The Sydney Multisite Intervention of LaughterBosses and ElderClowns (SMILE) study: cluster randomised trial of humour therapy in nursing homes.

0.42 MB·English

by  Low, Lee-Fay

Open Access Research
The Sydney Multisite Intervention of
LaughterBosses and ElderClowns
(SMILE) study: cluster randomised trial
of humour therapy in nursing homes
Lee-Fay Low,1 Henry Brodaty,1 Belinda Goodenough,1 Peter Spitzer,2
Jean-Paul Bell,3 Richard Fleming,4 Anne-Nicole Casey,1 Zhixin Liu,1
Lynn Chenoweth5
Tocite:LowL-F,BrodatyH, ABSTRACT
ARTICLE SUMMARY
GoodenoughB,etal.The Objectives:Todeterminewhetherhumourtherapy
SydneyMultisiteIntervention
reducesdepression(primaryoutcome),agitationand Article focus
ofLaughterBossesand
ElderClowns(SMILE)study: behaviouraldisturbancesandimprovessocial ▪ TheSydneyMultisiteInterventionofLaughterBosses
clusterrandomisedtrialof engagementandquality-of-lifeinnursinghome and ElderClowns (SMILE) study is a cluster rando-
humourtherapyinnursing residents. misedcontrolledtrialevaluatingtheeffectofhumour
homes.BMJOpen2013;3: Design:TheSydneyMultisiteInterventionof therapyon depression, agitation, behavioural distur-
e002072.doi:10.1136/ LaughterBossesandElderClownsstudywasasingle- bances, social engagement and quality-of-life in
bmjopen-2012-002072 blindclusterrandomisedcontrolledtrialofhumour nursinghomeresidents.
therapy.
Key messages
▸ Prepublicationhistoryfor Setting:35Sydneynursinghomes.
▪ Humour therapy was not shown to impact on
thispaperareavailable Participants:Alleligibleresidentswithingeographically
online.Toviewthesefiles definedareaswithineachnursinghomewereinvitedto depression (main outcome), behavioural distur-
pleasevisitthejournalonline participate. bances other than agitation, social engagement
(http://dx.doi.org/10.1136/ Intervention:Professional‘ElderClowns’provided orquality-of-life.
bmjopen-2012-002072). ▪ Humour therapy was successful in reducing agi-
9–12weeklyhumourtherapysessions,augmentedby
tation in the intervention at 13-week and
Received10September2012 residentengagementbytrainedstaff‘LaughterBosses’. 26-week follow-up relative to controls who
Revised15November2012 Controlsreceivedusualcare.
receivedusualcare.
Accepted10December2012 Measurements:DepressionscoresontheCornell ▪ Humour therapy should be considered as a psy-
ScaleforDepressioninDementia,agitationscoresonthe
chosocial intervention to reduce agitation, before
Thisfinalarticleisavailable Cohen-MansfieldAgitationInventory,behavioural
foruseunderthetermsof startingmedication.
disturbancescoresontheNeuropsychiatricInventory,
theCreativeCommons
socialengagementscoresonthewithdrawalsubscaleof Strengths and limitations of this study
AttributionNon-Commercial
MultidimensionalObservationScaleforElderlySubjects, ▪ SMILE offered a novel model of humour therapy
2.0Licence;see
http://bmjopen.bmj.com andself-ratedandproxy-ratedquality-of-lifescoresona that combined staff training (LaughterBosses)
health-relatedquality-of-lifetoolfordementia,the and 12 visits by professional performers
DEMQOL.Alloutcomesweremeasuredattheparticipant (ElderClowns).
levelbyresearchersblindtogroupassignment. ▪ The sample size was large involving 398 resi-
Randomisation:Siteswerestratifiedbysizeandlevel dents living in 35 Sydney nursing homes: 17
ofcarethenassignedtogroupusingarandomnumber homes in the intervention group and 18 in the
generator. usualcaregroup.
Results:Seventeennursinghomes(189residents) ▪ Data collection staff were unblinded to treatment
receivedtheinterventionand18homes(209residents) allocationin15ofthe35homes.
receivedusualcare.Groupsdidnotdiffersignificantly ▪ Therewasvariationinthe numberofElderClown
overtimeontheprimaryoutcomeofdepression,oron visits and LaughterBoss initiated humour
behaviouraldisturbancesotherthanagitation,social receivedbyindividualresidents.
engagementandqualityoflife.Thesecondaryoutcome ▪ Thestudyhadlowerpowertodetecteffectsthan
ofagitationwassignificantlyreducedintheintervention expectedduetohigherthanexpectedintracluster
Fornumberedaffiliationssee groupcomparedwithcontrolsover26weeks(timeby correlations.
endofarticle groupinteractionadjustedforcovariates:p=0.011).The
meandifferenceinchangefrombaselineto26weeksin
reducedepressionbutsignificantlyreducedagitation.
Blom-transformedagitationscoresafteradjustmentfor
Correspondenceto Trialregistration:AustralianNewZealandClinicalTrials
covariateswas0.17(95%CI0.004to0.34,p=0.045).
DrLee-FayLow;
Registry-ACTRN12611000462987.
[email protected] Conclusions:Humourtherapydidnotsignificantly
LowL-F,BrodatyH,GoodenoughB,etal.BMJOpen2013;3:e002072.doi:10.1136/bmjopen-2012-002072 1
A cluster randomised trial of humour therapy in nursing homes
INTRODUCTION TheprimaryaimoftheSydneyMultisiteInterventionof
Humour is a fundamental form of social play with many LaughterBosses and ElderClowns (SMILE) (Australian
psychological benefits: facilitating positive emotions that New Zealand Clinical Trials Registry number
in turn may increase problem solving and memory effi- ACTRN12611000462987) was to evaluate the effectiveness
ciency; facilitating social communication, social influ- of humour therapy in improving mood in individual resi-
ence and bonding and tension relief and coping with dents living in nursing homes. Secondary aims were to
anxiety.1 Humour therapy involves using humour to investigatetheeffectivenessofhumourtherapyondecreas-
facilitate laughter and happiness. The humour can be ing agitation and other behavioural disturbances, and
delivered by an individual (eg, by a clown) or recorded increasing quality-of-life and social engagement of resi-
stimuli (eg, a video) or participants can be shown how dents.Weutilisedaclustereddesignwithrandomisationat
to generate humour themselves. Anecdotally, humour thefacilitylevelastheinterventioninvolvedtrainingstaffto
interventions result in positive outcomes; however, the changetheirbehaviourtowardsresidents,whichisdifficult
research evidence is limited. Small experimental studies tocontaintoonlycertainresidentswithintheircare.
with randomised or quasi-randomised designs have
reported beneficial effects of humour interventions on a
METHODS
diverse range of outcomes, such as: pregnancy rates
during in vitro fertilisation,2 reducing preoperative A single-blind two-group longitudinal cluster rando-
anxiety in children,3 lowering diastolic blood pressure, mised controlled design was used in SMILE to evaluate
humour therapy in Australian nursing homes. The Study
respiratory frequency and temperature in children with
respiratory pathologies,4 and in decreasing depression protocol details have been reported previously.16 Ethics
and insomnia in older persons.5 6 While people living approval was obtained from the University of New South
Wales Human Research Ethics Committee (approval
with dementia appreciate and can express themselves
with humour,7 8 there are seldom opportunities to number 08345). Residents either provided written
experience humour when living in anursing home.9 consent, if judged by the researcher as being able to
Five of the six studies evaluating the efficacy of understand information about the study and make an
informed decision about participation, or verbal assent,
humour therapy in nursing home residents reported
some benefit,10–15 albeit limited by small sample sizes, if incapable of providing written consent, with written
relatively low-quality methodology and lack of flexibility consentobtained from aproxy.
in matching humour techniques to residents’ abilities
and preferences.10 15 A non-controlled study (n=21) of Participants
four clown sessions, evaluated using a modified demen- Invitations were made to 228 nursing homes located in
tia care mapping protocol, found an overall increase in Sydney, Australia. Eligible homes were: government
positive behaviours and decrease in negative behaviours accredited; located within greater metropolitan Sydney
during the sessions in persons with severe dementia; within 1h drive from the Universityof New South Wales,
however, therewas no follow-up outside the sessions.10 A Sydney; not enrolled in another intervention study; not
non-randomised controlled study (n=61) found that five catering for particular ethnic and minority groups or
sessions of comical singing and dancing were associated medical conditions except for dementia; not scheduled
with decreased self-rated anxiety and depression in for change in management, renovation or programme
nursing home residents.11 Fortnightly group humour delivery; able to secure site-specific governance permis-
therapy for psychiatric in-patients with either sion and in agreement with study terms. After screening,
Alzheimer’s disease, or late-life depression, did not the first 36 eligible homes were randomised to an inter-
improve quality-of-life for either group in addition to vention or a control group. Eight additional eligible
standard pharmacotherapy (n=20).15 Nursing home resi- homes on a waiting list replaced homes that withdrew
dents (n=87) randomly assigned to watch humorous before baseline. One home that began the intervention
movies three times a week showed decreased negative and was subsequently found to be ineligible because it
affect following the 12-week intervention compared with catered specifically to persons with mental illnesses and
both those residents who watched serious movies and drug and alcohol addiction was immediately excluded.
controls who received usual care.12 Residents (n=27) Data collection occurred in six rolling cycles between
randomly assigned to watch recordings of humorous July 2009 and May 2011.
story-telling weekly for 12weeks reported improved Within each home, a geographical SMILE neighbour-
quality-of-life compared to those who watched conven- hood was defined following discussion with facility man-
tional television and those who received usual care.13 agers. Criteria were that the area accommodated <40
Eight weekly sessions involving telling jokes, and funny residents, and permitted interaction between residents
stories and discussions on prioritising humour in daily and the staff member tobe trained in humour therapy.
life decreased reports of pain and perceived loneliness, All eligible residents within SMILE neighbourhoods
and increased reported happiness and life satisfaction in were invited to participate. Residents were ineligible if
residents in the intervention group (n=36) compared they were <50years of age, admitted to full-time care
withcontrols (n=34).14 <12weeks prior, exhibiting behaviour presenting a risk to
2 LowL-F,BrodatyH,GoodenoughB,etal.BMJOpen2013;3:e002072.doi:10.1136/bmjopen-2012-002072
A cluster randomised trial of humour therapy in nursing homes
Figure1 Patternofnursinghome(NH)andresidentrecruitmentandparticipation.
studypersonnel,foreshadowedtomove out oftheSMILE 1. One-day LaughterBoss training for each home’s
neighbourhood within 6months, experiencing severe nominated staff member.17 Training covered the evi-
communication obstacles, acutely unwell, under public dence linking humour and health, and practical ways
guardianship with no person responsible to consent on for including humour in daily care. LaughterBosses
their behalf, or if they had florid psychiatric symptoms or assisted during ElderClown visits and were encour-
aknownfearofclownsorstrangers(seefigure1). aged to continue the humour intervention between
and after ElderClown visits.
2. Between 9 and 12 humour therapy sessions by an
Intervention ElderClown;18 a trained performer experienced in
The humour therapy intervention comprised two healthcare settings, who at each session visited with
components: available and willing residents enrolled in the study.
LowL-F,BrodatyH,GoodenoughB,etal.BMJOpen2013;3:e002072.doi:10.1136/bmjopen-2012-002072 3
A cluster randomised trial of humour therapy in nursing homes
ElderClowns tailored their interactions to maximise collectors. Nevertheless nursing home staff, residents or
residentengagement,laughterandenjoyment,adapt- families revealed the blinding for 15 separate homes (5
ing to the background, personality, mood and phys- control, 10 intervention) to one or two data collection
ical and cognitive abilities of the resident. staff, the remaining three data collection staff remained
ElderClowns prepared for their work with individual blinded throughout.
residents,andalsoimprovisedbasedontheresidents’
reactions and to make the conversation and inter- Assessment
action light hearted and playful. Interactions could Datawerecollectedatthreetimepoints:baseline(week0),
be based around music—serenading the resident post (week 13) and follow-up (week 26). Information was
with their favourite song, or adapting a song to collectedfromtheresident (if ableand willing),areliable
include their name, encouraging them to dance with staffinformant,directobservationandfromclinicalfiles.
each other or the LaughterBoss; could involve asking The primary outcome measure was The Cornell Scale
the resident’s advice about a problem that the for Depression in Dementia (CSDD) which is a clinician-
ElderClown has such as whether to buy a cat or a rated depression scale.19 Secondary outcome measures
dog, or be based around jokes relating to a prop were: the Cohen-Mansfield Agitation Inventory (CMAI)20
such as a slip-on rubber thumb with a light at the and the Neuropsychiatric Inventory Nursing Home21 to
end. Interactions could occur in groups and individu- obtain informant ratings of the severity and frequency of
allydepending on the resident’s preferences. 12 behavioural disturbances to assess informant-rated fre-
For instance, a female resident with moderate demen- quency of agitated behaviours; the eight-item withdrawal
tia might have been engaged by an ElderClown asking subscale of the Multidimensional Observation Scale for
her opinion about which hat would suit her best. Hats Elderly Subjects (MOSES) to measure informant rated
varied from a formal dress hat, to a sailor’s hat to a silly social engagement,22 and the DEMQOL to provide proxy
costume hat with attached wig. The resident laughed at andself-ratedmeasureofhealth-relatedqualityoflife.23
howtheElderClownlookedinthehatsandatherexpres- Inaddition,demographicinformationwascollected,func-
sions and comments when wearing them, and made tional impairment was assessed using the Barthel Index,24
some joking comments about the hats. A male resident relative severity of dementia was rated using the Global
with severe dementia might have been engaged with a Deterioration Scale,25 and the appropriateness of the phys-
gameofpretendtennisusingplasticracketsandaballon ical environment for nursing home residents was rated at
a wire. The LaughterBoss held one racket, the resident thefacilitylevelusingtheEnvironmentalAuditTool.26
held the other and the ElderClown controlled the ball, After each ElderClown visit, the LaughterBoss and
making it easier or more difficult for the players to hit ElderClownindependently completed globalratingsof the
the ball, and also providing exaggerated commentaryon level of success in engaging each resident on a 10-point
the game. The resident tried to hit the ball, tracking it scale(1=extremelyunsuccessfulto10=extremelysuccessful).
visually and smiled when he succeeded and was praised
in the Elderclown’s commentary. Information about resi- Analyses
dents was obtained from resident charts and presession Asamplesizeof36homeswith9–10residentsineachwas
briefingsbytheLaughterBoss.Postsessiondebriefingses- estimated to provide over 80% power to detect a medium
sions between the ElderClown and LaughterBoss effect size (Cohen’s d=0.5) difference on continuous out-
involved a discussion on what did and did not work, and comeswithsignificancelevelsetattwo-sided0.05,27based
formulation of ideas for the next visit. The trailer of a on intraclass correlations on the outcome measures
documentary showing ElderClown visits filmed during obtainedfromapreviousmultisitenursinghomestudy.28
SMILEisavailableathttp://thesmilewithin.com.au/. Analyses were conducted on an intention-to-treat basis
by a statistician blind to group allocation using
Randomisation and masking intention-to-treat analysis in SAS V.9.2 software.29 The α
Enrolled homes were assigned a study number by the for significance was set at p=0.05 for all analyses, such
administrative assistant and deidentified characteristics that we had 5% chance of incorrectly rejecting the null
were used for randomisation by the first author. SMILE hypothesis for each test. Baseline differences in charac-
neighbourhoods were stratified by size (<25 beds vs ≥25 teristics and outcome measures between groups were
beds) and care level (high care vs low care). A random examined using t tests for normally distributed continu-
number generator in Excel was used to assign homes to ous data, χ2 tests for categorical data or Mann-Whitney
intervention and control groups. U tests for non-normally distributed continuous data.
Only one investigator (LFL) and the administrative Restricted maximum likelihood-based multilevel linear
assistant were aware of treatment allocation before base- mixed models were employed to estimate the interven-
line assessment at each facility. Nursing homes and tion effect, taking into account both within-resident cor-
humour intervention staff were notified of treatment relation (repeated measures) and within-cluster (nursing
group by the administrative assistant after baseline. home) correlation of the endpoints via random effect
Reminders were given to nursing home staff prior to specification.30 Baseline response was adjusted using an
each occasion of contact to maintain blinding of data approach recommended by Fitzmaurice et al.31 All
4 LowL-F,BrodatyH,GoodenoughB,etal.BMJOpen2013;3:e002072.doi:10.1136/bmjopen-2012-002072
A cluster randomised trial of humour therapy in nursing homes
outcome measures were continuous, and data from all (7 low care, 10 high care). Baseline characteristics of the
three time points were included as endpoints in the intervention and control groups are shown in Table 1.
model. Intervention by time interaction was estimated to There were no significant differences on demographic
test for global group differences and for global change characteristics between the groups. Intervention group
over time between groups. Potential confounding charac- residents were taking slightly more regular psychotropic
teristics of homes and residents were included as cluster- medications on average and were rated by staff as having
level or individual-level fixed or time-varying covariates higherlevelsofagitationontheCMAI.
and retained if evidence of confounding occurred (ie, if A total of 191 humour therapy sessions were delivered
estimates of treatment effect differed substantially in the (average of 11, SD=1 per facility), with individual resi-
adjusted vs unadjusted models, or they explained signifi- dentsreceivinganaverageof9(SD=3)ElderClownvisits.
cant variation in the outcomes). The Blom transform- Table 2 shows the raw mean scores by group and the
ationwasusedtoaccountfordataskewness.32 three assessment occasions for the five resident outcome
The effect of engagement dose on outcomes was measures, the model adjusted mean difference in change
explored further within the intervention group. Dose between groups, and the primary analysis results.
was calculated as the total of engagement score for all Depression and social engagement decreased and
visits, averaged across LaughterBoss and ElderClown resident-rateddementiaquality-of-lifeincreasedovertime,
ratings which were highly correlated (r =0.863, but the group by time interactions on depression, non-
(1475)
p<0.001). Statistical analysis was conducted in a similar agitation behavioural disturbance, social engagement or
manner tothe primaryanalysis as outlined above. resident-rated or proxy-rated quality-of-life were non-
significant (p>0.05). The group-by-time interaction was
significantforagitationmeasuredusingtheCMAI,before
RESULTS andafteradjustmentforcovariates(p<0.05).Theadjusted
Figure 1 shows trial recruitment and participant flow to mean differences of change based on Blom-transformed
achievethefinalanalysis sampleof209usual carecontrol scoresindicatesthatthehumourtherapygroupdecreased
groupresidentsfrom18homes(7lowcare,11highcare) on the CMAI by 0.17 (95% CI 0.004 to 0.34; p=0.045)
and 189 humour therapy group residents from 17 homes pointsmorethancontrolsbetweenbaselineandfollow-up,
Table1 BaselinecharacteristicsofresidentsandSMILEneighbourhoodsbygroup
Usualcarecontrols(n=209, Humourtherapy(n=189,
18facilities) 17facilities) Teststatistic
SMILEneighbourhoods
Numberofresidents 22.1±8.6 19.1±7.7 t =1.08,p=0.29
(33)
Numberofresidentswho 10.8±3.3 11.3±2.3 t =0.67,p=0.51
(33)
participated
Residents
Ageinyears 84.5±8.7 84.5±7.5 t =−0.09,p=0.93
(396)
Numberoffemales 161(77.0%) 146(77.2%) Χ2 =0.00,p=0.96
(1)
Numberwithdementiadiagnoses 165(78.9%) 145(76.7%) Χ2 =0.29,p=0.59
(1)
inchart
Yearslivedincare 2.7±2.8 2.8±3.1 U=19153.0,p=0.66
GlobalDeteriorationScale 5.0±1.2 5.0±1.2 U=19573.0,p=0.87
NumberwithEnglishasafirst 204(97.6%) 186(98.4%) Χ2 =0.32,p=0.57
(1)
language
Numberofregularpsychotropic 1.0±1.0 1.2±1.0 U=17484.0,p=0.045
medications
Barthel 41.5±24.5 42.3±25.2 U=19397.5,p=0.76
CornellScaleforDepressionin 7.8±5.6 8.5±6.1 U=18565.5,p=0.34
Dementia
Cohen-MansfieldAgitation 38.9±11.0 45.3±20.0 U=16897.0,p=0.012
Inventory
NeuropsychiatricInventory 18.7±16.9 22.3±21.7 U=18771.5,p=0.39
DEMQOL-resident 89.9±13.8 89.4±15.5 U=8818.5,p=0.89
DEMQOL-proxy 106.0±13.4 103.5±11.1 U=17266.0,p=0.075
MOSESsocialengagement 18.1±6.2 17.4±6.0 U=18401.5,p=0.24
subscale
FiguresaremeansandSDsofscoresunlessotherwisestated.
SMILE,SydneyMultisiteInterventionofLaughterBossesandElderClowns;MOSES,MultidimensionalObservationScaleforElderly
Subjects.
LowL-F,BrodatyH,GoodenoughB,etal.BMJOpen2013;3:e002072.doi:10.1136/bmjopen-2012-002072 5
6
A
c
l
u
s
t
e
r
r
a
n
Table2 Effectofhumourtherapyonoutcomemeasuresovertimeatresidentlevel d
o
Adjustedmean m
difference Adjustedmeandifference Intracluster i
s
Baseline Follow-up baseline-post (95% baseline-follow-up(95% correlation e
d
(n=398)* Post (n=371)* (n=343)* CI) pValue CI) pValue coefficient
t
r
i
Depression(CSDD) p =0.68,p <0.01,p =0.88;p =0.50,p <0.01,p =0.89 a
Control 7.G8±5.6 T 6.5±4.G5T Gc 6.3±5.4Tc 0G.0T0c6(−0.19to0.20) 0.95 0.046(−0.18to0.27) 0.69 0.12 lo
Intervention 8.5±6.1 6.9±5.2 6.4±4.8 f
h
Agitation(CMAI) p =0.33,p =0.22,p =0.01;p =0.20,p =0.02,p =0.01 u
G T GT Gc Tc GTc
Lo Control 38.9±11.0 37.9±10.0 39.0±11.7 −0.04(−0.18to0.11) 0.61 0.17(0.004to0.34) 0.045 0.15 m
w o
L Intervention 45.3±20.0 43.4±19.1 42.0±18.3 u
-F,Brod Bdieshtuarvbiaonucrael(NPI) pG=0.69,pT=0.63,pGT=0.07;pGc=0.47,pTc=0.49,pGTc=0.09 rthe
aty Control 18.7±16.9 19.3±15.7 18.1±16.8 0.05(−0.11to0.22) 0.52 −0.15(−0.34to0.04) 0.13 0.18 ra
H Intervention 22.3±21.7 20.0±20.3 23.2±22.0 p
,Goode S(MoOciSalEeSn)gagement pG=0.44,pT<0.01,pGT=0.62;pGc=0.41,pTc<0.01,pGTc=0.45 yin
nou Control 18.2±6.0 18.2±6.0 18.7±6.3 −0.046(−0.21to0.12) 0.58 0.049(−0.13to0.22) 0.59 0.12 nu
gh Intervention 17.3±6.0 17.6±6.4 18.1±6.1 rs
B,e Resident-ratedquality pG=0.72,pT<0.01,pGT=0.29;pGc=0.51pTc=<0.01,pGTc=0.41 in
t g
a oflife(DEMQOL-res)
.lBM Control 89.9±13.8 92.9±12.7 92.5±15.4 −0.10(−0.31to0.11) 0.34 0.05(−0.18to0.28) 0.67 0.06 ho
J Intervention 89.4±5.5 93.7±13.1 92.0±14.0 m
Ope Staff-rated pG=0.13,pT=<0.01,pGT=0.44;pGc=0.11,pTc=<0.01,pGTc=0.40 es
n
2 quality-of-life
0
1 (DEMQOL-proxy)
3
3;:e Control 106.0±13.4 104.5±16.3 103.2±11.8 0.07(−0.16to0.31) 0.53 −0.07(−0.28to0.13) 0.48 0.40
0
0 Intervention 103.5±11.1 100.6±14.9 101.4±11.7
2
0
7 Adjustedmeanscoresarebasedonthestandardised(Blom-transformed)scores.Positivescoresindicateimprovement.
2
.d pValuesfrommixedmodelswithallthreetimepointsincludedasoutcomesbutnocovariates:pGisformaineffectofintervention,pTisformaineffectoftime,pGTisinteractionofgroup×time.
oi:1 pValuesformixedmodelsincludingsignificantcovariates:pGcisformaineffectofintervention,pTcisformaineffectoftime,pGTcisinteractionofgroup×time.
0 SignificantcovariatesforCSDDwereage,GDS,Barthelandtimeincare;forCMAIwereageandGDS;forNPIwereage,GDS,dementiadiagnosis;forMOSESweregender,GDS,Barthel;for
.11 DEMQOL-reswereBarthel,andtimeincare;forDEMQOL-proxywastimeincare.
3
6 *Dataarerawnon-transformedscores.
/bm CMAI,Cohen-MansfieldAgitationInventory;CSDD,CornellScaleforDepressioninDementia;GDS,GlobalDeteriorationScale;NPI,NeuropsychiatricInventory.
jo
p
e
n
-2
0
1
2
-0
0
2
0
7
2
A cluster randomised trial of humour therapy in nursing homes
thedifferenceinrawscoreswas2.52(95%CI0.20to5.32, variations between homes in the amount of humour
p=0.07). The difference in change from post to follow-up initiated by LaughterBosses outside ElderClown sessions.
on the CMAI was statistically significant with the adjusted LaughterBosses were not tested for competency in deli-
mean difference being 0.21 (95% CI 0.07 to 0.35; vering humour. Fourth, the two groups were unbalanced
p=0.003) the difference in raw scores was 2.95 (95% CI at baseline on several outcome measures, these differ-
0.89to5.02,p=0.005).Thereweresignificantengagement enceswereadjustedforinourstatisticalmodelling.Fifth,
dose-by-time interactions for depression (F =6.72, adjustment for multiple comparisons was not made to
(2,496)
p=0.00), behavioural disturbance (F =3.49, p=0.03) the α for significance, as outcome measures were corre-
(2, 497)
and resident-rated quality-of-life (F =3.39, p=0.03), lated. Sixth, the intracluster correlations for this sample
(2,337)
but not for the other outcome measures. Residents who were much higher than reported in the study on which
experienced higher doses of engagement showed greater our power calculations were based, which resulted in
improvementon depression, behaviouraldisturbanceand lowerpowerthanplanned.
resident-ratedquality-of-life. Several explanations are possible for the lack of effect
No adverse events were reported after ElderClown ses- on outcome measures other than CMAI agitation.
sions or generally in relation to humour therapy with Depression, other forms of behavioural disturbances
LaughterBosses. and self-rated quality-of-life all improved more in resi-
dents who experienced higher doses of engagement as a
result of humour therapy, suggesting that humour
DISCUSSION therapy does change these outcomes, even though there
While there was no significant benefit on the primary was not a statistical advantage of intervention over
outcome measure, depression, agitation levels decreased control groups. There was a floor effect: only 29% of
significantly over time with humour therapy compared our sample was assessed as having probable or possible
with usual care with the mean adjusted change differ- depression on the CSDD, and 28% of our sample was
ence between baseline and follow-up being 2.52 points rated as not having any agitation symptoms on the
—this would be equivalent to two agitated behaviours CMAI, thereby limiting the potential for improvement.
decreasing in frequency from daily to once a week. The DEMQOL was developed for persons with mild to
Difference between treatment and control groups on severe dementia; however, 33% of residents were unable
change scores on the CMAI pooled across three rando- tocomplete theself-reportversion. Whileweinterviewed
mised controlled trials of risperidone were 3. (95% CI staff members who knew the resident well, it was difficult
1.78 to 4.22).33 In decreasing agitation, humour therapy for staff to be aware of the quality-of-life related con-
had a similar effect to risperidone, the most commonly cerns of many residents, particularly those with poor
used antipsychotic in Australia for the treatment of verbal skills. Our data and others showed that DEMQOL
behavioural disturbance in dementia.34 Humour therapy proxy scores correlate only mildly or moderately with
showed none of the side effects of risperidone. When DEMQOL self-report. The measures used might not
adjustments were made for the ‘dosage’ of humour have been sensitive to anecdotally reported positive
therapy engagement, humour therapy demonstrated effects such as increased positive mood, and increased
benefitson depression, behaviouraldisturbance and self- initiation of and participation insocialactivities.
reported dementia quality-of-life effects not reported We offered a novel model of humour therapy delivery
withmedication. combining staff training (LaughterBosses) and profes-
The strengths of SMILE include: a large sample, clus- sional humour therapists (ElderClowns). This was
tered design and relatively high follow-up rates. We were designedtominimisecostsanddeliversustainablepractice
abletoimplementourinterventioninreal-worldnursing change within nursing homes. We believe that the active
homes, despite initial reservations from some managers ingredients of the programme are engagement and play.
and staff. The sustained benefits in agitation at follow-up This is supported by our engagement dose analyses and
underscore the importance of recruiting staff members consistent with a needs-driven behaviour explanatory
into the programme. Limitations are noted. First, data model of behaviours.35 Increasing social contact and
collection staff became ‘unblinded’ over time for 15 of opportunities for play could have fulfilled a need for
the 35 homes despite constant reminders to nursing stimulation which is expressed through agitation. Other
homestaffaheadofdatacollectionperiodsoftheimport- tailored models of engaging and stimulating residents
ance of maintaining confidentiality of intervention allo- might also be successful.36 Challenges in delivering the
cation.Second,participatinghomesmightnothavebeen programme predominantly related to false expectations
representative of Australian nursing homes. Compared that, rather than attempting to elicit individual responses
with national data, residents in our sample were similar from residents enrolled in the study, the ElderClown
in average age and dependency level, but were 6% more would perform for any audience. Some suggest that the
likely to be female, required less help with activities of label‘ElderClown’mightnotbeappropriateindescribing
daily living and had higher levels of behavioural distur- theworkoftheprofessionalperformers.
bances. Third, therewerevariations between residents in SMILE demonstrated that there are benefits to
the number of ElderClown sessions they received, and increasing positive interactions for residents. If our
LowL-F,BrodatyH,GoodenoughB,etal.BMJOpen2013;3:e002072.doi:10.1136/bmjopen-2012-002072 7
A cluster randomised trial of humour therapy in nursing homes
resultsarereplicated,considerationshouldbegiventosys- 8. HenryJD,RendellPG,SciclunaA,etal.Emotionexperience,
expression,andregulationinAlzheimer’sdisease.PsycholAging
tematically introducing humour therapy as a psychosocial
2009;24:252–7.
intervention to reduce the level of agitation in nursing 9. IsolaA,Astedt-KurkiP.Humourasexperiencedbypatients
homeresidents.Moreresearchisneededintotheefficacy andnursesinagednursinginFinland.IntJNursPract
1997;3:29–33.
ofdifferentmodelsofdeliveringhumourtherapy. 10. ThomsonR.Evaluationoftheuseofaclowntherapygroupwith
dementiasufferers.NHSBordersPychologicalServ2005:3–4.
11. HoustonDM,McKeeKJ,CarrollL,etal.Usinghumourtopromote
Authoraffiliations
psychologicalwellbeinginresidentialhomesforolderpeople.Aging
1DementiaCollaborativeResearchCentre,CentreforHealthyBrainAgeing, MentalHealth1998;2:328–32.
SchoolofPsychiatry,UniversityofNewSouthWales,Sydney,Australia 12. BoydR,McGuireF.Theefficacyofhumorinimproving
2HumourFoundation,Chatswood,NewSouthWales,Australia psychologicalwell-beingofresidentsoflong-termcarefacilities.
3ArtsHealthInstitute,Newcastle, NewSouthWales,Australia JLeisurability1996;23:1–15.
4NSW&ACTDementiaTrainingStudyCentre,UniversityofWollongong, 13. RonnbergL.Qualityoflifeinnursing-homeresidents:anintervention
studyoftheeffectofmentalstimulationthroughanaudiovisual
Wollongong,NewSouthWales,Australia programme.AgeAgeing1998;27:393–7.
5HealthandAgeingResearchUnit,UniversityofTechnology,Sydney, 14. TseMY,LoAPK,ChengTLY,etal.Humortherapy:relievingchronic
NewSouthWales,Australia painandenhancinghappinessforolderadults(ArticleID343574).
JAgeingRes2010.doi:10.4061/2010/343574
15. WalterM,HänniB,HaugM,etal.Humourtherapyinpatientswith
Acknowledgements TheauthorswouldliketothanktheSMILEresearchand late-lifedepressionorAlzheimer’sdisease:apilotstudy.IntJGeriatr
humourtherapyteams,andtheparticipatingnursinghomes,residents, Psychiatry2007;22:77–83.
familiesandstaff. 16. GoodenoughB,LowL-F,CaseyA-N,etal.Studyprotocolfora
randomizedcontrolledtrialofhumortherapyinresidentialcare:the
Contributors Allauthorshadfullaccesstoallofthedata(includingstatistical
SydneyMultisiteInterventionofLaughterBossesandElderClowns
reportsandtables)inthestudyandtakeresponsibilityfortheintegrityofthe (SMILE).IntPsychogeriatr2012;24:2037–44.
dataandtheaccuracyofthedataanalysis.HB,PS,L-FL,LCandRF 17. SpitzerP.TheLaughterBoss.In:AdamsT, LeeH.eds.Creative
participatedinstudyconceptanddesign.BG,A-NC,PBandPSparticipated approachesindementiacare.NewYork:PalgraveMacMillan,
inacquisitionofdata.L-FL,ZLandHBparticipatedinstatisticalanalysisand 2011:32–53.
18. WarrenB,SpitzerP.Theartofmedicine:laughingtolongevity—the
interpretationofdata.L-FL,HB,BG,PS,PB,RF,A-NC,ZLandLCparticipated workofelderclowns.Lancet2011;378:562–3.
incriticalrevisionofthemanuscriptforimportantintellectualcontent.HB, 19. AlexopolousG,AbramsR,YoungR,etal.Scalefordepressionin
L-FL,LC,RFandPSobtainedfunding.HB,L-FLandBGparticipatedinstudy dementia.BiolPsychiatry1988;23:271–84.
supervision. 20. Cohen-MansfieldJ.Adescriptionofagitationinanursinghome.
JGerontol1989;44:M77–84.
Funding TheNationalHealthandMedicalResearchCouncilhadnorolein 21. CummingsJL,MegaM,GrayK,etal.Theneuropsychiatric
studydesign,collection,analysisorinterpretationofdata,writingordeciding inventory:comprehensiveassessmentofpsychopathologyin
tosubmitthispaperforpublication. dementia.Neurology1994;44:2308–14.
22. HelmesE,CsapoK,ShortJ-A.Standardizationandvalidationofthe
Competinginterests AllauthorshavecompletedtheUnifiedCompeting MultidimensionalObservationScaleforElderlySubjects(MOSES).
Interestformathttp://www.icmje.org/coi_disclosure.pdf(availableonrequest JGerontol1987;42:395–405.
fromthecorrespondingauthor)anddeclare:fundingforthesubmittedwork 23. BanerjeeS,SmithSC,LampingDL,etal.Qualityoflifeindementia:
morethanjustcognition.Ananalysisofassociationswithqualityof
fromtheNationalHealthandMedicalResearchCouncilgrants568787and
lifeindementia.JNeurolNeurosurgPsychiatry2006;77:146–8.
455377;PSandJPBbothreceivedpaymentsfordeliveryofthehumour
24. MahoneyFI,BarthelDW.Functionalevaluation:theBarthelIndex.
therapyinterventionfromthenot-for-profitorganisationTheHumour MdStateMedJ1965;14:61–5.
Foundation;PSistheMedicalDirectorandboardmemberofTheHumour 25. ReisbergB,FerrisSH,deLeonMJ,etal.TheGlobalDeterioration
Foundation.JPBandLFLareboardmembersforthenot-for-profitArtsHealth Scaleforassessmentofprimarydegenerativedementia.Am
Institutewhichalsoprovideshumourtherapyservices;nootherrelationships JPsychiatry1982;139:1136–9.
26. FlemingR.Anenvironmentalaudittoolsuitableforuseinhomelike
oractivitiesthatcouldappeartohaveinfluencedthesubmittedwork. facilitiesforpeoplewithdementia.AustralasJAgeing2011;30:108–12.
Ethicsapproval UniversityofNewSouthWalesHumanResearchEthics 27. HayesRJ,BennettS.Simplesamplesizecalculationfor
Committee—approvalno.08345; cluster-randomizedtrials.IntJEpidemiol1999;28:319–26.
28. ChenowethL,KingMT,JeonYH,etal.CaringforAgedDementia
Provenanceandpeerreview Notcommissioned;externallypeerreviewed. CareResidentStudy(CADRES)ofperson-centredcare,
dementia-caremapping,andusualcareindementia:a
Datasharingstatement Therearenoadditionaldataavailable. cluster-randomisedtrial.LancetNeurol2009;8:317–25.
29. SASInstituteInc.BaseSAS®V9.2proceeduresguide.Cary,NC:
SASInstituteInc,2011.
30. MurrayDM.Designandanalysisofgroup-randomisedtrials.Oxford,
REFERENCES UK:OxfordUniversityPress,1998.
1. MartinR.Thepsychologyofhumour:anintegrativeapproach. 31. FitzmauriceG,LairdN,WareJ.Appliedlongitudinalanalysis.New
Burlington,MA:ElsevierInc,2007. Jersey:JohnWiley&Sons,2004.
2. FriedlerS,GlasserS,AzaniL,etal.Theeffectofmedicalclowning 32. BlomG.Statisticalestimatesandtransformedbetavariables.
onpregnancyratesafterinvitrofertilizationandembryotransfer. NewYork:JohnWiley&Sons,1958.
FertilSteril2011;95:2127–30. 33. SchneiderLS,DagermanK,InselPS.Efficacyandadverseeffects
3. CostaFernandesS,ArriagaP.Theeffectsofclowninterventionon ofatypicalantipsychoticsfordementia:meta-analysisof
worriesandemotionalresponsesinchildrenundergoingsurgery. randomized,placebo-controlledtrials.AmJGeriatrPsychiatry
JHealthPsychol2010;15:405–15. 2006;14:191–210.
4. BertiniM,IsolaE,PaoloneG,etal.Clownsbenefitchildren 34. SnowdonJ,GalanosD,VaswaniD.Patternsofpsychotropicmedication
hospitalizedforrespiratorypathologies.EvidComplementAlternat useinnursinghomes:surveysinSydney,allowingcomparisonsover
Med2011.doi:10.1093/ecam/neq064. timeandbetweencountries.IntPsychogeriatr2011;23:1520–5.
5. ShahidiM,MojtahedA,ModabberniaA,etal.Laughteryogaversus 35. AlgaseDL,BeckC,KolanowskiA,etal.Need-driven
groupexerciseprograminelderlydepressedwomen:arandomized dementia-compromisedbehavior:analternativeviewofdisruptive
controlledtrial.IntJGeriatrPsychiatry2011;26:322–27. behavior.AmJAlzheimer’sDisOtherDemen
6. KoHJ,YounCH.Effectsoflaughtertherapyondepression, 1996;11:10–19.
cognitionandsleepamongthecommunity-dwellingelderly.Geriatr 36. GitlinLN,WinterL,BurkeJ,etal.Tailoredactivitiestomanage
GerontolInt2011;11:267–74. neuropsychiatricbehaviorsinpersonswithdementiaandreduce
7. MoosI.Humour,ironyandsarcasminsevereAlzheimer’sdementia caregiverburden:arandomizedpilotstudy.AmJGeriatrPsychiatry
—acorrectivetoretrogenesis?AgeingSoc2011;31:328–46. 2008;16:229–39.
8 LowL-F,BrodatyH,GoodenoughB,etal.BMJOpen2013;3:e002072.doi:10.1136/bmjopen-2012-002072