Early Hearing and Communication Development

Chapter IV: Screening

Definition of the Target Disorder

Author: Dr. Martyn Hyde

In the screening context, the "target disorder" is defined most conveniently as the set of all hearing impairments that the screening is intended to detect. Precise definition of the target disorder is fundamental to appropriate program design and evaluation.

The target disorder for a hearing screening program is most commonly quantified in audiometric terms, that is, in terms of parameters of impaired sensitivity to sound. The precise definition adopted affects many aspects of screening programs, including the prevalence and individual impact of the disorder so defined, as well as the nature, timing, intrinsic operating characteristics and actual field performance of screening tests. Generally, the more conservative the definition of the target disorder, the better the screening test performance will be. Some key dimensions of the target disorder are impairment severity, frequency range, laterality (one or both ears), and permanence, as well as the site of the disorder in the auditory system and the associated categories of impairment type (e.g., conductive, sensory, neural). Hearing disorders that are mild, frequency-specific, time-variant or located in the inner cochlear hair cells or afferent auditory pathway (e.g., auditory neuropathy) tend to give rise to higher rates of screening errors. The disorder definition must also address the child's age when the impairment is first expressed. For example, screening in the neonatal period alone will not detect progressive, late-onset or acquired disorders that express later in infancy.

The most common target disorder is congenital permanent childhood hearing impairment (PCHI). Most PCHI is of the "sensorineural" type, originating either in the inner ear (sensory) or in the nerve pathways to higher brain centres (neural), or a combination of the two. The PCHI definition may also include structural conductive impairment arising in the external ear or middle ear. There is a lack of consensus on the merit of detecting transient hearing impairment, which is usually conductive in nature, such as may arise from middle ear infection. Screening targets typically range from a lower sensitivity threshold limit of 25 dBHL (hearing level) at any of several frequencies in any ear (a liberal definition) through to a lower threshold limit of 40 dBHL in the better ear and affecting a wide range of frequencies (a conservative definition).¹

In general, the evidence base relating specific impairment characteristics to specific outcomes in infant development is limited, and the specification of target disorders for screening programs is often based on ad hoc rationales and emerging conventions of practice. On psychoacoustical grounds, it is certain that a child with a 40dB better-ear, permanent impairment would experience major difficulty hearing conversational speech.² The direct evidence base in support of targeting lesser degrees of impairment is less well established. However, it must be stressed that absence of such evidence does not logically equate to absence of impact for smaller degrees of impairment. Even a 25 dB loss of hearing sensitivity at important frequencies would be expected to confer significant limitations in perception of real world signals, on acoustical principles alone.³

In practice, the intrinsic operating characteristics of feasible screening tests strongly influence the target disorder criteria. The current, low limit of impairment that appears to be detectable with reasonable accuracy is typically reported to be about 30 dBHL.² It should be noted, however, that there are several unresolved technical issues relating to the meaning of hearing level in the context of newborn screening. The hearing level scale reference zero level is defined in relation to adult ears, and the effect of delivering a given stimulus to the ear of a newborn may differ from that in an adult, because of differences in anatomy and function of the immature ear.

The question of whether unilateral impairment should be targeted is important because, currently, there is no consensus, and the decision has a large effect on screening test performance and program resource requirements. Important considerations are the likelihood that a unilateral impairment will progress to a bilateral one, or if a child with a unilateral impairment will be seriously disadvantaged, even temporarily, by a middle ear disorder in the better ear. Even if there were no generally acknowledged, effective steps to ameliorate a unilateral impairment, such children may merit close observation and it can be argued that they should be detected by a screening program. This is an apparent departure from the World Health Organization (WHO) tenet that an effective intervention must be available,⁴ but an interpretation is that the intervention in this example is to monitor for adventitious or progressive exacerbation of the target disorder. It is the risk of significant functional limitation that could be considered a justifiable, immediate target.

Definitive (Gold Standard) Measures of Hearing Sensitivity in Infants

Hearing screening tests must be evaluated in relation to definitive tests of hearing. There are two general approaches to "definitive" assessment: behavioural and electrophysiologic; their strengths and limitations are discussed in more detail in Chapter V.

Behavioural tests are commonly quoted to be the gold standard measure. Behavioural observation audiometry (BOA) has been discredited because of poor reliability.¹ Behavioural testing using operant conditioning (visual reinforcement audiometry (VRA)) usually becomes feasible at a developmental age of 6-9 months.

For audiometry to be definitive it must be ear specific and frequency specific, that is, it must test hearing sensitivity for specific frequencies of sound, and in each ear separately. Also, it must be able to be done by both air-conduction (AC) and bone-conduction (BC) stimuli, where clinically indicated. Testing by AC (with an earphone) measures the sensitivity of the entire auditory system including the external and middle ears, and reflects both conductive and sensorineural impairment. Testing by BC (with a transducer usually placed behind the ear) stimulates the inner ear directly by skull vibration, bypassing the external and middle ears, and reflects only sensorineural impairment.

Above about 30 months of age in a developmentally normal child, conditioned play audiometry becomes practicable. Children between 18 and 30 months of age can be very difficult to test accurately by behavioural methods. In children with significant developmental delay, visual or motor impairment, or other co-morbidities, it may be difficult or impossible to obtain accurate hearing assessment by behavioural means at any age.

Auditory evoked potentials (AEPs) may be used to estimate perceptual hearing thresholds in infants and young children. Generally, these potentials are recorded in response to rapidly repeated sounds that are delivered by earphone or BC transducer. Electroencephalogram (EEG) recording electrodes register minute electrical responses from the neurones of the auditory pathway from the cochlea to the cerebral cortex. The AEPs are extracted from ongoing, spontaneous electrical activity of the brain and scalp musculature, using computer averaging of responses that is synchronized to many rapidly repeated stimuli.

The auditory brainstem response (ABR) is an AEP that is a widely accepted proxy gold standard measure of hearing sensitivity in newborns and infants.⁵ ABR measurements can yield reasonably accurate and ear-specific, frequency-specific estimates of perceptual threshold, as well as other information about the functional status of auditory neural pathways.

Measures of Screening Test Performance

A successful screening test yields a binary (pass or refer*) outcome in each ear, for some criterion set of stimulus parameters that are linked as closely as possible to the target disorder definition. If unilateral impairment is within the target disorder, then a refer result in either ear is sufficient to refer the child; if only bilateral impairment is targeted, each ear must refer for an overall refer result. Common measures of screening test performance include test sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and refer rate. Sensitivity and specificity are intrinsic operating characteristics of the screening test (or multi-stage test protocol), whereas the other measures depend strongly on the prevalence of the target disorder. For a given target disorder, sensitivity and specificity vary inversely as the screening refer criterion is altered, that is, made more liberal or more conservative. Therefore, it is necessary to specify both sensitivity and specificity in order to quantify intrinsic test performance adequately. More sophisticated measures include the relative operating characteristic (ROC), which is a curve relating test sensitivity to false-positive rate, as the screening refer criterion is changed.

A limitation of current evidence on screening test performance is that to estimate sensitivity accurately it is necessary to determine the true hearing status of all infants screened, regardless of their screening result. In real world screening programs, only infants who have a refer result from screening are followed up with confirmatory testing. There are very few studies in which the complete screened cohort approach was used.

Follow-up studies restricted to screening refers cannot assess true sensitivity, because the total number of true positive cases is not determined. The number of true cases found is a lower bound on the actual number present. Published reports vary in the appropriateness with which this issue is handled. Also, reliance on extant case-finding systems to identify false-negative screens is clearly questionable in all but the most controlled environments. Such errors or omissions in the literature will tend to lead to inflated estimates of screening test sensitivity.

Loss of cases to follow-up is a source of negative bias in sensitivity estimation from the referred group. In some reports, the true number of cases is estimated by prorating observed yield at follow-up on the basis of the proportion of cases successfully followed. This is a first approximation but is simplistic in that it assumes an equal probability of follow-up attendance for true cases and non-cases, which seems unlikely.

In contrast to the situation for sensitivity, screening test specificity can be estimated accurately from large screening programs. When the prevalence of the target impairment is low, as it is in the population at large, very few babies who pass the screen will have hearing impairment, so the error in assuming the specificity to approximately equal the proportion of infants who pass the screen will be small.

When evaluating screening tests, it is crucial that the tests be known to be conducted with appropriate conditions and techniques. In this regard, it is important to note that screening test performance depends on many variables, and that performance observed in the context of a research study may differ from that seen under field operational conditions. Other factors that may bias estimates of screening test performance include change in hearing status in the interval between screening and gold standard assessment, and less than perfect accuracy of the gold standard assessment itself.

In many of the studies reviewed, screening protocols were not optimized, and this accounts for some of the variation observed. Another important source of variation is sample size. Only large, well-designed studies using appropriate screening and confirmatory test protocols should be considered to yield representative data.

*The term "refer" is used to indicate that the appropriate response is diagnostic investigation and to minimize the use of the word "fail" which has a more negative connotation.

Automated Auditory Brainstem Response Screening

Conventional ABR testing for definitive hearing testing typically involves manual selection of stimulus and recording parameters, and subjective interpretation of averaged EEG/ABR waveforms.⁶ Automated auditory brainstem response (AABR) screening is an adaptation that usually involves a single stimulus condition, with computer-controlled stimulation and recording and computer-based response detection. The usual stimulus is a click at a sound pressure level that corresponds to about 35 dBnHL (normal hearing level), namely, 35 dB above the average subjective perceptual threshold for the click in a group of young adults with normal hearing. A successful test will yield a binary outcome, specifically a pass or refer result, for each ear.

AABR screening is fully automated, objective and non-invasive. It can be performed on any neonate or very young infant who is asleep or at least at quiet rest, in a moderately quiet test environment. Sensors are attached to the scalp and the stimuli delivered by insert earphone or supra-aural earphone. The test typically takes about 10 minutes per baby. The automation of the measurement and interpretation radically reduces the knowledge and skills required in the screening personnel, and renders the procedure much less costly and generally more accurate. Some skill is required of the tester, especially in choosing an appropriate behavioural state of the child at test, the placement of the recording electrodes, and the application of the earphone.

The studies that address the accuracy of screening ABR tests describe results for click ABR screening relative to subsequent definitive audiometry and are of fair to good quality.⁷ There are significant sensitivity limitations of the click ABR, because the click is a broadband stimulus that stimulates a wide range of frequencies. It appears that frequency-specific screening AABR tests are not currently available.

The accuracy of click AABR depends on the definition of the target disorder. It cannot detect hearing impairment at low frequencies or at isolated, specific frequencies, but it has reasonable performance for hearing impairments that are defined in terms of average impairment across the speech frequencies (between 500 and 4,000 Hz). Typical reported sensitivities are from 80-90%, with false-positive rates (FPRs) from 5-10%. It is likely that this is an underestimate of true performance, because factors such as transient impairment at the time of screening or emergent impairment between screening and confirmatory testing degrade apparent test accuracy. Failure to detect a hearing impairment that is not present at the time of screening is not a screening test error, nor is a refer result in the presence of a transient disorder that resolves prior to definitive assessment. The false negative screening error rates of the statistical algorithms for AABR response detection are typically set in the range 0.01-0.001. Discrepancies between these parameters and the higher apparent rates of missed cases in follow-up studies may reflect emergent impairment, frequency-specific impairment and confirmatory test error.

Automated Otoacoustic Emission Screening

Otoacoustic emissions (OAEs) are very faint sounds that arise from the inner ear and may be recorded by a miniature microphone in the external auditory canal.⁸ Evoked OAEs arise in response to controlled acoustical stimulation. There are two types of evoked OAE: transient- and distortion product-evoked OAE; these are denoted as TEOAE and DPOAE, respectively. TEOAE measurement involves delivery of a rapid series of click stimuli, with computer averaging and frequency analysis in order to extract the minute OAE signal from concurrent environmental sound or body-generated sounds. DPOAE measurement involves delivery of two simultaneous, continuous tones that have a specific relationship in intensity and frequency. In response, the inner ear generates a third tone that is related in frequency to the two stimulus tones. It arises because of non-linear distortion in the patterns of excitation within the cochlea, hence the name DPOAE.

Like the AABR, an automated otoacoustic emission (AOAE) screening test is fully automated, objective and non-invasive. It can be done on any neonate or infant who is asleep or at least at quiet rest, and it requires a quiet acoustical environment. A transducer is placed in the ear, and both delivers the stimuli and records the OAE for immediate computer processing. A binary (pass/refer) result is obtained for each ear. The test typically takes less than five minutes per baby.

For an OAE screen to be accurate, the earphone or earphone/microphone assembly must be placed appropriately in the external ear canal, and the canal must be free of vernix or debris. Also, OAEs are especially vulnerable to the presence of middle ear fluid and conductive hearing impairment, partly because the middle ear must first conduct the stimulus to the cochlea and then conduct the emission back to the external ear. It is probably for this reason that AOAE screening in the first few hours after birth is generally reported to yield elevated rates of test failure or false-positive refer results. The AABR, in contrast, appears to be less affected by most minor conditions of the middle ear or external ear.

The most definitive, experimental study to date involved screening of about 5,000 neonatal intensive care unit (NICU) graduates and well babies, using ABR, TEOAE and DPOAE screens with optimized test parameters and objective, statistical screening failure criteria.⁸ Screening results were compared with high quality, confirmatory VRA at 8-12 months corrected age in the complete cohort.⁹ For all optimized screens, with a 10% FPR the sensitivity was typically 80-90%. Because of the possibility of intercurrent or progressive impairment in longitudinal validation studies, these values should be considered as lower bounds on true sensitivity. ABR was found to be the most accurate predictor of hearing impairment at lower frequencies (1 kHz), probably because OAE is especially vulnerable to ambient noise at lower frequencies. At the higher frequencies more commonly associated with PCHI (2-4 kHz), all screening tests performed similarly, with less than a 6.4% refer rate. Multi-stage testing with DPOAE followed by ABR resulted in lower refer rates than TEOAE followed by ABR, the best rate being 2%.

The most definitive screening program report to date addressed 69,761 newborns screened using a two-stage screening prior to hospital discharge (usually OAE and AABR), and an outpatient AABR re-screen after four to six weeks.¹⁰ The program performance improved over three years, the overall refer rate improving from 5.9-2.6% and the pre-discharge FPR achieving less than 3% (specificity 97%). There is increasing evidence that multi-stage screening that includes AABR in at least one stage can achieve overall referral rates as low as 1-2%, which sets an upper bound for the true FPR.¹¹

A final issue to be considered is auditory neuropathy (AN), a condition that may affect up to 10% of all infants with PCHI.¹² Neuropathy is usually characterized by sensorineural hearing impairment (SNHI) of any degree, usually normal OAEs, normal cochlear microphonic potentials (CM), absent acoustic reflexes and an absent or grossly abnormal ABR. OAE screening will miss such babies, but AABR screening will detect them. Because the majority of babies with AN are likely to have attended an NICU, the use of AABR for screening in all NICU graduates is necessary in order to identify most babies with AN.

Potential Harms Associated with Screening Outcomes

False-positive screens increase the burden on follow-up diagnostic services, and may increase family anxiety and stress. These concerns figure prominently in reviews of universal newborn hearing screening (UNHS) and have led to a maximum FPR of 3% being proposed as a benchmark for UNHS programs.¹³ It is clear that high quality screening protocols can achieve overall FPRs of less than 2%. For hospital-based programs, low FPRs at hospital discharge usually require series screening protocols that involve at least two pre-discharge screening tests, the second screen being conditional on a refer outcome for the first screen. The AABR is especially effective at least as the second screen. The lowest overall rates of referral for diagnostic, audiologic assessment may be achieved by including a third screen with AABR, shortly after hospital discharge. These compound protocols reduce the immediate family burden of false positives, as well as the resource impact on both the family and the program, with respect to attendance for audiologic follow-up.

When examining parental responses to universal programs, it was found that negative attitudes were rare and positive attitudes common.¹⁴ The limited body of evidence available does not support the belief that parents generally suffer anxiety or stress due to early screening or to false-positive tests. A substantial majority of parents endorse early identification of hearing impairment. Any concerns are more closely related to timeliness and quality of professional interactions. The rate of significant family anxiety in screening referrals does not appear to differ from the population base pattern of anxiety scores on standard, psychometrically validated measures.

Screening Coverage and Follow-Up Compliance

The overall effectiveness of a screening program depends not only on the performance of the screening tests themselves but on the extent to which subjects are successfully accessed for screening and are successfully followed up after a screening refer result. The overall sensitivity of a program, for example, is the product of the screening coverage, the screening protocol sensitivity, and the follow-up rate. It is commonly reported that there is increased follow-up of screening failures and increased coverage over the first two to three years of screening program implementation.¹⁵ Unpredictability of discharge or transfer has been cited as a reason for higher miss rates in NICU groups. Reports generally indicate that incomplete follow-up coverage is a major area of concern, with reported follow-up rates rarely exceeding 80%.

Surveillance and Referral Components

The proportion of PCHI expressed in infancy but not present congenitally is poorly understood. Estimates lie in the 5-15% range but may be underestimates, due to over-attribution of congenital expression in the absence of comprehensive early detection programs. There is evidence that certain risk indicators, including cytomegalovirus (CMV) infection, persistent pulmonary hypertension of the newborn (PPHN) and several syndromes, are strongly associated with progressive or late-onset impairment. These impairments cannot be detected by neonatal screening and require a program component that includes tracking and screening at a later date, such as at 1 year of age and beyond. It should be noted that re-screening infants at risk (including NICU graduates) with OAE will miss hearing impairment associated with AN. AABR re-screening will not miss neuropathies but has the disadvantage that it will often require use of sedation, in infants older than about 6 months. Also, because not all such cases will be determined to be at risk perinatally, it is probable that education of both families and professionals (e.g., primary care physicians) about early warning signs of hearing impairment will be necessary to maximize overall detection performance. However, to date there appear to have been no systematic studies of the effectiveness of such efforts.

Finally, a comprehensive system for early identification would include a program component that ensures prompt referral for screening and/or audiologic assessment of infants with postnatal risk factors that are associated with acquired hearing impairment, such as meningitis.

Key Differences Between Universal Screening and High-Risk (Targeted) Screening

The target population for screening is usually defined as either all children (universal screening, UNHS) or children at risk for hearing impairment (targeted screening). Typically, the proportion of the general newborn population considered to be at risk for hearing impairment is reported to be in the 8-15% range.^1,16 It follows that it will usually require substantially fewer resources to screen only the high-risk group, relative to those required for universal screening.

Populations at risk for hearing impairment differ from the general population most obviously in the prevalence of the target disorder. The prevalence of hearing impairment in typical high-risk groups, as defined currently, is typically about eight to ten times greater than in babies currently considered to be not at low risk. Relative to universal screening, this increase in base prevalence substantially increases the positive predictive value (PPV) of a screening refer result (non-pass), and substantially reduces the number needed to screen (NNS) in order to identify an additional case.

The most obvious limitation of targeted screening in high-risk newborns and infants is that a substantial proportion of infants who actually have the target impairment will not be screened. It is reported that between 50-75% of young children with significant PCHI manifest a risk indicator, with the most common estimates being closer to 50%.^15,17 This means that even with a perfect screening test, the overall sensitivity of high-risk screening could be no better than 75%, and would probably be closer to 50%. Therefore, relative to targeted screening, the incremental yield of UNHS is the 25-50% of all infants who have hearing impairment and who would not be screened in a targeted program.

Another significant limitation of targeted screening is the difficulty of accurately identifying the sub-population genuinely at risk. Comprehensive and accurate risk identification is a difficult and time-consuming task. Many important risk indicators are not routinely determined or accurately documented. These include perinatal infections such as asymptomatic CMV, as well as manifestations of a variety of syndromes known to be associated with PCHI, including craniofacial anatomical anomalies. Global, proxy risk indicators, such as NICU attendance for over 48 hours, are simple to implement but are very inaccurate indicators of genuine risk. Familial childhood hearing impairment is an important indicator but is very difficult to determine accurately.

In a targeted program, the risk assessment can be viewed conceptually as a documentation-based screening test with very poor sensitivity and specificity, that is, in series with one or more of the physiologic screening tests. The sensitivity and specificity of a compound, series screening protocol are no better than the poorest performance parameter for each component screen. Furthermore, it should be noted that to actually perform a physiologic screening test in a given child may be more reliable and substantially less resource consumptive than to carry out a comprehensive risk determination, so it would seem potentially more accurate and efficient to screen all newborns. However, if risk indicator information were considered absolutely necessary for other activities, such as to define a subpopulation for targeted surveillance, then the risk assessment is required for all infants, and the efficiency argument is weakened. Alternatively, if a program were to incorporate routine surveillance of all infants, then the relevance of specific risk indicators is reduced and the relative accuracy and efficiency of universal, physiologic screening again become a dominant consideration.

Because only about 10% of newborns will manifest a risk indicator detectable with current methods in widespread use,^18,19 UNHS typically involves screening about 10 times as many babies as in targeted high-risk programs, but will increase the yield of true cases by a factor of 33-100%. The actual increase in yield will depend on many variables that reflect individual programs, among them the relative proportions of progressive and early-onset hearing impairment in the at-risk and no-risk groups. These proportions are currently unknown.

It is important to note that a large number of variables affect the optimal design of a screening program. While there are certain, generalizable principles that appear to affect the likelihood of good program performance, it is unlikely that any one particular program design will be optimal under all circumstances. Good screening program design must take proper account of local contextual and infrastructural factors, resources and constraints.

Conclusions

• Hearing screening is currently based on automated otoacoustic emission (AOAE) testing and automated auditory brainstem response (AABR) testing.
• The sensitivity of AABR is 85% and the specificity is 90-95%; for AOAE testing, the sensitivities are only slightly less.
• Current screening protocols frequently involve series or parallel combinations of tests that achieve overall referral rates of <2%.
• Loss to follow-up is the largest single factor limiting effective sensitivity and preventing delivery of hearing and communication development options. Future studies should include determination of predictive models of parental participation, such as optimal communication of screening results to families and psychosocial causes of non-participation.

Key References

1. Joint Committee on Infant Hearing. Year 2000 Position Statement: Principles and Guidelines for Early Hearing Detection and Intervention Programs. Am J Audiol. 2000;9:9-29.
2. Fortnum HM, Summerfield AQ, Marshall DH, et al. Prevalence of permanent childhood hearing impairment in the United Kingdom and implications for universal neonatal hearing screening: questionnaire based ascertainment study. Brit Med J. 2001 Sep 8;323(7312):536-40.
3. Crandell CC. Speech recognition in noise by children with minimal degrees of hearing loss. Ear Hearing. 1993;14:210-6.
4. World Health Organization. Principles and practices of screening for disease. Geneva: WHO; 1968.
5. Durieux-Smith A, Picton TW, Bernard P, et al. Prognostic validity of brainstem electric response audiometry in infants of a neonatal intensive care unit. Audiology. 1991;30(5):249-65.
6. Stapells DR, Oates P. Estimation of the pure-tone audiogram by the auditory brainstem response: a review. Audiol Neuro-Otol. 1997;2:257-80.
7. Hyde ML, Riko K, Malizia K. Audiometric accuracy of the click ABR in infants at risk for hearing loss. J Am Acad Audiol. 1990;1:59-66.
8. Norton SJ, Gorga M, Widen J, et al. Identification of neonatal hearing impairment: evaluation of transient evoked otoacoustic emission, distortion product otoacoustic emission and auditory brainstem response test performance. Ear Hearing. 2000;21:508-29.
9. Widen JE, Folsom RC, Cone-Wesson B, et al. Identification of neonatal hearing impairment: hearing status at 8 to 12 months corrected age using a visual reinforcement audiometry protocol. Ear Hearing. 2000 Oct;21(5):471-87.
10. Prieve B, Stevens F. The New York State universal newborn hearing screening demonstration project: Introduction and overview. Ear Hearing. 2000; 21:85-91.
11. Sokol J, Hyde ML. Preventive Health Care: June 2001 Update: Newborn Hearing Screening. A report to the Canadian Task Force on Preventive Health Care. 2001.
12. Starr A, Picton TW, Siniger Y et al. Auditory Neuropathy. Brain. 1996;119:741-53.
13. Thompson DC, McPhillips H, Davis RL, et al. Universal Newborn Hearing Screening. JAMA. 2001 Oct;286:2000-10.
14. Abdala de Uzcateguic C, Yoshinaga-Itano C. Parents' reactions to newborn hearing screening. Audiology Today. 1997;9:24-25.
15. Finitzo T, Albright K, O'Neal J. The newborn with hearing loss: detection in the nursery. Pediatrics. 1998 Dec;102(6):1452-60.
16. Wessex Universal Neonatal Hearing Screening Trial Group. Controlled trial of universal neonatal screening for early identification of permanent childhood hearing impairment. Lancet. 1998 Dec 19-26;352(9145):1957-64.
17. Galambos R, Wilson MJ, Silva PD. Identifying hearing loss in the intensive care nursery: a 20-year summary. J Am Acad Audiol. 1994 Mar;5(3):151-62.
18. Mehl AL, Thomson V. The Colorado newborn hearing screening project, 1992-1999: on the threshold of effective population-based newborn hearing screening. Pediatrics. 2002 Jan;109(1):E7.
19. Vohr BR, Carty LM, Moore PE, Letourneau K. The Rhode Island Hearing Assessment Program: experience with statewide hearing screening (1993-1996). J Pediatr. 1998;133:353-7.

Previous Page | Table of Contents | Next Page