Is It Time for the MMSE to Retire? Evidence of Better & Shorter Tests for Dementia Alex Mitchell Consultant & Hon SnR Lecturer in Liaison Psychiatry, Leicester Cardiff 20 th Anniversary Conference October 2007
Jul 06, 2015
Is It Time for the MMSE to Retire?
Evidence of Better & Shorter Tests for Dementia
Alex MitchellConsultant & Hon SnR Lecturer in Liaison Psychiatry, Leicester
Cardiff 20th Anniversary Conference October 2007
When is it time to retire?
• When your job is done?
• When you are no longer making a difference?
• When you have been superseded?
• When you have been around a long time?
• When you are fed up with the latest changes?
What is the most cited study in psychiatry?
• Folstein M, Folstein S, McHugh P. Mini-Mental State. A practical method for grading the cognitive state of patients forthe clinician. J Psych Res 1975;12:189–198.
2x2 =>
Where is the cut-off on the MMSE?
Severity on the MMSE
• mild Alzheimer’s disease: – MMSE 21 to 26
• moderate Alzheimer’s disease– MMSE 10 to 20
• moderately severe Alzheimer’s disease– MMSE 10 to 14
• severe Alzheimer’s disease– MMSE less than 10.
CG42 November 2006
1.4.1.3 Clinical cognitive assessment in those with suspected dementia should include examination of attention and concentration, orientation, short and long-term memory, praxis, language and executive function. As part of this assessment, formal cognitive testing should be undertaken using a standardised instrument. The Mini Mental State Examination (MMSE) has been frequently used for this purpose, but a number of alternatives are now available, such as the 6-item Cognitive Impairment Test (6-CIT), the General Practitioner Assessment of Cognition (GPCOG) and the 7-Minute Screen. Those interpreting the scores of such tests should take full account of other factors known to affect performance, including educational level, skills, prior level of functioning and attainment, language, and any sensory impairments, psychiatric illness or physical/neurological problems.
1.4.1.4 Formal neuropsychological testing should form part of the
assessment in cases of mild or questionable dementia.
Dementia : Supporting people with dementia and their carers in health and social care
Technology appraisal 111
1.1 The three acetylcholinesterase inhibitors donepezil, galantamine and rivastigmine are recommended as options in the management of patients with Alzheimer’s disease of moderate severity only (that is, subject to section 1.2 below, those with a Mini Mental State Examination [MMSE] score of between 10 and 20 points), and under the following conditions.
Expert Opinion
What year is it?
Or
Have you been having memory problems?
HealthyDementiaPatient Doesn’t Complain
AAMIMCIPatientComplains
Relative Doesn’t Complain
Relative Complains
Contents=>
Concepts of Screening
• Screening (possible case)– MMSE
• Case-Finding (probable case)– NINCDS-ADRDA criteria, accuracy in
autopsy-verified cases is approximately 90%
• Severity Rating
• Gold Standard (definite case)– Pathology => disease High accuracy
High convenience
Levels of Screening Competence
• Untrained, unassisted
• Untrained, assisted
• Trained, unassisted
• Trained, assisted
• Trained, assisted and monitoredHigh accuracy
High convenience
Who is Looking for Dementia?
willingness =>
GP Screening Preferences
• 74% of people consult a GP first after noticing symptoms of cognitive decline 3
• 82% of GPs say screening for dementia is worthwhile
– but 24% routinely screen (GPs)
– 39% psychiatrists use the MMSE1
• 93% would use a brief effective tool2
1 Gilbody, House Sheldon (2002) Br J Psychiatry2 Bush et al Can Fam Physician. 19973 Wilkinson et al (2004);
Gps neurol =>
A Look An Unassisted Detection
Symptoms spotted spontaneously
Gps
Neurologists
Patients
Relatives
How Accurate are GPs? (untrained, unassisted)
unass =>
Gold standard is probable dementia
Recognition of “Dementia” by GPsUsing documentation of dementia in the medical notes
PrevalenceSpecificitySensitivity
NPVTrue -VeFalse -VeTest -ve
PPVFalse +veTrue +veTest +ve
DementiaABSENT
DementiaPRESENT
Simple Measures of Accuracy
Recognition of “Dementia” by GPs
12601148112
1202114458No dementia in notes
58454Dementia in notes
DementiaABSENT
Dementia (DSMIV)
Sensitivity48%
PPV 93%
Specificity99.6%
NPV 95%
Prevalence 8%
Using documentation of dementia in the medical notes
By severity =>
How Accurate are Neurologists (trained, unassisted)
willingness =>
Gold standard is Alzheimer’s disease
Accuracy of Trained Clinicians
1000190810
160 Total _ve10060Test -ve
840 Total +ve90750Test +ve
ALZHEIMER’SABSENT
ALZHEIMER’SPRESENT
Sensitivity93%
PPV 90%
Specificity55%
NPV 64%
Prevalence81%True n= 2188,
GS = pathologyMayeux et al (1998)
Recognition Rate of Dementia by Severity
97%
73% 71%
46%
66%
33%
0
10
20
30
40
50
60
70
80
90
100
SevereDementia
(CI)
SevereDementia
(Dementia)
ModerateDementia
(CI)
ModerateDementia
(Dementia)
Milddementia
(CI)
Milddementia
(dementia)
Accuracy of Patient’s Impression?Do we need the Dr at all?
Single item =>
SMC => 1 Item Test (Ultra-screening)
• 1 Item Q for Depression– During the last month, have you often been bothered by feeling down,
depressed or hopeless?
• 1 Item Q for Dementia– Have you had memory problems in the last year?
1Item: “Have You Had Memory Loss in the last year?”
58.5% (se)
39
55
MCI
Prevalence = 10%
79% (Sp)61% (se)
95%115124No
10%31233Yes
DementiaAbsent
DementiaPresent
St. John & Montgomery, J Geriatr Psychiatr Neurol 2003 (n=1751)
clinician =>
Pooled Analysis: 3 + 3 Studies (n=5074)
32% (se)
271
133
Mild Cases
Prevalence = 7.7%
90% (Sp)52% (se)
96%4226187No
31%456205Yes
DementiaAbsent
DementiaPresent
clinician =>
“Have you had memory problems”
Simple Memory Complaints
63.8
25.5
35.1
70.2
16
68.2
39.4
48.5
80.3
30.3
73.3
41.3
58
88
28
73.2
45.1
67.6
87.3
43.7
0
10
20
30
40
50
60
70
80
90
100
Forgetting where things areplaced
Unable to recall the names ofgood friends*
Unable to follow and recallconversation**
Subjective memory problems* Consider own memory to beworse than others of a similar
age**
ControlsMCIMCI=>DementiaAD (CDR1)
Lam et al. Int J Geriatr Psychiatry 2005; 20: 876–882. (n=306)
What Makes Unassisted Detection Difficult?Or indeed easy?
Theory of Diagnostic Tests
PopulationNumber ofIndividuals
Cognitive Score
Theory of Diagnostic Tests
Cognitive Impairment
Dementia
Number ofIndividuals
Optimum Cut-off value
False +veFalse +veFalse -veFalse -ve
True -veTrue -ve
True +veTrue +ve
Point of Partial Rarity?
Cognitive Score
Animals named in 1 min (mms>19) - CERAD data set
0
2
4
6
8
10
12
0 10 20 30 40
number of animals named
perc
ent o
f tot
al
Normal Controls, CS = 1, n = 386
Alzheimer patients, CS = 0, n = 380
GP Testing by Actual MMSE Score (n=162)Ganguli M et al. Detection and Management of Cognitive Impairment in Primary Care: The Steel Valley Seniors Survey. JAGS 52:1668–1675, 2004.
methdos =>
Distribution of MMSE Scores
0
5
10
15
20
25
30Thir
tyTwen
try Nine
Twentry
E ight
Twentry
Seve
Twentry
S ixTwen
try Five
Twentry
FourTwen
try th
reeTwen
try Two
Twentry
One
Nineteen
Twentry
Nine
108 Controls54 with dementia
Funabiki et al (2002) Geriatrics Gerontol Int.
Types of Assisted Detection (tools)Can tools aid detection?
Types of Assisted Recognition
• Clinician Prompts– GDS, CDR
• Simple (Bedside) Single Item Cognitive Tests– Verbal fluency, Name & Address, Orientation
• Short Batteries– MMSE
• Long Batteries– CAMCOG
• Criterion Standard– DSMIV, ICD10
domains =>
Types of Recognition
• Clinician Prompts– GDS, CDR
• Simple (Bedside) Single Item Cognitive Tests– Verbal fluency, Name & Address, Orientation
• Short Batteries– MMSE
• Long Batteries– CAMCOG
• Criterion Standard
domains =>
New data on battery instruments
methods =>
Methods
• Stage 1 – Search (n=500)– 100 MMSE, 100 not short, 100 no validation, 100 n<200
• Stage 2 - Narrative Review– Authors own claims
• Stage 3 – Pooled Meta-analysis– Go to Excel =>?
• Stage 4 - Recommendations– Grades of Diagnostic Accuracy– 90% Accuracy = “Excellent”– 80% accuracy = “Good”– 75% accuracy = “Satisfactory”– <75% accuracy = “Unsatisfactory/Poor”
More methods =>
Methods
• Literature search– Electronic, manual, reverse (citation)
• Keywords– (cognitive, dementia) +(brief, short, rapid) +(screen,test)
• Inclusion criteria– Brief tests- max. 10 minutes– Sensitivity, specificity or raw data– Criterion Reference (DSM IIIR /IV, NINCDS-ADRDA)
• Exclusion– Study <200 participants
poster =>
Srinivasa Malladi & Alex J Mitchell Liaison Psychiatry, Brandon Unit, Leicester General Hospital, Gwendolen Road, Leicester LE5 4PW [email protected]
Aim: To identify brief and effective cognitive screening tools which may be practical and valid in the primary care settings.
Methods: An extensive literature search was performed electronically using keywords, abstract and citation searches. A full text search and manual search was also performed. A total of 56 articles were initially identified as “candidates.” Inclusion criteria were that the tool should be compared against a gold standard such as NINCDS-ADRDA, DSM IIIR/IV diagnostic criteria. A comparison with MMSE was desirable.
Brief Cognitive Assessment Tools To Detect Dementia - A Systematic Review
Results: To date 31 papers have been analysed in detail, 15 papers were excluded and another 10 are under review. Sensitivity and specificity findings have been collated where data was presented. Preliminary results suggest that very short and ultra-short tests (Combined JB +VB; Rapid dementia screening test, the Memory impairment screen or the Mini-Cog) are more accurate and quicker than MMSE alone. A variety of tools including telephone interviews and informant questionnaires appear to be promising alternatives to MMSE which is not thought to be ideal for cognitive assessment in general practice. Several of the tools are also thought to be user friendly and potentially free of biases that can affect MMSE. Performance in different types of dementia has not been adequately tested.
Conclusions: Our preliminary data suggest that there appear to be a number of brief tools that are as effective as MMSE and several that may be more effective. These tools can also be combined where required.
Authors/year Test tool Criterion standard Sensitivity Specificity
Erkinjuntti et al 1988 BDS DSM IIIR 105 90% 84%
Albert M et al.1991 SPMSQ NINCDS-ADRDA. 3811 34.4% 94.3%
Albert M et al. 1991 EBMT NINCDS-ADRDA. 3811 47.8% 95.1%
Hooijer et al. 1992 SPMSQ GMS/AGECAT 24 100% 96.8%
Hooijer et al. 1992 MSQ GMS/AGECAT 18 92.3% 98.3%
Hooijer et al. 1992 AMTS GMS/AGECAT 28 92.3% 95.4%
Glosser et al..1993 Extended CCCE Clinical diagnosis 115 94% 99%
Stuss et al. 1996 DRS NINCDS-ADRDA 283 87% 84%
Stuss et al 1996 OMC NINCDS-ADRDA 283 95% 77%
Wind et al.1996 Set of items GMS/AGECAT 533 64.9% 96.4%
Tanya et al. 1998 T&C Test Clinical diagnosis 100 62.5% 96.4%
Solomon et al. 1998 7MS NINCDS-ADRDA 90 93.3% 96.6%
Bushke et al. 1999 MIS NINCDS-ADRDA 483 90% 81%
Belle,S.H.,(2000). SASSI CERAD CDR 1178 94% 91%
De Koning 2000 R-CAMCOG NINDS-AIREN 284 91% 90%
Borson S et al 2000 CASI NINCDS-ADRDA 249. 92% 96%
Derrer et al. 2001 WL+ LM+ VM NINCDS-ADRDA 74 100% 94.5%
Derrer et al. 2001 Word list NINCDS-ADRDA 74 95% 89%
Kirby et al.. 2001 CDT GMS-AGECAT 648 76% 80.5%
Kirby et al. 2001 MMSE + CDT GMS-AGECAT 648 95% 95%
Scanlan et al 2001 Mini Cog NINCDA-ADRDA 249 99% 93%
Brodaty et al. 2002 AMT DSM IV CAMDEX 375 42% 93%
Brodaty et al. 2002 GPCOG DSM IV CAMDEX 380 82% 83%
Salib E et al. 2002 MAT NINCDS-ADRDA 113 95% 81%
Kuslansky et al. 2002 MIS NINCDS-ADRDA 240. 86% 97%
DeYebenes et al. 2003 PCL DSM IV/IPA-WHO 375 93.9% 94.7%
Lipton et al. 2003 CF-T + MIS-T NINCDS-ADRDA 355 89% 93%
Lipton et al.. 2003 TICS NINCDS-ADRDA 355 83% 86%
Lin et al2003 3 Item CDT NINCDS-ADRDA 403 67% 75%
Kalbe et al. 2003 RDST NINCDS-ADRDA 490 72% 89%
Borson S et al. 2003 Mini-Cog NINCDR CERAD 1119 76% 89%
Robert et al. 2003 SCEB DSM IV 123 93.8% 85%
Meulen et al. 2004 7MS NINCDS-ADRDA 331 92.6% 93.5%
Brodaty et al. 2004 GPCOG DSM IV 283 85% 86%
Storey J et al. 2004 RUDAS DSM IV 90 89% 98%
Kalbe et al. 2004. DemTect NINCDS-ADRDA 363 100% 92%
Mahoney et al. 2005 TE4D-Cog DSM IV 203 100% 84%
Kilada S et al. 2005 C’bined JB + VF NINCDS-ADRDA 456 87% 90%
Galvin et al 2005. AD8 CDR 85% 86%
Abbreviations:MMSE- Mini mental state examination RUDAS-Rowland universal dementia assessment scale MIS-Memory impairment screen SCEB- Short cognitive evaluation battery MAT- Mental alternation test CCCE-Cross cultural cognitive examination PCL-Prueba cognitiva de leganes 7 MS- 7 minute screen TICS- Telephone instrument for cognitive status CF-T – Category fluency test MIS-T- Memory impairment screen by telephone RDST- Rapid dementia screening test CDT- Clock drawing testCombined VB+ JB- Combined verbal fluency and John brown address SASSI- Short and sweet screening instrument DRS-Dementia rating scale BDS- Blessed dementia scale MSQ- Mental status questionnaire SPMSQ- Short portable mental status questionnaire OMC- Orientation memory concentration test T&C test- Time and change test TE4D-Cog- Test for early detection of dementia from depression CASI- Cognitive abilities screening test AMT- Abbreviated mental test GPCOG-General practitioners assessment of cognition EBMT- East boston memory test WL+LM+VM- Word list + Logical memory + verbal memory
Sample Size
Individual Results : MMSE (Mean) Sensitivity= 83% Specificity= 85.9%As accurate as MMSE=21As accurate + quicker than MMSE = 7: Mini-Cog Combined (VF+JB) MIS AD8 GPCOG Word list acquisition CCCE
As accurate + as quick as MMSE = 3: RUDAS DemTect 7MS
Head to head comparison with MMSE:More accurate + quicker = 6: MiniCog RDST MIS GPCOG Set of items(Time+ PM+ Address) CCCEMore accurate +as quick = 3 :TE4D-Cog DemTect 7MSAs accurate+ quicker =3:Word list acquisition MAT OMC As accurate but longer=3:SASSI (MMSE+CDT) DRS
Less accurate but quicker=1: CDT
Tools not compared =6: SCEB TICS (CF-T + MIS-T) AD8 RUDAS PCL
Royal College of Psychiatrists Faculty of Old Age Psychiatry Annual Residential Conference Radisson SAS Hotel, Glasgow, 1st – 3rd March 2006
Excel Spreadsheet
Results (63 analyses, 19 MMSE) n=26,600
1Telephonic interview for Cognitive Status
3Short portable mental status questionnaire (SPMSQ)
1Short and sweet screening instrument (SASSI) Includes MMSE Verbal fluency Temporal orientation test
1Saint Louis University Mental Status Examination
1Revised Blessed dementia rating scale (RDS)
1RDST (Rapid dementia screening test)
1R-CAMCOG
1PCL- Prueba cognitive de leganes
1MSQ
3MMSE (Modified Analysis)
19MMSE
1Mini-Cog
1Memory impairment screen (MIS-T) –Telephonic interview
1GPCOG
1Eurotest
1D-MMSE (age Independent)
1DemTect (RDST plus 3 items)
1Combined CF-T + MIS-T
1CASI (Cognitive Abilities Screening Instrument)
4Blessed dementia rating scale (BDRS)
1AMTS+MSQ
4AMTS
1AD8
17MS (7 minute screen)
36 Item Screen (3 word recall and 3 temporal orientation)
# of StudiesType of Battery
How does the MMSE Perform?
2x2 =>
6 Domains11 Instructions20-24 Items30 PointsTime = 8.5 minutes (well)Time = 13.5 minutes (unwell)
Pooled Accuracy of MMSE (n=10,400 x 20 >22)
Prevalence = 10%86% (Sp)76% (se)
90% (NPV)6534669MMSE No
68% (PPV)10052192MMSE Yes
DementiaAbsent
DementiaPresent
ceiling =>
AD all
0.00
0.10
0.20
0.30
0.40
0.50
0.60
-4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10
DISABILITY SCALE
TEST
INFO
RM
ATI
ON
MMSE Item-Response Analysis
Ceiling Effect Floor Effect
Acknowledgement: Ashford, Cost-Effectiveness Screening for Dementia Limitations =>
AD all (easiest to hardest at p=.5)
00.10.20.30.40.50.60.70.80.9
1
-4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10
DISABILITY ("time-index" year units)
PRO
BAB
ILIT
Y C
ORR
ECT
PENCILAPPL-REPWATCLOCATIONPENY-REPTABL-REPCLOS-ISRIT-HANDCITYFOLD-HLFSENTENCECOUNTYNO-IFSFLOORSEASONYEARPUT-LAPMONTHADDRESSDRAW-PNTDAYSPEL_ALLDATEAPPL-MEMPENY-MEMTABL-MEM
Mini-Mental State Exam itemsMMSEitems
Based on Ashford et al., 1989; 1995; applied to CERAD data setAcknowledgement: Ashford, Cost-Effectiveness Screening for Dementia
Can Single Items Outperform the MMSE?
Pooled Analysis of Single Items (primary care)
10,43978762563
6979891Test -ve
8971672Test +ve
DementiaABSENT
DementiaPRESENT
Sensitivity65%
PPV 65%
Specificity89%
NPV 89%
Prevalence24%
Pooled Analysis of Single Items (secondary care)
240512341171
969317Test -ve
265854Test +ve
DementiaABSENT
DementiaPRESENT
Sensitivity72%
PPV 76%
Specificity78%
NPV 75%
Prevalence48%
Relative risk meta-analysis plot (fixed effects)
0.2 0.5 1 2
Heun et al (1998) [Word List Delay] 0.97 (0.76, 1.22)
Kalbe et al. (2004) [Word List Delay] 1.01 (0.92, 1.11)
Heun et al (1998) [Word List Imm] 0.97 (0.76, 1.22)
Kalbe et al. (2005) [Verbal Fluency-S] 1.18 (1.06, 1.33)
Kalbe et al. (2004) [Verbal Fluency-S] 1.00 (0.91, 1.10)
Kilada S et al. (2005) [Verbal Fluency-A] 0.90 (0.82, 0.97)
Heun et al (1998) [Verbal Fluency-A] 0.97 (0.76, 1.22)
Heun et al (1998) [TMT] 0.97 (0.76, 1.22)
Kalbe et al. (2005) [Transcoding] 0.88 (0.76, 1.01)
Kilada S et al. (2005) [Orientation] 0.87 (0.80, 0.95)
Kilada S et al. (2005) [John Brown Recall] 0.95 (0.87, 1.02)
Heun et al (1998) [GDS] 0.65 (0.45, 0.88)
Kirby et al. (2001) [CDT] 0.86 (0.68, 1.06)
Borson S et al. (2003) [CDT] 0.82 (0.64, 1.05)
Kilada S et al. (2005) [3 Word Recall] 0.71 (0.64, 0.78)
Borson S et al. (2003) [3 Word Recall] 0.76 (0.58, 0.99)
Borson S et al (2000) [3 Word Recall] 1.00 (0.92, 1.09)
combined [fixed] 0.90 (0.87, 0.93)
relative risk (95% confidence interval)
More on Single Items
• Most successful in non-specialist settings (Youden 53.8 vs 51.4, respectively), they would correctly exclude 18 out of 20 people without dementia but correctly identify 13 out of 20 cases in non-specialist settings
• 2 Specific Tests perform well– Memory
• Memory Impairment Screen (MIS), the East Boston Memory Test (EBMT) and 3 word recall
– Verbal fluency• Animals, Supermarket
• On meta-analysis verbal fluency was no less accurate in terms of sensitivity than the MMSE itself and individual memory tests were no less accurate than the MMSE itself in terms of specificity.
Is Recall Alone better MMSE?
AUCs
RI 48 Test: 98.5CERAD immediate recall: 96.5CERAD delayed recall: 95.2MMSE: 85.3
9.7
6.7
10.5
12.3
13.2
12.9
58.8
76
64.2
49.9
62.7
31.5
30.4
13.5
23.5
36.9
24.4
62.9
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
MMSE
Battery No MMSE
6-CIT
Blessed
SPMSQ
AMTS
False NegativesAccurate DiagnosesFalse Positives
NPV
NPV
PPV
Where? =>
Multiple Validation
20
6
22
10
10
Items
10
2 min
5 min
2 min
2 min
Minutes
3192863MMSE (by comparison)
1365756-CIT
2444.461Blessed dementia rating scale (BDRS).
4425.157Abbreviated Mental Test Score (AMTS)
5318.754Short portable mental statusquestionnaire (SPMSQ)
RankingStudiesDORYoudenTest
Where Can I Find these scales?
• Many Scales are here
– www.neurotransmitter.net/alzheimerscales.html
– www.medal.org