|
|
Background Children with bilateral permanent hearing impairment often have impaired language and speech abilities. However, the effects of universal newborn screening for permanent bilateral childhood hearing impairment and the effects of confirmation of hearing impairment by nine months of age on subsequent verbal abilities are uncertain.
Methods We studied 120 children with bilateral permanent hearing impairment identified from a large birth cohort in southern England, at a mean of 7.9 years of age. Of the 120 children, 61 were born during periods with universal newborn screening and 57 had hearing impairment that was confirmed by nine months of age. The primary outcomes were language as compared with nonverbal ability and speech expressed as z scores (the number of standard deviations by which the score differed from the mean score among 63 age-matched children with normal hearing), adjusted for the severity of the hearing impairment and for maternal education.
Results Confirmation of hearing impairment by nine months of age was associated with higher adjusted mean z scores for language as compared with nonverbal ability (adjusted mean difference for receptive language, 0.82; 95 percent confidence interval, 0.31 to 1.33; and adjusted mean difference for expressive language, 0.70; 95 percent confidence interval, 0.13 to 1.26). Birth during periods with universal newborn screening was also associated with higher adjusted z scores for receptive language as compared with nonverbal ability (adjusted mean difference, 0.60; 95 percent confidence interval, 0.07 to 1.13), although the z scores for expressive language as compared with nonverbal ability were not significantly higher. Speech scores did not differ significantly between those who were exposed to newborn screening or early confirmation and those who were not.
Conclusions Early detection of childhood hearing impairment was associated with higher scores for language but not for speech in midchildhood.
Currently, screening for bilateral permanent childhood hearing impairment, with the use of transiently evoked otoacoustic emissions and automated measurement of auditory brain-stem responses, is recommended for all infants before the age of three months in the United States,6 the United Kingdom,7 and Europe.8 The value of these recommendations is supported by studies showing that enrollment in an intervention program by nine months of age, as compared with later intervention, is associated with improvements in the verbal ability quotient by as much as 19 points9 (equivalent to 0.5 to 0.6 SD) and that birth during periods in which universal hearing screening of newborns was in place is associated with a similar benefit.10 The U.S. Preventive Services Task Force, however, has rated the quality of evidence linking early treatment or birth during periods with universal newborn hearing screening with improved language function as fair or poor.3
In a previous controlled trial in the Wessex region of southern England, we showed that universal newborn screening increased the rate of early referral (i.e., before six months of age) for audiologic assessment of babies with bilateral permanent childhood hearing impairment, defined as a hearing loss of at least 40 dB hearing level (HL), on two assessments at least 12 months apart.11,12 In the present study, we assessed speech and oral language abilities in a sample of children with bilateral permanent childhood hearing impairment, including children enrolled in the earlier trial, and the relationship of these measures to the timing of confirmation of hearing impairment (by nine months of age or later) and to the availability of universal screening of newborns.
Methods
The study sample included all children with bilateral permanent childhood hearing impairment of at least 40 dB HL identified from a cohort of 157,000 children born in eight districts of southern England. We did not include children with a known postnatal cause of bilateral permanent childhood hearing impairment (e.g., bacterial meningitis). The children in the sample were born between 1993 and 1996 in four districts in the Wessex region or between 1992 and 1997 in two pairs of adjacent districts in the Greater London region.
The four districts in the Wessex subgroup had provided the birth cohort for the Wessex trial, in which a program of universal newborn screening was or was not in place in each pair of districts for birth cohorts born in alternate four- or six-month periods.11,12 The Greater London subgroup consisted of children born in the only two districts in the United Kingdom that provided universal newborn screening for permanent childhood hearing impairment in the early 1990s (Whipps Cross and Hillingdon13,14) and in two other districts, one adjacent to Whipps Cross and one adjacent to Hillingdon.15 Other than variation in providing the newborn screening, protocols for the identification and confirmation of bilateral permanent childhood hearing impairment, previously reported,11,12,13,14,15 were similar at all sites.
Follow-up of this birth cohort included audiologic screening at school entry and information both from multiple sources within the participating regions and from primary care teams in other regions. We obtained details of the detection and management of all cases of bilateral permanent childhood hearing impairment from pediatric audiologists, family practitioners, and other involved professionals and by review of the case records of the audiology service in each district. The severity of hearing loss was categorized from recent audiologic records as moderate (40 to 69 dB HL), severe (70 to 94 dB HL), or profound (
95 dB HL) (Table 1) according to four-frequency averaging of the pure-tone thresholds from 500 to 2000 Hz (or, if pure-tone thresholds were unavailable, sound fields and electrophysiologic-test results).
|
Two researchers unaware of the child's early hearing or audiologic history evaluated the child during a home visit. One researcher interviewed the principal caregiver (usually the mother), which included completion of the speech scale of the Children's Communication Checklist.17 Simultaneously, the child was assessed by the other researcher in a separate space on the following: the Test for Reception of Grammar,18 the British Picture Vocabulary Scale19 (receptive language); the Renfrew Bus Story Test20 (expressive language); and Raven's Progressive Matrices Test (nonverbal abilities).21 For all these measures, a higher score indicates better function. Normal receptive language is the ability to understand communication through gestures, facial expressions, and words, whereas expressive language is the ability to express needs with the use of gestures, vocalization, facial expressions, and words.
Other characteristics of the child and family, including maternal education according to the 2001 census in the United Kingdom, were also documented (Table 1). The mean age at assessment of language and speech was 7.9 years (range, 5.4 to 11.7).
Our study was approved by the South and West Multicenter Research Ethics Committee, United Kingdom. Principal caregivers provided written informed consent.
For the purpose of comparisons within the group of children with hearing impairment in this report, we used norms22 obtained from a group of 63 English-speaking children with normal hearing, matched for place of birth and age at assessment with our group of 120 children with hearing impairment. The group mean score and standard-deviation scores in children with normal hearing were used to derive z scores for the children with hearing impairment, equal to the number of standard deviations of the distribution of scores in children with normal hearing by which their age-adjusted score differed from the mean score in children with normal hearing. We also calculated aggregate scores as follows: the z score for receptive language was equal to half the sum of the z score for the Test for Reception of Grammar and the z score for the British Picture Vocabulary Scale; and the z score for expressive language equal to half the sum of the z score for sentence information and the z score for five longest sentences. We calculated difference scores as follows: the z score for a deficit of receptive (or expressive) language as compared with nonverbal skills was equal to the difference between the z score for receptive (or expressive) language and the z score for nonverbal ability.
We assessed the associations between exposure to universal newborn screening (i.e., birth during periods when universal newborn screening was in place), or to confirmation by nine months of age, and age-adjusted individual and aggregate language and speech scores with the use of a two-sample t-test. The preplanned primary outcomes of our study were language and speech scores and differences between language and nonverbal scores at primary-school age after adjustment in a multiple linear regression (Stata software,23 version 8) for severity of hearing impairment, maternal education, and (except in the case of difference scores) nonverbal ability, which were recognized as potential confounders of the primary outcomes. Normality and homogeneity of the residual variance were examined for all measures to ensure that the regression model was appropriate. All reported P values are two-sided. Adjustment was not made for multiple testing.
On the basis of an expected overall sample size of 154 children with bilateral permanent childhood hearing impairment (determined according to the expected rates in the general population), we anticipated a statistical power of 80 percent to detect a difference of 0.5 SD in verbal ability between the two groups, with a two-sided P value of 0.05. We also performed a subgroup analysis of children who were enrolled in the Wessex controlled trial of universal newborn screening.11,12 The subgroup was expected to be balanced with respect to both known and unknown confounding factors, although that trial was not powered for the end points of speech and language.
Results
Seventy-seven infants with bilateral permanent childhood hearing impairment (i.e., 40 dB HL) were identified among the 68,714 infants born during periods in which universal newborn screening was in place, and 91 infants were identified among the 88,019 infants born during periods without universal screening. These numbers are equivalent to an overall prevalence in our sample of 107 per 100,000, which is close to the expected population prevalence of 112 per 100,000.1 Estimates of the completeness of ascertainment in our study sample exceeded 95 percent for both the London and Wessex subgroups.12,13,15 Of the 168 cases identified, the principal caregivers of 120 children gave consent for participation in the study (Figure 1). Of these 120 children, 61 were born during periods with universal newborn screening and 59 during periods without the screening. Participants were similar to nonparticipants with respect to age, sex, and severity of hearing loss. Hearing impairment was confirmed in 72 infants (60 percent) by one year of age, 90 children (75 percent) by two years of age, 106 children (88 percent) by four years of age, and in all 120 children (100 percent) by six years of age.
|
Baseline characteristics, including the severity of hearing impairment, were similar between the 61 children who were exposed to universal newborn screening and the 59 who were not (data not shown), and between 57 children whose impairment was confirmed by nine months of age and 63 whose impairment was confirmed later (Table 1). Confirmation of impairment by nine months of age was significantly more common among children exposed to universal newborn screening (41 of 61 [67 percent]) than among children not exposed to such screening (16 of 59 [27 percent]) (absolute difference, 40 percent; 95 percent confidence interval, 24 to 56 percent; P<0.001).
Children whose impairment was confirmed by nine months of age had significantly higher adjusted mean aggregate scores for receptive language than did children whose hearing impairment was confirmed later (difference in mean z scores, 0.76; 95 percent confidence interval, 0.26 to 1.27) and for expressive language ability (difference in mean z scores, 0.50; 95 percent confidence interval, <0.01 to 1.01). Furthermore, their difference scores (i.e., the z score for language minus the z score for nonverbal ability) showed smaller deficits in receptive and expressive language relative to nonverbal ability (Figure 2A and Table 2). Children who were exposed to universal newborn screening also had higher adjusted mean aggregate scores for receptive language than those who were not exposed (difference in mean z scores, 0.56; 95 percent confidence interval, 0.03 to 1.08), and their difference scores showed a smaller deficit in receptive language relative to nonverbal ability (difference in mean z scores, 0.60; 95 percent confidence interval, 0.07 to 1.13) but no significant difference in expressive language ability (Figure 2B and Table 3).
|
|
|
Associations between early confirmation of hearing impairment or exposure to universal newborn screening and later language abilities were similar in the Wessex and Greater London subgroups (data not shown). In the Wessex subgroup, for whom the chance of confounding factors should have been reduced because the exposure to newborn screening occurred in a controlled trial, the unadjusted difference between receptive language and nonverbal scores in children born during periods with universal newborn screening as compared with those born during periods without newborn screening was 1.00 (95 percent confidence interval, 0.001 to 2.00; P=0.05) (see Table 2 in the Supplementary Appendix, available with the full text of this article at www.nejm.org). In addition, associations between early confirmation of impairment or newborn screening and higher language scores in this cohort were similar to, or higher than, those observed in the whole sample (Table 2 and Table 3, and the Supplementary Appendix).
Discussion
We observed significantly higher scores for language, but not for speech, in midchildhood among a population-based sample of children with bilateral permanent hearing impairment who were exposed to universal newborn screening or who had confirmation of hearing impairment by nine months of age than among those who were not exposed to newborn screening or whose impairment was confirmed after nine months of age. In the case of children whose hearing impairment was confirmed by nine months of age, this difference was equivalent to an increase of 10 to 12 points in the verbal as compared with the nonverbal intelligence quotient.
The estimated size of the benefit of newborn screening and of early confirmation of impairment in this sample may be conservative. This birth cohort was the first in the United Kingdom in which newborn screening was applied, and systems to ensure short intervals between positive results on newborn screening and confirmation of hearing impairment, now standard in the United Kingdom24 and the United States,25,26 were still evolving. Intervals between confirmation and the fitting of hearing aids were also longer than is the current standard of care in the United Kingdom.13,24 Delays in confirmation and intervention might have decreased the benefit to language that was associated with early detection.
We excluded children whose hearing impairment was of known postnatal cause, but a minority of cases may have been postnatal in onset. Children who lost hearing after infancy would be expected to benefit less from early confirmation than children with an unchanging congenital hearing impairment. On the basis of the Wessex cohort, the maximum proportion of children with progressive hearing loss during childhood (derived by adding the proportion with negative results on newborn screening to the proportion with a positive result and in which a subsequent increase in the severity of hearing impairment was documented) was 23 percent12; a lower percentage was derived for the Greater London subgroup. A sensitivity analysis (not shown) based on the estimate that impairment worsens after birth in 23 percent of children suggests that the true benefit to language acquisition for those with congenital and unchanging hearing impairment would have been larger by a factor of 1.05 to 1.30 than the benefit that we reported for the whole sample.
In contrast to the higher language scores observed among children whose hearing impairment was confirmed early, measures of speech did not differ significantly between groups. Speech was assessed on the basis of parental or professional report, rather than by direct measurement, and may reflect a lack of sensitivity to relevant aspects of speech. Objective analysis of audiotaped samples of speech, recorded for the assessments of language, is currently being performed to further evaluate speech in these children.
Data collected in the Greater London subgroup were observational and cannot prove causality. However, conclusions from these data are strengthened by the similarity of results between the Greater London and Wessex subgroups, since exposure or lack of exposure to newborn screening and the resulting variation in early confirmation was quasi-experimental in design in the latter subgroup. The trial among the Wessex subgroup was not statistically powered to detect differences in language and speech in midchildhood.
The severity of hearing impairment measured at the time of speech and language evaluation, rather than at the time of confirmation of impairment, was used in the regression models. This approach minimizes the chance of bias due to possible differences in the estimates of the severity of impairment between groups with early confirmation of impairment and those with late confirmation among children in whom severity increased with age. This adjustment did not in any case materially alter our estimate of the size of the benefits of early confirmation, since the groups with early confirmation and those with late confirmation were similar in their distribution of severity of hearing impairment.
Our data extend findings from previous studies of the relationship between early identification of hearing impairment and later outcomes. Adjusted mean vocabulary scores of children with hearing impairment, assessed at the age of 5 years, were higher in children enrolled before 11 months of age in an early intervention program in Nebraska than in those enrolled at 11 to 23 months of age (by 0.69 SD) or at 24 to 35 months of age (by 0.99 SD).27 Similar findings were reported from a study in Washington State comparing children enrolled before 24 months of age with those enrolled later.28 Both these studies lacked clear criteria for inclusion, selected as participants only those who adhered to the early intervention programs, and used unblinded assessment of children whose language ability may have been greater than that of children whose hearing impairment was identified early but who were not available for follow-up.3 In contrast to these studies, a population-based Australian study reported no benefit of early diagnosis before 6 or 12 months of age on speech, language, and other outcomes at 7 to 8 years of age among children with congenital hearing impairment.29 All three studies excluded children with developmental disabilities, and none included sufficient children enrolled before 12 months of age to estimate reliably the benefit of early intervention in that age group.
In the largest previous study,9 the language abilities of 150 children enrolled in the Colorado Home Intervention Program, assessed at 13 to 36 months by parental report, were 0.5 to 0.6 SD higher (scaled according to subsequently published norms30) in those in whom hearing impairment was identified by 6 months of age and who enrolled after a mean of 3 months more than in those whose impairment was identified later. In a smaller case–control study, these investigators used similar assessment methods and reported a benefit of 0.5 to 0.8 SD to language and speech associated with birth in hospitals providing universal newborn screening, as compared with those not providing it.10 However, these Colorado studies did not adjust for differences between the baseline characteristics of the groups, were subject to possible selection bias, and relied on unblinded assessments of parental reports of language abilities.3 The greater benefits reported, as compared with those reported in the present study, may have arisen from these potential sources of bias or from the differing ages at assessment.
Universal newborn screening and early confirmation of permanent childhood hearing impairment had clinically important benefits to the language abilities of children at primary-school age in this population-based study. Other data from this cohort suggest that such screening may be cost-effective.31 Longer follow-up is needed to establish whether these children have higher academic achievement and continue to show superior language skills at high-school age.
Supported by the Wellcome Trust (061839).
No potential conflict of interest relevant to this article was reported.
We are indebted to the children, families, school staff, specialist teachers of the hearing impaired, speech and language pathologists, and audiology professionals in the local area teams for their help and assistance; to Margaret Baldwin, Joy Bhattacharya, Alyson Bumby, Irene Curtis, Carol Hunter, David Reed, Scott Richards, Peter Savundra, Huw Thomas, Tim Williamson, Jan Nanor, Helen Davis, Shirley Golden, Eleanor Lutman, Kristen Paul, and Helen Ryder; to Julie Brinton for her advice on assessments of speech and language; and to Sue Robinson for her advice on audiology.
Source Information
From the Departments of Child Health (C.R.K., D.C.M., S.W.), Public Health Sciences and Medical Statistics (M.M., H.M.Y.), and Psychology (J.S.), University of Southampton, Southampton; the Institute of Primary Care and General Practice, University of Sheffield, Sheffield (M.J.C.); the Institute of Child Health, University College London, London (C.M.L.); the National Perinatal Epidemiology Unit, University of Oxford, Oxford (S.P.); and the Department of Audiology, Whipps Cross University Hospital, London (P.W.) — all in the United Kingdom.
Address reprint requests to Dr. Kennedy at Mailpoint 21, Child Health, Southampton General Hospital, Southampton SO16 6YD, United Kingdom, or at crk1{at}soton.ac.uk.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||
Related Letters:
Language Ability after Early Detection of Hearing Impairment
Grima J., Kennedy C., Watkin P., Worsfold S.
Extract |
Full Text |
PDF
N Engl J Med 2006;
355:734, Aug 17, 2006.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
