4-19-11 annals of internal medicine - off-label use of factor viia

Off-Label Use of Recombinant Factor VIIa in U.S. Hospitals: Analysisof Hospital RecordsAaron C. Logan, MD, PhD; Veronica Yank, MD; and Randall S. Stafford, MD, PhD

Background: Recombinant factor VIIa (rFVIIa) is approved fortreatment of bleeding in patients who have hemophilia with inhib-itors but has been applied to a wide range of off-label indications.

Objective: To estimate patterns of off-label rFVIIa use in U.S.hospitals.

Design: Retrospective database analysis.

Setting: Data were extracted from the Premier Perspectives data-base (Premier, Charlotte, North Carolina), which contains dischargerecords from a sample of academic and nonacademic U.S. hospitals.

Patients: 12 644 hospitalizations for patients who received rFVIIaduring a hospital stay.

Measurements: Hospital diagnoses and patient dispositions from1 January 2000 to 31 December 2008. Statistical weights for eachhospital were used to provide national estimates of rFVIIa use.

Results: From 2000 to 2008, off-label use of rFVIIa in hospitalsincreased more than 140-fold, such that in 2008, 97% (95% CI,96% to 98%) of 18 311 in-hospital uses were off-label. In contrast,

in-hospital use for hemophilia increased less than 4-fold and ac-counted for 2.7% (CI, 1.9% to 3.5%) of use in 2008. Adult andpediatric cardiovascular surgery (29% [CI, 21% to 33%]), bodyand brain trauma (29% [CI, 19% to 38%]), and intracranial hem-orrhage (11% [CI, 7.7% to 14%]) were the most common indi-cations for rFVIIa use. Across all indications, in-hospital mortalitywas 27% (CI, 19% to 34%) and 43% (CI, 26% to 59%) ofpatients were discharged to home.

Limitation: Accuracy and completeness of the discharge diagnosesand patient medication records in the database sample cannot beverified.

Conclusion: Off-label use of rFVIIa in the hospital setting farexceeds use for approved indications. These patterns raise concernabout the application of rFVIIa to conditions for which strongsupporting evidence is lacking.

Primary Funding Source: Agency for Healthcare Research andQuality.

Ann Intern Med. 2011;154:516-522. www.annals.orgFor author affiliations, see end of text.

To be approved by the U.S. Food and Drug Adminis-tration (FDA), medications must clear many regulatory

hurdles and demonstrate both efficacy and a lack of exces-sive harms in clinical trials. After a medication has receivedapproval, there are no further limitations on its use, eitheron- or off-label. In many cases, this leads to medicationsbeing approved for narrowly defined indications, followedby substantial use in areas that have not been well-studied(1).

The FDA approved recombinant factor VIIa (rFVIIa)(Novo Nordisk, Bagsværd, Denmark) in 1999 for treat-ment of spontaneous or surgical bleeding episodes in pa-tients with hemophilia A or B who have inhibitors to factorVIII or factor IX. When first introduced, rFVIIa was usedpredominantly for these indications. After it became widelyavailable, however, rFVIIa was rapidly used in the treat-

ment or prophylaxis of bleeding in other conditions. Al-though rFVIIa activity is thought to be confined to sites ofendothelial injury (2, 3), the potential for thromboemboliccomplications with its use has been demonstrated inseveral trials and retrospective analyses (4 –9), raisingconcern about potential harms with widespread, off-label application. To estimate patterns of off-label, in-hospital use of rFVIIa, we performed a retrospectiveevaluation of data from the Premier Perspectives data-base (Premier, Charlotte, North Carolina) of U.S. hos-pitals. This representative sample was used to projectnational usage patterns from 1 January 2000 to 31 De-cember 2008.

METHODS

DesignWe purchased access to data from the Premier Per-

spectives database. The data set encompassed all hospital-izations during which rFVIIa was ordered for a patientfrom 1 January 2000 to 31 December 2008. We analyzedthis data set to categorize the discharge diagnoses and pa-tient outcome for each rFVIIa-associated hospitalization.

Hospital SampleThe Premier Perspectives database contains data from

615 nonfederal U.S. hospitals. Institutional participationin the database is voluntary. Data included are patient de-mographic characteristics; primary and secondary diagno-ses (coded by International Classification of Diseases,

See also:

PrintEditors’ Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517Editorial comment. . . . . . . . . . . . . . . . . . . . . . . . . . 566Related article. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529

Web-OnlyAppendix TablesAppendix FigureConversion of graphics into slides

Annals of Internal MedicineOriginal Research

516 19 April 2011 Annals of Internal Medicine Volume 154 • Number 8 www.annals.org

Ninth Revision [ICD-9]); length of hospital stay; medica-tions used; and disposition for each hospitalization. Weidentified 12 644 hospitalizations in which rFVIIa was ad-ministered to patients from 2000 to 2008. A total of286 100 ICD-9 diagnosis and procedure codes were re-ported for this cohort.

Categorizations of UseWe constructed a descending hierarchy of ICD-9

codes to categorize each case into mutually exclusive indi-cation categories (Appendix Table 1, available at www.annals.org). This hierarchy started with the most relevant,most reliable, and most specific clinical diagnoses, followedby successively less relevant, less reliable, or less specificdiagnoses. A case was assigned to a diagnostic category onthe basis of the ICD-9 code that placed it in the highestcategory in the hierarchy.

The higher-order diagnoses in this hierarchy were theFDA-approved indications of hemophilia A and B, fol-lowed by unapproved indications similar to hemophilia orapproved in other countries (for example, Glanzmannthrombasthenia). If these diagnoses were noted, the hospi-talization was classified into that category regardless ofwhether other potential indications were also noted. Allremaining hospitalizations were then categorized succes-sively for an array of off-label indications (Appendix Table1). Sensitivity analyses of this coding hierarchy had littleeffect on the main findings. The greatest change in thedistribution between indications was seen when traumawas moved lower in the hierarchy, which suggests thattrauma frequently occurs in the setting of other diseaseprocesses. We chose to prioritize trauma as the indicationfor treatment in these instances because it was likelytrauma and not the associated diagnoses that instigated theuse of rFVIIa.

The hematologic conditions not associated with he-mophilia were divided into 2 groups. We gave high prior-ity to primary clotting disorders that represent distinct andusually isolated defects in the clotting process. We gavelower priority to secondary clotting disorders that are morelikely to be the end product of other pathologic conditions,as with traumatic bleeding causing consumptive coagu-lopathy or the disruption of clotting produced by liverdisease.

ProjectionsStatistical weights associated with each hospital by

quarter were applied for projection of estimated rFVIIa useat the national level. Hospitals providing data to the data-base represent a convenience sample and may differ fromhospitals nationwide; however, the Premier Perspectivesdatabase is the largest hospital-based, service-level compar-ative database in the country. Statistical weighting allowsnational projections to be made under the assumption thathospitals of specific types in the database resemble hospitalsnationwide with these same characteristics. For each typeof hospital (defined by size, region, teaching status, and

ownership), weights are calculated as the inverse of thefraction of national admissions for this subgroup repre-sented in the database. These aggregate national estimatesare made by using data from the American Hospital Asso-ciation. The statistical weight assigned to each type of hos-pital by quarter allows derivation of national projections(see reference 10 for additional details). Although this ad-justed projection strategy is relatively rudimentary, it en-ables national projections that account for some, but notall, selection biases in the Premier Perspectives databasesample. The statistical weights vary from around 10 percase documented by the database in 2000 to around 5.5 in2008 as a function of the increasing number of hospitalsincluded in the database over time. For our sample of12 644 hospitalizations, we projected 73 747 (95% CI,51 247 to 96 245) cases of rFVIIa use nationwide duringthe study period.

Statistical AnalysisOur analyses identified annual trends in national in-

hospital rFVIIa use. To characterize patterns of use by in-dication, we produced a crosstabulation of indication cat-egory by year. We performed similar analyses for patientage, discharge disposition, and hospital characteristics. Theunit of analysis was any “case” of rFVIIa use—any hospi-talization during which rFVIIa was administered 1 or moretimes. We chose this case-based unit of analysis because itcaptures the medical decision-making component of careabout whether to use or not use rFVIIa for a given patient.Alternative methods of analysis by dosing were considered(for example, volume of rFVIIa dispensed by the phar-macy) but were determined to have substantial disadvan-tages: possible discrepancies between dispensed rFVIIa andthe amount actually administered, lack of consistent hos-pital coding of rFVIIa dispensing, and outlier cases. Anal-

Context

Although recombinant factor VIIa (rFVIIa) is approved onlyfor treatment of bleeding in patients with hemophilia,physicians use it to prevent or treat bleeding in othersituations.

Contribution

Investigators examined 12 644 discharge records of pa-tients who received rFVIIa at U.S. hospitals from 2000 to2008. During that time, off-label use of rFVIIa increasedmore than 140-fold, whereas use for hemophilia increasedless than 4-fold. In 2008, 97% of the in-hospital use ofrFVIIa was for off-label indications, including cardiovascu-lar surgery, trauma, and intracranial hemorrhage.

Implication

There is troubling, widespread off-label use of rFVIIa inU.S. hospitals.

—The Editors

Original ResearchOff-Label Use of Recombinant Factor VIIa

www.annals.org 19 April 2011 Annals of Internal Medicine Volume 154 • Number 8 517

yses were done by using SAS software, version 9.2 (SASInstitute, Cary, North Carolina). We calculated 95% CIsfor counts of hospital cases and proportions by using theSURVEYMEANS procedure in SAS, which accounted forselection of a national sample of hospitals.

Role of the Funding SourceThe Agency for Healthcare Research and Quality funded

the study. The funding source had no role in the study design,data collection and interpretation, writing of the report, ordecision to submit the manuscript for publication.

RESULTS

From 2000 to 2008, there were an estimated 73 747(CI, 51 247 to 96 245) hospital cases in the United Statesin which rFVIIa use was reported. During this time, off-label rFVIIa cases increased 143-fold (Figure 1) from 125(CI, 0 to 543) cases in 2000 to 17 813 (CI, 14 381 to22 240) cases in 2008. Off-label use in hospitalized pa-tients represented 96% (CI, 66% to 100%) of cases from2000 to 2008 and 97% (CI, 96% to 98%) of cases in2008. Use increased most rapidly for cardiovascular surgeryand trauma. There have been more modest increases fornontraumatic intracranial hemorrhage (ICH), liver disease,gastrointestinal bleeding, and other off-label indications(Table 1).

Figure 1. Estimated annual in-hospital cases of recombinantfactor VIIa use for hemophilia and off-label indications.

2000

Cas

es, n

Year

All other off-label use

Cardiovascular surgery

Trauma

ICH

Hemophilia

2001 2002 2003 2004 2005 2006 2007 20080

2000

4000

6000

10 000

12 000

14 000

16 000

18 000

20 000

8000

Cases signify the number of hospitalizations during which recombinantfactor VIIa was used. All cases for each year are depicted. The width ofeach segment represents the number of cases for each category, as indi-cated by differential shading. Hemophilia includes hemophilia A and B,and trauma includes body and brain trauma. ICH � intracranialhemorrhage.

Table 1. Indications for In-Hospital Use of rFVIIa

Clinical Indication Hospitalizations

2000–2008 2008 Only

Estimated Cases, n Amount of Total, % Estimated Cases, n Amount of Total, %

FDA-approved indicationsHemophilia A and B 3121 4.2 498 2.7

Off-label indicationsAdult cardiovascular surgery 12 086 16.4 4862 26.6Body trauma 11 689 15.9 3214 17.6Intracranial hemorrhage 7755 10.5 2005 11.0Brain trauma 7158 9.7 2033 11.0Primary clotting disorders 4104 5.6 792 4.3Nonvariceal GI bleeding 3882 5.3 713 3.9Secondary clotting disorders 3744 5.1 766 4.2Other liver disease 2451 3.3 539 2.9Pediatric cardiovascular surgery 1684 2.3 387 2.1Other vascular procedures 1515 2.0 381 2.1Aortic aneurysm 1216 1.6 306 1.7Pulmonary hemorrhage 1119 1.5 275 1.5Cancer or stem cell transplant 1094 1.5 138 0.8Variceal bleeding 897 1.2 235 1.3Liver biopsy or resection 867 1.2 122 0.7Neonatal indications 729 1.0 135 0.7Obstetric hemorrhage 672 0.9 220 1.2Neurosurgery 500 0.7 108 0.6Liver transplant 132 0.2 58 0.3Prostatectomy 120 0.2 0 0.0Other procedures 3867 5.2 407 2.2Other diagnoses 3345 4.5 117 0.6

Total treated with rFVIIa 73 747 100 18 311 100

FDA � U.S. Food and Drug Administration; GI � gastrointestinal; rFVIIa � recombinant factor VIIa.

Original Research Off-Label Use of Recombinant Factor VIIa


Hemophilia and Related ConditionsInitial use of rFVIIa was predominantly for FDA-

approved indications, including hemophilia A and B, aswell as related (but not FDA-approved) conditions, such asother clotting factor deficiencies, von Willebrand disease,and Glanzmann thrombasthenia (Table 1). Use for hemo-philia A and B has increased 3.8-fold since 2000 but seemsto have plateaued, whereas there has been a 7.4-fold in-crease in use for the other primary clotting disorders. These2 groups remained the most common cluster of indicationsfrom 2000 to 2003, but their proportion of use decreasedfrom 93% (CI, 71% to 100%) in 2000 to 7.1% (CI, 5.3%to 8.8%) in 2008. Hemophilia A and B alone account for4.2% (CI, 2.6% to 5.9%) of use overall and 2.7% (CI,1.9% to 3.5%) of use in 2008.

Cardiovascular SurgeryUse of rFVIIa in cardiovascular surgery was initially

observed in 2002 and by 2008 was 1 of the most frequentand rapidly increasing indications. Although use in pediat-ric cardiovascular surgery (mostly for repair of congenitalanomalies) has been modest (2.3% of use overall and 2.1%in 2008), use in adult cardiovascular surgery (largely foraortic valve, mitral valve, and coronary artery bypass graftprocedures) has rapidly increased, accounting for 16% (CI,12% to 21%) of cases overall and 27% (CI, 21% to 33%)of cases in 2008 (Table 1 and Appendix Figure, availableat www.annals.org).

TraumaSizable use of rFVIIa for traumatic bleeding episodes

began in 2002, and it remained the dominant indicationfor off-label use until it was equaled in prevalence by car-diac surgery (adult and pediatric) in 2008. Although usefor body and brain trauma leveled off in 2008 (Table 1and Appendix Figure), trauma other than brain trauma(Table 1) remains the second most common indication forrFVIIa use and accounts for 16% (CI, 8.3% to 23%) ofoverall use and 18% (CI, 12% to 23%) of use in 2008. Usein brain trauma, particularly traumatic subdural hema-toma, constitutes 9.7% (CI, 5.2% to 14%) of overall useand 11% (CI, 7% to 15%) of use in 2008.

Intracranial HemorrhageUse of rFVIIa in nontraumatic ICH, particularly

intracerebral hemorrhage, reached a sizable scale in 2004.Use for this indication has also grown, with use in 2008nearly 8-fold higher than that reported in 2004, althoughuse decreased slightly from 2007 to 2008 (Table 1 andAppendix Figure). Intracranial hemorrhage accounts for11% (CI, 6.9% to 14%) of cases overall and 11% (CI,7.7% to 14%) of cases in 2008.

Liver DiseaseA range of indications related to liver disease account

for 6% (CI, 2.5% to 9.3%) of all uses, including liverbiopsy (1.2%), variceal bleeding (1.2%), liver transplanta-tion (0.2%), and other liver indications (3.3%).

Other ConditionsGastrointestinal bleeding other than that associated

with esophageal varices or liver disease accounted for 5.3%(CI, 3.2% to 7.4%) of cases overall. Management of aorticaneurysm, in the presence and absence of surgical interven-tion, represented 1.7% of use overall and in 2008. Otherconditions contributed minimally to rFVIIa use, includingpulmonary indications, particularly biopsy and lung trans-plantation (1.5% overall); cancer-related conditions, par-ticularly leukemia (1.5%); neonatal indications (1%); ob-stetric conditions, particularly postpartum hemorrhage(0.9%); and several surgical procedures, none of which wasindividually prominent (5.3% overall and 2.2% in 2008).Various secondary clotting disorders were associated withrFVIIa use (5.1% overall and 4.2% in 2008), includingsecondary thrombocytopenia and complications of warfa-rin anticoagulation.

Age DistributionConsistent with the growth of off-label indications,

the mean age of patients has increased from 3 years (CI,1.0 to 4.7 years) in 2000 to 59 years (CI, 56 to 61 years) in2008. The age distribution of rFVIIa use varies by indica-tion. Use for hemophilia A and B occurs predominantly inpatients aged 25 years or younger (63% [CI, 48% to77%]). For ICH, 58% (CI, 55% to 61%) of use is inpatients aged 65 years or older, with 36% (CI, 33% to39%) in persons aged 75 years or older. Other conditionsin which use in patients aged 65 years or older is promi-nent include aortic aneurysm (82% [CI, 77% to 87%]),

Table 2. Disposition of Hospitalizations Involving Use ofRecombinant Factor VIIa From 2000 to 2008

Indication Total,n

Died,%*

Other,%†

Home,%

Aortic aneurysm 1216 54.4 26.6 19.0Neonatal indications 729 46.7 15.0 38.3Other liver disease 2451 40.4 27.7 31.9Variceal bleeding 897 39.0 28.8 32.2Other vascular procedures 1515 38.5 30.3 31.2Liver transplant 132 38.1 14.3 47.6Liver biopsy 867 35.9 18.9 45.2Nontraumatic ICH 7755 34.4 49.7 15.9Body trauma 11 689 33.1 36.7 30.2Brain trauma 7158 33.1 44.2 22.7Secondary clotting disorders 3744 30.1 28.6 41.3Nonvariceal GI bleeding 3882 29.7 30.8 39.5Pulmonary hemorrhage 1119 24.5 29.7 45.8Adult cardiovascular surgery 12 086 23.3 28.5 48.2Pediatric cardiovascular surgery 1684 21.7 5.0 73.3Primary clotting disorders 4104 16.8 26.0 57.2Obstetric hemorrhage 672 14.6 7.8 77.6Cancer or stem cell transplant 1094 13.8 18.8 67.4Other procedures 3867 9.4 26.0 64.6Other diagnoses 3345 5.8 17.5 76.7Hemophilia A and B 3121 3.8 11.4 84.8Prostatectomy 120 0.0 19.1 80.9

GI � gastrointestinal; ICH � intracranial hemorrhage.* In-hospital deaths.† Interhospital transfers and transfers to skilled nursing facilities or hospice.



prostatectomy (66% [CI, 33% to 98%]), brain trauma(58% [CI, 54% to 63%]), cardiovascular surgery (57%[CI, 54% to 61%]), and nonvariceal gastrointestinal bleed-ing (57% [CI, 51% to 64%]).

Discharge DispositionOverall in-hospital mortality among patients who re-

ceived rFVIIa was 27% (CI, 19% to 34%). Only 43% (CI,26% to 59%) of patients were discharged directly home(Table 2). Most of the remaining patients were transferredto other facilities (for example, nursing homes and rehabil-itation hospitals). Mortality increased from 5% (CI, 0% to35%) in 2000 to a peak of 31% (CI, 24% to 37%) in 2004but declined slightly to 27% (CI, 25% to 30%) in 2008.Mortality among patients with hemophilia A and B was4% (CI, 1.8% to 5.8%). The highest mortality rates wereassociated with aortic aneurysm (54% [CI, 43% to 65%]),neonatal indications (47% [CI, 35% to 59%]), and non-variceal complications of liver disease (40% [CI, 32% to48%]). Regarding the most common indications, mortalitywas 34% (CI, 31% to 38%) for spontaneous ICH, 33%(CI, 20% to 47%) for brain and body trauma, and 23%(CI, 21% to 26%) for adult cardiovascular surgery.

Hospital CharacteristicsUse of rFVIIa was reported in 235 of the 615 hospitals

(38%) represented in the Premier Perspectives database.Most rFVIIa use occurred in nonacademic hospitals (68%[CI, 39% to 97%]), a pattern that was consistent for mostof the indications we studied (Figure 2). Over time, theproportion of rFVIIa use in nonacademic hospitals grewfrom 11% (CI, 0% to 43%) in 2000 to a peak of 73%(CI, 31% to 71%) in 2005, followed by 67% (CI, 57%to 77%) in 2008. Off-label use of rFVIIa was lower atacademic hospitals (92% [CI, 88% to 96%]) than at non-academic hospitals (97% [CI, 96% to 98%]).

DISCUSSION

The potential for rFVIIa to be used as an emergencyhemostatic agent outside the setting of hemophilia was rec-ognized soon after its clinical introduction (11, 12). Al-though this agent is currently approved by the FDA forcertain uncommon conditions, its in-hospital use for theseis now far exceeded by its use for off-label indications. Byusing a representative database of individual hospitaliza-tions across the United States, we found that 96% of allin-hospital cases of rFVIIa use from 2000 to 2008 and97% of cases in the year 2008 were off-label. These find-ings are consistent with past reports (13–21) (AppendixTable 2, available at www.annals.org).

Adult cardiovascular surgery is the most rapidlyemerging indication for rFVIIa use. In 2008, more than 1in 4 patients who received rFVIIa were treated in the con-text of cardiovascular surgery. There is limited evidence tosupport this use. Diprose and colleagues (22) reported arandomized, controlled trial (RCT) with 20 patients (9received rFVIIa) in which a single dose of rFVIIa was usedprophylactically at the termination of cardiopulmonary by-pass in noncoronary cardiovascular surgery. Although theneed for allogeneic blood transfusion was significantly re-duced after the administration of rFVIIa, there was noeffect on patient survival. More recently, Gill and col-leagues (23) reported a placebo-controlled RCT in whichpatients with bleeding episodes after cardiovascular surgerywere randomly assigned to receive a single dose of rFVIIaat 40 mcg/kg (n � 35) or 80 mcg/kg (n � 69) versusplacebo (n � 68). Significant decreases in the need forreoperation and allogeneic blood transfusions were seen inthe groups that received rFVIIa, but there were no differ-ences in mortality. Furthermore, there were increases inthromboembolic adverse events, particularly stroke, in therFVIIa groups, although they did not reach statisticalsignificance.

Body trauma accounted for 18% of rFVIIa use in2008 and was addressed by 2 simultaneously reportedRCTs. Although these trials demonstrated a significant de-crease in transfusion requirements and possible decrease inthe acute respiratory distress syndrome, they also censoredpatients who died in the first 48 hours of rFVIIa adminis-

Figure 2. Use of rFVIIa at academic and nonacademicinstitutions from 2000 to 2008.

0

Liver transplant

Hemophilia A and B

Adult cardiovascular surgery

Primary clotting disorders

Other vascular procedures

Aortic aneurysm

Neurosurgery

Body trauma

Neonatal indications

Brain trauma

Liver biopsy or resection

Obstetric hemorrhage

Secondary clotting disorders

Intracranial hemorrhage

Pediatric cardiovascular surgery

Other liver disease

Pulmonary hemorrhage

Nonvariceal GI bleeding

Cancer or stem cell transplant

Variceal bleeding

Other procedures

Other diagnoses

Prostatectomy

Use of rFVIIa, %

Academic

Nonacademic

20 40 60 80 100

GI � gastrointestinal; rFVIIa � recombinant factor VIIa.



tration and yet demonstrated no mortality benefit (24).The recently published CONTROL trial (25), which wasterminated early because of failure to demonstrate a mor-tality benefit with use of rFVIIa in blunt and penetratingtrauma at a preplanned interim analysis, suggests that otheraspects of trauma management have improved sufficientlythat mortality is not significantly affected by procoagulantadministration, even with severe hemorrhage. Brain traumaand nontraumatic ICH each constituted 11% of use in2008. Nontraumatic ICH is the subject of perhaps thebest-quality RCT in this field to date. Mayer and col-leagues (26) demonstrated no mortality or functional out-come benefit after early treatment of ICH with rFVIIa,despite evidence that this treatment reduced hematoma ex-pansion. The results of this study may have contributed tothe decline in rFVIIa use for ICH in 2008.

Together, 3 major indications account for 55% ofcases of in-hospital rFVIIa use from 2000 to 2008 and69% of cases in 2008: cardiac surgery (adult and pediatric),trauma (body and brain), and nontraumatic ICH. Remain-ing use is dispersed among a multitude of medical andsurgical indications. Most of these have not been well-studied. Thus, use of rFVIIa is growing in the absence ofclear evidence of therapeutic efficacy and without close sur-veillance for associated harms. In addition, we identified itsuse in 38% of hospitals, with a shift from academic (89%in 2000) to nonacademic hospitals (67% in 2008), suchthat the bulk of overall rFVIIa use has occurred in non-academic settings (Figure 2). This suggests wide adoptionof rFVIIa as a therapy despite concerns about its efficacyand safety.

We also conducted a comparative effectiveness reviewof in-hospital, off-label use of rFVIIa for 5 indications thatdemonstrated no evidence of mortality reduction but anincreased risk for arterial thromboembolic events in cardio-vascular surgery and nontraumatic ICH (27). Levi and co-workers (28) similarly confirmed an increase in arterialthrombotic adverse events among all published RCTs in-vestigating off-label use of rFVIIa. Previous analyses ofvoluntary reports to the FDA Adverse Event ReportingSystem identified deep venous thrombosis, ischemiccerebrovascular accident, and myocardial infarction as themost common adverse events associated with rFVIIa use(18, 19). As noted by Aledort (29), rare adverse events maynot be clearly associated with an intervention on the basisof an individual trial; however, meta-analysis of several tri-als may provide insight into the risk for rare events, such asthe thromboembolic adverse events that seem to be associ-ated with off-label rFVIIa use. In our study, we were un-able to estimate the incidence of patient harm associatedwith rFVIIa use because of insufficient information in thePremier Perspectives database on the temporal relationshipbetween thromboembolic events and the use of rFVIIa.

An indirect measure of the efficacy of rFVIIa in thecare of patients who receive it is survival to discharge. Inthe nationwide sample of hospitals that we analyzed, in-

hospital mortality for patients who received rFVIIa washigh: 27% overall and as high as 40% to 50% for severalindications (Table 2). In-hospital mortality has been notedto be higher in retrospective studies of rFVIIa use outsideof clinical trials than that seen in carefully selected clinicaltrial populations (7, 10, 13, 27, 30, 31).

Our study has several limitations. The Premier Per-spectives database gathers data from a voluntary alliance ofnonfederal hospitals and thus is not a random selection ofhealth care facilities nationwide. Our analysis of indica-tions associated with use of rFVIIa was confined to theprimary and secondary diagnoses (as listed by ICD-9 code)reported for each hospitalization. We cannot validate thesediagnoses or identify diagnostic codes that may have beenomitted. In addition, the indication hierarchy we used tocategorize rFVIIa use may not have captured the nuancesof each individual case. For instance, we are unable todetermine at what point in a disease process rFVIIa wasused (for example, as prophylaxis, as treatment, or for end-stage catastrophic bleeding episodes). We also did not haveaccess to data on patients with similar clinical scenarios inwhich rFVIIa was not used.

Although utilization of medications will change overtime as additional hypotheses about patient therapeuticsare examined, the key question is at what point has a newindication attained a sufficient evidence base to justifygreatly expanded use. Given the absence of supporting dataand the suggestion of patient harm, our analysis has iden-tified probable nationwide overuse of rFVIIa outside theconditions approved by the FDA. We hope that this andsimilar analyses of real-world use of medications will pro-vide guidance for the design of clinical trials and evidencereviews that adequately address the efficacy and associatedharms of individual agents in the contexts in which theyare most commonly used. If concerning patterns of off-label use exist for other medications, these analyses mayalso help highlight areas for improvement in systems re-sponsible for drug approval and subsequent surveillance.

From Stanford Prevention Research Center and Stanford UniversitySchool of Medicine, Stanford, California.

Disclaimer: The authors of this report are responsible for its content.Statements in the report should not be construed as endorsement by theAgency for Healthcare Research and Quality or the U.S. Department ofHealth and Human Services.

Acknowledgment: The authors thank C. Vaughan Tuohy, Dena M.Brevata, Kristan Staudenmayer, Robin Eisenhut, Vandana Sundaram,Donal McMahon, David Johnston, Christopher Stave, James Zehnder,Ingram Olkin, Kathryn M. McDonald, and Douglas K. Owens for theircontributions to an associated comparative effectiveness review commis-sioned by the Agency for Healthcare Research and Quality.

Grant Support: By the Agency for Healthcare Research and Quality,U.S. Department of Health and Human Services, under contract 290-02-0017. Dr. Stafford also was supported by a National Heart, Lung,and Blood Institute mentoring award (RSS, K24HL086703).



Potential Conflicts of Interest: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum�M10-2334.

Reproducible Research Statement: Study protocol: Available from www.ahrq.gov. Statistical code: Available from Dr. Stafford (e-mail, [email protected]). Data set: Available for purchase from Premier Perspec-tives (www.premierinc.com).

Requests for Single Reprints: Randall S. Stafford, MD, PhD, StanfordPrevention Research Center, Stanford University, Medical School OfficeBuilding X312, 251 Campus Drive, Stanford, CA 94304-5411; e-mail,[email protected].

Current author addresses and author contributions are available at www.annals.org.

References1. Stafford RS. Regulating off-label drug use—rethinking the role of the FDA. NEngl J Med. 2008;358:1427-9. [PMID: 18385495]2. Hedner U. Factor VIIa and its potential therapeutic use in bleeding-associatedpathologies. Thromb Haemost. 2008;100:557-62. [PMID: 18841276]3. Roberts HR, Monroe DM, White GC. The use of recombinant factor VIIa inthe treatment of bleeding disorders. Blood. 2004;104:3858-64. [PMID:15328151]4. Roberts HR, Monroe DM 3rd, Hoffman M. Safety profile of recombinantfactor VIIa. Semin Hematol. 2004;41:101-8. [PMID: 14872430]5. Thomas GO, Dutton RP, Hemlock B, Stein DM, Hyder M, Shere-WolfeR, et al. Thromboembolic complications associated with factor VIIa administra-tion. J Trauma. 2007;62:564-9. [PMID: 17414329]6. Diringer MN, Skolnick BE, Mayer SA, Steiner T, Davis SM, Brun NC,et al. Risk of thromboembolic events in controlled trials of rFVIIa in spontaneousintracerebral hemorrhage. Stroke. 2008;39:850-6. [PMID: 18239180]7. Hsia CC, Chin-Yee IH, McAlister VC. Use of recombinant activated factorVII in patients without hemophilia: a meta-analysis of randomized control trials.Ann Surg. 2008;248:61-8. [PMID: 18580208]8. Diringer MN, Skolnick BE, Mayer SA, Steiner T, Davis SM, Brun NC,et al. Thromboembolic events with recombinant activated factor VII in sponta-neous intracerebral hemorrhage: results from the Factor Seven for Acute Hemor-rhagic Stroke (FAST) trial. Stroke. 2010;41:48-53. [PMID: 19959538]9. Howes JL, Smith RS, Helmer SD, Taylor SM. Complications of recombi-nant activated human coagulation factor VII. Am J Surg. 2009;198:895-9.[PMID: 19969148]10. Yank V, Tuohy CV, Logan AC, Bravata DM, Staudenmayer K, EisenhutR, et al. Comparative effectiveness of recombinant factor VIIa for off-label indi-cations versus usual care. (Prepared by Stanford-UCSF Evidence-based PracticeCenter under contract no. 290-02-0017.) Rockville, MD: Agency for HealthcareResearch and Quality; 2010. Accessed at www.effectivehealthcare.ahrq.gov/reports/final.cfm on 1 March 2011.11. Goodnough LT. Treatment of critical bleeding in the future intensive careunit. Intensive Care Med. 2002;28 Suppl 2:S221. [PMID: 12583409]12. Mayer SA. Ultra-early hemostatic therapy for intracerebral hemorrhage.Stroke. 2003;34:224-229. [PMID: 12511778]13. Hsia CC, Zurawska JH, Tong MZ, Eckert K, McAlister VC, Chin-Yee IH.Recombinant activated factor VII in the treatment of non-haemophilia patients:physician under-reporting of thromboembolic adverse events. Transfus Med.2009;19:43-9. [PMID: 19302454]14. Magnetti S, Oinonen M, Matuszewski KA. An evaluation of off-label use ofrecombinant activated human factor VII (NovoSeven): patient characteristics,

utilization trends, and outcomes from an electronic database of US academichealth centers. P&T. 2007;32:218-30.15. Carcao MD, Webert KE. On-label versus off-label use of recombinant acti-vated factor VII: a comprehensive review of use in two Canadian centers. SeminHematol. 2008;45:S68-71. [PMID: 18544429]16. Heller M, Lau W, Pazmino-Canizares J, Brandao LR, Carcao M. A com-prehensive review of rFVIIa use in a tertiary care pediatric center. Pediatr BloodCancer. 2008;50:1013-7. [PMID: 17960639]17. Webert KE, Arnold DM, Carruthers J, Molnar L, Almonte T, Decker K,et al. Utilization of recombinant activated factor VII in southern Ontario in 85patients with and without haemophilia. Haemophilia. 2007;13:518-26. [PMID:17880438]18. Aledort LM. Comparative thrombotic event incidence after infusion of re-combinant factor VIIa versus factor VIII inhibitor bypass activity. J ThrombHaemost. 2004;2:1700-8. [PMID: 15456478]19. O’Connell KA, Wood JJ, Wise RP, Lozier JN, Braun MM. Thromboem-bolic adverse events after use of recombinant human coagulation factor VIIa.JAMA. 2006;295:293-8. [PMID: 16418464]20. MacLaren R, Weber LA, Brake H, Gardner MA, Tanzi M. A multicenterassessment of recombinant factor VIIa off-label usage: clinical experiences andassociated outcomes. Transfusion. 2005;45:1434-42. [PMID: 16131375]21. Goodnough LT, Lublin DM, Zhang L, Despotis G, Eby C. Transfusionmedicine service policies for recombinant factor VIIa administration. Transfu-sion. 2004;44:1325-31. [PMID: 15318856]22. Diprose P, Herbertson MJ, O’Shaughnessy D, Gill RS. Activated recombi-nant factor VII after cardiopulmonary bypass reduces allogeneic transfusion incomplex non-coronary cardiac surgery: randomized double-blind placebo-controlled pilot study. Br J Anaesth. 2005;95:596-602. [PMID: 16183679]23. Gill R, Herbertson M, Vuylsteke A, Olsen PS, von Heymann C, MythenM, et al. Safety and efficacy of recombinant activated factor VII: a randomizedplacebo-controlled trial in the setting of bleeding after cardiac surgery. Circula-tion. 2009;120:21-7. [PMID: 19546387]24. Boffard KD, Riou B, Warren B, Choong PI, Rizoli S, Rossaint R, et al;NovoSeven Trauma Study Group. Recombinant factor VIIa as adjunctive ther-apy for bleeding control in severely injured trauma patients: two parallel random-ized, placebo-controlled, double-blind clinical trials. J Trauma. 2005;59:8-15.[PMID: 16096533]25. Hauser CJ, Boffard K, Dutton R, Bernard GR, Croce MA, Holcomb JB,et al; CONTROL Study Group. Results of the CONTROL trial: efficacy andsafety of recombinant activated factor VII in the management of refractory trau-matic hemorrhage. J Trauma. 2010;69:489-500. [PMID: 20838118]26. Mayer SA, Brun NC, Begtrup K, Broderick J, Davis S, Diringer MN, et al;FAST Trial Investigators. Efficacy and safety of recombinant activated factor VIIfor acute intracerebral hemorrhage. N Engl J Med. 2008;358:2127-37. [PMID:18480205]27. Yank V, Tuohy CV, Logan AC, Bravata DM, Staudenmayer K, EisenhutR, et al. Systematic review: benefits and harms of in-hospital use of recombinantfactor VIIa for off-label indications. Ann Intern Med. 2011;154:529-40.28. Levi M, Levy JH, Andersen HF, Truloff D. Safety of recombinant activatedfactor VII in randomized clinical trials. N Engl J Med. 2010;363:1791-800.[PMID: 21047223]29. Aledort LM. Off-label use of recombinant activated factor VII—safe or notsafe? [Editorial]. N Engl J Med. 2010;363:1853-4. [PMID: 21047230]30. Ganguly S, Spengel K, Tilzer LL, O’neal B, Simpson SQ. Recombinantfactor VIIa: unregulated continuous use in patients with bleeding and coagulopa-thy does not alter mortality and outcome. Clin Lab Haematol. 2006;28:309-12.[PMID: 16999720]31. Hardy JF, Belisle S, Van der Linden P. Efficacy and safety of recombinantactivated factor VII to control bleeding in nonhemophiliac patients: a review of17 randomized controlled trials. Ann Thorac Surg. 2008;86:1038-48. [PMID:18721620]



Current Author Addresses: Dr. Logan: Division of Blood and MarrowTransplantation, Stanford University School of Medicine, 265 CampusDrive, Lokey Stem Cell Research Building, Room G3045, Stanford, CA94305.Drs. Yank and Stafford: Stanford Prevention Research Center, StanfordUniversity, Medical School Office Building X312, 251 Campus Drive,Stanford, CA 94304-5411.

Author Contributions: Conception and design: A.C. Logan, V. Yank,R.S. Stafford.Analysis and interpretation of the data: A.C. Logan, V. Yank, R.S.Stafford.Drafting of the article: A.C. Logan, R.S. Stafford.Critical revision of the article for important intellectual content: A.C.Logan, V. Yank, R.S. Stafford.Final approval of the article: A.C. Logan, V. Yank, R.S. Stafford.Provision of study materials or patients: R.S. Stafford.Statistical expertise: R.S. Stafford.Obtaining of funding: R.S. Stafford.Administrative, technical, or logistic support: R.S. Stafford.Collection and assembly of data: A.C. Logan, R.S. Stafford.

Appendix Table 1. Diagnostic Hierarchy for Analysis of the Premier Perspectives Database

Rank inHierarchy

Description Most Frequent Conditions or Procedures

1 Hemophilia A and B Hemophilia A and B2 Primary clotting disorders Other clotting factor deficiencies, Glanzmann thrombasthenia3 Brain trauma Subdural hemorrhage, subarachnoid hemorrhage4 Body trauma Motor vehicle accident, fall, assault5 Intracranial hemorrhage Intracerebral hemorrhage, subdural hemorrhage6 Neurosurgery Excision of lesion, craniotomy7 Pediatric cardiovascular surgery Transposition of the great vessels, atrial septic defect8 Adult cardiovascular surgery Aortic valve replacement, CABG, mitral valve replacement9 Obstetric hemorrhage Immediate postpartum hemorrhage, preeclampsia10 Neonatal indications The respiratory distress syndrome11 Aortic aneurysm Abdominal aortic aneurysm, thoracic aortic aneurysm12 Prostatectomy Retropubic prostatectomy13 Other vascular procedures Vascular bypass, intra-abdominal venous shunt14 Liver transplant Liver transplant15 Liver biopsy Closed biopsy of liver, open biopsy of liver16 Variceal bleeding Esophageal varices17 Other liver disease Nonalcoholic cirrhosis, alcoholic cirrhosis18 Nonvariceal GI bleeding Unspecified GI bleeding, ischemic bowel19 Secondary clotting disorders Unspecified coagulation defect, defibrination syndrome20 Pulmonary hemorrhage Closed bronchial biopsy, hemoptysis21 Cancer or stem cell transplant Acute lymphoid leukemia, acute myeloid leukemia22 Other surgical procedures Various23 Other diagnoses Various

CABG � coronary artery bypass graft; GI � gastrointestinal.

Annals of Internal Medicine

W-174 19 April 2011 Annals of Internal Medicine Volume 154 • Number 8 www.annals.org

Appendix Figure. Estimated annual in-hospital cases usingrecombinant factor VIIa for adult and pediatric cardio-vascular surgery (top) and for trauma and ICH (bottom).

Cas

es, n

Year

Body trauma

Brain trauma

Nontraumatic ICH

2000

2001

2002

2003

2004

2005

2006

2007

2008

0

5000

6000

4000

3000

2000

1000

Cas

es, n

0

3000

4000

2000

1000

Adult cardiovascular surgery

Pediatric cardiovascular surgery

ICH � intracranial hemorrhage.

Appendix Table 2. Prevalence of Off-Label Use of Recombinant Factor VIIa at Academic (Single or Multiple Institutions) andNonacademic Medical Centers

Study, Year (Reference) Latest Year Included Setting Cases, n Off-Label Use, %

This report 2008 Academic or nonacademic 12 644* 97.0Hsia et al, 2009 (13) 2007 Academic (single) 94 88.3Magnetti et al, 2007 (14) 2005 Academic (multiple) 2660 89.0Carcao and Webert, 2008 (15) 2005 Academic (multiple) 196 80.0Heller et al, 2008 (16) 2005 Academic (single) 111 79.0Webert et al, 2007 (17) 2005 Academic or nonacademic 85 83.6O’Connell et al, 2006 (19) 2004 Academic or nonacademic† 168 89.9MacLaren et al, 2005 (20) 2004 Academic or nonacademic 701 92.0Goodnough et al, 2004 (21) 2004 Academic (single) 130 93.8

* Number of unweighted observations.† Subset of sample used for this report.

www.annals.org 19 April 2011 Annals of Internal Medicine Volume 154 • Number 8 W-175

4-19-11 annals of internal medicine - off-label use of factor viia

Documents