Racial and Regional Differences in Venous Thromboembolism in the United States in 3 CohortsCLINICAL PERSPECTIVE
Background—Blacks are thought to have a higher risk of venous thromboembolism (VTE) than whites. However, prior studies are limited to administrative databases that lack specific information on VTE risk factors or have limited geographic scope.
Methods and Results—We ascertained VTE from 3 prospective studies: the Atherosclerosis Risk in Communities Study (ARIC), the Cardiovascular Health Study (CHS), and the Reasons for Geographic and Racial Differences in Stroke study (REGARDS). We tested the association of race with VTE using Cox proportional hazard models adjusted for VTE risk factors. Over 438 090 person-years, 916 incident VTE events (302 in blacks) occurred in 51 149 individuals (17 318 blacks) who were followed up. In risk factor–adjusted models, blacks had a higher rate of VTE than whites in the CHS (hazard ratio, 1.81; 95% confidence interval, 1.20–2.73) but not ARIC (hazard ratio, 1.21; 95% confidence interval, 0.96–1.54). In REGARDS, there was a significant region-by-race interaction (P=0.01): Blacks in the Southeast had a significantly higher rate of VTE than blacks in the rest of the United States (hazard ratio, 1.63; 95% confidence interval, 1.08–2.48) that was not seen in whites (hazard ratio, 0.83; 95% confidence interval, 0.61–1.14).
Conclusions—The association of race with VTE differed in each cohort, which may reflect the different time periods of the studies or different regional rates of VTE. Further studies of environmental and genetic risk factors for VTE are needed to determine which underlie racial and perhaps regional differences in VTE.
Venous thromboembolism (VTE), consisting of pulmonary embolism (PE) and deep venous thrombosis (DVT), is a common cardiovascular disease, affecting >300 000 individuals annually in the United States with 100 000 fatalities per year.1 Many studies suggest that black Americans have higher rates of VTE than white Americans.2 Reasons for these potential racial differences are unclear, with risk factors such as obesity,3 diabetes mellitus,4 and elevated factor VIII5 being more common in blacks and genetic polymorphisms such as factor V Leiden and the prothrombin gene 20210A mutation more common in whites.2
Editorial see p 1463
Clinical Perspective on p 1509
Prior studies on race and VTE in the United States have been limited in that they examined administrative databases without validation of VTE events,6–8 had limited numbers of blacks, were from discrete geographic areas, or excluded outpatient-treated DVTs.9 Furthermore, many studies were not able to evaluate whether differences in VTE risk factors explained any race association.6–8 To determine the association of race with VTE and to evaluate whether conventional VTE risk factors might mediate any observed differences, we assessed VTE incidence in blacks and whites in 3 large cohorts: the Cardiovascular Health Study (CHS), the Atherosclerosis Risk in Communities Study (ARIC), and the Reasons for Geographic and Racial Differences in Stroke study (REGARDS).10–13 Together, these studies have followed up 51 149 individuals over 439 090 person-years and include 17 318 blacks. They offer a unique opportunity to study the association of race with VTE and, in the case of REGARDS, to evaluate the association of region of residence with VTE in the United States. Our goal was to study the association of race with VTE risk in these 3 cohorts and whether common VTE risk factors affected any race association.
VTE events were ascertained in 3 longitudinal cohorts designed to study the causes and consequences of vascular disease (Table 1 and Table I in the online-only Data Supplement). ARIC12 recruited 15 792 individuals (4266 blacks) 45 to 64 years old in 1987 to 89 from 4 field centers: Forsyth County, North Carolina; Washington County, Maryland; suburban Minneapolis, MN; and Jackson, MS. CHS13 recruited 5201 individuals ≥65 years of age in 1989 to 90 and an additional 687 black men and women in 1991 to 92 from 4 field centers: Forsyth County, North Carolina; Sacramento County, California; Washington County, Maryland; and Pittsburgh, PA (924 blacks). The study of VTE in CHS and ARIC is called the Longitudinal Investigation of Thromboembolism Etiology (LITE) study. The methods for the LITE study, including VTE case ascertainment, have been described in detail elsewhere.10
REGARDS recruited 30 239 black and white individuals ≥45 years of age between 2003 and 2007 in the contiguous United States, oversampling blacks and individuals living in the Southeast (Alabama, Arkansas, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, and Tennessee).11 Exclusion criteria included self-reported race other than white or black, inability to converse in English, cognitive impairment as judged by the telephone interviewer, residence in or on the waiting list for a nursing home, or active cancer or current treatment for cancer. After enrollment, medical history (including self-reports of prior VTE events) and risk factors were assessed via computer-assisted telephone interview followed by an in-home visit, which included anthropomorphic and blood pressure measures, medication ascertainment, and phlebotomy (Examination Management Systems Inc, Irving, TX).
Participants in ARIC, CHS, and REGARDS gave written informed consent. This study was approved by the Institutional Review boards of all participating institutions.
In CHS and ARIC, VTE (consisting of DVT and PE) events were captured by review of hospital discharge codes and verified by 2 physicians (A.R.F. and M.C.). Briefly, participants in ARIC were followed up by clinic visits every 3 years and annual telephone calls. Further hospitalizations not captured by other methods were obtained from surveillance of community hospitals. In CHS, participants were followed up by alternating telephone calls and clinic visits every 6 months. Hospitalizations were also identified through Medicare records. For all hospitalizations, hospital discharge codes were used to identify possible cases of thrombosis. By design, hospital record review was similar to that reported in REGARDS below. A detailed description of event ascertainment has previously been published.10 VTE events were captured through December 31, 2001, in CHS and December 31, 2005, in ARIC.
REGARDS participants were contacted by telephone every 6 months, and deaths were ascertained by proxy report and through periodic searches of the National Death Index.14 VTE events were captured in 4 ways. First, at each telephone call, participants or their proxies were asked for an update on the participant’s medical status and for the reasons for any hospitalizations. A research nurse reviewed the text recorded for each reported hospitalization through February 2010. Any report of a blood clot in the legs, arms, or lungs was considered a potential case for physician review. Second, a telephone interview was developed and administered between February 2010 and February 2011 to ascertain participant-reported VTE events back to baseline. Similar questionnaires in epidemiological studies have 98% specificity and >70% sensitivity for ascertaining VTE.15 Third, the reasons for all deaths were reviewed by the use of any available data (National Death Index determination of death, exit interview with proxy/next of kin, or records from last hospital stay). Fourth, VTE events discovered from a review of other events (stroke and coronary heart disease) were abstracted. On the basis of all available information, we retrieved medical records up to 1 year before and 1 year after potential events. Retrieved records were used to help guide further record retrieval if they did not contain the primary VTE event. Primary inpatient and outpatient records, including history and physical examinations, discharge summaries, imaging reports, and outpatient notes, were retrieved using up to 3 attempts. If after review by a research nurse and confirmation by a physician (N.A.Z.) it was ascertained that no VTE occurred and no workup for VTE occurred, the record was closed as a nonevent. If separate events were judged by the research nurse and the physician to be 1 event (ie, a DVT and PE that occurred on the same day) or events were captured via >1 mechanism (ie, through reviewing reasons for hospitalizations and through the telephone interview), the events were consolidated. Each potential event was reviewed by 2 of 3 physicians (N.A.Z. reviewed all events; M.C. and A.R.F. each reviewed 60% of events). Major disagreements, defined as a disagreement in the level of evidence of a VTE event or whether the event was provoked or unprovoked, were adjudicated by blind review by the third reviewer. Telephone conferences were used in cases when all 3 reviewers disagreed. Minor disagreements (such as date of a VTE event) were resolved by a physician (N.A.Z.) and the research nurse.
Consistent definitions were used in REGARDS and LITE to define VTE events. PE was considered thrombus in the pulmonary arteries, and DVT was considered thrombus of the deep veins of the legs or arms (including distal veins). Provoked VTE was defined as a VTE event preceded within 90 days by major trauma, surgery, or marked immobility or associated with active cancer or chemotherapy. All other events were considered unprovoked. Definite VTE events required an autopsy, unambiguous imaging, or a healthcare provider’s description of positive imaging. Probable VTE events required a high clinical suspicion but without a record of definitive radiology evidence of VTE. Baseline VTE was defined as a self-reported history of PE or DVT before enrollment.
Body mass index was defined as weight in kilograms divided by the square of height in meters. Estimated glomerular filtration rate was defined with the Chronic Kidney Disease Collaboration equation.16 Diabetes mellitus was defined as fasting glucose ≥126 mg/dL, nonfasting glucose ≥200 mg/dL, or participant report of diabetes mellitus or the use of hypoglycemic medication. Hypertension was defined as blood pressure >140/90 mm Hg or self-report of current treatment for hypertension. Because of the differing years of recruitment and because participants only selected ranges for income, income was divided into cohort-specific percentiles by cohort with a category for refused (≤20th, 21st–50th, 51st–75th, and >75th percentile). Race was defined from participant self-report.
As a result of differences in methodology, follow-up, and start date, analyses were stratified by cohort. Differences between blacks and whites were tested by the use of t tests, Wilcoxon rank-sum tests, and χ2 tests of associations as appropriate. After the exclusion of participants with self-reported baseline VTE, Poisson regression was used to estimate VTE incidence rates accounting for age, sex, and race. Cox proportional hazard models tested the association of race with VTE, excluding individuals with baseline VTE (Table 1). Interaction terms of race with age, sex, and region (in REGARDS) were assessed, and values of P<0.10 were considered significant. Because of a significant region-by-race interaction, analyses in REGARDS were presented stratified by race or region. Sensitivity analyses were done by determining the probability that nonretrieved records in REGARDS represented VTE events and including the probabilities in analyses. The probability that a nonretrieved record would be a VTE was calculated on the basis of the percent of retrieved records that were VTE (stratified by region and race). Follow-up time for the probable events was determined from a random uniform distribution, the distribution most closely representing the temporal distribution of validated VTE events in REGARDS (Tables II and III in the online-only Data Supplement). Analyses were performed with SAS version 9.3 (SAS Institute Inc, Cary, NC).
The cohort characteristics by race are reported in Table 1 and Table I in the online-only Data Supplement. Briefly, ARIC had the lowest mean age (54 years), CHS had the highest mean age (73 years), and REGARDS had an intermediate mean age (65 years). The prevalence of obesity was lowest in CHS (20%) and highest in REGARDS (38%), with the highest prevalence of diabetes mellitus in REGARDS. In all cohorts, blacks had a higher body mass index than whites, had greater prevalences of diabetes mellitus and hypertension, were less likely to be in the top 25th percentile of income, and were more likely to have lower levels of education (Table 1).
VTE event ascertainment has previously been reported in detail for ARIC and CHS.10 In REGARDS, 936 potential VTE events were identified (Figure). Among the 785 events for which records were requested (after consolidation of duplicate events), 624 records (79.5%) were successfully retrieved, and 471 (75.5%) were reviewed by physicians. Among blacks, 231 of 321 records (72%) were retrieved, and among whites, 393 of 464 records (85%) were retrieved. Among the 471 records reviewed, there were 379 VTE events in 332 individuals; 268 were first-time VTE events in those not reporting VTE at baseline (123 in blacks). Retrieval rates by race and region in REGARDS and the percent of retrieved records that became cases are presented in Tables II and III in the online-only Data Supplement. Blacks had lower record retrieval than whites, although there were no differences by region. In CHS, there were 172 validated incident VTE events (37 in blacks), and in ARIC, there were 476 validated incident VTE events (163 in blacks; Table 2). In all cohorts, blacks had a higher percentage of VTEs that were DVTs than whites but a similar percentage of VTEs that were PEs, except in ARIC in which blacks had a lower percent of PEs than whites (26% versus 41%). The percent of VTEs that were provoked was similar in blacks and whites in each cohort (Table 2).
Table 3 presents VTE rates for ARIC, CHS, and REGARDS in blacks and whites normalizing to the mean age and sex distribution of each cohort. Overall, blacks had a higher incidence of VTE than whites in ARIC, in CHS, and in the Southeast in REGARDS but not in REGARDS outside the Southeast. These differences were statistically significant in CHS and ARIC but not in REGARDS. These patterns were similar for DVT. In contrast, PE incidence was not higher in blacks than whites in ARIC, CHS, or REGARDS. The rate of provoked VTE was higher in blacks than whites in CHS but not in ARIC or REGARDS.
Table 4 presents a series of sequentially adjusted Cox proportional hazard models demonstrating the association of black versus white race with incident VTE in each of the cohorts. In a model adjusted for age and sex, blacks had a greater risk of VTE than whites in ARIC and CHS. In REGARDS, blacks had a nonsignificantly increased hazard of VTE compared with whites in the Southeast (hazard ratio [HR], 1.33; 95% confidence interval [CI], 0.94–1.87) and a nonsignificantly decreased hazard of VTE than whites in the rest of the nation (HR, 0.78; 95% CI, 0.54–1.12; P for interaction=0.03). Adjusting for body mass index reduced the HR for race by >10% in ARIC, in CHS, and in participants living in the Southeast in REGARDS. Adjusting for body mass index, hypertension, diabetes mellitus, kidney disease, and baseline warfarin use decreased the association of race with VTE in ARIC (HR from 1.61 to 1.25) but had little effect on the HR in CHS (HR from 1.82 to 1.72) or REGARDS in the Southeast (HR from 1.33–1.34). Adjustment for socioeconomic factors had little effect on the association of race with VTE in CHS but decreased the black versus white HR in ARIC and in participants in the Southeast in REGARDS. In a final model including all risk factors, the association of race with VTE was no longer significant in ARIC (HR, 1.21; 95% CI, 0.96–1.54) and was little changed in CHS (HR, 1.81; 95% CI, 1.20–2.73) and REGARDS (outside the Southeast: HR, 0.68; 95% CI, 0.46–1.02; in the Southeast: HR, 1.34; 95% CI, 0.93–1.94). In REGARDS, a significant regional difference in the association of race with VTE was found in all models (all P for interactions ≤0.03). Among blacks in REGARDS, the HR of VTE for living in the Southeast versus elsewhere was 1.63 (95% CI, 1.08–2.48), but among whites, living in the Southeast was not associated with increased risk (Table 5).
Tables 5 and 6 break down the results by VTE type (PE or DVT, unprovoked or provoked). In fully adjusted models, blacks had a higher risk of DVT than whites in ARIC, in CHS, and in the Southeast in REGARDS but not outside the Southeast in REGARDS. After full multivariable adjustment, there was little evidence of an increased risk of PE for blacks versus whites in ARIC, CHS, or REGARDS. There was no association of black race with provoked or unprovoked VTE in ARIC, but the HR for blacks versus whites remained elevated in CHS for both unprovoked and provoked VTE. In REGARDS, again, there was evidence of a regional interaction, with blacks in the Southeast having a higher rate of both DVT and PE than whites in the Southeast but not outside the Southeast.
Sensitivity analyses excluding those on baseline warfarin or including only VTE events ascertained by the computer-assisted telephone interview in REGARDS did not materially affect the conclusions (data not shown). Table IV in the online-only Data Supplement presents one scenario of the potential effect that differential record retrieval by race or region may have had in REGARDS, which assumes that had all records been obtained, the percent validated as VTEs would have been the same for the percent actually validated among received records. From this analysis, we estimated that we missed ≈70 VTE events (44 in blacks) because of incomplete record retrieval. The region and race differences in REGARDS were preserved in a sensitivity analysis accounting for these potentially missing events, and the nearly significant inverse association of black race (versus white) with VTE outside the Southeast disappeared.
In 3 large US cohort studies including 51 149 individuals and 916 VTE events, we found at most a modest association of race with risk of VTE, particularly once comorbid conditions and socioeconomic status were accounted for. The studied cohorts were recruited with different methods spanning 20 years (from 1987 through 2007) and revealed different results for the association of race with VTE. In ARIC, blacks had a higher risk of VTE that was attenuated by VTE risk factors; in CHS, blacks had a higher risk of VTE that was not attenuated by risk factors; and in REGARDS, there was a significant region-by-race interaction whereby blacks in the Southeast were at significantly higher risk of VTE relative to blacks outside the Southeast, whereas in the rest of the country, there was no evidence that risk of VTE varied by race. Furthermore, no study demonstrated a racial association with PE alone.
Race in the United States is as much a social construct as a marker of continental origin, with self-identified blacks having on average 20% European ancestry but with great variation both within populations and between different locations in the United States.17 There are few studies in the United States in which the risk of VTE can be directly compared between blacks and whites. In hospital discharge registries from California7,18 and the National Hospital Discharge Survey,8 blacks had a higher risk of VTE compared with whites (relative risk, 1.37 and 1.18, respectively). These analyses, although powerful, have limitations, including potential misclassification of events and race, with reliance on discharge codes for events and on census data to classify the racial characteristics of the population. International Classification of Diseases, Ninth Revision coding of VTE is challenging; in 1 report, >20% of hospitalized International Classification of Diseases, Ninth Revision–reported VTEs were miscoded.19 The assumption with administrative databases is that miscoding is consistent by race and region, which may not be the case.20 Blacks also have a 50% greater risk of death resulting from PE compared with whites in the United States, but whether this results from an increased rate of VTE or a higher case fatality rate is not known.21–24 Apart from an earlier report from the LITE,9 no other national cohorts with physician-validated VTE events have reported racial differences in VTE in the United States.
Further discussion of the differences between ARIC, CHS, and REGARDS is warranted. Our original intent was to perform a pooled analysis; however, these cohorts had different recruitment ages, geographic scopes, and enrollment years, decreasing the scientific appropriateness of a pooled analysis. Differences in the race association between CHS and ARIC could be due to age differences; however, REGARDS encompassed the entire age spectrum of ARIC and CHS. In terms of geographic scope, REGARDS is the only study to include a substantial number of blacks outside the Southeast. The majority of blacks in ARIC resided in Jackson, MS, and Forsyth County, North Carolina; the majority of blacks in CHS resided in Forsyth County, North Carolina, and to a lesser extent in Pittsburgh, PA. No ARIC whites came from Jackson. In ARIC and CHS, there were only 2 and 21 VTE events, respectively, in blacks outside the Southeast, precluding an analysis by region in these cohorts. Although individual-level national data on VTE in the United States do not exist, the effect of race seems modest at best in an analysis from the National Hospital Discharge Survey, with blacks having an 18% higher risk than whites.8 Further data from the Centers for Disease Control show that blacks were at greater risk of PE death relative to whites in the Northeast, Midwest, and South, but not the West (see Table V in the online-only Data Supplement).25 Reasons for these differences are unclear but may be due to true biological differences and differences in the prevalence of comorbidities that may lead to VTE, result in complications among VTE patients that predispose to death, or may relate to access to medical care, quality of medical care, and quality of medical reporting. Further confounding the observations in the current studies, differences in VTE event rates among the studies may be due to secular trends in the evaluation of suspected VTE. The diagnosis of VTE has shifted from relying on ventilation-perfusion scans for PE to computed tomography angiography and from inpatient to outpatient treatment of DVT.26 If these changes in medical care differed by age, race, or region, this could have contributed to the different patterns observed here.
Possible limitations of our study include self-identification of race, lack of generalizability of the cohorts to the population of the United States, although the national reach of REGARDS mitigates this somewhat, and issues of underascertainment or biased ascertainment for VTE. In each cohort, race was self-defined and thus represents both a social and a genetic construct. Other phenotypes such as coronary artery calcium vary by genetic origin even within racial groups, and by using self-identified race, we may have missed these associations.17 LITE did not capture out-of-hospital VTE deaths (there are likely few) and outpatient treatment of VTE. During the time period of LITE, the proportion of DVT events treated in the outpatient setting was small because evidence for the safety of this practice did not emerge until the late 1990s.27 In REGARDS, there were no discrete field centers or local hospitals to search for discharges, and case ascertainment relied predominantly on participant report. Therefore, cases were missed, but VTE events treated in the outpatient setting were sought. Despite efforts, record retrieval was not 100%, and fewer records of blacks than whites were obtained. However, although record retrieval differed by race in REGARDS, it did not differ by region. When we accounted for potential cases resulting from missing records, blacks in the Southeast remained at greater risk than whites, whereas for the rest of the country, there was no evidence of a relation (Tables II–IV in the online-only Data Supplement). Furthermore, as shown in in the Centers for Disease Control data, there are clear regional and racial differences in PE mortality in the United States, demonstrating the need to define geography and race when studying racial differences in VTE (Table V in the online-only Data Supplement).25
We present the most detailed examination yet of the association of race with VTE in the United States. In contrast to prior studies, we were able to study the association of race with VTE using validated VTE events and individual-level data over a long time period. The differences seen by cohort may represent secular trends in the diagnosis or incidence of VTE, may reflect an unrecognized race-by-region interaction in which blacks in the Southeast have higher rates of VTE than whites and blacks outside the Southeast, or may be an artifact of bias. Whether differences in VTE by race are compared on a regional, national, or global scale will greatly influence associations of race on VTE. Study limitations necessitate caution in the interpretation of the results, but these results highlight the need for further studies of VTE in the United States in diverse geographic and racial populations.
We thank the staff and participants of ARIC, CHS, and REGARDS for their important contributions. The Executive Committee or Publications Committee of ARIC, CHS, and REGARDS reviewed and approved this manuscript for publication.
Sources of Funding
The ARIC study is carried out as a collaborative study supported by National Heart, Lung, and Blood Institute contracts (HHSN268201100005C, HHSN268201100006C, HHSN268201100007C, HHSN268201100008C, HHSN268201100009C, HHSN268201100010C, HHSN268201100011C, and HHSN268201100012C). CHS was supported by contracts HHSN268201200036C, HHSN268200800007C, N01 HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, and N01HC85086 and grant HL080295 from the National Heart, Lung, and Blood Institute, with additional contribution from the National Institute of Neurological Disorders and Stroke. Additional support was provided by AG023629 from the National Institute on Aging. LITE was funded by grant R01-HL59367 from the National Heart, Lung, and Blood Institute. REGARDS was funded by cooperative agreement NS 041588 from the National Institute of Neurological Disorders and Stroke, with additional funding from the American Recovery and Reinvestment Act grant RC1HL099460 from the National Heart, Lung, and Blood Institute.
The online-only Data Supplement is available with this article at http://circ.ahajournals.org/lookup/suppl/doi:10.1161/CIRCULATIONAHA.113.006472/-/DC1.
- Received September 24, 2013.
- Accepted January 13, 2014.
- © 2014 American Heart Association, Inc.
- 1.↵US Public Health Service, Office of the Surgeon General. The Surgeon General’s Call to Action to Prevent Deep Vein Thrombosis and Pulmonary Embolism. Rockville, MD: US Public Health Service, Office of the Surgeon General; 2008.
- 3.↵Quickstats: from the National Center for Health Statistics: prevalence of obesity* among adults aged ≥20 years, by race/ethnicity and sex: National Health and Nutrition Examination Survey, United States, 2009–2010. JAMA. 2012;308:1084–1084.
- Voeks JH,
- McClure LA,
- Go RC,
- Prineas RJ,
- Cushman M,
- Kissela BM,
- Roseman JM
- 12.↵The Atherosclerosis Risk In Communities (ARIC) Study: design and objectives: the ARIC Investigators. Am J Epidemiol. 1989;129:687–702.
- Wojcik NC,
- Huebner WW,
- Jorgensen G
- Frezzato M,
- Tosetto A,
- Rodeghiero F
- Wassel CL,
- Pankow JS,
- Peralta CA,
- Choudhry S,
- Seldin MF,
- Arnett DK
- 25.↵CDC WONDER. About Underlying Cause of Death, 1999-2010. http://wonder.cdc.gov/ucd-icd10.html. Accessed April 30, 2013.
Venous thromboembolism (VTE) is the third-leading cause vascular disease in the United States with >100 000 deaths annually. Prior studies have suggested that blacks have higher rates of VTE than whites in the United States, but these studies are limited in geographic or racial scope or rely on administrative databases. We studied the association of race (black versus white) with VTE in 3 large cohort studies to assess whether race and geography are associated with VTE in the United States. In the Cardiovascular Health Study (individuals ≥65 years of age recruited between 1989 and 1991 from 4 field centers), blacks had an increased risk of VTE not explained by conventional risk factors. In the Atherosclerosis Risk in Communities Study (individuals 45–64 years of age recruited between 1989 and 1991 from 4 field centers), blacks had a higher risk of VTE explained by conventional risk factors. In the Reasons for Geographic and Racial Differences in Stroke study (individuals ≥45 years of age recruited between 2003 and 2007 from throughout the contiguous United States), overall there was no association of race with VTE; however, blacks in the Southeast had a higher risk of VTE than blacks in the rest of the nation that was not explained by VTE risk factors, with no geographic differences apparent in whites. These data highlight that there are no unquestioned differences in VTE risk in blacks compared with whites and that geographic differences may be as important as race.