- Email: [email protected]
Auditory and cognitive abilities of children suspected of auditory processing disorder (APD)
International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600
Contents lists available at ScienceDirect
International Journal of Pediatric Otorhinolaryngology journal homepage: www.elsevier.com/locate/ijporl
Auditory and cognitive abilities of children suspected of auditory processing disorder (APD) Stuart Rosen a,*, Mazal Cohen b, Iyngaran Vanniasegaram c a
UCL Speech, Hearing and Phonetic Sciences, 2 Wakefield Street, London WC1N 1PF, England, United Kingdom Human Communication and Deafness, School of Psychological Sciences, The University of Manchester, Manchester, England, United Kingdom c Department of Audiological Medicine, St Georges Hospital, Hornchurch, Essex, England, United Kingdom b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 20 August 2009 Received in revised form 19 February 2010 Accepted 22 February 2010 Available online 26 March 2010
Objective: Auditory processing disorder (APD) is typically characterised by difficulties in ‘listening’, particularly to speech in a noisy environment, despite normal peripheral function. In school-age children, APD has attracted considerable interest because of suspicions that it may lead to learning difficulties, especially affecting language and literacy. Here, we evaluated auditory and cognitive abilities in a group of children referred for an auditory evaluation on the grounds of a suspected auditory processing disorder (susAPD), and in age-matched children who were typically developing, in order to determine the extent to which any deficits in cognitive abilities could be related to auditory deficits. Methods: A battery of auditory and cognitive tests was applied to 20 susAPD school-age children, all reported as having listening/hearing problems but performing within normal limits for standard audiometric assessments. Also tested was a group of 28 age-matched controls. The auditory tasks consisted of two simple same/different discrimination tasks, one using speech, and one nonspeech. The cognitive evaluation comprised a vocabulary test, a test of grammar and four non-verbal IQ measures. Symptoms of Attention Deficit Hyperactivity Disorder (ADHD) were assessed in the susAPD group through a standardised questionnaire. Results: A significant proportion of susAPD children appeared to display genuine auditory deficits evidenced by poor performance on at least one of the auditory tasks, although about 1/3 had no detectable deficit. Children in the susAPD group scored consistently lower than the controls on cognitive measures that were both verbal (vocabulary and grammar) and non-verbal. Strikingly, susAPD children with relatively good auditory performance did not differ in cognitive ability from susAPD children with poor auditory performance. Similarly, within-group correlations between auditory and cognitive measures were weak or non-existent. Measures of ADHD did not correlate with any aspect of auditory or cognitive performance. Conclusions: Although children suspected of having APD do show, on average, poorer performance on a number of auditory tasks, the presence or absence of an auditory deficit appears to have little impact on the development of the verbal and non-verbal skills tested here. ß 2010 Elsevier Ireland Ltd. All rights reserved.
Keywords: Auditory processing disorder (APD) CAPD Children
1. Introduction An auditory processing disorder (APD) is typically characterised by difficulties in understanding speech, often in a noisy listening environment, despite having a normal audiogram. In school-age
Abbreviations: APD, auditory processing disorder; susAPD, suspected auditory processing disorder; CCMP, Consonant Cluster Minimal Pair; TDT, Tallal Discrimination Task; BPVS, British Picture Vocabulary scale; TROG, Test for Reception of Grammar; WISC, Wechsler Intelligence Scale for Children; ADHD, Attention Deficit Hyperactivity Disorder; TOWRE, Test of Word Reading Efficiency. * Corresponding author. Tel.: +44 20 7679 4077; fax: +44 20 7679 4238. E-mail address: [email protected] (S. Rosen). 0165-5876/$ – see front matter ß 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ijporl.2010.02.021
children, APD has attracted considerable interest because of suspicions that it may lead to learning difficulties, especially affecting language and literacy, and hence to poor school performance. The diagnosis of APD remains a clinical challenge. Poor performance on APD tests does not provide sufficient evidence of an APD. The assessment should enable the clinician to delineate the cause of a listening problem and to separate auditory problems from language learning or attention deficits which may present in a similar way. Potential influences must be explored ranging from peripheral auditory dysfunction, in particular a history of otitis media with effusion in childhood associated with hearing loss [1], cross-modal sources such as attention [2], developmental factors
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600
[3] and language-learning problems [4]. A better understanding of the sequelae of APD and implications for school performance would provide a basis for selecting the appropriate remedial action that may be needed in each case. In an earlier study, we investigated auditory performance in a group of children suspected of having APD on the basis of a medical referral with concerns about difficulties in ‘listening’ [5]. A battery of auditory tests was applied to these children and a control group, in the hope of finding objective evidence of APD. As it turned out, two of the auditory tasks (one verbal and one non-verbal) statistically separated the two groups. In the current study we hoped to gain a further understanding of the characteristics of these children, by evaluating a subgroup on a number of measures of intelligence and language ability. We have also evaluated the children for signs of Attention Deficit and Hyperactivity Disorder (ADHD), because of the concern that ADHD may be frequently co-morbid with, or even indistinguishable from, APD [6,7]. A clinical history questionnaire was aimed at delineating the contribution of developmental factors, history of OME and events surrounding the birth on APD. Our goal was to establish whether the auditory processing deficits we found are associated with a more global cognitive deficit that may impact on the child’s educational achievements. 2. Methods 2.1. Subjects 48 children and adolescents, aged 6–14 years old, and attending mainstream schools, participated in the main part of this study. Twenty were normal hearing children referred to the Audiology Clinic because a teacher or parent expressed concerns about their hearing (13 males and 7 females; mean age 10.1 years 2.1). The remaining 28 were age-matched controls (15 males, 13 females; mean age 10.3 years 2.9), also attending mainstream schools in the area. The participants were all those willing to undergo further testing from a larger group (32 susAPD and 33 controls) who had undergone a detailed evaluation of their auditory function some months before [5]. A clinical history questionnaire was completed for all children in an interview with the child and his or her parent(s) or care giver. For none of the control children were concerns about hearing or listening problems expressed. No other criteria (except those relating to basic audiometry, as described in the next section) were applied in participant selection. 2.2. Auditory assessments In the earlier study, all 48 children had a basic audiometric evaluation comprising pure tone audiometry and tympanometry, to ensure that middle ear function was intact and hearing thresholds were in the normal range (mean hearing level at 0.5, 1, 2 and 4 kHz 20 dB HL). This was followed by four auditory processing tasks, chosen because they have previously been useful in the investigation of APD, and/or in distinguishing language-impaired groups from controls. Two of the tasks (simultaneous/backward masking and dichotic competing sentences) did not distinguish the controls from the susAPD group so will only be discussed in passing. The tests that distinguished the two groups were the Consonant Cluster Minimal Pairs (CCMP) and Tallal Discrimination Task (TDT), both presented by computer in a forced choice, same/different paradigm. The CCMP is a simple verbal discrimination task, similar to the popular SCAN test, consisting of 24 rhyming word pairs which differed in their initial consonant cluster (e.g., fog/frog, smack/snack). The word pairs, with equal numbers of same and different trials, were presented in a background of speech-shaped noise at a signal-to-noise ratio of 2.3 dB. The TDT is a non-verbal discrimination task which has been used extensively to test
595
auditory skills in children with language disorders [8]. The task comprised 20 short (50 ms) complex periodic tone pairs presented with a varying amount of time between the tones of each pair (0, 10, 50, 100 and 400 ms). Again, equal numbers of same and different trials were provided, with the two tones differing only in fundamental frequency (100 Hz vs. 305 Hz). This task is clearly aimed at testing a nonspeech ability, and thus has as much claim to be an APD test as any. Logistic regression was used to provide an age-corrected score for each listener on the two relevant auditory tasks. Using a model of the dependence of performance on age, estimated from the control data alone, a standardised residual was calculated for the score of each listener. These residuals are independent of age, with (approximately) a mean of 0 and standard deviation of 1. They are thus directly comparable to z scores calculated from normally distributed data. About 95% of the normal population should have residuals within a range of 1.96. A more detailed description of all the tests, the results from them, and the statistical procedures applied, is provided elsewhere [5]. 2.3. Assessments of cognitive skills, ADHD and clinical history In addition to the auditory evaluation, all children participating in this study had a psychometric evaluation of verbal and nonverbal cognitive abilities. The tests were chosen after consultation with the chief Child Clinical Psychologist at St Georges’ Hospital and carried out by a clinical psychologist. Those tests were comprised of the British Picture Vocabulary scale (BPVS), the Test for Reception of Grammar (TROG) and four non-verbal IQ measures from the WISC (picture completion, coding, block design and object assembly). The assessment of ADHD was based on the ADHDT questionnaire [9], demarcating the three components of ADHD: impulsivity, inattention and hyperactivity, plus an overall ADHD quotient. The questionnaire was completed in 17 of 20 children in the susAPD group, in an interview with the child, his/her parent/ carer and the Audiological Physician (author IV). Nine additional children from the original susAPD group, but without cognitive scores, had scores available from the ADHDT questionnaire. Given the centrality of literacy in the school curriculum, and claims regarding the possible role of an APD in the genesis of dyslexia, it would clearly be of interest to evaluate reading. Unfortunately, this was not part of our original design, but once the study was over, we were able to gather preliminary data over the telephone from eight controls and seven children from the susAPD group using the Test of Word Reading Efficiency (TOWRE, [10]). An additional three children from the larger susAPD group (without other cognitive tests) were also tested, and these values used when possible. Finally, a number of summary variables were collected from a clinical history questionnaire, documenting events surrounding the birth, history of OME and hearing and listening difficulties in an attempt to uncover any relevant factors that may contribute to the development or presentation of APD. 3. Results 3.1. Auditory function Fig. 1 displays a scatter plot of the standardised residuals of the two auditory discrimination tasks. The pattern of results is similar to that in the larger study. Performance in the susAPD group was significantly poorer for both measures (p < 0.001 and p < 0.03 for CCMP and TDT, respectively, in independent t-tests), with no correlation within groups between the two measures (p > 0.4). A high proportion of susAPD children performed quite poorly (in the
596
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600
Fig. 1. CCMP standardised residual z scores plotted against TDT z scores. Dashed lines (z = 1.64) indicate scores below which only about 5% of the normal population would be expected to score, whereas the solid lines (z = +1.64) indicate scores which only 5% of the normal population would be expected to exceed. Note the one susAPD data point obscured behind the control with the lowest CCMP score.
expected bottom 5% of the population) on one or both tasks (13/ 20 = 65%) whereas relatively few controls did (6/28 = 21%). The CCMP clearly provides the better separation between the two groups, with 9/20 (45%) susAPD listeners failing the test, but only 1/28 controls (about 4%, as would be expected). Strikingly, 1/4 of the susAPD group fail both tasks, whereas none of the controls did. The difference in performance between the two groups is even more clearly seen in the boxplots of Fig. 2, in which the distribution of scores in both auditory tasks is compared for the two groups. 3.2. Cognitive assessments As Fig. 2 shows, the cognitive abilities of the susAPD group were also worse than those of the controls, in all three main assessments (p 0.01). The susAPD listeners also had lower standard scores than the controls on all four of the non-verbal IQ subtests. For one of the tests (object assembly), the difference did not quite reach statistical significance (p < 0.07) but was robust for the other three tests (p < 0.01). Most of the difference between the two groups can be attributed to the control group performing at a substantially higher level than the population average, with scores of 109–114 on the three main assessments, nearly a full standard deviation (s.d.) higher than would be expected (p < 0.005 on all three onesample t-tests). Of the three tests, the susAPD group performed worst on the TROG, with a score of 93.5, nearly 1/2 of a s.d. below normal (p < 0.01). Vocabulary knowledge and non-verbal IQ were, however, within normal limits, at least on average (98.1 and 96.4, respectively, with p > 0.4). The correlations among the cognitive tasks revealed an interesting difference between the two groups, as shown in Table 1. Both groups show a reasonably strong correlation between the two language measures, BPVS and TROG, but only for the susAPD group did non-verbal IQ correlate with any of the language measures (TROG, in particular; however, the correlation with BPVS narrowly misses significance at p = 0.062, and deleting one case with the most extreme value of Cook’s d – a metric to identify influential outliers in a regression [11] – makes the correlation highly significant at p < 0.005). In spite of this difference in the patterns of correlation, the two groups showed much the same difference between their non-verbal IQ and language measures,
Fig. 2. Boxplots comparing the performance of the susAPD group to controls on the two auditory and three cognitive tasks. The box indicates the inter-quartile range of values obtained, with the median indicated by the solid horizontal line. The range of measurements is shown by the whiskers except for points more than 1.5 box lengths (indicated by ‘o’) or 3 box lengths (‘*’) from the upper or lower edge of the box. Performance on all 5 measures differed significantly for the two groups (p 0.03 for TDT, p 0.001 for CCMP, p 0.01 for BPVS, p 0.002 for non-verbal IQ and p 0.001 for TROG, with unequal variances assumed when required). Note that the z scores for the auditory tasks have been transformed to IQ-like scores (mean of 100 and s.d. of 15).
implying that the poorer cognitive performance by the susAPD group is not specific to language skills. Some interesting patterns were found in the reading task, even thought it was applied to relatively few children. Word and nonword reading scores were highly correlated over the entire 18 children tested (r = 0.942) so we use a simple average of the two as an overall measure of reading ability. Just as for all the cognitive tests, the controls score significantly higher than expected (by about 1 s.d. with a mean score of 116), and considerably higher than the susAPD group. In fact, the difference between the two groups is the highest of any of the cognitive measures, with the susAPD group also showing significantly lower scores (mean = 90) than the expected population mean of 100 (p 0.03). Although UK norms are not yet available for this test, initial data suggest that median scores in the UK population are somewhat higher than 100 (written communications from D.V.M. Bishop, Oxford University, June 2005 and N. Harlaar, Institute of Psychiatry, King’s College London, June 2005). Thus the susAPD group appear to be genuinely below average in reading, a fact which cannot only be attributed to better-than-average controls. As regards correlations with other cognitive tests, the small number of children tested means that even relatively large correlations do not reach statistical significance (e.g., r = 0.495 between reading and BPVS in the susAPD group). 3.3. Relationships between auditory and cognitive assessments Generally speaking, as Table 1 and Fig. 3 show, correlations between cognitive measures and listening skills were weak or nonexistent. Of 12 relevant within-group correlations (TDT/CCMP vs. non-verbal IQ/TROG/BPVS), only two Pearson correlations just reached significance (without correction for multiple comparisons) at p < 0.05 (TDT vs. non-verbal IQ in the controls, and TROG vs. TDT in the susAPD group). Neither of these correlations is very robust, as supported by the lack of significant correlations using the nonparametric Spearman’s rho. Both appear to be carried solely by one or two listeners.
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600 Table 1 Pearson correlation coefficients for the three main cognitive assessments, and the two measures of auditory function. Correlations significant at the 0.05 level in a two-tailed test are marked with * (none were significant at p 0.01). Values for the susAPD group are in bold italics in the lower left half of the table, while values for the control group are in the upper right half of the table in a normal font. nvIQ nvIQ BPVS TROG TDT CCMP
BPVS 0.087
0.425 0.505* 0.038 0.025
0.469* 0.102 0.419
TROG
TDT
0.230 0.447*
0.391* 0.206 0.097
0.455* 0.073
CCMP 0.085 0.063 0.147 0.035
0.161
For the theoretically interesting correlation in the susAPD group, excising the single listener with the highest value of Cook’s d (and the lowest TROG score) drops the variance accounted for from 20% to a non-significant 10% (p 0.18). Excising the next most extreme listener (again the highest Cook’s d and lowest TROG) drops further the variance accounted for to 2% (p > 0.5). Similarly for TDT vs. non-verbal IQ in the controls: excising the single listener with the highest value of Cook’s d (and the lowest TDT score) drops the variance accounted for from 15% to a non-significant 8% (p 0.18). In another exploration of the extent to which poor listening skills result in poor cognitive ability, we defined two subgroups within each group according to auditory performance as measured by the mean z score on the CCMP and TDT tests. In order to get a reasonable balance of numbers in the two subgroups, we used a cut-off of z = 1 in this instance. Thus listeners with mean z scores one standard deviation or more below the control mean were considered to be ‘impaired’, whereas the rest were considered ‘unimpaired’. As shown in Fig. 4, there was little difference in the cognitive ability of the two subgroups both for control and susAPD groups (p > 0.1 in independent ttests). Similarly, the difference between non-verbal IQ and language measures (not shown in the figure) did not differ for impaired and unimpaired listeners. Although it seemed unlikely that performance in the two auditory tasks that did not distinguish the susAPD group from the
Fig. 3. Scatter-plot matrix showing the relationships among the 3 cognitive and 2 main auditory assessments. Each subplot in the matrix is a simple scatter plot, whose x-axis variable is given by the column, and y-axis variable is given by the row. For example, the scatter plot in the extreme upper right depicts non-verbal IQ as a function of performance on the consonant cluster task. The scatter plot on the extreme lower left depicts the same relationship, but with x and y variables exchanged. Best-fit straight lines are shown separately for the two groups of children.
597
controls could be useful in predicting the different cognitive abilities that do distinguish the two groups, we also assessed a subgrouping based on all of the auditory tasks applied. In a parallel analysis to the one described above, we calculated the mean z score across the 4 auditory tests (with a single z-score for mean performance in the simultaneous and backward masking tasks), and classified listeners as impaired or not using a cut-off of z = 1. The results were very similar to those shown in Fig. 4, with no significant differences in any of the 3 cognitive tests between impaired and unimpaired listeners within each listener group. On the reading tasks, there was a trend for listeners with impaired auditory performance to score lower than listeners with normal auditory performance, but again the numbers in each group are too small to allow much statistical power. No convincing within-group correlations were found between reading abilities and the two auditory tasks. In fact, performance on the consonant discrimination task was negatively correlated with reading in both groups, and nearly reached significance for the control group (r = 0.63, p = 0.092). All other correlations had p > 0.2. As sparse as these data are, they do show that reading itself appears to pose problems for the susAPD group, although they provide little credence for the notion that auditory skills are an important factor in literacy.
Fig. 4. Boxplots comparing the performance, within susAPD and control groups, of impaired and unimpaired listeners on three measures of cognitive ability. None of the pairs of clustered boxplots are significantly different from one another.
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600
598
3.4. Relationships between ADHD and other assessments There is much concern about the extent to which ADHD cooccurs with, or is independent of, APD, because the lack of attention which characterises ADHD can, on its own, readily produce symptoms suggestive of APD (i.e., through poor performance on listening tasks [12]). Therefore, we investigated the relationships among the various measures of ADHD and performance on the listening and cognitive tasks. For the 26 susAPD children who had the ADHD assessment, all 3 measures (impulsivity, inattention and hyperactivity) correlated highly with one another (p < 0.001, with correlation coefficients around 0.8). The overall ADHD quotient covered a reasonable range (68–113, with a mean of 88.7 and s.d. of 12.7). However, there was no correlation between any of the ADHD measures (including the ADHD quotient) and performance on either of the two listening tasks (p > 0.5). Listeners were also split into two subgroups depending upon the mean standardised residual value of the two auditory scores mentioned above. As before, scores less than 1 were designated as ‘impaired’ (16 listeners) while those > 1 were designated ‘unimpaired’ (10 listeners). None of the ADHD measures differed between these two subgroups (p > 0.15 in independent t-tests). Finally, listeners were grouped into two approximately equal-sized sets based on their overall ADHD score but neither of the auditory measures differed between these two groups (p > 0.4). For the 17 children with the appropriate measures, correlations were also examined between ADHD scores and cognitive skills (BPVS, TROG and non-verbal IQ). Again, no measure of ADHD correlated significantly with any of the three cognitive measures (p > 0.1). Furthermore, children with relatively high scores for ADHD did not differ on any cognitive measure from children with relatively low scores for ADHD. We did not investigate relationships between reading abilities and ADHD scores as only 6 children were tested on both measures. 3.5. Analyses of clinical histories The clinical history, although relatively informal, did reveal some notable associations with the incidence of many problems varying significantly between the susAPD and control groups (summarised in Table 2). Strikingly, although there was a higher incidence of reported problems in the susAPD children with impaired auditory function than the susAPD children who were unimpaired (z 1 vs. z > 1 on the mean residual of the CCMP and TDT tests), none of these differences were statistically significant. As would be expected, the susAPD group reported a much higher incidence of auditory problems than controls (after all, a concern about hearing/listening was the cause for referral), with the most frequently reported being that of hearing the television, and speech in a noisy background (both reported by over half of the susAPD group and few controls). Neither of these features differs between our so-called ‘impaired’ and ‘unimpaired’ listeners.
There was a tendency for some association of a history of OME with susAPD and impaired hearing, but neither was statistically significant. The ability to attend or concentrate was noted as a problem for significantly more of the susAPD listeners, but again did not differ significantly for impaired and unimpaired listeners. When listeners were asked about what school subjects were particular weaknesses, 70% of the susAPD group reported literacy as a particular problem but only 17% of controls. Yet again, listeners with an objectively identified auditory impairment do not differ in this regard from those without. In a similar manner, 25% of susAPD listeners, but no controls, report some difficulty with spoken language (e.g., delayed speech, using the wrong word, mumbling, etc.). Interestingly, the difference in incidence of spoken language problems was the only feature even to approach statistical significance for groups defined by auditory task performance. 4. Discussion Over half of normal hearing children who were referred to the clinic because of hearing/listening difficulties were found to have an auditory impairment. The proportion was significantly higher than the incidence of poor listening skills found in the controls. Furthermore, the test which assessed the ability to discriminate speech-in-noise (a word discrimination task, CCMP), the problem which is often said to characterise APD, distinguished the susAPD group from the controls best. Because our chosen auditory tests were employed in a much larger battery that precluded any test taking too much time, each of the tests may be relatively weak in detecting impaired auditory performance. After all, there were only 20 trials in the nonspeech task, and 28 in the speech one, in both of which chance performance was 50%. In view of the limited test battery and the phenotypic heterogeneity of APD, it is surprising that the tests worked as well as they did in separating out the children in the susAPD group from the controls. In future work, it would certainly make sense to increase the number of items in both of these tests in order to make them more sensitive assays of ability, and to determine the extent to which they measure genuinely different processes. One of the key challenges for audiologists is to distinguish an auditory problem from a language one, as both may lead to similar results. The cognitive evaluation employed in this study comprised verbal and non-verbal tests, and our general approach was based on investigating correlations between skills primarily within as opposed to between groups. The underlying premise was that if auditory deficits are the prime cause of some cognitive deficit, for example in language, we should expect strong correlations between measures of auditory processing and language skills, not only across the entire population, but within languageimpaired groups as well [13]. Insofar as correlations only hold across groups, there are two obvious possibilities. One is that the auditory deficit is causal in some cases only, with different causes responsible for others. Alternatively, it may be that the auditory deficit is associated with the language deficit rather than being
Table 2 Incidence of problems noted from a clinical history and questionnaire, across groups defined by clinical status (controls vs. susAPD) or mean performance in the two auditory tasks (unimpaired vs. impaired). The numbers in the table indicate the percentage of children in each group who were reported to have a problem in each of the particular areas. The columns labelled ‘p’ indicate the degree to which the incidence in the two groups is statistically different, as determined by Fisher’s exact test. One, two and three asterisks indicates p < 0.05, p < 0.005 and p < 0.001, respectively. A blank indicates p > 0.1.
Any auditory problem Speech-in-noise Listening to the television Middle ear disease Attention/concentration Literacy Spoken language
Controls (of 23)
susAPD (of 20)
p
Unimpaired (of 27)
Impaired (of 16)
p
17 9 9 9 0 17 0
90 55 60 30 60 70 25
*** ** ***
41 26 26 11 22 41 4
69 37 44 31 37 44 25
0.06
*** *** *
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600
causal (see [14] for a similar argument regarding dyslexia and sensori-motor impairments). The analysis showed a complete dissociation between auditory and cognitive tests: performance on the word discrimination task did not correlate with either of two explicit measures of language ability (nor with reading), so it does not appear to be likely that impaired speech perceptual performance led to impaired language. (But see [15] for a claim that such relationships should be strong.) The high standardised scores of the control group in the current study raise reservations about the degree to which the children in it really are appropriate controls. Insofar as auditory skills appear to be unrelated to measures of non-verbal IQ and language function, the control group is probably a reasonable comparison. But when it comes to comparisons on standard cognitive measures, the controls are clearly inappropriate. Consider, for example, performance on the BPVS measure of vocabulary. The susAPD group mean fell well within normal bounds (at around 98) yet is significantly different from the controls with a mean just above 109. Such problems with control groups are far from rare, and are often attributed to the fact that controls cannot be randomly chosen from the population at large. It appears likely that the simple requirement of obtaining a signed permission form from a parent may be sufficient to secure an unrepresentative population. It is therefore impossible to know how to interpret the relatively low cognitive scores in the susAPD group except through comparisons within the susAPD group itself (see [16] and discussion below). Interestingly, the susAPD group appear to be genuinely below average in reading, which is consistent with questionnaire responses identifying literacy as a particular problem. Alongside literacy, the clinical questionnaires identified additional common factors and/or symptoms which were found in the susAPD group. These include difficulty understanding speech in noisy backgrounds, difficulty following the TV, history of OME, difficulties with attention and with spoken language. The exact nature of these associations was not elucidated here due to small numbers and to the informal nature of the clinical histories. Although no correlation was found between these factors and impaired listening or cognitive function, they contribute to the emergence of a clinical profile of a child referred to the clinic because of suspected APD. There was no correlation between auditory scores and the various ADHD measures obtained from the ADHD questionnaire, with the implication that poor performance on the auditory tests used in this study does not simply reflect inattention and/or hyperactivity, at least as measured by the particular questionnaire used. It is important to highlight that some effort was made to minimise the effects of inattention and/or hyperactivity on the performance on the auditory tasks by allowing participants frequent breaks. Similar conclusions to ours (although not concerning the extensive cognitive measures we assessed) can be extracted from a recent retrospective study from case notes of children referred for concerns about hearing [16]. In this group of children, 32 were diagnosed with APD (on the basis of a number of fairly typical clinically available tests like the SCAN), whereas 57 were considered not to have APD. However, no aetiological factor (e.g., history of otitis media) nor incidence of reported symptoms distinguished the two groups, just as in our within-group susAPD comparison. In summary, it is clear that the children in our study referred with a suspicion of APD, appear, at least on the group level, to have been referred with reason. Their listening abilities do seem to be impaired, on average, and it is natural enough to suppose this would have an impact on academic performance were the deficit severe enough. On the other hand, we found no relationship between the observed degree of auditory impairment and the
599
measures of cognitive skill in the susAPD group, and a similar independence of literacy and speech perceptual skills has been convincingly demonstrated in a large study of children entering school [17]. It may well be desirable to remediate an auditory deficit because, for example, classroom behaviour might improve, but without the expectation that improved auditory performance will necessarily result in improved cognitive abilities and/or academic performance [18]. However, impaired listening may not be the only reason for referral. Lower then average cognitive scores in the susAPD group may be leading to impaired academic performance, which in itself, with or without hard evidence of a listening problem, may cause concern. It is interesting to note that, even though the controls may be a poor comparison sample, the TROG scores of the susAPD group are significantly lower than the expected population mean, and there are indications that their literacy skills are relatively poor too. Furthermore, 4 of the susAPD children had non-verbal IQ scores at 76 or below, which puts them in the bottom 5% of the population (no control listener scored below 80). Referrals in the susAPD group may thus result from a variety of reasons, only some truly auditory. For the clinician, APD remains something of a minefield. One thing is quite clear. Children referred for suspected APD need to have a reasonably thorough psychometric work-up. Of the four susAPD children who had low language scores (<85 on the TROG), only one had seen a speech and language therapist, some 9 years before at age two, which ended in a diagnosis of ‘normal development’. One of the others had seen an educational psychologist recently. Even if it is not clear what remedial course children with APD should be directed to, a referral to professionals with more established remedial plans certainly seems in order when a problem is clearly identified through whatever route. Conflict of interest statement None declared. Acknowledgements We are very grateful to: Tatiyana Tairi and Saras Swaminathan who undertook the cognitive assessments; the Head Teachers and children of Harold Wood Primary School, King’s Wood School and Woodlands Preparatory School in Essex for their willingness to participate in the study; Andrew Faulkner for many helpful comments on the manuscript; Dorothy Bishop and Nicole Harlaar for providing unpublished information about the TOWRE. This research was partially funded by Barking and Dagenham Primary Care Trust. References [1] A. Zumach, E. Gerrits, M.N. Chenault, L.J.C. Anteunis, Otitis Media and speech-innoise recognition in school-aged children, Audiol. Neurootol. 14 (2009) 121–129. [2] E. Sussman, M. Steinschneider, Attention effects on auditory scene analysis in children, Neuropsychologia 47 (2009) 771–785. [3] R. Geva, R. Eshel, Y. Leitner, A. Fattal-Valevski, S. Harel, Verbal short-term memory span in children: long-term modality dependent effects of intrauterine growth restriction, J. Child Psychol. Psychiatry 49 (2008) 1321–1330. [4] L.R. Shapiro, J. Hurry, J. Masterson, T.N. Wydell, E. Doctor, Classroom implications of recent research into literacy development: from predictors to assessment, Dyslexia 15 (2009) 1–22. [5] I. Vanniasegaram, M. Cohen, S. Rosen, Evaluation of selected auditory tests in school-aged children suspected of auditory processing disorder (APD), Ear Hear. 25 (2004) 586–597. [6] G.D. Chermak, E. Tucker, J.A. Seikel, Behavioral characteristics of auditory processing disorder and attention-deficit hyperactivity disorder: predominantly inattentive type, J. Am. Acad. Audiol. 13 (2002) 332–338. [7] K.L. Purvis, R. Tannock, Phonological processing, not inhibitory control, differentiates ADHD and reading disability, J. Am. Acad. Child Adolesc. Psychiatry 39 (2000) 485–494.
600
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600
[8] P. Tallal, M. Piercy, Defects of non-verbal auditory perception in children with developmental aphasia, Nature 241 (1973) 468–469. [9] J.E. Gilliam, Attention-deficit/Hyperactivity Disorder Test: A Method for Identifying Individuals with ADHD, Pro-ed, Austin, TX, 1995. [10] J.K. Torgesen, R. Wagner, C. Rashotte, Test of Word Reading Efficiency (TOWRE), Psychological Corporation, New York, 1999. [11] R.D. Cook, S. Weisberg, Applied Regression Including Computing and Graphics, Wiley, New York, 1999. [12] P.A. Sutcliffe, D.V.M. Bishop, S. Houghton, M. Taylor, Effect of attentional state on frequency discrimination: a comparison of children with ADHD on and off medication, J. Speech Lang. Hear. Res. 49 (2006) 1072– 1084. [13] S. Rosen, Auditory processing in dyslexia and specific language impairment: is there a deficit? What is its nature? Does it explain anything?, J. Phon. 31 (2003) 509–527.
[14] F. Ramus, Developmental dyslexia: specific phonological deficit or general sensorimotor dysfunction? Current Opinion in Neurobiology. 13 (2003) 212–218. [15] J.C. Ziegler, C. Pech-Georgel, F. George, F.X. Alario, C. Lorenzi, Deficits in speech perception predict language learning impairment, Proc. Natl. Acad. Sci. U.S.A. 102 (2005) 14110–14115. [16] P. Dawes, D.V.M. Bishop, T. Sirimanna, D.E. Bamiou, Profile and aetiology of children diagnosed with auditory processing disorder (APD), Int. J. Pediatr. Otorhinolaryngol. 72 (2008) 483–489. [17] C.S. Watson, G.R. Kidd, D.G. Horner, P.J. Connell, A. Lowther, D.A. Eddins, et al., Sensory, cognitive, and linguistic factors in the early academic performance of elementary school children: the Benton-IU project, J. Learn. Disabil. 36 (2003) 165–197. [18] G.M. McArthur, D. Ellis, C.M. Atkinson, M. Coltheart, Auditory processing deficits in children with reading and language impairments: can they (and should they) be treated? Cognition 107 (2008) 946–977.
Contents lists available at ScienceDirect
International Journal of Pediatric Otorhinolaryngology journal homepage: www.elsevier.com/locate/ijporl
Auditory and cognitive abilities of children suspected of auditory processing disorder (APD) Stuart Rosen a,*, Mazal Cohen b, Iyngaran Vanniasegaram c a
UCL Speech, Hearing and Phonetic Sciences, 2 Wakefield Street, London WC1N 1PF, England, United Kingdom Human Communication and Deafness, School of Psychological Sciences, The University of Manchester, Manchester, England, United Kingdom c Department of Audiological Medicine, St Georges Hospital, Hornchurch, Essex, England, United Kingdom b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 20 August 2009 Received in revised form 19 February 2010 Accepted 22 February 2010 Available online 26 March 2010
Objective: Auditory processing disorder (APD) is typically characterised by difficulties in ‘listening’, particularly to speech in a noisy environment, despite normal peripheral function. In school-age children, APD has attracted considerable interest because of suspicions that it may lead to learning difficulties, especially affecting language and literacy. Here, we evaluated auditory and cognitive abilities in a group of children referred for an auditory evaluation on the grounds of a suspected auditory processing disorder (susAPD), and in age-matched children who were typically developing, in order to determine the extent to which any deficits in cognitive abilities could be related to auditory deficits. Methods: A battery of auditory and cognitive tests was applied to 20 susAPD school-age children, all reported as having listening/hearing problems but performing within normal limits for standard audiometric assessments. Also tested was a group of 28 age-matched controls. The auditory tasks consisted of two simple same/different discrimination tasks, one using speech, and one nonspeech. The cognitive evaluation comprised a vocabulary test, a test of grammar and four non-verbal IQ measures. Symptoms of Attention Deficit Hyperactivity Disorder (ADHD) were assessed in the susAPD group through a standardised questionnaire. Results: A significant proportion of susAPD children appeared to display genuine auditory deficits evidenced by poor performance on at least one of the auditory tasks, although about 1/3 had no detectable deficit. Children in the susAPD group scored consistently lower than the controls on cognitive measures that were both verbal (vocabulary and grammar) and non-verbal. Strikingly, susAPD children with relatively good auditory performance did not differ in cognitive ability from susAPD children with poor auditory performance. Similarly, within-group correlations between auditory and cognitive measures were weak or non-existent. Measures of ADHD did not correlate with any aspect of auditory or cognitive performance. Conclusions: Although children suspected of having APD do show, on average, poorer performance on a number of auditory tasks, the presence or absence of an auditory deficit appears to have little impact on the development of the verbal and non-verbal skills tested here. ß 2010 Elsevier Ireland Ltd. All rights reserved.
Keywords: Auditory processing disorder (APD) CAPD Children
1. Introduction An auditory processing disorder (APD) is typically characterised by difficulties in understanding speech, often in a noisy listening environment, despite having a normal audiogram. In school-age
Abbreviations: APD, auditory processing disorder; susAPD, suspected auditory processing disorder; CCMP, Consonant Cluster Minimal Pair; TDT, Tallal Discrimination Task; BPVS, British Picture Vocabulary scale; TROG, Test for Reception of Grammar; WISC, Wechsler Intelligence Scale for Children; ADHD, Attention Deficit Hyperactivity Disorder; TOWRE, Test of Word Reading Efficiency. * Corresponding author. Tel.: +44 20 7679 4077; fax: +44 20 7679 4238. E-mail address: [email protected] (S. Rosen). 0165-5876/$ – see front matter ß 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ijporl.2010.02.021
children, APD has attracted considerable interest because of suspicions that it may lead to learning difficulties, especially affecting language and literacy, and hence to poor school performance. The diagnosis of APD remains a clinical challenge. Poor performance on APD tests does not provide sufficient evidence of an APD. The assessment should enable the clinician to delineate the cause of a listening problem and to separate auditory problems from language learning or attention deficits which may present in a similar way. Potential influences must be explored ranging from peripheral auditory dysfunction, in particular a history of otitis media with effusion in childhood associated with hearing loss [1], cross-modal sources such as attention [2], developmental factors
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600
[3] and language-learning problems [4]. A better understanding of the sequelae of APD and implications for school performance would provide a basis for selecting the appropriate remedial action that may be needed in each case. In an earlier study, we investigated auditory performance in a group of children suspected of having APD on the basis of a medical referral with concerns about difficulties in ‘listening’ [5]. A battery of auditory tests was applied to these children and a control group, in the hope of finding objective evidence of APD. As it turned out, two of the auditory tasks (one verbal and one non-verbal) statistically separated the two groups. In the current study we hoped to gain a further understanding of the characteristics of these children, by evaluating a subgroup on a number of measures of intelligence and language ability. We have also evaluated the children for signs of Attention Deficit and Hyperactivity Disorder (ADHD), because of the concern that ADHD may be frequently co-morbid with, or even indistinguishable from, APD [6,7]. A clinical history questionnaire was aimed at delineating the contribution of developmental factors, history of OME and events surrounding the birth on APD. Our goal was to establish whether the auditory processing deficits we found are associated with a more global cognitive deficit that may impact on the child’s educational achievements. 2. Methods 2.1. Subjects 48 children and adolescents, aged 6–14 years old, and attending mainstream schools, participated in the main part of this study. Twenty were normal hearing children referred to the Audiology Clinic because a teacher or parent expressed concerns about their hearing (13 males and 7 females; mean age 10.1 years 2.1). The remaining 28 were age-matched controls (15 males, 13 females; mean age 10.3 years 2.9), also attending mainstream schools in the area. The participants were all those willing to undergo further testing from a larger group (32 susAPD and 33 controls) who had undergone a detailed evaluation of their auditory function some months before [5]. A clinical history questionnaire was completed for all children in an interview with the child and his or her parent(s) or care giver. For none of the control children were concerns about hearing or listening problems expressed. No other criteria (except those relating to basic audiometry, as described in the next section) were applied in participant selection. 2.2. Auditory assessments In the earlier study, all 48 children had a basic audiometric evaluation comprising pure tone audiometry and tympanometry, to ensure that middle ear function was intact and hearing thresholds were in the normal range (mean hearing level at 0.5, 1, 2 and 4 kHz 20 dB HL). This was followed by four auditory processing tasks, chosen because they have previously been useful in the investigation of APD, and/or in distinguishing language-impaired groups from controls. Two of the tasks (simultaneous/backward masking and dichotic competing sentences) did not distinguish the controls from the susAPD group so will only be discussed in passing. The tests that distinguished the two groups were the Consonant Cluster Minimal Pairs (CCMP) and Tallal Discrimination Task (TDT), both presented by computer in a forced choice, same/different paradigm. The CCMP is a simple verbal discrimination task, similar to the popular SCAN test, consisting of 24 rhyming word pairs which differed in their initial consonant cluster (e.g., fog/frog, smack/snack). The word pairs, with equal numbers of same and different trials, were presented in a background of speech-shaped noise at a signal-to-noise ratio of 2.3 dB. The TDT is a non-verbal discrimination task which has been used extensively to test
595
auditory skills in children with language disorders [8]. The task comprised 20 short (50 ms) complex periodic tone pairs presented with a varying amount of time between the tones of each pair (0, 10, 50, 100 and 400 ms). Again, equal numbers of same and different trials were provided, with the two tones differing only in fundamental frequency (100 Hz vs. 305 Hz). This task is clearly aimed at testing a nonspeech ability, and thus has as much claim to be an APD test as any. Logistic regression was used to provide an age-corrected score for each listener on the two relevant auditory tasks. Using a model of the dependence of performance on age, estimated from the control data alone, a standardised residual was calculated for the score of each listener. These residuals are independent of age, with (approximately) a mean of 0 and standard deviation of 1. They are thus directly comparable to z scores calculated from normally distributed data. About 95% of the normal population should have residuals within a range of 1.96. A more detailed description of all the tests, the results from them, and the statistical procedures applied, is provided elsewhere [5]. 2.3. Assessments of cognitive skills, ADHD and clinical history In addition to the auditory evaluation, all children participating in this study had a psychometric evaluation of verbal and nonverbal cognitive abilities. The tests were chosen after consultation with the chief Child Clinical Psychologist at St Georges’ Hospital and carried out by a clinical psychologist. Those tests were comprised of the British Picture Vocabulary scale (BPVS), the Test for Reception of Grammar (TROG) and four non-verbal IQ measures from the WISC (picture completion, coding, block design and object assembly). The assessment of ADHD was based on the ADHDT questionnaire [9], demarcating the three components of ADHD: impulsivity, inattention and hyperactivity, plus an overall ADHD quotient. The questionnaire was completed in 17 of 20 children in the susAPD group, in an interview with the child, his/her parent/ carer and the Audiological Physician (author IV). Nine additional children from the original susAPD group, but without cognitive scores, had scores available from the ADHDT questionnaire. Given the centrality of literacy in the school curriculum, and claims regarding the possible role of an APD in the genesis of dyslexia, it would clearly be of interest to evaluate reading. Unfortunately, this was not part of our original design, but once the study was over, we were able to gather preliminary data over the telephone from eight controls and seven children from the susAPD group using the Test of Word Reading Efficiency (TOWRE, [10]). An additional three children from the larger susAPD group (without other cognitive tests) were also tested, and these values used when possible. Finally, a number of summary variables were collected from a clinical history questionnaire, documenting events surrounding the birth, history of OME and hearing and listening difficulties in an attempt to uncover any relevant factors that may contribute to the development or presentation of APD. 3. Results 3.1. Auditory function Fig. 1 displays a scatter plot of the standardised residuals of the two auditory discrimination tasks. The pattern of results is similar to that in the larger study. Performance in the susAPD group was significantly poorer for both measures (p < 0.001 and p < 0.03 for CCMP and TDT, respectively, in independent t-tests), with no correlation within groups between the two measures (p > 0.4). A high proportion of susAPD children performed quite poorly (in the
596
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600
Fig. 1. CCMP standardised residual z scores plotted against TDT z scores. Dashed lines (z = 1.64) indicate scores below which only about 5% of the normal population would be expected to score, whereas the solid lines (z = +1.64) indicate scores which only 5% of the normal population would be expected to exceed. Note the one susAPD data point obscured behind the control with the lowest CCMP score.
expected bottom 5% of the population) on one or both tasks (13/ 20 = 65%) whereas relatively few controls did (6/28 = 21%). The CCMP clearly provides the better separation between the two groups, with 9/20 (45%) susAPD listeners failing the test, but only 1/28 controls (about 4%, as would be expected). Strikingly, 1/4 of the susAPD group fail both tasks, whereas none of the controls did. The difference in performance between the two groups is even more clearly seen in the boxplots of Fig. 2, in which the distribution of scores in both auditory tasks is compared for the two groups. 3.2. Cognitive assessments As Fig. 2 shows, the cognitive abilities of the susAPD group were also worse than those of the controls, in all three main assessments (p 0.01). The susAPD listeners also had lower standard scores than the controls on all four of the non-verbal IQ subtests. For one of the tests (object assembly), the difference did not quite reach statistical significance (p < 0.07) but was robust for the other three tests (p < 0.01). Most of the difference between the two groups can be attributed to the control group performing at a substantially higher level than the population average, with scores of 109–114 on the three main assessments, nearly a full standard deviation (s.d.) higher than would be expected (p < 0.005 on all three onesample t-tests). Of the three tests, the susAPD group performed worst on the TROG, with a score of 93.5, nearly 1/2 of a s.d. below normal (p < 0.01). Vocabulary knowledge and non-verbal IQ were, however, within normal limits, at least on average (98.1 and 96.4, respectively, with p > 0.4). The correlations among the cognitive tasks revealed an interesting difference between the two groups, as shown in Table 1. Both groups show a reasonably strong correlation between the two language measures, BPVS and TROG, but only for the susAPD group did non-verbal IQ correlate with any of the language measures (TROG, in particular; however, the correlation with BPVS narrowly misses significance at p = 0.062, and deleting one case with the most extreme value of Cook’s d – a metric to identify influential outliers in a regression [11] – makes the correlation highly significant at p < 0.005). In spite of this difference in the patterns of correlation, the two groups showed much the same difference between their non-verbal IQ and language measures,
Fig. 2. Boxplots comparing the performance of the susAPD group to controls on the two auditory and three cognitive tasks. The box indicates the inter-quartile range of values obtained, with the median indicated by the solid horizontal line. The range of measurements is shown by the whiskers except for points more than 1.5 box lengths (indicated by ‘o’) or 3 box lengths (‘*’) from the upper or lower edge of the box. Performance on all 5 measures differed significantly for the two groups (p 0.03 for TDT, p 0.001 for CCMP, p 0.01 for BPVS, p 0.002 for non-verbal IQ and p 0.001 for TROG, with unequal variances assumed when required). Note that the z scores for the auditory tasks have been transformed to IQ-like scores (mean of 100 and s.d. of 15).
implying that the poorer cognitive performance by the susAPD group is not specific to language skills. Some interesting patterns were found in the reading task, even thought it was applied to relatively few children. Word and nonword reading scores were highly correlated over the entire 18 children tested (r = 0.942) so we use a simple average of the two as an overall measure of reading ability. Just as for all the cognitive tests, the controls score significantly higher than expected (by about 1 s.d. with a mean score of 116), and considerably higher than the susAPD group. In fact, the difference between the two groups is the highest of any of the cognitive measures, with the susAPD group also showing significantly lower scores (mean = 90) than the expected population mean of 100 (p 0.03). Although UK norms are not yet available for this test, initial data suggest that median scores in the UK population are somewhat higher than 100 (written communications from D.V.M. Bishop, Oxford University, June 2005 and N. Harlaar, Institute of Psychiatry, King’s College London, June 2005). Thus the susAPD group appear to be genuinely below average in reading, a fact which cannot only be attributed to better-than-average controls. As regards correlations with other cognitive tests, the small number of children tested means that even relatively large correlations do not reach statistical significance (e.g., r = 0.495 between reading and BPVS in the susAPD group). 3.3. Relationships between auditory and cognitive assessments Generally speaking, as Table 1 and Fig. 3 show, correlations between cognitive measures and listening skills were weak or nonexistent. Of 12 relevant within-group correlations (TDT/CCMP vs. non-verbal IQ/TROG/BPVS), only two Pearson correlations just reached significance (without correction for multiple comparisons) at p < 0.05 (TDT vs. non-verbal IQ in the controls, and TROG vs. TDT in the susAPD group). Neither of these correlations is very robust, as supported by the lack of significant correlations using the nonparametric Spearman’s rho. Both appear to be carried solely by one or two listeners.
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600 Table 1 Pearson correlation coefficients for the three main cognitive assessments, and the two measures of auditory function. Correlations significant at the 0.05 level in a two-tailed test are marked with * (none were significant at p 0.01). Values for the susAPD group are in bold italics in the lower left half of the table, while values for the control group are in the upper right half of the table in a normal font. nvIQ nvIQ BPVS TROG TDT CCMP
BPVS 0.087
0.425 0.505* 0.038 0.025
0.469* 0.102 0.419
TROG
TDT
0.230 0.447*
0.391* 0.206 0.097
0.455* 0.073
CCMP 0.085 0.063 0.147 0.035
0.161
For the theoretically interesting correlation in the susAPD group, excising the single listener with the highest value of Cook’s d (and the lowest TROG score) drops the variance accounted for from 20% to a non-significant 10% (p 0.18). Excising the next most extreme listener (again the highest Cook’s d and lowest TROG) drops further the variance accounted for to 2% (p > 0.5). Similarly for TDT vs. non-verbal IQ in the controls: excising the single listener with the highest value of Cook’s d (and the lowest TDT score) drops the variance accounted for from 15% to a non-significant 8% (p 0.18). In another exploration of the extent to which poor listening skills result in poor cognitive ability, we defined two subgroups within each group according to auditory performance as measured by the mean z score on the CCMP and TDT tests. In order to get a reasonable balance of numbers in the two subgroups, we used a cut-off of z = 1 in this instance. Thus listeners with mean z scores one standard deviation or more below the control mean were considered to be ‘impaired’, whereas the rest were considered ‘unimpaired’. As shown in Fig. 4, there was little difference in the cognitive ability of the two subgroups both for control and susAPD groups (p > 0.1 in independent ttests). Similarly, the difference between non-verbal IQ and language measures (not shown in the figure) did not differ for impaired and unimpaired listeners. Although it seemed unlikely that performance in the two auditory tasks that did not distinguish the susAPD group from the
Fig. 3. Scatter-plot matrix showing the relationships among the 3 cognitive and 2 main auditory assessments. Each subplot in the matrix is a simple scatter plot, whose x-axis variable is given by the column, and y-axis variable is given by the row. For example, the scatter plot in the extreme upper right depicts non-verbal IQ as a function of performance on the consonant cluster task. The scatter plot on the extreme lower left depicts the same relationship, but with x and y variables exchanged. Best-fit straight lines are shown separately for the two groups of children.
597
controls could be useful in predicting the different cognitive abilities that do distinguish the two groups, we also assessed a subgrouping based on all of the auditory tasks applied. In a parallel analysis to the one described above, we calculated the mean z score across the 4 auditory tests (with a single z-score for mean performance in the simultaneous and backward masking tasks), and classified listeners as impaired or not using a cut-off of z = 1. The results were very similar to those shown in Fig. 4, with no significant differences in any of the 3 cognitive tests between impaired and unimpaired listeners within each listener group. On the reading tasks, there was a trend for listeners with impaired auditory performance to score lower than listeners with normal auditory performance, but again the numbers in each group are too small to allow much statistical power. No convincing within-group correlations were found between reading abilities and the two auditory tasks. In fact, performance on the consonant discrimination task was negatively correlated with reading in both groups, and nearly reached significance for the control group (r = 0.63, p = 0.092). All other correlations had p > 0.2. As sparse as these data are, they do show that reading itself appears to pose problems for the susAPD group, although they provide little credence for the notion that auditory skills are an important factor in literacy.
Fig. 4. Boxplots comparing the performance, within susAPD and control groups, of impaired and unimpaired listeners on three measures of cognitive ability. None of the pairs of clustered boxplots are significantly different from one another.
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600
598
3.4. Relationships between ADHD and other assessments There is much concern about the extent to which ADHD cooccurs with, or is independent of, APD, because the lack of attention which characterises ADHD can, on its own, readily produce symptoms suggestive of APD (i.e., through poor performance on listening tasks [12]). Therefore, we investigated the relationships among the various measures of ADHD and performance on the listening and cognitive tasks. For the 26 susAPD children who had the ADHD assessment, all 3 measures (impulsivity, inattention and hyperactivity) correlated highly with one another (p < 0.001, with correlation coefficients around 0.8). The overall ADHD quotient covered a reasonable range (68–113, with a mean of 88.7 and s.d. of 12.7). However, there was no correlation between any of the ADHD measures (including the ADHD quotient) and performance on either of the two listening tasks (p > 0.5). Listeners were also split into two subgroups depending upon the mean standardised residual value of the two auditory scores mentioned above. As before, scores less than 1 were designated as ‘impaired’ (16 listeners) while those > 1 were designated ‘unimpaired’ (10 listeners). None of the ADHD measures differed between these two subgroups (p > 0.15 in independent t-tests). Finally, listeners were grouped into two approximately equal-sized sets based on their overall ADHD score but neither of the auditory measures differed between these two groups (p > 0.4). For the 17 children with the appropriate measures, correlations were also examined between ADHD scores and cognitive skills (BPVS, TROG and non-verbal IQ). Again, no measure of ADHD correlated significantly with any of the three cognitive measures (p > 0.1). Furthermore, children with relatively high scores for ADHD did not differ on any cognitive measure from children with relatively low scores for ADHD. We did not investigate relationships between reading abilities and ADHD scores as only 6 children were tested on both measures. 3.5. Analyses of clinical histories The clinical history, although relatively informal, did reveal some notable associations with the incidence of many problems varying significantly between the susAPD and control groups (summarised in Table 2). Strikingly, although there was a higher incidence of reported problems in the susAPD children with impaired auditory function than the susAPD children who were unimpaired (z 1 vs. z > 1 on the mean residual of the CCMP and TDT tests), none of these differences were statistically significant. As would be expected, the susAPD group reported a much higher incidence of auditory problems than controls (after all, a concern about hearing/listening was the cause for referral), with the most frequently reported being that of hearing the television, and speech in a noisy background (both reported by over half of the susAPD group and few controls). Neither of these features differs between our so-called ‘impaired’ and ‘unimpaired’ listeners.
There was a tendency for some association of a history of OME with susAPD and impaired hearing, but neither was statistically significant. The ability to attend or concentrate was noted as a problem for significantly more of the susAPD listeners, but again did not differ significantly for impaired and unimpaired listeners. When listeners were asked about what school subjects were particular weaknesses, 70% of the susAPD group reported literacy as a particular problem but only 17% of controls. Yet again, listeners with an objectively identified auditory impairment do not differ in this regard from those without. In a similar manner, 25% of susAPD listeners, but no controls, report some difficulty with spoken language (e.g., delayed speech, using the wrong word, mumbling, etc.). Interestingly, the difference in incidence of spoken language problems was the only feature even to approach statistical significance for groups defined by auditory task performance. 4. Discussion Over half of normal hearing children who were referred to the clinic because of hearing/listening difficulties were found to have an auditory impairment. The proportion was significantly higher than the incidence of poor listening skills found in the controls. Furthermore, the test which assessed the ability to discriminate speech-in-noise (a word discrimination task, CCMP), the problem which is often said to characterise APD, distinguished the susAPD group from the controls best. Because our chosen auditory tests were employed in a much larger battery that precluded any test taking too much time, each of the tests may be relatively weak in detecting impaired auditory performance. After all, there were only 20 trials in the nonspeech task, and 28 in the speech one, in both of which chance performance was 50%. In view of the limited test battery and the phenotypic heterogeneity of APD, it is surprising that the tests worked as well as they did in separating out the children in the susAPD group from the controls. In future work, it would certainly make sense to increase the number of items in both of these tests in order to make them more sensitive assays of ability, and to determine the extent to which they measure genuinely different processes. One of the key challenges for audiologists is to distinguish an auditory problem from a language one, as both may lead to similar results. The cognitive evaluation employed in this study comprised verbal and non-verbal tests, and our general approach was based on investigating correlations between skills primarily within as opposed to between groups. The underlying premise was that if auditory deficits are the prime cause of some cognitive deficit, for example in language, we should expect strong correlations between measures of auditory processing and language skills, not only across the entire population, but within languageimpaired groups as well [13]. Insofar as correlations only hold across groups, there are two obvious possibilities. One is that the auditory deficit is causal in some cases only, with different causes responsible for others. Alternatively, it may be that the auditory deficit is associated with the language deficit rather than being
Table 2 Incidence of problems noted from a clinical history and questionnaire, across groups defined by clinical status (controls vs. susAPD) or mean performance in the two auditory tasks (unimpaired vs. impaired). The numbers in the table indicate the percentage of children in each group who were reported to have a problem in each of the particular areas. The columns labelled ‘p’ indicate the degree to which the incidence in the two groups is statistically different, as determined by Fisher’s exact test. One, two and three asterisks indicates p < 0.05, p < 0.005 and p < 0.001, respectively. A blank indicates p > 0.1.
Any auditory problem Speech-in-noise Listening to the television Middle ear disease Attention/concentration Literacy Spoken language
Controls (of 23)
susAPD (of 20)
p
Unimpaired (of 27)
Impaired (of 16)
p
17 9 9 9 0 17 0
90 55 60 30 60 70 25
*** ** ***
41 26 26 11 22 41 4
69 37 44 31 37 44 25
0.06
*** *** *
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600
causal (see [14] for a similar argument regarding dyslexia and sensori-motor impairments). The analysis showed a complete dissociation between auditory and cognitive tests: performance on the word discrimination task did not correlate with either of two explicit measures of language ability (nor with reading), so it does not appear to be likely that impaired speech perceptual performance led to impaired language. (But see [15] for a claim that such relationships should be strong.) The high standardised scores of the control group in the current study raise reservations about the degree to which the children in it really are appropriate controls. Insofar as auditory skills appear to be unrelated to measures of non-verbal IQ and language function, the control group is probably a reasonable comparison. But when it comes to comparisons on standard cognitive measures, the controls are clearly inappropriate. Consider, for example, performance on the BPVS measure of vocabulary. The susAPD group mean fell well within normal bounds (at around 98) yet is significantly different from the controls with a mean just above 109. Such problems with control groups are far from rare, and are often attributed to the fact that controls cannot be randomly chosen from the population at large. It appears likely that the simple requirement of obtaining a signed permission form from a parent may be sufficient to secure an unrepresentative population. It is therefore impossible to know how to interpret the relatively low cognitive scores in the susAPD group except through comparisons within the susAPD group itself (see [16] and discussion below). Interestingly, the susAPD group appear to be genuinely below average in reading, which is consistent with questionnaire responses identifying literacy as a particular problem. Alongside literacy, the clinical questionnaires identified additional common factors and/or symptoms which were found in the susAPD group. These include difficulty understanding speech in noisy backgrounds, difficulty following the TV, history of OME, difficulties with attention and with spoken language. The exact nature of these associations was not elucidated here due to small numbers and to the informal nature of the clinical histories. Although no correlation was found between these factors and impaired listening or cognitive function, they contribute to the emergence of a clinical profile of a child referred to the clinic because of suspected APD. There was no correlation between auditory scores and the various ADHD measures obtained from the ADHD questionnaire, with the implication that poor performance on the auditory tests used in this study does not simply reflect inattention and/or hyperactivity, at least as measured by the particular questionnaire used. It is important to highlight that some effort was made to minimise the effects of inattention and/or hyperactivity on the performance on the auditory tasks by allowing participants frequent breaks. Similar conclusions to ours (although not concerning the extensive cognitive measures we assessed) can be extracted from a recent retrospective study from case notes of children referred for concerns about hearing [16]. In this group of children, 32 were diagnosed with APD (on the basis of a number of fairly typical clinically available tests like the SCAN), whereas 57 were considered not to have APD. However, no aetiological factor (e.g., history of otitis media) nor incidence of reported symptoms distinguished the two groups, just as in our within-group susAPD comparison. In summary, it is clear that the children in our study referred with a suspicion of APD, appear, at least on the group level, to have been referred with reason. Their listening abilities do seem to be impaired, on average, and it is natural enough to suppose this would have an impact on academic performance were the deficit severe enough. On the other hand, we found no relationship between the observed degree of auditory impairment and the
599
measures of cognitive skill in the susAPD group, and a similar independence of literacy and speech perceptual skills has been convincingly demonstrated in a large study of children entering school [17]. It may well be desirable to remediate an auditory deficit because, for example, classroom behaviour might improve, but without the expectation that improved auditory performance will necessarily result in improved cognitive abilities and/or academic performance [18]. However, impaired listening may not be the only reason for referral. Lower then average cognitive scores in the susAPD group may be leading to impaired academic performance, which in itself, with or without hard evidence of a listening problem, may cause concern. It is interesting to note that, even though the controls may be a poor comparison sample, the TROG scores of the susAPD group are significantly lower than the expected population mean, and there are indications that their literacy skills are relatively poor too. Furthermore, 4 of the susAPD children had non-verbal IQ scores at 76 or below, which puts them in the bottom 5% of the population (no control listener scored below 80). Referrals in the susAPD group may thus result from a variety of reasons, only some truly auditory. For the clinician, APD remains something of a minefield. One thing is quite clear. Children referred for suspected APD need to have a reasonably thorough psychometric work-up. Of the four susAPD children who had low language scores (<85 on the TROG), only one had seen a speech and language therapist, some 9 years before at age two, which ended in a diagnosis of ‘normal development’. One of the others had seen an educational psychologist recently. Even if it is not clear what remedial course children with APD should be directed to, a referral to professionals with more established remedial plans certainly seems in order when a problem is clearly identified through whatever route. Conflict of interest statement None declared. Acknowledgements We are very grateful to: Tatiyana Tairi and Saras Swaminathan who undertook the cognitive assessments; the Head Teachers and children of Harold Wood Primary School, King’s Wood School and Woodlands Preparatory School in Essex for their willingness to participate in the study; Andrew Faulkner for many helpful comments on the manuscript; Dorothy Bishop and Nicole Harlaar for providing unpublished information about the TOWRE. This research was partially funded by Barking and Dagenham Primary Care Trust. References [1] A. Zumach, E. Gerrits, M.N. Chenault, L.J.C. Anteunis, Otitis Media and speech-innoise recognition in school-aged children, Audiol. Neurootol. 14 (2009) 121–129. [2] E. Sussman, M. Steinschneider, Attention effects on auditory scene analysis in children, Neuropsychologia 47 (2009) 771–785. [3] R. Geva, R. Eshel, Y. Leitner, A. Fattal-Valevski, S. Harel, Verbal short-term memory span in children: long-term modality dependent effects of intrauterine growth restriction, J. Child Psychol. Psychiatry 49 (2008) 1321–1330. [4] L.R. Shapiro, J. Hurry, J. Masterson, T.N. Wydell, E. Doctor, Classroom implications of recent research into literacy development: from predictors to assessment, Dyslexia 15 (2009) 1–22. [5] I. Vanniasegaram, M. Cohen, S. Rosen, Evaluation of selected auditory tests in school-aged children suspected of auditory processing disorder (APD), Ear Hear. 25 (2004) 586–597. [6] G.D. Chermak, E. Tucker, J.A. Seikel, Behavioral characteristics of auditory processing disorder and attention-deficit hyperactivity disorder: predominantly inattentive type, J. Am. Acad. Audiol. 13 (2002) 332–338. [7] K.L. Purvis, R. Tannock, Phonological processing, not inhibitory control, differentiates ADHD and reading disability, J. Am. Acad. Child Adolesc. Psychiatry 39 (2000) 485–494.
600
S. Rosen et al. / International Journal of Pediatric Otorhinolaryngology 74 (2010) 594–600
[8] P. Tallal, M. Piercy, Defects of non-verbal auditory perception in children with developmental aphasia, Nature 241 (1973) 468–469. [9] J.E. Gilliam, Attention-deficit/Hyperactivity Disorder Test: A Method for Identifying Individuals with ADHD, Pro-ed, Austin, TX, 1995. [10] J.K. Torgesen, R. Wagner, C. Rashotte, Test of Word Reading Efficiency (TOWRE), Psychological Corporation, New York, 1999. [11] R.D. Cook, S. Weisberg, Applied Regression Including Computing and Graphics, Wiley, New York, 1999. [12] P.A. Sutcliffe, D.V.M. Bishop, S. Houghton, M. Taylor, Effect of attentional state on frequency discrimination: a comparison of children with ADHD on and off medication, J. Speech Lang. Hear. Res. 49 (2006) 1072– 1084. [13] S. Rosen, Auditory processing in dyslexia and specific language impairment: is there a deficit? What is its nature? Does it explain anything?, J. Phon. 31 (2003) 509–527.
[14] F. Ramus, Developmental dyslexia: specific phonological deficit or general sensorimotor dysfunction? Current Opinion in Neurobiology. 13 (2003) 212–218. [15] J.C. Ziegler, C. Pech-Georgel, F. George, F.X. Alario, C. Lorenzi, Deficits in speech perception predict language learning impairment, Proc. Natl. Acad. Sci. U.S.A. 102 (2005) 14110–14115. [16] P. Dawes, D.V.M. Bishop, T. Sirimanna, D.E. Bamiou, Profile and aetiology of children diagnosed with auditory processing disorder (APD), Int. J. Pediatr. Otorhinolaryngol. 72 (2008) 483–489. [17] C.S. Watson, G.R. Kidd, D.G. Horner, P.J. Connell, A. Lowther, D.A. Eddins, et al., Sensory, cognitive, and linguistic factors in the early academic performance of elementary school children: the Benton-IU project, J. Learn. Disabil. 36 (2003) 165–197. [18] G.M. McArthur, D. Ellis, C.M. Atkinson, M. Coltheart, Auditory processing deficits in children with reading and language impairments: can they (and should they) be treated? Cognition 107 (2008) 946–977.