PEDIATRICS Vol. 110 No. 5 November 2002, pp. 855-864
Projected Cost-Effectiveness of Statewide Universal Newborn Hearing Screening
,
,**,||
* Department of Medicine
Clinical Effectiveness Program, Children’s Hospital, Boston, Massachusetts
Department of Pediatrics, Harvard Medical School, Boston, Massachusetts
|| Department of Ambulatory Care and Prevention, Harvard Pilgrim Health Care and Harvard Medical School, Boston, Massachusetts
¶ Oregon Health Sciences University and the Oregon Evidence-based Practice Center, Portland, Oregon
# Department of Pediatrics, University of Washington, Seattle, Washington
** National Initiative for Children’s Healthcare Quality, Boston, Massachusetts
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Objectives. Early identification of hearing impairment may improve language outcomes and subsequent school and occupational performance of the deaf. Universal newborn hearing screening (UNHS), currently mandated by 32 states, can reduce the median age of identification of hearing impairment from 12 to 18 months to 6 months or less. However, because false-negative tests must be minimized, the prevalence of congenital deafness is low, and screening tests are imperfect, UNHS results in many false-positive results and has a low positive predictive value (PPV). The objective of this study was to evaluate UNHS and selective screening in terms of both short- and long-term benefits, harms, and financial costs and to identify steps in the screening process that could be improved to increase cost-effectiveness.
Methods. The cost-effectiveness analysis, conducted from the societal perspective, compared the projected outcomes of 1) no newborn hearing screening, 2) selective newborn hearing screening, and 3) UNHS for a hypothetical state birth cohort of 80 000 infants. Probability and cost estimates for the decision model were obtained from published studies, expert opinion, and national and state sources. The main outcomes were incremental cost per infant whose deafness was diagnosed by 6 months, which included only the cost of screening and diagnostic evaluation; and incremental cost per deaf child with normal language, which also included the costs of medical care, education and assistive devices, and lost productivity over the lifetime of the deaf individual.
Results. Selective screening identified 62 of the 128 deaf infants in the birth cohort, referred 0.18% of all infants for diagnostic evaluation, and had a PPV of 43%. UNHS identified 116 of the 128 deaf infants, referred 1.6% of all infants, and had a PPV of 8.8%. Our model simulated real-world conditions in which some infants whose deafness is identified at screening do not receive a definitive diagnosis of being deaf before 6 months; and a portion of deaf and hard-of-hearing infants who 1) have false-negative screening test results, 2) are not screened, or 3) fail the hearing screen but are not immediately followed up with diagnostic evaluation nonetheless receive a diagnosis by 6 months of age. In the absence of newborn hearing screening, approximately 30 deaf infants were identified by 6 months of age by passive detection alone at a cost of $69 000. The selective screening protocol, when compared with no newborn hearing screening, resulted in an additional 36 infants whose deafness was diagnosed by 6 months at an additional cost of approximately $600 000, yielding an incremental cost-effectiveness of approximately $16 000 per additional infant whose deafness was diagnosed by 6 months. Compared with selective screening, the UNHS protocol resulted in 33 additional infants whose deafness was diagnosed by 6 months of age at an additional cost of approximately $1.5 million, yielding an incremental cost-effectiveness of approximately $44 000 per additional infant whose deafness was diagnosed by 6 months of age. Increasing the rate of follow-up to diagnostic evaluation from the base-case estimate of 77% to 100% decreased the incremental cost of UNHS to $38 000 per additional infant whose deafness was diagnosed by 6 months. Under the base-case assumptions about lifetime savings that result from normal language with early intervention, UNHS resulted in normal language achievement for more deaf children and was cost saving in the long term compared with both selective screening and no screening.
Conclusions. The short-term cost-effectiveness of UNHS is comparable to the cost per case diagnosed of other newborn screening programs and could be improved by increasing the rate of follow-up to diagnostic evaluation after positive screening test results. If early identification results in improved language abilities, lower educational and vocational costs, and increased lifetime productivity, then UNHS has the potential for long-term cost savings compared with selective hearing screening and no screening. To understand the actual long-term economic effects of UNHS, better evidence is needed regarding the impact of early intervention on language outcomes and subsequent changes in educational costs and lifetime productivity.
Key Words: newborn • deafness • neonatal screening • hearing tests • cost-benefit analysis • program evaluation
Abbreviations: UNHS, universal newborn hearing screening • PPV, positive predictive value • NICU, neonatal intensive care unit • TEOAE, transient evoked otoacoustic emissions • AABR, automated auditory brainstem response • QALY, quality-adjusted life year
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Approximately 5000 infants in the United States are born each year with congenital bilateral moderate to profound deafness. The overall prevalence is between 1 to 2 per 1000 newborns and is higher (approximately 8 per 1000) in infants with risk factors for deafness.1–10 Congenital hearing loss is associated with delayed language, learning, and speech development early in life and low educational and occupational performance in adulthood.11
Universal newborn hearing screening (UNHS), currently mandated in 32 states, has been shown to reduce the average age of identification of deaf infants from 12 to 18 months down to 6 months or less.2,10,12–18 Early identification and intervention with speech and language therapy, amplification, and family support optimizes communication during the early critical window of language acquisition and may improve language outcomes at 2 to 5 years of age.19–26 One limitation of UNHS is that because false-negative tests must be minimized, the prevalence of congenital deafness is low, and screening tests are imperfect, UNHS results in many false-positive results and has a low positive predictive value (PPV). The PPV of several recently described UNHS programs ranged from 1.4% to 68.4%.1–9 False-positive results may cause parental anxiety9,27,28 and result in unnecessary follow-up tests and occasionally unnecessary interventions.29 It is also unclear whether false-positive screening test results adversely influence the way parents treat their infant and the infant’s subsequent social development in the first few months of life.
Selective screening of newborns with risk factors for congenital deafness is one alternative to UNHS. Risk factors for congenital deafness, as defined by the Joint Committee on Infant Hearing, include an illness or condition requiring neonatal intensive care unit (NICU) admission for 
48 hours, stigmata or other findings associated with syndromes known to include deafness, family history of permanent childhood sensorineural deafness, craniofacial anomalies, and in utero infections associated with deafness.11 Although few published studies have documented the results of large-scale selective screening programs,30–33 this strategy was commonly used in the United States before the introduction of UNHS and is currently used by hospitals in states that have not mandated UNHS, as well as in many countries outside North America.30,34–36 Because selective screening focuses on a subset of high-risk infants, it is less expensive and has a higher PPV than UNHS. However, selective screening will not identify deaf newborns without risk factors and, because it relies on systematic and reliable identification of infants with risk factors, may be difficult to implement.
Several studies have evaluated the cost-effectiveness of UNHS protocols37–42 but only Turner43–45 and Kemper and Downs46 compared UNHS and selective screening. Their analyses were limited to the short-term costs of hearing screening and did not consider the potential impact of early identification on improved language outcomes, educational costs, and long-term productivity. The specific aims of this analysis were to 1) compare UNHS and selective screening in terms of both short- and long-term benefits, harms, and financial costs and 2) identify steps in the screening process that could be improved to increase cost-effectiveness.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Design
This cost-effectiveness analysis was conducted from the societal perspective for a hypothetical birth cohort of 80 000 infants in 1 state. We chose to model a statewide cohort because decisions to implement UNHS are typically made at the state level. The decision model compared no newborn hearing screening, selective newborn hearing screening, and UNHS (Fig 1). Under selective newborn hearing screening, only infants with risk factors for congenital deafness were screened, whereas under UNHS, all infants were screened. Although the term "deaf" may be used to refer to a wide spectrum of hearing loss, this analysis defined deafness as moderate to profound bilateral hearing loss (
40 decibels), the definition used by most of the UNHS programs from which we derived our probability estimates.1,3,5,6,8,10
|
In this model, all of the screening protocols used 2-step screening before discharge, which prevents the loss to follow-up seen with postdischarge screening programs and reduces the number of false-positive results seen with single-step predischarge screening. In the UNHS protocol, an automated transient evoked otoacoustic emissions (TEOAE) test was followed by an automated auditory brainstem response (AABR) test for infants who did not pass the TEOAE. This UNHS protocol was chosen because it is widely used and was recommended in a 1993 National Institutes of Health Consensus Development Conference Statement.18 Also, previous studies suggested that it is a cost-effective UNHS protocol.39,42 The selective screening protocol used an AABR test with repeat AABR testing for nonpasses. This is an inexpensive, accurate, and commonly used strategy for screening smaller numbers of high-risk infants. In our model, the diagnosis of deafness was made during a diagnostic evaluation by an otorhinolaryngologist and an audiologist. Because infants who fail screening do not always promptly attend the recommended diagnostic evaluation, we included this incomplete follow-up rate in our model. We also accounted for the fact that a fraction of deaf infants who 1) had false-negative screening test results, 2) were not screened under selective screening, or 3) failed the hearing screen but did not have immediate follow-up with diagnostic evaluation could still have their deafness diagnosed by 6 months of age. For each screening strategy, we modeled the proportion of infants whose deafness was diagnosed by 6 months of age and, of those, the proportion who began early intervention (amplification and/or enrollment in an early intervention program) by 12 months of age. Finally, we modeled the proportion of infants who, with intervention before or after 12 months, had normal language abilities at entry to grade school. We assumed that deaf children with normal language abilities at entry to grade school would have normal language as adults.
The main outcomes for the analysis were 1) the incremental cost per infant whose deafness was diagnosed by 6 months, which included only the cost of screening and diagnostic evaluation; and 2) the incremental cost per deaf child with normal language, which also included the costs of medical care, education, and assistive devices as well as lost productivity over the lifetime of the deaf individual. Incremental costs are rounded off and reported using 2 significant digits in the text; more significant digits are presented in the tables for verification of calculations.47 Intermediate outcomes included test results, the percentage of infants who failed the screening protocol and were referred for diagnostic evaluation (refer rate), the percentage of newborns referred who actually followed up with diagnostic evaluation, and the PPV of the test.
Probabilities
Most probabilities in the decision model (Table 1) were derived from published results of hospital and state newborn hearing screening programs. For questions on which the data were sparse or lacking, we convened a panel of 4 national experts on hearing screening and language development in the deaf and used a modified Delphi process to derive estimates. Expert panelists were selected by asking authorities in the field to identify individuals with extensive experience and a balanced perspective on the effectiveness of hearing screening and interventions to improve speech and language in the deaf.
|
Epidemiology and Screening Test Properties
The proportion of newborns with risk factors for deafness was based on 3 large population-based studies for which information on risk status was available.2,3,48 Ideally, infants should be risk stratified using the complete list of risk factors described in the Joint Committee on Infant Hearing’s 2000 position statement.11 However, the studies used to derive estimates of risk status in this analysis simplified assignment of high-risk status to infants who had been admitted to a NICU and thus may have underestimated the prevalence of high-risk status. The impact of increasing the prevalence of high-risk status was evaluated in sensitivity analysis. The estimates of the prevalence of deafness in low- and high-risk infants were weighted averages (by sample size) derived from studies that provided data on bilateral moderate to profound deafness stratified by risk status.1,3,6,9
Because no studies have compared newborn hearing screening results of all screened infants with a gold standard test result, the true sensitivity and specificity of the various hearing screening technologies are not known. We assessed the sensitivity and specificity of the TEOAE and AABR machines reported in studies that used those technologies.2,6,9,49,50 and used point estimates for these variables that produced referral rates close to those reported for 2-step inpatient screening protocols.1,4,51,52 Testing infants twice decreases the number of false-positive results and thus increases the overall specificity of the screening protocol. However, screening twice also increases the number of false negative results and thus decreases the overall sensitivity of the protocol. In our model, the UNHS protocol had an overall sensitivity and specificity of 90.3% and 98.5%, respectively. The selective screening protocol had an overall sensitivity and specificity of 90.3% and 99%, respectively. Follow-up rates for definitive audiologic evaluation were obtained from studies that tracked this parameter.2,6,8,9,53
For selective screening, we assumed that 80% of high-risk infants would be identified and screened. It is possible that some high-risk infants would be missed, for example, if providers failed to ask parents about a family history of congenital sensorineural hearing loss. However, by using admission to a NICU as a proxy for many of the risk factors (birth weight <1500 g, Apgar score of 0–4 at 1 minute or 0–6 at 5 minutes, prolonged mechanical ventilation, receipt of ototoxic medications, and hyperbilirubinemia requiring exchange transfusion), at least 60% of the high-risk infants could be identified.6 Thus, if providers were only 50% effective in screening for the other risk factors (family history of congenital sensorineural hearing loss, congenital syndromes associated with deafness, congenital infections associated with deafness, and craniofacial anomalies) in the well-infant nursery, then they would still identify 80% of all infants with risk factors for deafness.
Age at Diagnosis and Intervention
We modeled the probability that an infant with moderate to profound deafness would be identified by 6 months of age and receive early intervention by 12 months of age because these are the age targets reported in most published studies of UNHS3,10 and no screening.13 We also modeled the probability that a nondeaf infant would reach audiologic evaluation and be incorrectly diagnosed and treated as deaf.29 When data were lacking for estimates of certain conditional probabilities, we used known probabilities to mathematically derive these estimates and then checked them for face validity. For example, it is not known what proportion of infants who fail newborn hearing screening but do not follow up with diagnostic evaluation have their deafness diagnosed by 6 months. However, knowing the proportion of all deaf infants whose deafness is diagnosed by 6 months and the proportion who fail the screen and do follow up with audiologic evaluation, we were able to derive an estimate of that proportion.
Language Outcomes
Because the size of the effect of early identification on language outcomes in deaf infants is not established, we relied on our expert panel to derive estimates of the probability of normal language quotients on entering grade school in children who did and did not have risk factors and were given early (before 12 months) versus late (after 12 months) intervention. We performed extensive sensitivity analyses around these base-case estimates.
Costs
All costs were adjusted to 2001 US dollars using the medical component of the Consumer Price Index for medical costs, the employment cost index for wages, and the Consumer Price Index for the cost of other goods.54 Future costs were discounted at a rate of 3% per year, as recommended by the Panel on Cost-Effectiveness in Health and Medicine.47
Screening and Diagnostic Evaluation Costs
Screening cost data (Table 2) were derived primarily from the National Center for Hearing Assessment and Management55 and the audiology department of Children’s Hospital, Boston. Screening costs were broken down into fixed costs, which include the cost of machines, and variable costs, which include supplies and wages. For UNHS, we assumed that each of the state’s 53 birthing hospitals (with annual birth rates >100) required 1 AABR machine and 1 TEOAE machine for the first 2000 newborns plus an additional TEOAE machine for every additional 2000 newborns (total 53 AABR and 71 TEOAE machines for the state). For selective screening, we assumed that each of the state’s 53 birthing hospitals (with annual birth rate >100) needed only 1 AABR machine. Infants who fail newborn hearing screening (as well as those identified as deaf outside a screening program) require a complete audiological evaluation, which includes a formal ABR and examinations by an audiologist and otorhinolaryngologist, to confirm the diagnosis. We estimated the cost of a complete audiological evaluation to be $540.
|
For selective screening, we also included the cost of the screening procedure to decide which infants are high risk and thus require newborn hearing screening. The majority of infants with risk factors for deafness (as described by the Joint Committee on Infant Hearing) will have spent time in a NICU. For example, all infants who have a birth weight <1500 g, prolonged mechanical ventilation, and hyperbilirubinemia requiring exchange transfusion will be cared for in the NICU. Most infants who have congenital syndromes or infections associated with deafness, craniofacial anomalies, or low Apgar scores or who receive ototoxic medications will also be cared for in the NICU. Thus, admission to the NICU is a good proxy risk factor for most of the Joint Committee’s risk factors. Assuming that a hospital that uses selective screening would screen all infants who are admitted to a NICU and would recognized most craniofacial anomalies and congenital syndromes and infections associated with deafness, the only risk factors for which the hospital would have to screen actively in the well-infant nursery would be low Apgar scores, receipt of ototoxic medications, and a family history of hereditary childhood sensorineural hearing loss. We estimated that it would take 3 minutes for a nurse to question both parents about a family history of hearing loss and to identify the presence of the other 2 risk factors. Thus, at a salary of approximately $19/hour (same salary used for screeners in UNHS program), it would cost $0.95 per infant to perform the screen.
In the state we used for our model, only 3 birthing hospitals had annual birth rates <100. Under both selective screening and UNHS, arrangements would need to be made to screen after discharge some or all of the infants who were born in these hospitals. We did not include in our model the extra cost of screening and tracking this small group of infants.
Societal Costs
In our model, the lifetime societal costs of congenital deafness (Table 3) consisted of lost productivity, special education, vocational rehabilitation, medical costs, and assistive devices (Table 3). For an individual with preverbal onset of deafness, these have been estimated to total $1.1 million.56 In the base case, we assumed that improved language outcomes resulting from early intervention would result in a 75% decrease in lost productivity.57 We also assumed that improved language outcomes would result in a 10% decrease in special education needs and a 75% decrease in vocational rehabilitation needs. Thus, in our base-case estimate, improved language reduced the lifetime costs associated with deafness by approximately $430 000 per deaf individual. As the true impact of early intervention on long-term costs is not yet known, we performed sensitivity analyses around all of our cost reduction estimates.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Base-Case Results
Screening Test Results and Referral Rates
In our model, selective screening identified 62 of the 128 deaf infants in the birth cohort, referred 0.18% of all infants for diagnostic evaluation (145 of 80 000), and had a PPV of 43% (62 of 145). UNHS identified 116 of the 128 deaf infants, referred 1.6% of all infants (1314 of 80 000), and had a PPV of 8.8% (116 of 1314). Under selective screening, 112 of 145 infants with positive screening tests were followed up with a diagnostic evaluation, and of those, 49 received a diagnosis of being deaf. Under UNHS, 1015 of 1314 infants with positive screening tests were followed up with a diagnostic evaluation, and 97 received a diagnosis of being deaf (Table 4).
|
Diagnosis by 6 Months, Intervention by 12 Months, and Normal Language Outcomes at Entry to Grade School
Our model simulated real-world conditions in which a portion of deaf infants who 1) have false-negative screening test results, 2) are not screened under selective screening, or 3) fail the hearing screen but do not have immediate follow-up with diagnostic evaluation nonetheless have their deafness diagnosed by 6 months of age. Thus, although 49 infants under selective screening and 97 infants under UNHS received a diagnosis of being deaf through screening and prompt diagnostic evaluation, a total of 66 infants under selective screening and 99 infants under UNHS received a diagnosis as being deaf by 6 months of age. Our model also simulated that not all infants whose deafness is diagnosed by 6 months receive intervention before 12 months and that some infants whose deafness is diagnosed after 6 months receive intervention before 12 months. Thus, although 66 infants under selective screening and 99 infants under UNHS had their deafness diagnosed by 6 months, 75 infants under selective screening and 97 infants under UNHS received intervention by 12 months. Of these infants, 59 under selective screening and 65 under UNHS had normal language quotients at entry to grade school.
Incremental Cost Per Infant Whose Deafness Was Diagnosed by 6 Months of Age
In the absence of newborn hearing screening, 30 deaf infants were identified by 6 months of age by passive detection alone. The total cost of detection of hearing loss for these infants was $69 000. Compared with no newborn hearing screening, the selective screening protocol resulted in an additional 36 infants whose deafness was diagnosed by 6 months at an additional cost of approximately $600 000, yielding an incremental cost-effectiveness of approximately $16 000 per additional infant whose deafness was diagnosed by 6 months. Compared with selective screening, the UNHS protocol resulted in 33 additional infants whose deafness was diagnosed by 6 months of age at an additional cost of approximately $1.5 million, yielding an incremental cost-effectiveness of $44 000 per additional infant whose deafness was diagnosed by 6 months of age.
Incremental Cost per Deaf Child With Normal Language
When lifetime savings from normal language with early intervention were incorporated, both selective screening and UNHS resulted in normal language achievement for more deaf children and total cost reduction compared with no screening. For example, compared with no newborn hearing screening, selective screening resulted in 6 additional deaf children with normal language abilities and a cost reduction of approximately $1.5 million for the cohort of deaf children. Compared with selective screening, UNHS resulted in 6 more deaf children with normal language abilities and an additional cost reduction of $870 000. In cost-effectiveness analysis terms, UNHS was the dominant strategy in this analysis because it resulted in better outcomes and reduced costs compared with both selective screening and no screening.
Sensitivity Analyses
The base-case estimate of marginal cost per infant whose deafness was diagnosed by 6 months of age was robust to changes in most of the key probabilities and costs. Two variables to which the model was moderately sensitive were the screening success rate (the proportion of infants with risk factors actually screened) under selective screening and the rate of follow-up with diagnostic evaluation after referral for a positive screening test. As the screening success rate under selective screening was varied from 40% to 100%, the incremental cost of UNHS changed from $33 000 to $55 000 per additional infant whose deafness was diagnosed by 6 months. Increasing the rate of follow-up with diagnostic evaluation from 50% to 100% decreased the incremental cost of UNHS from $56 000 to $38 000 per additional infant whose deafness was diagnosed by 6 months.
Varying the probability of diagnosis by 6 months without newborn hearing screening had a smaller effect on the marginal cost per infant whose deafness was diagnosed by 6 months under selective screening and UNHS. When the proportion of high-risk infants whose deafness was diagnosed by 6 months in the absence of newborn hearing screening was decreased from the base-case estimate of 25% to 10%, the incremental cost per additional infant whose deafness was diagnosed by 6 months decreased from $16 000 to $13 000 for selective screening and $44 000 to $41 000 for UNHS. When the proportion of low-risk infants whose deafness was diagnosed by 6 months in the absence of newborn hearing screening was decreased from the base-case estimate of 20% to 10%, the incremental cost per additional infant whose deafness was diagnosed by 6 months was unchanged for selective screening and decreased from $44 000 to $39 000 for UNHS.
There is a paucity of evidence and thus considerable uncertainty around the base-case estimates of the proportion of low-risk infants with normal language abilities after early intervention and the percentage increase in lifetime productivity given normal language abilities. The cost-effectiveness of UNHS was very sensitive to changes in the estimates for these variables. A 2-way sensitivity analysis (Fig 2) showed that when the percentage increase in productivity given normal language and the proportion of low-risk infants with normal language after early intervention both were high, UNHS resulted in net savings (represented by the area to the right of and above the curve labeled "Universal screening cost saving compared with selective screening"). Under less favorable estimates of these probabilities, UNHS was not cost saving and the incremental cost (per additional child with normal language) increased significantly, as indicated by the isocontour curves to the left of and below the above-mentioned curve. For example, when the proportion of low-risk deaf infants with normal language after early intervention was 60% (compared with 40% after late intervention), UNHS was cost saving only when the percentage increase in lifetime productivity as a result of normal language was 64%. When the percentage increase in lifetime productivity decreased to 30%, the incremental cost (per additional child with normal language) of UNHS was somewhere between $50 000 and $200 000. In the range of estimates modeled, selective screening was cost saving compared with no screening as long as the percentage increase in lifetime productivity given normal language was >15%.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Major Findings
This study found that the incremental cost of diagnosing an infant with bilateral moderate to profound deafness by 6 months of age was approximately $16 000 using selective screening and $44 000 using UNHS. These costs are comparable to the cost per case diagnosed of screening programs for other newborn conditions such as congenital hypothyroidism and phenylketonuria, estimated in a recent economic evaluation of screening for inborn errors of metabolism at $27 000 and $42 000, respectively.58 (These figures are converted from 1997 British pounds into 1997 US dollars and inflated from 1997 to 2001 US dollars using the medical component of the US Consumer Price Index). Under the base-case assumptions about the impact of early intervention on language outcomes and increased productivity associated with normal language, both selective screening and UNHS were cost saving. A 2-way sensitivity analysis around these assumptions revealed that UNHS was cost saving only under a limited range of probability estimates. Under less favorable probability estimates, the incremental cost (per additional deaf child with normal language abilities) of UNHS increased significantly.
Unfortunately, the true impact of early intervention on language outcomes and subsequent increases in productivity are not yet known. Several studies suggest that early intervention improves language outcomes of deaf children in the first few years of life,19–26 but these studies have methodologic limitations59–61 and none actually quantifies the probability of normal language abilities given early versus late intervention. Because UNHS and early intervention for deaf infants before 6 months of age are relatively new phenomena, no studies have specifically quantified the long-term impact of early intervention and normal language abilities on the educational and vocational costs and lifetime productivity of the deaf.
Comparisons With Other Studies
This study differs from recent economic evaluations of UNHS in that the decision model incorporated the improved sensitivity of 2-step screening before discharge from the hospital as well as imperfect follow-up for diagnostic evaluation after failed screens, both of which have been shown to influence the effectiveness of UNHS programs.51,52 Our model also simulated real-world conditions in which infants whose deafness is diagnosed by 6 months do not necessarily receive intervention by 12 months and infants (particularly high-risk ones with other developmental problems) who receive intervention by 12 months do not necessarily have normal language abilities at entry to grade school. The model also allowed for the fact that some deaf infants who are not screened, are not diagnosed by 6 months, or do not receive intervention by 12 months may nevertheless have normal language abilities at entry to grade school. These opposing processes—attrition after early diagnosis and intervention on the one hand and attainment of normal language in the absence of screening on the other-explain the convergence seen in the number of children with normal language outcomes under the 3 different screening strategies. It is possible that our estimates of the proportion of deaf infants who are diagnosed by 6 months, receive early intervention by 12 months, or attain normal language abilities in the absence of newborn hearing screening and early intervention were too optimistic. However, when the proportion of deaf infants whose deafness was diagnosed by 6 months in the absence of newborn hearing screening was decreased in sensitivity analysis, the incremental cost per infant whose deafness was diagnosed by 6 months with selective screening and UNHS did not decrease significantly. Conversely, decreasing the proportion of deaf infants with intervention by 12 months in the absence of newborn hearing screening increased the total incremental savings of selective screening and UNHS.
This is also the first cost-effectiveness analysis of newborn hearing screening that estimated the incremental cost of both UNHS and selective screening and incorporated the lifetime societal costs of deafness. A few previous cost analyses estimated the cost per infant screened under a UNHS program,37,40,41 and several cost-effectiveness analyses compared alternative UNHS protocols in terms of cost per infant whose deafness is diagnosed.38,39,42–45 However, only Kemper and Downs46 compared UNHS and selective screening and calculated the incremental cost-effectiveness of UNHS. They found that, compared with selective screening, UNHS would cost an additional $24 000 for each additional deaf infant detected. Our finding of an incremental cost of $44 000 is consistent with theirs but somewhat higher, most likely because we modeled the imperfect follow-up rate for diagnostic evaluation, which was shown in sensitivity analyses to affect incremental cost-effectiveness. This study extends Kemper and Downs’ findings in that we estimated the incremental cost-effectiveness of selective screening compared with no screening and extended the time horizon of the model to incorporate lifetime societal costs of deafness.
We chose to include selective screening in our decision model for several reasons. First, in the context of a fixed budget for health care expenditures, the true cost of UNHS can be understood only in incremental terms in relation to a less expensive, albeit less effective, strategy. Second, although UNHS has been legislatively mandated in 32 states, questions about its long-term effect on language outcomes, cost, and quality of life have not been firmly resolved, and critics continue to worry about the potential harm of false-positive results in low-risk infants.59–61 Last, but perhaps most important, selective screening was and continues to be practiced throughout the United States and the rest of the world.30–32,35,36 For hospitals, states, or countries with constrained health care budgets, information on incremental cost may assist in decisions about how to allocate resources.62
Policy Implications
The results of our model have important policy implications. For states that are interested in implementing or improving a UNHS program, this analysis demonstrates that a program’s short-term cost-effectiveness depends not only on the test properties of the screening machines and protocols used but also on the ability to ensure follow-up of infants who do not pass screening tests. This analysis also shows that UNHS will be cost saving in the long run only if a high proportion of deaf infants achieve normal language outcomes and increased productivity after early intervention. Thus, health care providers and educators of the deaf and hard of hearing must work with states to ensure that early detection results in early intervention, without the delays reported in previous studies,12,13,16 and to demonstrate that early intervention programs are successful in improving the language abilities and long-term productivity of early identified deaf children.
Limitations
The results of this study must be interpreted in the context of our model’s assumptions. This study analyzed only one commonly used hearing screening protocol, but US hospitals use diverse protocols. Although other screening protocols may have different costs and test properties, the results of our model were robust to changes in these variables. Our model was also restricted to infants with bilateral moderate to profound deafness. Including infants with mild or unilateral deafness might increase the PPV of screening, and if early intervention has a similar effect on language outcomes for these children, then UNHS would seem even more cost-effective. We chose to limit our analysis to infants with bilateral moderate to profound deafness because it is unclear that infants with mild or unilateral deafness would have the same degree of benefit from early intervention as those with bilateral moderate to profound deafness.60
It is also important to note that the base-case estimates of the impact of early identification and intervention on language outcomes and subsequent educational needs and lifetime productivity were based not on controlled clinical trials but rather on expert opinion and estimates from other cost-effectiveness models. Limitations in the evidence concerning these key assumptions must be considered in interpreting our base-case calculations of incremental long-term cost-effectiveness. However, 2-way sensitivity analyses (Fig 2) allowed us to define conditions under which the different screening strategies would be more or less incrementally cost-effective.
Finally, the denominators in the cost-effectiveness ratios are expressed not in measures of utility, such as quality-adjusted life years (QALYs), but rather in number of infants whose deafness is diagnosed by 6 months or number of deaf children with normal language abilities. QALYs would capture the utility of earlier successful communication between parents and infants and the long-term social and psychological benefits of normal communication for deaf children and adults, as well as the disutility of false-positive tests. Using QALYs would also enable comparisons of cost-effectiveness between hearing screening and other interventions whose cost per QALY has been estimated. Future research should be aimed at measuring the utilities of deaf children and adults given early or late identification and normal or delayed language abilities, as well as parental utilities concerning hearing screening test results.
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Maximizing the rate of follow-up to diagnostic evaluation after positive screening test results could improve the short-term cost-effectiveness of UNHS programs. UNHS has the potential for long-term cost-savings compared with selective hearing screening or no screening. To understand the actual long-term economic effects of UNHS programs, better evidence is needed on how early intervention affects the language, educational and vocational costs, and lifetime productivity of deaf individuals.
|
ACKNOWLEDGMENTS |
|---|
Dr Keren was supported by grant T32 PE 10018 from the Health Resources and Services Administration, US Department of Health and Human Services.
We thank Janet Farrell, Deborah Klein Walker, EdD, and Tracy Osbahr at the Massachusetts Department of Public Health for sharing their experience with and knowledge about universal newborn hearing screening in Massachusetts. We also thank Marilyn Neault, PhD, at Children’s Hospital, Boston, for educating us on hearing screening technologies and protocols, and Scott Smith, MD, and Terrell Clark, PhD, at the Boston Center for the Deaf and Hard of Hearing, for their insight into the educational, social, and cultural issues in the deaf community. We are grateful to our expert panelists, who provided feedback on many of the probability estimates in the model and helped us to derive estimates of the impact of early intervention on language outcomes in the deaf. Finally, we thank Robert Davis, MD, MPH, for thoughtful comments on the manuscript, and Diane Thompson, MS, for sharing the results of her systematic evidence review of newborn hearing screening.
|
FOOTNOTES |
|---|
Received for publication Oct 10, 2001; Accepted Apr 9, 2002.
Reprint requests to (R.K.) Division of General Pediatrics, Children’s Hospital of Philadelphia, 34th and Civic Center Blvd, Philadelphia, PA 19104. E-mail: keren{at}email.chop.edu
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
- Controlled trial of universal neonatal screening for early identification of permanent childhood hearing impairment. Wessex Universal Neonatal Hearing Screening Trial Group. Lancet.1998; 352 :1957 –1964[CrossRef][Web of Science][Medline]
- 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 –357[CrossRef][Web of Science][Medline]
- Mason JA, Herrmann KR. Universal infant hearing screening by automated auditory brainstem response measurement.
Pediatrics.1998; 101
:221
–228
[Abstract/Free Full Text] - Finitzo T, Albright K, O’Neal J. The newborn with hearing loss: detection in the nursery.
Pediatrics.1998; 102
:1452
–1460
[Abstract/Free Full Text] - Mehl AL, Thomson V. Newborn hearing screening: the great omission. Pediatrics.1998; 101(1) . Available at: www.pediatrics.org/cgi/content/full/101/1/e4
- Aidan D, Avan P, Bonfils P. Auditory screening in neonates by means of transient evoked otoacoustic emissions: a report of 2,842 recordings. Ann Otol Rhinol Laryngol.1999; 108 :525 –531[Web of Science][Medline]
- Barsky-Firkser L, Sun S. Universal newborn hearing screenings: a three-year experience. Pediatrics.1997; 99(6) . Available at: www.pediatrics.org/cgi/content/full/99/6/e4
- Watkin PM. Outcomes of neonatal screening for hearing loss by otoacoustic emission.
Arch Dis Child Fetal Neonatal Ed.1996; 75
:F158
–F168
[Abstract/Free Full Text] - Clemens CJ, Davis SA, Bailey AR. The false-positive in universal newborn hearing screening. Pediatrics.2000; 106(1) . Available at: www.pediatrics.org/cgi/content/full/106/1/e7
- Dalzell L, Orlando M, MacDonald M, et al. The New York State universal newborn hearing screening demonstration project: ages of hearing loss identification, hearing aid fitting, and enrollment in early intervention. Ear Hear.2000; 21 :118 –130[CrossRef][Web of Science][Medline]
- Joint Committee on Infant Hearing, American Academy of Audiology, American Academy of Pediatrics, American Speech-Language-Hearing Association, and Directors of Speech and Hearing Programs in State Health and Welfare Agencies. Year 2000 position statement: principles and guidelines for early hearing detection and intervention programs.
Pediatrics.2000; 106
:798
–817
[Free Full Text] - Elssmann SF, Matkin ND, Sabo MP. Early identification of congenital sensorineural hearing impairment. Hear J.1987; 40 :13 –17
- Harrison M, Roush J. Age of suspicion, identification, and intervention for infants and young children with hearing loss: a national study. Ear Hear.1996; 17 :55 –62[Web of Science][Medline]
- Lyon M, Lyon D. Early detection of hearing loss: a follow-up study. Can J Public Health.1986; 77 :221 –224[Web of Science][Medline]
- Mace AL, Wallace KL, Whan MQ, Stelmachowicz PG. Relevant factors in the identification of hearing loss. Ear Hear.1991; 12 :287 –293[Web of Science][Medline]
- Stein LK, Jabaley T, Spitz R, Stoakley D, McGee T. The hearing-impaired infant: patterns of identification and habilitation revisited. Ear Hear.1990; 11 :201 –205[Web of Science][Medline]
- Stein LK. On the real age of identification of congenital hearing loss. Audiol Today.1995; 7 :10 –11
- National Institutes of Health Consensus Development Conference Statement. Early identification of hearing impairment in infants and young children. Int J Pediatr Otorhinolaryngol.1993; 27 :215 –227[CrossRef][Web of Science][Medline]
- Yoshinaga-Itano C, Coulter D, Thomson V. The Colorado Newborn Hearing Screening Project: effects on speech and language development for children with hearing loss. J Perinatol.2000; 20 :S132 –S137[CrossRef][Medline]
- Yoshinaga-Itano C. Benefits of early intervention for children with hearing loss. Otolaryngol Clin North Am.1999; 32 :1089 –1102[CrossRef][Web of Science][Medline]
- Yoshinaga-Itano C, Apuzzo ML. The development of deaf and hard of hearing children identified early through the high-risk registry. Am Ann Deaf.1998; 143 :416 –424[Web of Science][Medline]
- Yoshinaga-Itano C, Apuzzo ML. Identification of hearing loss after age 18 months is not early enough. Am Ann Deaf.1998; 143 :380 –387[Web of Science][Medline]
- Yoshinaga-Itano C, Sedey AL, Coulter DK, Mehl AL. Language of early- and later-identified children with hearing loss.
Pediatrics.1998; 102
:1161
–1171
[Abstract/Free Full Text] - Calderon R, Naidu S. Further support for the benefits of early identification and intervention for children with hearing loss. Volta Rev.2000; 100 :53 –84
- Mayne AM, Yoshinaga-Itano C, Sedey AL, Carey A. Expressive vocabulary development of infants and toddlers who are deaf or hard of hearing. Volta Rev.2000; 100 :1 –28
- Moeller MP. Early intervention and language development in children who are deaf and hard of hearing. Pediatrics.2000; 106(3) . Available at: www.pediatrics.org/cgi/content/full/106/3/e43
- Barringer D, Mauk G. Survey of parents perceptions regarding hospital-based newborn hearing screening. Audiol Today.1997; 9 :18 –19
- Weichbold V, Welzl-Mueller K. Maternal concern about positive test results in universal newborn hearing screening.
Pediatrics.2001; 108
:1111
–1116
[Abstract/Free Full Text] - Kennedy CR. Controlled trial of universal neonatal screening for early identification of permanent childhood hearing impairment: coverage, positive predictive value, effect on mothers and incremental yield. Wessex Universal Neonatal Screening Trial Group. Acta Paediatr Suppl.1999; 88 :73 –75[CrossRef][Medline]
- Eden D, Ford RP, Hunter MF, Malpas TJ, Darlow B, Gourley J. Audiological screening of neonatal intensive care unit graduates at high risk of sensorineural hearing loss. N Z Med J.2000; 113 :182 –183[Web of Science][Medline]
- Van Riper LA, Kileny PR. ABR hearing screening for high-risk infants. Am J Otol.1999; 20 :516 –521[Web of Science][Medline]
- Wynne MK, Molloy T, Bloom BT. The high-risk register for hearing loss in Kansas: some preliminary data. Kans Med.1992; 93 :198 –202[Medline]
- Mauk GW, White KR, Mortensen LB, Behrens TR. The effectiveness of screening programs based on high-risk characteristics in early identification of hearing impairment. Ear Hear.1991; 12 :312 –319[Web of Science][Medline]
- Meyer C, Witte J, Hildmann A, et al. Neonatal screening for hearing disorders in infants at risk: incidence, risk factors, and follow-up.
Pediatrics.1999; 104
:900
–904
[Abstract/Free Full Text] - McClelland RJ, Watson DR, Lawless V, Houston HG, Adams D. Reliability and effectiveness of screening for hearing loss in high risk neonates. BMJ.1992; 304 :806 –809
- Hess M, Finckh-Kramer U, Bartsch M, Kewitz G, Versmold H, Gross M. Hearing screening in at-risk neonate cohort. Int J Pediatr Otorhinolaryngol.1998; 46 :81 –89[CrossRef][Web of Science][Medline]
- Johnson JL, Mauk GW, Takekawa KM, Simon PR, Sia CCJ, Blackwell PM. Implementing a statewide system of services for infants and toddlers with hearing disabilities. Semin Hear.1993; 14 :105 –118
- Kezirian EJ, White KR, Yueh B, Sullivan SD. Cost and cost-effectiveness of universal screening for hearing loss in newborns. Otolaryngol Head Neck Surg.2001; 124 :359 –367[CrossRef][Web of Science][Medline]
- Gorga MP, Preissler K, Simmons J, Walker L, Hoover B. Some issues relevant to establishing a universal newborn hearing screening program. J Am Acad Audiol.2001; 12 :101 –112[Medline]
- Maxon AB, White KR, Behrens TR, Vohr BR. Referral rates and cost efficiency in a universal newborn hearing screening program using transient evoked otoacoustic emissions. J Am Acad Audiol.1995; 6 :271 –277[Medline]
- Dort JC, Tobolski C, Brown D. Screening strategies for neonatal hearing loss: which test is best? J Otolaryngol.2000; 29 :206 –210[Web of Science][Medline]
- Grein AJ, Lemons M, Weiner GM. What is the most cost effective strategy for implementing universal newborn hearing screening [abstr 1107]. Pediatr Res.2001; 49 :195A
- Turner RG. Modeling the cost and performance of early identification protocols. J Am Acad Audiol.1991; 2 :195 –205[Medline]
- Turner RG. Comparison of four hearing screening protocols. J Am Acad Audiol.1992; 3 :200 –207[Medline]
- Turner RG. Factors that determine the cost and performance of early identification protocols. J Am Acad Audiol.1992; 3 :233 –241[Medline]
- Kemper AR, Downs SM. A cost-effectiveness analysis of newborn hearing screening strategies.
Arch Pediatr Adolesc Med.2000; 154
:484
–488
[Abstract/Free Full Text] - Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost-Effectiveness in Health and Medicine. New York, NY: Oxford University Press; 1996
- Spivak L, Dalzell L, Berg A, et al. New York State universal newborn hearing screening demonstration project: inpatient outcome measures. Ear Hear.2000; 21 :92 –103
- van Straaten HL. Automated auditory brainstem response in neonatal hearing screening. Acta Paediatr Suppl.1999; 88 :76 –79
- Norton SJ, Gorga MP, Widen JE, et al. Identification of neonatal hearing impairment: evaluation of transient evoked otoacoustic emission, distortion product otoacoustic emission, and auditory brain stem response test performance. Ear Hear.2000; 21 :508 –528[CrossRef][Web of Science][Medline]
- Mehl AL, Thomson MA. The Colorado newborn hearing screening project, 1992–1999: on the threshold of effective population-based universal newborn hearing screening. Pediatrics.2002; 109(1) . Available at: www.pediatrics.org/cgi/content/full/109/1/e7
- Clemens CJ, Davis SA. Minimizing false-positives in universal newborn hearing screening: a simple solution. Pediatrics.2001; 107(3) . Available at: www.pediatrics.org/cgi/content/full/107/3/e29
- Prieve B, Dalzell L, Berg A, et al. The New York State universal newborn hearing screening demonstration project: outpatient outcome measures. Ear Hear.2000; 21 :104 –117
- Bureau of Labor Statistics. Consumer Price Index: all urban consumers. Available at: 146.142.4.24/cgi-bin/surveymost?cu. Accessed November 2000
- National Center for Hearing Assessment and Management. Universal newborn hearing screening: summary statistics of UNHS in the United States. Available at: www.infanthearing.org/status/unhsstate.html
- Mohr PE, Feldman JJ, Dunbar JL, et al. The societal costs of severe to profound hearing loss in the United States. Int J Technol Assess Health Care.2000; 16 :1120 –1135[CrossRef][Web of Science][Medline]
- Cheng A, Rubin H, Powe N, Mellon N, Francis H, Niparko J. Cost-utility analysis of the cochlear implant in children.
JAMA.2000; 284
:850
–856
[Abstract/Free Full Text] - Pollitt RJ. Neonatal screening for inborn errors of metabolism: cost, yield and outcome. Health Technol Assess.1997; 1 :i –iv[Medline]
- Thompson DC. Universal newborn hearing screening: summary of evidence.
JAMA.2001; 286
:2000
–2010
[Abstract/Free Full Text] - Bess FH, Paradise JL. Universal screening for infant hearing impairment: not simple, not risk-free, not necessarily beneficial, and not presently justified.
Pediatrics.1994; 93
:330
–334
[Abstract/Free Full Text] - Paradise JL. Universal newborn hearing screening: should we leap before we look?
Pediatrics.1999; 103
:670
–672
[Free Full Text] - Mencher GT, Devoe SJ. Universal newborn screening: a dream realized or a nightmare in the making? Scand Audiol Suppl.2001; 53 :15 –21
PEDIATRICS (ISSN 1098-4275). ©2002 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
This article has been cited by other articles:
|
|
 |
|
  J. A. F. Zupancic, A. M. Hibbs, L. Palermo, W. E. Truog, A. Cnaan, D. M. Black, P. L. Ballard, S. R. Wadlinger, R. A. Ballard, and and the NO CLD Trial Group Economic Evaluation of Inhaled Nitric Oxide in Preterm Infants Undergoing Mechanical Ventilation Pediatrics, November 1, 2009; 124(5): 1325 - 1332. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. W Chang, T J O-Lee, and M. Price Evaluation of unilateral referrals on neonatal hearing screening J Med Screen, March 1, 2009; 16(1): 17 - 21. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. L. Kamholz, C. H. Cole, J. E. Gray, and J. A. F. Zupancic Cost-effectiveness of Early Treatment for Retinopathy of Prematurity Pediatrics, January 1, 2009; 123(1): 262 - 269. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. H. Schultz, A. R. Localio, B. J. Clark, C. Ravishankar, N. Videon, and S. E. Kimmel Epidemiologic Features of the Presentation of Critical Congenital Heart Disease: Implications for Screening Pediatrics, April 1, 2008; 121(4): 751 - 757. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. D. Grosse and D. S. Ross Cost Savings From Universal Newborn Hearing Screening Pediatrics, August 1, 2006; 118(2): 844a - 845. [Full Text] [PDF]
|
|
|
|
|
|
L. Schroeder, S. Petrou, C. Kennedy, D. McCann, C. Law, P. M. Watkin, S. Worsfold, and H. M. Yuen The Economic Costs of Congenital Bilateral Permanent Childhood Hearing Impairment Pediatrics, April 1, 2006; 117(4): 1101 - 1112. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Grant State Strategies to Contain Costs in the Early Intervention Program: Policy and Evidence Topics in Early Childhood Special Education, January 1, 2005; 25(4): 243 - 250. [Abstract] [PDF]
|
|
|
|
|
|
L. N. Venditti, C. P. Venditti, G. T. Berry, P. B. Kaplan, E. M. Kaye, H. Glick, and C. A. Stanley Newborn Screening by Tandem Mass Spectrometry for Medium-Chain Acyl-CoA Dehydrogenase Deficiency: A Cost-Effectiveness Analysis Pediatrics, November 1, 2003; 112(5): 1005 - 1015. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
