Using visual reinforcement audiometry in the assessment of hearing in infants

Abstract

We recall the referral from a fellow audiologist many years ago, when auditory brainstem response (ABR) testing was a relatively new procedure. He had seen an infant for ABR testing and asked if we would repeat the test for a second opinion. As we took down the information necessary to set up the appointment, we asked for the child's birth date and age. “Eight months,” he said. “How does the baby respond to sound?” The colleague replied, “I don't know. When he's been in our office, he's been asleep.” For us, this raised an important question: Is a hearing evaluation complete if the child has slept through the whole thing? The initial diagnosis of hearing loss in infants identified through newborn hearing screening programs is expected to occur by 3 months of age and thus relies heavily on electrophysiologic tests rather than behavioral tests.1–3 However, we contend that the evaluation is not complete unless some form of behavioral assessment is included in those early evaluations. BEHAVIORAL OBSERVATION AUDIOMETRY Prior to the advent of immittance measures, ABR, and OAE, audiologists had to rely on their observations of a young infant's responses to sound to estimate the audiogram. Called behavior observation audiometry (BOA), this procedure required an observer to judge if a behavioral response (i.e., a change of state such as a startle, eye widening, a grimace, cessation or initiation of sucking) was related to the presentation of an auditory stimulus. If the observed behavior was time-locked with the presentation, it implied that the infant was responding to the sound. Observers noted systematic behaviors that tended to occur in response to different stimulus types (speech, noisemakers, tones) and different stimulus levels at different ages.4–6 Responses ranged from arousal from sleep in the newborn period to rather precise localization of sound by 2 years of age. Younger infants required greater sound levels to elicit responses than did older infants. And, speech was the most effective signal at eliciting responses at low levels. These observations culminated in the Auditory Behavior Index.7 Illustrations from every edition of Hearing in Children by Northern and Downs8,9 have become the hallmark of the index. Copies of the figures and the index can be found taped to the walls of sound-treated test booths across the country. What's important to remember about the index is that it is a developmental index of responsiveness to sound, not sensitivity to sound. The levels included in the index are not norms for audiometric threshold. Numerous investigators have noted other limitations of BOA. First, babies with normal hearing range greatly in their responsiveness.6 For example, at 3 months of age, some normally hearing babies will respond at levels as low as 20 dB HL and others not until levels of 80 dB HL. This range of “normal” hearing effectively eliminates the possibility of detecting mild and moderate degrees of hearing loss (between 30 and 70 dB HL). Another problem is the wide range in responsiveness within individual babies, depending on state, alertness, and attention. Also, the responses are prone to habituation. That is, an infant may respond to the presentation of a novel stimulus, but subsequent presentations are less likely to evoke a response, especially at low levels.6 Another problem is that infants respond differently to different types of stimuli. Speech signals are the most evocative; they will typically elicit the most responses at the lowest levels. In order of likelihood to evoke infant responses are broad-band complex noises, narrow bands of noise, and pure tones.4,5 Thus, the most frequency-specific signals, those used for audiometry, especially at low stimulus levels (near threshold), are the ones least likely to result in responses from infants. A final limitation on the clinical use of BOA is that our judgment of responses tends to be colored by our expectations.10–12 If we know the stimulus level is high, we're more likely to expect, and thus to perceive, a baby's behavioral response. If we know the child has previously failed a hearing screening, we're less likely to expect a response. These expectations can result in underestimating or overestimating hearing loss. Many audiologists have abandoned the use of BOA in any formal sense.13–15 Others report good agreement between BOA and later audiograms.16,17 Those who do use BOA have added considerable, needed restrictions to the protocol. They require that the evaluation be done when the baby is hungry, awake, and quiet—a rare occurrence! They limit the acceptable response to one: only a sucking response, either cessation or initiation. They require that two observers agree on the responses, and that there must be three responses noted at one level for an interpretation of minimal response level (MRL). An alternative to BOA, which we prefer, is the use of functional scales, which can be completed by parents, early interventionists, speech-language pathologists, teachers of the deaf, early childhood specialists, and audiologists. Examples of scales that may supplement the electrophysiologic test results are the Developmental Index of Audition and Listening (DIAL)18 and the Infant-Toddler Meaningful Auditory Integration Scale (IT-MAIS).19 VISUAL REINFORCEMENT AUDIOMETRY When an infant has reached a developmental age of approximately 6 months, the behavioral evaluation may become the primary test in the battery. Visual reinforcement audiometry (VRA) can be used to obtain behavioral thresholds for pure-tone stimuli that approximate adult values.20,21 VRA capitalizes on a child's natural instinct to respond to auditory stimuli by turning toward the stimuli. When a child responds by turning, reinforcement is provided for that response with the momentary lighting and activation of a motorized toy.22,23 VRA is an operant-conditioned response. That is, the desired behavior is rewarded, thus increasing the likelihood that the behavior will continue. With VRA, we hope the head-turn response will continue until we have completed the audiogram. Conditioning the child There are many ways to optimize the chances of getting a complete, valid test. One of the most important is to adhere to the principles of operant conditioning. Typically in operant conditioning, there are conditioning or training trials in which the discriminative stimulus is paired with the reinforcement to shape the desired response. This means that initial conditioning or training must be done for a signal that is clearly perceptible/audible to the child. Because localization toward sounds, especially novel ones, is a natural response for most infants, conditioning often occurs rapidly. For normally hearing infants, conditioning may consist of a couple of trials in which the toy reinforcement is provided immediately upon spontaneous head turn in the direction of signals presented slightly above threshold (e.g., 30 dB HL). For other children, particularly those with hearing loss who lack experience with sound, the head-turn behavior associated with the sound signal may need to be shaped. For children with severe hearing loss, a vibratory signal (e.g., a 250-Hz or 500-Hz bone-conducted signal at 40 or 50 dB HL) may need to be used in conjunction with the acoustic signal so that the infant will associate the test stimulus with the reward. Once the child has demonstrated understanding of the task (i.e., test stimulus…head turn…reward), then stimulus levels can be decreased and threshold can be bracketed in standard audiometric fashion. Research has shown that once conditioning is established the MRL or threshold is not influenced by type of stimulus (e.g., complex signals vs. pure tones), as it is in when behavioral responsiveness is simply observed.24 Infants will respond at near-threshold levels for tones, as they will for noise bands and speech. Another important finding is that VRA does not require localization skills.25 Once conditioned, the infant continues to look to the reinforcement, regardless of the signal source. Thus, earphone testing, especially with light-weight insert earphones, can be quite successful.26,27 It is likely, however, that localization (and its counterpart under earphones, lateralization) may provide cues that aid in the ease of conditioning and testing, so it is wise to have reinforcement available for whichever side or ear is being tested.23,25 Effective reinforcement In operant conditioning, reinforcement is provided only for correct responses; that is, head turns that are time-locked to the stimulus presentations. Two conditions are required to ensure this: (1) stimulus duration should be consistent and short, and (2) the response must be unambiguous. We suggest a signal duration of 1 to 2 seconds, with a response window slightly longer than that (e.g., 4 seconds) to allow for responses to both onset and offset of the signal. Inexperienced examiners may tend to keep the signal on too long, as if its presence will eventually alert the child. Most assuredly, if one waits long enough the infant will turn during his normal wiggling-around behavior. The best way to ensure a clear-cut response is to place the reinforcement to the baby's side, where a 90-degree head turn will be required to view the toy. If the reinforcers are placed on top of the loudspeakers in the corners of the test booth, the audiologist must judge a 45-degree head turn of a child whose head and body are rarely stable between signal trials. Regarding the reward or reinforcement, there are other principles of operant conditioning to remember. One, we aim to provide positive reinforcement, so it is important to choose toys that are likely to please, not scare, children. The infamous monkey that bangs cymbals has ended more than one VRA session prematurely. Two, if the child's reward is to see the toys, let's put them at the child's eye level where they can be seen easily (again, not on top of a distant loudspeaker). Finally, several different toys, presented in random order, are likely to keep the child on task longer than a single toy. Two examiners recommended The optimal set-up for VRA involves two examiners. One examiner is at the audiometer in the control room presenting the test signals and activating the reinforcer when correct responses are observed. The other examiner is in the test booth with the parent and child, keeping the child centered and mildly distracted from the reinforcer. Usually this distraction is achieved by showing the baby a series of objects or toys that do not compete with the reinforcer (i.e., plastic rings, balls, puppets, etc. that can be manipulated quietly). A fairly large number of distracting items may be needed to keep the child's attention throughout a test session. The role of the in-room examiner is important in achieving a valid and reliable audiogram. It takes skill and practice to keep the child in what we assume is a listening posture, so that the child wants to see the reinforcer toy more than anything else while still allowing her/his attention to be drawn away so that a head turn can be easily judged. The examiner and the examiner's actions must not be more interesting than the reinforcer toys. The examiner should refrain from talking or smiling at the baby, except to provide praise for correct responses. A secondary role of the in-room examiner is to maintain a quiet environment and rapport with the parent, as well as to be available for quick insertion (and possibly re-insertion) of earphones. The in-room examiner's role is deemed so important that several investigators have refined procedures and equipment so that one examiner can do it all: Control the stimuli and reinforcers while located within the exam room with the child and parent.28–30 Behavioral testing of infants and young children requires efficient use of time, because the child's latest response may well be his or her last response. Therefore, for example, if conditioning doesn't occur quickly, change the stimulus (e.g., increase level, switch to speech or a different frequency) or change ears. Don't waste time above threshold (e.g., quickly descend in level using a 10-dB step size rather than the traditional 5 dB). For many infants, VRA thresholds can be obtained sequentially, first one frequency and then another, as in standard audiometry. Because thresholds for normally hearing infants tend to be 10 to 15 dB above adult values, some people prefer the term “minimal response level” to describe the results. Whatever term is used, it is especially important to note that the levels noted in the Auditory Behavior Index are not to be used as normative data for conditioned responses, such as those obtained with VRA. For example, VRA thresholds of normally hearing infants should be obtained at levels of 20 dB HL or better as young as 6 months. MRL or VRA thresholds of 30 to 40 dB HL indicate a hearing loss and will correlate with elevated ABR thresholds, abnormal tympanograms, or absent otoacoustic emissions. Until recently, there were no formal, well-defined VRA protocols. Widen et al.27 developed a fairly structured VRA protocol that was used successfully for the systematic hearing assessment of more than 3000 infants who were enrolled in a national, multi-site research project. This protocol may be modified for clinical use. Control trials One component of the protocol that we feel is important for routine clinical testing is the systematic inclusion of control trials (i.e., observation intervals in which the examiners judge whether a head turn occurs in the absence of sound stimulation). Control trials are to be conducted in precisely the same way that stimulus trials are when the child is appropriately attentive at midline. The responses during control trials are to be judged in the same way and for the same length of time as stimulus trials. The only difference is that reinforcement is not provided if the child turns. Control trials have several purposes. The first, of course, is to determine if the responses (head turns) you are judging are truly responses to the test stimuli and not just random motions. If the child turns as often during control trials as during stimulus trials, then your test is not valid. A second purpose of control trials is to provide feedback to the examiner in the test room about how effectively the child's attention is being maintained. Although adding control trials might seem to increase the test time, it appears that the critical factor in determining how long infants remain on task is the number of times they have seen the reinforcement toys. On average, 10-month-old infants will last 15–20 minutes for an average of 45 stimulus trials.27 There are a few practical strategies that may help delay habituation. Scheduling the appointment at a good time for the baby is one. Older infants (18 months and up) who often habituate more quickly than younger babies may respond well to variety; thus, changing test frequencies often or changing test ears may help keep the baby on task.31,32 Limiting the duration of the reinforcement or providing intermittent reinforcement, especially later in the test session, may be helpful.31,33 Giving the child a break, for a snack or playtime or even a diaper change or tympanometry, may yield a few extra responses in a subsequent test session.34 As infants approach 2 years of age, the test procedure of choice becomes a guessing game.35 Many toddlers are ready for more active involvement in the test procedure, such as tangible reinforcement operant conditioning audiometry (TROCA) or conditioned play audiometry.