Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Circulation. Author manuscript; available in PMC 2013 May 1.
Published in final edited form as:
PMCID: PMC3342773

Triggers of Hospitalization for Venous Thromboembolism



The rate of hospitalization for venous thromboembolism (VTE) is increasing in the United States. While predictors of hospital-acquired VTE are well-known, triggers of VTE prior to hospitalization are not as clearly defined. The objective of this study was to evaluate triggers of hospitalization for VTE.

Methods and Results

A case-crossover study was conducted. Subjects were participants in the Health and Retirement Study, a nationally-representative sample of older Americans. Data were linked to Medicare files for hospital and nursing home stays, emergency department visits, outpatient visits including physician visits, and home health visits from years 1991-2007 (n=16,781). The outcome was hospitalization for venous thromboembolism (n=399). Exposures during the 90-day period prior to hospitalization for VTE were compared to exposures occurring in 4 comparison periods. Infection was the most common trigger of hospitalization for VTE, occurring in 52.4% of the risk periods prior to hospitalization. The adjusted incidence rate ratios (IRR; 95% CI) were 2.90 (2.13, 3.94) for all infection, 2.63 (1.90, 3.63) for infection without a prior hospital or skilled nursing facility stay, and 6.92 (4.46, 10.72) for infection with a prior hospital or skilled nursing facility stay. Erythropoiesis-stimulating agents and blood transfusion were also associated with VTE hospitalization (IRR=9.33, 95% CI: 1.19, 73.42; IRR=2.57, 95% CI: 1.17, 5.64; respectively). Other predictors included major surgeries, fractures (IRR=2.81), immobility (IRR=4.23) and chemotherapy (IRR=5.70). These predictors, combined, accounted for a large proportion (69.7%) of exposures prior to VTE hospitalization as opposed to 35.3% in the comparison periods.


Risk prediction algorithms for VTE should be reevaluated to include infection, erythropoiesis-stimulating agents and blood transfusion.

Keywords: deep vein thrombosis, infection, pulmonary embolism, thromboembolism, venous

Risk prediction algorithms are widely used for anticoagulation therapy to prevent venous thromboembolism (VTE) during hospitalization1-3. Such algorithms employ several well-known risk factors for hospital-acquired VTE. Less is known regarding the etiology of venous blood clots that form prior to hospitalization for VTE. This is of consequence because there are more than 330,000 hospital admissions for VTE annually in the United States4. From 2004-2009, the rate of hospitalization for VTE has increased from 7.0 to 8.4 per 1000 discharges in the United States4. Triggers (or potential proximal causes) of such hospitalizations for VTE have not been fully explored although acute-event triggers have been characterized for outcomes such as myocardial infarction, stroke, and drug side effects5-7.

In order to explore triggers that may precipitate hospitalization for VTE, we designed a case-crossover study using the linked databases of the Health and Retirement Study, a longitudinal study of a nationally representative sample of older Americans, and files from the Centers for Medicare and Medicaid Services (CMS). This design incorporates a within-person comparison and is particularly useful in evaluating transient exposures while controlling for confounding by fixed covariates such as race and gender, as well as relevant comorbidities and hereditary factors that influence the development of VTE8-10.


Data Sources

The Health and Retirement Study is an ongoing longitudinal study of Americans 51 years of age and older11. Participants were selected from a nationally representative sample of households in the United States and biennial interviews are conducted to collect information regarding economic circumstances, occupations and employment, health and health care, cognition, living and housing arrangements, demographics and family relationships (1992 to present). Data from participants who were fee-for-service Medicare beneficiaries were linked to files from the Centers of Medicare and Medicaid Services (CMS), 1991-2007 (n=16,781). The CMS files used for the purposes of this study were Inpatient Standard Analytical Files, Outpatient Standard Analytical Files, Skilled Nursing Facility Standard Analytical Files, Home Health Agency Standard Analytical Files, Carrier (Part B) Standard Analytical Files, and Denominator files. In the US, adults generally qualify for Medicare insurance at the age of 65 or, at younger ages, if they have been receiving Social Security or Railroad Retirement Board disability benefits for 2 or more years.


We identified all subjects who were admitted to a hospital with a principal diagnosis of either deep vein thrombosis or pulmonary embolism during 1991-2007 using valid ICD-9-CM codes (451.11, 451.19, 451.2, 451.81, 451.9, 453.40, 453.41, 453.42, 453.8, 453.9, 415.1, 415.11, 415.19). This approach has a positive predictive value of approximately 95%12.

Study Design

A case-crossover design was used. In this study design, each person serves as his/her own control and factors or characteristics that do not vary within an individual (i.e., history of comorbid conditions, behaviors, inherited thrombophilia) are fixed effects. The risk period of interest was the 90-day period prior to the date of hospital admission for VTE. Exposures during this 90-day period were compared with exposures during 4 previous 90-day periods, with a 90-day wash-out period between risk and comparison periods (Figure 1).

Figure 1
Design of the Case-Crossover Study.


Triggers under investigation were acute, transitory exposures that were potentially amenable to change. These included several injectable medications that were extracted as Healthcare Common Procedure Coding System (HCPCS) codes from Carrier and Outpatient files. Erythropoiesis-stimulating agents (epoetin alpha, darbepoetin alfa), chemotherapeutic agents, and antipsychotics (haloperidol, droperidol, chlorpromazine, fluphenazine, perphenazine, prochlorperazine, triflupromazine, promazine, promethazine, risperidone, ziprasidone, and aripiprazole) were extracted. Codes used in this study are listed in the Supplemental Appendix.

Other exposures of interest included injuries, surgeries, blood transfusion, central venous catheterization, infection and immobility. Injuries were ascertained using ICD-9-CM diagnosis codes from the Carrier and Outpatient files, including fractures, open wounds, sprains and dislocations. Lower limb amputations were identified from ICD-9-CM procedure codes in the Inpatient file. Major surgical procedures were identified through Berenson-Eggers Type of Service (BETOS) codes in the Carrier file and were categorized by CMS as cardiovascular, orthopedic and other major surgeries. Blood transfusion was ascertained through codes available in the Inpatient, Outpatient, Carrier, and Skilled Nursing Facility files. Central venous catheterization was determined using HCPCS codes in the Outpatient and Carrier files. Infections were determined using ICD-9-CM diagnosis codes that explicitly stated infection or provided evidence of infection (purulent, suppurative, septic, pyogenic or abscess) from the Outpatient, Inpatient, Carrier, Home Health Agency and Skilled Nursing Facility files. For purposes of this study, immobility was defined as any hospitalization or skilled nursing facility stay during the 90-day risk and comparison periods. Only nonsurgical hospitalizations were included since major surgeries were investigated separately, as described above. Dates for all exposures of interest were extracted. If the exposure occurred over several days (such as during hospitalization), the exposure was considered to have occurred if any of the days within such hospitalization fell within the specific 90-day time period.

Participant characteristics were available in the Health and Retirement Study and included age at the time of VTE hospitalization, gender, race (Caucasian, other), body mass index (BMI, kg/m2) at the time of the first interview, smoking (ever/never with smoking defined as more than 100 cigarettes in a lifetime), and region of residence at the time of the first interview (northeast, midwest, south, west). There were 0.6% (n=3) missing values for BMI and these were imputed prior to analysis. Diagnoses of diabetes, heart disease or cancer were ascertained through questions on the biennial interviews. For example, the participant was asked, “Has a doctor ever told you that you have diabetes or high blood sugar?” Each of these was coded as ever versus never (i.e., no diagnosis at any time).

Statistical Analyses

Characteristics of the participants were initially tabulated both for the sample and, with survey weighting, for the overall fee-for-service Medicare population in the United States. Statistical analyses followed the methods described for case-crossover designs8-10. For those patients who had multiple hospitalizations with a principal diagnosis of VTE, the first hospitalization was chosen. Incidence rate ratios (IRR) were calculated with 95% confidence intervals (CI) using a conditional fixed-effects Poisson model for panel data. The final Poisson model included simultaneous adjustment for major surgeries, infection, blood transfusion, hemodialysis, central venous catheterization, injuries, lower limb amputation, immobility, erythropoiesis-stimulating agents, injectable antipsychotics, and chemotherapy. Secondary analyses were performed to evaluate respiratory and non-respiratory tract infections, as well as specific chemotherapeutic agents. In addition, secondary analyses were conducted to assess whether the infection occurred with or without a prior hospitalization or skilled nursing facility stay. Alpha was set 0.05, 2-tailed and all analyses were conducted within Stata/MP 11.2 (College Station, TX).

This study received human subjects approval from the Institutional Review Board at the University of Michigan and the Privacy Board at CMS.

Role of the Funding Source

The National Heart, Lung, and Blood Institute and the National Institute on Aging had no role in the design and conduct of the study; collection, management, analysis and interpretation of the data; or preparation, review, or approval of the manuscript.


Of the 16,781 participants in the linked database, there were 480 patients who were hospitalized for VTE. Two subjects with VTE hospitalization had fewer than 1.5 years of observation and were excluded as this length of follow-up was necessary to determine exposures in the comparison periods. Of the 478 patients, 79 had a diagnosis of VTE at a date previous to the index hospitalization and therefore were excluded. This yielded 399 patients who were hospitalized for VTE without prior evidence of VTE and were included in the study.

The mean age of the 399 subjects at the time of the hospitalization for VTE was 76.9 years (SD 9.0). There were 4.8% of subjects who were Medicare beneficiaries and were less than 65 years of age when the outcome occurred. Most of the patients hospitalized for VTE were women and Caucasian (Table 1). Percentages are given for both the HRS sample used in this study and, with survey weighting, the reference population (fee-for-service Medicare beneficiaries in the United States). Twenty-nine percent of older Americans hospitalized for VTE were obese (BMI >30.0 kg/m2) and 41.8% were overweight. Heart disease and smoking were common in patients hospitalized for VTE and 26.3% had a diagnosis of cancer. Most patients hospitalized for VTE lived in the South or Midwest, accounting for nearly 3/4ths of such hospitalizations nationwide.

Table 1
Characteristics of Older Americans who were Hospitalized for Venous Thromboembolism

Frequencies of potential triggers during the risk and comparison periods are given in Table 2, as are the results from the regression models. Cardiovascular, orthopedic, and other major surgeries were significant independent triggers of hospitalization for VTE in the final model. The prevalence of major surgeries ranged from 6.8% to 8.8% in the 90 days prior to hospitalization for VTE while the prevalence was 1.0% to 2.6% in the comparison periods.

Table 2
Incidence Rate Ratios of Hospitalization for Venous Thromboembolism

Infection was the most common exposure during the 90-days prior to hospitalization for VTE, occurring in 52.4% of the patients compared to 28.2% in the comparison periods. An infection occurred 2.90 times more often before a VTE hospitalization than in the comparison periods (p<0.001). Respiratory tract infections occurred in 21.8% of the risk periods preceding VTE hospitalization and in 9.0% of the comparison periods. The unadjusted IRR for respiratory tract infection was 3.35 (95% CI: 2.43, 4.64). After adjustment in the final model (for the covariates in Table 2), the IRR was 2.75 (95% CI: 1.91, 3.96). Non-respiratory tract infections occurred 26.8% of the risk periods preceding VTE hospitalization and in 18.5% of the comparison periods. The unadjusted IRR for non-respiratory tract infection was 1.87 (95% CI: 1.39, 2.51) and the adjusted IRR was 1.45 (95% CI: 1.04, 2.03). There were 165 hospitalizations (41.4%) for pulmonary embolism and 234 hospitalizations (58.6%) for deep vein thrombosis in the study. The adjusted IRR for the association between infection and hospitalization for pulmonary embolism was 3.32 (95% CI, 2.04, 5.39), while the adjusted IRR for respiratory infection and hospitalization for pulmonary embolism was 2.95 (95% CI, 1.70, 5.11). The adjusted IRR for the association between infection and hospitalization for deep vein thrombosis was 2.72 (95% CI, 1.81, 4.08).

Transfusion occurred in 8.3% of the periods just prior to hospitalization and in 0.9% of the comparison time periods. The unadjusted IRR for transfusion was 9.99 and, with adjustment for surgery, infection and other covariates (Table 2), blood transfusion was a significant predictor of hospitalization for VTE with an IRR of 2.57 (p=0.018). Central venous catheterization was infrequent during the risk and comparison periods and, although significantly related to hospitalization for VTE in the unadjusted model, it was not associated with the outcome after adjustment for other factors.

Of the injuries studied, only fractures were independently associated with hospitalization for VTE, yielding an adjusted IRR of 2.81 (p<0.001). Fractures were 2.8 times more common prior to VTE hospitalization than in the comparison periods. Other types of injury such as sprains, dislocations, and open wounds were not associated with VTE hospitalization. The IRR for lower limb amputation was 8.00 in the unadjusted model (p=0.090) and was not significant in the final model including other covariates, although the frequency of this exposure was very low.

Immobility as defined by any nonsurgical hospitalization or skilled nursing home stay was also a significant trigger of hospitalization for VTE. The risk of VTE hospitalization was 4.2 fold greater in the time period when immobility occurred. Of note, immobility was a relatively common risk factor and occurred in 20.6% of the risk periods just prior to hospitalization for VTE compared to only 6.4% of the comparison risk periods.

We found a strong relationship between erythropoiesis-stimulating agents and hospitalization for VTE, with an adjusted IRR of 9.33 (p=0.034). In this cohort, 3% of the subjects received erythropoiesis-stimulating agents in the 90-day period prior to VTE hospitalization versus 0.8% in the comparison periods. The use of erythropoiesis-stimulating agents was more common in the weeks just prior to hospitalization for VTE (Figure 2).

Figure 2
Use of Erythropoiesis Stimulating Agents by Time of Administration Prior to Hospitalization for Venous Thromboembolism

Injectable antipsychotic medications were not significantly related to VTE hospitalization in the unadjusted model (p=0.070) or the adjusted model (p=0.577). Chemotherapeutic agents were given 5.7 times more often preceding a VTE hospitalization in contrast to the comparison periods (p=0.001). In particular, fluorouracil yielded an unadjusted IRR of 21.0 (95% CI: 2.53, 174.10) and an adjusted IRR of 28.47 (95% CI: 2.56, 317.06) when all covariates (from Table 2) were included in the regression model. Use of fluorouracil was relatively low, however, with 1.8% of patients exposed in the 90-days prior to the hospitalization for VTE compared to a mean percentage exposed of 0.3% in the comparison periods. Taxanes (docetaxel, paclitaxel), gemcitabine, and leuprolide did not yield significant IRRs in adjusted models although the unadjusted IRR for taxanes was 8.36 (95% CI: 0.78, 89.27) and the unadjusted IRR for gemcitabine was 17.70 (95% CI: 2.04, 153.32).

To determine whether our results differed when patients diagnosed with cancer were excluded, we repeated our analysis excluding this subset. This yielded 294 individuals (Table 3). In general, the results were similar. Major surgeries, infection, transfusion, fractures and immobility remained significant triggers of hospitalization for VTE. Excluding patients with cancer, respiratory tract infections occurred in 22.8% of the risk periods preceding VTE hospitalization and in 8.9% of the comparison periods. The unadjusted IRR for respiratory tract infection was 3.68 (95% CI: 2.52, 5.37) and the adjusted IRR was 2.96 (95% 1.94, 4.52) in participants who were never diagnosed with cancer. Non-respiratory tract infections occurred 26.5% of the risk periods preceding VTE hospitalization and in 19.2% of the comparison periods when patients with cancer were excluded. The unadjusted IRR for non-respiratory tract infection was 1.76 (95% CI: 1.24, 2.50) in subjects who were never diagnosed with cancer and the adjusted IRR was 1.51 (95% CI: 1.02, 2.22).

Table 3
Incidence Rate Ratios of Hospitalization for Venous Thromboembolism in Individuals without Cancer

We further explored whether or not the infection was accompanied by a prior hospital or skilled nursing facility stay. The results are shown in Table 4. Of the 399 patients hospitalized for VTE, there were 104 (26.1%) who had an infection without a hospital or skilled nursing facility stay in the 90-day risk period prior to hospitalization and there were 105 (26.3%) who had an infection with a stay during this time period. Using no infection as the reference category, the IRRs for infection with and without a prior stay were calculated. After adjustment for other covariates, the risk of hospitalization for VTE was 2.63 times greater with an infection that did not occur during a prior hospital or skilled nursing facility. The risk of hospitalization for VTE was 6.92 times greater when infection occurred during a prior hospital or skilled nursing facility. These analyses were repeated after excluding patients with cancer. The results were similar. The adjusted IRR was 2.69 for infection without prior hospitalization or a skilled nursing facility stay and 8.03 for infection with prior hospitalization or a skilled nursing facility stay.

Table thumbnail
Table 4. Incidence Rate Ratios for Infection and Hospitalization for Venous Thromboembolism by Prior Hospitalization or Skilled Nursing Facility Stay

Of the 399 subjects in the study, 69.7% (n=278) had at least one of the following exposures during the 90 days prior to hospitalization for VTE: major surgery, infection, blood transfusion, fracture, nonsurgical immobility, erythropoiesis-stimulating agent, and/or chemotherapy. This was in contrast to 35.3% exposure in the comparison periods, yielding a difference in exposure rates of 34.4% (i.e., “attributable exposure”). These factors, combined, yielded an IRR of 7.04 (95% CI, 5.25, 9.44).

In this cohort, 4.0% of patients died during the hospitalization or were transferred to a hospice after discharge, 20.1% were transferred to a skilled nursing facility or other care facility after discharge, 61.2% went home, and 14.5% were discharged home with services of a home health agency.


In this nationally-representative sample of older Americans, infection was the most frequent trigger of hospitalization for VTE, preceding more than half of all such hospitalizations. Moreover, respiratory tract infection appeared more strongly related to hospitalization for VTE than non-respiratory infections. Infection is not currently included on many clinical VTE risk assessment tools13-16 or on patient and consumer-related information sources17. Findings from our population-based study are consistent with population-based studies from England and Denmark18-20 and with a secondary data analysis of the MEDENOX (prophylaxis in MEDical patients with ENOXaparin) clinical trial21. In the Danish study, infection from any source was found to be a strong predictor of VTE, with an elevated VTE risk most predominant within 2 weeks after the onset of infection18. In the British investigations, the hypotheses centered on respiratory and urinary tract infections; results from one study indicated that both types of infection predicted VTE19 while the other found that respiratory tract infections were more strongly related to VTE20. The physiological underpinnings of the interconnection between the inflammatory response and coagulation have been well described22-24. Venous stasis is a recognized component of the inflammatory response which enables the migration of leukocytes to the site of infection. Inflammation initiates the tissue factor pathway in the coagulation cascade, through induction of tissue factor by complement activation, endotoxin, C-reactive protein and/or inflammatory cytokines that contribute to hemostasis and thrombus formation23. A patient’s ability to resolve an infection while not provoking overstimulation of coagulation would be an important avenue for future research.

Blood transfusion is also not often incorporated in VTE risk prediction tools or patient-education materials13-17. Our finding that blood transfusion precedes VTE hospitalization is consistent with several clinical and laboratory studies. With increased storage, transfused red blood cells have been shown to exhibit greater adhesion to endothelial cells with sequestration in the lungs, resulting in decreased blood oxygenation and exacerbating microvascular pathology25. Hemolysis and microparticle formation occur with increased length of storage and when transfused, is associated with a reduction in vasodilation, platelet adhesion and aggregation, and exacerbation of inflammation26. Unfortunately, length of storage was not available in our database for red cell transfusion nor was there information regarding the quantity of transfused red cells, platelets or plasma. However, both red blood cell and platelet transfusions were found to be risk factors for VTE in a study of 504,208 hospitalizations using the University Health System Consortium data27. Blood transfusion was also a significant predictor of VTE in a study of 21,943 patients undergoing colorectal resection using data from the National Surgical Quality Improvement Program28. Clinical trials comparing different transfusion thresholds have been too small to evaluate VTE risk (2 trials with a total of 204 patients) 29.

A potent trigger of hospitalization for VTE is the use of erythropoiesis-stimulating agents. A meta-analysis of Phase 3 trials of erythropoiesis-stimulating agents in patients with cancer demonstrated an increased risk of VTE and mortality30. There have been several trials suggesting concerns with administration in patients with chronic kidney disease as well31. Labeling of these drugs was revised by the FDA with a black box warning in 2007 for treatment in patients with cancer and in 2011, the FDA issued a safety communication regarding their use in patients with chronic kidney disease32, 33. Nevertheless, one of the erythropoiesis-stimulating agents was still among the top 10 prescribed drugs in the United States in 2010 based on sales34. Furthermore, there is considerable variability in use across providers35. Unfortunately, our findings suggest that patients who receive either erythropoiesis-stimulating agents or blood transfusions, two of the most common therapies available for anemia management, are at markedly higher risk of VTE.

Antipsychotic medications were found to increase the risk of VTE in a large primary care population-based study in the United Kingdom36. In their results, the adjusted odds ratio for VTE was most elevated for antipsychotics given by injection (odds ratio, 3.24). Since injectable medications were available through Medicare data in our study, we evaluated injectable antipsychotics as well. The adjusted IRR was not significant in our study, although utilization of antipsychotic medications was rather low (1.0% in the risk period and 0.3% in the comparison periods) and therefore, it is possible that we did not have sufficient power to detect such differences.

Our findings that established risk factors of VTE are triggers of hospitalization for VTE (e.g., surgery, fractures, immobility, chemotherapy) indicate that this particular study design is suitable for evaluating such predictors. Moreover, the case-crossover design implicitly controls for genetic determinants, demographic characteristics and past comorbidities since the comparisons are made within the individual. Therefore, the findings are not due to differences in the genetic profile of patients or medical history of various chronic conditions; such factors are fixed in this study.

There are some limitations of this study. First, we would have preferred a larger database in order to investigate less common exposures, particularly since exposure occurrence was within specific time ranges. Second, oral medications were not available in the existing database so we were unable to evaluate hormone replacement therapy or other medications that may trigger or prevent VTE. Likewise, we were unable to completely characterize immobility in this database. It is possible that there were other factors that changed between the risk period and the comparison periods. There was a maximum of 1.5 years from the beginning of the first comparison period and the end of the risk period. Therefore, changes in factors related to vascular health could have occurred that were not recorded in the medical visits and stays available through the hospital, skilled nursing home, home health, physician, or outpatient files. Moreover, interviews in the Health and Retirement Study occurred every two years and therefore, changes in habits within shorter periods of time (e.g., 6 months) were not known. While we considered the possibility of diagnostic bias (e.g., VTE found during a diagnostic work-up for infection), the date of the visit/stay when the infection was recorded was prior to the date of hospitalization for VTE. Moreover, at the time when the infection was recorded (either during a clinic visit, home health visit, previous hospital stay, etc), VTE was not recorded. This provides evidence that the infection occurred prior to the VTE.

In conclusion, infection is a common potential trigger of hospitalization for VTE and patients should be made aware of this risk. Exposure to erythropoiesis-stimulating agents and blood transfusion are not as common but, if administered, these therapies confer significantly elevated risk of hospitalization for VTE. Current risk algorithms for VTE need reassessment and possible updating to include infection, erythropoiesis-stimulating agents and blood transfusion.


There are more than 330,000 hospital admissions for venous thromboembolism annually in the United States. The objective of this study was to evaluate triggers, or acute transitory exposures, of such hospitalizationswhich potentially areamenable to change. We designed a case-crossover study using the databases of the Health and Retirement Study, a longitudinal study of a nationally representative sample of older Americans, linked toMedicare files. This design incorporates a within-person comparison (90 days prior to the hospitalization versus other time periods) and is particularly useful in controlling for factors such as hereditary conditions, gender, race and past comorbidities. We found that over half (52%) of all hospitalizations for venous thromboembolism were preceded by an infection and that the risk of such hospitalizationswas 2.9 fold greater within 90 days of an infection than at other times. Both respiratory and non-respiratory infections were associated with greater risk. Individuals were at 9.3 fold greater risk of hospitalization for venous thromboembolism within 90 days of receiving erythropoiesis-stimulating agents than at other times. In addition, the results suggest that blood transfusion is a significant trigger of hospitalization for venous thromboembolism, independent of major surgeries, infection, injuries and immobility. We recommend that clinical prediction rules and patient informational sources be updated to include infection, erythropoiesis- stimulating agents and blood transfusion as risk factors for venous thromboembolism.

Supplementary Material

Supplementary Material


Funding Sources: This study was supported by a grant from the National Heart, Lung, and Blood Institute, 5R21HL093129-02 (Rogers, PI). The Health and Retirement Study is sponsored by the National Institute on Aging (grant number NIA U01AG009740) and is conducted by the University of Michigan. Dr. Blumberg is supported in part by a grant from the National Heart, Lung, and Blood Institute RO1 HL095467.


Conflict of Interest Disclosures: Dr. Blumberg received lecture fees from Fenwal, Caridian and Pall (<$10,000), and was a consultant to Fenwal (<$10,000). Dr. Flanders has received honoraria for presentations as a Visiting Professor and at a medical specialty society meeting, testified as an expert witness (<$10,000) and served as a consultant for the Institute of Healthcare Improvement and the Centers for Disease Control and Prevention (>$10,000).

Journal Subject Codes: [8] Epidemiology; [121] Primary prevention; [173] Deep vein thrombosis

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


1. Wells PS, Anderson DR, Rodger M, Ginsberg JS, Kearon C, Gent M, Turpie AG, Bormanis J, Weitz J, Chamberlain M, Bowie D, Barnes D, Hirsh J. Derivation of a simple clinical model to categorize patients probability of pulmonary embolism: increasing the models utility with the SimpliRED D-dimer. Thromb Haemost. 2000;83:416–420. [PubMed]
2. van Belle A, Büller HR, Huisman MV, Huisman PM, Kaasjager K, Kamphuisen PW, Kramer MH, Kruip MJ, Kwakkel-van Erp JM, Leebeek FW, Nijkeuter M, Prins MH, Sohne M, Tick LW, Christopher Study Investigators Effectiveness of managing suspected pulmonary embolism using an algorithm combining clinical probability, D-dimer testing, and computed tomography. JAMA. 2006;295:172–179. [PubMed]
3. Spyropoulos AC, Anderson FA, Jr, Fitzgerald G, Decousus H, Pini M, Chong BH, Zotz RB, Bergmann JF, Tapson V, Froehlich JB, Monreal M, Merli GJ, Pavanello R, Turpie AG, Nakamura M, Piovella F, Kakkar AK, Spencer FA, IMPROVE Investigators Predictive and associative models to identify hospitalized medical patients at risk for VTE. Chest. 2011;140:706–714. [PubMed]
4. Healthcare Cost and Utilization Project (HCUP) HCUP Databases. Agency for Healthcare Research and Quality; Rockville, MD: [Accessed 11 November 2011].
5. Mittleman MA, Mostofsky E. Physical, psychological and chemical triggers of acute cardiovascular events: preventive strategies. Circulation. 2011;124:346–354. [PMC free article] [PubMed]
6. Douglas IJ, Smeeth L. Exposure to antipsychotics and risk of stroke: self controlled case series study. BMJ. 2008;337:a1227. [PubMed]
7. Lee GM, Greene SK, Weintraub ES, Baggs J, Kulldorff M, Fireman BH, Baxter R, Jacobsen SJ, Irving S, Daley MF, Yin R, Naleway A, Nordin JD, Li L, McCarthy N, Vellozzi C, Destefano F, Lieu TA. Vaccine Safety Datalink Project. H1N1 and seasonal influenza vaccine safety in the vaccine safety datalink project. Am J Prev Med. 2011;41:121–128. [PubMed]
8. Maclure M, Mittleman MA. Should we use a case-crossover design? Annu Rev Public Health. 2000;21:193–221. [PubMed]
9. Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol. 1991 Jan 15;133:144–153. [PubMed]
10. Whitaker HJ, Farrington CP, Spiessens B, Musonda P. Tutorial in biostatistics: the self-controlled case series method. Stat Med. 2006;25:1768–1797. [PubMed]
11. Juster FT, Suzman R. An overview of the Health and Retirement Study. J Hum Resour. 1995;30(Suppl):S7–S56.
12. White RH, Garcia M, Sadeghi B, Tancredi DJ, Zrelak P, Cuny J, Sama P, Gammon H, Schmaltz S, Romano PS. Evaluation of the predictive value of ICD-9-CM coded administrative data for venous thromboembolism in the United States. Thromb Res. 2010;126:61–67. [PubMed]
13. Hippisley-Cox J, Coupland C. Development and validation of risk prediction algorithm (QThrombosis) to estimate future risk of venous thromboembolism: prospective cohort study. BMJ. 2011;343:d4656. [PubMed]
14. Wells PS, Owen C, Doucette S, Fergusson D, Tran H. Does this patient have deep vein thrombosis? JAMA. 2006;295:199–207. [PubMed]
15. Eichinger S, Heinze G, Jandeck LM, Kyrle PA. Risk assessment of recurrence in patients with unprovoked deep vein thrombosis or pulmonary embolism: the Vienna prediction model. Circulation. 2010;121:1630–1636. [PubMed]
16. Gearhart MM, Luchette FA, Proctor MC, Lutomski DM, Witsken C, James L, Davis K, Jr, Johannigman JA, Hurst JM, Frame SB. The risk assessment profile score identifies trauma patients at risk for deep vein thrombosis. Surgery. 2000;128:631–640. [PubMed]
17. Your Guide to Preventing and Treating Blood Clots. Agency for Healthcare Research and Quality; Rockville, MD: [Accessed 12 November 2011]. May, 2009. AHRQ Publication No. 09-0067-C.
18. Schmidt M, Horvath-Puho E, Thomsen RW, Smeeth L, Sørensen HT. Acute infections and venous thromboembolism. J Intern Med. 2011 doi: 10.1111/j.1365-2796.2011.02473.x. [PMC free article] [PubMed]
19. Smeeth L, Cook C, Thomas S, Hall AJ, Hubbard R, Vallance P. Risk of deep vein thrombosis and pulmonary embolism after acute infection in a community setting. Lancet. 2006;367:1075–1079. [PubMed]
20. Clayton TC, Gaskin M, Meade TW. Recent respiratory infection and risk of venous thromboembolism: case-control study through a general practice database. Int J Epidemiol. 2011;40:819–827. [PMC free article] [PubMed]
21. Alikhan R, Cohen AT, Combe S, Samama MM, Desjardins L, Eldor A, Janbon C, Leizorovicz A, Olsson CG, Turpie AG, MEDENOX Stud Risk factors for venous thromboembolism in hospitalized patients with acute medical illness: analysis of the MEDENOX Study. Arch Intern Med. 2004;164:963–968. [PubMed]
22. Wagner DD. New links between inflammation and thrombosis. Arterioscler Thromb Vasc Biol. 2005;25:1321–1324. [PubMed]
23. Esmon CT. The impact of the inflammatory response on coagulation. Thromb Res. 2004;114:321–327. [PubMed]
24. van der Poll T, de Boer JD, Levi M. The effect of inflammation on coagulation and vice versa. Curr Opin Infect Dis. 2011;24:273–278. [PubMed]
25. Zhu H, Zennadi R, Xu BX, Eu JP, Torok JA, Telen MJ, McMahon TJ. Impaired adenosine-5′-triphosphate release from red blood cells promotes their adhesion to endothelial cells: A mechanism of hypoxemia after transfusion. Crit Care Med. 2011;39:2478–2486. [PMC free article] [PubMed]
26. Lee JS, Gladwin MT. Bad blood: the risks of red cell storage. Nat Med. 2010;16:381–382. [PubMed]
27. Khorana AA, Francis CW, Blumberg N, Culakova E, Refaai MA, Lyman GH. Blood transfusions, thrombosis, and mortality in hospitalized patients with cancer. Arch Intern Med. 2008;168:2377–2381. [PMC free article] [PubMed]
28. Xenos ES, Vargas HD, Davenport DL. Association of blood transfusion and venous thromboembolism after colorectal cancer resection. Thromb Res. 2011 doi:10.1016/j.thromres.2011.07.047. [PubMed]
29. Carless PA, Henry DA, Carson JL, Hebert PP, McClelland B, Ker K. Transfusion thresholds and other strategies for guiding allogeneic red blood cell transfusion. Cochrane Database Syst Rev. 2010;10 CD002042. [PubMed]
30. Bennett CL, Silver SM, Djulbegovic B, Samaras AT, Blau CA, Gleason KJ, Barnato SE, Elverman KM, Courtney DM, McKoy JM, Edwards BJ, Tigue CC, Raisch DW, Yarnold PR, Dorr DA, Kuzel TM, Tallman MS, Trifilio SM, West DP, Lai SY, Henke M. Venous thromboembolism and mortality associated with recombinant erythropoietin and darbepoetin administration for the treatment of cancer-associated anemia. JAMA. 2008;299:914–924. [PubMed]
31. Unger EF, Thompson AM, Blank MJ, Temple R. Erythropoiesis-stimulating agents--time for a reevaluation. N Engl J Med. 2010;362:189–192. [PubMed]
32. U.S. Food and Drug Administration [Accessed 12 November 2011];Public Health Advisory: Erythropoiesis-Stimulating Agents (ESAs): Epoetin alfa (marketed as Procrit and Epogen), Darbepoetin alfa (marketed as ranesp) 11/08/207.
33. U.S. Food and Drug Administration [Accessed 12 November 2011];FDA Drug Safety Communication: Modified dosing recommendations to improve the safe use of Erythropoiesis-Stimulating Agents (ESAs) in chronic kidney disease. 6/24/2011.
34. Bartholow M. [Accessed 12 November 2011];Top 200 drugs of 2010. Pharmacy Times.
35. Wright JD, Neugut AI, Wilde ET, Buono DL, Malin J, Tsai WY, Hershman DL. Physician characteristics and variability of erythropoiesis-stimulating agent use among Medicare patients with cancer. J Clin Oncol. 2011;29:3408–3418. [PMC free article] [PubMed]
36. Parker C, Coupland C, Hippisley-Cox J. Antipsychotic drugs and risk of venous thromboembolism: nested case-control study. BMJ. 2010;341:c4245. [PubMed]