Author + information
- Received September 4, 2015
- Revision received September 24, 2015
- Accepted September 25, 2015
- Published online December 22, 2015.
- Vivek Y. Reddy, MD∗∗ (, )
- Ronald L. Akehurst, MFPHM†,
- Shannon O. Armstrong, BA‡,
- Stacey L. Amorosi, MA§,
- Stephen M. Beard, MSc‖ and
- David R. Holmes Jr., MD¶
- ∗Mount Sinai Medical Center, New York, New York
- †University of Sheffield, Sheffield, United Kingdom
- ‡GfK, Wayland, Massachusetts
- §Boston Scientific, Marlborough, Massachusetts
- ‖BresMed, Sheffield, United Kingdom
- ¶Mayo Clinic, Rochester, Minnesota
- ↵∗Reprint requests and correspondence:
Dr. Vivek Y. Reddy, Helmsley Trust, Electrophysiology Center, Icahn School of Medicine at Mount Sinai, One Gustave L. Levy Place, Box 1030, New York, New York 10029.
Background Left atrial appendage closure (LAAC) and nonwarfarin oral anticoagulants (NOACs) have emerged as safe and effective alternatives to warfarin for stroke prophylaxis in patients with nonvalvular atrial fibrillation (AF).
Objectives This analysis assessed the cost-effectiveness of warfarin, NOACs, and LAAC with the Watchman device (Boston Scientific, Marlborough, Massachusetts) for stroke risk reduction in patients with nonvalvular AF at multiple time points over a lifetime horizon.
Methods A Markov model was developed to assess the cost-effectiveness of LAAC, NOACs, and warfarin from the perspective of the Centers for Medicare & Medicaid Services over a lifetime (20-year) horizon. Patients were 70 years of age and at moderate risk for stroke and bleeding. Clinical event rates, stroke outcomes, and quality of life information were drawn predominantly from PROTECT AF (Watchman Left Atrial Appendage System for Embolic Protection in Patients with Atrial Fibrillation) 4-year data and meta-analyses of warfarin and NOACs. Costs for stroke risk reduction therapies, treatment of associated acute events, and long-term care following disabling stroke were presented in 2015 U.S. dollars.
Results Relative to warfarin, LAAC was cost-effective at 7 years ($42,994/quality-adjusted life-years [QALY]), and NOACs were cost-effective at 16 years ($48,446/QALY). LAAC was dominant over NOACs by year 5 and warfarin by year 10. At 10 years, LAAC provided more QALYs than warfarin and NOACs (5.855 vs. 5.601 vs. 5.751, respectively). In sensitivity analyses, LAAC remained cost-effective relative to warfarin ($41,470/QALY at 11 years) and NOACs ($21,964/QALY at 10 years), even if procedure costs were doubled.
Conclusions Both NOACs and LAAC with the Watchman device were cost-effective relative to warfarin, but LAAC was also found to be cost-effective and to offer better value relative to NOACs. The results of this analysis should be considered when formulating policy and practice guidelines for stroke prevention in AF.
Atrial fibrillation (AF) affects nearly 6 million Americans, a number expected to double by 2050 (1). Stroke is the most debilitating and costly consequence of AF. Medicare spends an estimated $16 billion annually on AF, with AF-related stroke accounting for nearly one-half of this expense (2,3). For decades, warfarin has been the standard of care for stroke prophylaxis in AF. Although inexpensive and effective, warfarin is associated with an increased risk of bleeding, lower quality of life (QoL), and high patient nonadherence.
More recently, 4 nonwarfarin oral anticoagulants (NOACs) (dabigatran, rivaroxaban, apixaban, and edoxaban) were approved by the U.S. Food and Drug Administration (FDA). As a group, they are noninferior to warfarin for ischemic stroke reduction and superior for hemorrhagic stroke and all-cause mortality, but they have an increased risk for gastrointestinal bleeding (4). As with warfarin, the effectiveness of these drugs is contingent on patient adherence.
For patients who are poor candidates for long-term oral anticoagulation, percutaneous left atrial appendage closure (LAAC) is a device-based alternative for stroke prophylaxis in AF (5). The first FDA-approved LAAC device, the Watchman device (Boston Scientific, Marlborough, Massachusetts), was demonstrated to be noninferior to warfarin for ischemic stroke reduction and superior for hemorrhagic stroke and all-cause mortality. However, it is associated with procedure-related complications that predictably diminish in frequency with operator experience (6). Importantly, most patients can discontinue lifelong anticoagulation following Watchman device implantation (6).
Both clinical and economic value are important in evaluating new therapies. Published U.S. economic analyses have largely focused on the cost-effectiveness of oral anticoagulants, and these reports invariably use a lifetime analysis spanning 20 to 35 years (7–9). Herein, we estimate the cost-effectiveness of all stroke prevention strategies available in the United States at multiple points over a lifetime horizon. Although both NOACs and LAAC would initially be expected to be more costly than warfarin, their clinical benefits may supervene with time, thereby altering cost-effectiveness. This comprehensive assessment may aid clinicians and health care decision makers in making informed choices regarding patient treatment.
A Markov model was developed in Excel to evaluate the cost-effectiveness of 3 treatment strategies:
1. LAAC with the Watchman device (Boston Scientific)
2. NOACs as a class
3. Adjusted-dose warfarin
The model was constructed from the perspective of the Centers for Medicare & Medicaid Services (CMS), with a lifetime horizon (defined as 20 years) and a 3-month cycle length. Within each cycle, patients could experience clinical events leading to death, disability, and/or therapy discontinuation and could incur associated costs and QoL adjustments.
Cost-effectiveness was evaluated using the U.S. accepted willingness-to-pay threshold of $50,000 per quality-adjusted life-year (QALY) gained, and it was reported as the incremental cost-effectiveness ratio (ICER), which provides a standardized approach to measure cost per unit of health improvement in and across health states. Cost-effectiveness was assessed annually to determine whether there was an observable time horizon over which treatment options reached accepted levels of cost-effectiveness.
Model structure and clinical pathways
The model began with patients assigned to 1 of 3 treatments (Figures 1A and 1B). Patients in the LAAC arm faced a 1-time procedure-related risk, including ischemic stroke resulting from air embolism (1.10%), major bleeding (0.60%), and pericardial effusion (4.80%). Patients undergoing LAAC could experience a successful or failed implantation procedure. In accordance with the PROTECT AF (Watchman Left Atrial Appendage System for Embolic Protection in Patients With Atrial Fibrillation) trial, patients who had successful device implantation were assumed to receive warfarin for 45 days, aspirin plus clopidogrel from 46 days to 6 months, and aspirin thereafter. Following a failed procedure, patients were assumed to continue warfarin therapy. Patients receiving warfarin or NOACs could discontinue therapy following a bleeding event or for nonclinical reasons. Patients who discontinued primary drug therapy were assumed to switch to aspirin. Discontinuation of second-line therapy was assumed to result in no treatment.
On entering the model, patients were assumed to be “well,” or in normal, good health. Patients transitioned to new health states following a clinical event or death. Only ischemic and hemorrhagic stroke affected disability outcomes. Patients who had a second stroke could either remain in the same health state or worsen to greater disability. Transient ischemic attack (TIA) and systemic embolism led to patients being well with a history of stroke, which increased their risk of subsequent stroke. All events, except TIA, could lead to death.
Clinical inputs were taken from several sources (Table 1). For LAAC, procedural complications and event probabilities were drawn from PROTECT AF at 4-year follow-up (6). Relative risks for post-procedural stroke and bleeding were used to apply a standard efficacy estimate to the derived baseline risks. For warfarin, the relative risk of stroke was derived from a meta-analysis of warfarin trials (10). Risk data for all other warfarin-related clinical events were estimated from a pooled analysis of multiple AF clinical trials involving warfarin (4,16–20). NOAC clinical event rates were derived from a NOAC meta-analysis and from individual NOAC trials (4,12,13,21,22).
Probabilities of death following systemic embolism, extracranial hemorrhage, or myocardial infarction were derived from Healthcare Utilization Project mortality rates for these diagnoses (23). Risk of death from unrelated causes was obtained from U.S. life tables, with disabled patients facing a 2.3-fold greater risk of death (24,25).
Baseline risk of stroke was assigned according to CHA2DS2-VASc (congestive heart failure, hypertension, age 75 years or older, diabetes mellitus, stroke–vascular disease) scores, and risk of bleeding was derived from HAS-BLED (hypertension, abnormal renal/liver function, stroke, bleeding history or predisposition, labile international normalized ratio, elderly [>65 years], drugs/alcohol concomitantly) scores (11,26). Because neither measure was prospectively collected in the PROTECT AF or NOAC trials, both were estimated using available baseline characteristic data from the trials. Patients were assumed to be 70 years of age, with a mean CHA2DS2-VASc score of 3.2 (annual stroke risk 3.4%), and a HAS-BLED score of 2 (annual bleeding risk 1.88%) (11,26). To account for increasing risk with age, rates of ischemic and hemorrhagic stroke were increased by 1.4- and 1.97-fold per decade, respectively (27,28). Patients experiencing a TIA or systemic embolism were assumed to have a 2.6-fold higher risk of experiencing a second ischemic event (28).
Health state utilities and stroke outcomes
As is conventional in cost-effectiveness analyses, QoL was captured in the model as health utility. Health utility values were on a scale of 0 to 1, with 1 representing perfect health and 0 representing death. Utility values for well with warfarin (0.987), well with NOAC (0.994), and well with aspirin (0.998) are consistent with values used in other AF stroke reduction analyses (28–32). The utility value for well with LAAC (0.999) was calculated by applying the Nichol ordinary least squares (OLS) algorithm to Short Form-12 (SF-12) data collected during PROTECT AF (33,34). The utility weights for all well-based health states were applied as a multiplying factor to an underlying baseline utility of 0.82, representing QoL at age 70 years (29). Baseline utility was decremented by 2% per decade to account for general decline in QoL with advancing age (35).
Because stroke is the most debilitating consequence of AF, QoL also was assessed by stroke severity. Stroke outcomes were assigned using the modified Rankin score (MRS) and were characterized as nondisabling (MRS 0 to 2), moderately disabling (MRS 3), severely disabling (MRS 4 to 5), and fatal (MRS 6) (Table 2). Stroke outcomes for LAAC were gathered from additional analyses of PROTECT AF at 4 years. Warfarin stroke outcomes were estimated using a weighted average of outcomes from 4 warfarin trials (16,17,19,20). NOAC stroke outcomes were derived from 2 of the 4 pivotal trials (12,13). Because only the rate of nondisabling strokes was reported, the inverse represented disabling and fatal strokes, with the distribution of moderately disabling, severely disabling, and fatal strokes assumed to be the same as with warfarin.
Additionally, a series of disutilities was applied in the model for acute clinical events, representing a 1-time decrement to QoL experienced for a finite length of time. Utility decrements were assessed for stroke (−0.139), extracranial hemorrhage (−0.181), TIA (−0.103), systemic embolism (−0.120), and myocardial infarction (−0.125) (36). A value of −0.0315 was used for the LAAC procedure itself, which assumed a 2-week disruption to healthy life.
QALYs were calculated by multiplying the health state utility value of each health state by the mean time spent in the health state. Future QALYs were discounted at an annual rate of 3% (37).
The model incorporated all direct health care costs for the therapies and treatment of associated acute events, as well as costs for long-term care following a disabling stroke (Table 3). Costs for acute events were taken from U.S. 2015 diagnosis-related group (DRG) national average values, and costs for post-stroke inpatient rehabilitation were from 2015 case-mix group (CMG) reimbursement rates (38,39). Long-term stroke disability costs were obtained from previous publications (30–32,40–42).
LAAC procedure costs were calculated as a weighted average of the 2 newly created DRGs for percutaneous intracardiac procedures (273 and 274), plus the cost of 2 follow-up transesophageal echocardiograms (43,44). The annual cost of warfarin therapy was also applied for LAAC-treated patients who were unable to discontinue warfarin. Warfarin costs were from U.S. pharmaceutical wholesale acquisition data, in combination with reimbursement rates for Current Procedural Terminology (CPT) codes related to international normalized ratio monitoring (44,45). NOAC costs were calculated as an average of U.S. wholesale acquisition costs for the first 3 approved drugs: dabigatran, rivaroxaban, and apixaban (45). All costs, with the exception of LAAC procedure costs, are in 2015 U.S. dollars and are discounted at an annual rate of 3%.
To assess the impact of parameter uncertainty on model results, 1-way sensitivity analysis and probabilistic sensitivity analysis (PSA) were undertaken. One-way sensitivity analysis reveals which inputs have the greatest impact on results, and PSA generates thousands of outcomes by using varied inputs to allow estimation of the effect of individual parameter uncertainty on uncertainty around model results (46). Inputs were varied within 95% confidence intervals (CIs), where available, and by ±20% where CIs were not published. Additionally, the impact of LAAC procedure cost was explored independently, with 100% variation. Table 1 shows ranges and distributions used for clinical inputs. Stroke outcomes assumed a Dirichlet distribution, and health state utilities assumed a beta distribution. All costs were varied by ±20% and assumed a gamma distribution. The PSA followed a standard Monte Carlo approach on the basis of 5,000 randomly drawn simulations of parameter values.
LAAC versus warfarin
As expected, LAAC was more expensive than warfarin in the first post-procedure year, and patients had fewer QALYs because of the disutility applied for the procedure (Table 4, Central Illustration). By year 3, LAAC-treated patients gained more QALYs than did warfarin-treated patients (0.015). LAAC became cost-effective relative to warfarin by year 7, with a cost per QALY gained of $42,994. By year 10, LAAC was dominant (more effective and less costly) over warfarin and remained so throughout the remainder of the 20-year time horizon. Over a lifetime, LAAC provided an additional 0.506 life-years and 0.638 QALYs relative to warfarin. Online Figure 1 shows the total QALYs for all 3 treatment groups over 20 years.
NOACs versus warfarin
NOACs were more effective than warfarin in year 1 and achieved cost-effectiveness relative to warfarin at year 16 ($48,446/QALY). NOACs were not cost saving relative to warfarin over 20 years, although the ICER continued to decrease over time, such that the cost per QALY at 20 years was $40,602.
LAAC versus NOACs
When comparing LAAC with NOACs, LAAC was expectedly more expensive than NOACs in the first post-procedure year. However, by year 5, LAAC was less expensive ($20,892 vs. $20,924) and more effective (3.455 vs. 3.448 QALYs) than NOACs. LAAC remained dominant over NOACs throughout the modeled time horizon. Over a patient’s lifetime, LAAC was estimated to provide an additional 0.298 life-years and 0.349 QALYs relative to NOACs.
One-way sensitivity analysis
Tornado diagrams illustrating the 10 most impactful variables in descending order of influence at 20 years are depicted in Figure 2 (Online Figures 2A to 2C for 10-year results). One-way sensitivity analyses of LAAC versus warfarin demonstrated that the health utility values used for well with LAAC had a significant impact on model results, with this being the only variable for which the upper threshold increased the ICER at 20 years beyond $50,000/QALY (Figure 2A). The baseline risk of ischemic stroke also influenced the ICER, with higher risks resulting in more favorable cost-effectiveness for LAAC. All other variables had a <1% impact on model outcomes. When comparing LAAC with NOACs (Figure 2B), the 20-year results were most sensitive to the utility values for well with LAAC and well with NOACs along with the baseline risk of bleeding. However, only varying well with LAAC increased the ICER beyond the $50,000 threshold. In the NOAC versus warfarin analysis (Figure 2C), the baseline risk of bleeding had the greatest impact on modeled results, followed by the utility value for well with NOAC, baseline risk of stroke, and the proportion of NOAC ischemic strokes that were severely disabling.
Additional analyses demonstrated that, even with a doubling of LAAC procedure cost, LAAC remained cost-effective. As expected, time to cost-effectiveness increased, with LAAC becoming cost-effective at 11 years relative to warfarin ($41,470/QALY) and at 10 years relative to NOACs ($21,964/QALY); however, LAAC remained dominant over both drug comparators over the lifetime analysis.
Probabilistic sensitivity analysis
PSA simulations (Figure 3) demonstrated that LAAC had lower average total costs than warfarin and NOACs at 20 years: $31,127 (95% CI: $17,350 to $51,067) versus $48,817 (95% CI: $28,719 to $77,484) versus $58,254 (95% CI: $44,865 to $77,763). Over 20 years, there was a 94% probability that LAAC was cost saving relative to warfarin, a 97% probability that LAAC provided more QALYs, and a 100% probability that it provided more life-years. The overall probability of cost-effectiveness was 98%, using a willingness-to-pay threshold of $50,000/QALY. At 20 years, there was a 95% probability that LAAC was cost-effective relative to NOACs and a 75% probability that NOACs were cost-effective relative to warfarin. Similar outcomes, albeit with greater scatter, are seen when the PSAs were conducted at 10 years (Online Figures 3A to 3C).
This analysis demonstrates that both novel therapies, NOACs and LAAC with the Watchman device, are cost-effective relative to warfarin for stroke risk reduction in patients with nonvalvular AF. When compared over a lifetime, LAAC proved to be the most cost-effective treatment. LAAC becomes cost-effective relative to warfarin at 7 years, with an ICER of $42,994. In contrast, NOACs require twice as long to achieve cost-effectiveness relative to warfarin, with an ICER of $48,446. LAAC was less costly than warfarin at 10 years and NOACs at 5 years, with cost savings generated annually thereafter. NOACs are more effective than warfarin, but they remain more expensive throughout the lifetime horizon.
Sensitivity analyses suggest that these cost-effectiveness results are robust and not overly sensitive to variation in individual parameters, with the exception of well with LAAC, well with NOACs, and baseline risk of bleeding. Specifically, 1-way sensitivity analyses found that LAAC is more cost-effective in patients at higher risk of stroke and bleeding. Given the importance of baseline utility values in this analysis, future prospective studies of these novel therapies should evaluate QoL over time. PSA indicates a high probability that LAAC is cost saving relative to warfarin and NOACs over 20 years.
Although costs of any new health intervention must be considered, patient outcomes remain of paramount importance. Cost-effectiveness modeling uses 1 metric to account for cost, clinical effectiveness, and patient outcomes. For stroke prevention, the risk and cost of stroke are highly important, but so too are QoL and functional outcomes following stroke. Indeed, a preference study of stroke outcomes revealed that most patients rate severe disability as worse than death (47).
LAAC-treated patients experienced fewer disabling strokes than did patients taking warfarin or NOACs: 79% of strokes in patients who had undergone LAAC resulted in an MRS score of 0 to 2 versus 24% of strokes in patients taking warfarin and 44% of strokes in patients taking NOACs. This finding suggests that most LAAC-treated patients can return to daily life without assistance following stroke, whereas more than one-half the patients who have a stroke while taking oral anticoagulants require lifetime assistance. Furthermore, previously published SF-12 data collected during PROTECT AF found an overall improvement in QoL and physical functioning in LAAC-treated patients relative to warfarin-treated patients at 12 months (34). Disability following a stroke has long-term implications for both patients and health care costs, thus warranting further study.
NOACs offer significant advantages over warfarin, but the problem of therapy adherence persists. Analysis of Medicare MarketScan data found that only 38% to 40% of patients with AF who had a moderate to high risk of stroke received a prescription for anticoagulation therapy, and 46% of these patients had a gap in anticoagulation therapy during the 2 years analyzed (48). Additionally, 63% of all stroke-related hospitalizations occurred in patients who were not receiving oral anticoagulation. With nearly $8 billion spent annually on AF-associated strokes (3), improving adherence is crucial to reducing this cost burden.
By its very nature, LAAC is not subject to patient adherence issues. Once implanted, the device provides lifelong stroke prophylaxis without the risk of complications associated with oral anticoagulants. Although the procedure has risks, they are reduced with operator experience. In recognition of the clinical risk/benefit relationship, the 2014 American Heart Association/American Stroke Association updated guidelines include consideration of LAAC in patients at high risk for stroke who are “unsuitable” for long-term anticoagulation, if they can tolerate the risk of at least 45 days of post-procedural anticoagulation (49).
To the best of our knowledge, the only other study assessing the cost-effectiveness of LAAC in the United States was published in abstract form. In this study, LAAC dominated all NOACs, except for the 110-mg dose of dabigatran, and was cost-effective, but not cost saving relative to warfarin (50). This difference from our analysis is likely explained by our detailed patient-level accounting of stroke severity and QoL. A cost-effectiveness analysis for the Canadian health system compared LAAC with Watchman, dabigatran, and warfarin (51). Despite overall conclusions directionally consistent with our analysis (particularly both LAAC and dabigatran cost-effectiveness relative to warfarin and LAAC dominance over dabigatran), the earlier analysis had several limitations. First, only 1.5-year follow-up PROTECT AF data were used, versus 3.8-year follow-up data in our present analysis. Second, whereas only dabigatran outcomes were available at that time, the present analysis additionally incorporates outcomes with rivaroxaban, apixaban, and edoxaban. Third, whereas that earlier analysis examined only a lifetime horizon, the present analysis considers cost-effectiveness at multiple time points. Finally, the present analysis uses a U.S. payer perspective.
Our model allowed for only 1 clinical event per 3-month cycle, whereas multiple events may occur. Data were taken from multiple sources, including meta-analyses and randomized controlled trials. Individual trials had different time horizons, and all data were extrapolated to 20 years. By necessity, an indirect comparison method was used, with warfarin as the common control value. Although this methodology is well established in health economic analyses (52), these data are not derived from a direct, randomized comparison. However, it will likely be many years before an LAAC versus NOAC randomized trial is performed. Additionally, the datasets from which probabilities were drawn were all on the basis of intention-to-treat analyses, whereas the model allowed for the possibility of therapy change. Finally, treatments administered in clinical practice may vary in effectiveness compared with that observed in randomized trials, which generally enroll selected patients, achieve higher levels of adherence, and monitor patients more intensively.
Both NOACs and LAAC with the Watchman device have the potential to change the treatment paradigm for patients with nonvalvular AF who are at risk of stroke. Both novel therapies demonstrated cost-effectiveness relative to warfarin, but only LAAC demonstrated cost savings by year 10 relative to warfarin and by year 5 relative to NOACs.
COMPETENCY IN SYSTEMS-BASED PRACTICE: As an alternative to long-term warfarin therapy for stroke prophylaxis in patients with AF who are poor candidates for long-term anticoagulation, percutaneous LAAC become cost-effective at 7 years. On the basis of current cost estimates, LAAC would be associated with more QALYs gained by 10 years than warfarin or NOACs.
TRANSLATIONAL OUTLOOK: Longer-term clinical studies are needed to define more clearly the return on investment in nonpharmacologic strategies to prevent stroke in patients with AF and incorporate the findings into clinical practice guidelines and reimbursement policies.
For supplemental figures, please see the online version of this article.
Dr. Reddy, Prof. Akehurst, Ms. Armstrong, and Mr. Beard are paid consultants to Boston Scientific, manufacturer of the Watchman device discussed in this paper. Ms. Amorosi is a full-time employee of Boston Scientific. Dr. Holmes and the Mayo Clinic have a financial interest in the Watchman device. Dr. Reddy is a paid consultant for and has received grant support from Coherex and St. Jude Medical.
Bruce D. Lindsay, MD, served as Guest Editor for this paper.
- Abbreviations and Acronyms
- atrial fibrillation
- incremental cost-effectiveness ratio
- left atrial appendage closure
- modified Rankin score
- nonwarfarin oral anticoagulant
- probabilistic sensitivity analysis
- quality-adjusted life-year
- quality of life
- transient ischemic attack
- Received September 4, 2015.
- Revision received September 24, 2015.
- Accepted September 25, 2015.
- 2015 American College of Cardiology Foundation
- Go A.S.,
- Mozaffarian D.,
- Roger V.L.,
- et al.
- Harrington A.R.,
- Armstrong E.P.,
- Nolan P.E. Jr..,
- et al.
- ACTIVE Writing Group of the ACTIVE Investigators
- Connelly S.J.,
- Laupaucis A.,
- Gent M.,
- et al.
- Stroke Prevention in Atrial Fibrillation investigators
- HCUP Databases. Healthcare Cost and Utilization Project (HCUP)
- ↵Social Security Administration. Actuarial Life Table 2011. Available at: http://www.ssa.gov/oact/STATS/table4c6.html. Accessed October 4, 2015.
- Dennis M.S.,
- Burn J.P.S.,
- Sandercock P.A.G.,
- et al.
- Ariesen M.J.,
- Claus S.P.,
- Rinkel G.J.E.,
- et al.
- Shah S.V.,
- Gage B.F.
- Alli O.,
- Doshi S.,
- Kar S.,
- et al.
- ↵Centers for Medicare & Medicaid Services. Acute Inpatient PPS FY 2015 Final Rule Tables. 2014. Available at: https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/FY2015-IPPS-Final-Rule-Home-Page-Items/FY2015-Final-Rule-Tables.html. Accessed October 4, 2015.
- ↵Centers for Medicare & Medicaid Services. Inpatient Rehabilitation Facility PPS Data Files. 2014. Available at: https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/InpatientRehabFacPPS/Data-Files.html. Accessed October 4, 2015.
- Cipriano L.E.,
- Steinberg M.L.,
- Gazelle G.S.,
- et al.
- Caro J.J.,
- Huybrechts K.F.
- ↵Centers for Medicare & Medicaid Services. Acute IPPS FY 2016 Final Rule and Correction Notice Tables. 2015. Available at: https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/FY2016-IPPS-Final-Rule-Home-Page-Items/FY2016-IPPS-Final-Rule-Tables.html. Accessed October 4, 2015.
- ↵The Coding Institute. Available at: http://www.codinginstitute.com/supercoder. Accessed October 5, 2015.
- ↵DMD America. AnalysourceBiweekly.com. Available at: https://www.analysource.com/pricing.html. Accessed October 5, 2015.
- Solomon N.A.,
- Glick H.A.,
- Russo C.J.,
- et al.
- Meschia J.F.,
- Bushnell C.,
- Boden-Albala B.,
- et al.
- Lee V.W.,
- Yan B.,
- Chow I.,
- et al.
- Singh S.M.,
- Micieli A.,
- Wijeysundera H.C.