View
223
Download
3
Category
Tags:
Preview:
Citation preview
Klinimetrie bij het stellen van de functionele prognose na een CVA:
hulp of last?
Dr. G. Kwakkel
Identify patient’s problem
Define a meaningful question
Determine the prognosis
Select most appropriate therapy
Evaluate efficacy
Doel van het (klinisch) meten
om te onderscheiden (diagnostiseren en/of klassificeren)
Kirshner & Guyatt, J. Chron. Dis 1985; 38: 27
om te voorspellen
om (verandering) te evalueren
home at 6 months ?
outcome of ADL ?
needs further help ?
clinical decision making
pa
tie
nts
fu
ture
home at 6 months?
outcome of ADL ?
needs further help?
pattern recognition
hypothetico-deductive reasoning?
problem solving ?
intuition ?
pa
tie
nts
fu
ture
determinantOutcome of
what?
(1) Prediction of what?
Health Condition Health Condition ((disorder/diseasedisorder/disease))
Interaction of ConceptsInteraction of ConceptsICF 2001ICF 2001
Environmental Environmental FactorsFactors
Personal Personal FactorsFactors
Body Body function&structure function&structure
(Impairment(Impairment))
ActivitiesActivities(Limitation)(Limitation)
ParticipationParticipation(Restriction)(Restriction)
? ?
Neuro-physiology
Neuroradiology
Clinical neurology
Neuro-psychology
Demographic factors
determinant
Functional outcome (e.g., dexterity, walking ability, (I)ADL-independency
Which construct (at level of activity) do we exactly like to predict?
Basic ADLArm-
handvaardigheid
Barthel Index ARAT
?
Loopvaardigheid
Functional Ambulation Categories
Construct validity of BI (N=89)
0
0,2
0,4
0,6
0,8
1
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26
FAC
weeks
Correlation coefficient (rs)
Construct validity of BI (N=89)
0
0,2
0,4
0,6
0,8
1
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26
FACARA
weeks
Correlation coefficient (rs)
Neuro-physiology
Neuroradiology
Clinical neurology
Neuro-psychology
Demographic factors
determinantFunctional outcome of basic ADLs
(2) Which determinants are valid?
I . Key methodological criteria for prognostic research
internal validity reliable and valid measurements inception cohort appropriate end-points of observation control of patient drop-out
statistical validity control for statistical significance adequate estimation of sample-size control for multicollinearity
Kwakkel et al, Age & Ageing 1996;25:479-489
Factor x2
Factor x1
Outcome Y(explained variance)
r y.x2
r y.x1
r x1.x2
Predictive value of volume of stroke according to MRI scan for outcome of ADL-independency at 6 months post stroke
Schiemanck et al, Stroke. 2006;37:1050-1054
Copyright ©2006 American Heart AssociationSchiemanck, S. K. et al. Stroke 2006;37:1050-1054
Receiver operating curves of model 1 (clinical) and model 2 (clinical imaging) (N=75)
Model 1 (AUC=0.83) = age and IBI
Model 2 (AUC =0.87) = Model 1 + volume MRI
Schiemanck et al, Stroke. 2006;37:1050-1054.
II. Key methodological criteria for prognostic research
external validity specification of inclusion and exclusion criteria description of additional treatment effects during period of
observation cross-validation of the prediction model
Kwakkel et al, Age & Ageing 1996;25:479-489
Internal validity
78
Statistical validity
external validity
13 (8)
0
Valid predictors for ADL (and walking ability)
admission ADL (i.e., assessment specific) urinary (in)continence age* previous stroke (and other disabling co-morbidity) consciousness at onset severity of paresis sitting balance orientation in time and place level of social support inattention depression
Possible negative predictors for ADL
homonymous hemianopia conjugate deviation of the eyes fatigue dyspraxia dysphasia ??
Variables not related to outcome of ADL
gender ethnic origin side of stroke
Kwakkel et al., Age & Ageing, 1996: 25:479-489
Individual recovery patterns of Barthel Index (n=13)
0
2
4
6
8
10
12
14
16
18
20
22
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26
weeks
BI-score
weeks
Mean recovery pattern of Barthel Index
0
10
20
30
40
50
60
70
80
90
100
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26
weeks
Barthelindex
prediction
outcome
Regression model statistics for outcome of BI
Intercept Initial BI Sitting balance Soc. Support age *
Model Nh Pooled
R-square
0.51 0.61 0.64 0.67
0.52 0.57
Eigenvalue 4.291 0.363 0.226 0.110 0.023
0.106 0.120
CI 1.0 3.44 4.36 6.24 13.66
5.71 5.34
Final regression model for outcome of Barthel Index at 6 months post stroke
BI = 42.29 + (0.51 * IBI) + (20.93 * SB) + (10.35 * SS)
BI= (Initial) Barthel Index (ranging from 0-100)
SB=initial sitting balance (yes/no on the TCT)
SS= Social Support (yes/no: having a partner able to support)
Increment in explained variance for outcome of BI score (N=102)
0
10
20
30
40
50
60
70
80
90
100
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26weeks
model retesting
Explained variance (%)
Effects of initial BI on outcome at 6 months post stroke (N=89)
0123456789
1011121314151617181920
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26
weeks
Barthelindex
Adjusted R2=0.50 (Initial BI)
0
20
40
60
80
100
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26
weeks
Barthel In
dex
stairs
bowel
grooming
bladder
feeding
transfer
toilet use
mobility
dressingbathing
Coefficient of Scalability: 0.72 (week 26) < CS <0.85 (week3)
Rest. Neurology & Neuroscience 2004;22: 281-299
Copyright ©2006 American Heart Association
van Hartingsveld, F. et al. Stroke 2006;37:162-166
Logit item step difficulties ({beta}I) of the Rasch homogeneous 8-item Barthel scale
Van Hartingsveld et al, Stroke. 2006;37:162-166.
Logit item step difficulties (I) of the Rasch homogeneous 8-item Barthel scale (N=559).
1.4
1.21.0
0.80.6
0.40.2
0-0.2
-0.4-0.6
-0.8-1.0
-1.2-1.4 feeding
transfer step 1
groom
dress step 1
toilet
mobility step 2
dress step 2
stairs
bathing
easy
difficult
Take home message:
• Barthel Index gemeten in de eerste 2 weken na een CVA is een robuuste determinant voor het uiteindelijk te verwachten herstel op de BI na 6 maanden.
• Een klinimetrische testuitslag krijgt pas een prognostische
(meer)waarde wanneer men deze relateert aan het moment waarop het CVA heeft plaatsgevonden.
• Functionele prognostiek is pas mogelijk wanneer men eveneens kennis heeft over de psychometrische eigenschappen van gebruikte meetinstrumenten.
Mijn dank voor uw aandacht!
Clinical assessments increase the transparency in making client-related decisions within a team of professionals working together as a stroke team.
Advantage of clinimetrics (2):
Increment in explained variance for outcome of BI, FAC and ARA score (N=102)
0
10
20
30
40
50
60
70
80
90
100
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26
weeks
explained variance of model
consensus in clinimetrics:What do we measure systematically?How do we measure systematically?Who is measuring what?When do we measure the stroke patient?
Steps to follow for getting relation coordination
0
10
20
30
40
50
60
70
80
90
100
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26
weken
% h
ers
tel
Hospital Rehabilitation centre
Home/ Nursing house
0
10
20
30
40
50
60
70
80
90
100
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26
weken
% h
ers
tel
?
Hospital Rehabilitation centre
Home/ Nursing house
0
10
20
30
40
50
60
70
80
90
100
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26
weken
% h
ers
tel ?
‘learning from making functional prognosis’
Hospital Rehabilitation centre
Home/ Nursing house
Clinimetrics (ICF 2001)
pathology impairment disability handicap
OCSP
Stroke type
Number of strokes
Epilepsy
HSP, GHS
SHS
MI-score
FM-motor score
Ashworth Scale
Thumb-Finding Test
Letter cancellation task, line-bisection task
MMSE
Scan. Stroke Scale
Trunk Control Test
Berg-Balance Scale
Timed-Balance test
Timed-Get-up & Go-Test
FAC, 10-meter walking test
ARA, Frenchay Arm
Barthel Index
FAI, EADL
SIP-68
NHP-part 1
Post-stroke Depression
Carer Strain Index
Satisfaction Questionnaires
activities participation
Construct validiteit van de BI (N=89)
0
0,2
0,4
0,6
0,8
1
1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 26
ARA
weeks
Pearson correlation coefficients with Barthel Index
Clinimetrics
objectivity
communication
relia
bility validity
hiera
rchy
responsiveness
CONSENSUS
From perspective of a health care professional:
Assessment contribute to set realistic and therapeutically attainable treatment goals (i.e., improves objectivity).
Advantage of clinimetrics:
From perspective of a (stroke) team:
Clinical assessments increase the transparency in making client-related decisions within a team of professionals working together as a stroke team (i.e., improves communication).
Multidisciplinary Multidisciplinary guidelines for guidelines for
stroke financed by stroke financed by the Dutch Heart the Dutch Heart
FoundationFoundation
Stroke Stroke GuidelinesGuidelines
http://www.kngf.nl/
Praktijkrichtlijn
Samenvattingskaart
Deskundigheidsbevorderingspakket
Verantwoording en toelichting
Samenvattingskaart
Conclusion
Not only differences in heterogeneity in stroke patients are responsible for lack of accuracy in predicting functional outcome, but also the methodological shortcomings in published prognostic research
Kwakkel et al., Age & Ageing, 1996: 25:479-489
Recommended