Editorial
Improving the Management of Warfarin May Be Easier
Than We Think
Adam J. Rose, MD, MSc, FACP
P
Downloaded from http://circ.ahajournals.org/ by guest on June 18, 2017
The authors deftly used a feature of their dataset to
demonstrate that percent time in therapeutic range (TTR)
really is in the causal pathway to adverse events. Because
their dataset was drawn from a randomized trial of dabigatran
versus warfarin, they were able to show that, whereas sites
with more algorithm-consistent warfarin dosing had lower
rates of adverse events among patients receiving warfarin,
they did not have lower rates of adverse events among
patients receiving dabigatran. Some have expressed doubts
about whether TTR really is in the causal pathway to
outcomes,5 suggesting that instead, sites with higher TTR
also might be delivering high-quality care in other ways. If
this were true, then quality improvement efforts aimed at
increasing TTR might not achieve the desired benefits in
terms of preventing adverse events. The finding by Van Spall
and colleagues4 serves as a strong refutation for this line of
reasoning, because if high-TTR sites were truly delivering
other interventions responsible for preventing adverse events,
we would have seen a similar reduction in adverse events
among patients receiving dabigatran. The present study therefore serves as a strong endorsement of efforts to improve TTR
at the site level and thereby prevent adverse events.
The algorithm studied by Van Spall and colleagues4 was
similar to others that are in widespread use. The algorithm
suggested no dose change for an in-range international
normalized ratio (INR) value, a 10% dose change when the
INR was somewhat out of range (1.51–1.99 or 3.01– 4.00),
and a 15% dose change for greater deviations from the target
range. The authors note that, although they distributed this
algorithm, they cannot know to what extent the sites of care
actually used it—in essence, their study examined the difference between care that would have been algorithmconcordant, as the authors put it, “whether intentionally or
not.” In a study by Kim et al,6 a very similar (but not
completely identical) dosing algorithm was shown to improve
TTR. The Kim study complements the current study in that it
shows that the introduction of a standardized dosing algorithm improved TTR considerably using a before–after design.6 Although Van Spall and colleagues were able to show
that algorithm-concordant care was important, they were
unable to examine whether the introduction of the algorithm
had actually increased the likelihood that such care would be
delivered. Also, in the Kim study, algorithm-consistent dosing was highly correlated with patient-level TTR, which
complements the site-level findings of the current study.
Taken together, the 2 studies suggest that the algorithm is
both valid (in the sense of predicting important outcomes)4
and feasible (in the sense that it can successfully be introduced and change clinical practice).6
erformance variation is easy to find—all one needs to do
is to look for it. We have long known that some nations
achieve better control of hypertension than others,1 some
hospitals have shorter door-to-balloon times than others,2 and
some cardiac surgeons have better risk-adjusted mortality
after coronary artery bypass graft surgery than others.3 In
fact, it is difficult to recall an instance when performance was
found not to vary. Given that performance variation is
ubiquitous, it is no longer shocking to find it; the more
interesting question is why performance varies so much. The
answer to this question would likely be a key step along the
pathway to improving performance.
Article see p 2309
It is rather uncommon to find a single answer to the
question of why performance variation exists. However, in
the current issue of Circulation, Van Spall and colleagues4
have found an unusually straightforward explanation for
performance variation, at least in the context of the management of warfarin. The authors found that site-level adherence
to a relatively simple algorithm regarding when to change the
dose of warfarin and when not to change the dose predicted
fully 87% of between-center variance. Adding patient-level
clinical variables (ie, risk-adjustment), center-level variables,
and country-level variables only increased the amount of
explained variation to 89%. In short, management of warfarin
doses appeared to be almost deterministic regarding the
anticoagulation control that was achieved. Remarkably, the
authors also found that greater adherence to the algorithm
also predicted a reduced rate of the combined primary end
point of stroke, major hemorrhage, or death at the site level.
For each 10% increase in algorithm-consistent dosing at the
center level, the annual rate of the combined end point was
8% lower, even after adjusting for a host of patient-level
predictors. The algorithm has additional attractive features as
well, not least of which is that it does not require a computer
or proprietary software to use. In fact, it would be equally
suitable to use in developing countries.
The opinions expressed in this article are not necessarily those of the
editors or of the American Heart Association.
From the Center for Health Quality, Outcomes, and Economic Research, Bedford VA Medical Center, Bedford, MA; and the Department
of Medicine, Section of General Internal Medicine, Boston University
School of Medicine, Boston, MA.
Correspondence to Adam J. Rose, MD, MSc, FACP, Center for Health
Quality, Outcomes, and Economic Research, 200 Springs Road, Mail
Stop 152, Bedford, MA 01730. E-mail [email protected]
(Circulation. 2012;126:2277-2279.)
© 2012 American Heart Association, Inc.
Circulation is available at http://circ.ahajournals.org
DOI: 10.1161/CIRCULATIONAHA.112.141887
2277
2278
Table.
Circulation
November 6, 2012
Some Key Challenges to Improving Anticoagulation Control
Topic
Issue
Challenge or Knowledge Gap
Warfarin dose management
Limited use of computerized dose support or
paper-based algorithms, both of which are shown
to improve control
Promoting wider adoption of computerized dose
support or paper-based algorithms
Loss to follow-up
Lack of functioning systems to track patients and
detect loss to follow-up
Need to develop such systems, particularly for
sites without proprietary anticoagulation
management software
Use of nonstandard target INR ranges
Lower target INR ranges such as 1.5 to 2.0 remain
in use, despite having been shown to produce
harms without additional benefit
Finding effective ways to promote the use of
standard rather than nonstandard target ranges
Timely follow-up after deranged INR values
Prompt follow-up after high (⬎4) or low (ⱕ1.5) INR
improves control
Clinicians may resist changing practice; patients
may resist frequent follow-up
Patient education
Patient education may lack standardization and may
have limited effectiveness.
Developing new and innovative ways to educate
and re-educate patients about warfarin therapy
Patient adherence to therapy
Limited adherence to pill-taking, dietary consistency,
on-time follow-up, and other matters
Need to identify and address various patient-level
barriers to improved adherence
Each clinic needs strong leadership to ensure continuous
quality improvement
Individuals may not feel empowered to identify and
implement strategies to improve outcomes
Anticoagulation therapy is clearly important enough to
deserve a program of performance measurement
Performance measures have been developed, but
still need to be used more widely, and may need
to be amended over time in response to feedback
Anticoagulation clinic leadership
Downloaded from http://circ.ahajournals.org/ by guest on June 18, 2017
Performance measurement
INR indicates international normalized ratio.
Any algorithm for warfarin dose management, whether
paper-based or computer-based, needs to be introduced with
a caveat, namely that the clinician must always have the
power to override the algorithm. Any decision about warfarin
dose changes must occur in the context of the visit, and some
information that is divulged by the patient (for example,
recent dietary intake) may well prompt a decision that is
discordant with the algorithm.7 However, both the Van Spall
and Kim studies strongly suggest that clinicians managing
warfarin should ask themselves whether they really have a
compelling reason to deviate from the algorithm—and then
ask again for good measure.
The present study also adds to a growing discussion about
how to manage warfarin doses when the INR is only slightly
out of range. In the past, several studies have suggested that
dose changes are not necessary for mildly deranged INR, and
in fact may begin a cycle of overcorrection and rebound that
may worsen control.8 –9 Based on this limited evidence, the
latest consensus guidelines for managing warfarin suggest not
changing the dose when the INR deviates by ⱕ0.5 from the
target range, and instead merely rechecking in 1 to 2 weeks.10
Notably, a deviation of up to 0.5 is an even wider tolerance
range than would be suggested by either of the previous
studies.8 –9 With this issue in mind, Van Spall and colleagues
examined whether their findings would also be true when the
tolerance range was widened from 2.0 to 3.0 to 1.9 to 3.1, 1.8
to 3.2, 1.7 to 3.3, and 1.6 to 3.4. They found that TTR was
higher with increased adherence to any of these ranges, but
was best with the strict construction of the target range (ie,
2.0 –3.0). In addition, only the strict construction predicted
the primary combined end point in a statistically significant
fashion. Thus, the findings of the present study cast at least
some doubt on the idea of a widened tolerance range, within
which slightly deranged INR values do not prompt a dose
change. Ideally, a well-designed prospective study could
address this issue more definitively. For now, although some
clinicians may still opt to have a tolerance range, it may be
prudent not to extend it beyond approximately 1.8 to 3.2,8 –9
as opposed to the much wider range of 1.5 to 3.5 recommended by the most recent guidelines.10
One noteworthy limitation of the present study stems from
its origins in a clinical trial.11 The patients represented in this
study may have been unusually adherent to their medication,
and gaps in INR monitoring were likely minimal. In a
real-world setting, dose management may not predict 87% of
site variation, and other issues, like gaps in INR monitoring,
also may play an important role. Our group has shown that
nonstandard target ranges, gaps in INR monitoring, and
failure to recheck INR promptly after an out-of-range value
are also strong predictors of site-level TTR.12–14 Together,
these 3 measures predicted 48% of TTR variation among sites
of care in the Veterans Health Administration (our unpublished finding). It is likely, therefore, that although adherence
to the algorithm examined by Van Spall and colleagues may
be the single most important determinant of site-level TTR, it
is not the only performance measure worth addressing. In
fact, improving TTR is likely to require a multifactorial
approach, as there are many related issues that contribute to
TTR (Table).
Another limitation, mentioned above, is that the present
study was not an attempt to actually implement the algorithm
under study. Kim and colleagues successfully implemented a
similar algorithm,6 but they did so at a single site of care,
which may not be representative of other sites. An important
next step would be a prospective test of this algorithm in a
real-world setting, preferably across an integrated health
system. Such an effort would require great attention to the
principles of implementation science to promote successful
adoption of the algorithm.15 In fact, we have recently begun
such a study in the New England region of the Veterans
Rose
Downloaded from http://circ.ahajournals.org/ by guest on June 18, 2017
Health Administration. One challenge for our effort, and for
any effort occurring in the real world, is that automated data
may not reliably identify dose changes in warfarin, which is
usually prescribed as use as directed. A need to rely on chart
review to track adherence to the algorithm may hamper
efforts to measure its uptake.
In summary, Van Spall and colleagues4 have added several
important new pieces of information to the literature. First, in
combination with other studies,6 the present study convincingly demonstrates that managing warfarin according to a
standard algorithm can improve patient outcomes. Second, it
should lay to rest any lingering doubts about whether TTR is
truly in the causal pathway to adverse events. Finally, it
suggests that changing the dose of warfarin for any out-ofrange INR may in fact be best, although other studies have
found differently,8 –9 and further evidence is needed. The
study by Van Spall and colleagues certainly provides a
powerful argument for greater adoption of any sort of system
to promote standardized dose management for warfarin—
either computer support when it is available, or paper-based
algorithms if not. Improving outcomes for patients receiving
warfarin will not be easy, but it may be easier than we think.
Disclosures
Dr Rose is supported by a career development award from the
Veterans Health Administration, Health Services Research and
Development Service (CDA-2-08-017). The opinions expressed are
those of the author and do not necessarily represent the official views
or policies of the Veterans Health Administration.
References
1. Wolf-Maier K, Cooper RS, Banegas JR, Giampaoli S, Hense HW, Joffres
M, Kastarinen M, Poulter N, Primatesta P, Rodriguez-Artalejo F,
Stegmayr B, Thamm M, Tuomilehto J, Vanuzzo D, Vescio F. Hypertension prevalence and blood pressure levels in 6 European countries,
Canada, and the United States. JAMA. 2003;289:2363–2369.
2. Bradley EH, Herrin J, Wang Y, McNamara RL, Radford MJ, Magid DJ,
Canto JG, Blaney M, Krumholz HM. Door-to-drug and door-to-balloon
times: where can we improve? Time to reperfusion therapy in patients
with ST-segment elevation myocardial infarction (STEMI). Am Heart J.
2006;151:1281–1287.
3. Jha AK, Epstein AM. The predictive accuracy of the New York State
coronary artery bypass surgery report-card system. Health Aff
(Millwood). 2006;25:844 – 855.
Warfarin Dosing by Algorithm
2279
4. Van Spall HGC, Wallentin L, Yusuf S, Eikelboom JW, Nieuwlaat R,
Yang S, Kabali C, Reilly PA, Ezekowitz MD, Connolly SJ. Variation in
warfarin dose adjustment practice is responsible for differences in the
quality of anticoagulation control between centers and countries: an
analysis of patients receiving warfarin in the Randomized Evaluation of
Long-Term Anticoagulation Therapy (RE-LY) trial. Circulation. 2012;
126:2309 –2316.
5. van Walraven C, Oake N, Coyle D, Taljaard M, Forster AJ. Changes in
surrogate outcomes can be translated into clinical outcomes using a
Monte Carlo model. J Clin Epidemiol. 2009;62:1306 –1315.
6. Kim YK, Nieuwlaat R, Connolly SJ, Schulman S, Meijer K, Raju N,
Kaatz S, Eikelboom JW. Effect of a simple two-step warfarin dosing
algorithm on anticoagulant control as measured by time in therapeutic
range: a pilot study. J Thromb Haemost. 2010;8:101–106.
7. Ansell J, Hirsh J, Hylek E, Jacobson A, Crowther M, Palareti G. Pharmacology and management of the vitamin K antagonists: American
College of Chest Physicians Evidence-Based Clinical Practice Guidelines
(8th Edition). Chest. 2008;133:160S–198S.
8. Banet GA, Waterman AD, Milligan PE, Gatchel SK, Gage BF. Warfarin
dose reduction vs watchful waiting for mild elevations in the international
normalized ratio. Chest. 2003;123:499 –503.
9. Rose AJ, Ozonoff A, Berlowitz DR, Henault LE, Hylek EM. Warfarin
dose management affects INR control. J Thromb Haemost. 2009;7:
94 –101.
10. Holbrook A, Schulman S, Witt DM, Vandvik PO, Fish J, Kovacs MJ,
Svensson PJ, Veenstra DL, Crowther M, Guyatt GH. Evidence-based management of anticoagulant therapy: antithrombotic therapy and prevention of
thrombosis, 9th ed: American College of Chest Physicians Evidence-Based
Clinical Practice Guidelines. Chest. 2012;141:e152S–e184S.
11. Connolly SJ, Ezekowitz MD, Yusuf S, Eikelboom J, Oldgren J, Parekh A,
Pogue J, Reilly PA, Themeles E, Varrone J, Wang S, Alings M, Xavier
D, Zhu J, Diaz R, Lewis BS, Darius H, Diener HC, Joyner CD, Wallentin
L. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl
J Med. 2009;361:1139 –1151.
12. Rose AJ, Berlowitz DR, Miller DR, Hylek EM, Ozonoff A, Zhao S,
Reisman JI, Ash AS. INR targets and site-level anticoagulation control:
results from the Veterans AffaiRs Study to Improve Anticoagulation
(VARIA). J Thromb Haemost. 2012;10:590 –595.
13. Rose AJ, Hylek EM, Berlowitz DR, Ash AS, Reisman JI, Ozonoff A.
Prompt repeat testing after out-of-range INR values: a quality indicator
for anticoagulation care. Circ Cardiovasc Qual Outcomes. 2011;4:
276 –282.
14. Rose AJ, Miller DR, Ozonoff A, Berlowitz DR, Ash AS, Zhao S,
Reisman JI, Hylek EM. Gaps in monitoring during oral anticoagulation:
insights into care transitions, monitoring barriers, and medication nonadherence. Chest. In press.
15. Stetler CB, Damschroder LJ, Helfrich CD, Hagedorn HJ. A guide for
applying a revised version of the PARIHS framework for implementation.
Implement Sci. 2011;6:99.
KEY WORDS: Editorials 䡲 anticoagulants 䡲 medication therapy management
䡲 outcome assessment (health care) 䡲 quality of health care
Improving the Management of Warfarin May Be Easier Than We Think
Adam J. Rose
Circulation. 2012;126:2277-2279; originally published online October 1, 2012;
doi: 10.1161/CIRCULATIONAHA.112.141887
Downloaded from http://circ.ahajournals.org/ by guest on June 18, 2017
Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 2012 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://circ.ahajournals.org/content/126/19/2277
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial
Office. Once the online version of the published article for which permission is being requested is located,
click Request Permissions in the middle column of the Web page under Services. Further information about
this process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Circulation is online at:
http://circ.ahajournals.org//subscriptions/