As shown in slide #17, there are at least five different methods that are frequently used for early identification of hearing loss. What do we know about the accuracy of those methods as a tool for universal newborn hearing screening? Even though there are many articles in the literature which use the terms sensitivity and specificity with regard to each of these techniques, there are no studies of universal newborn hearing screening where there are sufficiently large sample sizes and sufficiently good follow up to definitively establish the sensitivity and specificity of any of those techniques. As summarized in slide #18, most studies which refer to sensitivity and specificity have used very small sample sizes, have focused only on high-risk babies, or have not followed all of the babies who passed the screening test to determine their true hearing status. A frequent problem is that studies that have allegedly examined the sensitivity and specificity of a particular screening technique have done that by comparing one screening technique to another screening technique. In other words, instead of comparing otoacoustic emissions in a screening program to behaviorally confirmed hearing loss, a study might compare otoacoustic emissions to conventional ABR. Thus, data are not available to definitively establish the sensitivity and specificity of any of the techniques.
Auditory Brainstem Response (ABR)
We do, however, have other types of information that provide reasonably good evidence about the accuracy of each of these screening techniques. For example, Hyde, Riko, and Malizia evaluated 713 high-risk babies who were screened with auditory brainstem response (ABR) prior to hospital discharge, and then were assessed by uninformed diagnosticians when they were an average of almost four years old. The behaviorally-confirmed hearing status was compared to the results of the hearing screening test. For this high-risk population, the sensitivity and specificity was 98% and 96% when the ABR screening threshold was set at 40 dB HL and 100% and 91% when the ABR screening threshold level was set at 30 dB HL (see slide #19). Additional data about the accuracy of ABR is shown in slide #20. From a UNHS program at Saint Barnabus Medical Center in New Jersey, over a three-year period with more than 5,000 babies screened each year, an average of 3.3 infants per thousand were identified with a congenital hearing loss and referred into intervention programs. Sedation was not used, the referral rate for further diagnostic testing at the time of hospital discharge was 3.1%, and most screening was done by audiologists within 24 hours of the baby's birth.
Until just recently, the most frequent method used in the United States to identify hearing loss in very young children was based on the high-risk indicators recommended by the Joint Committee on Infant Hearing (JCIH). About 10% of all children born will exhibit one or more of these indicators, including family history of congenital hearing loss, very low birth weight, congenital malformations of the head and neck, hyperbilirubinemia, etc. Since 1974, the JCIH has recommended that infants who exhibit one of these risk indicators be screened for hearing loss using auditory brainstem response. The rationale for this approach was that by focusing on a subset of the population which was at higher risk, hospitals would be able to afford to use auditory brainstem response, whereas most people did not believe hospitals could afford to use ABR to screen all infants.
Although such high-risk-based screening programs were the most frequent method used to identify hearing loss in very young children during the late 70's and early 80's, most hospitals are no longer using this approach for two reasons. First, as shown in slide #21, even though children who exhibit one of the risk factors are at higher risk of having a congenital hearing loss, only about half of all children with congenital hearing loss will exhibit one or more of these indicators. Thus, even if a high-risk-based screening program were to work perfectly, about half of all children with PCHL would be missed. Secondly, high-risk-based screening programs found that it was very difficult to get parents to come back for the necessary diagnostic evaluations. For example, in 1987, Mahoney and Eichwald reported the results of the Utah- based high-risk screening program over a seven-year period (1978 to 1984). During that time, the JCIH indicators were incorporated into the legally required birth certificate, so they were reported for every child. The program included computerized mailings and follow up, free diagnostic assessments at regional offices, and/or a mobile van that went to parents' homes. As summarized in slides #22 and #23, only about half the parents whose children had a high-risk indicator ever made an appointment for a diagnostic evaluation, and only about half of those parents completed the evaluation. Only .36 children per thousand were identified with sensorineural hearing loss. Other high-risk indicator-based programs experienced similar or even worse difficulties. Even though complete follow-up data is not available for any of the programs, it is clear, based on the very small number of children identified, that such programs have very low sensitivity; and given the relatively high number of children exhibiting risk factors, such programs also have very low specificity.
Behavioral Evaluations at Seven to Nine Months of Age
Although it is not used very often in the United States, another alternative for early identification of hearing loss is to do behavioral assessments of children when they are 7-9 months old. Such programs are used extensively in Europe and have sometimes been advocated for implementation in the U.S. As they operate in Europe, the screening is done by home visitors who are already making routine visits as a part of the well-child health care system. Because the implementation of such a program in the U.S. would require the creation of a very expensive infrastructure of home visitors, some people suggest that such behavioral evaluations could be done by physicians as a part of well-baby care when babies are 3-9 months of age using techniques similar to what is used in those European countries.
Although often advocated as an alternative for this country, the data on the success of those home-based behavioral screening programs, often called the Home Visitor Distraction Test, is very disappointing. For example, Peter Watkin and his colleagues (slide #24) did a retrospective analysis of over 55,000 children in one geographic district in England. For each of the 171 two- to fifteen-year-old children who had a hearing loss, Watkin and his colleagues determined whether the child was first identified through a home visitor or school-age screening program, a parent, or someone else, such as a doctor or teacher. More than a third of the children were missed by both the nine-month and the school-age screening program.
Slide #25 shows the results for just the home-based screening program. Of the 39 children with severe profound bilateral losses, only 44% were identified from the Home Visitor Distraction Test. For children with mild moderate bilateral losses and children with unilateral losses, only 25% and less than 10% were identified with the Home Visitor Distraction Test, respectively. So even with home visitors who were specifically trained to do that type of behavioral assessment in a home setting and were given a great deal of support and monitoring to do it well, most of the children are being missed. If such home visitors are unable to do behavioral screening at nine months of age, it is very unlikely that it can happen as a part of the routine in a busy doctor's office.
Otoacoustic Emissions (OAE)
A fourth newborn hearing screening approach is the use of otoacoustic emissions, either transient evoked or distortion product otoacoustic emissions. Numerous small-scale studies, such as those summarized in slides #26 and #27, have demonstrated that transient evoked otoacoustic emissions (TEOAE) have a very high rate of agreement with auditory brainstem response. The first large-scale evaluation of otoacoustic emissions in a universal newborn hearing screening program was the Rhode Island Hearing Assessment Program (RIHAP) led by Betty Vohr between 1990 and 1994. As shown in slide #28, of the first cohort of 1,850 infants from well-baby and special-care nurseries, 11 were identified with sensorineural hearing loss. Because ABR screening was also done with each of these babies, the study showed that there was very high agreement between TEOAE screening and auditory brainstem response. Furthermore, four of the babies would have been missed if screening had only been done with high-risk infants or with babies in the NICU. Although it is impossible to determine sensitivity since follow up was not done with all of the children who passed, there was a questionnaire study done later, when children were 2-3 years old, to which about 40% of the "passes" responded, and none of the parents believed that their child had a sensorineural loss. If we assume that all of the original passes had normal hearing, then the sensitivity in this study was 100%, and the specificity was 95% (slide #29).
Automated Auditory Brainstem Response (AABR)
The fifth and final technique for newborn hearing screening considered here is the use of automated ABR. Although several other units are now available, the data in these slides were collected the most widely used AABR equipment produced by Natus Medical. As can be seen in slides #30 and #31, four different studies in which results of the Algo1 were compared to conventional ABR show high levels of sensitivity and specificity for the two techniques. The combined data are summarized in slide #32. Even though the results do not provide information on the sensitivity for determining hearing loss since behavioral evaluations of all the children were not done, given what we know about the accuracy of conventional ABR, these data provide convincing evidence about the accuracy of automated ABR.
University of Washington Research
One other study that is now underway should be mentioned. Sponsored by the National Institutes of Health and being coordinated by the University of Washington, this is a multi- centered study which is evaluating the effectiveness of newborn hearing screening using conventional ABR, transient evoked otoacoustic emissions, and distortion product otoacoustic emissions. Over 7,000 infants (4,500 NICU babies and 2,600 normal-care nursery babies) were screened prior to discharge using all three of the techniques in random order. Babies were behaviorally assessed at 8-12 months of age. All of the behavioral assessments were completed in October 1997, and final results from the study are expected to be available by April 1998. As summarized in slide #33, these data will provide us with more definitive information than we have had about the true sensitivity and specificity of ABR and evoked otoacoustic emissions. It's important to note, however, that the screening methods used in this study are now five years old, so even though the study will provide important information about the underlying mechanism, the equipment used with each of these techniques is quite different today than at the time data were collected for this study. Parenthetically, it is important to note that newborn hearing screening equipment continues to evolve quite rapidly. Whether ABR or EOAE is used, the equipment is becoming faster and more accurate with each passing year. The last five years have seen dramatic changes in both AABR and EOAE equipment, and it is safe to predict that similar advances will happen during the next five years.
Back | Intro | Next