poststroke population Lauren E Cipriano Jeff healey Omar Akhtar Karen Lee Luciano A Sposato April 12 2016 Disclosures Funding CADTH Conflicts None to report Atrial fibrillation in patients with a history of stroke or TIA ID: 921174
Download Presentation The PPT/PDF document "Cost effectiveness of AF monitoring stra..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Cost effectiveness of AF monitoring strategies in a post-stroke population
Lauren E.
Cipriano
Jeff
healey
Omar Akhtar
Karen Lee
Luciano A.
Sposato
April 12, 2016
Slide2DisclosuresFunding: CADTHConflicts: None to report
Slide3Atrial fibrillation in patients with a history of stroke or TIA
~ 20% of stroke patients have a known history of AF
~ 20 - 25% of stroke patients without a prior AF diagnosis, have AF
History of stroke & AF
10-15% risk of recurrent
stroke in year 1
2
Slide4Oral anticoagulation treatmentReduce the risk of recurrent stroke (HR = 0.3-0.65)
Increase the risk of bleeding (HR = 1.5-3.0)
NOAC lower the risk of ICH but increase the risk of non-brain bleeding compared to warfarin
Warfarin is inexpensive, but has many contraindications and requires regular visits for patients
NOACs are expensive, similarly effective to warfarin, and require less follow-up visits to ensure safety /efficacy
3
Slide5Post discharge AF monitoring
4
Patient with
stroke or TIA
ECG
Hospital
discharge
Age 70
CHADS-VASC ~4
24+
hr
m
onitoring
(sometimes)
Focus of our analysis
24
hr
– 7 days
monitoring?
30 days
monitoring?
2-3 years
m
onitoring?
(implantable device)
Slide6Objective To evaluate the cost effectiveness of outpatient cardiac monitoring devices for the evaluation of atrial fibrillation in discharged patients with a recent history of stroke or
TIA
Variation in clinical evaluations of monitoring strategies made many desired comparisons difficult to evaluate.
5
Slide7Research QuestionThree separate cost effectiveness analyses
6
Study
Higgins et al.
(
2013)
Gladstone et al. (
2014)
Sanna
et al. (2014)
Cohort:
Ischemic
stroke or TIA within 7 days
Cryptogenic
stroke or TIA within 6 months
Cryptogenic
stroke or TIA within 90
days
Prior AF monitoring:
12-lead ECG
12-lead ECG
and
≥
24-hour
Holter
12-lead ECG and
≥
24-hour Holter
Intervention:
7 day cardiac event monitoring (ELR)
30-day event triggered recorder (ELR)
3-yr
insertable
cardiac
monitor
(ILR
)
Comparator:
Standard practice
(
60%
received
24-hour
Holter
)
24-hour
Holter
Standard practice
(
30%
≥ 1 ECG
and
8
%
24-hour
Holter
)
Slide8Markov Model
7
Undiagnosed AF / No AF
Diagnosed AF
May initiate OAC
May discontinue OAC
+ GI bleed
(any cycle)
Slide9Model assumptionsMonthly cycles & Lifetime horizon
Perspective: Public payer and
Soceital
Input parameters
Medical literature
Expert opinion
Validated to Oxford Vascular Study
Life-expectancy, QALYs, and 5-year recurrent strokes
8
Slide10OAC assumptions
61% of patients initiate OAC after diagnosis
16% of patients on OACs quit each year
OACs reduce the risk of recurrent stroke
Warfarin: 36% reduction compared to aspirin
Apixiban: 45% reduction compared to aspirin
Annual Cost
Disutility
Patient time
*
Warfarin
$396
- 0.013
2 hours every 3 weeks
Dabigatran
$1288
- 0.006
-
Rivaroxaban
$1157
- 0.006
-
Apixaban
$1288
- 0.006
-
* Assume a caregiver attends all visits; patient and caregiver time @ $25/
hr
Slide11Model-based outcomes10
Outcomes
AF
without diagnosis
AF with diagnosis
Warfarin
Apixaban
Within 2
years (per 10,000)
Recurrent
IS,
severe or fatal
1,087
709
636
Recurrent stroke, any
1,724
1,223
1,093
ICH
47
99
42
GI bleed
171
276
264
AF diagnosis:
Non-fatal strokes
Fatal strokes
Bleeds
Slide12Model-based outcomes11
Outcomes
AF
without diagnosis
AF with diagnosis
Warfarin
Apixaban
LE
4.31
4.53
4.60
Discounted LY
3.59
3.75
3.80
Discounted QALYs
2.30
2.39
2.45
Costs
Baseline
229,025
238,458
241,473
OAC
0
843
2,870
Acute events
19,876
17,223
15,952
Total
248,901
256,525
260,294
Life expectancy
Quality-adjusted life expectancy
AF diagnosis:
Costs
Slide13Minimum diagnostic yield to be cost effective
12
61% of AF patients initiate OAC
100% of AF patients initiate OAC
Slide14Minimum diagnostic yield
to be cost effective
13
61% of AF patients initiate OAC
100% of AF patients initiate OAC
7d
7d
30d
30d
Slide15Evaluation 1: 7-day monitoring vs standard practice
Patients: Stroke and TIA patients within 7 days of discharge
Previous evaluation: ECG
Comparison: 7-day ELR vs Standard practice
(60% 24-hr Holter) diagnostic yield (sustained AF) = 16% (4.7% - 27.3%)
Results: ICER $50,000 - $80,000 per QALY gained
14
Slide16Base case analysis
15
Health economic outcomes
OAC treatment
Warfarin
Apixaban
Lifetime Costs
Cost of
FN
diagnosis
$151,148
$151,435
Cost of
TP diagnosis
$153,490
$155,140
Incr.
cost of correct diagnosis
$2,342
$3,705
Incr.
cost of monitoring
$139
$139
Lifetime benefits
QALYs of
FN diagnosis
2.30
2.31
QALYs of
TP
diagnosis
2.36
2.40
Incr.
QALYs of correct diagnosis
0.055
0.088
Incr.
cost effectiveness ratio (ICERs) ($/QALY gained)
$58,800
$
52,200
Slide17Sensitivity analysis7-day monitoring ICER <$100,000 / QALY gained with
incremental diagnostic yield
as low as 5%
Patient characteristics that increase cost effectiveness
Higher
risk of recurrent stroke (>10% py)Lower baseline risk of bleeding
Healthier / Fewer comorbidities (lower baseline mortality risk; lower baseline health care costs; higher baseline utility)
Higher likelihood of OAC uptake if diagnosed
Societal perspective (apixaban
)
ICER $75,000 – 85,000 / QALY gained
ICER
<$70,000 if
OAC
if patient is above average baseline health
16
Slide18Cost effectiveness improves with connection to OAC treatment
17
Slide19Cost of monitoring is critical factor influencing cost effectiveness
18
Slide20Summary: 7-day monitoring vs standard practice
ICER likely between $40,000-$80,000 per QALY gained
Patient population
Unselected stroke and TIA patients within 7 days
Prior monitoring: only ECG
Targeted patient populations ICER <$50,000 per QALY gained
> 20% incremental diagnostic yield
high rate of connection to OAC treatment
low OAC discontinuation rateselection of relatively healthy patients
Unable to evaluate 7-day monitoring vs. universal 24-hour monitoring
Optimal duration of monitoring is unknown
19
Slide21Evaluation 2: 30-day ELR vs 24-hour Holter
Patients: Cryptogenic stroke patients within 6 months
Previous evaluation: ECG and at least 24-hrs of
Holter
Comparison: 30-day ELR vs 24-hr
HolterIncremental diagnostic yield = 12.9% (8.0% - 17.6%)Results: ICER $90,000 - $120,000
per QALY gained
20
Slide22Base case analysis
21
Health economic outcomes
OAC treatment
Warfarin
Apixaban
Lifetime Costs
Cost of
FN
diagnosis
$151,148
$151,435
Cost of
TP diagnosis
$153,490
$155,140
Incr.
cost of correct diagnosis
$2,342
$3,705
Incr.
cost of monitoring
$508
$508
Lifetime benefits
QALYs of
FN diagnosis
2.30
2.31
QALYs of
TP
diagnosis
2.36
2.40
Incr.
QALYs of correct diagnosis
0.055
0.088
Incr.
cost effectiveness ratio (ICERs) ($/QALY gained)
$ 114,979
$ 87,130
Slide23Sensitivity analysis30-day monitoring followed by treatment with warfarin and apixaban
has an ICER < 100,000 per QALY gained
Incremental diagnostic yield > 20%
Incremental cost of monitoring < $300
OAC uptake > 85%
Patients with above average baseline healthPatient and system factors alone insufficient to make 30-day monitoring cost effective compared to 24-hour
Holter
without greater diagnostic yield and/or lower
incremental costSocietal perspective: base case >$110,000 per QALY gained
22
Slide24Cost effectiveness improves with connection to OAC treatment
23
Slide25Summary30-day ELR vs 24-hour Holter
Unlikely to be cost
effective (ICER > $
100,000/QALY gained)
Patient population
Cryptogenic stroke and TIA patients within 6 monthsPrior monitoring: ECG and 24-hour Holter
Unable to compare 24-hour
Holter
to no monitoringUnclear if any monitoring after ECG and 24-hour Holter is cost effective
Unable to evaluate
whether evaluating patients sooner (within 30 or 90 days of stroke) would improve cost effectiveness
24
Slide26Evaluation 3: 3-years of ILR vs standard practice
Patients: Cryptogenic stroke patients within 90 days
Previous evaluation: ECG and at least 24-hrs of
Holter
Comparison: 3-years of ILR vs. standard practice
(38% ECG or 24-hour Holter within 6 months)
30% of patients in ILR group diagnosed with AF
vs. 3% in the standard practice group
Incremental cost: EV ~$3400 ($2800 implantation + monitoring)Results
: ICER
> $
2
50,000
per QALY gained
25
Slide27Connection to OAC treatment is important but not sufficient
No single or reasonable combination of factors achieves ICER < $100,000 per QALY gained
26
Slide28Summary3-years of ILR vs standard practice3-year ILR is not cost effective compared to standard care
Patient population
Cryptogenic stroke and TIA patients within
90 days
Prior monitoring: ECG and 24-hour
Holter
27
Slide29Post discharge AF monitoringEvaluation 1: 7-day ELR vs. standard care
28
Patient with
stroke or TIA
ECG in Hospital
Hospital
discharge
Age 70
CHADS-VASC ~4
7-day ELR
Standard care
Conclusion:
Likely cost effective
$40,000-80,000/QALY gained
∆ diagnostic yield > 10-20%
Patient factors
↑
risk of stroke;
↓
risk of bleeding
Relatively healthy
↑ connection to OAC
∆ diagnostic yield = 16%
∆ cost = $140
Slide30Post discharge AF monitoringEvaluation 2: 30-day ELR vs. 24-hr Holter
29
Patient with
cryptogenic
stroke or TIA
ECG in Hospital
Hospital
discharge
Age 70
CHADS-VASC ~4
30-day ELR
24-hour
Holter
Conclusion:
Not
cost
effective
$
9
0,000-120,000
/ QALY gained
<$100,000
if ∆
diagnostic
yield >
20%
<$100,000
if ∆ cost < $300Patient or system factors alone are unlikely
to be sufficient
∆ diagnostic yield = 13%
∆ cost = $476
24-hr
Holter
Slide31Post discharge AF monitoringEvaluation 3: ILR vs. standard care
30
Patient with
cryptogenic
stroke or TIA
ECG in Hospital
Hospital
discharge
Age 70
CHADS-VASC ~4
3-year ILR
Standard care
Conclusion:
Not cost effective
ICER > $250,000
/ QALY gained
∆ diagnostic yield = 27%
∆ cost = $3400
24-hr
Holter
Slide32SummaryIn a stroke or TIA patient who received ECG in
hospital
7-days
of monitoring is likely cost effective compared to standard care
Unclear if cost effective vs. universal 24-hour monitoring
Unclear if 7-days is the optimal duration of monitoringUnlikely to be cost effective in a patient who has received in-patient Holter
/CICT/
cECG
monitoring (needs large incremental diagnostic yield)In a cryptogenic
stroke
or TIA
patient
who
received
ECG
and 24-hr Holter
30-day monitoring is unlikely cost effective compared to 24-hr Holter
ILR is not cost effective compared to standard careUnclear 24-hour Holter
is cost effective compared to no further monitoringUnclear if 30-days of monitoring is cost effective compared to 7-days in a patient cohort without prior 24-hr Holter
31
Slide33LimitationsMany comparisons of interest were not evaluated
Optimal sequence of monitoring technologies remains unknown
Many simplifying structural assumptions
Assume a constant rate of recurrent stroke
AF diagnosis after subsequent event
OAC adherence, quitting after a bleeding event, and re-starting after a clotting event
Model considers patients at average risk
Explored the influence of patient factors on cost effectiveness of monitoring
Do not consider correlation between patient factorsDo not consider the incidence of AF after stroke
Incident AF will be captured by longer-monitoring technologies (ILR)
32
Slide3433
Slide35Evaluation 1: 7-day ELR vs. standard careOAC Adherence
34
Slide36Evaluation 2: 30-day ELR vs. 24-hr Holter
OAC Adherence
35
Slide37Comparisons to other modelsKamel et al.
Cost effectiveness of dabigatran,
apixaban
vs. warfarin in post-stroke patients with AF
36
Kamel
et al.
QALY
CADTH
Base case OAC discontinuation
No OAC
discontinuation
QALY LY QALY LY
Dabigatran
4.27
2.57
4.01
2.64
4.12
Apixaban
4.19
2.63
4.08
2.74
4.24
Warfarin
3.91
2.56
4.02
2.63
4.14
* 3% annual discount rate
Utility weight for initial health state = 0.994-0.987 vs. 0.68
Slide38Comparisons to other modelsCoyle et al.Cost effectiveness of NOACs vs. warfarin in post-stroke patients with AF
Lifetime QALYs with warfarin
Coyle et al. Mild stroke (utility = 0.75): 4.45 QALYs
Coyle et al. Severe stroke (utility = 0.33): 2.2 QALYs
Our model Mixed cohort (utility = 0.68): 2.4 QALYs (3.75 LY)
37
* 5% annual discount rate
Our rate of recurrent stroke on warfarin is greater.
Coyle et al. = 0.035 (base rate of 0.016 x HR of 2.2)
Our model = 10.2% annual rate on aspirin x HR of 0.63 = 0.068
Slide39Model-based outcomes38
Outcomes
AF
without diagnosis
AF with diagnosis
Warfarin
Apixaban
Within 2
years (per 10,000)
Recurrent
IS,
severe or fatal
1,087
709
636
Recurrent stroke, any
1,724
1,223
1,093
ICH
47
99
42
GI bleed
171
276
264
Within
5 years (per 10,000)
Recurrent
IS,
severe or fatal
2,002
1,455
1,345
Recurrent stroke, any
3,189
2,459
2,266
ICH
87
170
80
GI bleed
316
487
471
LE
4.31
4.53
4.60
Slide40Model-based outcomes39
Outcomes
AF
without diagnosis
AF with diagnosis
Warfarin
Apixaban
LE
4.31
4.53
4.60
Discounted LY
3.59
3.75
3.80
Discounted QALYs
2.30
2.39
2.45
Costs
Baseline
229,025
238,458
241,473
OAC
0
843
2,870
Acute events
19,876
17,223
15,952
Total
248,901
256,525
260,294
Slide41Model Validation
40
Slide42Comparison to Oxford Vascular StudyOxford vascular study
440 TIA and 748 stroke patients (2002-2007) followed up for 5-10 years
17% prior history of stroke
18% prior diagnosis of AF
41
Life years
QALYs
TIA
4.3
3.3
Stroke
3.6
2.5
Mild
4.1
2.9
Moderate
2.8
1.7
Severe1.40.7
Overall3.82.7
Adjusted (12%)4.43.1
Life years
QALYs
No AF
7.2
4.73
AF without diagnosis
4.3
2.75
AF with diagnosis
Warfarin
4.5
2.88
Apixaban
4.6
2.95
Our model
Oxford Vascular Study
Slide43Comparison to Oxford Vascular StudyOxford vascular study
440 TIA and 748 stroke patients (2002-2007) followed up for 5-10 years
17% prior history of stroke
18% prior diagnosis of AF
42
5-year risk of recurrent stroke
TIA
16%
Stroke
20%
5-year risk of recurrent stroke
No AF
8.1%
AF without diagnosis
31.9%
AF with diagnosis
Warfarin
24.6%
Apixaban
22.7%
Our model
Oxford Vascular Study
Slide44Inputs: Natural history
Recurrent stroke
MI
ICH
GI bleed
Annual risk of event
No AF
2.2%
2.6%
0.3%
1.1%
AF
no OAC
10.2%
5.1%
0.3%
1.1%
HR
with OAC (compared to aspirin*)
Warfarin
0.38
0.89
2.60
2.50
Dabigatran
0.38
1.42
1.08
4.11
Rivaroxaban
0.36
0.72
1.74
3.65
Apixaban
0.35
0.78
1.09
2.23
43
Slide45Inputs: Costs and Utilities
Recurrent
stroke
Death
GI bleed
MI
ICH
TIA/Mild*
Mod-
Sev
Month of
Mortality (no OAC/OAC)
0%
11%
31%/44%
39% / 29%
Cost
$43,600
$9,975
$18,400
$36,000
$16,200
$55,000
QOL
- 0.03
- 0.01
- 0.05
- 0.02
- 0.13
History of
Annual mortality
(no AF/AF)
12 % / 16
%
16 % / 22 %
Direct health costs
18,400
18,400
18,400
30,900
Unpaid caregiver costs
15,000
15,000
15,000
25,000
QOL
0.65
0.62
0.68
0.31
Slide46Minimum diagnostic yield
to be cost
effective (100K)
45
61% of AF patients initiate OAC
100% of AF patients initiate OAC
7d
7d
30d
30d
Slide47Comparison to Yong et al.Cost effectiveness of 30-day ELR vs 24-hr Holter
46
* 5% annual discount rate
** assumes 16.1% AF prevalence
Key differences
Yong et al. are modeling a healthier cohort
Annual mortality rate = 5.4% per year vs. 12.1% per year
Baseline utility = 0.93 vs. 0.68
Baseline costs associated with history of stroke = $9301 vs. $43,200
Yong et al.
CADTH
30-day ELR
LE = 8.158 years
LE = 6.8 years
24-hr
Holter
LE = 8.138 years
LE = 6.2 years
ICER
Cost saving
$151,208 / QALY gained
Slide48AF diagnosis occurs over time
Assumed
30% AF prevalence
While undiagnosed, patients (with and without AF) may die or have recurrent stroke
Considered censored
Calculated the rate of AF identification in undiagnosed patients to match KM curves in Sanna
(NEJM 2014)
47
Slide49Cohort with 30% AF prevalence
48
Outcomes
Warfarin
Standard
practice ILR
Apixaban
Standard
practice ILR
Within 2
years (per 10,000)
Recurrent
IS,
severe or fatal
475
449
474
443
Recurrent stroke, any
780
746
777
735
ICH
51
54
50
49
GI bleed
183
190
183
189
LE
6.31
6.33
6.31
6.34
Discounted LY
4.82
4.83
4.82
4.84
Discounted QALY
3.18
3.19
3.18
3.19
Slide50Cohort with 30% AF prevalence
49
Outcomes
Warfarin
Standard
practice ILR
Apixaban
Standard
practice ILR
Costs
Baseline
$292,815
$293,792
$292,959
$294,14
Testing
$40
$1,129
$40
$1,129
OAC
$41
$118
$138
$402
Acute Events
$11,651
$11,404
$11,591
$11,227
TOTAL
$304,547
$306,443
$304,727
$306,971
Incr.
cost of ILR monitoring
$1,896
$2,244
Incr. QALYs
0.009
0.015
Incr. cost effectiveness ratio (ICERs) ($/QALY gained)
$205,169
$151,208
Slide51Sensitivity analysisILR monitoring vs. standard care
ICER >
100,000 per QALY gained is robust
Doubling the diagnosis rate (diagnosing patients faster)
W: $175,400 per QALY gained A: $132,600 per QALY gainedImmediate diagnosis in ILR arm
W: $154,300 per QALY gained A: $120,500 per QALY gained
Cost of implantation = $0 ICER > $100,000 per QALY gained
Cost of
monitoring =$0
ICER > $100,000 per QALY
gained
ICER < 100,000 per QALY gained Cost of implantation and monitoring 25% of base case
50
Slide52Comparison to Diamantopoulos et al.Cost effectiveness of ILR vs standard care
51
* 3.5% annual discount rate ** 5% annual discount rate
Key differences
Baseline mortality rate in
Diamantopoulos et al.
is based on UK life tables
(no increase in mortality rate for history of stroke)
Baseline utility = 0.76 vs. 0.68
Low costs of long-term care
e
.g., £578 per year after a mild stroke (vs. $43,200)
e
.g., £1,712 per year after a recurrent major stroke (vs. $65,000)
Diamantopoulos et al.*
CADTH**
ILR
7.367 QALY
4.84 QALY
Standard care
7.216 QALY
4.82 QALY
ICER
£ 17,175 / QALY gained
$151,208 / QALY gained