Screening of the hearing of newborns - Update

Abstract

Introduction

Permanent congenital bilateral hearing loss (CHL) of moderate or greater degree (≥40 dB HL) is a rare disease, with a prevalence of about 1 to 3 per 1000 births. However, it is one of the most frequent congenital diseases. Reliance on physician observation and parental recognition has not been successful in the past in detecting significant hearing loss in the first year of life. With this strategy significant hearing losses have been detected in the second year of life. With two objective technologies based on physiologic response to sound, otoacoustic emissions (OAE) and auditory brainstem response (ABR) hearing screening in the first days of life is made possible.

Objectives

The objective of this health technology assessment report is to update the evaluation on clinical effectiveness and cost-effectiveness of newborn hearing screening programs. Universal newborn hearing screening (UHNS) (i), selective screening of high risk newborns (ii), and the absence of a systematic screening program are compared for age at identification and age at hearing aid fitting of children with hearing loss. Secondly the potential benefits of early intervention are analysed. Costs and cost-effectiveness of newborn hearing screening programs are determined. This report is intended to make a contribution to the decision making whether and under which conditions a newborn hearing screening program should be reimbursed by the statutory sickness funds in Germany.

Methods

This health technology assessment report updates a former health technology assessment (Kunze et al. 2004 [1]). A systematic review of the literature was conducted, based on a documented search and selection of the literature using predefined inclusion and exclusion criteria and a documented extraction and appraisal of the included studies. To assess the cost-effectiveness of the different screening strategies in Germany the decision analytic Markov state model which had been developed in our former health technology assessment report was updated.

Results

Universal newborn hearing screening programs are able to substantially reduce the age at identification and the age at intervention of children with CHL to six months of age in the German health care setting. High coverage rates, low fail rates and - if tracking systems are implemented – high follow-up-rates to diagnostic evaluation for test positives were achieved. New publications on potential benefits of early intervention could not be retrieved. For a final assessment of cost-effectiveness of newborn hearing screening evidence based long-term data are lacking. Decision analytic models with lifelong time horizon assuming that early detection results in improved language abilities and lower educational costs and higher life time productivity showed a potential of UNHS for long term cost savings compared to selective screening and no screening. For the short-term cost-effectiveness with a time horizon up to diagnostic evaluation more evidence based data are available. The average costs per case diagnosed range from 16,000 EURO to 33,600 EURO in Germany and hence are comparable to the cost of other implemented newborn screening programs. Empirical data for cost of selective screening in the German health care setting are lacking. Our decision analytic model shows that selective screening is more cost-effective but detects only 50% of all cases of congenital hearing loss.

Discussion

There is good evidence that UNHS-Programs with appropriate quality management can reduce the age at start of intervention below six months. Up to now there is no indication of considerable negative consequences of screening for children with false positive test results and their parents. However, it is more difficult to prove the efficacy of early intervention to improve long-term outcomes. Randomized clinical trials of the efficacy of early intervention for children with CHL hearing losses are inappropriate because of ethical reasons. Prospective cohort studies with long-term outcomes of rare diseases are costly, take a long time and simultaneously substantial benefits of early intervention for language development seem likely.

Conclusions

A UNHS-Program should be implemented in Germany and be reimbursed by the statutory sickness funds. To achieve high coverage and because of better conditions for obtaining low false positive rates UNHS should be performed in hospital after birth. For outpatient deliveries additionally screening measures in an outpatient setting must be provided.

Executive Summary

1. Introduction

Permanent congenital bilateral hearing loss of moderate or greater degree is a rare disease, with a prevalence of about 1 to 3 per 1000 births. However, the impact of such hearing loss on the development of language and communication can be substantial if effective intervention is delayed. The mean age at identification of a permanent congenital hearing loss in Germany was 47 months of age for moderate and 22 months for profound hearing loss for birth cohorts after 1987. During the nineties automated screening devices were developed for two objective technologies based on physiologic response to sound, the auditory brainstem response (ABR) and otoacoustic emissions (OAE). These devices permit screening of newborns by non-professional personnel within short test times and without considerable discomfort for the child. Since 1994 several statements of Scientific Medical Societies were published recommending the implementation of universal newborn hearing screening programs (UNHS). The goal of universal detection of hearing loss in infants before three months of age, with appropriate intervention not later than six months of age is endorsed. Universal hearing screening is supported, because screening of high-risk populations alone could only identify about 50% of newborns with significant hearing loss. Reliance on physician observation and/or parental recognition is judged as not have been successful in the past in detecting hearing loss in the first year of life. Economic consequences of an early detection and intervention of congenital hearing impairment are also considerable. While health care costs might be increased by screening programs because screening costs and costs for diagnostic evaluation for false positive test results, educational costs probably could be saved if the proportion of children with hearing loss, who can be taught in mainstream classes, could be increased. Indirect costs (productivity losses) may also be influenced.

2. Objectives

The objective of this health technology assessment was to evaluate the clinical effectiveness and cost-effectiveness of a UNHS compared to selective screening of high risk newborns or no systematic screening. This report is intended to make a contribution to the decision making whether and under which conditions a newborn hearing screening program should be reimbursed by the statutory sickness funds in Germany. The background of connatal hearing disorders and its epidemiology is described and the evidence concerning age at identification of a congenital hearing impairment, age at hearing-aid fitting and start of accompanying therapies with and without a systematic newborn hearing screening and the coverage of screening programs is collected. In addition the question is addressed whether an earlier diagnosis of a permanent hearing impairment and an earlier application of available therapies are clinically more effective than a delayed start of treatment making newborn hearing screening an alternative to the existing practice. Available data on the consequences of delayed detection and treatment of permanent hearing impairment concerning language and speech development, but also educational and occupational achievements are described. Evidence on sensitivity and specificity of screening tests based on evoked otoacoustic emissions (EOAE) or brainstem evoked response audiometry (BERA) is analyzed. The clinical effectiveness of UNHS is compared to selective screening of high risk children and to no systematic screening. Furthermore, the consequences of false positive and false negative tests are addressed.

The objective of the economic evaluation and decision analytical modelling is to compare UNHS, risk screening and the existing practice of no systematic screening for cost-effectiveness. Evaluated outcomes are: costs per screened child, costs for early and late detected children until hearing-aid fitting, costs for education and also indirect costs by productivity losses caused by hearing impairment. Costs are related to health outcomes using cost-effectiveness ratios.

3. Clinical evaluation

3.1 Methods

A systematic review of clinical and economic effectiveness has been conducted based on a literature search in 18 electronic databases and a search of internet sites related to the topic. The time horizon of the literature search was limited from 2001 to March 2005. Two independent reviewers selected studies to be entered in this assessment using a priori defined inclusion- and exclusion criteria specifying study population, intervention, comparison technology, outcomes and study types. Study quality was assessed with instruments developed by the German Scientific Working Group Technology Assessment for Health Care and by the authors. Data of included studies were extracted and summarized in evidence tables.

3.2 Results

The literature search retrieved 1950 hits. After application of inclusion and exclusion criteria 35 publications were included in the syntheses of information of clinical effectiveness: ten publications on the epidemiology of permanent congenital hearing loss, two on potential benefits of early intervention, two publications on sensitivity and specificity of screening devices, two publications on comparisons of screening devices, 15 publications on newborn hearing screening programs (NHS), and four publications on the attitude of parents to NHS or on consequences of false test results.

Age at identification and intervention without Newborn Hearing Sreening

The median age at identification ranged from 18 to 30 months of age (seven studies). The median age at hearing-aid fitting ranged from 22 to 34 months (four studies). Only two German studies were identified, one with poor study quality and one with to short follow-up time to identify all cases with hearing loss.

Age at identification and intervention with Universal Newborn Hearing Screening

See “Effectiveness of Newborn Hearing Screening-Programs”.

Potential benefits of early intervention

Since our former HTA-Report no new studies could be identified comparing language development, or educational attainment in children with hearing loss receiving early or late intervention with a prospective study design. Two studies examined language development or quality of life respectively of a birth cohort of seven to eight years old children with hearing loss. The influence of age at identification on language development (mean 21.6 months) was analysed by regression analysis. No association between age at identification and language development has been found. However, the proportion of children with an early treatment start was very small. Language development was 25% beneath the values that could be expected according to nonverbal intelligence and quality of life in the domain of psychosocial functioning was statistically significant lower than an age matched cohort of hearing children.

Test sensitivity and specificity of OAE- and BERA-Screening devices

The test sensitivity of four different automated screening devices (Echoscreen/S-TOAE, GSI60/D-DPOAE, Blitzbera/AABR, Eroscan/S-DPOAE modus, S-TOAE modus) ranged from 99.4% to 100%. The test specificity for three devices (Echoscreen/S-TOAE, GSI60/D-DPOAE, Blitzbera/AABR) ranged from 82.4% to 92.3%.

Effectiveness of Newborn Hearing Screening-Programs

The achieved coverage rates of 13 included NHS screening programs ranged from 65% to 100% with a median coverage of 94.2%. For regional UNHS in Germany (three studies) coverage ranged from 87.4% to 95.3%. Fail rates for bi- and unilateral fails after the first stage of screening in relation to all screened newborns ranged from 2.5% to 20% with a median of 4.3% (nine studies). For bilateral fails they ranged from 0.5% to 10.6% with a Median of 2.3%. The median fail rate for bilateral fails after the last screening stage was 1.5% (range 0.08%-3.9%). The percentage of false positive test results after the first stage of screening in relation to all screened newborns ranged from 0.13% to 9.3% with a median of 2.4% (nine studies). The median of the percentage of false positive test results after the last stage of screening in relation to the number of newborns with diagnostic evaluation was 72.7% (range 0%-93.9%, nine studies). In only two studies the number of identified CHL in the whole region was collected to compare it to the number identified by the UNHS. The proportion of false negative test results was 8.4% and 5.5% respectively. Until to the last stage of screening 0.3% to 2% (median 0.72%, nine studies) of test positive newborns in relation to all newborns screened were lost to follow-up. The range of the proportion of newborns not getting a diagnostic evaluation after having completed the last stage of screening was 0%-23.8% (median 0%, ten studies). The mean age at identification was reduced to about three months of age (range of means 1.2–4.3 months). The proportion of children with bilateral hearing losses fitted with hearing-aids or treated by surgery ranged from 77.8% to 100% (median 98.4%, six studies). The mean age at intervention was reduced to six months of age in UNHS-programs reporting this outcome (six studies).

Consequences of false screening results for children and their parents

Four studies reported on the attitude of parents to NHS. The parents in these studies had different experiences with NHS: Mothers without personal experience, mothers of screened babies where the test result was not yet known, mothers of children taking part in a rescreening because they failed the first screening test, parents of children with diagnosed hearing loss. The majority of mothers and parents did not feel substantial anxiety after a positive test result.

4. Economic evaluation

4.1 Methods

Literature search and selection of the studies to be included in the systematic review of cost-effectiveness has been performed analogous to the methods in the systematic review of clinical effectiveness. Costs, cost components and incremental cost-effectiveness-ratios (IKER) are systematically summarized in evidence tables. For currency conversions to EURO 2004 values purchasing power parities of the WHO are used. Adjustments for inflation are made using the general consumer price index of the Statistisches Bundesamt.

4.2 Results

Five studies were included in the systematic review of cost-effectiveness, three cost-analyses and two cost-effectiveness analyses. With the exception of one study with a lifelong time horizon, the studies had only a short term time horizon covering merely the period until diagnostic evaluation of hearing loss. Furthermore, these studies were heterogeneous regarding sample size, screening procedures, and included cost components. The average costs per screened child ranged from 14.3 EURO to 49.37 EURO (2004 values, four studies). The average costs per identified case of hearing impairment in the well baby care unit were 25,311 EURO and 43,837 EURO respectively (two studies) and 1237 EURO for high risk babies in an UNHS-Program and 10,151 EURO in selective screening (two studies). In UNHS-Programs the average costs for the whole population (WBN and high risk) ranged from 4888 EURO to 33,613 EURO per case (three studies). The IKER of selective screening versus no screening and of UNHS versus risk screening in one study were 16,648 EURO and 44,971 EURO respectively per additionally early detected child (until the sixth month of life).

5. Decision analysis

The short-term submodel including diagnostic outcomes of the former developed two-component Markov state model was updated with new data from the present systematic review. IKER were calculated. The long-term component could not be updated because no new data were available on long-term outcomes and costs.

5.1 Model for diagnostic outcomes

Because screening aims at an early detection of CHL, the primary outcome of the diagnostic submodel is the number of detected child months after six, twelf and 120 months and the number of children detected until six months of age. For the diagnostic model, cost-effectiveness was expressed as costs per additionally detected child month and cost per additionally detected child at the age of six months. The primary cost-effectiveness outcome of the education submodel was saved educational costs per additionally detected child. Time horizon for diagnostic model was 120 months.

5.2 Results

The model parameters of the diagnostic model were derived from the published literature. Assuming a prevalence of congenital hearing impairment of 0.15%, UNHS detected 75% of children with hearing impairment within the first six months of life, compared to 49% detected with risk screening and 24% detected without systematic screening. UNHS yields 72% of all possible child months after six months and 77% after 12 months. Risk screening yielded 43% after six months and no systematic screening (the alternative which is the most common in Germany at the moment) yields only 23% child months after six months. Sensitivity analysis shows that results strongly depend on the prevalence of hearing impairment. The costs of detection are varying considerably between strategies. Based on the decision analysis, total costs for screening of 100,000 newborns until six months of life are 1.52 Mio. EURO for UNHS, 0.21 Mio. EURO for risk screening, and 13,370 EURO for the absence of systematic screening. The costs per detected child are 10,306 EURO for UNHS, 2592 EURO for selective screening, and 1402 EURO without systematic screening respectively. The costs for detection strongly depended on the costs and sensitivity of screening tests, the probability of a false suspicion of hearing impairment among healthy children, and the coverage of the screening program.

6. Comprehensive discussion

Is congenital permanent hearing loss a target for screening?

The new identified studies confirmed the results of the former HTA-report on the epidemiology of congenital permanent hearing loss: CHL is with a prevalence of 1 of 1000 newborns one of the most frequent congenital diseases. Without screening the age of detection is in the second year of life. The most frequent (50%) moderate degree of CHL is detected in the third year of life. The evidence for benefits of early intervention for long-term outcomes (language development, quality of life, educational attainment) could not be improved. The evidence that attaining a normal language development with early intervention is basically possible although the influence of age at intervention might be over- or underestimated has already been shown in studies on this item in the former HTA-report. Evidence for the efficacy of early intervention by randomized trials can not be provided because of ethical reasons. Prospective cohort studies with long-term outcomes of rare diseases are costly, take a long time and simultaneously substantial benefits of early intervention for language development seem likely according to the theory of language development.

Do appropriate screening tests exist?

There is only limited evidence about sensitivity and specificity of automated screening devices. The two diagnostic studies identified, showed considerable limitations in study quality. Furthermore program sensitivity can only be measured if all cases of CHL in the region of the NHS are identified. Only one study with a sufficiently long follow-up has been identified. Besides test accuracy, coverage and follow-up rate contribute to program sensitivity and a negative screening test in a case of late identified hearing loss may also be due to late onset hearing impairment and has not to represent a false test result. After the present experience with large size hearing screening programs the specificity and sensitivity of the screening devices seems to be high enough with the above discussed limited evidence for sensitivity.

Are Newborn Hearing Screening-Programs effective in reducing age at detection and age at appropriate intervention (fitting of hearing Aids, Cochlea Implant)?

As distinguished to the former HTA-report in the present report five publications of UNHS-programs in Germany could be identified. They report high coverage rates, low fail rates and - if tracking systems are implemented – high follow-up-rates to diagnostic evaluation for test positives. Only two of the German studies report on age at diagnosis and hearing-aid fitting. They achieved the goal of starting intervention until six months of age. The feasibility study in Hannover has shown that test positive rates have been strongly influenced by the health care setting. Screening tests in medical practices with babies with a higher mean age at time of testing show significantly higher test positive rates of about 10% vs. 2.3% or 3.9% in the well baby nursery or the children’s hospital respectively. Another important issue is that regional differences in capacity for diagnostic evaluation must be taken into account.

Negative consequences of screening

CHL is a rare disease. Of 740,000 newborns per year about 740 babies are affected by an at least moderate CHL. Assuming a fail-rate of 2% 14,800 children would have to undergo further testing, 14,060 children of them will not benefit from the screening. Therefore it is important, that these healthy children will not be harmed by the screening program. The screening tests themselves are not invasive and are normally performed during sleep. For diagnostic evaluation a proportion of children receive an oral sedative and furthermore it cannot be completely excluded that a hearing child will be fitted with hearing-aids. Up to the present on a basis of some 100,000 screened newborns in the published literature no adverse events during diagnostic evaluation were reported. Concerns that positive test results might raise anxiety in parents and have a negative impact on the bonding process are not confirmed. The majority of parents in two studies did not feel substantial anxiety due to positive test results. Two qualitative studies give a more differentiated picture of the feelings of parents after receiving positive test results during the screening process, but they had small sample sizes and low response rates thus the generalisability of these studies are questionable. On the other hand it has to be considered that the majority of hearing impaired would get the opportunity of a normal language and psychosocial development and the consequences of lifelong disability might be prevented. Ultimately ethical principles as well will determine which risk and how much inconveniences seem to be acceptable to healthy persons to improve the opportunities of persons affected by diseases or disabilities. In any case quality management seems to be indicated to optimise the effectiveness and acceptability of screening.

Is a Newborn Hearing Screening Program cost-effective?

The assessment of the cost-effectiveness of UNHS depends strongly on the selection of the perspective and time horizon of costs and benefits to be included into the evaluation. From the health care perspective screening could only improve effectiveness. Early detected cases of CHL are not expected to save costs for the health care system in comparison to late detected cases. Besides from the perspective of the affected children a big positive effect both on costs and effects can be expected for the educational system and for the national economy regarding productivity losses. However, in comparison to the former HTA-report the evidence concerning societal and long-term perspective of cost-effectiveness analysis did not improve. On the other hand new data for the short term horizon for Germany are now available from the feasibility study in Hannover at least concerning average cost of screening and per detected case of CHL. An incremental cost-effectiveness analysis comparing additional cost of screening per additionally early detected case of CHL to selective screening or no screening has not been done. Rather an incremental analysis comparing different health care settings to the setting of the model project has been given. The lowest average costs of screening (16.57 EURO) and per case detected (15,961 EURO) were reported from model 1, where all newborns in well baby nurseries and children’s hospital were screened by a two-stage TEOAE-screening program and only newborns missed by screening and outpatient deliveries were screened in certified ear-nose-throat practices. The highest costs were calculated for model 2 (34.67 EURO per baby screened, 33,613 EURO per case detected) where only babies in the children’s hospital have been screened there and all other newborns were screened during the routinely performed examination between the fourth and sixth week of life in pediatric practices (U3). In model 3 all newborns in well baby nurseries and children’s hospital were screened in hospital, only the children missed by screening in the hospital were screened in pediatric practices during U3-examination. Intermediate average costs of 23.02 EURO per baby screened and 22,417 EURO per case detected were calculated. A comparison of incremental costs of these different settings shows that additionally detected case by model 3 versus model 1 are accompanied by additional costs of 139,568 EURO.

Unfortunately the data provided in cost-effectiveness studies of other newborn screening programs do not permit to compare incrementel cost-effectiveness ratios (IKER) for the short-term horizon until diagnostic evaluation. Only IKER provides the information about additional costs for an additionally by screening identified case, but solely average costs are reported. The average costs per detected case of UNHS lie in the range of other newborn screening programs in European countries already implemented. But for Germany no data on the cost-effectiveness of other newborn screening programs are available.

7. Conclusions and recommendations

It has to be assumed that the mean age at detection of CHL in regions without UNHS in Germany lies still in the second half of the second year of life. A spontaneous improvement cannot be assumed without directed measures. Evaluation of the clinical and economic evidence shows that UNHS supports early intervention. Despite the lack of data and the flaws in study design on the issue of the benefit of early intervention, it seems likely that early intervention improves language development and the opportunity for a normal educational and vocational development increases. From a long-term societal perspective it can be assumed that UNHS is the most cost-effective strategy and selective screening is more cost-effective than no screening, but the evidence is uncertain. However, short-term cost-effectiveness of UNHS is in the range of other accepted newborn screening programs. In due consideration of the evaluation of clinical and economic evidence and decision analysis the authors of the present report recommend the implementation of a UNHS-program in Germany reimbursed by the statutory sickness funds. To achieve high coverage and because of better conditions for obtaining low false positive rates UNHS should be performed in hospital after delivery, for outpatient deliveries additionally screening measures in an outpatient setting must be provided. Institutions performing UNHS should apply quality management measures of which a general framework has to be established by the institutions of the joint self-governing body of the German health care system. Babies failing the screening tests must be reliably referred to diagnostic evaluation.