---

PREDICTIVE VALIDITY OF BEHAVIOURAL OBSERVATION AUDIOMETRY (BOA)

Z H Baig
Lecturer cum Speech Therapist; Topiwala National Medical College / BYL Nair Charitable Hospital, Mumbai Central, Mumbai 400 008.

---

A comparative study of the auditory brainstem response (ABR).

Estimated hearing sensitivity levels with that of behavioural observation audiometry (BOA) results is being reported for 1110 ears of 555 children, whose hearing sensitivity levels were unknown at the time of BOA and who had been referred to the department with inability or ability to hear and/or speak.

Correct diagnosis was made on 876 (79%) ears on both the tests, out of which 325 (29%) ears showed no hearing impairment.

The number of ears that could be identified without error was 387 (70%) and ears with one-degree error were 98 (18%). 150 (13%) ears were not identified on BOA and 84 (7.6%) ears were not found to be impaired on ABR and hence were marked false negative (-ve) and false positive (+ve) outcomes respectively.

Discrepancies in test results were analysed according to the degree of error giving reasons for the same.

Thus easy and simple administration through BOA designed for early assessment of hearing impairment in children (age 0 to 3 years) has been validated with ABR with sensitivity 79% and specificity 80% of BOA.

INTRODUCTION

One of the challenging and rewarding aspects of clinical audiology is evaluation of infants and young children.

Literature shows contradictory views such as -

For example, while Thompson and Folson (1981) described behavioural observation audiometry (BOA) as having limited application and could not reliably estimate auditory sensitivity.

Hayes and Pashley (1991) stated that BOA cannot provide diagnostic information, McCorwick (1994) believes that behavioural measures provide important quantitative and qualitative information regarding an infants hearing status.

Jerger and Hayes (1976) caution that simple behavioural observation of auditory behaviour in children can be misleading and ultimately leading to mismanagement. Therefore behavioural test results be cross-checked with ABR and immittance measures.

Most of the researchers, however agree that BOA in combination with biophysical measures may lead to development of an initial clinical impression of hearing status in infants.

Today the recognised goal for infants assessment is identification and rehabilitation including amplification by 6 months. (American Speech and Language Hearing Association - 1987).

While evaluation techniques are available that permit accurate behavioural assessment, a major challenge of paediatric audiology is to bring these procedures into routine use.

The author here has attempted to determine the sensitivity and specificity of BOA against auditory brainstem response (ABR) threshold testing, to see if BOA can be used for predicting the hearing loss.

A comparative study of the ABR estimated hearing sensitivity levels with that of behaviouralobservation results is being reported for 1110 ears of 555 children.

MATERIAL

Amongst the paediatric cases that attended our clinic, 555 children were tested for both BOA and ABR (age group 0 months to 3 years at the time of testing).

The data were collected over a period of three years from 1995 - 1998. By and large the two test were performed by different teams of clinicians, under the supervision of experienced and skilled clinicians.

TABLE 1 Age wise break up of 555 cases
	0-6 months	6-12 months	1-2 years	2-3 years	Total
Male	80	94	105	57	336
Female	46	58	72	43	219
					555

After taking into account the case history, parental observations regarding child’s behaviour and administering rapid developmental check list, case was taken up for BOA and subsequently ABR.

PROCEDURE

Behavioural assessment of auditory thresholds of infants is based on observation of overt responses to controlled auditory signals, the examiner observes changes in behaviours which are time-locked to auditory signal. (Jerger).

Accordingly the procedure follows -

Test is done in a quiet room preferably sound treated (with a sound level meter showing 40 dB (A) or less).

0 month to 5 month

Child is in lying down position preferably in light sleep state as the prestimulus state for neonates. Change in behaviour to the auditory signal is observed such as eye-blink, startle, cessation of activity, rudimentary head-turn.

6 month and 2 year

Distraction test procedure was administered as given by Ewing and Ewing (1944). Child is made to sit on the mother’s lap, facing the tester I. Tester I will take the child’s attention on the object, which is made to roll over the table. Tester II will present the stimuli through Paediatric audiometer, the distracting object is hidden by hand before the stimulus is presented. The expected response is the direct localisation towards the sound source.

2 year to 3 year

The child is also taken up to play therapy session for observation regarding his comprehension and expressive ability.

Accordingly the degree of hearing loss is predicted i.e. Estimation of threshold on BOA was made on following points.

1. Intensity level of the warble tone at which positive response was elicited comparing it with other stimuli such as speech sounds, noise makers.
2. Parental observations regarding child’s hearing status.
3. Mental age of the child on which depends,

   (a) the test procedure

   (b) the presentation level of the stimuli

   (c) expected response.

4. Observations made in play therapy session regarding child’s comprehensive and expressive ability.

Procedure for ABR

The auditory brainstem evoked response (ABR)

test was administered using the instrument dantec evomatic 4000XT. This instrument has unique facility of allowing the tester to ascertain the thresholds of the two ears concurrently using dual trigger mechanism. The ABR threshold was determined by one of the several testers using click stimuli presumably calibrated in dB p.e. SPL. The test was started at higher intensity level, either at 132,120,108,99 p.e. SPL, depending on the discretion of the tester concerned. The levels are then decreased in 12 or 24 dB steps till no replicable response is observed, criteria for response being the delectability of presumed Wave V. The estimated threshold is the midpoint of the level at which no response is observed and the next 12 dB higher level which elicited replicable / unambiguous response. Accordingly the expected pure tone average of frequencies 1000,2000,4000 Hz (PTA 2) in dBHL is taken from the table given below which is empirical findings based on the clinical data and is presently used in the department and which needs further validation.

RESULTS

The results of ABR are correlated with BOA given in the tabular form on the basis of abnormal and normal findings. 551 ears show abnormal findings with hearing impairment ranging from mild to profound on both the tests. 325 ears showed normal findings. 150 ears showed abnormal findings on ABR and normal findings on BOA. 84 ears showed abnormal findings on BOA and normal findings on ABR. The correct diagnosis or hit rate of 551 (50%) ears and correct rejection of 325 ears (29%). False positive (+ve) outcomes (i.e. those with presumably no hearing impairment but diagnosed as having hearing loss) are 84 ears (8%). False negative (-ve) outcomes (i.e. those with hearing impairment passed on BOA) are 150 ears (13.5%). Accordingly the specificity and sensitivity of BOA as given by the following definitions was found out.

Specificity : The proportion of individuals without the disorder who have negative (-ve) test for the disorder. Hence, Specificity = d/d+c = 325/409 = 80%.

Sensitivity : The proportion of individuals with a disorder who have a positive (+ve) test for the disorder. Hence, Sensitivity = a / a+b = 551 / 701 = 79%.

Even though these 551 ears have been correctly diagnosed as hearing impaired on BOA detailed analysis of these ears becomes necessary to determine the extent of agreement between the two tests as regards the exact degree of hearing loss.

TABLE 2
Click Thresholdin dB p.e.SPL	Predicated degree of loss	Expected PTA2 in DBHL	Range
0-45	Nil	=0-25	+10 dB
46-60	Mild	=26-40	-10 to + 15 dB
61-80	Moderate	=41-55	- 10 dB
81-90	Moderately severe	=56-70	- 5 dB to + 5 dB
91-100	Severe	=71-90	- 10 dB to + 5 dB
> 100	Profound	> 90	- 10 dB to + 5 dB

 

Table 3 1110 ears comparing BOA and ABR Findings
	ABR
		Abnormal	Normal	Total
B O A	Abnormal	551 (50%)	84(8%)	635(57%)
	Normal	150 (135.5%)	325 (29%)	475 (43%)
		701(63%)	409 (37%)	1110(100%)

 

TABLE 4 Analysis of 1110 ears
		Normal	Mild	Moderate	Moderately severe	Severe	Profound	Total
B O A	Normal	325	107	31	7	1	4	475
	Mild	18	17	—	4	1	7	40
	Moderate	21	28	29	11	3	9	101
	Moderately severe	8	8	4	5	3	6	34
	Severe	18	6	6	10	17	39	96
	Profound	19	13	8	2	3	319	364
	Total	409	179	78	39	28	377	1110

The degree of hearing loss on ABR and is compared with BOA and a detailed analysis is made. The diagonal passing through the matrix shows that these ears tally with the findings on both the tests.

E.g. 325 ears showed no hearing impairment on both the test.
17 ears shows mild hearing loss.
29 ears showed moderate hearing loss.
5 ears showed moderately severe hearing loss.
17 ears showed severe hearing loss.
319 ears showed profound hearing loss.

The number of ears on both sides of the diagonal show the disagreement between the two tests. Hence error analysis is made stating probable reasons for discrepancies.

TABLE 5 Error analysis given by Jerger et al (1974)
Degree of error	No. of ears
Nil	387 (70%)
1	98 (18%)
2	29 (5.2%)
3	24 (4.3%)
> 4	13 (2.3%)
	551 (100%)

In 39 ears the two test disagree by one category,

TABLE 6 98 (18%) ears show 1-degree error
BOA	ABR	No. of ears
Mild	—/ Moderate	—
Moderate	Mild / Moderately	39
Severe
Moderately severe	Moderate / Severe	07
Severe	Moderately Severe /	49
Profound
Profound	Severe / —	03 98 (18%)

BOA shows moderate hearing loss and ABR shows mild hearing loss i.e. ABR findings were better than BOA findings for 28 ears as against this ABR shows moderately severe hearing loss i.e. ABR findings are worse than BOA findings in 11 ears.

Probable reasons for the discrepancies in these 98 ears

ABR click level calibration and accuracy of prediction criteria. Intensity of the stimuli is not monitored while administering BOA, since the distance of the paediatric audiometer from the child’s ear is not maintained. However, first degree error is the least serious / important from the clinical point of view.

A. Errors in the 16 ears where ABR is better than BOA can be attributed to age of the child and mental retardation.

TABLE 7 29 ears (2.6%) show 2 error
BOA	ABR		Total no. of ears
	Better	Worse
Mild	—	4	4
Moderate	—	3	3
Moderately severe	8	6	14
Severe	6	—	6
Profound	2	—	2
	16	13	29

B. For 13 ears where ABR is worse than BOA, the reason can be better ear taking part while doing BOA, low frequencies could be better, intensity not monitored while administering BOA.

TABLE 8 24 ears (4.3%) show 3rd degree error
BOA	ABR	No. of ears
Mild	Severe	1
Moderate	Profound	9
	Total	10

Reasons for discrepency are same as in ‘B’.

ABR	BOA	No. of ears
Mild	Severe	6
Moderate	Profound	8
	Total	14

Most of These Cases had Multiple Problems

e.g. Mental retardation, cerebral palsy, neuroregression.

If the diagnosis for these 13 ears was considered to be correct on BOA, may be attempted to fit a powerful hearing aid which could be an error on the part of audiologist. Discrepancies were seen in neuroregression cases and patients on anticonvulsive drugs.

The reasons could be same as those discussed at

TABLE 9 13 ears (2.3%) serious degree error BOA ABR No. of ears Profound Mild 13 Mild Profound —   Total 13

 

TABLE 10 Analysis of false negative (-ve) outcomes
BOA	ABR	Degree of error	No. of ears
Normal	Mild	1	107
Normal	Moderate	2	31
Normal	Moderately	3	07
	Severe
Normal	Severe	Serious	01
Normal	Profound	Serious	04
			Total :150

‘B’ under Table 7. High frequency hearing losses show good hearing sensitivity at low frequency and hence not detected on ABR due to its limitations. Also, BOA being a sound field procedure better ear participation takes place which may lead to misdiagnoses of the worse ear but this is not serious from the clinical point of view.

TABLE 11 Analysis of false positive (+ve) outcomes
ABR	BOA no. of ears	Degree of error	No. of ears
Normal	Mild	1	18 (25%)
Normal	Moderate	2	21 (25%)
Normal	Moderately severe	3	08 (9.5%)
Normal	Severe	Serious	18 (44%)
Normal	Profound	Serious	19
			84

Most of the above mentioned cases show complications such as mental retardation, cerebral palsy, neuroregression, deep sleep as the prestimulus state of the child.

CONCLUSION

From the current study it has been observed that:

Although diagnostic audiology evaluation of infants and young children requires a test battery approach, BOA if done under optimum conditions such as skills in child handling, timely observations of responses and clinical decision making can lead to valid audiological evaluation.

BOA to some extent can be frequency specific test procedure.

Profound hearing loss cases can be effectively ruled out on BOA irrespective of age and associated problems, same holds true for cases with no significant hearing loss.

False -ve type errors which are clinically most serious accounted for only 150 (13.2%) of 1110 ears. Even among this 3rd and higher degree error were confined to only 14 ears (< 2%). It is possible that at least in some of these cases BOA might have been more valid rather than ABR.

BOA is a comfortable, inexpensive and short duration procedure.

REFERENCES

ASHA 1982.

Jerger, Hayes. The crosscheck principle in paediatric audiology. Archives of otolaryngology 1976; 102 : 614-20.

Jerger, Pilips Barney, Larry Mauldin, Betsy Crump. Predicting hearing loss from acoustic reflex. Journal of Speech and Hearing Disorders 1974.

Jagdeesh, Sanghi. Validity of ABR click threshold. ISHA Journal.

Jerger. Paediatric audiology.

McCorwick. Paediatric Audiology 0-5 yrs. 1994.

Northern, Downs. Hearing in children.

Thompson, Folson. Hearing assessment of at risk infants. Clinical Paediatrics 1981; 20,257,261.