Authors: Jennifer J. Lister (1Communications Sciences & Disorders, University of South Florida, Tampa, FL, USA), Elizabeth M. Hudak (2Department of Psychiatry and Behavioral Neurosciences, University of South Florida, Tampa, FL, USA), Ross Andel (3School of Aging Studies, University of South Florida, Tampa, FL, USA), Jerri D. Edwards (2Department of Psychiatry and Behavioral Neurosciences, University of South Florida, Tampa, FL, USA)
Categories: Article, Cognitive aging, Music listening, Cognitive interventions, Music therapy
Source: Journal of cognitive enhancement : towards the integration of theory and practice
Authors: Jennifer J. Lister, Elizabeth M. Hudak, Ross Andel, Jerri D. Edwards
Learning to play a musical instrument is commonly recommended to avoid cognitive decline and dementia, but experimental evidence is lacking. In this Keys to Staying Sharp study, we investigated the efficacy of piano training as compared to music listening instruction to improve auditory processing, cognition, and everyday function among older adults with and without mild cognitive impairment (MCI). Older adult participants with and without MCI (n=268) included 58% females; 16% identified as Black race and 8.2% reported Hispanic ethnicity. Education ranged from 11 to 20 years. Participants were randomly assigned to either piano training (n=133) or an active control group of music listening instruction (n=135). Group training sessions were led by an instructor for 90 min twice a week for 20 sessions. Measures of auditory processing (time compressed speech, words-in-noise, dichotic digits test, dichotic sentence identification, adaptive tests of temporal resolution), cognition (trail making test, digit coding, verbal fluency), and everyday function (timed instrumental activities of daily living, test of everyday attention) were administered at baseline and immediately post the intervention phase. Analyses were registered at Open Science Framework https://osf.io/sh27y/ on April 25, 2018. Relative to music listening, no significant effects of piano training on auditory processing, cognition, or everyday function were found (ps>0.265). Future research should continue to examine the connection of impaired auditory processing with subsequent dementia and investigate whether effectively enhancing auditory processing by intervention may reduce dementia risk.
Identifying effective interventions to prevent or delay cognitive decline and dementia is imperative to public health. Verghese et al. (2003) observed that older adults who frequently played musical instruments were 69% less likely to be diagnosed with dementia, spurring interest in music training as a potential cognitive intervention. Although learning to play a musical instrument has since been commonly recommended to stave off cognitive decline and dementia (Seidler et al., 2010; Wan & Schlaug, 2010), evidence from randomized clinical trials investigating the efficacy of these claims has only recently emerged. In this study, Keys to Staying Sharp (Hudak et al., 2019), we examined the efficacy of piano training to improve auditory processing, cognition, and everyday function among older adults with and without mild cognitive impairment (MCI).
Research has established that auditory processing is fundamentally linked with memory and executive functioning (e.g., Albers et al., 2015). Furthermore, auditory processing performance longitudinally predicts dementia (Gates, Beiser, Rees, D’Agostino, & Wolf, 2002). Auditory processing may be described as the ability to use supra-threshold sound (i.e., sound that is clearly audible and presented well above the softest sound a person can perceive) and includes understanding degraded speech (e.g., speech in background noise), binaural processing, and auditory processing speed. When auditory processing is impaired, information received by the cognitive systems is distorted and incomplete, which negatively affects cognition (Albers et al., 2015). Prior research indicates that adults with MCI show significant auditory processing deficits (Edwards, Lister, et al., 2017; Gates, Anderson, McCurry, Feeney, & Larson, 2011; Idrizbegovic et al., 2011). Although auditory processing is a strong longitudinal predictor of dementia (Gates, Anderson, Feeney, McCurry, & Larson, 2008; Gates et al., 2011; Gates et al., 2002; Gates et al., 1996), studies have not examined whether enhancing auditory processing through intervention improves the cognitive and functional abilities of older adults. A potential way to enhance auditory processing and cognition may be through music training.
Interestingly, research has shown that music experience is associated with better auditory processing (Kraus & Chandrasekaran, 2010; Merten et al., 2021; Mishra, Panda, & Herbert, 2014; Zendel & Alain, 2012). Correlational studies suggest that adults with formal music training have better auditory processing abilities as compared to non-musicians (e.g., Merten et al., 2021; O’Brien, Nikeh, & Lister, 2015; Parbery-Clark, Strait, Hittner, & Kraus, 2013). This potential positive effect of music training on the processing of supra-threshold sound is in contrast to the associated negative effect on threshold hearing due to noise-induced hearing loss (Potts, Liotti, Tucker, & Posner, 1996; Pourryaghoub, Mehrdad, & Pourhosein, 2017; Schnick, Kreutz, Busch, Pigeot, & Ahrens, 2014).
Merten and colleagues (2021) examined the longitudinal associations of music training, auditory processing (i.e., word recognition in competing message), cognition (i.e., Trail Making Test, Auditory Verbal Learning Test, Digit Symbol Substitution Test, and verbal fluency), and incident dementia in an epidemiological study among 2938 older adults. Results indicated that the relationships among music training, auditory processing, cognition, and dementia were significantly moderated by sex and education. As examples, female musicians showed less auditory processing decline over time than non-musicians, but effects were not significant after adjusting for education. Male musicians performed better overall on speech perception than non-musicians, but effects were not significant after adjusting for the male musicians’ better pure tone audiometric thresholds. Music training was not significantly associated with cognitive performance across time in either males or females. Finally, incidence of dementia such as Alzheimer’s disease was lower among male musicians, but effects were not significant after considering education. Thus, experimental evidence is needed to elucidate if music training enhances auditory processing or cognition or reduces dementia risk.
Systematic reviews and meta-analyses have further examined the effects of various types of music interventions on global measures of cognitive status (Dorris, Neely, Terhorst, VonVille, & Rodakowski, 2021; Li, Wang, Fan-Hao, & Chen, 2015). Li et al. (2015) quantified evidence across 5 randomized trials involving a total of 234 participants including both those with and without cognitive impairment or dementia. The music intervention approaches varied widely (e.g., singing in groups, playing a musical instrument, music care, music-stimulated activity, or receptive music therapy such as music listening) and were categorized as either active (e.g., playing an instrument) or passive (e.g., music listening). Results indicated no overall significant improvements on global cognition either across all types of music interventions or among only the active music interventions. On the other hand, Dorris et al. (2021) showed different results when quantifying effects of only active music interventions on cognition among older adults with MCI or dementia. A meta-analysis of 9 studies including 495 participants indicated a significant small effect (d=0.30) of active music interventions relative to controls to improve global cognitive status. Limitations of these meta-analytic findings are that global cognition is an insensitive outcome measure, and the effects of music training may vary for healthy older adults as compared to those with MCI or dementia. Neither of these reviews examined the effects of music interventions on auditory processing.
The effects of piano training on cognitive performance of older adults has been studied in prior research (Bugos, 2010; Bugos, Perlstein, McCrae, Brophy, & Bedenbaugh, 2007; Seinfeld, Figueroa, Ortiz-Gil, & Sanchez-Vives, 2013). For example, Bugos et al. (2007) investigated the effects of 6 months of individual piano training (involving 30 min of instruction and 3 h of practice weekly) on cognition among 31 older adults. Participants were not screened for MCI, but those reporting dementia were excluded. Results among those adherent to complete the piano training regimen indicated significant group × time interactions for Digit Symbol and Trails B performance. Thus, piano training improved some aspects of cognitive performance immediately post-training relative to no contact controls. The authors indicated that 3 months after piano training ended, the cognitive improvements were not maintained.
In another study, Bugos (2010) compared older adults without dementia (n=70) who were matched on age, education, and intelligence and assigned to 16, 45-min sessions of either group piano training or music listening instruction combined with 15 min of socialization. Analyses compared pre- to post-performance across cognitive measures of verbal fluency, processing speed, and cognitive control among 46 participants who were adherent. No significant group × time interactions were evident indicating no differential benefits of group piano training to improve cognition as compared to music listening.
Seinfeld et al. (2013) compared older adults (n=41) without signs of MCI or dementia who completed either 4 months of weekly piano training lessons with 45 min of daily practice or various leisure activities. Random assignment was not applied, but groups were matched on age and education. Data from 13 participants completing piano training and 16 participants completing leisure activities were analyzed. Significant improvements within the piano training group were reported on one measure of cognitive performance (i.e., Stroop test), but no improvements were found on several other cognitive assessments. Group × time interactions were not reported; so it is not clear if there were differential benefits of piano training as compared to leisure activities.
Bugos and Wang (2022) investigated the effects of piano training relative to either computerized cognitive training targeting auditory processing and working memory (i.e., Posit Science Brain Fitness) or no contact controls among 155 older adults without signs of dementia (Montreal Cognitive Assessment (MoCA) scores >19). The training conditions included 16 weeks of classes meeting twice a week for 90 min. Participants were instructed not to practice outside of class. Analyses were conducted among 115 participants including the no contact controls and those and who were adherent to complete 20 or more sessions of either piano or cognitive training. The authors reported significant effects of piano training to improve verbal fluency relative to either cognitive training or controls as indicated by significant group × time interactions. No significant improvements subsequent to piano training were evident on other measures of cognition as compared to either the cognitive training or no contact control groups.
In summary, the aforementioned piano training studies conducted analyses among only those who were adherent. Although beneficial effects for those completing piano training were found on a few cognitive tests as compared to either no contact controls, cognitive training, or leisure activities, no significant benefits of piano training were evident relative to music listening. Only Seinfeld and colleagues excluded those with MCI and focused on cognitively healthy older adults, but this was not a randomized trial. None of these studies examined the effects of piano training specifically among those with MCI as compared to those who were cognitively healthy, and none examined the effects of piano training on auditory processing or everyday functioning.
With regard to the effects of piano training on auditory processing, Worschech et al. (2021) examined older adults without dementia (n=156) who were randomized to either a piano playing or musical culture course stratified by age, sex, and cognitive status. Participants completed either piano training in dyads or music culture classes in small groups for 60-min weekly sessions across 6 months and were asked to complete 30 min of homework each day. Auditory processing was assessed by a speech in noise test, and although there were significant improvements across time, no significant group × time interactions were evident. Piano training was not more efficacious than the music course to improve auditory processing. The authors concluded that both the piano playing and music culture course groups showed improved speech in noise perception.
Although prior studies indicate the potential efficacy of piano training to improve auditory processing and cognition, it remains unclear if piano training is more beneficial than other types of music instruction. None of the piano training studies reviewed examined effects on everyday function, and the effects of piano training among persons with clinically diagnosed MCI, who are at higher risk for dementia, have not been delineated. The primary objective of this Keys to Staying Sharp study was to examine the efficacy of piano training to improve auditory processing, cognition, and everyday function among older adults with and without MCI in a rigorous, randomized clinical trial. We further examined the moderating effects of MCI on piano training efficacy. We hypothesized that piano training would enhance auditory processing, which would in turn lead to improved cognition and everyday function.
Community dwelling older adults (n=366) completed an in-person screening visit to determine eligibility; they were, on average, 69.79 (SD=5.83) years of age, with an average of 15.79 (SD=2.37) years of education, and included 58% females, with 71% identifying as White race and 8.7% indicating Hispanic ethnicity. Participants were primarily recruited through our participant registry, mass mailings, newspaper advertisements, and community events. The study took place at the University of South Florida Cognitive Aging Lab and the Music Research and Testing Lab. Ethical rules of the Declaration of Helsinki, including obtaining approval from the University of South Florida Institutional Review Board, were followed. Participants freely provided informed consent including permission to publish deidentified study results.
Inclusion and exclusion criteria for participation are summarized here and are detailed elsewhere (Hudak et al., 2019). These criteria were chosen based on prior research and clinician input with the goal to enroll those capable of benefitting from the intervention who were novice to music training. As shown in Fig. 1, 98 participants were not randomized; 23 refused further participation and 75 were deemed ineligible. The number in parentheses after each of the following inclusion/exclusion criterion indicates the number deemed ineligible by reason. Inclusion criteria were age 60 or older (n=0), willing, available, and capable of completing study procedures (n=16), English speaking (n=0), MoCA score >19 (n=24), near visual acuity of 20/50 or better (*n=*0), pure tone hearing thresholds of <70 dB HL at 1000 and 2000 Hz in at least one ear (*n=*1), and a Music Reading Assessment score of 18 or lower (*n=*7). Those with Geriatric Depression Scale short form (GDS) scores greater than 4 (n=12), who had difficulty with hand/finger movements (n=7), or were planning to undergo medical procedures such as anesthesia, chemotherapy, or radiation (n=2) were excluded. Persons who self-reported neurological disorders (n=3) or enrollment in another research study (n=1) were excluded. Other criteria considered participants’ previous experiences in music such that those with four or more years of formal music training or who were actively involved in music training were excluded (n=1). Finally, those who reported completing 10 or more hours of cognitive training (i.e., a computerized cognitive training program involving targeted exercises designed to improve specific cognitive abilities) were excluded (n=1).
Two hundred and sixty-eight participants were randomized to piano training (n=133) or an active control group of music listening instruction (n=135). The randomized sample (n=268) was, on average, 70 years of age (median=69, range 60–88, SD=5.73), included 58% females, and 8.2% reported Hispanic ethnicity. Most randomized participants identified as white (76%), 16% identified as Black, and 2% identified as Asian. Race was recoded into white versus other for subsequent analyses. Education ranged from 11 to 20 years with an average of 15 years (SD=2.20). See Table 1 for the characteristics of participants with and without MCI by randomized group who were included in analyses.
We administered inclusion and exclusion measures of near visual acuity, hearing, cognitive status, music reading, and depressive symptoms according to standard procedures; these measures were also considered covariates for analyses. Near visual acuity was measured at 40 cm using standard procedures (Good-Lite, 2010). Hearing and speech recognition thresholds were assessed for each ear using standard procedures (American Speech-Language-Hearing Association, 2005). Pure tone air conduction thresholds were measured and pure-tone averages (PTAs; average threshold at 500, 1000, and 2000 Hz) were calculated for each ear. The MoCA was used to examine participants’ cognitive status (Nasreddine et al., 2005). The short form GDS was used to screen depressive symptoms (Sheikh & Yesavage, 1986). To ensure that prospective participants were musically naïve, only participants scoring 18 or lower on the Music Reading Assessment were included (Bugos & Groner, 2009).
The Neuropsych Influences on Cognitive Training Questionnaire (NICT, Rabipour & Davidson, 2015) was adapted to refer to the randomized groups (i.e., piano training or music listening) rather than “cognitive training.” The questionnaire measures participants’ expectations for the assigned condition to improve cognition, memory, concentration, distractibility, multi-tasking, and everyday abilities. Ratings are on a 7-point scale ranging from completely unsuccessful (1) to completely successful (7). Answers were averaged for a total score. The NICT was administered at session 17 of training for most participants. If participants had not yet completed the NICT, it was administered at their final study visit prior to any other assessments.
We chose standard outcome measures of auditory processing, cognition, and everyday function used in our pilot studies, prior research, and/or clinical practice with evidence of reliability and validity. We aimed to include indices of auditory processing across the domains of auditory processing speed, degraded speech understanding, and binaural speech processing. Cognitive outcomes were chosen to tap speed of processing and executive function. Our everyday functional outcomes were chosen based on sensitivity to age-related declines in speed of processing and executive function. Alternate forms were used at post-test for all applicable measures. Please see Hudak et al. (2019) for additional detail on the measures.
Five measures were administered binaurally to examine auditory Time Compressed Speech (TCS), Words-in-Noise (WIN), Dichotic Digits Test (DDT), Dichotic Sentence Identification (DSI), and the Adaptive Tests of Temporal Resolution (ATTR). The TCS (Wilson, Salomon, Sperry, & Bornstein, 1994) is a word recognition task for which the timing is manipulated to resemble rapid speech. A total of 50 words are presented with 65% compression applied and percent correct is calculated. WIN is a word recognition task for which two, 35-word lists are presented in a background of multi-talker babble, and the participant is asked to repeat the word that was heard. The signal-to-noise ratio (SNR) for 50% accuracy is calculated (Wilson, Abrams, & Pillion, 2003). For DDT, 25 sets of four monosyllabic numbers between one and 10 are presented to the right and left ear simultaneously, the participant is asked to repeat all four numbers heard from the closed set, and percent correct is calculated (Strouse & Hall, 1995). For DSI, 30 pairs of nonsense sentences are presented one sentence is presented to one ear and a different sentence is presented to the contralateral ear, the participant is asked to identify which two sentences were heard from a list of six sentences, and percent correct is calculated (Fifer, Jerger, Berlin, Tobey, & Campbell, 1983). The adaptive tests of temporal resolution (ATTR; Lister, Roberts, Shackelford, & Rogers, 2006) measured 70.7% correct gap detection thresholds (in ms) for within- and across-channel subtests. Participants are asked to select which of three sound intervals contained a silent gap.
Cognition was measured with Trail Making (Reitan & Wolfson, 1985), Digit Symbol Coding, and verbal fluency tests (Delis, Kaplan, & Kramer, 2001). In Trails A, participants draw a line sequentially between numeric stimuli, and in Trails B, a similar task requires alternation between visual alphabetic and numeric stimuli. Completion times were analyzed. In Digit Symbol Coding, the numbers 1–9 are each paired with a unique geometric symbol in a reference key at the top of the test form. To learn and practice the task, participants are first asked to complete 10 sample numbered boxes by writing the correct symbol associated with the number from the key in the box. On the same page as the key, 135 boxes are associated with the numbers 1–9 in random order. The participant is given 120 s to write the correct geometric symbol associated with the number in the reference key into each of the 135 boxes. The number correct is recorded. Alternate forms were not available for Trails or Digit Symbol Coding. Verbal fluency was measured in 60-s trials to evaluate letter, category, and category switching according to standard procedures. Participants are asked to say as many words as possible that begin with a specific letter that fit within a specific category, or to alternate providing words that fit within two different categories.
Two measures examined everyday the Timed Instrumental Activities of Daily Living Test (Timed IADL; Owsley, Sloane, McGwin, & Ball, 2002) and the Test of Everyday Attention (TEA; Robertson, Ward, Ridgeway, & Nimmo-Smith, 1996). Timed IADL includes five tasks in the domains of communication (finding a telephone number in a phone book), finance (counting change), cooking (reading the first three ingredients listed on a can of food), shopping (locating two food items on a shelf), and medication management (reading the directions printed on the label of a medicine bottle) using real-world stimuli. Performance is summarized by time and accuracy. The TEA consists of eight subtests designed to measure attention and mimic everyday tasks that include visual and auditory domains. We analyzed data from four of the eight TEA subtests that simulate everyday visual elevator (participants count the number of “floors” up and down using a series of visually presented elevator doors and arrows), elevator counting with reversal (participant count up and down floors using visually presented elevator doors and auditory tones to instruct whether to go up or down floors), telephone search (participants identify symbols on a telephone directory page), and telephone search while counting (participants search a telephone directory and count strings of tones at the same time). Performance is summarized by scaled scores.
The intervention and control groups completed manualized courses led by instructors with a music performance background. The randomized conditions were equivalent in terms of frequency and duration of each session. Classes were conducted in groups of up to 10 persons. Both conditions were described to the study participants as “music training”.
The intervention involved piano training using the Alfred Basic All-in-One Method and consisted of learning basic piano technique, completing finger dexterity exercises, playing music pieces, and learning music theory. At each bi-weekly training session, participants were required to perform technique such as scales and finger dexterity exercises, to play music on the piano (i.e., piano repertoire), and to complete music theory assignments. Each class session was structured with a short review followed by instruction to acquire new piano skills (e.g., scales, chord progressions) and to learn music concepts (e.g., intervals).
The active control condition consisted of music listening instruction using the text, Music Listening Today. The participants read about, listened to, discussed, viewed diagrams of, and answered questions about music. Lessons included information about composition styles, composers, as well as organization and structure of melodic compositions.
Please see Hudak et al. (2019) for details on the protocol. This study was a parallel, two-arm, randomized clinical trial (ClinicalTrials.gov identifier NCT03528486 May 17, 2018) conducted by blinded assessors. Participants provided informed consent and completed assessments to determine initial eligibility. Interested individuals were telephone screened and those enrolled completed baseline and post-test assessments of auditory processing, cognition, and everyday function per standard procedures in a specified order. Testing visits did not exceed 2.5 h. Recruitment and enrollment began in January 2018 and continued through December 2020, with data collection completed in April 2021.
Participants who scored <26 on the MoCA (Nasreddine et al., 2005) were further assessed for MCI at a clinical evaluation by a physician and completed the National Alzheimer’s Coordinating Center Clinical Dementia Rating scale and neuropsychological battery. Informants were interviewed, if available. Criteria outlined by the Alzheimer’s Association/National Institute on Aging guidelines were applied for MCI diagnosis.
Eligible participants with (n=75) and without MCI (n=193) were randomized stratified by MCI status with a 1 allocation to either piano training or music listening instruction. The study coordinator applied statistician-generated random allocations to assign participants while investigators and assessors remained blinded. Classes were conducted in groups that met twice a week, for 90 min per class, across 10 weeks. Immediately after the intervention phase (M=20, SD=13 days), participants completed a post-test visit to again complete assessments of auditory processing, cognition, and everyday function. Following the principle of intent-to-treat, all participants were encouraged and invited to complete post-testing regardless of intervention adherence.
Statistical analyses were pre-registered at Open Science Framework https://osf.io/sh27y/. Statistical power analysis (Faul, Erdfelder, Lang, & Buchner, 2007) informed by prior piano training research (e.g., Bugos & Kochar, 2017; Bugos et al., 2007) indicated that to detect a small effect size for a group × time interaction (f=.010, which is roughly equivalent to Cohen’s d=0.20), a final n of 198 participants (n=99 per arm) would achieve 95% power with an alpha of 0.05. Power analyses further indicated that to examine the moderating effects of MCI on intervention efficacy, a final n of 168 would have 0.95 power at alpha 0.05 to detect medium effects (f=0.25, roughly equivalent to Cohen’s d=0.50) for a significant group-×-time by moderator three-way interaction.
We conducted chi-square and one-way ANOVA analyses to compare the randomized groups at baseline on age, race, education, sex, MoCA, GDS, hearing, and vision. Variables with marginally significant (p<0.1) differences were included as covariates in subsequent analyses. Data reduction by principal component factor analysis was used to inform creation of composite outcomes of auditory processing, cognition, and everyday function for analyses.
Mixed effects models were conducted in SAS (SAS Institute, Cary, NC) procedure MIXED to examine change in performance from baseline to post-test by randomized group. Mixed effects models are an advanced form of repeated measures ANOVA that account for random effects (i.e., variance in individual baseline values and change in scores) and thus tend to provide more accurate estimates. We report effects for baseline scores (intercept), the overall rate of change (time), the estimate for the difference between baseline values for the two groups, and the group-×-time interaction, which is indicative of changes in performance attributable to intervention. Primary analyses included randomized participants and secondary analyses were conducted among those who were adherent (i.e., completed >15 sessions). Secondary analyses were also conducted including MCI, group-×-MCI, time-×-MCI, and group-×-time-×-MCI interactions to examine if intervention effects varied by cognitive status. Our funding agency requires analyses to be stratified by sex. Furthermore, published findings since our registration indicated that the relationships among music training, auditory processing, and cognition were moderated by sex (i.e., Merten et al., 2021). Thus, we also conducted sensitivity analyses examining if effects varied by sex. Additional sensitivity analyses were added as suggested by reviewers and are detailed below.
Of the 268 participants randomized, 216 returned for the post-test visit; 76% of the randomized participants completed post-test. Per the study protocol, 19 participants were excluded from primary analyses due to a head injury (n=1), undergoing anesthesia (n=10), or hospitalization (n=8) between baseline and post-test. Please see Fig. 1 for details.
The randomized groups did not differ in sex, χ^2^(1)=0.787, p=0.375, ethnicity χ^2^(1)=0.167, p=0.683, or race, χ^2^(1)=0.045, p=0.832. The Music Reading Assessment scores did not differ between the two groups, t(266)=−0.958, p=0.338. The randomized groups did not significantly differ in age, F(1,266)=0.293, p=0.589, education, F(1,266)=0.369, p=0.544, MoCA, F(1,266)=0.427, p=0.514, or GDS, F(1,266)=0.943, p=0.332. However, there were marginally significant group differences for PTA in the left ear, F(1,266)=3.31, p=0.070 and statistically significant group differences for PTA in the right ear, F(1,266)=6.69, p=0.010, and visual acuity, F(1,266)=8.99, p=0.003. The music listening group tended to have worse hearing, while the piano training group tended to have worse visual acuity. See Table 1. The groups also differed significantly in their expectations regarding the potential benefits of their randomized group, F(1,214)=15.07, p<0.001. The piano training group had an average rating of 5.55 while the music listening group had an average rating of 4.99, indicating the piano training condition had higher (i.e., better) expectations about the potential effects of their randomized condition.
Principal component analysis with varimax rotation was conducted to inform our creation of composites; composites were used to reduce the number of variables for analyses (i.e., reduce the number of multiple comparisons). Please see Table 2. The first factor reflected cognitive speed of processing and included the Trail Making Test, Digit Coding, and the TEA visual elevator and telephone search subtests. The second factor included all three verbal fluency subtests. The third factor reflected auditory processing and included DDT, DSI, TCS, and WIN. The fourth factor included two indices from the TEA visual elevator accuracy and dual task decrement. The fifth outcome included ATTR subtests of across- and within-channel performance. Based on these results, baseline composites of cognitive performance, verbal fluency, ATTR, auditory processing, and everyday function were created by averaging z scores after the reverse-scaling of items with negative factor loadings. Post-test scores were standardized by baseline means and SDs. Timed IADL performance did not significantly load on any factor and was thus examined as an outcome separately.
In the primary analyses, to investigate the efficacy of piano training as compared to music listening to improve auditory processing, cognition, and everyday function among older adults with and without mild cognitive impairment, we examined whether performance changed differentially by random assignment across composite outcomes from baseline to post-test as indicated by a significant group-×-time interaction. Covariates included hearing PTA, visual acuity, and expectations. The group-×-time interactions were not statistically significant for any of the outcomes (ps>0.194). See Table 3. The results did not change when we included only those who were adherent (*n=*207, ps>0.180). See Table 4. Piano training did not significantly enhance auditory processing, cognition, or everyday function as compared to music listening. We further examined if intervention effects varied by MCI status (Table 4). There were no significant group-×-time-by-MCI status interactions indicating those with and without MCI did not show differential benefit (ps>0.245). Thus, MCI did not significantly moderate the effects of piano training.
Results showed significant effects of time for ATTR in all models and significant effects of time for auditory processing in the primary and adherent analyses indicating improved performance from pre- to post-training for both groups. Significant effects of time were also found for cognition and TEA in the primary and adherent analyses indicating a tendency for decline across time for both groups. No other main effects of time were significant.
Effect sizes for piano training as compared to music listening for both composite outcomes and individual assessments are shown in Table 5. Improvements were not evident in the primary analyses, adherent analyses, or the subsample without MCI on measures of auditory processing, cognition, or everyday function. Participants with MCI randomized to piano training showed potential small effect sizes for improvement relative to music listening on Trails A, Digit Symbol Coding, and one subtest of the TEA (ds≥0.25). Raincloud plots of individual assessment and composite variables are available at https://osf.io/fcphq.
With respect to covariate effects, auditory processing, ATTR, and cognitive performance were significantly impacted by hearing (i.e., PTA). Similarly, Timed IADL and cognitive performance varied significantly by visual acuity. Those with poorer hearing or vision tended to perform worse on these outcomes.
Interestingly, expectations about the assigned condition were significantly related to composite outcomes of auditory processing and cognition in the randomized sample. However, having greater expectations that the randomized condition would positively affect abilities was associated with poorer performance. See Table 3.
In addition to our pre-registered analytic plan, several sensitivity analyses were conducted in response to reviewers’ requests. In response, we repeated the primary and secondary (among adherent and by MCI status) mixed models analyses including the 19 participants who were excluded in the aforementioned results due to a head injury (n=1), undergoing anesthesia (n=10), or hospitalization (n=8) and the pattern of results was the same. Among those randomized, we examined whether covariates of age, race, ethnicity, vision, hearing, or randomization arm predicted who did and did not complete the post-test visit using binary logistic regression. None of these factors significantly predicted study completion, χ^2^(7)=8.31, p=0.31. Additional sensitivity analyses examined if there were differential effects by sex. No significant group-time-sex interactions were evident (ps>0.119), and the pattern of results was the same when analyses were stratified by sex. Sensitivity analyses also examined if class size (i.e., the number of persons in a class) was correlated to change across outcomes. No significant correlations with class size and change on the outcomes were evident in the overall sample or within the piano training group (ps>0.09). Class size was significantly associated with changes in verbal fluency within the music listening group. Larger class sizes were significantly associated with greater improvements in verbal fluency among those in the music listening condition, r=0.26, p=0.009.
We examined the efficacy of piano training as compared to music listening to improve auditory processing, cognition, and everyday function among older adults with and without MCI. We further examined whether MCI status moderated intervention efficacy. We applied more rigorous methodology than prior studies of piano training by conducting a randomized controlled trial with an active control group of music listening, pre-registering analyses, and applying an intent-to-treat principle. We did not find improved auditory processing, cognition, or everyday function from piano training relative to music listening. Our hypothesis that piano training would enhance auditory processing resulting in improved cognition and everyday function was not supported. We also did not find significant effects among those who were adherent (defined as attending 15 or more classes), and we did not find significant moderating effects of MCI status or sex. Nevertheless, our results extend prior research by including participants with clinically diagnosed MCI, applying objective measures of music reading for screening, examining a broader array of auditory processing measures, assessing everyday functional performance outcomes, and accounting for expectations of perceived benefits.
Unlike prior studies of piano training, we compared effects among older adults with and without a clinical diagnosis of MCI. Prior piano training studies have either included combined samples of participants who were cognitively normal and with probable MCI based on their inclusion/exclusion criteria (Bugos, 2010; Bugos et al., 2007; Bugos & Wang, 2022; Worschech et al., 2021) or intentionally excluded those with MCI (Seinfeld et al., 2013). We found no significant group×time×MCI status interactions indicating those with and without MCI did not show differential benefit. Thus, MCI did not significantly moderate the effects of piano training. Although we did not find MCI status to be a significant moderator, this could be due to lack of statistical power. The effect sizes among participants with MCI were larger than the overall effect sizes and showed potential small effects (ds≥0.25) reflecting improvements for Trails A, Digit Symbol Coding, and some aspects of TEA performance. Interestingly, a meta-analysis (Dorris et al., 2021) found that active music interventions (such as playing an instrument) had a similar small, significant effect (d=0.30) on global cognitive status among persons with MCI or dementia. Bugos and Wang (2022) found that cognitive effects were significantly stronger for participants with low MoCA scores (potentially indicative of MCI) than those with high MoCA scores. Thus, piano training may potentially be efficacious among persons with MCI with small effect sizes to improve cognitive speed of processing, but further investigation with a larger sample is needed.
We used principal components analyses to inform our creation of composites to reduce the number of comparisons made in statistical analyses. In addition to analyzing the composite data, we reported the effect sizes for each individual measure and the supplemental materials show the scatterplots, box plots, and distributions for the individual assessments. The data for the individual assessments confirm our overall pattern of results. The ATTR tended to show improvements across time with small effect sizes. The largest effect size observed for ATTR was a Cohen’s d effect of 0.24 indicating potential improvements among those with MCI who were randomized to piano training relative to controls. Across the analytic sample, there were no effect sizes indicating potential improvements on the cognitive assessments, but rather effects show a tendency for decline. The only exception is that among those adherent to complete 15 or more sessions, category fluency had a small effect size (d=0.20) indicating potential improvement from piano training relative to music listening. As discussed above, only within the MCI subsample did effect sizes show improvements on some cognitive assessments. Thus, we conclude that piano training may potentially be beneficial among older adults with MCI. However, an experimental study with greater statistical power is needed to investigate this possibility.
It is possible that a longer duration of piano training or additional practice may be necessary to obtain benefits. Our study examined 30 h of group piano training and did not include outside practice. At the same time, when Bugos (2007) et al. examined 40 h of individualized piano instruction with practice, results indicated that 3 months after practice ended the cognitive benefits on Trails B and Digit Symbol were no longer evident relative to a no contact control condition. When examining 48 h of piano training without practice, Bugos and Wang (2022) found improved verbal fluency relative to either cognitive training or a no-contact control arm, but other aspects of cognition were not significantly enhanced relative to controls. Worschech et al. (2021) included 60-min sessions weekly and practice of 30 min per day across 6 months but did not find differential benefits of piano training relative to music listening on auditory processing outcomes. Thus, the required dose of music interventions to facilitate improvements remains unclear.
Alternatively, it is possible that music interventions overall are beneficial for auditory processing rather than piano training specifically. Like some prior studies (Bugos, 2010; Worschech et al., 2021), we compared piano training to music listening rather than a no-contact control group. Our study as well as these prior investigations comparing piano training to music listening have not found that piano training is more efficacious than music listening as indicated with significant group × time interactions. Our results also did not indicate any significant improvements on cognition or everyday function from participating in either music listening or piano training. In comparison, studies finding potential benefits of piano training on cognition have used a no contact control group or non-musical comparison group (Bugos et al., 2007; Bugos & Wang, 2022; Seinfeld et al., 2013).
Interestingly, Dubinsky et al. examined whether 10 weeks of choir participation could improve auditory processing among older adults relative to a no contact comparison group matched for age and peripheral hearing (randomization was not applied). Those who completed choir participation showed improved auditory processing relative to the comparison condition, but no effects were observed for cognition as indicated by listening span or Flanker effect. Our results showed significant effects of time for ATTR in all models and significant effects of time for auditory processing in the primary and adherent analyses indicating improved auditory processing from pre- to post-training for both groups. Overall, research suggests that music participation in general, rather than piano training specifically, may enhance auditory processing among older adults.
On the other hand, the role of socialization cannot be ruled out as a contributing factor to prior research findings that music interventions enhance auditory processing or cognition among older adults. In a study by Bugos and Wang (2022), piano training enhanced verbal fluency relative to either computerized cognitive training or a no contact control condition. Piano training classes may involve more social interaction than computerized cognitive training, but less socialization than music listening courses. To explore the effects of socialization, we conducted sensitivity analyses to examine if class size was associated with improvements across outcomes. No significant correlations with class size and change on the outcomes were evident in the overall sample or within the piano training group (ps>0.09). However, results indicated that within the music listening group, larger class sizes were significantly associated with greater improvements in verbal fluency, r=0.26, p=0.009. Our music listening course did involve more social interaction in the form of class discussions than did our piano training course. Thus, it could be that improvements in verbal fluency subsequent to music training relative to no contact control conditions found in prior studies were due to socialization. A limitation of our study is that we did not include a no contact control group as recommended by experts on cognitive intervention (Green et al., 2019). Thus, we could not directly test if music listening effectively enhanced auditory processing. To do so, a randomized clinical trial to compare music listening to an appropriate control condition is needed.
Unlike prior studies of piano training (Bugos, 2019; Bugos et al., 2007; Seinfeld et al., 2013; Worschech et al., 2021), our analyses considered participants’ expectations for their randomized condition. A critique of cognitive interventions is that participants’ beliefs drive improvements (Simons et al., 2016). Our results indicated that higher expectations were significantly associated with worse auditory processing and cognitive performance; expectations were not significantly associated with ATTR, verbal fluency, or everyday functional performance. Thus, it is unlikely that improvements observed from music interventions in prior research can be attributable to participants’ expecting beneficial results.
Prior research has established that auditory processing performance longitudinally predicts dementia (Gates et al., 2011; Gates et al., 2002). Several studies indicate significant associations between enhanced auditory processing and musical training (Dubinsky, Wood, Nespoli, & Russo, 2019; Merten et al., 2021; Worschech et al., 2021), and prior correlational data indicate that frequently playing a musical instrument is associated with lower dementia risk (Verghese et al., 2003). Although our study did not find immediate benefits of piano training on auditory processing, cognition, or everyday function relative to music listening, longitudinal follow-up could reveal long-term benefits of piano training. As an example, in the ACTIVE study, 10-year follow-up data indicated that functional decline and risk of dementia were reduced across 10 years, despite everyday functional improvements not being evident immediately post-training (Edwards, Clark, et al., 2017a; Rebok et al., 2014). Similarly, although Merten and colleagues (2021) found reduced risk of incident dementia in their epidemiological study among male musicians as compared to non-musicians, no significant associations of music training were found with cognitive decline across 15–20 years using objective cognitive measures.
In summary, longitudinal follow-up on the efficacy of piano training, in particular, and music training, in general, is warranted. Finally, future research is necessary to further examine the connection of poor auditory processing with subsequent dementia (Gates et al., 2002) and explore whether effectively enhancing auditory processing by intervention may reduce dementia risk.