You are on page 1of 21

Ja<unoI 01 _

02007by ...

'IJJV (4j, 2007, 308-328


......-

The Temporal Limits of Cognitive Change


from Music Therapy in Elderly Persons
with Dementia or Dementia-Like
Cognitive Impairment: A Randomized
Controlled Trial
Robert A. Bruer, ARCT, MPE, MTA',2

Edward SpitZnagel, PhD2


C. Robert Cloninger, MD2
Health Cflntre Penetangulehene,
PMetangulllilene, ON
'WlIlIhington UnIvenIIty SChool of Medicine,
St. Loulll, MO
1 Mental

This study explored the temporal limits of cognitive change


from an intention-to-tleat with group music therapy. Eldet1y
cognitively-lmpBired psychiatric inpatients IN =28) participated in an B-week randomized controi trial using a crossover design, Once a week, subjects Wen!l assigned alther to
music therapy or a control treatmant (age-appropriate
movie). The MiniMental State Exam (MMSE) assessed cognition 3 times every week: prior to the intervention, immediatelyafter the mid-afternoon intervflntion, and the morning
foHowing the intervantion. Comparisons between conditions
included weekly changes in individual subject's MMSE
SCOfSS from weekly basaline tD both the 2 follow-ups and the
foliowing week's basaline. Significant next morning improvements in MMSE scores were found within intant-to-tleat music therapy cases as Comparsd tD control cases. While all the
subjects in this study Wen!l cognltlvely impaired, only 17 hed
been fotmaIIy diagnosed with dementia, Based on a Cochrane
Collaboration suggestion that music therapy studias within
geriatric populations iook specifically at the treatment of dementia, a final generalized estimating equation modal con
sidered only the change within the 17 dementia-diagnosed
subjects, Immediately after the intervention. MMSE scores in
the dementia-diagnosed subjects assigned to music therapy
The authors gratefully acknowledge the significant assistance of the control group
facilitator, Marlene Stewart.

Vol. XLIV; No.4, Winter 2007

309

improved 2.00 points compared to the dementia-diagnosed


subjects assigned to the control group (Z = 1.99, P < .05).
Next-day MMSE test SCOnlS in the dementia-diagnosed subjects assigned to music therapy showed averege improvements of 3.69 points compared to the control subjects (Z =
3.38, P < .001). By the folloWing week, no significant cognitive differences remained between the two groups. It was
concluded thet a reasonable music therapy intervention faciOtated by a trained and accredited music therapist significantfy
improved next-morning cognitive functioning among dementia patients. With meny music therapists worldng in geriatric
settings, more research is justified to both raplicete this
study end provide better guidance into the effective use of
music therapy in the treatment of dementia.

Informed prevalence estimates of dementia within elderly music


therapy clients have risen over time. In 1969 for example, it was believed that only 11 % of US nursing home residents had any psychiatric disorder; while sixteen years later, the rate of dementia alone
ranged between 23% and 43% (Htng, 1985). Future prevalence estimates for dementia can be expected to continue growing as the
medical community develops and refines strategies for earlier and
more reliable diagnosis. In addition to future increases in dementia diagnoses within nursing home clients, music therapists can also
expect many clients in Assisted Living (AL) settings to have dementia. A2oo0 study reported that at least half of the one million
Americans living within AL settings had either dementia or some
form of cognitive impairment (Kopetz et aI., 2000).
Historically, a large proportion of music therapists have worked
within geriatric settings (Lathom, 1982; Register, 2002). At the
same time, such therapists have arguably suffered from a paucity of
research relating specifically to music therapy in elderly populations. A 1964 bibliography of existing music therapy research by
Schneider contained only six articles relating to geriatrics. In comparison, Schneider found 44 articles relating to mental retardation
and 8 articles for the deaf. The timely nature of the Schneider synopsis of existing literature was commendable, in that it appeared
within the first volume of the jfJUmal ofMusic Therapy (Schneider,
1964). On the other hand, publication of the first geriatric article
in'the juumai ofMusic Therapy did not occur until more than 3 years
later (i.e., Iiederman, 1967).

310

Journal of Music Therapy

Based on a comparison of publishing dates for the 21 studies included in a 1999 meta-analysis of music therapy for dementia
(Koger, Chapin, & Brotons, 1999), research interest in this area appears to have increased markedly by the mid 1990s. That is, while
only tWo articles (l 0%) were published within the first half of the
12-year review (i.e., 1985 to 1991), eight (38%) were published in
the final 2 years (i.e., 1996 and 1997).
The body of literature is now extensive enough to provide a
broad range of insight to practitioners. With respect to effective
session planning, significant differences have been identified in
the rates of participation between the varied types of musical expression offered to persons with dementia. Brotons and PickettCooper (1994) for example, while remaining resolute that all
modalities appeared reasonable for clients with dementia, at the
same time noted significantly less participation in the areas of composing and improvising. The authors also suggested that dementia
could impair clients' ability to verbally articulate modality preferences. A subsequent study by Hanson, Gfeller, Woodworth, Swanson, & Gerand (1996) found significantly greater client response
during movement as compared to singing.
Outcome studies suggest music therapy with clients with dementia leads to a wide range of overt effects including increased levels of
melatonin (Kumar et al., 1999); decreased agitation (Brotons &
Pickett-Cooper, 1996) and depression symptomology (Ashida,
2000); and improved attention, social interaction (Gregory, 2002)
and language skills (Brotons & Koger, 2000). At least one study
found no benefits relating to cognition (Groene, 1993).
Within psychiatry, it has generally been assumed that music therapy in dementia care is most beneficial for the improvement of behavior. Such an assumption was reinforced more recently by a 2005
American Journal of Psychiatry systematic review of 1,632 studies
involving a variety of psychological approaches to dementia management (Livingston, Johnston, Katona, Paton, & Lyketsos, 2005).
The reviewers concluded that "evidence suggests music therapy decreases agitation during and immediately after sessions."
Psychiatric treatments for dementia have recently focused on
cholinesterase-inhibiting (CEI) medications, which have been
shown to slow down or stop cognitive decline in dementia (Birks,
2006). Despite having significant adverse side effects, CEls are
widely used by persons with mild or moderately-severe dementia
(Lopez, Becker, Wisniewski, Saxton, Kaufer, & DeKosky, 2002).

Vol. XLIV, No.4, Winter 2007

311

Based on the popularity of these "cognition-enhancing" drug treatments, persons with dementia might similarly favor "cognition-enhancing" music therapy.
Improved cognition should be expected with the use of music
therapy treatments for dementia, based on empirical relationships
known to exist between anxiety, cognition, and music. Increased
anxiety in the elderly is correlated with decreased cognition (Jorm,
Christensen, KoTten,Jacomb, & Henderson, 2001). Interventions
which reduce anxiety cause significant short-term reductions in
cognitive impairment (Yesavage,1984; Yesavage &Jacob, 1984). Finally, music stimulus is effective in quickly reducing anxiety (Hirokawa, 2004; Panksepp & Bernatzky, 2002), suggesting that some
rapid cognitive improvement from music therapy should be expected through an anxiolytic effect.
In spite of the popularity of music therapy in dementia and general assumptions as to its effectiveness, the Cochrane Collaboration's
current review of evidence-based outcome research in this area
(Vink, Birks, Bruinsma, & Scholten, 2004) maintains that four necessary components are still missing from the current literature: (a)
randomization of subjects, (b) blinding of assessors, (c) the use of a
standardized assessment tool, and (d) adequate length of trial in
which to establish the longitudinal effects.
Thus, this study's goal was to examine the promising area of
cognitive benefits in music therapy for dementia, while adhering to
the requested standards oCthe Cochrane protocol (Vink et al.,
2004). Specifically, subjects were randomized to either music therapy
or a control condition, assessors were blinded to subjects' assigned
intervention, the we1l-studied Mini-Mental State Exam (MMSE) was
used to assess cognitive change, and the study extended over 8 weeks
in order to explore longitudinal variations in therapeutic efficacy.
So as to be clinically relevant, the music therapy interventions were
held to a level which we believed most accredited music therapists
could facilitate within a normal work schedule.
In addition to our adherence to Cochrane standards, an intentta-treat design was used. Such designs are commonly found in drug
trials, where rates of treatment compliance are markedly lower
than prescribing rates. Intent-ta-treat studies thereby estimate the
practical limits of proposed treatment interventions. l Having simi1Statistically, intenHo-treat designs attenuate the true effect of a treatment,. and
therefore do not increase the chance of Type I errors.

Journal of Music Therapy

312

lar intentions to estimate practical limits of music therapy, the wellunderstood and respected term intent-to-treat" was retained. albeit using it to identify an approach more synonymous with scheduled-to-attend." Such an approach forced music therapy to rise
above realistic session events such as late arrivals, early departures
and completely missed appointments. It was felt by the authors that
positive findings from this approach could provide highly compelling evidence for music therapists to present to administrators
and service providers.
Three hypotheses for testing were posited. First. it was predicted
that an intent-to-treat decision for music therapy would produce a
measurable short-term cognitive improvement as compared to a
reasonable control condition. Secondly, it was predicted that the
cognitive improvement from an intent-to-treat decision for music
therapy would extend into the day following the session. Finally. it
was predicted that the cognition improvement from an intent-totreat decision for music therapy would extend into the week(s) following music therapy.
Method

Subjects
Subjects represented 90% of the total inpatient population on a
geriatric service ward within a government-run Ontario psychiatric
hospital. The ward deals exclusively with disabilities arising from
chronic cognitive impairment, and it was therefore assumed that
all of the ward's patients met the research inclusion requirement of
significant cognitive impairment. Two exclusion criteria were nonelderly status (i.e. less than 60 years of age) and/or severe hearing
loss. Out of all inpatients (31) present on the ward during the
study, three were excluded from the final analysis due to being less
than 60 years of age. No potential subjects had severe hearing loss.
IRB approval was granted from both the studied hospital and the
American medical school to which all the authors are affiliated.
Hospital social workers met with each patient or their legally authorized representative to explain the proposed research and to
obtain signed consent for participation.
Within the province of Ontario. all residents are both eligible and
restricted to an equal level of mental health care regardless of socioeconomic status. Furthermore. the hospital is mandated to provide the same level of mental health to everyone living within a geo-

Vol. XU'/, No.4. Winter 2007

313

graphical area. Thus, it is possible that the research participants were


representative of geriatric inpatients found in a real population.2
When the study began, the 26-bed geriatric ward was running at
slightly less than full capacity with 23 patients. Over the eight week
course of the research trial, discharges and admissions resulted in
a total of 28 patients eventually participating in the study. Subjects'
mean age was 74.1 years (SD= 7.64).
While all patients had been admitted to hospital to address problems arising from chronic cognitive impairment, only 17 of the 28
(61 %) study participants had a dementia diagnosis. Of the remaining 11 patients without a dementia diagnosis, two had bipolar m
ness and three had schizophrenia. One subject did not have a formal psychiatric diagnosis, but did have an ICD-9 general medical
diagnosis of delirium (780.0).s

Procedure
Prior to the study, group music therapy sessions had been avaIlable to all patients for 7 years. Team leaders on the ward had come
to believe all of their patients benefit from music therapy attendance. More than 90% of the ward's patients regularly attended
the weekly music therapy program. Sessions lasted 45 minutes and
had always taken place Thursday afternoon at 2 p.m. The starting
time was moved to 3 p.m. during the research trial, with the hope
that the study might explore the effects of Sundown Syndrome,"
that is the increased agitation and confusion in dementia patients
which occurs in the late afternoon (Burney-Puckett, 1996).
Assessors were trained in small groups by a member of the geriatric team skilled in dementia assessment. Training sessions lasted
an hour and each assessor received take-home review notes. For logistic reasons it was difficult to get a full team of assessors for much
of the study. Instead, random samples of 12 subjects (6 per group)
were assessed for 6 of the 8 weeks. A full assessment was achieved
on weeks 3 and 8.
The day before the start of the study, the patients were randomly
It should be noted that 17 oIthe 28 (61%) subjects were male. while the actual
Canadian senior population is only 43% male. However. the proportional difference was not significant (XI - 3.6. p- .06).
Of the remaining five subjects. primary diagnoses were (i) Unspecified Pl1I1>noia (ii) Depressive Disorder Unspecified (iii) Unspecified Psychosis (iv) Transient
Organic Unspecified. and (v) Affective Psychosis.

314

Journal of Music Therapy

split into two groups. Mter the study had begun, new admissions
were randomly assigned (coin-toss) by an independent party to one
of the two treatment groups.
The control condition was a video presentation planned and facilitated by the ward's activity director. Videos such as "Some Like
it Hot" and "For the Boys" were expected to maximize participants'
enjoyment. Detailed notes pertaining to the control condition
were kept by the facilitator.
The study used a croslH>ver design, with subjects acting as their
own controls. One group per week was offered the opportunity to
attend music therapy while the other group was offered the control
activity. The following week, the group previously offered music
therapy was offered the opportunity to attend the control activity,
while the other group was offered music therapy.
CroSlH>ver designs are commonly used in drug trial research to
provide robust assessment of intersubject variance in drug absorption and elimination. In this study, the use of a croslH>ver design
was expected to similarly assess wide intersubject variance expected
in response to the music therapy treatment.
As previously noted, ward staff have traditionally encouraged all
the patients to attend music therapy. This allowed the study to consider the more relevant effect of "intent-to-treat" with music therapy.
Staff and patients rentained free to decide on a case-by-case basis
whether it was appropriate for patients to attend an assigned intervention, be it music therapy or the control program. However, it is
noted that the converse was not true and patients were never permitted to attend an intervention to which they were not assigned.
Baseline assessments (Time 0) took place at 10 a.m. each Thursday morning. Interventions took place at 3 p.m. and the first follow-up assessments occurred at 4 p.m. (Time I). The next morning
(Friday) at 10 a.m., the second follow-up assessments took place
(Time 2). Each assessor was assigned four patients per assessment
period, allowing all testing to be completed within an hour. Assessors began each test with a reminder to the subjects as to their right
to refuse assessment. Patient assignments were rotated among the
assessors so that each assessor never evaluated the same patient
more than once per week.
Instrument

The Mini-Mental State Exam (MMSE) (Folstein, Folstein, &


McHugh, 1975) was used to assess cognitive change. The MMSE is

Vol. XU'/, No.4, Wint&r 2007

315

a brief (less than 20 min) tool and requires a minimal amount of


training. With a maximum score of 30. the MMSE assesses cognitive ability within a wide range of domains. including calculation,
memory, language, and orientation. Since its introduction in 1966,
the MMSE has become widely recognized and respected within the
healthcare professions for its reliable assessment of cognitive impairment associated with dementia. A score below 23 on the MMSE
is generally thought to indicate cognitive impairment. In individuals with dementia, the MMSE is effective for documenting changes
in cognition which occur over time (Tombaugh. 2005; Tombaugh
& McIntyre, 1992).
Music Therapy Approach

The music therapy treatment condition was based on Reality Orientation methodology, modified musically to contain at least 15
musical selections per session. Gibbons (1977) has suggested the
most appropriate therapeutic music selections are songs made popular when clients were about 25 years old. Thus. many of the songs
used in the study were originally made famous by Nat King Cole
and Elvis Presley. during the mid 1950s. With vocal ranges known
to become more restricted with age (Greenwald & Salzberg. 1979),
all melodies intended for singing were transposed to the most accessible keys.
In persons with dementia. cognitive functioning required for fa
cial recognition is reported to decline less rapidly than functioning
required for verbal communication (Bucks & Radford, 2004). As a
result. the earlier portion of sessions dwelt on facial features and
gestures rather than names. Also early in each session. much atten
tion was given to the weather and a eurhythmic component intended to recall spatial concepts such as left, right, up, and down.
Roughly 15 minutes into the session. at least 5 minutes were used
to individually welcome attendees with a song (i.e., a personalized
'hello song'). All attendees were encouraged to sing along and
greet each other by shaking hands.
Greene (1982) and Schmitt (1990) found elderly patients favor
humorous medical staff. Humor was therefore employed thera
peutically, with the regular inclusion of a component devoted to
telling jokes. Beginning with a clear disclosure that "this is therapeutic laughter," unison laughter was then rehearsed, followed by
jokes and clear prompts to "laugh!" The therapist often described
laughter to be an aid to effective oxygen intake. Jokes were selected

316

JoumsI of Music Thetapy

to thematically relate to concepts such as hospitalization, the season, and issues related to aging and memory failure. Following the
jokes, the therapist moved around the group and had each attendee "test" their lungs by blowing on a flag, while a laughter song
played in the background.
It is expected that hospitalization and dementia contribute to
stress within patients. Cloninger (2004) believes stress attenuates
sensory awareness. Natural props such as tree sprigs, flowers, and
spices were used to stimulate the oft-neglected senses of touch, vision, and smell, accompanied by background music recordings.
Goswami (2002) suggested that phoneme-level language units,
such as those found in simple rhyming poetry, contribute to both
learning and recalling information. Neuropsychological processing of poetry and music are reportedly highly similar (Lerdahl,
2001). Over time, these theories had lent support to an increased
use of poetry within the studied music therapy program.
Appropriate "call and response" and "lining out" techniques were
used for learning new songs, accompanied by comments such as
"you're never too old to learn." The group was commonly divided
into smaller sections to sing rounds such as "Row, Row, Row, Your
Boat." Time was always allotted to reflect on the extent to which
these "performance-focused" components had been processed.
Clair and Bernstein (1990) noted that severely demented individuals do not usually sing. Hanson et al. (1996) found such persons very willing to participate in components involving movement. Brotons & Pickett-Cooper (1994) reported instrument
playing in particular to be favored by severely-impaired clients.
Thus, a playing component occurred every week near the end of
each session, accompanied by ongoing prompts (e.g., modeling),
as stlggested by Cevasco &: Grant (2003).

Statistical Analyses
Data were analyzed using SASQi> version 9.1 for Windows (SAS,
2004). The repeated use of subjects (i.e., clustering) was accounted
for within all SAS procedures except the preliminary use of the t
test procedure. For the final Generalized Estimating Equation
(GEE) model, the GENMOD procedure specified a normal distribution and a linear (identity) link. A Wald's chi-square test determined the goodness of fit. A Pvalue of less than .05 indicated a statistically significant result for all hypothesized effects.

Vol.

xuv. No.4, Winter 2007

317

Results
Rates of confirmed attendance by those scheduled for music
therapy fell over the course of the study, from an average of81 % attendance in the first half of the study, to 58% in the second half.
The largest decline occurred in the 11 subjects without a dementia
diagnosis, from 86% confirmed attendance in the first half to 44%
in the second half. Attendance by the 17 subjects with a dementia
diagnosis also fell over the course of the study, from 79% in the first
half to 67% in the second half. In light of the fact that this study
sought to research the effectiveness of music therapy in the treatment of dementia, the change in attendance between the first and
last half of the study for those subjects with a dementia diagnosis
was tested. The decline in attendance by those with dementia between the first and second half of the study was significant (p - .03).
Over the course of the study, 66% of the scheduled MMSEs (199
of 300) were successfully completed. Reasons for incompletion
were recorded by assessors. Factors such as absence from ward. being asleep. or being occupied with toileting or bathing were assumed to be unassociated with systematic changes arising from the
treatment conditions. No similar assumption was made regarding
44 assessments (15% of total) marked by assessors as being "refused". Proportional analyses were carried out on several possible
predictors for refusing assessment and the results are summarized
in Table 1. Three factors significantly predicted assessment refusal: malegemkr(Xl-15.84. p= .00), more than 1 year at the same address (Xl - 8.54, P= .00) and follow-up assessments (Xl - 17.50. P=
.04). Within the context of this investigation into the effects of
music therapy. it is important to note that actual music therapy attendance had no significant effect on later refusing to be assessed
(Xl - 0.01. P- .92). That is. in follow-up assessments. similar proportions of music therapy attendees and non-attendees refused to
be assessed.
All MMSE scores greater than zero were deemed valid and used
in the final analysis. Only 5% of the tests (9 of 199) had a perfect
score of 30. therefore. any "ceiling effect" was assumed to be insignificant. The mean score for all tests was 18.3 (SD = 8.2). Tests
completed by subjects with a dementia diagnosis had an average
score of 14.5. as compared to an average score of 22.5 by those
without a dementia diagnosis.

318

Journal of Music Therapy

Hypothesis I predicted that "intent-to-treat with music therapy'


would significantly improve short-term cognitive functioning as
compared to the control condition. For the purposes of this study,
short-term improvement was represented by the change between
each subject's baseline MMSE score (i.e., Thursday at 10 a.m.) and
their respective score in the MMSE administered immediately after
the intervention (i.e., Thursday at 4 p.m.). Subjects assigned to music therapy improved an average of 1.46 points between baseline
and immediately after the intervention, while control scores wonr
ened an average of 0.61 points. A t test which assumed unequal
variance between the two treatment groups found the resulting difference of 2.07 points in favor of music therapy to be not significant (t=1.48, p< .13).
Hypothesis 2 predicted that "intent-to-treat with music therapy'
would significantly improve next day cognitive functioning as compared to the control condition. Next-day improvement was represented by the change between each subject's baseline MMSE score
(i.e., Thursday at 10 a.m.) and their respective score in the followup MMSE test administered the morning after the intervention
(i.e., Friday at 10 a.m.). In subjects assigned to music therapy,
MMSE scores improved an average of 1.70 points between baseline
and the morning after the intervention, while control scores wonr
ened an average of I. 76 points. 4 The resulting difference of 3.46
points in favor of music therapy was significant. (t = 2.34, P< .02).
With the intent-to-treat protQ(:ol essentially counting some cases
where treatment did not actually occur, the following logical expectations were investigated:
I. The difference in score changes found in subjects who were
known to have attended music therapy versus those known to
have not attended music therapy should be greater than the significant +3.46 difference found in the above intent-to-treat versus control comparison.

4 It has been assumed that the decline in control scores between Thursday and
Friday morning results from a shift rotation in hospital staff which takes place between these 2 days. Subjects wake up Friday morning to many new faces (i.e., staff).
At the same time, the new team of caregivers is expected to be preoccupied on Friday morning with understanding the changes which occurred during their off-days.
It seems reasonable therefore, that Fridays are a time of significant distress (and increased cognitive disturbance) for patients.

319

Vol. XLIV, No.4, Winter 2007


TABLE 1

PossiIJle Pn!diaive FiJ.dors fer lIJifwing AsJmment


p

Weeks 5-8 (51%)


Male gender (61%)
Private residence (63%)
Married/common-law (50%)
More than 1 year at the same address (48%)
Dementia diagnosis (64%)
Follow-up assessments (67%)
Music therapy attendance
(within follow-up assessments) (35% of 67% =23%)

15.0%
20.8%
15.3%
16.0%
20.8%
14.6%
17.5%
17.1%

0.0
15.8
0.2
M
8.5
0.0
4.1
0.0'

.85
.00*

.67
.51
.00*
.96
.04*

.92

The overall refusal rate ofl4.7% is based on a total of 44 refusals out of 300.
, 17.1 % compared to 17.5% refusal rate for follow-up assessments.
I

2. In subjects with a dementia diagnosis, the difference in score


changes between those who actually attended music therapy
versus those who had not attended music therapy should show
a still greater difference in next-day cognitive change as compared to both the +3.46 difference already found in the intentto-treat versus control comparison and the expected improvement to be found within those who actually attended music
therapy. Such expectation was built on the fact that the studied
music therapy approach was developed by a trained clinician
over 7 years. From the start, the main therapeutic goal was to
improve cognition in persons having a dementia diagnosis. Therefore it was expected that the treatment should be most effective
for those diagnosed with dementia.
Both of the above postulates were confirmed, as detailed in
Table 2. Scores for subjects who actually attended music therapy
improved 2.74 points while scores of those not attending worsened
1.50 points (for a total difference of 4.24 points in favor of music
therapy). Furthermore, looking only at subjects with a dementia diagnosis, scores for those who actually attended music therapy improved 3.92 points while those not attending worsened 1.69 (for a
total difference of 5.60 in favor of music therapy).
Analyses of variance (ANOVAs) tested the validity of the study
design, with no significant variance found in group assignment, F
(1,18) = .70, p> 0041, study week, F (7,21) = 1.29, p> .30, or followup testing, F (2,189) = 0.27, P> .77.

Journal of Music Therapy

320

TABLE 2

UJgkal Hitrtm:hy Of 0IJservIId NexWay Cognitive ImfJroutm<nt Based on Tuled Group


Difference in

Tested grouf

[ntenHo-treat with Music Therapy

Control Group
Confirmed Attendance in Music Therapy
\'S. Non-Attendance in Music Therapy
Confirmed Attendance in Music Therapy by
with Subjects Dementia \'S. Non-Attendance
by Subjects with Dementia

me>m

+3.46

\'S.

+4.24
+5.60

,
(dl)

2.34
(47)
2.87
(47)
3.18
(26)

0.024*
0.006*
0.004"

3
Cross-Compari.wn:r ofMeans in NexWay Cognitive Chang<

TA8LE

Within First Halfof Study:


Control Group-Q.33
\'S.
Music Therapy Group +2.55

+2.88

0.188

Within Control Group:


First HalfofStudy-Q.33 \'S. Second HalfofStudy-3.18

-2.85

0.193

Within Music Therapy group:


First Half of Study +2.55 \'S. Second Halfof Study +1.20

-1.35

0.515

Witbin Second Half ofStudy:


Control Group -3.18
\'S.
Music Therapy Group +1.20

+4.38

0.038*

In contrast, a cross-comparison of intent-to-treat with study halves


found an unusual difference in cognitive change within the control
group across the study. As detailed in Table 3, next..<Jay cognitive
change in the control group shifted from an average decline of
0.33 points in the first half of the study to a larger decline of 3.18
points in the second half of the study. 5
While noting that the difference was not significant (i.e., p <
.193), clinical notes from the control group were nonetheless reviewed to identify possible causes for the larger average decline in
the second half of the study. Supported by the unusual decline,
one logical concern related to whether the control group might
5 At the same time, it was noted that music therapy between the first and second
half of the study remained the most stable of all cross-comparisons. showing a non~
significant change between an average improvement of 2.55 points in the first half
down to a less robust improvement of 1.20 points in the second half,

321

Vol. XUV; No.4. Wintl!lr 2007

TABLE

~p Noles in Sectmd Half of-,

Week 5
Week 6
Week 7
Week 8'
J

"My best yet ... popcorn ... great movie...


... they really enjoyed watching 'Some Uke It Hot' ...
"The best run group to date ...

No part of the notes from Week 8 mentioned the effectiveness of the 5esaion.

5
AMIysis of Cii: EmpitiaJi Slandtml Estimatts ofDiff- in AWn MMSE &mes J1dr-.
1 . . - - Mwi< 7'1Im>J1J tIS. Control in Individuals with 0 Dtmmtio Diognosis.

TABLE

Intercept
Music therapy va. control
at Tune 1 (same day)

1'>111

0.0000

0.0000

0.0000

0.0000

2.0002

1.0036

0.0331

3.9673

1.99

.0463"

3.6928

1.0920

1.5524

5.8332

3.38

.0007"

Music therapy VI. control

at TIme 2 (next day)


Music therapy va. control

atTune 3 (next week)

-0.2958

0,4949 -1.2657

0.6741 -0.60

.5500

No. Wald Statistic!! for Type 3 GEE Anal)'llis of source intent-to-treat " thne (allowing for the repeated use of subject through assuming an autoregressive correlation):
X' = 17.02 (df, 3). p< .0007.

have functioned less well over time. However. as highlighted by relevant excerpts from notes taken by tbe control-group facilitator
and reproduced in Table 4, tbe control group in fact appeared better run in Weeks 5 to 8.
The 2004eochrane protocol for geriatric music tbempy was restricted to persons "formally diagnosed as having dementia" (Vink
et al" 2004). Thus. a final analysis compared tbe difference in cognitive change witbin the 17 subjects having a dementia diagnosis,
using a GEE to model MMSE change. The interaction between intmt-to-treat and time was a significant predictor of MMSE change
(X2 -17.02. dj= 3. P-.0007). The exact statistics as provided by tbe
$AS GENMOD procedure are provided in Table 5 witb Estimate
representing tbe difference in means between intent-to-treat music
tbempy and tbe control condition. Baseline MMSE scores were
fixed to zero (i.e. tbe intercept was fixed to O). Times 1, 2. and 3
represent respectively: (a) change from the 10 a.m. Thursday

Journal of Music Thelapy

322

scores (baseline) to 4 p.m. Thursday (same-day), (b) change from


baseline to 10 a.m. Friday (next-day), and (c) change from baseline
to the following Thursday at 10 a.m. (next week). Results showed a
significant 2.00 improvement from music therapy as compared to
control within the same day (Time 1), growing to 3.69 by the next
day (TIme 2), and returning to a nonsignificant -{).30 difference by
the next week (TIme 3).
A post hoc analysis was carried out on next-day standardized
changes in MMSE subscale scores for those subjects with a dementia diagnosis who were known to have actually attended music therapy. As detailed in Figure 1, the three-stage language question relating to folding a piece ofpaper showed the greatest improvement
(t = 2.53). A Multiple ANOVA (MANOVA) using confirmed music
therapy as the dependant variable was not significant, F (11, 15) =
1.03, p> .47. The nonsignificant MANOVA result left this study
constrained from attaching significance to the large t score found
on the MMSE question relating to a 3-stage command.
Discussion
An intention to treat all of the patients on a geriatric ward with
music therapy led to significantly improved cognitive functioning
the day after treatment as compared to a well-run and age-appropriate video control condition. Overall implications for these robust effects are constrained by wide variations in cognitive levels
observed within the control condition. While such control variation remains poorly understood, the lack of similar variation within
the music therapy participants (i.e., cognitive change from music
therapy was as hypothesized) supports the possibility of a protective
role for music therapy during critical times of the day (e.g., mid-aftemoon within this study).
As requested by the Cochrane Collaboration's protocol for future reports of music therapy efficacy in dementia (Vink et a!.,
2004), this study identified an average next-day improvement in
MMSE scores of 3.69 points (SE = 1.09) for individuals known to
have a dementia diagnosis. As a comparison, a Swedish populationbased study estimated the average rate of MMSE-scaled cognitive
decline in persons with mild to moderate dementia to be 2.4 points
per year (Aguero-Torres, Fratiglioni, Guo, Viitanen, & Winblad,
1998). Therefore, by the next day, the course of dementia in music
therapy subjects' appeared to have moved back in time approxi-

~
~
:<:

2._....

1.11
1.04

3. _

'.

8 olljocIo.

'00.",,,,_1.'

8. IclII1lIIy a

_1lIId-

5. _

$.

~
~

1.57

"rio.

"'*"'" _ "'*"""

"

-0.7

by

7.Aapaa!""10.-''''_
5. Talcaa_InItl.loIdhall

.-011 _

.0.69
2.53

1.54
1.67

O. -llIIdolla\l 'ClOSE YOUR E'IEll.

100_a_

.... ~._-~_ .. _._-~~--

0.69

FrGURE 1.

Standardized next"day MMSE item change within confirmed music therapy attendees having dementia diagnosis.

324

JoumaJ ofMusic Therapy

mately 1.5 years. Practically, it seems reasonable that many persons


with dementia might derive much benefit from music therapy the
day prior to situations requiring their maximum cognitive abilities
(e.g., legal appointments).
Such single-session benefits were not permanent, in that followup tests carried out a week after the treatment found cognitive improvements from music therapy to have disappeared. The observed rise and fall in cognition over a week is congruent with
theory to suggest that music therapy acts primarily by way of shortterm anxiety reduction.
A recent revision to the Cochrane Collaboration review of music
therapy in schizophrenia and schizophrenia-like illness (Gold, HeJ..
dal, Dahle, & Wigrarn, 2005) was qualified by the comment: specific techniques of music therapy, including, among others, musical improvisation and the discussion of personal issues related to
the musical processes, require specialized music therapy training."
(our italics). likewise, while it is acknowledged that this study used
very different techniques from those used by music therapists to
treat schizophrenia, it should be nonetheless noted that treatment
was facilitated by a trained and accredited (Canadian Association for
Music Therapy) music therapist. Thus, these results are not intended to infer general efficacy for therapeutic interventions involving music by persons outside the music therapy profession. On
the contrary, it is expected that this type of treatment requires a
trained music therapist.
Previously, Brotons and Pickett-Cooper (1994) found 20% of
their elderly music therapy subjects disliked singing modalities
while expressing no similar dislike for instrument playing and
dance!movement modalities. Similarly, participants in this study
rarely sang within the music therapy sessions. Instead, most subjects restricted their musical involvement to movement and instrumental playing. In light of the fuct that such modalities of musical
functioning coincided with improvement in next-day cognition, it
is reasonable to state that movement and playing are not only
client-preferred modalities, but also therapeutically-effective
modalities. Thus, it is recommended that music therapy intended
to improve cognition in clients with dementia should commonly
engage such clients in movement and playing.
Music therapy had no apparent effect on whether a subject later
refused to be assessed, suggesting subjects' level of complacency or
happiness remained unchanged within a music therapy interven-

Vol.

xuv, No.4, Winter 2007

325

tion designed specifically to improve cognition. At the same time,


the stability of affect occurred in a treatment clearly derived from
Reality Orientation methodology. This study did not consider popular geriatric music therapy techniques such as reminiscence and
sensory stimulation. Such approaches commonly set goals relating
to the promotion of positive emotion and self-awareness (Grasel,
Wiltfang, & Komhuber, 2003), and therefore remain more appropriate for reducing sadness and depression.
It is noted with caution the results of the analysis of change by
MMSE subscales, in which the greatest improvement occurred in
the "behavioral" component involving the three-stage task of taking
a piece of paper, folding it in half and placing it on the floor. On
the one hand, the large improvement found in this task coincides
with findings from both psychiatric and music therapy studies
where the greatest effect seen in music therapy occurred within the
behavioral domain (Groene, 1993; Livingston et al., 2005). On the
other hand, the significance of the post hoc findings remains questionable in light of a non-significant MANOVA test result.
Limitatians and Strengths
A wide variety of factors limit the implications of these findings.
Both males and subjects who had a more stable address prior to
hospitalization (i.e., more than one year at the same address) were
significantly more likely to refuse to be assessed. During the latter
half of the study, actual attendance by those with dementia declined significantly from 79% to 67%. Also during the second half
of the study, a robust decline in cognition within the control group
remains poorly understood.
With the low 22.5 average for MMSE scores in those without a
dementia diagnosis, it is not clear whether many of them might
also have dementia, albeit less severe. It is therefore noted that the
greater effectiveness found in those with a dementia diagnosis does
not necessarily mean that the intervention was specifically effective
for dementia, but rather that it was most specifically effective in the
more severe forms of dementia.
As for the strengths of this study, a wide range of factors add support to the findings. In particular, this research took place in
Canada, where socioeconomic status does not affect the mental
health care provided to citizens. Furthermore, the hospital where
the research took place serves a complete geographical area, with
expectation for admission theoretically the same for every citizen.

326

Journal of Musk: Therapy

Of those citizens admitted to the ward with behavioral concerns


arising from cognitive impairments associated with aging, 97% participated in the study. The actual rate of attendance in music therapy sessions over the course of the study was 62%. Accordingly, the
subjects appear to be a true population sample.
Over the 7 years during which music therapy has been provided
on the ward, the program has become increasingly popular, with
more than 90% of the patients regularly attend the weekly sessions.
By way of contrast, literature has found other widely-used interventions such as (nonmusic therapy) Reality Orientation occasionally
judged unpleasant by participants (Livingston et al., 2005). Finally,
in stark contrast to the very common risks associated with many
geriatric medications, staff members on the ward have noticed no
adverse events associated with their music therapy program.

Future Direaions
Beyond a fundamental need for these findings to be replicated,
many issues indicate a need for future research to remain clinicallyfocused. With many similar sessions for dementia expected to occur in the morning, future research should soon determine
whether morning music therapy sessions are as effective as the afternoon sessions considered in this study. Further studies should also
determine the most effective frequency (i.e., dose level) for music
therapy in dementia.
In conclusion, this study sought to fill a crucial gap which we perceived to exist in outcome research. We hope our efforts provide encouragement and guidance to front-line clinicians. We also hope
these results inspire future investigations so as to gain increased understanding into the popular use of music therapy in dementia care.
References
Aguero-Torres, R., Fratiglioni, L., Guo, Z., Vlitanen, M.. 8< WmbIad, B. (1998). Prognostic factors in very old demented adults: A seven-year follow-up from a population-based survey in Stockholm. journal af the American Geriatrics SocUty,
46(4),444-452.
Ashida. S. (2000). The effect of reminiscence music therapy sessions on changes in
depressive symptoms in elderly persons with dementia. journal afMusic T""'apy.
37,17(>",182.
Birks,]. (2006). Cholinesterase inhibitors for Alzheimer's disease. Oxkram DaJQ./;ase
<if Systemo.tic &views (I), CD005593.
Brotons, M., 8< Koger, S. M. (2000). The impact of music therapy on language functioning in dementia journal afMusic T""'apy, 37, 183-195.

Vol.

xuv. No.4, Winter 2f)()7

327

Brotons, M., & Pickett-Cooper, P. (1994). Preference of Alzheimer's disease patients


for music activities: Singing. instruments., dance/movement, games, and com~
position/improvisation. Journal ofMusic Therapy, 31, 220-233.
Brotons, M., & Pickett-Cooper, P. K. (1996). The effects of music therapy interven
tion on agitation behaviors of Alzheimer's disease patients. Journal ofMusic T""'
apy, 33, 2-18.
Bucks, R. S., & Radford, S. A. (2004). Emotion processing in Alzheimer's disease. Aging&! MmtalHea/th, 8(3),222-232.
Burney-Puckett. M. (1996). Sundown syndrome: Etiology and managemenLJournal
ofPsycIwociJJJ Numng "nd Mmtal Heahh Strvicts, 34(5), 40-43.
Cevasco, A. M., & Grant. R. E. (2003). Comparison of different methods for elicit
ing exercise-to-music for clients with Alzheimer's disease. journal of Musi TMr-apy, 40,41-56.
Clair. A. A.. & Bernstein. B. (1990), A comparison of winging, vibrotaetile and nonvibroractile instrumental playing responses in severely regressed persons with
dementia of the Alzheimer's Type. Journal of Music Therapy, 27, 119-125.
Cloninger, C. R. (2004). FNiing good:TM sciena of w<IU1eing. Oxford; New York: Ox
ford Univemty Press.
Folstein, M. F. Folstein. S. E.. & McHugh, P. R. (1975). "Mini-mental state". A prac
tical method for grading the cognitive state of patients for the clinician. Journal
of Psychiatric 1IJtseaTch, 12(3), 189-198.
Gibbons, A. C. (1977). Popular music preferences of elderly people. Journal of Mu
sil: TMropy, Xlv, 180-189.
Gold. C., Helda!, T. 0., Dahle, T.. & Wigram. T. (2005). Music therapy for schizophrenia or schizophrenia-like illnesses. Cochrane Database ofSystematic Reviews
(2), CD004025.
Goswami. U. (2002). In the beginning was the rhyme? A reflection on Hulme.
Hatcher, Nation. Brown, Adams. & Stuart (2002). Journal ofExperimmtal Child
Psych.ology, 82(1),47-64.
Grasel. E.. Wiltfang,j.. & Kornhuber,j. (2003). Non-drug therapies for dementia:
An overview of the current situation with regard to proof of effectiveness. D~
mentia and Gmiatric Cognitive DisurtUrs. 1.5(3), 115-125.
Greene, G. j. (1982). The practical unit administrator's toolkiL The Hospital and
Heahh Strviw Review, 78(8), 227.
Greenwald, M. A., & Salzberg, R. S. (1979). Vocal range assessment of geriatric
clients. Jouma/ of Music Therapy, 16, 172-179.
Gregory, D. (2002). Music listening for maintaining attention of older adults with
cognitive impairments. Journal ofMusic TMropy, 3, 244-264.
Groene. R. W. (1993). Effectiveness of music therapy }:1 intervention with individuals having senile dementia of the Alzheimer's type. Journal ofMusic TMrapy, 30.
138-157.
Hanson. N. Gfeller. K., Woodworth. G., Swanson. E. A. & Gerand, L. (1996). A
comparison of the effectiveness of differing types and difficulty of music activities in programming for older adults with Alzheimer's disease and related disorders. Journal ofMusic T'herapy, 33,93-123.
Hirokawa. E. (2004). Effects of music listening and relaxation instructions on
arousal changes and the working memory task in older adults. Journal ofMusic
TMropy, 41. 107-127.

328

Journal of Music Thampy

Htng, E, (1985). Nursing home utilization by current residents: United States. 1985.
Retrieved November 27. 2005, from http://www.cdc.gov/nchs/data/series/sr_
13/srI3_102.pdf
Jorm. A. F.. Christensen. H., Korten. A. E.Jacomb. P. A.. &. Henderson. A. S. (2001).
Memory complaints as a precursor of memory impairment in older people: A
longitudinal analysis over 7-S years. PsydwWgical Medici,.., 31(3). 441-449.
Koger, S. M., Chapin, K., &. Brotons. M. (1999). Is music therapy an effective intervention for dementia? A meta-analytic review of literature. Journal ofMwic Therapy. 36. 2-15.
Kopetz, S. Steele. C. D. Brandt,]..l3aker, A., Kronberg, M. Galik, E., et al. (2000).
Characteristics and outcomes of dementia residents in an assisted living facility.
1nttrnatUmalJournal ofGmatri< Psychiatry. 15(7).586--593.
Kumar, A. M.. TIms, F.. Cruess, D. G., Mintzer, M.]., lronson, G., Loewenstein, D., et
aL (1999). Music therapy increases serum melatonin levels in patients with
AJzheimer's disease. A1krnative Therapies in Health and Medici,.., 5(6), 49-57.
Lathom, W. B. (1982). Survey of current functions of a music therapisLJournal of
Music Therapy. 19, 2-27.
Lerdahl, F. (2001). The sounds of poetry viewed as music. Annals of the Ntw y",*
Academy of&iences, 930,337-354.
Uederman, P. C. (1967). Music and rhythm group therapy for geriatric patients.
Journal of MILl;'; Therapy. 4, 126-127.
Uvingston, G.,Johnston, K., Katona, c., Paton,]., &. Lyketsos, C. G. (2005). Systematic
review of psychological approaches to the management of neuropsychiatric
symptoms of dementia. The AmericanJournal ofPsychiatry, 162(1l), 1996-2021.
Lopez. O. L.. Becker,]. T., Wisniewski, S., Saxton,].. Kaufer. D. I., 8< DeKosky, S, T.
(2002). Cholinesterase inhibitor treatment alters the natural history of
AJzheimer's disease. Journal of NeurolnlfJ, Neurosurgt>:ry, and Psychiatry, 72(3),
31!>-314.
Panksepp.].. &. Bernatzky, G. (2002). Emotional sounds and the brain: The neumaffective foundations of musical appreciation. Behavioural ProasSe5t 60(2),
133-155.
Register. D. (2002). Collaboration and consultation: A survey of board certi6ed music therapists. Journal ofMwic Therapy. 39, 305-321.
SAS. (2004). SAS OnlineDoc$ 9.1.3. (Version 9.1). Cary, NC: SAS Institute Inc.
Schmitt. N. (1990). Patients' perception oflaughter in a rehabilitation hospital. &habilitation Nursing, 15(3).143-146.
Schneider. E. H. (1964). Selected articles and research studies relating to music
therapy. Journal of MILl;'; Therapy. 83-110.
Tombaugh. T. N. (2005). Test-retest reliable coefficients and 5-year change scores
for the MMSE and 3M5. Archiv<s of Clinical Neuropsyclwlogy, 20(4), 485-503.
Tombaugh. T. N. 8< McIntyre, N.]. (1992). The mini-mental state examination: A
comprehensive review. Journal of the American Geriatrics Society, 40(9),922-935.
Vink. A. C. Birks.J. S. Bruinsma, M. S. 8< Scholten. R.]. (2004). Music therapy for
people with dementia. ClJChrane Database ofSystmultU Rroitws (3), CD003477.
Yesavage.]. A. (1984). Relaxation and memory training in 39 elderly patients. The
American]ournal ofPsychiatry, 141(6),778-781.
Yesavage,]. A. &. Jacob, R. (1984). Effects of relaxation and mnemonics on memory.
attention and anxiety in the elderly. ExperimnUa1Agjng_h. 10(4), 211-214.

You might also like