cost model for a new acoustic diagnostic aid to rule … · time (ms) e heart sound recording:...
TRANSCRIPT
COST MODEL FOR A NEW ACOUSTIC DIAGNOSTIC AID TO RULE OUT CORONARY ARTERY DISEASE
Winther S1, Böttcher M2, Wahler S3, Bolin K4
1Aarhus University Hospital, Aarhus N, Denmark,2Hjertemedicin Hjerteforskningsklinikken, Herning, Denmark,3St. Bernward GmbH, Hamburg, Germany,4University of Gothenburg, Gothenburg, Sweden
06.11.2017 1
• The technical and methodical principles of acoustic diagnosing coronary heart disease (CAD)
• Cardiologic care system in Denmark
• Clinical evidence of the method
• Application of the cost model and results
06.11.2017 2
AGENDA
• The development of acoustics in CAD:
The technical and methodical principles of acoustic diagnosing coronary heart disease
Diagnosis of CAD using an electronic stethoscope
• As in other arteries a stenosis causes flow turbulence, which makes noise
• The sounds are weak, but detectable using advanced signal processing
Master thesis in 2005
(Claus Graff & Samuel Schmidt)
Stenosis in the coronary arteries
.
1967
Heart sounds and CAD murmurs
5
What you listens for today: Murmurs from defective heart valves
Noise from stenotic coronary arteries•Not audible in most cases •Up to 1000 times weaker than other heart sounds
1000 1500 2000 2500 3000 3500 4000-0.1
-0.05
0
0.05
0.1
Time (ms)
Am
plit
ude
Heart sound recording: Normal subject
1000 1500 2000 2500 3000 3500 4000-0.4
-0.2
0
0.2
0.4
Time (ms)
Am
plit
ude
Heart sound recording: Subject wiht systolic murmur
1000 1500 2000 2500 3000 3500 4000
siDia
siSys
S1
S2
Sta
te
Time (ms)
State sequence
1000 1500 2000 2500 3000 3500 4000
siDia
siSys
S1
S2
Sta
te
Time (ms)
State sequence
Manual segmented states (qt) Segmentation results (q*
t)
S1 S2 S1 S2 S1 S2 S2S1S2S2S1 S1
Frequency spectrum (M. Akey et. al.)
The CADScore® AlgorithmCADScore Algorithm
Diastole
Identify
acoustic
features
FilterCombine
features
- 99: High risk
- 0: Low risk
CADscore®
Main clinical studies
CE Mark August 2015Registrationstudy
AC003SH
(2013)
N=228
Symptomatic, no CAD diagnosed
One center (DK)
VALIDATE
(2016)
Dan-NICAD
(2014-2016)
Screening
Studies
N=231Symptomatic, sceduled
for CAG
Two centers (GER)
PlannedStress-ECG
N=1,675Symptomatic, no CAD
diagnosed
Two centers (DK)
PlannedDan-NICAD II
N=1,000Symptomatic, undiagnosed
Multiple centers (GER/UK)
N=2,000New Algorithm
Multiple centers (DK/SWE)
Suspected CAD flow through the system
Percutaneous Coronary
Intervention
Medical Management
StenosisAppropriate for
Intervention
No Interventional Target
Explore Noncardiac
Causes
Medical Management
Medical Management
Low pretest likelihood of CAD
Identification of Chest Pain
Stress Test (Treadmill)
Stress-Echo/MPI/Nuclear
Imaging
cCTA
Invasive Coronary Angiography
Low Risk
Indeterminate CAD
Likely severe CAD
Severe fun
ction
al steno
sis
Intermediate or High Risk
Low RiskHigh Risk
Intermediate or High Risk
Physician Key
Primary Care Practitioner
General Card.
Radiologist
Interventional Card.
The CADScor® system for diagnosing CAD risk using acoustic properties
• The CADScor ® System consists of a disposable patch and reusable sensor/touch-screen interface
• The product is based on the concept that atherosclerosis in coronary vessels will produce unique murmurs detectable by advanced acoustic equipment during diastole
• Minimal training requirements• Measurement takes around 5-8 min. • Noninvasive • No patient exposure to radiation or
contrast dye• High negative predictive value • Relatively low capital and consumable
costs
Characteristics of tests commonly used to diagnose thepresence of coronary artery disease
Sensitivity Specificity NPV
Stress ECG 45-50 85-90 93%
Stress Echo 80-85 80-88 97%
Stress SPECT 73-92 63-87 95%
Dobutamin Stress Echo 79-83 82-86 97%
Dobutamin Stress MRI 79-88 81-91 97%
Vasodilatator Stress Echo 72-79 92-95 96%
Vasodilatator Stress SPECT 90-91 75-84 98%
Coronary CT 95-99 64-83 99%
CADScore* 81-90 45-53 97%
ESC SCAD guidelines 2013
*not in guideline
99 % Socialized medicine so few private clinics
General practitioners refer to outpatient clinics for workup
Workup means: Cardiologist – Echo - ECG bloodtest and history.
Activity:
• Invasive angiography 7,000 per year- 1.4 per 1,000 (Germany: 8.6)
• Cardiac CT 5,000 per year.
• Perfusion test ( Rb-PET) 3,000 per year
Excercise test not used
Stress Echo not used
Health care set up in Denmark
06.11.2017 12
Flow of study AC003SH
Patients referred toCCTA or ICAFor CAD diagnostics
Excl.: Pregnancy, diastolic murmur,heart surgery, arrhythmias, etc.
13
54%18%
28%
Result of diagnostics
Non CAD
Non obstructive CAD
Obstructive CAD
0
10
20
30
40
50
60
70
≤20 >20-30< ≥30
0
5
10
15
20
25
30
35
40
≤20 >20-30< ≥30
No CAD CAD
Results of AC008SH
Pre-test probability of CADDiamond-Forrester updated
2014 KBVGuidelines
15%> Intermediate risk <85% >85%<15%Nofurthertesting
CAGCTA15%< X-ECG<30%
>50% SPECT, C-
MRT, D-MRT, X-Echo
06.11.2017 15
255 personsenrolled
CAG
MPI
N-Contr.CT
High risk 41
Interm. risk
Low. risk
Very low risk
68
90
29
5 11
No-CAD
35
MPI
9
3
Cor. CTRelapse
3
MPI
1
CAG
2
71
Patient flow by DF assessment
Cor. CT
CAD
184
• Danish hospital tariff2016 - Cardiology
– Non contrast CTDKK 906
– Contrast CTDKK 2,621
– MPI DKK 3,015
– CAG DKK 5,997
– (CADSCORE DKK 350)
• Risk changes after combination of scores
06.11.2017 16
Price tag and change of probabilities
06.11.2017 17
255 personsenrolled
CAG
MPI
N-Contr.CT
High risk 61
Interm. risk
Low. risk
Very low risk
48
49
70
2 6
No-CAD
25
MPI
5
2
Cor. CTRelapse
7
MPI
2
CAG
6
71
184
Patient flow by new assessment
Cor. CT
CAD
06.11.2017 18
0
20
40
60
80
100
120
Non-contrast CT Contrast CT MPI CAG
Frequencies in the model
DF Combination
Changes in examination
-37.9% -43.1% -29.0% +4.9%
• Overall costs for diagnostics was DKK 6,096 applying acoustic testing and DKK 6,379 without; saving DKK 274 per patient.
06.11.2017 19
Results
• 28% CAD is not the typically expected population
• Free combination of DF and Score is not anymore in the focus of development
• Newly published study has 8-times more patients
• But the main consideration remains: How to improvethe pre-test probabilities to avoid futile diagnostic
06.11.2017 20
Drawbacks of the model