An Auditory Training Program for the Recognition of Innocent and Pathological Heart Murmurs
Rachel Caissie PhD
Halifax, Nova Scotia
John Finley MD CM
Halifax, Nova Scotia
Pam Nicol RN, BSc, MPH
University of Western Australia
Perth, Western Australia
This chapter examines the effectiveness of a novel heart murmur recognition program that was designed using auditory training techniques borrowed from audiological rehabilitation principles and practices. During the initial trial of the training program, it was administered to both medical and non-medical students to help teach the auditory distinction between innocent and pathological heart murmurs. Preliminary data obtained pre- and post-auditory training showed a significant improvement in the non-medical students’ ability to distinguish murmurs immediately post-training. Subsequent trials with Australian and Canadian medical students showed similar improvement. However, that improved performance declined two months after training. One year later a brief reinforcement program restored skills to the high levels achieved with initial training. Issues pertaining to the implementation of the program and the maintenance of skills over time are discussed.
A computer-assisted auditory training program to help teach auscultation of innocent and pathological heart murmurs was developed based on the principles of auditory training described in Chapter 5. The auditory training program teaches students how to distinguish between innocent and pathological heart murmurs as acoustic events; that is, cardiac diagnostic information is not provided with the training program, nor are students expected to diagnose the reason for the murmurs. Rather, the focus is on learning to distinguish between the acoustic characteristics of innocent versus pathological heart murmurs through hearing only. A second important feature of the program is repeated comparison of heart sounds, which are clearly normal, clearly abnormal, and those that share some features of both. The training program is based on the principle that the recognition of subtle differences between acoustic events can be improved with practice that involves listening to many occurrences of the target sounds.1-6 Similar to learning to play a musical instrument, over time one learns to “train one’s ears” to the musical notes after repeated exposure to the sounds. Auditory training improves the brain’s ability to recognize or identify sounds that have been detected by the peripheral auditory system7.
Description of the Auditory Training Program
The auditory training program developed for heart auscultation consists of several listening exercises (or listening trials) that automatically increase in difficulty based on the individual learners’ performance. The auditory training program employs a pool of 26 innocent and 30 pathological heart murmurs recorded from 56 patients. The heart sounds were recorded on children between the ages of several months and 18 years who were either normal or exhibited a variety of heart conditions. In order to construct auditory training exercises that varied in their level of difficulty, it was first necessary to group the recorded sounds according to whether they provided an easier listening task (dissimilar sounds) or a harder listening task (similar sounds). To achieve this, the 56 heart sounds were first classified into three categories: (1) typical innocent murmurs, (2) distinctly pathological murmurs (i.e., dissimilar sounds to target gross auditory discrimination), and (3) “atypical” pathological murmurs with some auditory features of innocent murmurs (i.e., similar sounds to target fine auditory discrimination). The computer-assisted program has three levels of training with multiple steps embedded within each level. The listening exercises become progressively more difficult with each training step, and one must achieve a certain correct performance criterion before proceeding to the next step. The software randomly selects murmurs from appropriate sound categories for the various training steps. Each sound recording is approximately 6 to 8 seconds in duration, and students can listen to a recording as often as they want to prior to providing a response. Feedback on performance is provided following each listening trial.
Level I of the training program employs a same-different sound discrimination task where the listener is required to pick the odd sound out of a set of four sounds. Four heart murmurs (recorded from four patients) appear on the computer screen; they consist of either three innocent murmurs and one pathological murmur, or three pathological murmurs and one innocent murmur. The program is designed so that, at first, the odd sound is quite different from the other three while the other three sounds resemble each other. The student’s task is to identify the one that does not belong (the odd sound). At this stage, the students are not required to identify the sound as either innocent or pathological murmur. Feedback is provided immediately after each answer, and the student continues trying as necessary until he or she is able to identify the odd sound in the set of four. Initially, Level I only utilizes sets involving pathological murmurs that sound very different from innocent murmurs in order to target gross sound discriminations (Step 1). Once the student is able to correctly identify the odd sound from six sets in a row, the auditory task increases in difficulty; that is atypical pathological murmurs that sound similar to innocent murmurs are used to target more challenging finer sound discriminations (Step 2). Once the students exhibit correct performance for six consecutive listening trials in Step 2, the software automatically brings them to Level II.
Level II of the training program uses a sound identification task with a closed-set response format (see Chapter 5 for more details on sound identification using closed versus open sets). As with Level I, four heart murmurs appear on the computer screen, with one of them being either innocent or pathological depending on the set. The students’ task is to either find the innocent murmur or the pathological murmur in each set. As in Level I, they are provided with immediate feedback after each listening trial, and if unsuccessful, they continue to try until they determine the correct answer for that set. Level II is comprised of three progressively more difficult steps, and six correct responses in a row are needed to be able to proceed to the next step. At first, only distinctly pathological murmurs that sound quite different from innocent murmurs are used (Step 3), and only one sound within each set needs to be identified (Steps 3 and 4). As students progress through the steps of Level II, they are presented with a mix of innocent murmurs, distinctly pathological murmurs, and atypical pathological murmurs, in each listening trial (Steps 4 and 5) and are required to identify each of the four sounds in the set as either an innocent murmur or a pathological murmur (Step 5). When they are unsuccessful in correctly identifying all four sounds in a set, the feedback provided will indicate which of the four sounds were incorrectly identified. This allows them to re-listen to a sound after an incorrect response while knowing what type of murmur it should be. In Step 5, once they can correctly identify, on their first attempt, each of the four sounds in six consecutive sets, in other words when they can successfully identify 24 murmurs in a row, they may proceed to Level III.
Level III of the auditory training program involves the identification of heart murmurs presented in isolation. That is, only one heart murmur appears on the screen at a time, and the student is asked to label it as innocent or pathological. This listening task does not represent a true open-set task as there are only two possible answers; therefore it is closer to a two-alternative forced-choice procedure. Nevertheless, the increased difficulty level arises from the fact that when making a judgment about a particular sound, the student no longer has access to other sounds, for comparison to the target sound, to facilitate judgment. Similar to previous training steps, first, only those pathological murmurs that sound markedly different from innocent murmurs are used (Step 6) before progressing to the identification of pathological murmurs that sound similar to innocent murmurs (Step 7). In this final step of the training program, students are required to identify heart murmurs presented in isolation and randomly derived from the entire pool of recorded sounds.
The auditory training program was first piloted on a group of 21 Canadian non-medical student volunteers, and then tested on a group of medical students at the University of Western Australia. Further testing was conducted on medical students at Dalhousie University. This chapter reviews of the results for the non-medical students and the University of Western Australia medical students. Results for the full group of medical students, including those from Dalhousie University, are described in Finley, Caissie, Nicol, and Hoyt13.
A cohort of 120 medical students from the University of Western Australia were invited to participate in the study and randomly assigned to either the intervention group or the control group. The non-medical students were between the ages of 18 to 34 years while the medical students’ ages ranged from 20 to 50 years. Prior to beginning the auditory training program, all participants underwent a hearing screening at 25 dB HL at audiometric frequencies between 250 Hz and 4000 Hz to ensure that hearing was within normal limits in each ear.
Students were asked to listen to the recorded heart sounds in a quiet room, using stethophones that were connected to a computer. A research assistant was available during auditory training so that guidance about the software could be provided when necessary. First, a brief demonstration of typical sound murmurs was provided by having the students listened to the recordings of an innocent and a pathological murmur. Then, in order to measure their pre-training skills, students performed a two-alternative force-choice identification task, where they listened to 20 heart murmurs and identified each one as either an innocent or a pathological murmur. Next, non-medical students and the medical students assigned to the intervention group completed the training program. Immediately after training, they performed a post-test of 20 recordings similar to the pre-training test. Another similar assessment was performed by the students two months after completing the training program. The medical students who were assigned to the control group were re-tested following two months of clinical activity in pediatrics but without structured heart sound teaching.
All pre- and post-training tests included 20 heart murmurs randomly selected by the software from the entire pool of heart murmurs, thus specific test items differed from test to test and from participant to participant. Immediately after completion of these tests, students were provided with their correct percentage scores.
In the non-medical group all 21 participants completed the full training program with the 2-month follow-up test. Of the 60 medical students randomized to the intervention group, 47 completed the training and the immediate post-training test, and 36 of these students completed the two-month follow-up test. Out of the remaining 13 students, 11 started the auditory training program but did not complete it, and 2 participants’ data were lost at the beginning of the study due to a server problem. Of the sixty medical students who were invited to participate in the control group, 54 completed the pre-test and 41 completed the post-test two months later. Wilcoxon Signed Rank tests were performed to determine if the pre- and post-training mean scores were significantly different. A level of significance of 0.05 was retained for the comparisons.
The non-medical students had a mean pre-training test score of 73% (95% CI=67, 78%; range=55-95%) and their immediate post-training test score improved to 92% (95% CI=88, 95%; range=70-100%). There was a significant difference between pre-training testing and post-training testing (p<0.001, 95% CI =13, 24%). After two months, the follow-up test score for non-medical students was 80% (95% CI=75, 85%; range=55-100%). The difference between the 2-month post-training and pre-training test score, (95% CI -0.05, 13.9%) was of borderline significance (p=0.052). The medical students' scores were: pre-training test, 77% (95% CI=74, 81%; range=45-100%); and immediate post-training test, 92% (95% CI=90, 94%, range 70-100%). This also represented a significant improvement between pre- and post-training tests (p<0.001, 95% CI =10, 18%). For medical students, the two-month follow-up score was 82% (95% CI=78, 86%; range=50-100%), which was a significant decrease from the immediate post-training test score (p<0.001) but a non-significant increase over the pre-training test score (p=0.109, 95% CI=-1.04, 9.93%). The control group who received no structured heart sounds teaching had scores of 78% (95% CI=75, 81%; range=50-95%) on the pre-test, and 81% at the two-month follow-up test (95% CI=78, 84%) which was a statistically significant difference from the pre-test score (p=0.04, 95% CI=0.11, 7.2). This follow- up score however did not differ from the two-month follow- up mean score for the study group who had taken the training test. Student reaction to the program was largely positive, with students agreeing that the instructions were clear; the program was interesting and taught a core skill which appeared to be transferable to clinical settings. At the beginning of the study, the students were asked about any previous musical training; however no relationship was found between scores and musical training. As they progressed through the auditory training program, students were required to listen to numerous repetitions of innocent and pathological heart murmurs. Exposure to a large number of repetitions of the target sounds is essential to facilitate auditory perceptual learning2,7. It is difficult to determine exactly how many repetitions of heart beats students listened to because they could re-play a sound if they needed to hear it again. An estimate of the minimum number of repetitions can be made as follows: (1) Each recording was a minimum of 6 seconds in duration, with an average of 8 heart beats per recording. (2) Exercises in Levels I and II included four recordings per listening trial, and Level 3 had one recording per listening trial. (3) Using the average number of listening trials needed to complete all three levels of the program, it may be estimated that students listened to at least 1230 repetitions of both normal and abnormal murmurs (See Appendix for calculation).
The training program took 40 to 150 minutes (an average of one hour) to complete depending on the individual’s auditory skills and previous experience with heart auscultation. Most students were able to complete the program in one or two training sessions. Medical education was not necessary to achieve success with the program, although non-medical students took longer to complete the program. They required an average of 108.4 trials compared to 88.8 trials for medical students, where a “trial” is an attempt at a screen with 4 murmurs (Levels I and II) or 1 murmur (Level III).
Results for the medical students from Dalhousie University showed trends similar to those observed with medical students from the University of Western Australia.13 In addition, long-term effects of the auditory training program were evaluated one year later for a subgroup of medical students from Dalhousie University.13 These results showed a further decline in performance compared to the two-month post training scores; however performance returned to a high level after a 15-minute refresher training program.
The initial results from our trial of this new auditory training program indicate considerable success with teaching murmur recognition. Over 90% accuracy is achieved with an average of one hour of training. This is a considerable improvement over the typical performance of medical students and residents8,9, and medical education is not necessary for success with the program. The high levels of murmur recognition are likely related to several factors: (1) the simple, practical question being asked: Is the murmur normal or abnormal?, (2) considerable repetition, and (3) repeated comparison of normal and abnormal sounds. These factors should be the foundation of learning programs related to heart sounds. All too frequently murmur recognition is made difficult for students by the presentation of too many types of sounds with little repetition and practice. Repetition has been shown to be crucial in auditory training2,7 and in the work of Barrett et al.10 in relation to heart sounds recognition. The practical skill required by the general physician is the ability to distinguish normal from abnormal, not necessarily to be able to identify large numbers of different murmurs.
It should be pointed out that chance performance level for our pre- and post-training tests is 50%, and pre-training test scores tended to be relatively high. For some students with high performance pre-training, there was limited room remaining to measure the full extent of any improvement. Moreover, a statistically significant improvement of less than 5%, such as that shown by the control group of medical students, is equivalent to 1 more correct response out of 20, which is not likely to be clinically relevant.
Both non-medical and medical students exhibited a decline in performance two months after completing the auditory training program. However, these post-training test scores should be interpreted with caution. Follow-up testing tended to occur two months into the school semester, often coinciding with heavier workloads such as during exam time. Indeed a number of students who had completed the pre-test and post-training test did not return for the follow-up test two months later due to a lack of time. Thus student fatigue or increased stress may have negatively affected follow-up test scores. Increased stress and heavy workload issues may also explain why 11 participants from the intervention group started the auditory training program but were not able to successfully complete it.
The decline in performance two months later may also have resulted from a lack of continued or regular exposure to heart sounds during everyday educational activities after training, especially in the case of non-medical students. Even for the medical students, there was very limited and unstructured exposure to heart sounds in the two-month interval before the follow-up post-training test. Especially since this auditory training program was their first encounter with these murmurs, it is perhaps not surprising that their recognition performance was not sustained. Continued exposure to the target sounds facilitates the maintenance of newly acquired auditory skills11. For example, in the rehabilitative audiology of adults with acquired hearing loss who are fitted with cochlear implants, the focus of auditory training may be on re-learning the recognition of speech sounds and meaningful environmental sounds (such as door bells, car horns, etc.) during structured listening exercises in the clinic, but learning to listen is also integrated with acoustic experiences that occur during everyday activities. Hence, the frequent exposure to speech and environmental sounds during any given day provides the cochlear implant user ample opportunity to continue practicing their auditory skills.
The reasons for the observed decline of performance two months post-training are currently under study. It appears that providing some form of reinforcement or continued exposure to heart sounds would be necessary to help with the maintenance of heart auscultation skills, especially for those students whose education program may not involve heart auscultation practice for extended periods of time. However, it is difficult to predict the optimal amount of exposure time that may be required for skills maintenance. For students who have limited opportunities to practice heart auscultation in the clinic, listening to recorded heart sounds for a few minutes a couple of times per week might be sufficient to facilitate skills maintenance. More research is needed to explore potential reasons for the decline post-training and the steps that may be taken to reduce it.
One might argue that any decline in performance post-training would rapidly reverse and the previously attained level be recaptured after a short refresher practice. This would be a scenario similar to the drop in performance that may be experienced by a musician who does not maintain regular practice, in that their previous performance level is rapidly regained after resuming practice. Preliminary results from a reinforcement program one year after initial auditory training suggest a return to the high post training levels after only fifteen minutes12,13. The auditory training program that was developed for medical students may also be used by physicians, nurses, and other health professionals who want to improve their auditory skills for heart auscultation, or want access to a practice tool for maintaining their current ability or regaining a previous performance level.
The data presented in this chapter provide support for further investigation of auditory training as an innovative teaching approach that could be integrated in the curricula of medical and other professional schools or used as an independent learning tool by both students and practitioners13. Early results from further study of the auditory training program in the Canadian context, at Dalhousie University, are similar to the Australian experience, despite differing curricula12,13.
The study outlined in this chapter was extended with a one year post-test and reinforcement12: 22 Canadian medical students who had performed the training were retested 1 year later on 20 random recordings including 10 additional novel items in a mastery-style reinforcement programme; any student scoring less than 90% took another 20-item test, and if that test score was less than 90%, the student took a final 20-item test. Each test provided feedback and implicit retraining.
The 1-year follow-up test median was 81% (55–100%) a significant decline from the 2-month post-test (P < .005). Only six out of 22 students achieved the 90% level at this test, but after first and second reinforcement tests, an additional six and two students, respectively, reached 90%. The median final score achieved after late reinforcement was 90% (70–100%), an improvement achieved after only 10-15 minutes.
Student reaction to the programme was largely positive, judging from a questionnaire, with most participants expressing satisfaction with skill improvement and even transferablility to patient assessment.
1. Sweetow R, Palmer C. Efficacy of individual auditory training in adults: A systematic review of the evidence. J Am Acad Audiol 2005;16:494-504
2. Sweetow R, Henderson-Sabes J. The need for and development of an adaptive listening and communication enhancement (LACETM) program. J Am Acad Audiol 2006;17:538-558
3. Fu QJ, Galvin JJ. Perceptual learning and auditory training in cochlear implant recipients. Trends Amplification 2007;11:193-205
4. Burk MH, Humes LE. Effects of long-term training on aided speech recognition performance in noise in older adults. J Speech Lang Hear Res 2008;51:759-771
5. Stacey PC, Raine CH, O’Donoghue GM, Tapper L, Twomey T, Summerfield AQ. Effectiveness of computer-based auditory training for adult users of cochlear implants. Int J Audiol 2010;49:347-356
6. Tremblay K. Training-related changes in the brain: Evidence from human auditory-evoked potentials. Semin Hear 2007;28:120-132
7. Sweetow R, Henderson-Sabes J. Auditory training. In: Montano J, Spitzer J, eds. Adult Audiologic Rehabilitation. San Diego, CA: Plural Publishing; 2009
8. Mangione S, Neiman LZ. Cardiac auscultatory skills of internal medicine and family practice trainees. A comparison of diagnostic proficiency. J Amer Med Assoc 1997;278:717-722
9. Gaskin PR, Owens SE, Talner NS, Sanders SPP, Li JS. Clinical auscultation skills in pediatric residents. Pediatrics 2000 Jun; 105(6):1184-7
10. Barrett MJ, Kuzma MA, Seto TC, Richards P, Mason D, Barrett DM, Gracely EJ. The power of repetition in mastering cardiac auscultation. Amer J Med 2006;119:73-75
11. Tye-Murray N. Foundations of aural rehabilitation: Children, adults, and their family members (2nd ed). Delmar Learning: Clifton Park, NY; 2004
12. Finley J, Caissie R, Hoyt B. 15 Minute Reinforcement Test Restores Murmur Recognition Skills in Medical Students. Canadian J Cardiology 2012; 28:S102
13. Finley J, Caissie R, Nicol P, Hoyt B. International Trial of Online Auditory Training Program for Distinguishing Innocent and Pathological Murmurs. In Press, J Pediat Child Health 2015:51(8); 815-819
There were an average of 72.8 tries in Level I and II and 16 tries for Level III. Assume each student listens to a minimum of 6 sec of each recording, with an average heart rate of 8 beats in 6 sec (averaged over 10 recordings).
There are 4 examples per screen for Levels I and II for each of the 72.8 tries so 4 murmurs are heard for each try. Each murmur is heard for 8 beats.
Thus the number of repetitions is 72.8 x 4 x 8 =2330 for Levels I and II and 16 x 1 x 8 = 128 for Level III. The total is 2458. There are 2 types of murmur, so the estimated minimum number of repetitions of each murmur is about 1230.