a monosyllabic-word test and 2 percent on a sentence test) and who were fluent in English. All underwent complete medical and psychological evaluations before entry. They were randomly assigned to receive one of the three devices (implant 1, Ineraid [Smith and Nephew Richards, Bartlett, Tenn.]; implant 2, Nucleus [Cochlear Corp., Englewood, Colo.]; or implant 3, 3M/ Vienna [3M Co., Minneapolis]). Thirty patients were assigned to implant 1, and 30 to implant 2; however, only 20 were assigned to implant 3 before it was withdrawn by its manufacturer. All the patients were counseled about the possible results, which ranged from perceiving only sound to understanding speech. Furthermore, they understood that they would be randomly assigned to receive a particular device. The study was approved by the appropriate institutional review boards of each of the participating centers. Informed consent was obtained from all the patients.
Implantation surgery consisted of raising a scalp flap, drilling a well for the electronic components, performing a mastoidectomy to gain access to the posterior middle ear, and inserting the electrode array into the scala tympani of the cochlea. The ear with poorer hearing or the cochlea with less obstruction was chosen as the implantation site. The surgeons were instructed to use a single-channel implant (implant 3 or a single-channel Nucleus device) if they were unable to insert at least half the electrodes of an assigned multichannel implant.
Approximately five weeks after surgery, the device was activated. Because of the differences in the external hardware and equipment needed to stimulate the devices, the audiologists were necessarily aware of which implant a patient had received.
Audiologic Evaluation
Twenty-four individual auditory tests were administered with audiotapes and videotapes. Although the results of each test could be compared at different intervals, as well as across implants, for ease of comparison of the overall results the tests were grouped into five categories according to their content, and a weighted composite index of the results of all the tests was calculated by principal-components analysis6. With the use of the pooled variance-covariance matrixes, a composite score for each category was obtained by multiplying a patient's best scores recorded at a particular evaluation by the coefficients of the first principal component and then summing the products. These measures, used as the primary study outcome, compared the patients' performances over time and across implants. Category 1 tests -- measuring prosodic characteristics -- included tests that evaluated the nonverbal aspects of speech, such as timing, rhythm, and stress patterns. Category 2 -- measuring lip-reading enhancement -- consisted of tests that evaluated the ability to lip read. This was the only group of tests administered in a "sound plus vision" mode; the four other groups of tests were performed with sound alone. Category 3 tests -- measuring phonetic level -- evaluated the ability to identify vowels and consonants in different positions in words. Category 4 tests -- spondee tests -- evaluated the ability to identify two-syllable words with equal stress on each syllable, such as "baseball." In the tests in categories 1 to 4, all results were based on a closed set, from which the patient chose a correct response from a finite number of possibilities. Category 5 tests -- measuring open-set speech recognition -- included monosyllabic-word and sentence tests in which the patient had to repeat the word or sentence without being given a choice of possibilities. This category, involving the most difficult hearing task, was heavily weighted and had the greatest effect on a person's ability to communicate with others in daily life. The scores among devices varied the most on the tests that measured open-set speech discrimination. By the nature of the principal-components analysis, these tests were given the greatest weight, to account for the large differences among the groups of implant recipients.
The following are the maximal composite scores for each of the categories and the composite index: category 1, prosodic characteristics, 52; category 2, lip-reading enhancement, 201; category 3, phonetic level, 77; category 4, spondee tests, 28; category 5, open-set speech recognition, 231; and composite index, 250. The maximal composite score for each category and the composite index were not additive since the variances and resulting component weights differed within each category. In addition, the results of testing involving sound only were used to calculate the composite index, thereby eliminating scores for category 2 tests from the equation. A person with normal hearing would easily obtain maximal scores in all categories and achieve a composite index of 250.
Statistical Analysis
Analysis of variance with multiple-comparisons testing was used to evaluate the changes in the composite scores and the composite-index values from preimplantation values to values recorded 24 months after the implant was activated. A P value of less than 0.05 was considered to indicate statistical significance. Stepwise multiple regression analysis was used to identify factors associated with improvement in the composite index at 24 months. Variables were included in the initial model only if they were associated with improvement in the composite index (P<0.10) according to univariate analysis.
Design Addendum
Late in 1989, a new speech processor incorporating several improvements became available for the Nucleus device (implant 2). Since this Mini-Speech Processor (MSP) was said to have an improved pitch extractor and noise-suppression circuit in addition to a modified speech-encoding strategy, and since preliminary data7 suggested that the new processor improved performance, we compared it both with the older Nucleus processor (the Wearable Speech Processor [WSP]) and with the other two implants. After the results recorded at 24 months were collected, 29 patients who had received the Nucleus implant were offered the MSP; 24 agreed to change processors and be retested. The other five patients had medical problems, refused to change processors, or were unable to return for testing at the appropriate times. Complete evaluations were performed at the time of initial stimulation with the new processor and after three months of use. The results were compared with those obtained at 24 months in the same patients just before the processors were changed and in the patients who received implant 1.
Results
There were no significant differences between the groups of implant recipients in their characteristics and performance at base line (Table 2). There were very few surgical or medical problems related to the cochlear implants (Table 3). Notably, no patient had a facial-nerve injury, major flap necrosis, or compression or misplacement of the electrodes, all of which have been reported in other series of cochlear-implant recipients8,9,10,11. In the implant 3 group, two devices had to be replaced: one failed, and the other caused persistent facial-nerve stimulation. In the implant 1 group, one device was removed at the patient's request because it was not being used although it was functioning, and two devices were replaced because percutaneous pedestals had fractured. One Nucleus single-channel backup device was implanted because obstruction of the scala tympani prevented insertion of the multichannel electrode array. Similarly, an implant 3 was given to one patient in whom implant 1 could not be inserted. These two patients are excluded from the data analysis. In all the other patients, all the active electrodes could be fully inserted.
|
|
The mean changes in the composite index from base line to 24 months in all three implant groups were significant (P = 0.03). Both multichannel devices (implants 1 and 2) were superior to the single-channel device (implant 3) (Figure 1). In addition, the changes in the composite scores for categories 1 and 5 were significantly larger in the patients who received either multichannel implant than in those who received the single-channel implant. The large difference (P = 0.03) between the recipients of single-channel implants and the recipients of multichannel implants in the score for open-set word understanding (category 5) was reflected in the composite index; the multichannel-implant recipients could understand some conversational speech without lip reading, whereas only one recipient of a single-channel implant had this level of benefit. On category 3 tests, only the implant 1 group performed significantly better than the implant 3 group. No individual audiologic test showed that the patients who received the single-channel implant had more improvement than those who received the multichannel implants. The composite index at 24 months, as well as the increases from base line, was similar in the patients who received implant 1 and those who received implant 2.
|
Performance with implant 2 improved after the processor was changed. The mean composite index of the 24 patients who received MSP processors increased from 56 after they had used the WSP processor for 24 months to 86 (P<0.001) after they had used the MSP processor for 3 months (Figure 2). When the MSP processor was used, the mean score for open-set word recognition was 25 percent (range, 6 to 58 percent), and the score for sentence recognition was 58 percent (range, 9 to 97 percent). The improvement in the composite index was also significantly greater in the patients who received implant 2 with the MSP processor than in those who received implant 1, but these results are not based on randomized groups studied simultaneously.
|
All base-line data were examined to identify factors associated with improvement in audiologic performance after implantation. Regardless of the device implanted, the factors that were predictive of better performance as measured by the composite index included younger age at the time of implantation, a higher IQ, a normal cochlea on computed tomography, and better ability to lip read (Table 4). This model accounted for 30 percent of the variability in the composite scores in each category among the patients, suggesting that unidentified factors also influenced prognosis.
|
Patients typically note a benefit immediately after the activation of a cochlear implant, but their performance improves over time. Nonrandomized studies have indicated that multichannel implants are superior to single-channel implants5. The preliminary results of this study (i.e., the findings at one year) showed that the two multichannel implants had a significant advantage over the single-channel implant12. The data for two years confirmed those for the first year, in demonstrating the superiority of both multichannel implants. This was most evident in open-set speech recognition and in the patients' ability to understand some everyday conversation without lip reading. Implant recipients who can understand some speech without lip reading can perform the activities of daily living more effectively. In social situations, conversing with others becomes far less stressful and more effective. The ability to carry out simple tasks, such as understanding questions, enables recipients to function far more independently. At work, they communicate more easily and, depending on the nature of work and the level of benefit from the implant, may be able to undertake greater responsibilities.
Implant recipients who do not reach this level of functioning receive less benefit. The implant functions as an assistive listening device that conveys aspects of the speech signal, such as the number of syllables, intensity, intonation, and stress patterns. These features, when combined with lip reading, enhance the ability to communicate in social and work situations.
It has been of ongoing concern that the development of new devices may render current implants obsolete. This has led some patients to ask that a functioning single-channel device be replaced by an improved, multichannel implant. Although the mean level of performance improves among the patients who receive multichannel implants, there are large differences in the levels of performance of all patients regardless of the type of implant. Some recipients of single-channel implants perform at a higher level than some recipients of multichannel implants, and it is impossible to anticipate whether a patient will do better if given an implant with another design. This study has clearly demonstrated, however, that changing the external hardware and speech-processing strategy of an existing implant can have a substantial effect on open-set speech understanding in an individual patient.
Another concern about cochlear implants has been their safety. Previous studies have reported a substantial but acceptable rate of surgical complications associated with a variety of implants8,9,10,11. The present study, which involved a limited number of surgeons and operations carried out within a protocol, demonstrated that these devices can be implanted with a high degree of safety.
Since the performance of a recipient of any cochlear implant probably depends to a large extent on the number of surviving neuronal components (dendrites, spiral ganglion cells, and vestibulocochlear-nerve fibers), it would be highly desirable to be able to identify the patients likely to have a favorable outcome. Although we found several factors that seemed to influence performance in general, none would suffice to serve as a prognostic indicator.
This study demonstrates the superiority of multichannel implants (implants 1 and 2) to a single-channel implant (implant 3), especially in their ability to allow patients to understand speech without lip reading. Furthermore, the use of an improved speech processor for implant 2 led to significantly better performance in open-set speech recognition, as compared with the performance with the original speech processor or implant 1.
Supported by the Cooperative Studies Program, Medical Research Service, and the Rehabilitation Research and Development Service of the Department of Veterans Affairs.
Source Information
From New York University School of Medicine, New York (N.L.C., S.B.W.), and the Hines Veterans Affairs Medical Center, Hines, Ill. (S.G.F.). The members of the Cochlear Implant Study Group are listed in the Appendix.
Address reprint requests to Dr. Cohen at the Department of Otolaryngology, New York University School of Medicine, 550 First Ave., New York, NY 10016.
References
The following members of the Cochlear Implant Study Group also participated in the study. Executive Committee: R. Tyler, R.A. Dobie, H. Jenkins, and C. Haakenson; ex officio -- D. Deykin, M. Giannini, W. Henderson, H. Spuehler, and A. Boysen.
Cooperative Studies Program Study Participants: otologists -- C.B. Jackson, D.E. Brackmann, N. Coker, R.A. Dobie, B.J. Gantz, M. Glenn, A.D. Hillel, R.A. Hoffman, H. Jenkins, J. Kemink, J.C. Kirchner, J. Kveton, P.J. Moloy, J. Niparko, and R. Rosenberg; audiologists -- H. Fryauf-Berpshy, B. Bromberg, R. Chmiel, L. Crump, S.D. Sigman, L. Eisenberg, K. Filson, G. Haskell, H.J. Jordan, M. Kessler, C. Mailett, D. Oviatt, J. Preece, W. Shapiro, C. Snead, J. Spitzer, L. Sutton, L. Turner, R.S. Tyler, S. Weston, and M. Wofford.
Cooperative Studies Program Coordinating Center: W.G. Henderson, S.G. Fisher, J. Rowe, L. Weber, M.E. Vitek, and D.L. Zullo.
Cooperative Studies Program Clinical Research Pharmacy: P. Day, C. Haakenson, M. Sather, and S. Buchanan.
Data Monitoring Board: F. Black (chairman), H. Levitt, G. Loeb, J.L. Northern, and S. Fowler.
Cooperative Studies Program Central Administration: D. Deykin, J. Gold, and P. Huang.
Rehabilitation Research and Development Administration: M. Giannini, A. Boysen, H. Spuehler, and C. Bazel.
Glossary
Bipolar: Describing an electrical current flowing from an active electrode to a nearby electrode within the cochlea
Channel: A pathway through which information is transmitted from the implant to the auditory nerve
Closed-set speech recognition: The ability to understand a word without lip reading when a finite number of choices are given
Filter-bank encoding: When the speech stimulus is divid ed into its frequency components by filters, the outputs of these filters are transmitted to the appropriate electrodes implanted in the cochlea; low frequencies activate the implanted electrodes in the apical end of the cochlea, and frequencies activate electrodes in the basal end
Monopolar: Describing an electrical current flowing from an active electrode to a ground electrode over some distance, usually outside the cochlea
Open-set speech recognition: The ability to understand words and sentences when no choices are given and lip reading is not used
Phonetic (phoneme): Describing the speech sounds, consisting of vowels and consonants, that make up a word
Postlingual deafness: Deafness occurring after the acquisition of language and speech
Prosodic characteristics: Aspects of speech, such as timing, stress, rhythm, intonation, and number of syllables; these are more readily perceived than phonemes
Scala tympani: The lowermost fluid-filled compartment of the cochlea, into which the electrode is inserted
Speech-feature encoding: Extracting and analyzing from the incoming speech signal only the aspects critical to the understanding of speech; this information about frequency and intensity is then transmitted to the appropriate electrodes implanted in the cochlea
Speech processor: The external portion of a cochlear implant that analyzes and encodes sound stimuli and then transmits them to the implant
Spondee words: Two-syllable words with equal stress on each syllable, such as "baseball" and "toothbrush"
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
