Assessing Patient Management and Outcomes in Atrial Fibrillation: Does your health
insurance plan know more than your doctor?
Quick View
Credits : Sanjeev Saksena MD FHRS and April Slee, MS
- Electrophysiology Research Foundation, Warren, New Jersey, and
- Department of Medicine Robert Wood Johnson
Medical School, New Brunswick, New Jersey,
- Axio Research Corporation, Seattle, Washington
Corresponding Address : Sanjeev Saksena, MD, Medical Director, Electrophysiology Research Foundation, 161 Washington Valley Rd, Suite 201, Warren, New Jersey, 07059, USA.
Short Title – Outcomes of Atrial Fibrillation
Assessing the landscape of any major
public health challenge and the effectiveness of existing health care practices
is a difficult proposition in any circumstance for health care planners and
providers. To do so with relatively current health care data has not been a
feasible reality. Too often health care planners have been relegated to use of
venerable but dated clinical information. Equally often, clinical trial data
collected for a purpose other than outcomes research have been extrapolated
well beyond their original intent. The field of atrial fibrillation is no
exception. The durable and well-reported Framingham study data have provided
modern day framework for a natural history base of the disease over many
decades [1, 2].
More recent analyses have shown worldwide similarity in patterns and increasing
prevalence [3, 4]. The cascade of
anticoagulant trials in the nineties with their metanalyses and methodology
also provided outcome endpoints that have been widely used as a benchmark [5, 6]. More recently, NIH clinical trials such
as the AFFIRM trial have provided some outcomes analyses [7].
Yet these tools provide information that may have been captured some time ago
and significantly lag current medical experiences and practice.
With the establishment of the Medicare
program has come the progressive development of its prospective reimbursement
scheme over the last two-decade period. This program has increasingly matured,
acquired serious complexity with a substantial underlying methodology. In a
digital age, the Medicare database has become a substantial alternate resource
for researchers seeking to examine modern health care in the United States of America. These data now provide source information for a vast range of
medical subjects and specialties. Medicare data is often made available for
analysis within a preceding three to five year period, and in some instances
even within a two-year capture window. It has identified atrial fibrillation as
one of the top diagnoses responsible for hospitalization in the USA [8]. As far aback as 1999, a total of 1,765,304 hospitalizations
(137.1 per 1,000 Medicare enrollees) were reported among persons with AF in the
Medicare population [8].
The incidence and prevalence of atrial
fibrillation have been a staple subject for several large epidemiologic reports
and widely analyzed [3, 4]. The risks of
adverse cardiovascular outcomes including mortality, stroke and impaired
quality of life have been well documented from these and clinical trial reports
[1 - 4, 7]. These results
have been used to justify a range of medical therapies from antiarrhythmic
drugs, rate control drugs, anticoagulation and non pharmacologic approaches
such as ablation and device therapy. Clinical reports using both observational
and clinical trial data have emerged with increasing swiftness. Clinical
practice guidelines have evolved from both sources of information and enshrine
best medical practice with currently available information [9].
These guidelines have been promulgated to influence physician management of
atrial fibrillation along best medical practice. The impact of these guidelines
has not been systematically assessed and real world physician behaviour is
still poorly defined. A smattering of market research reports, often based on
physician prescribing practices, provide limited insight. Figure
1 shows prescription practices of antiarrhythmic drugs from such data and
suggests that rate control is the dominant practice strategy in atrial
fibrillation. Rate control mandates antithrombotic therapy in the vast majority
of patients. However, obtaining widely applicable data from a large health care
segment of the population that is current and broadly applicable or reliably
analyzable in its component groups remains a major health care information
challenge.
Figure 1:Prescriptions of antiarrhythmic drugs in atrial fibrillation patients. Note that rate control therapy dominates the market as defined by this parameter.
Source: December 2007 Verispan PDDA.Drug Use by USC.
|
One inevitable limitation of the Medicare
data is the age segment. While many chronic diseases extend into the population
of patients older than 65 years, the earlier stages of disease, where
intervention may be most effective, may not be captured. Atrial fibrillation is
dubbed as a “disease of the elderly”, but may often develop in patients who are
in their 50s or 60s – as many as 15 years before patients become Medicare
eligible. Early management of AF can influence later outcomes, but data from
this period of the disease are remarkably sparse. In recent years, private
health insurance coverage provided by employers or by the self-employed has
become fairly standard, despite increasing concerns of absent coverage for many
Americans in their working years. These plans now collect and review many
aspects of health care provided to their clients including demographic
profiles, disease patterns, health care resource utilization, pharmacotherapy,
compliance and provider behaviour. Mining this database can provide
researchers with another look at health care in a different segment of
society.
In a recent report, Walker and Bennett
undertake an ambitious analysis of epidemiologic outcome aspects of atrial
fibrillation in the United States using a proprietary health insurance plan
database [10]. Patient records were selected using ICD 9
diagnosis codes and pharmacy claim data. Remarkably, available data was very
broad based; available information included physician provider claims and
hospitalizations in covered patients. Outpatient laboratory service provider
care was analyzed in 30% of the population using plan-contracted laboratories.
Patients included a broad range of demographics including Medicare age groups.
Unusual strengths of this database included drug use data even when the drug
was priced below the copay amount. Limitations of the database include
veterans’ benefits supplementing health care usage in this segment, and
inpatient laboratory data. Interestingly, the database analysis was exempt from
institutional review board requirements as study data were anonymized according
the Health Insurance
Portability and Accountability Act standards.
Walker and Bennett analyzed a population
of plan members aged 40 years or older who had coverage for both medical and
pharmacy services over a six year period from 1999-2005. Both non-valvular
atrial flutter and atrial fibrillation were analyzed and a diagnosis of mitral
valve disease was an exclusion criterion. The base analysis was a descriptive
analysis of medical practice in the broad population of 116,969 patients making
this one of the largest data sets analyzed. A nested subgroup analysis of
laboratory and patient data was performed in patients using contracted
laboratory services. Principal diagnoses were used for endpoints of stroke,
cerebral hemorrhage and thromboembolism, which may have underestimated event
rates relegated to secondary diagnosis status. The descriptive demographic data
is unique in capturing incidence and prevalence data in the 40-59 year age group
of over 30,000 insured AF patients. Approximately, 36% of patients were in the
> 75 year age group. These two extreme demographic age groups provided very
large patient numbers for analysis. Cardiac dysrhythmia was the most common
diagnosis for management in the overall group and prevalent cases, suggesting
that coding and clinical presentation were well matched. This finding is not
always seen in cardiac arrhythmias, for example in patients with ventricular
tachycardia or ventricular fibrillation where it may remain a secondary
diagnosis. Better coding procedures at the provider level may be responsible
for this improvement. While confirming hypertension as the major disease
diagnosis in new incident cases in this large swath of age groups, respiratory disease
and dyslipidemia emerged as next in line, in stark contrast to clinical trial
data such as AFFIRM. Reasons for this difference may include a change in a
point of engagement of the health care system, perhaps at an earlier high risk
population stage.
Few other clinical data are reported that
describe cardiovascular status in particular. Risk factors and prevalence data
for stroke risk were more detailed with expected factors of advanced age and
hypertension having prevalence rates of >30%. Coronary disease, diabetes and
heart failure followed with substantial prevalence rates of 14 to 21%. Most
patients had one risk factor, while 34% had more than one risk factor. Risk
factors for bleeding were quite uncommon with only 3% having renal failure,
giving a truly different perspective from anticoagulant trials in the elderly [7]. Stroke rates exceeded bleeding complications by a factor of
10. Stroke incidence in the population was 1.3 per 100 pt years and
gastrointestinal bleeding incidence was 0.8% with intracranial bleed averaging
0.3%. Thromboembolic events including pulmonary embolism had incidence ranging
from 0.18 to 0.2%. There was an average of one AF hospitalization in the
follow-up period.
The major focus of the report rotates on
anticoagulant drug use patterns and their correlation with INR values in the
nested analysis. Actual rates of prescription of warfarin and antiplatelet
agents were lower than expected (45% and 6%, respectively). Of particular
interest beyond the rates of prescription
were the duration and variables involved in their
use. The average time to discontinuation was about 4 months for both agents.
The AFFIRM trial published at the turn of the millennium, has promoted
anticoagulation as a lifelong strategy when feasible. It would have been
interesting for the authors to have performed a time-dependent analysis on this
aspect. Importantly, women were less likely to receive anticoagulation and
common risk factors such as hypertension, diabetes and coronary disease were
only weak predictors (odds ratio 1.11 to 1.18) of its use. Poor management of
warfarin was seen in one-third of patients who spent >80% of their time
outside the therapeutic INR range and only 19% remained in the recommended
range virtually all of the time, underscoring the difficulties in warfarin
management and effective implementation. In current reports, warfarin usage has
plateaued between 50 – 60% [6]. Therapeutic INR rates
improved from 40-45% early on to around 60%. As expected, inability to comply
with warfarin follow-up or high risk bleeding situations often precluded
warfarin use, but surprisingly a history of a fall or intracranial vascular
malformation did notpreclude use. Not surprisingly, the data confirmed prior
studies showing a doubled risk of stroke and five fold risk of embolism with
sub therapeutic INR values and a doubled risk of bleeding with supratherapeutic
values [7].
The most unique aspects of this data set
are its size and its existence outside the framework of a clinical trial. The
size in relationship to other studies has been described previously. A
randomized clinical trial of this size with this level of detail would not
currently be feasible. Unlike other large retrospective studies, this analysis
was performed on very recent data, so inference is not affected by new
therapeutic agents and treatment strategies. Since these data emerge from a
non-clinical trial setting, the treatment patterns are regulated by clinical
judgment and patient adherence rather than a protocol. The absence of inclusion
criteria implies that this analysis contains a broader spectrum of patients
with AF that a clinical trial would contain – the sickest patients are often
underrepresented or absent from clinical trial data. These data could offer an
accurate reflection of real-world treatment patterns, and as in prior reports,
we note a marked divergence from clinical guidelines [9, 11, 12]. While lifelong use of warfarin is
recommended, the reality in this data set is that treatment is intermittent
rather than continuous, and the median time to discontinuation was just over 4
months. In the NABOR study and other analyses, only persistent/permanent
atrial fibrillation and age > 80 years influenced warfarin usage rather than
risk factors for stroke [13]. For the subset of patients
with INRs, only 19% spent most of their time within the therapeutic range. In a
recent review, clinical trials show rates of 60 – 65%, raising a potential
dichotomy with practice guidelines and trial data.
Despite the authors’ efforts to identify
a robust population for health care resource use analysis using exclusively
private insurance, there is no guarantee that all resource use, and especially
prescription drug use, were reported. This concern is a major limitation of
this analysis. The authors do not show enough information about the age
distribution to determine what percent of the cohort was eligible for Medicare.
It would be interesting to know how Medicare-eligible patients were using their
private insurance plans [14]. For patients enrolled in
Medicare, private insurance would have been the secondary payor on prescription
drugs. The claims may or may not have been received by the private insurance
company. Government employees who are veterans may have accessed care through
the VA system, and is not possible to know what percent of claims were filled
by payors other than private insurance. Walker et al found that 48% of patients
who should have been anticoagulated had no claims for warfarin, but this could
be an overestimate of untreated patients due to claims filed to other payors.
As the authors note, diagnosis information is difficult to obtain. It is
inevitable that some patients had conditions for which warfarin was
contraindicated. It would be interesting to know if any patients without
warfarin claims did have claims for other drugs contraindicated with warfarin.
The lab data presented is also probably not a random
selection of data from participants in the health plan, and systematic
differences between patients who did and did not have INR data could bias the
results. Only labs qualified to be in the network reported data, and facilities
serving certain demographic groups or employment status may have been more
likely to use a network lab than other sites. The authors were aware of this
problem and examined differences in patients who did and did not have INRs in
the database. Since INRs could have been conducted at a non-network lab, the
absence of INR data does not necessarily imply that INR estimations were not
performed.
The authors correctly point out many of
the strengths, weaknesses and appropriate interpretations of their data. While
retrospective, the data do modify our understanding of patient presentation in
engagement of outpatient health care systems and move us away for randomized
clinical trial settings that often start in the hospital. The large sample size
exceeded by only one other report, along with real world pharmacy usage data is
a useful addition to our information base [3]. As the authors
note, these data originated outside of a randomized clinical trial, and may
more accurately describe standard practice. As expected, the limitations of
outpatient diagnosis coding, which is less likely to be as robust as hospital
coding, and the index INR values as reflective of the whole experience in some
patients, drug compliance and out of system drug purchases remain issues in the
data set. Yet overall, the data do confirm the difficulties in implementing a
warfarin based anticoagulant strategy in the community with lack of penetration
of practice guidelines being impeded, most probably not by physician or patient
education, but by real-life single drug related limitations and interactions
and compliance with follow-up testing procedures.
All in all, this report is a relevant and
current look at a very broad AF population from an epidemiologic and health
care delivery viewpoint. Findings should stimulate health care planners to
focus to a greater extent on the point of engagement of the health care
delivery system, i.e. the outpatient entry point, and seek better techniques to
improve risk recognition by patients and warfarin usage. The recent
availability of home monitoring kits for warfarin, anticoagulation clinics and
dietary instruction, along with potential genetic typing for metabolic enzyme
status offer new options for improving best medical practice with existing
therapy [15]. It is increasingly likely that more insurance
plans will provide this type of preventative and proactive intervention to
reduce health care expenditures due to preventable complications. For
pharmaceutical drug development, this report co-authored by an epidemiologist
from this industry, clearly sets the stage for the need for new therapeutic
agents such as factor 10a inhibitors that are in the pipeline. However, the
uptake of new therapeutic paradigms is often beset with new challenges in
acceptance and implementation at the provider, patient and health care delivery
system levels. In the immediate future, adjustment to warfarin management is
likely to yield more immediate results for health care planners, physicians and
patients.
-
Wolf P, Abbott R, Kannel W. Atrial fibrillation as an independent risk factor for stroke: The Framingham Study. Stroke 1991: 22:983–988.
-
Benjamin EJ, Wolf PA, D'Agostino RB, Silbershatz H, Kannel WB, Levy D.: Impact of atrial fibrillation on the risk of death: the Framingham Heart Study. Circulation. 1998 : 98(10):946-52.
-
Dagres N, Nieuwlaat R, Vardas P, et al. Gender-related differences in presentation, treatment, and outcome of patients with atrial fibrillation in Europe. A report from the Euro Heart Survey on Atrial Fibrillation. J Am Coll Cardiol 2007: 49:572–577.
-
Go AS, Hylek EM, Phillips KA, Chang Y, Henault LE, Selby JV, Singer DE :Prevalence of diagnosed atrial fibrillation in adults: national implications for rhythm management and stroke prevention: the AnTicoagulation and Risk Factors in Atrial Fibrillation (ATRIA) Study. JAMA. 2001 : 285(18):2370-5
-
Connolly SJ, Eikelboom J, O’Donnell M, Pogue J and Yusuf S: Challenges of Establishing New Antithrombotic Therapies in Atrial Fibrillation. Circulation 2007: 116;449-455
-
Reynolds MW, Fahrbach K, Hauch O, Wygant G, Estok R, Celia C, Nalysnyk: Warfarin anticoagulation and outcomes in patients with atrial fibrillation. Chest 2004: 126: 1938-1945.
-
Wyse DG, Waldo AL, DiMarco JP, Domanski MJ, Rosenberg Y, Schron EB, Kellen JC, Greene HL, Mickel MC, Dalquist JE, Corley SD; Atrial Fibrillation Follow-up Investigation of Rhythm Management (AFFIRM) Investigators. A comparison of rate control and rhythm control in patients with atrial fibrillation. N Engl J Med. 2002;347(23):1825-33.
-
Centers for Disease Control and Prevention (CDC). Atrial fibrillation as a contributing cause of death and Medicare hospitalization--United States, 1999. MMWR Morb Mortal Wkly Rep. 2003 :;52(7):128, 130-1
-
Fuster V, Rydén L, Cannom D, et al. ACC/AHA/ESC 2006 guidelines for the management of patients with atrial fibrillation— executive summary: a report of the American College of Cardiology/American Heart Association Task Force and the European Society of Cardiology Committee for Practice Guidelines (Writing Committee to Revise the 2001 Guidelines for the Management of Patients With Atrial Fibrillation). J Am Coll Cardiol 2006: 48:854 – 906.
-
Walker AM, Bennett D Epidemiology and outcomes in patients with atrial fibrillation in the United States: Heart Rhythm 2008, Vol 5, 1365 - 1372
-
Stafford R, Singer D. Recent national patterns of warfarin use in atrial fibrillation. Circulation 1998;97:1231–1233.
-
Stafford R, Singer D. National patterns of warfarin use in atrial fibrillation. Arch Intern Med 1996;156:2537–2541.
-
Tapson VF, Hyers TM, Waldo AL,Ballard DJ, Becke RC, Caprini JA, Khetan R, Wittkowsk AK, Colgan KJ, Shillington AC, et al for the NABOR (National Anticoagulation Benchmark and Outcomes Report) Steering Committee Antithrombotic Therapy Practices in US Hospitals in an Era of Practice Guidelines. Arch. Int. Med. 2005: 165: 1458-1464
-
Evans-Molina C, Regan S, Henault LE, Hylek EM, Schwartz GR. The new Medicare Part D prescription drug benefit: an estimation of its effect on prescription drug costs in a Medicare population with atrial fibrillation. J Am Geriatr Soc. 2007 l;55(7):1038-43. .
-
Higashi M, Veenstra D, Kondo L, et al. Association between CYP2C9 genetic variants and anticoagulation-related outcomes during warfarin therapy. JAMA 2002: 287:1690 –1698.