An Evaluation of Thrombophilia Screening in an Urban Tertiary Care ...

Coagulation and Transfusion Medicine / THROMBOPHILIA SCREENING

An Evaluation of Thrombophilia Screening in an Urban Tertiary Care Medical Center A “Real World” Experience Jonathan Somma, MD, Ira I. Sussman, MD, and Jacob H. Rand, MD Key Words: Thrombophilia; Hypercoagulability; Hypercoagulable; Thrombosis; Thromboembolism; Stroke; Pregnancy; Screening; Overutilization; Utilization DOI: 10.1309/KV0632LJ8EDMEWQT

Abstract To evaluate the utilization of thrombophilia screening at a large urban academic tertiary care center, we retrospectively examined the indications, appropriateness, and results of 200 consecutive thrombophilia panels. Of the panels, 103 (51.5%) were ordered for venous thromboembolism; 124 (62.0%) were ordered during acute thrombotic episodes, and at least 40 (20.0%) had abnormal protein C and S results, of which 25 (63%) were attributable to anticoagulation and the remainder to pregnancy. Of the 200 panels, 46 (23.0%) had a significant abnormality; the most common abnormal result was a lupus anticoagulant, occurring in 23 cases (11.3%). Thrombophilia screening seems to be overutilized in our population, especially considering that the majority of tests are ordered during suboptimal conditions, eg, acute thrombosis, pregnancy, or anticoagulation. At present, outside the research setting, thrombophilia panels should be reserved for special circumstances, targeted to factors for which there would be a specific clinical impact, and performed in the absence of confounding clinical variables.

120 120

Am J Clin Pathol 2006;126:120-127 DOI: 10.1309/KV0632LJ8EDMEWQT

The increased recognition of the potential roles of thrombophilic factors in the development of thromboembolic disorders in recent years has been accompanied by a marked increase in the screening for these abnormalities in clinical laboratories. A major factor that has motivated physicians to order these tests has been the belief that the results could help guide clinical decisions on the appropriate duration of anticoagulant treatment, eg, the idea that patients with identifiable thrombophilias may require longer-term treatment than patients without these risk factors. This concept has been challenged by 2 recent articles. The first, by Christiansen et al,1 questioned the significance of laboratory-determined, prothrombotic risk factors in predicting the recurrence of venous thromboembolism (VTE). The other article, a meta-analysis by Ost et al,2 demonstrated a benefit for prolonged anticoagulation in patients with VTE generally, ie, without defining a specific thrombophilic cause. These articles led us to investigate, for quality assurance purposes, the current physician ordering patterns for thrombophilia screening at Montefiore Medical Center, Bronx, NY, a large (1,062-bed), urban, academic tertiary care facility. Although there have been previous reports about the prevalence of prothrombotic conditions in a variety of populations3-8 and on the use of thrombophilia testing during unreliable clinical circumstances,9,10 there has not, to our knowledge, been an extensive study evaluating the appropriateness of thrombophilia screening in a “real world” setting. We, therefore, undertook a retrospective review of a consecutive series of thrombophilia workups to determine the clinical context in which thrombophilia panels are ordered, the results of these tests, and whether testing would be expected to be valuable in guiding clinical decision making. © American Society for Clinical Pathology

Coagulation and Transfusion Medicine / ORIGINAL ARTICLE

Our subjective impressions during daily reviews and interpretations of the results at the Montefiore Advanced Coagulation Laboratory led us to hypothesize, before undertaking this study, that thrombophilia panels are overutilized for the following reasons: the apparent low prevalence of thrombophilias in our population; the generally small potential for results to influence patient management decisions; the presence of known confounding clinical variables at the time of testing, including acute thrombosis, pregnancy, and anticoagulation9,10; and the significant expense associated with testing.

Materials and Methods The results of all consecutive, complete thrombophilia panels ordered between May 1 and October 31, 2005, by Montefiore Medical Center physicians were obtained and reviewed, and the following patient data were compiled from the electronic databases available at Montefiore: age; sex; patient location; clinical indication for thrombophilia workup; ordering service (defined as the service affiliation of the physician who recommended the panel); and, when relevant, record of medication administration and results of additional laboratory studies. For outpatients whose electronic record was insufficient to provide the indication for the workup, the physician requesting the panel was contacted for additional information. The accurate acquisition of these data was made possible by the implementation, as of May 1, 2005, of an electronic medical record archive system (Horizon Patient Folder, version 6.1.1010, McKesson, San Francisco, CA) in which patients’ written medical records are scanned and accessible on the Montefiore network within 24 hours of the conclusion of their visit. In addition, the clinical information system that contains a variety of information, including patients’ medication records, laboratory data, and imaging results, was consulted. The indications for ordering thrombophilia panels were categorized as part of a workup for the following: VTE, stroke (CVA), intrauterine fetal demise (IUFD), other, or unknown. The thrombophilia panel at Montefiore Medical Center evaluates 6 putative prothrombotic risk factors and is performed using commercially available methods as follows: (1) lupus anticoagulant (LA), using the LA Screen-25 for Detection of Lupus Anticoagulant Kit and the LA Confirm Kit (both from Life Diagnostics, Clarkston, GA) and performed on the Dade Behring Blood Coagulation System (BCS; Deerfield, IL); (2) antithrombin (also known as antithrombin-III) deficiency, using the Berichrome Antithrombin III Kit (functional assay; Dade Behring) and performed on the BCS and the ATIII Quiplate Kit (immunologic assay; Helena Laboratories, Beaumont, TX) and performed manually; (3) protein C deficiency, using the Berichrome Protein C Kit (functional assay;

Dade Behring) and performed on the BCS and the Protein C ELISA (enzyme-linked immunosorbent assay) Kit (immunologic assay; Helena Laboratories) and performed manually; (4) protein S deficiency, using the Staclot Protein S Kit (functional assay) and the LIATest Protein S Kit (total immunologic assay; both from Diagnostica Stago, Parsippany, NJ) and performed on the BCS; (5) plasminogen deficiency, using the Plasminogen Quiplate Kit (Helena Laboratories) and performed manually; and (6) activated protein C resistance, using the Coatest APC Resistance V Kit (Chromogenix, West Chester, OH) and performed on the BCS. Prices for equivalent testing at Quest Diagnostics Nichols Institute were determined from telephone quotations from its billing department at the time this article was written.

Results A total of 200 patient samples were sent for complete thrombophilia panels during the 6-month period from May 1 to October 31, 2005, at Montefiore Medical Center. It was possible to establish the indication for the panel in 197 cases. The following 16 cases did not fit into the 3 main categories (VTE, CVA, and IUFD) and were classified in the “other” category: central retinal vein occlusion, 2 cases; and 1 case each of the following: prepregnancy hematology consult in a known factor V Leiden heterozygote with a family history of thrombosis and factor V Leiden; incidentally found positive lupus anticoagulant before wisdom tooth extraction; positive phospholipid antibodies and family history of thrombosis; bullous leukocytoclastic vasculitis; shortened activated partial thromboplastin time, family history of thrombosis, and taking oral contraceptives; family history of positive methylenetetrahydrofolate reductase (MTHFR) polymorphism and homozygous for MTHFR; atherothrombotic aortoiliac disease, 2 deep vein thromboses (DVTs), and recurrent thrombosis of hemodialysis access; recurrent thrombosis of hemodialysis access; left common iliac artery aneurysm; extensive arterial thromboses involving the right kidney, spleen, superior mesenteric artery, loops of small bowel, and right femoral artery and a history of cystectomy for bladder cancer 1 year earlier; extensive arterial peripheral vascular disease and history of DVT, myocardial infarction, and cocaine abuse; acute pancreatitis, 7 episodes of pancreatitis during the previous 7 years, and pulmonary embolism; sepsis with secondary disseminated intravascular coagulation and a history of cervical cancer, DVT, and pulmonary embolism; and myocardial infarction and a history of extensive atherosclerotic cardiovascular disease. The demographic characteristics of the patients are shown in ❚Table 1❚. Of the 200 patients, 140 (70.0%) were females and 60 (30.0%) were males. Even after excluding the Am J Clin Pathol 2006;126:120-127

© American Society for Clinical Pathology 121

DOI: 10.1309/KV0632LJ8EDMEWQT

121 121

Somma et al / THROMBOPHILIA SCREENING

thrombophilia panels ordered for IUFD, tests ordered for female patients still markedly outnumbered those ordered for male patients (65.5% vs 34.5%). The overall mean and median age distributions were 42 and 42 years (range, 6-92 years) for females and 45 and 45 years (range, 2-82 years) males. The mean and median ages of the CVA group (42 and 46.5 years) were younger than would have been expected given the high prevalence of cerebral ischemia in elderly populations. A total of 9 medical departments were responsible for ordering the panels ❚Table 2❚, the largest number of which were ordered by Hematology. VTE was the indication in the majority of cases, and 56 (54.4%) of these panels were ordered by Hematology. The second most common indication was CVA, and the majority of these panels were ordered by Neurology (43 [83%]). Overall, 124 panels (62.0%) were performed on inpatients (Table 2), and the majority in the VTE and CVA subgroups

were inpatients as well. Of note, included in the outpatient group were 5 patients seen in the Emergency Department who were not subsequently admitted to the hospital. In determining the appropriateness of ordering thrombophilia panels, situations that could be anticipated to yield false-positive results should be taken into account ❚Table 3❚. There were 25 results of decreased protein S or decreased protein C and S levels in patients known to be taking warfarin; these were considered normal for the purposes of this study and represented 63% of the known false-positive results. There were 15 decreased protein S results in patients who were known to be pregnant or who had an IUFD diagnosed on the day of thrombophilia screening; these were considered normal as well. There were other miscellaneous abnormalities that were not included in the data analysis. One case was a septic patient with secondary disseminated intravascular coagulation who

❚Table 1❚ Demographic Features of Patients Undergoing Thrombophilia Evaluation* Indication† VTE Female Male Total Mean/median age (y) All patients Females Males

69 (67.0) 34 (33.0) 103 (51.5) 46/45 (6-92) 45/45 (6-92) 48/44.5 (17-82)

CVA 34 (65) 18 (35) 52 (26.0) 42/46.5 (2-78) 43/46.5 (13-78) 40/46 (2-76)

Other‡

IUFD 26 (100) 0 (0) 26 (13.0) — 29/29 (19-39) —

9 (56) 7 (44) 16 (8.0) 46/48 (14-78) 45/52 (14-78) 47/44 (23-66)

Unknown 2 (67) 1 (33) 3 (1.5) 58/53 (51-71) 61/61 (51-71) —

Total 140 (70.0) 60 (30.0) 200 43/43 (2-92) 42/42 (6-92) 45/45 (2-82)

* Values

are expressed as number (percentage) or number (range). The indications for ordering thrombophilia panels were as part of a workup for venous thromboembolism (VTE), stroke (CVA), intrauterine fetal demise (IUFD), other, or unknown. ‡ See Results section. †

❚Table 2❚ Requests for Thrombophilia Evaluation by Ordering Service and Patient Location* Indication†

Ordering service Cardiology Dermatology Hematology General medicine Neurology Obstetrics Renal General surgery Vascular surgery Patient location Inpatient Outpatient Total

IUFD

Other‡

0 (0) 0 (0) 5 (10) 4 (8) 43 (83) 0 (0) 0 (0) 0 (0) 0 (0)

0 (0) 0 (0) 0 (0) 1 (4) 0 (0) 25 (96) 0 (0) 0 (0) 0 (0)

40 (77) 12 (23) 52 (26.0)

12 (46) 14 (54) 26 (13.0)

VTE

CVA

4 (3.9) 0 (0.0) 56 (54.4) 36 (35.0) 2 (1.9) 0 (0.0) 1 (1.0) 1 (1.0) 3 (2.9) 64 (62.1) 39 (37.9) 103 (51.5)

Unknown

Total

1 (6) 1 (6) 10 (63) 4 (25) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)

0 (0) 0 (0) 2 (67) 1 (33) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)

5 (2.5) 1 (0.5) 73 (36.5) 46 (23.0) 45 (22.5) 25 (12.5) 1 (0.5) 1 (0.5) 3 (1.5)

8 (50) 8 (50) 16 (8.0)

0 (0) 3 (100) 3 (1.5)

124 (62.0) 76 (38.0) 200

* Values

are expressed as number (percentage). Some percentage totals do not equal 100% because of rounding. The indications for ordering thrombophilia panels were as part of a workup for venous thromboembolism (VTE), stroke (CVA), intrauterine fetal demise (IUFD), other, or unknown. ‡ Includes the other diagnoses described in Table 1. †

122 122


© American Society for Clinical Pathology


❚Table 3❚ False-Positive Protein C and S Results* Indication† Reason for False-Positive Result Warfarin Pregnancy Total

VTE

CVA

IUFD

Other‡

Unknown

Total

22 1 23

2 0 2

0 14 14

1 0 1

0 0 0

25 (63) 15 (38) 40

* Values

are expressed as number (percentage). The percentage total does not equal 100% because of rounding. The indications for ordering thrombophilia panels were as part of a workup for venous thromboembolism (VTE), stroke (CVA), intrauterine fetal demise (IUFD), other, or unknown. ‡ Includes the other diagnoses described in Table 1. †

would have been expected to have a decreased protein C result. In addition, results of the following 3 panels were considered unlikely to be clinically significant: (1) an increased plasminogen level, a decreased total immunologic protein S level, and a normal functional protein S level; (2) increased functional and immunologic protein C and S levels; and (3) an increased functional and immunologic protein C level. Five of the panels that were positive had abnormalities involving more than one prothrombotic risk factor. One case was an outpatient with a DVT (prothrombin time [PT], 30 seconds; no evidence of liver disease) who had a LA, decreased protein C (functional and immunologic) level, and decreased protein S (functional and immunologic) level in whom warfarin use was likely but could not be confirmed. The second was an inpatient with a DVT and a decreased functional antithrombin level who was not taking warfarin despite a decreased protein C (functional and immunologic) level and decreased protein S (functional and immunologic) level. This patient likely had vitamin K deficiency secondary to antibiotic treatment; the patient had had a PT of 10.5 seconds on admission that, after 9 days of broad-spectrum intravenous antibiotics and no oral feeding, rose to 12.2 seconds and later corrected with resumption of oral diet and multivitamins. The third patient was the inpatient from the other category (described earlier) with extensive arterial thromboses. This patient’s panel revealed a LA, decreased functional antithrombin level, and slightly decreased functional protein C and S levels. The patient recently had started receiving broad-spectrum intravenous antibiotics and had a PT that had correspondingly increased from 11.4 to 15.2 seconds. The patient, therefore, was presumed to have an early vitamin K deficiency. All decreased protein C and S components of these 3 cases were excluded from the data analysis. The fourth case of multiple prothrombotic abnormalities was an inpatient who was being evaluated for a pulmonary embolism. The patient had a borderline LA and decreased functional protein S level. The fifth patient was an inpatient admitted for a DVT who had a borderline LA and decreased functional antithrombin level.

As shown in ❚Table 4❚, 154 (75.9%) of the 203 results were within normal limits (WNL); the total number of results (203) exceeds the total number of panels (200) because the former includes 3 panels (the third, fourth, and fifth cases described in the previous 2 paragraphs) that each were counted as having 2 prothrombotic abnormalities. Further breakdown of the results by indication revealed that 69.5% of the VTE results (73/105) were WNL, 87% of the CVA results (45/52) were WNL, and 96% of the IUFD results (25/26) were WNL. Of the 49 positive results, the most common thrombophilic abnormality was a LA, occurring in 23 (11.3%) of 203 cases, followed by antithrombin (5.9%), protein S (4.9%), and activated protein C resistance (2.0%). LAs were the most common abnormality in the VTE group, occurring in 17 (16.2%) of 105 cases. It is noteworthy that, of all the abnormal LA results, 15 of 23 were borderline; at Montefiore, a borderline LA result is reported when the dilute Russell viper venom time (DRVVT) mixing test/confirmation ratio is 1.1, whereas a definite positive is reported with a ratio of 1.2 or greater. There also were 2 variant LA results in which, although the DRVVT was negative, the activated partial thromboplastin time was prolonged with a LA-sensitive reagent but normal with a LA-insensitive reagent. The expense associated with thrombophilia screening may be estimated based on the rate billed by a large commercial laboratory such as Quest Diagnostics, a large provider of diagnostic testing in the United States performing more than 250 million diagnostic laboratory tests annually.11 As of September 8, 2005, the Quest patient rate (market rate) for the tests represented on Montefiore’s thrombophilia panel is $2,083.25. At this rate, the annual thrombophilia panel cost for the approximately 400 complete panels performed each year at Montefiore would be $833,000, and the expense associated with ordering thrombophilia screens on all of the estimated 275,000 new cases of VTE that occur in the United States annually12 would be $573 million. It should be noted that Quest has several of its own suggested thrombophilia panels, some of which include additional tests, for which the overall retail cost per panel ranges from $1,311.75 (Thrombophilia Am J Clin Pathol 2006;126:120-127



123 123


❚Table 4❚ Results of Thrombophilia Panels* Indication† Panel Results

VTE

LA+ 17 (16.2) Decreased PS§ 5 (4.8) Decreased AT 7 (6.7) APCR 3 (2.9) Decreased plasminogen 0 (0.0) WNL 73 (69.5) Total 105 (51.7)

IUFD

Other‡

Unknown

Total

0 (0) 0 (0) 1 (4) 0 (0) 0 (0) 25 (96) 26 (12.8)

4 (24) 1 (6) 1 (6) 1 (6) 0 (0) 10 (59) 17 (8.4)

1 (33) 0 (0) 1 (33) 0 (0) 0 (0) 1 (33) 3 (1.5)

23 (11.3) 10 (4.9) 12 (5.9) 4 (2.0) 0 (0.0) 154 (75.9) 203

CVA 1 (2) 4 (8) 2 (4) 0 (0) 0 (0) 45 (87) 52 (25.6)

APCR, activated protein C resistance; Decreased AT, decreased functional or decreased functional and immunologic antithrombin; Decreased PS, decreased functional and/or immunologic protein S; LA+, lupus anticoagulant positive; WNL, within normal limits. * Values are expressed as number (percentage). Percentage totals do not equal 100% because of rounding. The total number of results (203) exceeds the total number of panels (Tables 1 and 2) because the results include 3 panels that each have 2 prothrombotic abnormalities. † The indications for ordering thrombophilia panels were as part of a workup for venous thromboembolism (VTE), stroke (CVA), intrauterine fetal demise (IUFD), other, or unknown. ‡ Includes the other diagnoses described in Table 1. § Does not include the 40 false-positive results in Table 3.

Screen, Inherited) to $2,429 (Thrombophilia Comprehensive Panel). It also should be noted that the actual payments by individual providers may vary from the “list” prices, depending on the health coverage status of individual patients.

Discussion An important aspect of this study is that, in contrast with other reported studies, including those from thrombosis clinics or centers, we reviewed a large, consecutive, real world series of unselected patients who were screened for thrombophilias. To our knowledge, such an evaluation of real world indications for thrombophilia screening has not been reported. In this series of 200 thrombophilia panels, 46 (23.0%) had a significant abnormality. The most frequent indication for screening was VTE (51.7%), of which 69.5% of results were negative. Of the panels, about 62% were ordered inappropriately at the time of hospitalization for the acute event, and at least 20.0% of panels (40/200) had abnormal protein C and S results, of which 63% were attributable to anticoagulation with warfarin and 38% to pregnancy; these represent false-positive results and were considered normal in the analysis. Of the abnormal results that may represent true thrombophilic states, the most common finding was a LA occurring in 23 (47%) of 49 cases. Among the other interesting findings is the preponderance of females undergoing screening. It is possible that our sample includes a large proportion of females who have had provoked thrombotic events, perhaps often as a result of oral contraceptive use, because females are approximately 3 times more likely to have a provoked VTE than are males.1 However, this alone is unlikely sufficient to explain the female/male ratio of more than 2:1. Another demographic result of interest is the relatively young age of the CVA group (mean and median, 42 and 46.5 years). A likely explanation is 124 124


that these panels were ordered as workups for strokes of unknown cause, which may represent up to 40% of stroke cases.9 The stroke patients selected for thrombophilia testing presumably are less likely to have established clinical risk factors such as chronic hypertension, diabetes, hypercholesterolemia, cigarette smoking, and atherosclerosis. It is important, however, to recognize that the prevalence of a thrombophilic cause for stroke is low and that screening in cases of cerebral ischemia is of questionable value because of the limited data on the optimal management of stroke patients with thrombophilias.9 The appropriate use of anticoagulation after idiopathic VTE is controversial owing to the uncertainty of the role of genetic and biochemical thrombophilic risk factors in predicting recurrence and the unclear benefit of extending anticoagulation. Christiansen et al1 assessed the former issue by prospectively following up a cohort of 474 patients with VTE to determine the risk of recurrence in the context of specific prothrombotic abnormalities. They concluded that there is no major effect on VTE recurrence in patients with only one of the studied prothrombotic risk factors, and, therefore, if these data are confirmed, thrombophilia workups may not have practical usefulness in predicting VTE recurrence or in determining the need for long-term anticoagulation for such patients. The data presented by Christiansen et al1 are at odds with much of the preceding literature,13-20 which overall demonstrated a significantly higher relative risk of VTE recurrence associated with several putative prothrombotic risk factors, including deficiency of the natural anticoagulants (protein C, protein S, and antithrombin), prothrombin 20210A mutation, and elevated levels of factor VIII.21 In contrast with the findings of Christiansen et al,1 in which 67% of patients had at least 1 prothrombotic abnormality, only about 23% showed an abnormality in our sample. Although this large discrepancy probably is multifactorial and © American Society for Clinical Pathology


related to the 2 significantly different study designs, a portion of it may be explained by the considerably different patient populations; a Northern European population, likely largely Caucasian, was evaluated in Christiansen et al1 vs the largely non-Caucasian population of the Bronx, NY, served by Montefiore Medical Center, which in the calendar year 2004 had the following patient profile: Hispanic, 36%; African American/black, 27%; unknown, 22%; Caucasian/white, 10%; other, 4%; and Asian/Pacific islander, 1%. In view of the lower prevalence of genetic prothrombotic risk factors among non-Caucasians,7,8 general thrombophilia screening would be of even less clinical usefulness for predicting VTE recurrence in medical centers like ours, with large non-Caucasian populations, than for the population studied by Christiansen et al.1 Furthermore, there are only a limited number of situations in which positive thrombophilia results would be expected to influence clinical management. Patients who have had an idiopathic first VTE are likely to receive an extended period of anticoagulation irrespective of the results of the thrombophilia workup compared with patients whose VTE was provoked by environmental risk factors. This is because of the 1.9-fold increased risk of recurrence associated with idiopathic VTE that is independent of prothrombotic genetic or biochemical risk factors and sex1 and because of the effectiveness of long-term anticoagulation in reducing the risk of recurrent VTE.2 Although the recent meta-analysis by Ost et al2 did not precisely define an optimal period of anticoagulation, it demonstrates that a period of at least 6 months for such highrisk patients would be warranted and that, even though the prophylactic effect of anticoagulation lessens after cessation of treatment, a clinical benefit persists. It, therefore, is reasonable to suggest that, other than in research settings, physicians reserve thrombophilia testing for situations in which a particular result will influence patient management decisions. The most notable candidate is LA testing for the purpose of diagnosing the antiphospholipid syndrome. It is apparent not only that this syndrome is associated with an increased risk of VTE recurrence on the order of 2- to 9-fold21 but also that these patients will benefit from long-term, possibly lifelong, anticoagulation.5 It is interesting to note that Christiansen et al1 did not include an assessment of the risk that LAs (or antiphospholipid antibodies) contribute to VTE recurrence, and confirmation of the recurrence risk associated with the antiphospholipid syndrome in a large sample would be prudent. If testing for prothrombotic abnormalities after VTE were limited to LA assays, there would be a potential savings, based on the market rate for our thrombophilia panel vs that for LA testing (Quest’s LA Profile, $663.25), of approximately $391 million annually in the United States. There may, however, be special circumstances under which general thrombophilia screening would be indicated.

Because having more than 1 prothrombotic risk factor is associated with a mildly increased, 1.6-fold, risk of VTE recurrence,1 screening may be warranted in cases in which more than 1 thrombophilic condition is considered. Other cases for screening might include a striking family history of thrombosis or the occurrence of thrombosis at unusual sites. In addition, general screening could become more justifiable if the cost of testing were to decline with the emergence of new and less expensive laboratory technologies. In the aforementioned cases, it would be sensible to schedule testing for times when potentially confounding clinical variables would not contribute to overdiagnosis. Among such variables, pregnancy and anticoagulation with warfarin explained the decreased protein C and S results in at least 20% of cases in our series. Despite accounting for these cases, approximately 5% of samples still had a decreased protein S result. It is possible that a portion of these remaining results represent other false-positive results given that the prevalence of protein S deficiency in patients with VTE is approximately 3%3,8; unlike more common prothrombotic risk factors such as activated protein C resistance (due to factor V Leiden), deficiencies of the natural anticoagulants occur with similar prevalence in Caucasian and non-Caucasian populations.8 Likewise, the relatively large proportion of decreased antithrombin results (~6%) is expected to consist predominantly of false-positive results, considering that the estimated prevalence of antithrombin deficiency is about 1%.3,8 Because assays for thrombophilic conditions may be unreliable in the setting of acute thrombosis and inflammation,9,10 these unexpectedly high numbers of potentially protein S– and antithrombin-deficient patients may be explained, in part, by the large proportion of patients (~62%) undergoing screening at the time of acute thrombosis. In fact, 7 of 10 decreased protein S results and 11 of 12 decreased antithrombin results occurred during the acute event. Antithrombin assay results may be falsely positive in other clinical conditions as well, including during heparin use (9 of 12 positive results occurred in heparinized patients). Thus, given the overutilization of thrombophilia testing during the acute thrombotic event, anticoagulation, and pregnancy, it would be more rational and cost-effective to defer testing in these cases. The high frequency of testing during inpatient hospitalization may be due partly to physicians’ anticipation of poor compliance with outpatient follow-up appointments. The appropriate use of laboratory testing in the diagnosis of hypercoagulability was evaluated on a limited basis in McKenzie et al,10 who studied a group of tests similar to those in our panel. In their study, 15 (44%) of 34 patients had an abnormal thrombophilia test result, most tests (55%) were ordered within 6 days of the thrombotic event, and up to one third of tests were ordered while the patient was undergoing anticoagulation. These findings are consistent with ours, particularly in Am J Clin Pathol 2006;126:120-127



125 125


light of the authors’ acknowledgment of the significant number of false-positive results that were included among the 44% abnormal results. Their study10 was, however, subject to significant limitations. Their sample was small (34 patients), and their inclusion criteria required patients to have had only 1 of the set of thrombophilia tests performed. In contrast, we excluded all patients who had less than a complete thrombophilia panel performed (ie, our definition for thrombophilia screening) in an effort to generate complete data regarding the prevalence of thrombophilic abnormalities. In addition, McKenzie et al10 did not report the indication for testing beyond the broad category of thrombosis (eg, no distinction was made for VTE, CVA, IUFD, and other subcategories). Although they documented the inappropriate use of thrombophilia testing during acute thrombosis and anticoagulation, they did not consider the effect of pregnancy on false-positive protein S results or the effect of race and ethnicity on the general appropriateness of testing. Their conclusion that 86% of thrombophilia testing was appropriate was made solely on the basis that the patients had a history of “familial” thrombotic risk factors (eg, younger than 45 years, recurrent thrombosis, and/or family history of thrombosis). This is not an adequate definition of appropriateness in view of the unreliable conditions during which the majority of their tests were ordered, the lack of consideration of the racial and ethnic characteristics of their population, and the recent evidence regarding the limited predictive value of prothrombotic risk factors in VTE recurrence and the benefit of prolonged anticoagulation in these cases.1,2 On the basis of our findings, it is clear that to use medical resources appropriately and improve the level of patient care through the practice of evidence-based medicine, corrective interventions must be taken. An example would be the design of a series of “pop-up” screens to be inserted into the thrombophilia panel ordering algorithm in the clinical information system that would guide ordering physicians to use these screening tests more appropriately. This modification could include a pop-up screen stating that various clinical circumstances can adversely affect the results of thrombophilia testing, including acute thrombosis, inflammation, anticoagulation, and pregnancy, and a reference to relevant medical literature on the predictive value of prothrombotic risk factors and their appropriate clinical utilization. A combination educational and interactive approach also may be used when, as part of the ordering algorithm, interactive prompts would inquire about the relevant, potentially confounding clinical variables so that the interpreting pathologist could more precisely interpret the results and recommend retesting as necessary. Moreover, specialized thrombophilia panels should be created (eg, separate arterial and venous thrombosis panels) in light of the association between different clinical manifestations and thrombophilic causes. We expect that through 126 126


these types of interventions we would be able to more closely attain our goal of clinical best practice. There were several limitations to this study. Although the study was retrospective, we made an effort to eliminate any associated bias by including every panel that had been performed during the research period. It could, however, be argued that a retrospective study has advantages over a prospective study in providing data on real world ordering patterns because physician awareness that they are under scrutiny could modify and bias their behavior for the study period. Another possible issue is the institutional variability with regard to the specific thrombotic risk factors that comprise a panel. However, it is likely that our results would be comparable to those seen in other large urban academic medical centers. Because the prevalence of risk factors varies in different populations, results of screening in different settings may vary as well. Although it would have been desirable to more directly assess the impacts of race and ethnicity in our study, Montefiore’s racial data are unreliable because patients are not queried routinely about their race or ethnic origin at the time of registration.

Conclusions Thrombophilia screening is overutilized in our population considering the low prevalence of prothrombotic abnormalities in non-Caucasians and in cases of stroke. Even when abnormalities are detected, the results are of limited value, regardless of the clinical setting, because the risk of recurrent thrombosis associated with most putative prothrombotic risk factors likely has been overestimated previously and because decisions about long-term anticoagulation after idiopathic VTE are unlikely to be affected by the thrombophilia workup results. The frequent ordering of thrombophilia testing at times when these test results are known to be unreliable represents another component of screening overutilization that is pertinent irrespective of the patient population. These real world findings cast further doubt on the usefulness of thrombophilia panels. Although additional outcome studies similar to that by Christiansen et al1 are needed, we believe that, for the present, thrombophilia screening panels should be reserved for special circumstances as outlined in the preceding text. In light of its substantial expense, testing should be targeted only to factors for which there would be a specific clinical impact and should be performed in the absence of confounding clinical variables. Considerable interventions may be required to shift physician thinking regarding screening given the overutilization of thrombophilia panels in current practice. From the Department of Pathology, Montefiore Medical Center, Bronx, NY.

© American Society for Clinical Pathology


Address reprint requests to Dr Rand: Hematology Laboratory, N8, Silver Zone, Montefiore Medical Center, 111 E 210th St, Bronx, NY 10467. Acknowledgments: We are grateful to Mojgan Raoufi for assistance with data collection, to Susan Romita and Joanne Critchlow for performing the thrombophilia assays, and to the Montefiore physicians and their patients who were involved in this study.

References 1. Christiansen SC, Cannegieter SC, Koster T, et al. Thrombophilia, clinical factors, and recurrent venous thrombotic events. JAMA. 2005;293:2352-2361. 2. Ost D, Tepper J, Mihara H, et al. Duration of anticoagulation following venous thromboembolism. JAMA. 2005;294:706715. 3. Kearon C. Epidemiology of venous thromboembolism. Semin Vasc Med. 2001;1:7-25. 4. Kraaijenhagen RA, in’t Anker PS, Koopman MM, et al. High plasma concentration of factor VIIIc is a major risk factor for venous thromboembolism. Thromb Haemost. 2000;83:5-9. 5. Levine J, Branch DW, Rauch J. The antiphospholipid syndrome. N Engl J Med. 2002;346:752-763. 6. Cattaneo M. Hyperhomocysteinemia, atherosclerosis and thrombosis. Thromb Haemost. 1999;81:165-176. 7. Dowling NF, Austin H, Dilley A, et al. The epidemiology of venous thromboembolism in Caucasians and AfricanAmericans: the GATE study. J Thromb Haemost. 2003;1:80-87. 8. Patel RK, Ford E, Thumpston J, et al. Risk factors for venous thrombosis in the black population. Thromb Haemost. 2003;90:835-838. 9. Bushnell C, Goldstein LB. Screening for hypercoagulable syndromes following stroke. Curr Atheroscler Rep. 2003;5:291298. 10. McKenzie SB, Clare CN, Smith LA, et al. Laboratory test utilization in the diagnosis of hypercoagulability. Clin Lab Sci. 2000;13:215-221.

11. Quest Diagnostics. About Quest Diagnostics Inc. Available at http://www.questdiagnostics.com/brand/index.html. Accessed November 22, 2005. 12. The National Alliance for Thrombosis and Thrombophilia. Research page. Available at http://www.nattinfo.org/research. htm. Accessed November 22, 2005. 13. Miles JS, Miletich JP, Goldhaber SZ, et al. G20210A mutation in the prothrombin gene and the risk of recurrent venous thromboembolism. J Am Coll Cardiol. 2001;37:215-218. 14. Ridker PM, Miletich JP, Stampfer MJ, et al. Factor V Leiden and risks of recurrent idiopathic venous thromboembolism. Circulation. 1995;92:2800-2802. 15. Simioni P, Prandoni P, Lensing AWA, et al. The risk of recurrent venous thromboembolism in patients with an Arg506→Gln mutation in the gene for factor V (factor V Leiden). N Engl J Med. 1997;336:399-403. 16. Simioni P, Prandoni P, Lensing AWA, et al. Risk for subsequent venous thromboembolic complications in carriers of the prothrombin or the factor V gene mutation with a first episode of deep-vein thrombosis. Blood. 2000;96:3329-3333. 17. Kyrle PA, Minar E, Hirschl M, et al. High plasma levels of factor VIII and the risk of recurrent venous thromboembolism. N Engl J Med. 2000;343:457-462. 18. Weltermann A, Eichinger S, Bialonczyk C, et al. The risk of recurrent venous thromboembolism among patients with high factor IX levels. J Thromb Haemost. 2003;1:28-32. 19. den Heijer M, Blom HJ, Gerrits WBJ, et al. Is hyperhomocysteinaemia a risk factor for recurrent venous thrombosis? Lancet. 1995;345:882-885. 20. Eichinger S, Stümpflen A, Hirschl M, et al. Hyperhomocysteinemia is a risk factor of recurrent venous thromboembolism. Thromb Haemost. 1998;80:566-569. 21. Weitz JI, Middeldorp S, Geerts W, et al. Thrombophilia and new anticoagulant drugs. Hematology (Am Soc Hematol Educ Program). 2004:424-438.

Am J Clin Pathol 2006;126:120-127



127 127