Venous thromboembolism (VTE), including deep vein thrombosis and pulmonary embolism,
is common, killing up to 100,000 Americans annually.[1] The overall incidence of VTE has increased over the past decade, particularly in
those with advanced age.[2] Diagnosis and management of VTE also generate large medical costs, as high as US$15,000
for a single patient with VTE in the United States and up to US$33,000 when factoring
in subsequent care and sequelae.[3]
Though evidence-based guidelines for diagnosis and treatment of VTE exist, mismanagement
is common and can lead to serious negative consequences for patients due to either
inadequate or excessive treatment.[4] In a large observational study, nearly half of new VTE cases are not managed based
on evidence-based recommendations.[5] Thrombophilia testing is often ordered inappropriately, leading not only to higher
costs but also potential harm through unnecessary work-up, treatment, and psychological
stress to the patient.[6] Inappropriate VTE treatment lengths are also common, likely contributing not only
to increased rates of recurrent thrombosis but also to bleeding complications.[7] Even with this information, describing the true costs of VTE management, both monetary
and otherwise, is difficult.
Several factors may be responsible for this discordance between data and practice,
including lack of provider awareness of guidelines (cumbersome guideline design),
infrequent guideline updates to include the latest high-quality data, low-quality
data leading to varying study interpretations, and conflicts between appropriate management
and patient wishes. Thrombosis management differs from that of hematological malignancies,
as it frequently involves the participation of general practitioners as well as providers
in most other fields of medicine, often without direct involvement of a hematologist.
Management of hematological malignancies also often benefits from user-friendly and
routinely updated evidence-based guidelines, lacking for nonmalignant hematology.[8] A noteworthy measure aimed at reducing some of the more frequent areas of VTE mismanagement
was the creation of the American Society of Hematology's (ASH) “Choosing Wisely” guidelines.
These include at least five statements pertinent to diagnosis and management of VTE,
including recommendations against routine thrombophilia testing, inferior vena cava
filter placement, and aggressive use of prothrombin complex concentrates for anticoagulation
reversal, among others.[9] While these recommendations, as well as those found in major society guidelines,
are high-yielding and backed by high-quality evidence, locating and integrating these
recommendations into daily clinical practice remains challenging.
Simplified VTE management guidelines could potentially reduce the burden of this disease
on our health-care system. We therefore attempted to address this problem by designing
an evidence-based thrombosis and hemostasis treatment algorithm, modeled after the
National Comprehensive Cancer Network guidelines for malignancies and designed to
be user-friendly and easily accessible.[8] We then prospectively tested this tool's effect on provider and trainee clinical
decision-making for VTE, hypothesizing that our tool would lead to improved VTE management
decisions.
Using the latest thrombosis guidelines from the American College of Chest Physicians
and other major hematological societies, and supplemented by additional high-level
data in the fields of thrombosis and hemostasis, we created an electronic diagnostic
and treatment algorithm tool for VTE (www.anticoag.net). We then designed a survey consisting of several demographic questions as well as
11 clinical questions describing commonly and uncommonly encountered clinical scenarios
pertaining to thrombosis and hemostasis, with answers in multiple-choice form. The
correct answers were formulated around the most strongly evidence-based data used
to design our guidelines. We implemented our algorithms and survey in a prospective,
observational, randomized, single-blinded study of health-care providers and medical
students from Oregon Health & Science University during July 2016. Study recruitment
was arbitrarily capped after this 1-month period. Eligible participants included attending
physicians, fellows, medical residents, nurse practitioners (NPs), and physician assistants
(PAs) in the fields of internal medicine, family medicine, hematology, and oncology,
as well as medical students of all levels of training. Providers working in both inpatient
and outpatient settings were included. Participants were recruited through e-mail;
at the time of enrollment, participants were randomized to either have access to our
algorithmic tool (in digital PDF form) or not; those without access were encouraged
to use any other resource they might typically use in their respective clinical settings.
The study was single-blind in nature; study coordinators were blinded to participant
randomization. Participants were asked to answer the clinical scenarios as well as
rate their confidence in each answer. Postassessment feedback was solicited. The location
and manner in which participants completed the exercise were not specified. The 11
clinical questions were scored, and an unpaired t-test was performed to determine if any significant difference existed in scores between
participants with and without the use of our algorithmic tool.
During the study period, 101 individuals participated: 48 medical students, 23 medicine
residents, 17 attending physicians, 9 fellows, and 4 NPs/PAs. Across all participants,
those with access to the algorithms on average answered 3.84 (34%) more questions
correctly (95% confidence interval [CI]: 3.08–4.60; p < 0.0001) ([Fig. 1], [Table 1]). The significantly increased number of correct answers was consistent across all
subgroups. Participant-reported confidence in their answers was also significantly
higher in those who were randomized to use our algorithm (mean difference 0.95 on
a 5-point confidence scale; 95% CI: 0.50–1.39; p < 0.0001).
Table 1
A summary of survey responses
|
All participants
|
Nonmedical students
|
Medical students
|
Residents
|
Fellows
|
Residents + fellows
|
Attending physicians
|
NPs/PAs
|
|
With (N = 52)
|
Without (N = 49)
|
With (N = 27)
|
Without (N = 26)
|
With (N = 25)
|
Without (N = 23)
|
With (N = 13)
|
Without (N = 10)
|
With (N = 3)
|
Without (N = 6)
|
With (N = 16)
|
Without (N = 16)
|
With (N = 8)
|
Without (N = 9)
|
With (N = 3)
|
Without (N = 1)
|
|
Mean score (out of 11)
|
7.08
|
3.24
|
7.07
|
4.04
|
7.08
|
2.35
|
7.46
|
3.5
|
7.33
|
3.83
|
7.44
|
3.63
|
7.25
|
4.67
|
4.67
|
5
|
|
Standard deviation
|
2.20
|
1.56
|
2.16
|
1.37
|
2.29
|
1.27
|
1.76
|
1.43
|
2.08
|
1.47
|
1.75
|
1.41
|
2.49
|
1.12
|
2.52
|
–
|
|
Mean difference of scores
|
3.84
|
3.03
|
4.73
|
3.96
|
3.5
|
3.81
|
2.58
|
–0.33
|
|
95% confidence interval
|
3.08–4.60
|
2.03–4.03
|
3.64–5.82
|
2.54–5.38
|
0.71–6.29
|
2.66–4.96
|
0.63–4.53
|
–
|
|
p-Value
|
<0.0001
|
<0.0001
|
<0.0001
|
<0.0001
|
0.021
|
<0.0001
|
0.0131
|
–
|
Abbreviations: NPs, nurse practitioners; PAs, physician assistants.
Fig. 1 Survey performance using thrombosis guidelines versus standard resources.
In conclusion, this prospective, observational, randomized, single-blinded study demonstrates
that an evidence-based algorithmic tool significantly improved clinical decision-making
abilities and confidence of all medical providers in the areas of thrombosis and hemostasis.
Specific elements of our tool that likely contribute to its positive effect include
easy and rapid access, a commonly used file format (PDF), a familiar, visually appealing,
and uncluttered presentation style, and clear citations and links to primary data.
Similar tools in the form of online algorithms and smartphone apps are currently available
from the American Society of Hematology (ASH “Pocket Guides”), Thrombosis Canada,
and the United Kingdom's Guy's and St. Thomas NHS Foundation Trust, though we feel
that the narrow breadth of topics and lack of robust supporting data in these resources
limit their clinical utility.[10]
[11]
[12] To the best of our knowledge, our guidelines are the first to demonstrate a potential
for improved clinical outcomes. Our guidelines have since been greatly expanded to
include additional primary data and new algorithms for several other areas of thrombosis
and hemostasis. These guidelines are now hosted on a custom domain (www.anticoag.net). Anecdotally, the algorithms on this Web site have seen widespread adoption among
practitioners across many subspecialties at our institution. A summary of key points
addressed by this correspondence is listed in [Table 2].
Table 2
A summary of key points addressed by this correspondence
|
VTE is often mismanaged despite evidence-based guidelines
|
|
We implemented a streamlined algorithm for VTE treatment in a prospective, single-center
study
|
|
Our algorithm led to statistically significant improvement in VTE management in all
training levels
|
|
These algorithms will be expanded to include additional thrombosis and hemostasis
topics
|
Abbreviation: VTE, venous thromboembolism.
We acknowledge several weaknesses of our study, including single-institution design,
small sample and survey size, limited breadth of medical and surgical specialties
included, and small number of clinical vignettes in the survey. The sample size was
limited due to difficulties in study recruitment. In the interest of simplicity, our
guidelines also omit some of the more nuanced data that might be required to manage
complex cases. Our algorithms, however, are designed primarily with nonhematology
subspecialists, internists, and other primary care providers in mind. We aim to expand
the content of our guidelines and test them prospectively in other medical centers
and in a larger variety of controlled clinical settings to further explore its impact
on medical decision-making. This research would provide a valuable feedback to help
modify the tool to include the most relevant information. Should this model prove
effective, it could be appropriated for other areas of nonmalignant hematology.
