0 1990 Gordon and Breach, Science Publishers, Inc.
Intern. J. Neuroscience, 1990, Vol. 50, pp. 233-242 Reprints available directly from the publisher Photocopying gmnitted by license only
Printed in the United Kingdom
BRAINSTEM EVOKED POTENTIALS IN LEARNING DISABLED CHILDREN ERZSE BET MAROSI, THALIA HARMONY, JACQUELINE BECKER
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
ENEP Iztacala, Universidad Nacional Autonoma de Mexico Department of Neurosciences (Received August 31, 1989)
Averaged brai.istem auditory responses (BAER) were recorded using left and right ear stimulation with clicks of 90 dB!; in two groups of primary school children (control and learning disabled). A linear multiple regression moc.el was used in an attempt to demonstrate the effects of sex, high risk factors related to brain damage and lexning disability on the evoked responses. Sex showed a strong influence in the latencies of the first five peaks, with girls having shorter latencies. Risk factors had an effect on the latency of peak V, the I-V interva: and the V/I amplitude ration, but only when the left ear was stimulated. Learning disability had no significint influence according to this analysis. Multivariate test of complete homogeneity showed highly significant differences between LD and control boys when the left ear was stimulated and between control and LT) girls when the right ear was stimulated. Principal component analyses revealed differences between the two groups: the BAER components of the control subjects showed a minor source of variance when the right ear was stimulated. A contrary effect was observed in LD children. Such differences might be related to ear preference and hemispheric dominance. Keywords: brai.rstem evoked potentials, learning disability, latency, interval, amplitude. multivariate statistics
BAERs have been used for estimating the amount of hearing loss (Hecox & Galambos, 1974), l’or early detection of acoustic neuroma and other neurological disorders (Starr, 1978).as well as to monitor changes due to maturation of the auditory system (Eggermont, 1986). Children with different types of neuropsychological problems have also been studied: Mason & Mellor (1 984) found smaller amplitudes for waves I, I11 and V in children with language and motor speech disorders. Squires et al. (1980) observed a prolongation of the central conduction time in mentally retarded children, and Tanguay et al. (1982) reported larger wave intervals in autistic children. Sohmer & Student (1978), comparing the recordings of four groups of children (control, autistic, mentally retarded and minimally brain dysfunctional children), found an increase in latencies and central conduction time that was proportional to the severity of the pathology. They consider these conditions a continuum ranging from mental retardation through autism and minimal brain dysfunction to normal functioning. Rumsey et id. (1984) recorded BAERs from atuistic children and found as much increase as decrease of central conduction time, denying the possibility of an interpretation related to any specific pathology and concluding that the results were due to concomitant neurological lesions or to peripheral hearing loss as a result of poorly controlled subject selction. We were interested in the study of learning disabilities (LD), and we have a well-selected sample of children without hearing impairment, neurologically normal and with normal IQ. Welsh et al. (1982) were not able to demonstrate any significant
Reprint request: Erzsebet Marosi, ENEP-UNAM-Iztacala, Apartado 3 14, 54030 Tlalnepantla, Estado de mexico. 233
2-34
t. MAROSl, T. HARMON\. A N D J. BECKER
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
abnonnality of the BAERs in dyslexic children. Children with LD are characterized by attention problems and distractability. As the brainstem reticular formation is directly related to learning and attention, we thought that brainstem dysfunctions might be responsible for some aspects of learning difficulties and that they might also produce BAER alterations. As L,D children very frequently have antecedents of risk factors associated with brain damage, we also wanted to evaluate the effect of such factors in school-age children, since it has been demonstrated that perinatal distress affects BAER latencies and intervals (Salamy et al.. 1985; Kileny et al., 1980: Ken Dror et al. 1987). As sex differences have also been described (Picton et al.. 1981; Moch17uki et al.. 1983), we wanted to analyze them also as a control for the effects that could be expected in our data. MATERIAL AND METHODS
A careful selection of the experimental subjects took place. The original number of 143 subjects was reduced to 82 after eliminating those children who had an IQ below 85 (measured by the Wechsler Intelligence Scale for Children n = 3), an auditory threshold above 25 db ( n = 2), or an atypical waveform of the BAER which made the identification of the components ambiguous ( n = 56). The selection of the artifactfree potentials was made by visual inspection. Of the remaining 82 subjects, 47 belonged to the control group ( 2 3 girls and 24 boys \+ith a mean age of 8.9 years) and 35 were children with learning disabilities ( 1 2 girls and 23 boys with a mean age of 9.2 years). The control children were free of any type of learning. emotional or behavioral problems and had a normal pediatric and neurological examination. The children of both groups were completely healthy and free of any medication at the time of recording. The learning disabled group was selected according to three basic criteria: a ) academic achievement: b) teachers‘ and parents’ opinion; and c) a specific pedagogical test on reading and writing (Hinojosa & Rocha, 1985). The control group was assigned from children who had not failed any courses at school with normal behavior at home and in school, and normal scores on the pedagogical test. The presence or absence of pathological antecedents that might be considered as possible cause of brain damage was estimated on the basis of detailed clinical history. The number of pre-, peri- and postnatal incidents was ascertained and a cumulative TABLE 1 Scale of scores for each antecedent of risk 2 points a.ere giken to each o f the following factors: Pregnancy, Malnutrition Pcrinatal: Forcepa. anoxia, circular umbilical cord. weight at birth smaller than 7.700 g. incubator Ibi w era1 da) $. Pr)stiiatal: Se~erc.malnutrition. severc dehydration. hcad trauma with loss of cnnsciousnery I point was added for each of the following factors: l’rcgnnncy: Bleeding. diabetes. trauma. drug use. mother’s agc older than 40. histor? o f ;ibortiony. L O I ~ \ulsi\ i‘ at.itcs Pcrinatal: Weight at birth bet\veen 2.200 and 7.500 g. incubator for hours, health1 piciiia~iiic. Iiypciinaturc. c!anosis. uncontrolled delivery at home. unprogrammed Cesarean or pelvic delivery, prolonged deli vcry Postnatal: Febrile convulsiona. weeping spasms. prolonged k b d e states duriiig the tirst year ot Iifc. ‘ibsence of breastfeeding with n o t adequate substitute for breastmilk. late introduction ol food other than hi es\tniilh ~~~
~
BAER IN LEARNING DISABLED CHILDREN
235
TABLE 2 Characteristics of the sample
Group
Sex
Risk
n
control
girls
with risk without risk with risk without risk with risk without risk with risk without risk
12 I1 12 12 9 3
boys
LD
girls
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
boys
19
4
score of the number of risk factors was created (Table 1). All our subjects came from low-income families and as a consequence had a high frequency of risk factors. The distribution of the children according to group, sex and risk factors can be seen in Table 2. Recordings were made with a MEDICID computer system. The left and right auditory brainstem responses were recorded between vertex (Cz) and the left (Al) and between ver:ex and the right ear (A2) according to the 10-20 International System. The biopotentials were amplified (x 500,000), filtered (100 Hz-3 KHz), sampled at 40 microsecono intervals and automatically averaged (3000 trials). The auditory brainstem responses were elicited monaurally, stimulating the ipsilateral ear with a series of rarefaction clicks. The characteristics of the stimulus were: repetition rate of 9/s and duration of 100 ps with an intensity of 90 dB. The contralateral ear was masked by white noise af 60 dB, Peaks were detected by visual inspection; recordings without a conventional waveform were eliminated. In every response the following parameters were calculated: latencies of the first five waves (VI and VII were eliminated from the statistical treatment because of their inconsistency), intervals I-V, 11-V, and I-IV, and the V/I amplitude ratio. Unfortunately, due to a change in the calibration of the amplifiers, we could not calculate the absolute amplitude values. The data were examined in accordance with the exploratory analysis of Tukey to assure the normal distribution of the residuals (Tukey, 1977.; Curts, 1984). Five statistical analyses were performed: 1. Univariatl: statistics: to determine differences between the mean values of the latencies and intervals for the control and learning-disabled groups. Comparisons between L D and control children were made independently for each sex. 2. Analysis of Variance: to reveal the effect of sex and of risk in the control group. Two different models were used: in one, risk was considered present or absent; in the other, the cumulative score of risk was taken as a covariable. 3. Multiple Linear Regression: to reveal the simulataneous influence of the independent variables (sex, the cumulative score of antecedents of risk, and the presence of learning disz bility) on the dependent variables (latencies, intervals, V/I amplitude ratio) of the whole sample. 4. Principal Component Analysis: to observe the interaction and the internal structure of the (different variables. Starting from the matrix of correlation, different principal component analyses were performed for the whole data matrix and for the latencies of t be first five waves for each ear stimulated in the control and LD groups. In the latter, Varimax rotation was used. The chi-square statistics (Bartlett's approxi-
E. MAROSI, T HARMONY A N D J BECKER
236
mation. Mardia et al., 1979) was used to test the hypothesis that the last eigenvalues were equal. 5 . Multivariate Test of Complete Homogeneity: To test the hypothesis of the equality of mean vectors and the matrices of variance and covariance between the control and the learning disabled groups, independently for each sex (Mardia et a].. 1979).
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
RESULTS 1 . In each sex the average values and the standard deviations for the control and LD groups were calculated for each variable (Tables 3 and 4). Mean values of latencies were similar to those previously published by John et al. (1982). Univariate statistical analysis revealed no significant differences between boys. Control and LD girls showed differences only in the latency of wave I1 stimulating the left ear (p = .02) and in the latency of wave V stimulating the right ear (p = .04). Those differences were not in the expected direction: control girls had longer latencies than LD. 2. ANOVA for sex and risk in the control group: Sex made for highly significant differences (p = ,002 for multivariate Wilks' lambda) in the latencies when stimulating the left ear. Sex explained 31 YOof the variance of wave 111, where the difference according to sex was very significant (p = .OOO). Waves IV (p = .012) and V (p = .024) were also TABLE 3 Mean and SD Values of Latencies and V/I Amplitude Ratio to the Stimulation of the Left Ear. Meas uremeii ts Group
I
11
Control Boys
*1.15 .24
2.24
3.40
-.I7
.I9
1.05 i 13
2.24 k.17
3.30
i .21
Learning Disabled Boys Control Girls
1.07
2.16
+ .12
t .I4
1.05 &.I3
+.I1
~~
Learning Disabled Girlc
+
2.03
V
V l
- .I8
+
5.26 +.I5
2 1.13
4.30 2.28
t .29
IV
111
3.20
+
- .I0
3.20 - .I4 +
4.40
4.30 +.I?
4.23
+
- .21
5.27
5.15
+.IX
I .54
1.81
+ 2.36
~~
I .29
*.?I
5.07
.99
+ .I3
i .5x
..~
TABLE 4 Mean and SD Values of Latencies and VII Amplitude Ratio to the Stimulation of the Right Ear Measurements Group Control Boys
111
I 1.12
i. I 8 Learning Disabled Boys Control Girls
1.02
-c .21 1.10 1.14
Learning Disahled Girls
1.10 1.I8
2.24 +.I6
*3.30
2.23 .I7
3.26 & .18
2.14
+
- .I3
2.19
+ .21
.I6
3.20
Iv
4.38
V
V/I
5.23
I.26
+.I9
I.96
4.33
5 29
i .25
-2 5 +
1-14 & .55
5.18 .1R
1.11 f .73
+ .25
_.
4.21 .20
+ .14
f
t
*3.18.20
4.26 5.21
5.05
.9x
& .08
& 20
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
BAER IN LEARNING DISABLED CHILDREN
231
affected with girls having shorter latencies than boys. With right ear stimulation, only waves I1 (p = .015) and TI1 0, = .007) showed significant differences between sexes, girls having, shorter latencies than boys. When considered as present or absent, no risk effect was observed. When the cumulative score was taken as covariable great interactions between sex and risk were observed with left ear stimulation. Therefore, risk effect was separately evaluated in girls and in boys. In females, the latency of wave I11 when stimulating the left ear had a positive (correlation with the risk score (Y = .44). Responses to the right ear stimulation were not affected by risk. 3. Table 5 shows the results of multiple linear regression, where R-square is the percentage of the explained variance and the t the significant values of Student’s test. When all data were considered, sex was the strongest independent variable, confiming previously-mentioned results. The variable “sex” explained, at most, 14% of the variation in the latencies. Girls had shorter latencies than boys (negative values). The presence of risk factors produced larger wave V latency and I-V interval and smaller V/I amplitude ratio for the left ear stimulation. The R-square values shown in Table 6 wcre lower than those obtained when only the control group was analyzed. As the values of R-square for the total model indicated that at most 21% of the variance was explained, we consider that the multiple regression model was not optimal. 4. The princl pal component analyses (PCA) of all parameters (latencies, intervals and V/I ratios for both left- and right-ear stimulation) for all subjects showed that from a total of 18 variables, 17 were significantly different. The first 8 eigenvectors explained 90% of the variance. The first eigenvector (explaining 32% of the variance) was mainly correlated with wave V latency and I-V interval for both left- and right-ear stimulation; the second eigenvector (19’/0) with wave I latency for both left- and right-ear stimulation; third eigenvector (10%) was related only to variables when the left ear was stimulated: I1 latency and 11-V interval; the forth eigenvector (9%) was highly correlated with the latency of wave I when the left ear was stimulated and 11-V interval and VjI amplitude ratio of the response to right-ear stimulation; the fifth eigenvector (6%) was related to V/I amplitude ratio during left-ear stimulation and with interval I-IV during right ear stimulation. The sixth eigenvector (5%) was TABLE 5 Multiple Linear Regression (Only Dependent Variables with Significant Effects Are Shown) Dependent Variables
Lat. I1 Left Stirn. Lat. I1 Right Stim.
Risk
Sex
Multiple
R2
t
R2
t
R2
.13
- 3.53
-
-
.16
- 2.28
-
-
.09
-
.16
.06
Lat. 111 Left Stim.
.14
- 3.65
-
Lat. 111 Right Stim.
.08
- 2.69
-
-
.15
Lat. IV Left Stim.
.07
- 2.50
-
-
.10
-
2.27
.21 .ll
.09
2.12
.I5
.05
- 1.99
.10
Lat. V Left Stini.
.08
- 2.56
Lata. V Right Stirn.
.09
- 2.74
.06 -
I-V Interval Left Stim.
-
-
V/l Ratio Left Stirn.
-
-
E. MAROSI. T. HARMONY AN-D J . BECKER
238
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
related to \'/I amplitude ratio during left ear stimulation and to the latency of wave IV during right-ear stimulation. The seventh eigenvector (4%) was related to latencies of wave TV during left-ear stimulation and of wave I1 during right-ear stimulation. The eighth eigenvector (3%) was mainly related by latency of wave 111 to left-ear stimulation. These results show us multiple sources of variance and make interpretation difficult, since only waves V and I were explained by the same eigenvector when the left and right ear were stimulated; therefore, we calculated the principal components separTABLE 6 Principal Component Analysis of Latencies of Control Group to the Stimulation of the Left Ear Me;rsurenicnts
I II I11 I\'
1.atency Latency Latency Latency latency
V
Accun1.
"0
Eigenvector I
Eigenvector
Eigenvector 3
Eigenvector 4
Eigenvector
?
0.47 0.40 0.44 0.45 0.38 4x.57
-0.50 0.41 -- 0.55 0.22 0.37 67.11
0.17 0.26 0.36 0.50 - 0.72 82.97
0.48 0.47 -- 0.35 - 0.66 0.00 94.08
0.52 ~~055 - 0 49 0 25 0.34 100.00
E I gen \ ect o r
€1
-
>
Rotated \ alues ~
.21en < urernen t
I
Latency latencq Litency Latency latency
r II Ill IV
V
0.38 0.8 I 0.04 0.84 0.19
gen vec t or
E r gen vect or 7
7
0.80 -0.13 0.89 - 0.20 -0.13 .-
~
0.06 0.33 0.26 - 0.03 - 0.95 -
TABLE 7 Princip:il C'omponent Analysis o f Latencies of Control Group to the Stimulation of the Right Ear Measurements ______ Latency I latency 11 Latencq I l l Latency IV 1-atency L,'
Accum.
'
Rleasurenients
Eigenvcctor 0.32 0.46 0.47 0.49 0.47 4X.25
Eigenvector Eigenvector Eigenvector 2 3 4 ____~ ______ 0.03 0 . I3 0.11 0.06 - 0.72 0.09 0.14 0.50 - 0.70 0.1 I - 0.26 0.01 0.3 I 0.39 0.70 65.19 78.49 X9.37
Eigenvector I
-3
I
~~
Figenvector
l'itenc> I1 Latency 111
0.16 0.69 0 .I 3.
Latency IV Latenq V
0.7 5
0.98 0.20 -0.14 -0.18
0.7x
0.00
latcncq I
-
EigenLcctor 5
_____
o o2 0.51 0 1.1
0.82 0.70 IOO.00
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
BAER I N LEARNING DISABLED CHILDREN
239
ately for the different groups and ear stimulated, taking into consideration the latencies of waves I to V. The results were as follows: a) There were three different components in the control group during stimulation of the left ear ((Table6). Varimax rotation of these 3 components showed that eigenvector 1 accounted for the latencies of waves I1 and IV, eigenvector 2 explained the latencies of waves I and 111, and the latency of wave V was explained by eigenvector 3. These three components represented 83% of total variance. b) Two difirent components could be observed in the control group when stimulating the right ear, explaining 65% of total variance (Table 7). (The same results were obtained with and without Varimax rotation): eigenvector 2 explained the variance of wave I and the remaining latencies were explained by eigenvector 1. c) In the learning-disabled group, two different principal components were identified after stimulation of the left ear (Table 8), representing 73% of the total variance. The rotated eigenvector 1 explained all the latencies with the exception of wave I which was represented by eigenvector 2. Wave I1 was equally represented in both factors. d ) In the learning-disabled group, when stimulating the right ear, three independent components could be distinguished (Table 9), explaining 88% of the total variance. Rotated eigenvector 1 explained the variance of the latencies of waves IV and V. Eigenvector 2 represented the latency of wave I and eigenvector 3 explained the latencies of waves I1 and 111. 5. Multivariate test of Complete Homogeneity: As our main purpose was to try to find differences between the control and the LD groups, we also performed multivariate tes1.s of complete homogeneity which take into consideration not only the mean vector differences but also inequality of variance and of covariance matrices. Comparing the actual values of the latencies of the first five waves and the V/I amplitude ratio according to sex and ear stimulated, we obtained the following results: a) When the left ear was stimulated, we found highly significant differences between the control and LD boys (p = .OOOl). In the case of the girls, the same procedure did TABLE 8 Principal Component Analysis of Latencies of the Learning Disabled Group to the Stimulation of the Left Ear Measurements
Eigenvector 1
Latency I Latency 11 Latency 111 Latency IV Latency V Accum. %
Eigenvector 2
Eigenvector 3
0.10 0.44 0.54 0.52 0.49 49.07
- 0.82
0.45
- 0.32
- 0.67 - 0.07
Measurements
Eigenvector 1
Eigenvector 2
Latency Latency Latency Latency Latency
- 0. LO
-0.17 0.38 0.24 72.99
Rotated Values
I 11 111 IV
V
0.56 0.76 0.91 0.82
0.90 0.53 0.40
- 0.08 - 0.06
0.00 0.59 87.05
Eigenvector 4
Eigenvector 5
0.17 0.43 - 0.82 0.2 I 0.26 94.35
-0.31 0.27 0.06 - 0.74 0.54 100.00
E. MAROSI. T. HARMONY .4ND J . BECKER
240
TABLE 9 Principal Component Analysis of Latencies of the Learning Disabled Group to the Stimulation of the Right Ear Measurements
Eigenvector 1
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
Latency Latency Latency Latency Latency Accum.
I 11
I11 IV V O/O
Eigenvector 2
Eigenvector 3
Eigenvector 4 0.09
0.28 0.47 0.54 0.49 0.42 55.28
- 0.74
0.60
0.26 - 0.05 0.28 0.54 73.54
- 0.60
- 0.10
- 0.30
-0.12 0.79 - 0.59 96.03
Eigenvector I
Eigenvector 2
Eigenvector 3
0.09 0.13 0.43 0.78 0.91
0.98 0.13 0.16 0.16 0.01
-0.16 0.94 0.80 -0.35 -0.13
-
0.23 0.38 88.16
-
Eigenvector 5 0.03 0.59 - 0.78 0.10 0.20 100.00
Rotated Values Measurements Latency Latency Latency Latency Latency
I I1
IT1 IV V
-
-
not give significant results, although the probability of rejecting the null hypothesis was .18. b) When the right ear was stimulated, no significant differences were found between the control and learning-disabled boys. In the case of the girls, the difference between the control and the learning disabled group was significant at a p = .01 level. When the scores of the first eight principal components were compared between control and LD children according to sex, boys presented significant differences 0, = .0004), but girls did not 0, = ,117). DISCUSSION First we shall discuss the results obtained in relation to sex and risk, and later those on LD. Our results confirm the reports of many authors (Picton et al.. 1981; Mochizuki et al., 1983) in relation to sex differences, i.e., females displaying shorter latencies of all waves. In relation to risk antecedents we observed that in control subjects latency of wave I11 during stimulation of the left ear in girls had a positive correlation with the risk score. When the whole sample was considered, by multiple regression, we observed longer latency of wave V, longer I-V interval and smaller V/I amplitude ratio in the left ear in children with risk antecedents, which suggest the existence of BAER abnormality due to risk. Differences in BAER latencies and amplitudes between premature and high-risk neonates were observed by Salamy et al. (1985) before the age of two years. We consider our observation important given that many children for whom BAER study is recommended have antecedents of risk factors. As the trace left by risk on BAERs continues to be present for many years, it is important to have a complete clinical history in order to make more appropriate evaluations. In relation to LD. our results showed no evident differences. By univariate analysis we found two unexpected differences between control and LD girls, which we can only
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
BAER IN LEARNING DISABLED CHILDREN
24 1
explain as clue to chance, given that multiple comparisons were made. The absence of obvious differences agrees with the results obtained by Welsh et al. (1982) & Finley & Johnson (1983). These authors found amplitude differences in BAER between normal and LD children, but as we mentioned earlier, we had calibration problems and were unable to confirm their results. Application of multivariate statistics revealed clear, significant differences in the mean vector and variance and covariance matrices for left ear stimulation between control and LD boys and for right ear stimulation between control and LD girls. The comparisor of the first eight factor loadings taking into account both ears also demonstrates significant differences in boys. This result focused on boys might be due to the sample composition, since we had many more LD boys than LD girls. The differences observed between the covariance matrices are understandable in light of principal component analysis results. The detailed PCA of the wave latencies by ear and sex groups produced interesting results: Coritrol children had three differentiable components during left-ear stimulation and t u o when the right ear was stimulated. The opposite was observed in LD children. If it is assumed that a conductive system is more efficient with less relays, for control children the right ear looked more efficient, while for the LD the left ear was more efficient. A differential hemispheric involvement in reading of left- and right-ear dominant children was demonstrated by Bakker et al. in 1980. Primary-school children who showed left- or right-ear advantage subsequent to a dichotic listening task were required to name monosyllabic words presented at a fixation point. Authors found different strategies of reading depending on the ear preference observed. Those showing lef :-ear preference and supposed right hemispheric control were sensitive to the semantic and linguistic aspects of the test. Although we had not determined ear preference, the electrophysiological results suggest that this was the key to the differences observed between control and LD children. We want to emphasize that we have described such differences in a sample of children selected by clinical and electrophysiological criteria: most subjects had all peaks detecable and therefore a clear waveform composition. Our initial hypothesis of a brainst1:m dysfunction producing learning problems and BAER alterations was not confirmed, since the differences observed between normal and LD children were related to ear lateralization and therefore to hemispheric dominance, and no signs of brainstem abnormality were found.
REFERENCES Bakker, D. J., Licht, R., Kok, A., & Bouma, A. (1980). Cortical responses to word reading by right- and left-eared normal and reading disturbed children. Journal of Clinical Neuropsychology. 2; 1-12. Curts, J. B. (15184). Introduction al analisis de residuos en biologia. Bihtica, 9, 271-278. Eggermont, J. .I_ (1986). Evoked potentials as an indicators of the maturation of the auditory system in V. Gallai (Ed.) Maturation pf the CNS and evoked potentials, Elsevier, 177-182. Finley, W., & Johnson, G. (1983). Operant control of auditory brainstem potentials in man. International Journal of Neuroscience, 21, 161-170. Galaburda, A. M., Sherman, G. F., Rosen, G. D., Aboitiz, F., & Geschwind, N. (1985). Developmental dyslexia: four consecutive patients with cortical anomalies. Annals of Neurology, 18(2), 222-233. Galambos, R., Hecox, K. E. (1978). Clinical applications of the auditory brain sem response. Clincal Otolaryngology, North America 11, 709-718. Hinojosa, G., tt Rocha, C. (1985). Evaluacion de lectura y escritura en niliios de primaria. V. Coloquio Interno de Investigacion ENEP, Iztacala. Ken-Dror, A., Pratt, H., Zeltzer, P., Sujov, P., Katzir, J., & Benderley, A. (1987). Auditory BAEP to clicks at differen presentation rates: estimating maturation of pre-term and full term neonates. EEG and Clinical Ntwrophysiology, 68, 209-2 18.
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
242
E. MAROSI, T. HARMONY A N D J . BECKER
Kileny. P.. Connelly. C., & Robertson, C. (1980). Auditory brainstem responses in perinatal asphyxia. International Journal of Pediatric Otorhinolaryngology, 2, 147-1 59. Mardia, K. V., Kent, J. 1.. & Bibby, J. M. (Eds.) (1979). Multivariate analysis London: Academic Press. Mason, S. M., Mellor, D. H . (1984). Brain stem middle latency and late cortical evoked potentials in children with speech and language disorders. EEG and Clinical Neurophysiologj. 59.. 297-309. Mochizuki. Y., Go. T., Ohkubo, H., & Motomura, T. (1983). Development of human brainstem auditory evoked potentials and gender differences from infants to young adults. Progress in Neurohiologj, 20. 273-285. Picton, T.. Stapells. D. R., & Campbell, K. B. (1981). Auditory evoked potentials from human cochlea and brainstem. Le Journal d' Oto-Rhino-Laryngologie Supl. 10. Rumsey, J . M., Grimes, M., Pikus, A. M., Duara, R., & Ismond, D. R. (1984). Auditory brainstem responses in pervasive developmental disorders. Biological Psychiatry, 19, L-17. Salamy, A., Eldredge, L.. & Wakeley, A. (1985). Maturation of contralateral brain stem responses in preterm infants. EEG and Clincal Neurophysiology. 62. 117-123. Sohmer, H., Student, M. (1978). Auditory nerve and BAER in normal, autistic, minimal brain dysfunction and psychomotor retarded children. EEG and Clincal Neurophysiology. 44, 380-388. Squires, N., Aine, C., Buchwald, J., Norman, R., & Galbraith, G. (1980). Auditory brain stem response abnormalities in severely and profoundly retarded adults. EEG and Clinical Neurophjvsiologj 50., 172-185. Starr. A. (1978). Sensory evoked potentials in clinical disorders of the nervous system. Anzericaiz Review qf'Neuroscience. 1, 103-127. Tanguay. P. E., Edwards, R. M., Buchwald, J., Schwafel, J., & Allen, V. (1982). Auditory brainstem evoked responses in autistic children. Archives of General Psychiatry. 39., 174-180. Trevarthen, C. (ed) (1974). Functional relationship of disconnected hemisphere with the brainstem and with each other: monkey and man. Hemispheric disconnection and cerebral function. Springfield. 187-207. Tukey, J. W. (ed). (1977). Exploratory data analysis. Addison-Wesley Publishing Co. Welsh, L. W., Welsh, J. J., Healy, M., & Cooper, B. (1982). Cortical, subcortical and brainstem dysfunction: a correlation in dyslexic children. Annals of Otology, Rhinology and Lar.vngo1og.v. 91, 3 10-3 15.
Intern. J. Neuroscience, 1990, Vol. 50, pp. 233-242 Reprints available directly from the publisher Photocopying gmnitted by license only
Printed in the United Kingdom
BRAINSTEM EVOKED POTENTIALS IN LEARNING DISABLED CHILDREN ERZSE BET MAROSI, THALIA HARMONY, JACQUELINE BECKER
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
ENEP Iztacala, Universidad Nacional Autonoma de Mexico Department of Neurosciences (Received August 31, 1989)
Averaged brai.istem auditory responses (BAER) were recorded using left and right ear stimulation with clicks of 90 dB!; in two groups of primary school children (control and learning disabled). A linear multiple regression moc.el was used in an attempt to demonstrate the effects of sex, high risk factors related to brain damage and lexning disability on the evoked responses. Sex showed a strong influence in the latencies of the first five peaks, with girls having shorter latencies. Risk factors had an effect on the latency of peak V, the I-V interva: and the V/I amplitude ration, but only when the left ear was stimulated. Learning disability had no significint influence according to this analysis. Multivariate test of complete homogeneity showed highly significant differences between LD and control boys when the left ear was stimulated and between control and LT) girls when the right ear was stimulated. Principal component analyses revealed differences between the two groups: the BAER components of the control subjects showed a minor source of variance when the right ear was stimulated. A contrary effect was observed in LD children. Such differences might be related to ear preference and hemispheric dominance. Keywords: brai.rstem evoked potentials, learning disability, latency, interval, amplitude. multivariate statistics
BAERs have been used for estimating the amount of hearing loss (Hecox & Galambos, 1974), l’or early detection of acoustic neuroma and other neurological disorders (Starr, 1978).as well as to monitor changes due to maturation of the auditory system (Eggermont, 1986). Children with different types of neuropsychological problems have also been studied: Mason & Mellor (1 984) found smaller amplitudes for waves I, I11 and V in children with language and motor speech disorders. Squires et al. (1980) observed a prolongation of the central conduction time in mentally retarded children, and Tanguay et al. (1982) reported larger wave intervals in autistic children. Sohmer & Student (1978), comparing the recordings of four groups of children (control, autistic, mentally retarded and minimally brain dysfunctional children), found an increase in latencies and central conduction time that was proportional to the severity of the pathology. They consider these conditions a continuum ranging from mental retardation through autism and minimal brain dysfunction to normal functioning. Rumsey et id. (1984) recorded BAERs from atuistic children and found as much increase as decrease of central conduction time, denying the possibility of an interpretation related to any specific pathology and concluding that the results were due to concomitant neurological lesions or to peripheral hearing loss as a result of poorly controlled subject selction. We were interested in the study of learning disabilities (LD), and we have a well-selected sample of children without hearing impairment, neurologically normal and with normal IQ. Welsh et al. (1982) were not able to demonstrate any significant
Reprint request: Erzsebet Marosi, ENEP-UNAM-Iztacala, Apartado 3 14, 54030 Tlalnepantla, Estado de mexico. 233
2-34
t. MAROSl, T. HARMON\. A N D J. BECKER
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
abnonnality of the BAERs in dyslexic children. Children with LD are characterized by attention problems and distractability. As the brainstem reticular formation is directly related to learning and attention, we thought that brainstem dysfunctions might be responsible for some aspects of learning difficulties and that they might also produce BAER alterations. As L,D children very frequently have antecedents of risk factors associated with brain damage, we also wanted to evaluate the effect of such factors in school-age children, since it has been demonstrated that perinatal distress affects BAER latencies and intervals (Salamy et al.. 1985; Kileny et al., 1980: Ken Dror et al. 1987). As sex differences have also been described (Picton et al.. 1981; Moch17uki et al.. 1983), we wanted to analyze them also as a control for the effects that could be expected in our data. MATERIAL AND METHODS
A careful selection of the experimental subjects took place. The original number of 143 subjects was reduced to 82 after eliminating those children who had an IQ below 85 (measured by the Wechsler Intelligence Scale for Children n = 3), an auditory threshold above 25 db ( n = 2), or an atypical waveform of the BAER which made the identification of the components ambiguous ( n = 56). The selection of the artifactfree potentials was made by visual inspection. Of the remaining 82 subjects, 47 belonged to the control group ( 2 3 girls and 24 boys \+ith a mean age of 8.9 years) and 35 were children with learning disabilities ( 1 2 girls and 23 boys with a mean age of 9.2 years). The control children were free of any type of learning. emotional or behavioral problems and had a normal pediatric and neurological examination. The children of both groups were completely healthy and free of any medication at the time of recording. The learning disabled group was selected according to three basic criteria: a ) academic achievement: b) teachers‘ and parents’ opinion; and c) a specific pedagogical test on reading and writing (Hinojosa & Rocha, 1985). The control group was assigned from children who had not failed any courses at school with normal behavior at home and in school, and normal scores on the pedagogical test. The presence or absence of pathological antecedents that might be considered as possible cause of brain damage was estimated on the basis of detailed clinical history. The number of pre-, peri- and postnatal incidents was ascertained and a cumulative TABLE 1 Scale of scores for each antecedent of risk 2 points a.ere giken to each o f the following factors: Pregnancy, Malnutrition Pcrinatal: Forcepa. anoxia, circular umbilical cord. weight at birth smaller than 7.700 g. incubator Ibi w era1 da) $. Pr)stiiatal: Se~erc.malnutrition. severc dehydration. hcad trauma with loss of cnnsciousnery I point was added for each of the following factors: l’rcgnnncy: Bleeding. diabetes. trauma. drug use. mother’s agc older than 40. histor? o f ;ibortiony. L O I ~ \ulsi\ i‘ at.itcs Pcrinatal: Weight at birth bet\veen 2.200 and 7.500 g. incubator for hours, health1 piciiia~iiic. Iiypciinaturc. c!anosis. uncontrolled delivery at home. unprogrammed Cesarean or pelvic delivery, prolonged deli vcry Postnatal: Febrile convulsiona. weeping spasms. prolonged k b d e states duriiig the tirst year ot Iifc. ‘ibsence of breastfeeding with n o t adequate substitute for breastmilk. late introduction ol food other than hi es\tniilh ~~~
~
BAER IN LEARNING DISABLED CHILDREN
235
TABLE 2 Characteristics of the sample
Group
Sex
Risk
n
control
girls
with risk without risk with risk without risk with risk without risk with risk without risk
12 I1 12 12 9 3
boys
LD
girls
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
boys
19
4
score of the number of risk factors was created (Table 1). All our subjects came from low-income families and as a consequence had a high frequency of risk factors. The distribution of the children according to group, sex and risk factors can be seen in Table 2. Recordings were made with a MEDICID computer system. The left and right auditory brainstem responses were recorded between vertex (Cz) and the left (Al) and between ver:ex and the right ear (A2) according to the 10-20 International System. The biopotentials were amplified (x 500,000), filtered (100 Hz-3 KHz), sampled at 40 microsecono intervals and automatically averaged (3000 trials). The auditory brainstem responses were elicited monaurally, stimulating the ipsilateral ear with a series of rarefaction clicks. The characteristics of the stimulus were: repetition rate of 9/s and duration of 100 ps with an intensity of 90 dB. The contralateral ear was masked by white noise af 60 dB, Peaks were detected by visual inspection; recordings without a conventional waveform were eliminated. In every response the following parameters were calculated: latencies of the first five waves (VI and VII were eliminated from the statistical treatment because of their inconsistency), intervals I-V, 11-V, and I-IV, and the V/I amplitude ratio. Unfortunately, due to a change in the calibration of the amplifiers, we could not calculate the absolute amplitude values. The data were examined in accordance with the exploratory analysis of Tukey to assure the normal distribution of the residuals (Tukey, 1977.; Curts, 1984). Five statistical analyses were performed: 1. Univariatl: statistics: to determine differences between the mean values of the latencies and intervals for the control and learning-disabled groups. Comparisons between L D and control children were made independently for each sex. 2. Analysis of Variance: to reveal the effect of sex and of risk in the control group. Two different models were used: in one, risk was considered present or absent; in the other, the cumulative score of risk was taken as a covariable. 3. Multiple Linear Regression: to reveal the simulataneous influence of the independent variables (sex, the cumulative score of antecedents of risk, and the presence of learning disz bility) on the dependent variables (latencies, intervals, V/I amplitude ratio) of the whole sample. 4. Principal Component Analysis: to observe the interaction and the internal structure of the (different variables. Starting from the matrix of correlation, different principal component analyses were performed for the whole data matrix and for the latencies of t be first five waves for each ear stimulated in the control and LD groups. In the latter, Varimax rotation was used. The chi-square statistics (Bartlett's approxi-
E. MAROSI, T HARMONY A N D J BECKER
236
mation. Mardia et al., 1979) was used to test the hypothesis that the last eigenvalues were equal. 5 . Multivariate Test of Complete Homogeneity: To test the hypothesis of the equality of mean vectors and the matrices of variance and covariance between the control and the learning disabled groups, independently for each sex (Mardia et a].. 1979).
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
RESULTS 1 . In each sex the average values and the standard deviations for the control and LD groups were calculated for each variable (Tables 3 and 4). Mean values of latencies were similar to those previously published by John et al. (1982). Univariate statistical analysis revealed no significant differences between boys. Control and LD girls showed differences only in the latency of wave I1 stimulating the left ear (p = .02) and in the latency of wave V stimulating the right ear (p = .04). Those differences were not in the expected direction: control girls had longer latencies than LD. 2. ANOVA for sex and risk in the control group: Sex made for highly significant differences (p = ,002 for multivariate Wilks' lambda) in the latencies when stimulating the left ear. Sex explained 31 YOof the variance of wave 111, where the difference according to sex was very significant (p = .OOO). Waves IV (p = .012) and V (p = .024) were also TABLE 3 Mean and SD Values of Latencies and V/I Amplitude Ratio to the Stimulation of the Left Ear. Meas uremeii ts Group
I
11
Control Boys
*1.15 .24
2.24
3.40
-.I7
.I9
1.05 i 13
2.24 k.17
3.30
i .21
Learning Disabled Boys Control Girls
1.07
2.16
+ .12
t .I4
1.05 &.I3
+.I1
~~
Learning Disabled Girlc
+
2.03
V
V l
- .I8
+
5.26 +.I5
2 1.13
4.30 2.28
t .29
IV
111
3.20
+
- .I0
3.20 - .I4 +
4.40
4.30 +.I?
4.23
+
- .21
5.27
5.15
+.IX
I .54
1.81
+ 2.36
~~
I .29
*.?I
5.07
.99
+ .I3
i .5x
..~
TABLE 4 Mean and SD Values of Latencies and VII Amplitude Ratio to the Stimulation of the Right Ear Measurements Group Control Boys
111
I 1.12
i. I 8 Learning Disabled Boys Control Girls
1.02
-c .21 1.10 1.14
Learning Disahled Girls
1.10 1.I8
2.24 +.I6
*3.30
2.23 .I7
3.26 & .18
2.14
+
- .I3
2.19
+ .21
.I6
3.20
Iv
4.38
V
V/I
5.23
I.26
+.I9
I.96
4.33
5 29
i .25
-2 5 +
1-14 & .55
5.18 .1R
1.11 f .73
+ .25
_.
4.21 .20
+ .14
f
t
*3.18.20
4.26 5.21
5.05
.9x
& .08
& 20
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
BAER IN LEARNING DISABLED CHILDREN
231
affected with girls having shorter latencies than boys. With right ear stimulation, only waves I1 (p = .015) and TI1 0, = .007) showed significant differences between sexes, girls having, shorter latencies than boys. When considered as present or absent, no risk effect was observed. When the cumulative score was taken as covariable great interactions between sex and risk were observed with left ear stimulation. Therefore, risk effect was separately evaluated in girls and in boys. In females, the latency of wave I11 when stimulating the left ear had a positive (correlation with the risk score (Y = .44). Responses to the right ear stimulation were not affected by risk. 3. Table 5 shows the results of multiple linear regression, where R-square is the percentage of the explained variance and the t the significant values of Student’s test. When all data were considered, sex was the strongest independent variable, confiming previously-mentioned results. The variable “sex” explained, at most, 14% of the variation in the latencies. Girls had shorter latencies than boys (negative values). The presence of risk factors produced larger wave V latency and I-V interval and smaller V/I amplitude ratio for the left ear stimulation. The R-square values shown in Table 6 wcre lower than those obtained when only the control group was analyzed. As the values of R-square for the total model indicated that at most 21% of the variance was explained, we consider that the multiple regression model was not optimal. 4. The princl pal component analyses (PCA) of all parameters (latencies, intervals and V/I ratios for both left- and right-ear stimulation) for all subjects showed that from a total of 18 variables, 17 were significantly different. The first 8 eigenvectors explained 90% of the variance. The first eigenvector (explaining 32% of the variance) was mainly correlated with wave V latency and I-V interval for both left- and right-ear stimulation; the second eigenvector (19’/0) with wave I latency for both left- and right-ear stimulation; third eigenvector (10%) was related only to variables when the left ear was stimulated: I1 latency and 11-V interval; the forth eigenvector (9%) was highly correlated with the latency of wave I when the left ear was stimulated and 11-V interval and VjI amplitude ratio of the response to right-ear stimulation; the fifth eigenvector (6%) was related to V/I amplitude ratio during left-ear stimulation and with interval I-IV during right ear stimulation. The sixth eigenvector (5%) was TABLE 5 Multiple Linear Regression (Only Dependent Variables with Significant Effects Are Shown) Dependent Variables
Lat. I1 Left Stirn. Lat. I1 Right Stim.
Risk
Sex
Multiple
R2
t
R2
t
R2
.13
- 3.53
-
-
.16
- 2.28
-
-
.09
-
.16
.06
Lat. 111 Left Stim.
.14
- 3.65
-
Lat. 111 Right Stim.
.08
- 2.69
-
-
.15
Lat. IV Left Stim.
.07
- 2.50
-
-
.10
-
2.27
.21 .ll
.09
2.12
.I5
.05
- 1.99
.10
Lat. V Left Stini.
.08
- 2.56
Lata. V Right Stirn.
.09
- 2.74
.06 -
I-V Interval Left Stim.
-
-
V/l Ratio Left Stirn.
-
-
E. MAROSI. T. HARMONY AN-D J . BECKER
238
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
related to \'/I amplitude ratio during left ear stimulation and to the latency of wave IV during right-ear stimulation. The seventh eigenvector (4%) was related to latencies of wave TV during left-ear stimulation and of wave I1 during right-ear stimulation. The eighth eigenvector (3%) was mainly related by latency of wave 111 to left-ear stimulation. These results show us multiple sources of variance and make interpretation difficult, since only waves V and I were explained by the same eigenvector when the left and right ear were stimulated; therefore, we calculated the principal components separTABLE 6 Principal Component Analysis of Latencies of Control Group to the Stimulation of the Left Ear Me;rsurenicnts
I II I11 I\'
1.atency Latency Latency Latency latency
V
Accun1.
"0
Eigenvector I
Eigenvector
Eigenvector 3
Eigenvector 4
Eigenvector
?
0.47 0.40 0.44 0.45 0.38 4x.57
-0.50 0.41 -- 0.55 0.22 0.37 67.11
0.17 0.26 0.36 0.50 - 0.72 82.97
0.48 0.47 -- 0.35 - 0.66 0.00 94.08
0.52 ~~055 - 0 49 0 25 0.34 100.00
E I gen \ ect o r
€1
-
>
Rotated \ alues ~
.21en < urernen t
I
Latency latencq Litency Latency latency
r II Ill IV
V
0.38 0.8 I 0.04 0.84 0.19
gen vec t or
E r gen vect or 7
7
0.80 -0.13 0.89 - 0.20 -0.13 .-
~
0.06 0.33 0.26 - 0.03 - 0.95 -
TABLE 7 Princip:il C'omponent Analysis o f Latencies of Control Group to the Stimulation of the Right Ear Measurements ______ Latency I latency 11 Latencq I l l Latency IV 1-atency L,'
Accum.
'
Rleasurenients
Eigenvcctor 0.32 0.46 0.47 0.49 0.47 4X.25
Eigenvector Eigenvector Eigenvector 2 3 4 ____~ ______ 0.03 0 . I3 0.11 0.06 - 0.72 0.09 0.14 0.50 - 0.70 0.1 I - 0.26 0.01 0.3 I 0.39 0.70 65.19 78.49 X9.37
Eigenvector I
-3
I
~~
Figenvector
l'itenc> I1 Latency 111
0.16 0.69 0 .I 3.
Latency IV Latenq V
0.7 5
0.98 0.20 -0.14 -0.18
0.7x
0.00
latcncq I
-
EigenLcctor 5
_____
o o2 0.51 0 1.1
0.82 0.70 IOO.00
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
BAER I N LEARNING DISABLED CHILDREN
239
ately for the different groups and ear stimulated, taking into consideration the latencies of waves I to V. The results were as follows: a) There were three different components in the control group during stimulation of the left ear ((Table6). Varimax rotation of these 3 components showed that eigenvector 1 accounted for the latencies of waves I1 and IV, eigenvector 2 explained the latencies of waves I and 111, and the latency of wave V was explained by eigenvector 3. These three components represented 83% of total variance. b) Two difirent components could be observed in the control group when stimulating the right ear, explaining 65% of total variance (Table 7). (The same results were obtained with and without Varimax rotation): eigenvector 2 explained the variance of wave I and the remaining latencies were explained by eigenvector 1. c) In the learning-disabled group, two different principal components were identified after stimulation of the left ear (Table 8), representing 73% of the total variance. The rotated eigenvector 1 explained all the latencies with the exception of wave I which was represented by eigenvector 2. Wave I1 was equally represented in both factors. d ) In the learning-disabled group, when stimulating the right ear, three independent components could be distinguished (Table 9), explaining 88% of the total variance. Rotated eigenvector 1 explained the variance of the latencies of waves IV and V. Eigenvector 2 represented the latency of wave I and eigenvector 3 explained the latencies of waves I1 and 111. 5. Multivariate test of Complete Homogeneity: As our main purpose was to try to find differences between the control and the LD groups, we also performed multivariate tes1.s of complete homogeneity which take into consideration not only the mean vector differences but also inequality of variance and of covariance matrices. Comparing the actual values of the latencies of the first five waves and the V/I amplitude ratio according to sex and ear stimulated, we obtained the following results: a) When the left ear was stimulated, we found highly significant differences between the control and LD boys (p = .OOOl). In the case of the girls, the same procedure did TABLE 8 Principal Component Analysis of Latencies of the Learning Disabled Group to the Stimulation of the Left Ear Measurements
Eigenvector 1
Latency I Latency 11 Latency 111 Latency IV Latency V Accum. %
Eigenvector 2
Eigenvector 3
0.10 0.44 0.54 0.52 0.49 49.07
- 0.82
0.45
- 0.32
- 0.67 - 0.07
Measurements
Eigenvector 1
Eigenvector 2
Latency Latency Latency Latency Latency
- 0. LO
-0.17 0.38 0.24 72.99
Rotated Values
I 11 111 IV
V
0.56 0.76 0.91 0.82
0.90 0.53 0.40
- 0.08 - 0.06
0.00 0.59 87.05
Eigenvector 4
Eigenvector 5
0.17 0.43 - 0.82 0.2 I 0.26 94.35
-0.31 0.27 0.06 - 0.74 0.54 100.00
E. MAROSI. T. HARMONY .4ND J . BECKER
240
TABLE 9 Principal Component Analysis of Latencies of the Learning Disabled Group to the Stimulation of the Right Ear Measurements
Eigenvector 1
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
Latency Latency Latency Latency Latency Accum.
I 11
I11 IV V O/O
Eigenvector 2
Eigenvector 3
Eigenvector 4 0.09
0.28 0.47 0.54 0.49 0.42 55.28
- 0.74
0.60
0.26 - 0.05 0.28 0.54 73.54
- 0.60
- 0.10
- 0.30
-0.12 0.79 - 0.59 96.03
Eigenvector I
Eigenvector 2
Eigenvector 3
0.09 0.13 0.43 0.78 0.91
0.98 0.13 0.16 0.16 0.01
-0.16 0.94 0.80 -0.35 -0.13
-
0.23 0.38 88.16
-
Eigenvector 5 0.03 0.59 - 0.78 0.10 0.20 100.00
Rotated Values Measurements Latency Latency Latency Latency Latency
I I1
IT1 IV V
-
-
not give significant results, although the probability of rejecting the null hypothesis was .18. b) When the right ear was stimulated, no significant differences were found between the control and learning-disabled boys. In the case of the girls, the difference between the control and the learning disabled group was significant at a p = .01 level. When the scores of the first eight principal components were compared between control and LD children according to sex, boys presented significant differences 0, = .0004), but girls did not 0, = ,117). DISCUSSION First we shall discuss the results obtained in relation to sex and risk, and later those on LD. Our results confirm the reports of many authors (Picton et al.. 1981; Mochizuki et al., 1983) in relation to sex differences, i.e., females displaying shorter latencies of all waves. In relation to risk antecedents we observed that in control subjects latency of wave I11 during stimulation of the left ear in girls had a positive correlation with the risk score. When the whole sample was considered, by multiple regression, we observed longer latency of wave V, longer I-V interval and smaller V/I amplitude ratio in the left ear in children with risk antecedents, which suggest the existence of BAER abnormality due to risk. Differences in BAER latencies and amplitudes between premature and high-risk neonates were observed by Salamy et al. (1985) before the age of two years. We consider our observation important given that many children for whom BAER study is recommended have antecedents of risk factors. As the trace left by risk on BAERs continues to be present for many years, it is important to have a complete clinical history in order to make more appropriate evaluations. In relation to LD. our results showed no evident differences. By univariate analysis we found two unexpected differences between control and LD girls, which we can only
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
BAER IN LEARNING DISABLED CHILDREN
24 1
explain as clue to chance, given that multiple comparisons were made. The absence of obvious differences agrees with the results obtained by Welsh et al. (1982) & Finley & Johnson (1983). These authors found amplitude differences in BAER between normal and LD children, but as we mentioned earlier, we had calibration problems and were unable to confirm their results. Application of multivariate statistics revealed clear, significant differences in the mean vector and variance and covariance matrices for left ear stimulation between control and LD boys and for right ear stimulation between control and LD girls. The comparisor of the first eight factor loadings taking into account both ears also demonstrates significant differences in boys. This result focused on boys might be due to the sample composition, since we had many more LD boys than LD girls. The differences observed between the covariance matrices are understandable in light of principal component analysis results. The detailed PCA of the wave latencies by ear and sex groups produced interesting results: Coritrol children had three differentiable components during left-ear stimulation and t u o when the right ear was stimulated. The opposite was observed in LD children. If it is assumed that a conductive system is more efficient with less relays, for control children the right ear looked more efficient, while for the LD the left ear was more efficient. A differential hemispheric involvement in reading of left- and right-ear dominant children was demonstrated by Bakker et al. in 1980. Primary-school children who showed left- or right-ear advantage subsequent to a dichotic listening task were required to name monosyllabic words presented at a fixation point. Authors found different strategies of reading depending on the ear preference observed. Those showing lef :-ear preference and supposed right hemispheric control were sensitive to the semantic and linguistic aspects of the test. Although we had not determined ear preference, the electrophysiological results suggest that this was the key to the differences observed between control and LD children. We want to emphasize that we have described such differences in a sample of children selected by clinical and electrophysiological criteria: most subjects had all peaks detecable and therefore a clear waveform composition. Our initial hypothesis of a brainst1:m dysfunction producing learning problems and BAER alterations was not confirmed, since the differences observed between normal and LD children were related to ear lateralization and therefore to hemispheric dominance, and no signs of brainstem abnormality were found.
REFERENCES Bakker, D. J., Licht, R., Kok, A., & Bouma, A. (1980). Cortical responses to word reading by right- and left-eared normal and reading disturbed children. Journal of Clinical Neuropsychology. 2; 1-12. Curts, J. B. (15184). Introduction al analisis de residuos en biologia. Bihtica, 9, 271-278. Eggermont, J. .I_ (1986). Evoked potentials as an indicators of the maturation of the auditory system in V. Gallai (Ed.) Maturation pf the CNS and evoked potentials, Elsevier, 177-182. Finley, W., & Johnson, G. (1983). Operant control of auditory brainstem potentials in man. International Journal of Neuroscience, 21, 161-170. Galaburda, A. M., Sherman, G. F., Rosen, G. D., Aboitiz, F., & Geschwind, N. (1985). Developmental dyslexia: four consecutive patients with cortical anomalies. Annals of Neurology, 18(2), 222-233. Galambos, R., Hecox, K. E. (1978). Clinical applications of the auditory brain sem response. Clincal Otolaryngology, North America 11, 709-718. Hinojosa, G., tt Rocha, C. (1985). Evaluacion de lectura y escritura en niliios de primaria. V. Coloquio Interno de Investigacion ENEP, Iztacala. Ken-Dror, A., Pratt, H., Zeltzer, P., Sujov, P., Katzir, J., & Benderley, A. (1987). Auditory BAEP to clicks at differen presentation rates: estimating maturation of pre-term and full term neonates. EEG and Clinical Ntwrophysiology, 68, 209-2 18.
Int J Neurosci Downloaded from informahealthcare.com by University of Otago on 12/27/14 For personal use only.
242
E. MAROSI, T. HARMONY A N D J . BECKER
Kileny. P.. Connelly. C., & Robertson, C. (1980). Auditory brainstem responses in perinatal asphyxia. International Journal of Pediatric Otorhinolaryngology, 2, 147-1 59. Mardia, K. V., Kent, J. 1.. & Bibby, J. M. (Eds.) (1979). Multivariate analysis London: Academic Press. Mason, S. M., Mellor, D. H . (1984). Brain stem middle latency and late cortical evoked potentials in children with speech and language disorders. EEG and Clinical Neurophysiologj. 59.. 297-309. Mochizuki. Y., Go. T., Ohkubo, H., & Motomura, T. (1983). Development of human brainstem auditory evoked potentials and gender differences from infants to young adults. Progress in Neurohiologj, 20. 273-285. Picton, T.. Stapells. D. R., & Campbell, K. B. (1981). Auditory evoked potentials from human cochlea and brainstem. Le Journal d' Oto-Rhino-Laryngologie Supl. 10. Rumsey, J . M., Grimes, M., Pikus, A. M., Duara, R., & Ismond, D. R. (1984). Auditory brainstem responses in pervasive developmental disorders. Biological Psychiatry, 19, L-17. Salamy, A., Eldredge, L.. & Wakeley, A. (1985). Maturation of contralateral brain stem responses in preterm infants. EEG and Clincal Neurophysiology. 62. 117-123. Sohmer, H., Student, M. (1978). Auditory nerve and BAER in normal, autistic, minimal brain dysfunction and psychomotor retarded children. EEG and Clincal Neurophysiology. 44, 380-388. Squires, N., Aine, C., Buchwald, J., Norman, R., & Galbraith, G. (1980). Auditory brain stem response abnormalities in severely and profoundly retarded adults. EEG and Clinical Neurophjvsiologj 50., 172-185. Starr. A. (1978). Sensory evoked potentials in clinical disorders of the nervous system. Anzericaiz Review qf'Neuroscience. 1, 103-127. Tanguay. P. E., Edwards, R. M., Buchwald, J., Schwafel, J., & Allen, V. (1982). Auditory brainstem evoked responses in autistic children. Archives of General Psychiatry. 39., 174-180. Trevarthen, C. (ed) (1974). Functional relationship of disconnected hemisphere with the brainstem and with each other: monkey and man. Hemispheric disconnection and cerebral function. Springfield. 187-207. Tukey, J. W. (ed). (1977). Exploratory data analysis. Addison-Wesley Publishing Co. Welsh, L. W., Welsh, J. J., Healy, M., & Cooper, B. (1982). Cortical, subcortical and brainstem dysfunction: a correlation in dyslexic children. Annals of Otology, Rhinology and Lar.vngo1og.v. 91, 3 10-3 15.
