CINAHL indicates Cumulative Index to Allied Health and Nursing Literature; FVL, factor V Leiden; VTE, venous thromboembolism.aNumber of excluded articles sums to more than total number because articles could be excluded for more than 1 reason at this level.bForty articles addressed the predictive value of testing and 9 addressed the effect of testing; 3 of these articles addressed both.
Size of the data markers is proportional to study size. CI indicates confidence interval.
Size of the data markers is proportional to study size. CI indicates confidence interval. Data not shown for Simioni et al,42 which had no events in either group.
Jodi B. Segal, Daniel J. Brotman, Alejandro J. Necochea, Ashkan Emadi, Lipika Samal, Lisa M. Wilson, Matthew T. Crim, Eric B. Bass. Predictive Value of Factor V Leiden and Prothrombin G20210A in Adults With Venous Thromboembolism and in Family Members of Those With a MutationA Systematic Review. JAMA. 2009;301(23):2472–2485. doi:10.1001/jama.2009.853
Author Affiliations: Department of Medicine, Johns Hopkins University School of Medicine (Drs Segal, Brotman, Emadi, Samal, and Bass, Ms Wilson, and Mr Crim), Johns Hopkins Evidence-Based Practice Center (Drs Segal and Bass and Ms Wilson), and General Preventive Medicine, Bloomberg School of Public Health (Dr Necochea), Baltimore, Maryland.
Context Testing for genetic risks for venous thromboembolism (VTE) is common, but the safety and utility of such testing need review.
Objectives To define rates of recurrent VTE among adults with VTE with a factor V Leiden (FVL) or prothrombin G20210A mutation compared with those without such mutations; to define rates of VTE among family members of adults with a FVL or prothrombin G20210A mutation according to presence or absence of a mutation; and to assess whether testing adults with VTE for FVL or prothrombin G20210A improves outcomes.
Data Sources We searched MEDLINE, EMBASE, the Cochrane Library, the Cumulative Index to Nursing and Allied Health Literature, and PsycInfo through December 2008.
Study Selection Studies were included if they assessed rates of VTE in individuals with a history of VTE who were tested for FVL or prothrombin G20210A or in family members of individuals with these mutations. Studies assessing the harms and benefits associated with testing were also included.
Data Extraction Two investigators abstracted data and assessed study quality. We pooled the odds of VTE associated with the mutations using random-effects models. We assessed the strength of the evidence using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) criteria.
Results We reviewed 7777 titles and included 46 articles. Heterozygosity (odds ratio [OR], 1.56; 95% confidence interval [CI], 1.14-2.12) and homozygosity (OR, 2.65; 95% CI, 1.2-6.0) for FVL in probands are predictive of recurrent VTE compared with individuals without FVL. Heterozygosity for FVL predicts VTE in family members (OR, 3.5; 95% CI, 2.5-5.0), as does homozygosity for FVL (OR, 18; 95% CI, 7.8-40) compared with family members of adults without FVL. Heterozygosity for prothrombin G20210A is not predictive of recurrent VTE in probands compared with individuals without prothrombin G20210A (OR, 1.45; 95% CI, 0.96-2.2). Evidence is insufficient regarding the predictive value of prothrombin G20210A homozygosity for recurrent VTE and the risk of VTE in family members of individuals with prothrombin G20210A. High-grade evidence supports that anticoagulation reduces recurrent VTE events in probands with either mutation. Low-grade evidence supports that this risk reduction is similar to that in individuals with a history of VTE and without mutations.
Conclusions Patients with FVL are at increased risk of recurrent VTE compared with patients with VTE without this mutation. However, it is unknown whether testing for FVL or prothrombin G20210A improves outcomes in adults with VTE or in family members of those with a mutation.
Quiz Ref IDClinicians test for prothrombotic genetic mutations including factor V Leiden (FVL) (612309.001) and prothrombin G20210A (176930.0009) when treating patients who have had or are at risk of venous thromboembolism (VTE). Factor V Leiden refers to a single base change in the factor V gene (G1691A) that eliminates 1 of its 3 activated protein C cleavage sites. Consequently, factor V is inactivated at a lower rate, leading to more thrombin generation.1 A single FVL allele is present in about 5%, 2.2%, and 1.2% of white, Hispanic and African American populations in the United States.2 The prothrombin (factor II) mutation is the second most common inherited risk factor for VTE. This allele is present in 1.1% of non-Hispanic whites and Mexican Americans and in 0.3% of African Americans.3 The G20210A mutation is associated with an increase in prothrombin levels by approximately 30% in heterozygotes and by 70% in homozygotes. In 2003, the US Food and Drug Administration approved the first DNA-based laboratory tests specifically for FVL and prothrombin G20210A detection (Roche LightCycler Tag-IT Mutation Detection Method; Roche Diagnostics, Indianapolis, Indiana). Testing for these mutations is widely offered in the United States.
The Evaluation of Genomic Applications in Practice and Prevention (EGAPP) initiative was developed by the Centers for Disease Control and Prevention to address the need for timely and objective evidence that allows health care providers and payers, policy makers, and consumers to identify genetic tests that are safe and useful. On their behalf, and as a part of an Agency for Healthcare Research and Quality–supported Evidence-Based Practice Center, we conducted a broad review of testing for FVL and prothrombin G20210A.4 We systematically reviewed the published literature on the predictive value of these genetic tests for future VTE in 2 populations: individuals with a history of VTE (probands) and family members of individuals with VTE and 1 of the mutations. We also assessed whether testing adults with VTE for these mutations improves outcomes.
We performed our search electronically, by hand, and through discussion with experts. We searched 5 databases, MEDLINE (1950 through December 2008), EMBASE (1974 through December 2008), the Cochrane Library (Issue 2, 2008), the Cumulative Index to Nursing and Allied Health Literature (CINAHL; 1982 through December 2008), and PsycInfo, to identify primary, published literature.
Two authors independently reviewed titles and abstracts to identify eligible articles. Abstracts were excluded when both investigators agreed they were not relevant, did not study adults, included no original data, or were not published in English. Full articles underwent independent parallel review to determine their appropriateness. Studies about the predictive value of the genetic tests were excluded if they did not report results separately for individuals with each mutation, did not objectively confirm VTE, studied fewer than 10 probands or 10 family members of individuals with mutations, or did not prospectively study probands. Retrospective studies of family members were acceptable. Studies of pregnant women were excluded. Studies using qualitative methods were excluded if less than 80% of the participants had either mutation or a history of VTE.
A primary reviewer completed all data abstraction forms. A second reviewer with clinical expertise checked the results for completeness and accuracy. Reviewers were not masked to the articles' authors, institutions, or journal.5
We abstracted information on study characteristics, population characteristics, objective of the study, and results. When available, we noted whether the index VTEs were idiopathic events; ie, without identifiable precipitants. Information was entered into the SRS 4.0 database (TrialStat! Corp, Ottawa, Ontario, Canada).
Two investigators independently assessed articles' internal validity based on study setting, inclusion/exclusion criteria, key characteristics of the enrolled participants, losses to follow-up, and funding source. Our quality assessment of qualitative studies included items from the Joanna Briggs Institute Qualitative Assessment and Review Instrument.6 A senior reviewer adjudicated the discrepancies in the quality assessment. Quality assessments are presented descriptively rather than quantitatively to highlight potential sources of bias.
We quantitatively pooled data from studies that assessed the predictive value of the tests when there were sufficient data and the studies were qualitatively homogeneous with respect to key variables. We calculated a pooled estimate of the odds ratio (OR) for VTE in probands and separately in family members. We used a random-effects model with the DerSimonian and Laird method for calculating between-study variance.7 Fixed-effects models yielded very similar results.
We assessed heterogeneity among the studies using a standard χ2 test and a significance level of α≤.10 and with an I2 statistic. The I2 statistic describes variability in effect estimates that is due to heterogeneity rather than chance.8 A value greater than 50% suggests substantial variability.
We used the Duval and Tweedie nonparametric “trim-and-fill” method of accounting for publication bias.9 We also used the Egger test to evaluate the likelihood of missing studies, with P<.05 for the bias estimate as an indicator of potentially missing studies10 Finally, we sequentially removed each study from the calculation of the pooled estimates and recalculated the pooled ORs with a 95% confidence interval (CI). We used STATA (Intercooled, version 9.0, Stata Corp, College Station, Tex).
We graded the quantity, quality, and consistency of the evidence by adapting a grading scheme recommended by the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) Working Group.11 For each question, we assessed the strength of the study designs; the quality and consistency of the body of evidence, including limitations affecting individual study quality; certainty regarding the directness of the observed effects across studies; and the precision and strength of the findings.
The literature search identified 7777 unique titles, including 165 found by hand searching. We included 46 articles (Figure 1).
Twenty-three articles addressed this question.12- 34 Study quality was moderate (Table 1). There was often insufficient description of modifiers of the relationship between the mutation and recurrent VTE, such as age and other thrombotic risk factors. However, in studies using multivariate models, statistical adjustment did not generally attenuate the mutation-specific effect size.12- 16 Funding sources were rarely described.
Thirteen studies described rates of recurrent VTE in probands heterozygous for FVL compared with probands without mutations13- 25 (Table 2). One presented data separately for 2 treatment groups (ximelagatran extended prophylaxis vs placebo).22 One study tested whether FVL homozygosity is associated with a higher rate of VTE recurrence than FVL heterozygosity.26 Overall, 979 heterozygous individuals experienced 161 recurrent VTE events and 3751 mutation-free individuals had 473 thrombotic events. The pooled OR for recurrent thrombosis was 1.56 (95% CI, 1.14-2.12) for individuals heterozygous for FVL compared with patients without the mutation (Figure 2A). The I2 value was 48%, but there was no significant evidence of publication bias. When each study was sequentially removed, the ORs changed little. In the 3 studies reporting annualized event rates in individuals heterozygous for FVL, event rates ranged from 2.8% to 7.5%.16,19,26 Annualized event rates in individuals without FVL ranged from 1.1% to 3.1%.16,19,23 When we included 3 studies that did not specify whether the probands were heterozygous or homozygous for FVL, the OR changed little (OR, 1.61; 95% CI, 1.24-2.10).16,19,27
Seven studies described rates of recurrent VTE in probands homozygous for FVL compared with those without mutations.13,14,17,20,22- 24 Overall, 49 homozygous individuals had 7 recurrent thrombotic events and 2333 mutation-free controls had 225 thrombotic events. The pooled OR was 2.65 (95% CI, 1.18-5.97) (Figure 2B and Table 2). There was no evidence of heterogeneity (I2 = 0%). When each study was sequentially removed from analysis, the ORs changed little.
Among 6 studies that included only patients with idiopathic index thromboses16,18,23,24 or reported results separately for individuals with idiopathic index events,15,28 the pooled OR for recurrence of VTE was 1.17 (95% CI, 0.63-2.18) for individuals with FVL relative to individuals without. One of these 6 did not specify if the individuals were homozygous or heterozygous for FVL28; the others studied heterozygous individuals. A single study described recurrence rates of VTE among individuals with vs without FVL who had a clearly provoked (nonidiopathic) VTE. Results of this study showed an OR of 6.5 (95% CI, 2.5-18).15
Three articles described rates of recurrent VTE in individuals with both the FVL and the prothrombin G20210A mutation compared with mutation-free control patients.15,17,25 Overall, 10 double-heterozygous individuals had 4 recurrent events and 833 mutation-free controls had 95 recurrent thromboses. The pooled OR was 4.81 (95% CI, 0.50-46.3). In 1 study, all 3 double heterozygotes developed recurrent thrombosis.25 Annual incidence rates were not reported in these studies. One additional study described recurrent VTE rates in double-heterozygous individuals but did not include a mutation-free comparison group.29
Eighteen articles examined rates of recurrent VTE in probands with the prothrombin G20210A mutation12- 15,17,18,21- 25,27- 33 (Table 2). Nine articles compared rates of recurrent VTE between probands heterozygous for prothrombin G20210A and mutation-free probands.14,15,17,19,21- 25 The number of probands with a heterozygous prothrombin G20210A mutation ranged from 3 to 58 (median, 26). The number of comparison probands ranged from 72 to 724 (median, 307). Among 281 probands heterozygous for the prothrombin G20210A mutation, 38 recurrent thrombotic events occurred compared with 385 recurrent events among 3355 mutation-free probands. The pooled OR was 1.45 (95% CI, 0.96-2.21) (Figure 2C). The I2 for heterogeneity was 8% and there was no evidence of publication bias. In our sensitivity analysis, removal of 1 study with an OR of 4.015 decreased the pooled OR to 1.30. Annual recurrence rates were reported in 2 studies and were 0% and 2.9%.19,24 One additional study reported a relative rate that did not differ significantly from unity but did not report numbers of recurrent VTE events.32 When 4 studies were included that did not specify whether probands were heterozygous or homozygous, the combined OR was 1.23 (95% CI, 0.87-1.7).27
Only 2 articles quantified event rates in individuals with VTE who were homozygous for prothrombin G20210A compared with mutation-free probands.22,24 There were only 3 probands homozygous for prothrombin G20210A in these studies, and none developed recurrent thrombosis.
Seventeen articles35- 51 assessed first venous thrombosis in family members of individuals with the FVL or prothrombin G20210A mutations (Table 3). Study quality was mixed. Inclusion criteria other than that participants be family members of probands with VTE were not always stated. Three of the 5 prospective studies conducted clinical surveillance for VTE either annually or biannually.35- 37 Studies varied in their descriptions of the population from which participants were identified. Several had little description of eligibility criteria for family members.38- 41 Age and sex were inconsistently described.
Nine studies described rates of VTE in relatives with heterozygosity for FVL compared with relatives without the mutation.35- 37,41- 46 One of these included a single person homozygous for FVL among the 91 studied.43 Two studies did not include a comparison group without a mutation.35,46 Seven studies with comparison groups enrolled 1066 family members heterozygous for FVL and 940 family members without the FVL mutation (Table 4). The ORs across these studies were similar except for 1 small study in which there were no events among 8 heterozygous family members.41 The pooled OR was 3.49 (95% CI, 2.46-4.96) with little heterogeneity between studies (I2 = 0%) (Figure 3A).
Removal of the study by Couturaud et al37 increased the pooled OR to just over 4.0. In this study, the annualized event rate among family members heterozygous for FVL was 0.36% (95% CI, 0.24%-0.49%), lower than the rates in other studies.36,43,45 It is unclear why VTE rates were lower in this high-quality study. There was no evidence of publication bias. The inclusion of 5 studies in which relatives were not differentiated as to whether they were homozygous or heterozygous for FVL did not substantially change the OR for VTE (OR, 3.30; 95% CI, 2.49-4.39).38- 40,47- 49
Six studies described results for 48 relatives who were homozygous for FVL compared with family members without the mutation.36,37,41,42,44,45 One additional study had no comparison group35 (Table 3). Annualized rates were described for only 3 studies.35- 37 The pooled OR was 18 (95% CI, 7.8-40). The I2 for heterogeneity was 0% (Figure 3B6). One small study was excluded from pooled analyses because it had only 1 individual in each group.42 Removal of 1 study44 decreased the pooled OR to 16. In this study, 5 of the 6 relatives who were homozygous for FVL had venous thromboembolic events.44 The annualized rate was not described for the mutation-free comparison group. There was no evidence of publication bias.
Four studies described venous thromboembolic events in 59 family members with double heterozygosity37,41,42,50 compared with 674 family members without mutations. One study observed no events in either group42 (Table 4). The pooled OR was 6.7 (95% CI, 2.9-16), with an I2 value of 0%. The OR for family members with double heterozygosity in 1 study was higher than in the remaining studies (OR, 8.0; 95% CI, 2.8-23).50 This may be due to low event rates among family members without mutations (0.07% per year). This compares with somewhat higher rates in the mutation-free relatives in the other studies, ranging from 0% to 0.34%.34- 37,39,42,44,46- 48 Analysis suggested potential publication bias, with a paucity of small studies reporting large effect sizes.
Family members who were heterozygous for prothrombin G20210A were compared with family members without the mutation in 3 studies37,41,51 (Table 4). In 1 study, the age-adjusted relative event rate was 3.4 (95% CI, 0.2-56) compared with family members without the mutation.51 Remaining studies were small.37,41 Pooling these small studies yielded an OR for events of 1.89 (95% CI, 0.35-10), with an I2 of 0%, suggesting that family members who are heterozygous for this mutation do not have an increased odds of thrombosis.
A single study with a comparison group described 1 relative who was homozygous for prothrombin G20210A.41 This individual was identified among 44 family members from 4 families. None reported a venous thromboembolic event.
Three studies described age-specific event rates.43,45,49 One additional study described the age of the individuals at the time of their first thrombosis among family members with and without mutations.44 Middeldorp et al45 described age-specific relative event rates for family members who were homozygous or heterozygous for FVL mutation or who had both FVL and prothrombin G20210A mutations compared with family members without mutations. The relative risk associated with a mutation was highest in the youngest patients (aged 15-30 years; relative rate of approximately 15) and was lowest in the older patients (relative rate of approximately 2.5). However, other similarly designed studies did not demonstrate this age × mutation interaction.43,49
No studies directly addressed the effect of testing on outcomes. Four studies described VTE recurrence rates during anticoagulation among probands with FVL or prothrombin G20210A.22,24,29,52 This is relevant to our question if clinicians alter anticoagulation management based on a positive test result. Three of these studies consisted of individuals participating in randomized controlled trials.22,24,52 The fourth was within a prospective cohort study.29 Two studies investigated the effect of warfarin on recurrence rates24,52 and 1 the effect of ximelagatran,22 and 1 did not specify the treatment29 (Table 5).
Ridker et al52 assessed thromboembolism recurrence rates among individuals with FVL or prothrombin G20210A treated with low-intensity warfarin or placebo. Of 77 patients with FVL or prothrombin G20210A in the placebo group, 14 had recurrences (8.6 events per 100 person-years) compared with 3 of 66 assigned to the warfarin group (2.2 events per 100 person-years). Low-intensity warfarin reduced the rate of recurrence among patients with thrombophilia by 75% (hazard ratio, 0.25; 95% CI, 0.07-0.85). This risk reduction was not significantly different than the 58% reduction seen among patients without either mutation (hazard ratio, 0.42; 95% CI, 0.20-0.86; P = .51 for interaction).
Kearon et al24 studied whether VTE recurrence rates differed among individuals with vs without thrombophilic defects receiving low-intensity (international normalized ratio goal, 1.5-2.0) or conventional (international normalized ratio, goal, 2.0-3.0) warfarin therapy. Among 171 patients with FVL, 3 had recurrences. None occurred in the 60 patients with prothrombin G20210A. The recurrence rate for patients with FVL receiving low- or conventional-intensity warfarin therapy was 0.8% per year (95% CI, 0.2-2.2). This rate was not statistically different from the rate among participants without mutations or coagulopathies on anticoagulants (hazard ratio, 0.7; 95% CI, 0.2-2.6).
Wåhlander et al22 assessed the risk of VTE recurrence in probands receiving ximelagatran, an oral direct thrombin inhibitor (not presently available) vs placebo. Among 111 FVL carriers in the treatment group, 2 had recurrent VTE compared with 16 among 125 with mutations assigned to placebo. This difference in recurrence rates was statistically significant (hazard ratio <0.25; P value not reported). The relative reduction in recurrence with treatment was similar for individuals with vs without FVL (P = .92 for interaction). Results were similar for individuals with prothrombin G20210A (P = .98 for interaction).
Vossen et al29 studied the recurrence rates of VTE among probands belonging to families with thrombophilia according to whether they were receiving long-term anticoagulation. Of 304 patients, 124 were receiving long-term anticoagulation and 180 were not. Fewer FVL carriers were in the long-term anticoagulation group than in the comparison group (13% and 44%, respectively). Of 79 patients with FVL who did not receive long-term anticoagulation, 13 had recurrences during 366 person-years (incidence rate, 3.5% per year; 95% CI, 1.9-6.1). Among the 13 FVL carriers who received anticoagulation, none had recurrences during 43 person-years. There were important differences in the baseline characteristics and duration of follow-up between the 2 groups and few details about the anticoagulation regimens.
Four studies addressed how probands' and family members' knowledge, behaviors and health care experiences were affected by their being tested for FVL or prothrombin G20210A (Table 6).53- 57 Heshka et al53 surveyed the perception of VTE risk and changes in behavior following testing for FVL or prothrombin G20210A among first-degree relatives of probands. Kaptein et al54 investigated whether the type of thrombophilic mutation and history of VTE affect perception of risk and worry among probands or their relatives with FVL compared with individuals with other thrombophilic mutations. Bank et al57 conducted a qualitative study of asymptomatic relatives of probands with FVL to assess their experience with testing and how results affected their daily lives. Saukko et al55,56 assessed the level of understanding of the testing process and the implications of the results among probands and relatives referred for FVL. Testing had minimal impact on knowledge or behavior and was associated with no more than modest distress.
Quiz Ref IDModerate evidence supports that both homozygosity and heterozygosity for FVL are predictive of recurrent VTE among individuals who have had a prior VTE. Homozygosity for prothrombin G20210A is a rare genotype and its association with recurrent venous thrombosis in probands is little known. Moderate evidence supports that heterozygosity for prothrombin G20210A in probands is not predictive of recurrent VTE. Evidence is insufficient about double heterozygosity in probands.
Quiz Ref IDHigh-grade evidence supports that homozygosity for FVL in family members predicts a higher rate of incident VTE. Moderate evidence supports that heterozygosity for FVL predicts a higher rate of incident VTE. Evidence is insufficient about risk of VTE among family members who are homozygous or heterozygous for prothrombin G20210A. Low-grade evidence supports that double heterozygosity among relatives is associated with higher risk of VTE.
Quiz Ref IDThere is no direct evidence that testing for these mutations, and the resultant management, reduces VTE related-outcomes in individuals who have had VTE or in the probands' family members who have been tested. We identified studies demonstrating that treatment reduces recurrent events in patients with FVL or prothrombin G20210A; however, the magnitude of this relative reduction is comparable with that seen in individuals without mutations. This suggests that other nongenetic factors may be as important as the presence or absence of the FVL or prothrombin G20210A mutation in determining the risk of recurrence and the absolute magnitude of benefit conferred by anticoagulation. This may be especially so in individuals with idiopathic VTE. We conclude that the incremental value of testing individuals with VTE for these mutations is uncertain. The literature does not conclusively show that testing individuals with VTE or their family members for FVL or prothrombin G20210A confers other harms or benefits. If testing is done in conjunction with education, it may increase knowledge about risk factors for VTE.
This review has limitations, including the use of only English-language articles. Furthermore, we included observational studies knowing that confounding is a possibility, particularly if there were factors that might have been associated with both the exposure (mutation status) and the outcome (thrombosis). We were limited to the covariates reported in each study, and the studies did not often present patient-level data that would allow testing for potential confounders. For the family studies, the expected confounder would be genes that are coinherited with the mutation of interest. For both the proband studies and family studies, there might be nongenetic factors such as more aggressive surveillance for events that might bias the estimate of the association between the mutation and outcome. In the studies that did report group-level data (eg, mean duration of anticoagulation or mean age at diagnosis by mutation status), we did not see any consistent evidence that these factors affected the outcomes, but we cannot definitively exclude confounding.
Studies are needed to measure how practice actually changes in response to results from FVL or prothrombin G20210A testing and whether that improves patient outcomes. Future studies should focus on whether management decisions based on testing results affect the recurrence rates (as well as complication rates) in carriers of each of these mutations. Quiz Ref IDIdeally, future trials would randomize patients with thrombosis or family members of individuals with mutations to a test group or a no-test group, and individuals would be managed by their physicians based on the results of the testing, perhaps with evidence-based guidance. Studies of clinical validity should include event rates over time (and relative rates of recurrence between specified groups) rather than just the number of events. Studies should consistently differentiate between heterozygosity and homozygosity. By examining specific subsets of patients, it may be possible to clarify whether there are any interactions between mutation status and clinical variables in predicting recurrence. The data presented in the studies were insufficient to definitively exclude the effect of confounders on the relationship between the mutations and the outcome.
Uncertainty remains about the magnitude of risk for family members with mutations given the very wide CIs surrounding the ORs and the small number of studies the reported event rates (rather than counts). The studies that we included were exclusively studies of European populations. Future research would be appropriate in white populations outside of Europe or in other populations with appreciable frequencies of mutations.
Corresponding Author: Jodi B. Segal, MD, MPH, 1830 E Monument St, Room 8047, Baltimore, MD 21287 (firstname.lastname@example.org).
Author Contributions: Dr Segal had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Segal, Brotman, Necochea, Emadi, Wilson, Crim, Bass.
Acquisition of data: Segal, Brotman, Necochea, Emadi, Samal, Wilson.
Analysis and interpretation of data: Segal, Brotman, Necochea, Emadi, Bass.
Drafting of the manuscript: Segal, Brotman, Necochea, Emadi, Samal.
Critical revision of the manuscript for important intellectual content: Segal, Brotman, Necochea, Emadi, Wilson, Crim, Bass.
Statistical analysis: Segal, Necochea.
Obtained funding: Segal, Bass.
Administrative, technical, or material support: Brotman, Emadi, Wilson, Crim.
Study supervision: Segal, Bass.
Financial Disclosures: None reported.
Funding/Support: This article was prepared under contract HHSSA 290-2007-10061-I Task Order No. 4 for the Agency for Healthcare Research and Quality (AHRQ). The report on which this article is based was requested and funded by the Centers for Disease Control and Prevention for the EGAPP working group.
Role of the Sponsors: The funding agency reviewed the final work plan prior to the project's initiation; reviewed the manuscript for the purpose of approval to assert copyright; and reviewed the acknowledgement of funding statement and the disclaimer. Aside from these exceptions, the funding source had no role in the design and conduct of the study, in the collection, analysis, and interpretation of the data, or in the preparation, review, or approval of the manuscript.
Disclaimer: Statements in the report should not be construed as endorsement by the Agency for Healthcare Research and Quality, the Centers for Disease Control and Prevention, or the US Department of Health and Human Services.
Additional Contributions: We gratefully acknowledge the editorial assistance of Renee Wilson, MSc, Olaide Odelola, MBBS, MPH, and Ritu Sharma, BS, Johns Hopkins University School of Medicine (who all received only regular salary compensation for their contributions), and the input of all of our technical experts.
Additional Information: This report is being published simultaneously by AHRQ and is available at http://www.ahrq.gov/clinic/tp/fvltp.htm.