ORIGINAL INVESTIGATION

Risk Factors for Deep Vein Thrombosis

and Pulmonary Embolism
A Population-Based Case-Control Study
John A. Heit, MD; Marc D. Silverstein, MD; David N. Mohr, MD;
Tanya M. Petterson, MS; W. Michael O’Fallon, PhD; L. Joseph Melton III, MD

Background: Reported risk factors for venous throm- or nursing home confinement (OR, 8.0; 95% CI, 4.5-14.2),
boembolism (VTE) vary widely, and the magnitude and malignant neoplasm with (OR, 6.5; 95% CI, 2.1-20.2) or
independence of each are uncertain. without (OR, 4.1; 95% CI, 1.9-8.5) chemotherapy, central
venous catheter or pacemaker (OR, 5.6; 95% CI, 1.6-19.6),
Objectives: To identify independent risk factors for deep superficial vein thrombosis (OR, 4.3; 95% CI, 1.8-10.6), and
vein thrombosis and pulmonary embolism and to esti- neurological disease with extremity paresis (OR, 3.0; 95%
mate the magnitude of risk for each. CI, 1.3-7.4). The risk associated with varicose veins dimin-
ished with age (for age 45 years: OR, 4.2; 95% CI, 1.6-11.3;
Patients and Methods: We performed a population- for age 60 years: OR, 1.9; 95% CI, 1.0-3.6; for age 75 years:
based, nested, case-control study of 625 Olmsted County, OR, 0.9; 95% CI, 0.6-1.4), while patients with liver disease
Minnesota, patients with a first lifetime VTE diagnosed had a reduced risk (OR, 0.1; 95% CI, 0.0-0.7).
during the 15-year period from January 1, 1976, through
December 31, 1990, and 625 Olmsted County patients Conclusion: Hospital or nursing home confinement, sur-
without VTE. The 2 groups were matched on age, sex, gery, trauma, malignant neoplasm, chemotherapy, neu-
calendar year, and medical record number. rologic disease with paresis, central venous catheter or
pacemaker, varicose veins, and superficial vein throm-
Results: Independent risk factors for VTE included sur- bosis are independent and important risk factors for VTE.
gery (odds ratio [OR], 21.7; 95% confidence interval [CI],
9.4-49.9), trauma (OR, 12.7; 95% CI, 4.1-39.7), hospital Arch Intern Med. 2000;160:809-815

V
ENOUS thromboembolism ample, studies that identified cases solely
(VTE) is a major national by autopsy4-6 or only included patients who
health problem, with at were enrolled in clinical trials7,8 or of one
least 201 000 first life- sex9 may not have identified important risk
time cases reported each factors, since the full clinical spectrum of
year in the United States.1 Of these, about disease was not represented. Moreover,
25% die within 7 days of VTE onset; for previous prospective cohort studies were
From the Division of about 22% of all patients with VTE, death limited by the relatively low incidence of
Cardiovascular Diseases and is so rapid, there is insufficient time for in- VTE and the correspondingly small sample
the Section of Hematology
tervention.2 Thus, to improve survival, pa- sizes.6,9 Because these cohort studies were
Research (Dr Heit), the
Division of Area General tients at risk must be identified and given not designed to determine risk factors for
Internal Medicine appropriate prophylaxis in order to re- VTE, the baseline characteristics avail-
(Drs Silverstein and Mohr), duce the incidence of VTE. Despite im- able for analysis did not include all po-
Department of Internal proved prophylaxis regimens,3 however, tential characteristics thought to place
Medicine, and the Sections of the annual incidence of VTE has been rela- patients at risk. Finally, previous case-
Clinical Epidemiology tively constant, at about 1 event per 1000 control studies either included an inap-
(Drs Silverstein and Melton) person-years since 1979.1 The failure to re- propriate control group7,8 or only ad-
and Biostatistics (Ms Petterson duce this rate may be a result of uncer- dressed the risk among women receiving
and Dr O’Fallon), Department tainty regarding risk factors for VTE and oral contraceptives10 or hormone replace-
of Health Sciences Research,
the associated difficulty in recognizing in- ment therapy.11-15
Mayo Clinic and Mayo
Foundation, Rochester, Minn. dividuals at risk.
Dr Silverstein is now with the Reported risk factors vary widely, and See also page 761
Center for Health Care the independence and magnitude of each
Research, Medical University of are uncertain. Several study design issues We have identified the inception co-
South Carolina, Charleston. may account for this variability. For ex- hort of Olmsted County, Minnesota, resi-

(REPRINTED) ARCH INTERN MED/ VOL 160, MAR 27, 2000 WWW.ARCHINTERNMED.COM
809

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a Purdue University User on 05/21/2015

 PATIENTS AND METHODS condition on part II of the death certificate. Pulmonary em-
bolism events that were first identified on autopsy exami-
STUDY SETTING AND DESIGN nation but not specifically labeled as a cause of death in
the autopsy report or listed on the death certificate were
Using the data resources of the Rochester Epidemiology categorized as “noncausal for death.”
Project,16 we identified the inception cohort of Olmsted
County residents with a first lifetime DVT or PE during the BASELINE CHARACTERISTICS
25-year period from 1966 through 1990 as previously de-
scribed.1 We then performed a case-control study nested A large number of baseline characteristics were tested as
within the Olmsted County population. All Olmsted County risk factors for VTE. Data were obtained by review of all
residents with a first lifetime definite DVT or PE diag- medical records (inpatient and outpatient) in the commu-
nosed during the 15-year period from January 1, 1976, nity for each subject17; consequently, it was not possible
through December 31, 1990, were included in the present to blind the nurse abstractors to case or control status. The
study. The Rochester Epidemiology Project also provides overall mean duration of prior medical record documen-
an enumeration of the population from which controls can tation was 34.7 years (34.7 years for patients with DVT or
be sampled, as described elsewhere.16 Using this system, PE and 34.7 years for controls). The characteristics as-
the Olmsted County resident matched for age (±1 year), sessed included type of incident event (PE, DVT, or both);
calendar year (±1 year), and sex whose medical record num- age at incident event; sex; year of incident event; patient
ber was closest to the medical record number of each pa- location at incident event onset (community, community
tient with DVT or PE was selected as a control. The study but hospitalized in the previous 90 days, hospital, or nurs-
was approved by the Mayo Clinic Institutional Review Board. ing home); body mass index (BMI) (calculated as weight
in kilograms divided by the square of height in meters:
DEFINITION OF DVT AND PE weight [kg]/[height (m)]2); chronic heart disease (conges-
tive heart failure vs other heart disease [ie, congenital heart
A DVT was categorized as definite when confirmed by veno- disease, cardiomyopathy, ischemic heart disease, or valvu-
gram, computed tomographic scan, magnetic resonance im- lar heart disease]); active malignant neoplasm (excluding
age, or pathologic examination of thrombus removed at sur- nonmelanoma skin cancer) with or without chemo-
gery or autopsy. A PE was categorized as definite when therapy (cytotoxic or immunosuppressive therapy for ma-
confirmed by pulmonary angiogram, computed tomo- lignant neoplasm, excluding tamoxifen); serious neuro-
graphic scan, magnetic resonance image, or pathologic ex- logic disease (stroke or other disease affecting the nervous
amination of thrombus removed at surgery or autopsy. system with associated extremity paresis, or acute stroke
Mayo Clinic pathologists performed all autopsy ex- with extremity paresis requiring hospitalization within the
aminations and completed the death certificates of per- previous 3 months); surgery requiring anesthesia (gen-
sons who died within Olmsted County during the study eral, orthopedic, neurologic, or gynecologic surgery); trauma
period. Autopsy-discovered PE events were classified as a requiring hospital admission (major fracture or severe soft-
cause of death only if the pathologist labeled them as such tissue injury); chronic lung disease (chronic obstructive pul-
in the autopsy report or if the death certificate listed PE as monary disease, emphysema, chronic bronchitis, bronchi-
an immediate or underlying cause of death or included PE ectasis, interstitial lung disease, pulmonary hypertension
on part I of the death certificate. Autopsy-discovered PE [asthma was included only if there was documented evi-
events were classified as a “contributory cause of death” if dence of fixed airflow obstruction]); serious liver disease
PE was listed as a contributing cause or other significant (including active hepatitis within the previous 3 months);

dents with a first lifetime deep vein thrombosis (DVT) topsy-discovered PE was classified as an immediate, un-
or pulmonary embolism (PE) diagnosed during the 25- derlying, or contributory cause of death, leaving 115 defi-
year period from 1966 through 1990.1 To address the limi- nite PE events that were classified as noncausal for death.
tations of previous studies, we performed a nested case- In univariate analyses of more than 25 baseline char-
control study to identify independent risk factors for DVT acteristics tested as potential risk factors for VTE, age;
and PE and to estimate the magnitude of risk associated BMI; congestive heart failure; active malignant neo-
with each. plasm; chemotherapy; previous superficial vein throm-
bosis; previous varicose vein procedure; chronic renal dis-
RESULTS ease; neurologic disease with extremity paresis; previous
central venous catheterization or transvenous pace-
Six hundred twenty-seven Olmsted County residents were maker placement; trauma; any surgery; orthopedic sur-
diagnosed with a definite first lifetime episode of DVT gery; neurosurgery; anesthesia; and hospital, nursing
or PE during the period from January 1, 1976, through home, or previous hospital admission within 90 days were
December 31, 1990. Two patients were missing risk- all significant risk factors (Table 1). Among the 326 fe-
factor information because of lost medical records and male case-control pairs, only the postpartum state and
were excluded. After these exclusions, the study popu- gynecologic surgery were additional significant risk fac-
lation consisted of 625 cases (326 women [52%]), and tors for VTE in the univariate analyses.
625 age- and sex-matched Olmsted County residents with- In the multivariate analysis, independent risk fac-
out VTE. For 237 cases (38%), the day of VTE onset was tors for VTE included surgery, trauma, hospital or nurs-
the day of their death. For 122 of these patients, the au- ing home inpatient status, malignant neoplasm with and

(REPRINTED) ARCH INTERN MED/ VOL 160, MAR 27, 2000 WWW.ARCHINTERNMED.COM
810

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a Purdue University User on 05/21/2015

 chronic renal disease (physician’s diagnosis and either cre- cluding interactions, since only a subset of the cases could
atinine level ⬎175 µmol/L [2 mg/dL] for at least 3 months be used. In the subset of cases with complete smoking in-
or nephrotic syndrome; arteriovenous fistula thrombosis formation, we verified that the final multivariate model vari-
excluded); inflammatory bowel disease; previous superfi- ables did not have odds ratios differing from those com-
cial vein thrombosis; varicose veins (varicose veins or treated puted using all cases. We then categorized those patients
varicose veins [injection sclerotherapy or stripping]); cen- with missing smoking status as nonsmokers, reasoning that
tral venous catheter or transvenous pacemaker place- this would be the most conservative approach. Because preg-
ment; anticoagulation therapy or prophylaxis immedi- nancy and the postpartum period, oral contraceptive use,
ately preceding or at the time of the incident event; smoking hormone replacement therapy, tamoxifen therapy, and gy-
status (none, former, or current [cigarettes only]); and (for necologic surgery could only be evaluated in women, these
women only) pregnancy or postpartum at the time of the variables were assessed after determining the otherwise fi-
incident event, oral contraceptive use, hormone therapy (es- nal model, including interactions. In the univariate analy-
trogen or progesterone), gynecologic surgery, and tamox- ses, the surgery variables were important risk factors. In
ifen therapy. It was necessary that active malignant neo- the multivariate modeling, however, the surgery variables
plasm, chemotherapy, serious neurologic disease (acute were nonsignificant after adjusting for patient location at
stroke or temporary lower-extremity paresis), all surgery VTE onset, and it was not possible to address interactions
variables, anesthesia, trauma, serious liver disease (active of surgery and hospitalization. Therefore, we created a new
or chronic hepatitis only), hormone therapy, central vein variable with the following categories: community onset
catheterization, oral contraceptive use, and tamoxifen with no recent (within 90 days) institutionalization (con-
therapy be documented in the 3 months prior to the inci- finement in hospital or nursing home), onset while insti-
dent event. Congestive heart failure and other heart dis- tutionalized but without recent (within 90 days) surgery,
ease, serious neurologic disease (with unresolved lower- and onset while institutionalized with recent surgery.
extremity paresis), chronic lung disease, serious liver disease
(cirrhosis only), chronic renal disease, nephrotic syn- ANALYSIS
drome, inflammatory bowel disease, previous superficial
vein thrombosis, varicose veins, and permanent transve- Baseline characteristics were assessed as potential risk fac-
nous pacemaker placement could be documented any time tors for VTE using conditional logistic regression. We used
prior to the incident event. Body mass index was based on stepwise and backwards conditional logistic regression to
the most recent height and weight measurements prior to identify a “final model.” Pⱕ.05 was required to enter the
the incident event. model. The variables selected were validated using a boot-
Body mass index could not be calculated for 16 cases strap method18 in which 500 random samples of the case-
and 176 controls, mainly because of missing height mea- control pairs were selected with replacement. A stepwise
surements. For these cases and controls, we imputed BMI regression was run for each sample. Variables were con-
values by assigning to each the respective average height sidered validated and retained in the final model only if they
or weight of the same sex and 10-year age group of cases entered more than 70% of the 500 logistic regression mod-
or controls. For the one child younger than 15 years with els. When the list of main effect variables was finalized, all
a missing measurement for height, we used the 50th height 2-way interactions of the main variables in the model were
percentile from the 1976 National Center for Health Sta- validated using the bootstrap technique. Discrete vari-
tistics growth chart. Because smoking status was missing ables with multiple components could not be assessed us-
in 88 case-control pairs, this variable was evaluated only ing the usual stepwise technique. Such variables were vali-
after determination of the otherwise final model, in- dated separately, as were all interactions.

without concurrent chemotherapy, previous central vein ous liver disease were 90% less likely to develop VTE.
catheterization or transvenous pacemaker placement, pre- Congestive heart failure was not a significant risk factor
vious superficial vein thrombosis, varicose veins, neu- for VTE when it was discovered before death or was dis-
rologic disease with extremity paresis, and congestive heart covered at autopsy and categorized as a cause of death;
failure (Table 2 and Figure). The risk of VTE was 22- congestive heart failure was only a significant risk fac-
fold higher for patients with recent surgery, more than tor for autopsy-discovered VTE categorized as non-
12-fold higher for patients with recent trauma, and nearly causal for death. Varicose veins were a significant risk
8-fold higher for patients confined to a hospital or nurs- factor for VTE, but the risk varied with age. The risk was
ing home. Malignant neoplasm alone was associated with highest among younger patients with varicose veins and
a 4-fold increased risk of VTE, and cytotoxic or immu- diminished with age.
nosuppressive chemotherapy increased the malignant neo-
plasm–associated risk to more than 6-fold. Patients with COMMENT
previous superficial vein thrombosis were more than 4
times more likely to develop DVT or PE. Placement of This population-based case-control study identified a num-
either a current or recent central venous catheter and ber of independent risk factors for VTE. Our study is the
placement of any previous transvenous pacemaker were first to identify hospital, nursing home, or other chronic
strong risk factors for upper-extremity venous throm- care facility confinement as an independent risk factor for
bosis, with more than a 5-fold increased risk. Patients with VTE. It is likely that the increased risk associated with hos-
neurologic disease with extremity paresis were at a 3-fold pital confinement reflects relative immobilization4,7,20,21 and
increased risk for VTE. Interestingly, patients with seri- the acuity and severity of illness. However, it is less clear

(REPRINTED) ARCH INTERN MED/ VOL 160, MAR 27, 2000 WWW.ARCHINTERNMED.COM
811

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a Purdue University User on 05/21/2015

 Table 1. Univariate Logistic Analyses of Potential Risk Factors for Definite Deep Vein Thrombosis
or Pulmonary Embolism Among Olmsted County, Minnesota, Residents With a First Lifetime
Venous Thomboembolism Diagnosed From 1966 Through 1990 Compared With Community Controls

Case-Control Pairs, No.*

Odds Ratio
Baseline Characteristics +/+ +/− −/+ −/− (95% Confidence Interval)
Age, y 1.38 (1.09-1.74)
Body mass index, kg/m2† 0.98 (0.96-1.00)
Smoking (ever)‡ 138 124 108 169 1.15 (0.89-1.49)
None 1.00
Current 1.30 (0.91-1.85)
Former 1.07 (0.79-1.44)
Institutionalization within previous 90 days 73 332 18 203 18.44 (11.48-29.64)
Community acquired 1.00
Community acquired/hospitalized within 90 days 10.03 (5.78-17.42)
Nursing home acquired 10.64 (5.55-20.38)
Hospital acquired 464.95 (62.68-3448.83)
Previous superficial vein thrombosis 4 40 16 565 2.50 (1.40-4.46)
Previous central venous catheter or pacemaker 0 71 6 548 11.83 (5.14-27.23)
Varicose veins 23 117 84 401 1.39 (1.05-1.84)
All cardiac diseases 65 124 79 357 1.57 (1.18-2.08)
No cardiac disease 1.00
Congestive heart failure 2.78 (1.85-4.17)
Other heart conditions 0.99 (0.69-1.41)
Chronic lung disease 15 81 66 463 1.23 (0.89-1.70)
Serious liver disease 0 5 6 614 0.83 (0.25-2.73)
Inflammatory bowel disease 0 4 5 616 0.80 (0.21-2.98)
Chronic renal disease 0 18 6 601 3.00 (1.19-7.56)
Systemic lupus erythematosus 0 3 1 621 3.00 (0.31-28.84)
Myeloproliferative disease 0 8 2 615 4.00 (0.85-18.84)
Coagulation disorders 0 2 1 622 2.00 (0.18-22.06)
Neurologic disease with extremity paresis 2 62 12 549 5.17 (2.78-9.59)
Malignant neoplasms 4 138 18 465 7.67 (4.69-12.53)
No malignant neoplasm 1.00
Malignant neoplasm without chemotherapy 6.90 (3.92-12.17)
Malignant neoplasm with chemotherapy 9.90 (3.89-25.18)
General surgery 0 75 5 545 15.00 (6.07-37.09)
Orthopedic surgery 0 70 6 549 11.67 (5.07-26.86)
Neurosurgery§ 0 20 0 605 41.00 (2.48-677.92)
Any anesthesia 2 175 10 438 17.50 (9.25-33.10)
No anesthesia 1.00
Regional anesthesia 11.50 (2.71-48.73)
General anesthesia 19.00 (9.33-38.68)
Trauma 0 82 4 539 20.50 (7.51-55.93)
Pregnancy㛳 1 2 2 321 1.00 (0.14-7.10)
Postpartum§㛳 0 9 0 317 19.00 (1.11-326.46)
Oral contraceptive therapy㛳 2 10 7 307 1.43 (0.54-3.75)
Estrogen replacement therapy㛳 1 13 9 303 1.44 (0.62-3.38)
Progesterone replacement therapy§㛳 0 2 0 324 5.00 (0.28-90.43)
Estrogen/progesterone replacement therapy㛳 1 14 9 302 1.56 (0.67-3.59)
Tamoxifen therapy㛳 0 8 3 315 2.67 (0.71-10.05)
Gynecologic surgery㛳 0 11 1 314 11.00 (1.42-85.20)

*+/+ indicates that both the case and the control had the characteristic, +/−, that the case had the characteristic while the control did not, −/+, that the case did
not have the characteristic while the control did, and −/−, that neither the case nor the control had the characteristic.
†Missing height, weight, and body mass index values were imputed.
‡Data on smoking status were missing among 88 case-control pairs.
§Methods were adapted from Fleiss19 to estimate the odds ratio and 95% CI.
㛳Women only.

that the risk associated with nursing home or chronic care ously reported,5,7 likely a result of differences in study
facility confinement represents immobilization, since many design. Clinical trial data support the high risk of VTE
such residents remain mobile. Further studies of the risk associated with surgery.3 Recent trauma was the next most
associated with these residents are warranted, since pro- potent risk for VTE in our study, with nearly a 13-fold
phylaxis is seldom provided. increase in risk. Both autopsy4,5 and cohort22 studies sup-
The VTE risk was highest among patients who were port an increased risk of VTE with trauma.
hospitalized with previous surgery, with nearly a 22- We found a 4-fold increased risk of VTE among pa-
fold increased risk. This risk is much higher than previ- tients with malignant neoplasm alone, which is similar

(REPRINTED) ARCH INTERN MED/ VOL 160, MAR 27, 2000 WWW.ARCHINTERNMED.COM
812

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a Purdue University User on 05/21/2015

 Surgery
Table 2. Multivariate Logistic Analysis of Potential Risk Trauma
Factors for Definite Deep Vein Thrombosis or Pulmonary Inpatient
Embolism Among Olmsted County, Minnesota, Residents Malignant Neoplasm With Chemotherapy
With a First Lifetime Venous Thromboembolism (VTE) Malignant Neoplasm Without Chemotherapy
Diagnosed From 1966 Through 1990 Compared Central Venous Catheter or Pacemaker
With Community Controls* Neurologic Disease
Superficial Vein Thrombosis
Odds Ratio Varicose Veins (Age 45 y)
(95% Confidence Varicose Veins (Age 60 y)
Baseline Characteristic Interval) Varicose Veins (Age 70 y)
CHF, VTE Incidental on Autopsy
Institutionalization with or without
CHF, Antemortem VTE, or VTE Causal of Death
recent surgery
Liver Disease
No institutionalization or 1.00
recent surgery 0 5 10 15 20 25 50
Institutionalization without 7.98 (4.49-14.18) Odds Ratio (95% Confidence Interval)
recent surgery
Odds ratios (and 95% confidence intervals) of risk factors for definite deep
Institutionalization with recent 21.72 (9.44-49.93)
vein thrombosis or pulmonary embolism among Olmsted County,
surgery Minnesota, residents with a first lifetime venous thromboembolism (VTE)
Trauma 12.69 (4.06-39.66) diagnosed from 1976 through 1990. CHF indicates congestive heart failure.
No malignant neoplasm 1.00
Malignant neoplasm without 4.05 (1.93-8.52)
chemotherapy logic disease and paralyzed legs 2 6 - 2 8 and among
Malignant neoplasm with chemotherapy 6.53 (2.11-20.23) nonambulatory patients who had strokes is increased.29
Prior central venous catheter or 5.55 (1.57-19.58) Our study is the first to document the magnitude of
transvenous pacemaker
risk for PE or upper-extremity DVT among patients with
Prior superficial vein thrombosis 4.32 (1.76-10.61)
Neurologic disease with extremity 3.04 (1.25-7.38)
a current or recent central venous catheter or a transve-
paresis nous pacemaker, as well as the first to identify superficial
Varicose veins vein thrombosis and varicose veins as independent risk
Age 45 y factors for VTE. Although the risk associated with cen-
No varicose veins 1.00 tral venous catheterization may reflect risk from comor-
Varicose veins 4.19 (1.56-11.30)
bid conditions, we controlled for most conditions in which
Age 60 y
No varicose veins 1.00
such catheters likely would be used. We found that the
Varicose veins 1.93 (1.03-3.61) risk associated with varicose veins varied by patient age.
Age 75 y For example, 45-year-old patients with varicose veins had
No varicose veins 1.00 a 4-fold increased risk of VTE compared with a 2-fold in-
Varicose veins 0.88 (0.55-1.43) creased risk for 60-year-old patients and no increased risk
CHF and VTE for 75-year-old patients. In contrast, varicose veins were
Antemortem-discovered VTE or
postmortem-discovered VTE
not an independent predictor of major PE discovered at
categorized as a cause of death autopsy in the Framingham Study.6 Similarly, varicose veins
No CHF 1.00 were not an independent risk factor for DVT among out-
CHF 1.36 (0.69-2.68) patients with no recent trauma, surgery, or immobiliza-
Postmortem-discovered VTE tion.8 Aside from differences in study population and de-
categorized as not a cause of death
sign, we have no explanation for this discrepancy. It is likely
No CHF 1.00
CHF 9.64 (2.44-38.10) that varicose veins among the young are caused by an in-
Serious liver disease 0.10 (0.01-0.71) herited connective-tissue defect.30
Interestingly, serious liver disease was associated with
*CHF indicates congestive heart failure. a 90% decrease in risk for VTE in our population. This
finding is biologically plausible. Patients with serious liver
disease often have prolonged clotting times, reduced clear-
to the increase in PE prevalence among patients with can- ance of fibrin degradation products, and thrombocyto-
cer who underwent autopsy.4,5,23 Moreover, malignant penia. Together, these impairments of normal hemosta-
neoplasm has been reported as an independent risk fac- sis may act to protect patients from VTE.
tor for outpatient-acquired DVT.8 Patients with cancer Congestive heart failure was a risk factor for post-
receiving immunosuppressive or cytotoxic chemo- mortem VTE that was not a cause of death, but not for
therapy were at an even higher risk for VTE. Previous VTE that was either manifest before death or catego-
studies have shown an increased risk of VTE among pa- rized as a cause of death. Other cardiac disease was also
tients with breast cancer who were receiving chemo- not an independent risk factor for VTE. Our findings con-
therapy.24,25 trast with autopsy studies that demonstrate an in-
Patients with neurologic disease and extremity pa- creased prevalence of PE among patients who are dying
resis or plegia had a 3-fold increased risk for VTE that from cardiac disease, especially cardiac disease causing
was independent of hospital confinement. Consistent with congestive heart failure.4,5 In addition, Cogo et al8 found
this finding, the prevalence of PE is increased among pa- heart failure to be an independent risk factor for DVT
tients with paraplegia or quadriplegia who underwent au- among outpatients with no recent trauma, surgery, or im-
topsy,5 and the risk of DVT among patients with neuro- mobilization. We cannot exclude the possibility of de-

(REPRINTED) ARCH INTERN MED/ VOL 160, MAR 27, 2000 WWW.ARCHINTERNMED.COM
813

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a Purdue University User on 05/21/2015

 tection bias and misclassification of cause of death among cluded in our final model be present in 70% or more of
patients with congestive heart failure whose PE was dis- separate bootstrap validations, thus reducing the chance
covered on autopsy. In our study, patients who were dy- of a type I error. Finally, we evaluated a large number of
ing of congestive heart failure may have been more likely baseline characteristics as potential risk factors for VTE,
to undergo autopsy and thus have an autopsy- including all interactions.
discovered PE. However, the attending pathologist may It is also important to address potential limitations
have attributed the cause of death to congestive heart fail- of our study. We were unable to evaluate age or sex as in-
ure rather than PE. Alternatively, congestive heart fail- dependent risk factors for VTE because of our matching
ure or other cardiac disease may not be a risk factor for strategy. Several studies have either shown an increased
VTE independent of hospital confinement. incidence of VTE with increased age1,5,37-40 or identified in-
Neither BMI nor current or past tobacco smoking creased age as an independent risk factor for DVT.8 The
was an independent risk factor for VTE. While previous risk associated with sex remains uncertain; previous stud-
studies reported increased risk caused by obesity4,5,8,9 and ies have reported either no difference in autopsy-
smoking,9 most of these earlier studies failed to control discovered PE by sex,5 an increase in the risk for PE and
for hospital confinement or other risk factors. Although DVT among men,7,8 or an increase in the risk for PE among
our ability to identify an above-normal BMI as a risk fac- obese women.6 Because of concern regarding diagnostic
tor may have been limited because of missing weight or suspicion bias, we did not analyze the potential risk as-
height data among controls, we do not believe our re- sociated with immobilization independent of hospital or
sults are biased since most of the missing data were body nursing home confinement (eg, immobilization during pro-
height measurements. Moreover, additional analyses of longed travel). We required that all risk factors be docu-
weight as a potential risk factor did not change our con- mented in the medical record prior to the onset of the VTE
clusions. We also confirmed that chronic obstructive pul- event, and data on such immobilization could not be re-
monary disease and renal failure were not independent liably ascertained from the medical record for controls.
risk factors for VTE.7,8 A previous study found that the It is likely that our study had insufficient power to
risk of VTE among patients who underwent surgery was exclude hormone therapy, tamoxifen therapy, inflam-
less with regional (spinal or epidural) anesthesia com- matory bowel disease, systemic lupus erythematosus, or
pared with general anesthesia.31 In our univariate analy- myeloproliferative diseases as risk factors for VTE. Al-
ses, regional anesthesia was associated with an 11.5- though we collected all available data on coagulation dis-
fold increased risk, while general anesthesia was associated orders, most cases and controls were not tested for these
with a 19-fold increased risk. However, in the multivar- disorders. Moreover, the most common coagulation dis-
iate analysis, type of anesthesia was not an independent order (eg, activated protein C resistance)41 had not been
risk factor for VTE after controlling for surgery. described at the inception of our study.
Among women, pregnancy, postpartum period, oral In summary, we have described independent base-
contraceptive use, hormone therapy, and tamoxifen line characteristics that identify a population at risk for
therapy were not independent risk factors for VTE. While VTE and estimated the magnitude of risk associated with
several cohort studies showed no significant increase in each characteristic. For many of these risk factors, avail-
VTE incidence among pregnant women compared with able prophylaxis has clear benefit,3 and all affected indi-
the general population, the incidence during the post- viduals should receive appropriate prophylaxis. Addi-
partum period was increased about 2-fold.32-35 In addi- tional data regarding genetic and acquired disorders as
tion, the vast majority of evidence suggests that both oral VTE risk factors are needed in order to better assess risk
contraceptive use10 and hormone replacement ther- in the individual and to target prophylaxis.
apy11-15,36 are significant risk factors for VTE. Since only
24 patients who had definite VTE events were pregnant Accepted for publication July 14, 1999.
(n = 3), postpartum (n = 9), or taking oral contracep- This study was funded in part by grants HL46974 and
tives (n = 12) at baseline, it is likely that the power was AR30583 from the National Institutes of Health, Bethesda,
insufficient to identify these variables as independent risk Md; and by the Mayo Foundation, Rochester, Minn.
factors. We thank C. Mary Beard, RN, for assistance in manag-
We believe our results are valid because we avoided ing the study; Janet Ebersold, RN, Kay Traverse, RN, Mary Lou
the potential distortions associated with referral bias by Notermann, RN, and Susan Stotz, RN, for their untiring medi-
performing a population-based study with both cases and cal record review; Randall Stick, BS, for programming; Diana
controls selected from residents in the community. More- Rademacher, BS, and Christine Lohse, BS, for assistance with
over, all cases met strict criteria for VTE; the diagnosis data analysis; and Stephanie Wellik for secretarial support.
was confirmed by either venogram, pulmonary angio- Corresponding author: John A. Heit, MD, Hematol-
gram, or postmortem examination. We separated VTE ogy Research, Plummer 549, Mayo Clinic, 200 First St SW,
events that were detected on postmortem examination Rochester, MN 55905 (e-mail: heit.john@mayo.edu).
into clinically important VTE events that were the im-
mediate, underlying, or a contributory cause of death and REFERENCES
incidentally noted VTE events (essentially all PE events)
that were discovered on postmortem examination but not
1. Silverstein MD, Heit JA, Mohr DN, Petterson TM, O’Fallon WM, Melton LJ III.
judged by the pathologist as sufficiently important to be Trends in the incidence of deep vein thrombosis and pulmonary embolism: a
listed in either part I or part II of the death certificate. 25-year population-based study. Arch Intern Med. 1998;158:585-593.
Our modeling strategy required that risk factors in- 2. Heit JA, Silverstein MD, Mohr DN, Petterson TM, O’Fallon WM, Melton LJ III.

(REPRINTED) ARCH INTERN MED/ VOL 160, MAR 27, 2000 WWW.ARCHINTERNMED.COM
814

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a Purdue University User on 05/21/2015

 Predictors of survival after deep vein thrombosis and pulmonary embolism: a 23. Svendsen E, Karwinski B. Prevalence of pulmonary embolism at necropsy in pa-
population-based, cohort study. Arch Intern Med. 1999;159:445-453. tients with cancer. J Clin Pathol. 1989;42:135-139.
3. Clagett GP, Anderson FA Jr, Geerts W, et al. Prevention of venous thromboem- 24. Goodnough LT, Saito H, Manni A, Jones PK, Pearson OH. Increased incidence
bolism. Chest. 1998;114(5 suppl):531S-560S. of thromboembolism in stage IV breast cancer patients treated with a five-drug
4. Coon WW, Coller FA. Some epidemiologic considerations of thromboembo- chemotherapy regimen. Cancer. 1984;54:1264-1268.
lism. Surg Gynecol Obstet. 1959;109:487-501. 25. Levine MN, Gent M, Hirsh J, et al. The thrombogenic effect of anti-cancer drug
5. Coon WW. Risk factors in pulmonary embolism. Surg Gynecol Obstet. 1976; therapy in women with stage II breast cancer. N Engl J Med. 1988;318:404-407.
143:385-390. 26. Bors E, Conrad CA, Masselli TB. Venous occlusion of lower extremities in paraple-
6. Goldhaber SZ, Savage DD, Garrison RJ, et al. Risk factors for pulmonary em- gic patients. Surg Gynecol Obstet. 1954;99:451-454.
bolism: the Framingham Study. Am J Med. 1983;74:1023-1028. 27. Warlow C, Ogston D, Douglas AS. Deep venous thrombosis of the legs after stroke.
7. Quinn DA, Thompson BT, Terrin ML, et al. A prospective investigation of pul- BMJ. 1976;1:1178-1181.
monary embolism in women and men. JAMA. 1992;268:1689-1696. 28. Myllynen P, Kammonen M, Rokkananen P, et al. Deep venous thrombosis and
8. Cogo A, Bernardi E, Prandoni P, et al. Acquired risk factors for deep-vein throm- pulmonary embolism in patients with acute spinal cord injury: a comparison with
bosis in symptomatic outpatients. Arch Intern Med. 1994;154:164-168. non-paralyzed patients immobilized due to spinal fractures. J Trauma. 1985;25:
9. Goldhaber SZ, Grodstein F, Stampfer MJ, et al. A prospective study of risk fac- 541-543.
tors for pulmonary embolism in women. JAMA. 1997;277:642-645. 29. Bromfield EB, Reding MJ. Relative risk of deep venous thrombosis or pulmo-
10. Chasan-Taber L, Stampfer MJ. Epidemiology of oral contraceptives and cardio- nary embolism post-stroke based on ambulatory status. J Neurol Rehab. 1988;
vascular disease. Ann Intern Med. 1998;128:467-477. 2:51-57.
11. Devor M, Barrett-Connor E, Renvall M, et al. Estrogen replacement therapy and 30. Goldman MP, Fronek A. Anatomy and pathophysiology of varicose veins. J Der-
the risk of venous thrombosis. Am J Med. 1992;92:275-282. matol Surg Oncol. 1989;15:138-145.
12. Daly E, Vessey MP, Hawkins MM, et al. Risk of venous thromboembolism in us- 31. Sharrock NE, Haas SB, Hargett MJ, et al. Effects of epidural anesthesia on the
ers of hormone replacement therapy. Lancet. 1996;348:977-980. incidence of deep-vein thrombosis after total knee replacement. J Bone Joint Surg
13. Jick H, Derby L, Wald M, et al. Risk of hospital admission for idiopathic venous Am. 1991;73:502-506.
thromboembolism among users of postmenopausal estrogens. Lancet. 1996; 32. Treffers PE, Huidekoper BL, Weenink GH, et al. Epidemiological observations of
348:981-983. thromboembolic disease during pregnancy and in the puerperium in 56,022
14. Pérez-Gutthann S, Garcı́a-Rodrı́guez LA, Castellsague-Piqué J, et al. Hormonal women. Int J Gynaecol Obstet. 1983;21:327-331.
replacement therapy and the risk of venous thromboembolic events: a population- 33. Weekes LR, Deukmedjian AG. Thromboembolic disease in pregnancy. Am J Ob-
based case-control study. BMJ. 1997;314:796-800. stet Gynecol. 1970; 107:649-650.
15. Varas-Lorenzo C, Garcı́a-Rodrı́guez LA, Cattaruzzi C, et al. Hormone replace- 34. Husni EA, Pena LI, Lenhert AE. Thrombophlebitis in pregnancy. Am J Obstet Gy-
ment and the risk of hospitalization for venous thromboembolism: a population- necol. 1967;97:901-905.
based study in southern Europe. Am J Epidemiol. 1998;147:387-390. 35. Kierkegaard A. Incidence and diagnosis of deep vein thrombosis associated with
16. Melton LJ III. History of the Rochester Epidemiology Project. Mayo Clin Proc. pregnancy. Acta Obstet Gynecol Scand. 1983;62:239-243.
1996;71:266-274. 36. Grodstein F, Stampfer MJ, Goldhaber SZ, et al. Prospective study of exogenous
17. Kurland LT, Molgaard DA. The patient record in epidemiology. Sci Am. October hormones as risk of pulmonary embolism in women. Lancet. 1996;348:983-987.
1981;245:54-63. 37. Morrell MT, Truelove SC, Barr A. Pulmonary embolism. BMJ. 1963;2:830-835.
18. Saerbrei W, Schumacher M. A bootstrap resampling procedure for model build- 38. Coon WW, Willis PW III, Keller JB. Venous thromboembolism and other venous
ing: application to the Cox regression model. Stat Med. 1992;11:2093-2109. disease in the Tecumseh community health study. Circulation. 1973;48:839-846.
19. Fleiss JL. Statistical Methods for Rates and Proportions. 2nd ed. New York, NY: 39. Kierkegaard A. Incidence of acute deep venous thrombosis in two districts: a phleb-
John Wiley & Sons Inc; 1981:48-49. ographic study. Acta Chir Scand. 1980;146:267-269.
20. Gibbs NM. Venous thrombosis of the lower limbs with particular reference to 40. Anderson FA, Wheeler HB, Goldberg RJ, et al. A population-based perspective
bed rest. Br J Surg. 1957;191:209-236. of the hospital incidence and case-fatality rates of deep vein thrombosis and pul-
21. Gross JS, Neufeld RR, Libow LW, Gerber I, Rodstein M. Autopsy study of the monary embolism: the Worcester DVT Study. Arch Intern Med. 1991;151:933-
elderly institutionalized patient: review of 234 autopsies. Arch Intern Med. 1988; 938.
148:173-176. 41. Koster T, Rosendaal FR, de Ronde H, et al. Venous thrombosis due to poor an-
22. Geerts WH, Code KI, Jay RM, et al. A prospective study of venous thromboem- ticoagulant response to activated protein C: the Leiden Thrombophilia Study. Lan-
bolism after major trauma. N Engl J Med. 1994;331:1601-1606. cet. 1993;342:1503-1506.

(REPRINTED) ARCH INTERN MED/ VOL 160, MAR 27, 2000 WWW.ARCHINTERNMED.COM
815

©2000 American Medical Association. All rights reserved.

Downloaded From: http://archinte.jamanetwork.com/ by a Purdue University User on 05/21/2015