![Economic Evaluation of Insertable Cardiac Monitors to
Detect Atrial Fibrillation and Subsequently Moderate Stroke
Risk in a High-Risk Population in the United Kingdom
AUTHORS
Rinciog C1, Sawyer L1, Diamantopoulos
A1, Elkind M2, Reynolds M3, Tsintzos S4,
Ziegler PD5, Quiroz M5, Wolff C4,
Witte K6
AFFILIATIONS
1 Symmetron Ltd., London, England, United Kingdom
2 Columbia University, New York City, NY, United States
3 Baim Institute for Clinical Research, Health Economics and
Outcomes Research, Boston, United States
4 Global CRHF Health Economics and Outcomes Research,
Medtronic International Trading Sàrl, Tolochenaz, Switzerland
5 Medtronic Global CRHF Headquarters, Mounds View, MN, United
States
6 University of Leeds, Leeds, England, United Kingdom
INTRODUCTION
Atrial fibrillation (AF) increases the risk of
cerebrovascular events including stroke and transient
ischaemic attack (TIA). Oral anticoagulants (OACs) are
proven to reduce this risk in patients with confirmed AF.
In the recent REVEAL AF study, patients with risk
factors for AF and stroke fitted with an insertable cardiac
monitor (ICM) were found to have a high incidence of
AF despite negative baseline external monitor (≥24
hours). Currently it is unknown whether ICM monitoring
to detect AF in this population is cost-effective.
PURPOSE
The objective of this analysis was to evaluate the cost-
effectiveness of using ICM vs. annual 24h Holter
monitoring (standard of care, SoC) to detect AF and
guide oral anticoagulation administration for stroke
prevention in patients with elevated risk profiles
(CHADS2>2), in the UK National Health Service (NHS)
setting. AF detection guides anticoagulation decisions
that ultimately leads to better stroke prevention.
METHODS
The analysis used a Markov model to estimate lifetime
quality-adjusted life years (QALYs) and diagnostic,
monitoring and treatment costs from a UK NHS
perspective, all discounted at 3.5% annually1. The
model also estimated life-years gained and stroke
events avoided. A three-month cycle length was used.
The model separately simulated the detection of AF as
well as the incidence of clinical events (treatment-
related and otherwise) associated with the condition
(Figure 1). In the model’s hypothetical cohort, all
patients with underlying AF are assumed to be
undiagnosed at baseline. In each cycle, AF patients may
remain undiagnosed, have existing AF detected, or
develop AF. Once AF is detected, patients are assumed
to receive OAC therapy (Novel OAC [NOAC] therapy
was used in the base-case, warfarin in sensitivity
analysis) that reduces stroke risk by an average of
61.5%; by contrast, patients with undetected AF will
have no change in treatment.
The model was populated based on the prospective,
single arm, multicenter REVEAL AF study2,3. In this
study patients underwent monitoring with ICM for 18 to
30 months and AF episodes were recorded if they
lasted for 6 minutes or longer. The patient cohort’s
baseline characteristics, AF episode duration and rates
of AF detection by the ICM also came from this study.
The incidence of cerebrovascular events was predicted
based on patients’ AF status, baseline CHADS2 risk
score, any treatment being administered and age, with
probabilities synthesised from a range of published
sources.
Utilities and costs were sourced from existing
publications including a prior economic evaluation of
Reveal® ICMs in patients with cryptogenic stroke4.
Deterministic and one-way sensitivity analysis were
performed for key model parameters; probabilistic
sensitivity analysis (PSA) was performed as well (1,000
samples). Scenario analyses explored the impact of
alternative model assumptions or inputs on the results.
RESULTS
The total incremental cost for the ICM device for the average patient (aged 71.3 years and with a virtual CHADS2 score of 2.94)
was £2,891. When considering treatment and downstream events, the total per-patient costs were £13,370 and £12,946 for the
ICM and SoC strategy respectively. Base-case deterministic analysis found that ICM was associated with 0.20 more QALYs
over the patient lifetime (6.51 vs. 6.31) and 48 fewer strokes per 1,000 patients. With an incremental total cost of £1,424
compared to SoC, the base-case ICER of ICM was £7,140 per QALY gained. When assuming a willingness-to-pay (WTP) of
£20,000, the net monetary benefit of the ICM versus SoC was £2,564. This suggests ICM is a cost-effective intervention for use
in the UK NHS.
By enabling patients to start targeted OAC and thereby reducing the incidence of embolic stroke events, ICM generated cost-
savings both from the events themselves as well as their long-term costs (total stroke-related costs for ICM were £3,783 versus
£4,270 for SoC).
The ICER remained cost-effective across a wide range of scenarios and sensitivity analyses (Table 1 and Figure 2). Importantly,
in an alternative scenario analysis, that considered episodes longer than 5.5h to be diagnostic, the ICER rose to £17,659, still
below UK threshold limits. Amongst model parameters, ICERs were most sensitive to the assumed proportions of patients who
initiate and discontinue OAC after AF diagnosis, type of OAC used, and the intensity of traditional monitoring methods assumed
to be SoC.
When considering a £20,000 WTP threshold in PSA, the ICM’s probability of being cost-effective was 78.4%. This percentage
rose to 86.4% at a threshold of £30,000.
DISCUSSION
Overall, base-case model settings and all but one of the sensitivity analysis scenarios (when multiple inputs took extreme
values simultaneously) produced results indicating ICM monitoring as a cost-effective intervention for a high-risk AF population
in the UK. Our results may vary if one or more of these assumptions change: duration of detected AF required to initiate OAC,
type of OAC used (NOAC or Warfarin), patient adherence to OAC therapy following diagnosis and SoC monitoring detection
relative to ICM. Regardless, no single input value adopted to test uncertainty in the analysis tipped the ICER over the WTP
threshold of £20,000.
CONCLUSIONS
This analysis shows that long-term cardiac monitoring with an ICM is a cost-effective means of detecting AF and thus
preventing strokes in high-risk patients in the UK NHS.
Figure 1. Model Flow
Figure 2. Tornado Diagram. Abbreviations: AF, Atrial Fibrillation; CI,
Confidence Interval; ICM: Insertable Cardiac Monitor; OAC, Oral
Anticoagulant; QALY: Quality-Adjusted Life Year. Red=Changes to input
value that lead to decreased ICERs; Blue=Changes to input value that
lead to increased ICERs.
Table 1. Sub-group analysis by CHADS2 score and various scenario analyses
Analysis Description Incremental Costs (£) Incremental QALYs
ICER
ICM vs SoC
Base-Case = NOAC after 6mins of AF in 100% of patients £1,424 0.20 £7,140
Sub-group Analyses
CHADS2 score 2 £1,677 0.16 £10,735
CHADS2 score 3 £1,461 0.17 £8,425
CHADS2 score 4, 5, and 6 £1,442 0.15 £9,463
Scenario Analyses
Choice of OAC = warfarin £1,075 0.14 £7,900
Disallow treatment discontinuation due to reasons
unrelated to bleeding (i.e. Full Adherence to NOAC)
£1,611 0.36 £4,427
HR ICM vs. SoC (diagnostic yield) = 8.78 (CRYSTAL-AF) £1,494 0.17 £8,793
Monitoring costs for SoC = assume pulse check (zero
cost) & HR of ICM vs. SoC = 1/24th of 24h Holter (33.9/24)
£2,170 0.21 £10,323
Monitoring costs for SoC = assume pulse check (zero-
cost) & base-case HR of ICM vs. SoC = 33.9
£2,168 0.20 £10,874
OAC initiation only for AF episode duration lasting for
≥5.5 hours
£1,722 0.10 £17,693
OAC uptake after AF diagnosis = 66.35% £1,592 0.14 £11,145
REFERENCES
1National Institute for Health and Care Excellence. Guide to the
methods of technology appraisal. London. 2013
2Medtronic. REVEAL AF Data on File. 2018.
3Reiffel JA, Verma A, Kowey PR, Halperin JL, Gersh BJ, Wachter R, et
al. Incidence of Previously Undiagnosed Atrial Fibrillation Using
Insertable Cardiac Monitors in a High-Risk Population: The REVEAL AF
Study. JAMA cardiology. 2017;2(10):1120-7
4Diamantopoulos A, Sawyer LM, Lip GY, Witte KK, Reynolds MR,
Fauchier L, et al. Cost-effectiveness of an insertable cardiac monitor
to detect atrial fibrillation in patients with cryptogenic stroke.
International journal of stroke. 2016;11(3):302-12](https://image.slidesharecdn.com/suspectedafrevealposterpmd92final-181126183014/85/Insertable-cardiac-monitors-to-detect-AF-and-subsequently-moderate-stroke-risk-in-a-high-risk-population-1-320.jpg)
Insertable cardiac monitors to detect AF and subsequently moderate stroke risk in a high-risk population
- 1. Economic Evaluation of Insertable Cardiac Monitors to Detect Atrial Fibrillation and Subsequently Moderate Stroke Risk in a High-Risk Population in the United Kingdom AUTHORS Rinciog C1, Sawyer L1, Diamantopoulos A1, Elkind M2, Reynolds M3, Tsintzos S4, Ziegler PD5, Quiroz M5, Wolff C4, Witte K6 AFFILIATIONS 1 Symmetron Ltd., London, England, United Kingdom 2 Columbia University, New York City, NY, United States 3 Baim Institute for Clinical Research, Health Economics and Outcomes Research, Boston, United States 4 Global CRHF Health Economics and Outcomes Research, Medtronic International Trading Sàrl, Tolochenaz, Switzerland 5 Medtronic Global CRHF Headquarters, Mounds View, MN, United States 6 University of Leeds, Leeds, England, United Kingdom INTRODUCTION Atrial fibrillation (AF) increases the risk of cerebrovascular events including stroke and transient ischaemic attack (TIA). Oral anticoagulants (OACs) are proven to reduce this risk in patients with confirmed AF. In the recent REVEAL AF study, patients with risk factors for AF and stroke fitted with an insertable cardiac monitor (ICM) were found to have a high incidence of AF despite negative baseline external monitor (≥24 hours). Currently it is unknown whether ICM monitoring to detect AF in this population is cost-effective. PURPOSE The objective of this analysis was to evaluate the cost- effectiveness of using ICM vs. annual 24h Holter monitoring (standard of care, SoC) to detect AF and guide oral anticoagulation administration for stroke prevention in patients with elevated risk profiles (CHADS2>2), in the UK National Health Service (NHS) setting. AF detection guides anticoagulation decisions that ultimately leads to better stroke prevention. METHODS The analysis used a Markov model to estimate lifetime quality-adjusted life years (QALYs) and diagnostic, monitoring and treatment costs from a UK NHS perspective, all discounted at 3.5% annually1. The model also estimated life-years gained and stroke events avoided. A three-month cycle length was used. The model separately simulated the detection of AF as well as the incidence of clinical events (treatment- related and otherwise) associated with the condition (Figure 1). In the model’s hypothetical cohort, all patients with underlying AF are assumed to be undiagnosed at baseline. In each cycle, AF patients may remain undiagnosed, have existing AF detected, or develop AF. Once AF is detected, patients are assumed to receive OAC therapy (Novel OAC [NOAC] therapy was used in the base-case, warfarin in sensitivity analysis) that reduces stroke risk by an average of 61.5%; by contrast, patients with undetected AF will have no change in treatment. The model was populated based on the prospective, single arm, multicenter REVEAL AF study2,3. In this study patients underwent monitoring with ICM for 18 to 30 months and AF episodes were recorded if they lasted for 6 minutes or longer. The patient cohort’s baseline characteristics, AF episode duration and rates of AF detection by the ICM also came from this study. The incidence of cerebrovascular events was predicted based on patients’ AF status, baseline CHADS2 risk score, any treatment being administered and age, with probabilities synthesised from a range of published sources. Utilities and costs were sourced from existing publications including a prior economic evaluation of Reveal® ICMs in patients with cryptogenic stroke4. Deterministic and one-way sensitivity analysis were performed for key model parameters; probabilistic sensitivity analysis (PSA) was performed as well (1,000 samples). Scenario analyses explored the impact of alternative model assumptions or inputs on the results. RESULTS The total incremental cost for the ICM device for the average patient (aged 71.3 years and with a virtual CHADS2 score of 2.94) was £2,891. When considering treatment and downstream events, the total per-patient costs were £13,370 and £12,946 for the ICM and SoC strategy respectively. Base-case deterministic analysis found that ICM was associated with 0.20 more QALYs over the patient lifetime (6.51 vs. 6.31) and 48 fewer strokes per 1,000 patients. With an incremental total cost of £1,424 compared to SoC, the base-case ICER of ICM was £7,140 per QALY gained. When assuming a willingness-to-pay (WTP) of £20,000, the net monetary benefit of the ICM versus SoC was £2,564. This suggests ICM is a cost-effective intervention for use in the UK NHS. By enabling patients to start targeted OAC and thereby reducing the incidence of embolic stroke events, ICM generated cost- savings both from the events themselves as well as their long-term costs (total stroke-related costs for ICM were £3,783 versus £4,270 for SoC). The ICER remained cost-effective across a wide range of scenarios and sensitivity analyses (Table 1 and Figure 2). Importantly, in an alternative scenario analysis, that considered episodes longer than 5.5h to be diagnostic, the ICER rose to £17,659, still below UK threshold limits. Amongst model parameters, ICERs were most sensitive to the assumed proportions of patients who initiate and discontinue OAC after AF diagnosis, type of OAC used, and the intensity of traditional monitoring methods assumed to be SoC. When considering a £20,000 WTP threshold in PSA, the ICM’s probability of being cost-effective was 78.4%. This percentage rose to 86.4% at a threshold of £30,000. DISCUSSION Overall, base-case model settings and all but one of the sensitivity analysis scenarios (when multiple inputs took extreme values simultaneously) produced results indicating ICM monitoring as a cost-effective intervention for a high-risk AF population in the UK. Our results may vary if one or more of these assumptions change: duration of detected AF required to initiate OAC, type of OAC used (NOAC or Warfarin), patient adherence to OAC therapy following diagnosis and SoC monitoring detection relative to ICM. Regardless, no single input value adopted to test uncertainty in the analysis tipped the ICER over the WTP threshold of £20,000. CONCLUSIONS This analysis shows that long-term cardiac monitoring with an ICM is a cost-effective means of detecting AF and thus preventing strokes in high-risk patients in the UK NHS. Figure 1. Model Flow Figure 2. Tornado Diagram. Abbreviations: AF, Atrial Fibrillation; CI, Confidence Interval; ICM: Insertable Cardiac Monitor; OAC, Oral Anticoagulant; QALY: Quality-Adjusted Life Year. Red=Changes to input value that lead to decreased ICERs; Blue=Changes to input value that lead to increased ICERs. Table 1. Sub-group analysis by CHADS2 score and various scenario analyses Analysis Description Incremental Costs (£) Incremental QALYs ICER ICM vs SoC Base-Case = NOAC after 6mins of AF in 100% of patients £1,424 0.20 £7,140 Sub-group Analyses CHADS2 score 2 £1,677 0.16 £10,735 CHADS2 score 3 £1,461 0.17 £8,425 CHADS2 score 4, 5, and 6 £1,442 0.15 £9,463 Scenario Analyses Choice of OAC = warfarin £1,075 0.14 £7,900 Disallow treatment discontinuation due to reasons unrelated to bleeding (i.e. Full Adherence to NOAC) £1,611 0.36 £4,427 HR ICM vs. SoC (diagnostic yield) = 8.78 (CRYSTAL-AF) £1,494 0.17 £8,793 Monitoring costs for SoC = assume pulse check (zero cost) & HR of ICM vs. SoC = 1/24th of 24h Holter (33.9/24) £2,170 0.21 £10,323 Monitoring costs for SoC = assume pulse check (zero- cost) & base-case HR of ICM vs. SoC = 33.9 £2,168 0.20 £10,874 OAC initiation only for AF episode duration lasting for ≥5.5 hours £1,722 0.10 £17,693 OAC uptake after AF diagnosis = 66.35% £1,592 0.14 £11,145 REFERENCES 1National Institute for Health and Care Excellence. Guide to the methods of technology appraisal. London. 2013 2Medtronic. REVEAL AF Data on File. 2018. 3Reiffel JA, Verma A, Kowey PR, Halperin JL, Gersh BJ, Wachter R, et al. Incidence of Previously Undiagnosed Atrial Fibrillation Using Insertable Cardiac Monitors in a High-Risk Population: The REVEAL AF Study. JAMA cardiology. 2017;2(10):1120-7 4Diamantopoulos A, Sawyer LM, Lip GY, Witte KK, Reynolds MR, Fauchier L, et al. Cost-effectiveness of an insertable cardiac monitor to detect atrial fibrillation in patients with cryptogenic stroke. International journal of stroke. 2016;11(3):302-12