|Home | About | Journals | Submit | Contact Us | Français|
While diabetes mellitus is reported as a risk factor for venous thromboembolism (VTE), persons with diabetes are frequently hospitalized for medical illness or surgery, or confined to a nursing home, all major VTE risk factors. Consequently, we tested diabetes for an independent association with incident VTE.
Using Rochester Epidemiology Project resources, we identified all Olmsted County, MN residents who met objective criteria for incident VTE over the 25-year period, 1976-2000 (n=1922),and one to two resident controls per case, matched on age, gender and length of medical history (n=2115). Complete medical histories in the community were reviewed for previously identified independent VTE risk factors and diabetes-related variables. We tested diabetes and diabetes complications (retinopathy, nephropathy or neuropathy; and ketoacidosis) as potential VTE risk factors, both alone and after adjusting for VTE risk factors. Univariately, diabetes by clinical diagnosis or by stricter criteria (fasting ambulatory blood glucose ≥140 mg/dL and/or anti-diabetic drug therapy), and diabetes complications, were associated with overall incident VTE. However, after controlling for hospitalization for major surgery or medical illness and nursing home confinement, diabetes was no longer associated with VTE.
Diabetes mellitus and diabetes complications are not independent risk factors for incident VTE.
Diabetes mellitus is a reported risk factor for VTE1-3 and pulmonary embolism (PE),4 and a meta-analysis estimated a 1.4-fold increased risk of VTE for persons with diabetes.5 However, analyses of data from the Nurses' Health Study and from the Heart and Estrogen/progestin Replacement Study (HERS) failed to find an association between diabetes and PE or VTE, respectively.6,7 This could be explained by the fact that persons with diabetes are frequently hospitalized for major surgery or acute medical illness, or confined to a nursing home or chronic rehabilitation facility, all of which are major risk factors for incident VTE.8,9 Consequently, we hypothesized that diabetes is not an independent risk factor for incident VTE after controlling for previously identified VTE risk factors.
We performed a population-based case-control study to test diabetes mellitus for an association with incident VTE, both alone and after controlling for other VTE risk factors. We also tested the association of diabetes with incident idiopathic VTE. We wished to test the entire spectrum of diabetes mellitus occurring in the community for an association with VTE. Therefore, for the purposes of this study, we used a very broad definition of diabetes (i.e., any mention of diabetes within the complete medical record prior to the incident VTE), and a very strict but possibly less sensitive definition of diabetes (i.e., evidence of any ambulatory [outpatient] fasting blood glucose ≥140 mg/dL and/or anti-diabetic drug therapy prior to the incident VTE).10-12
Diabetes may be complicated by microvascular occlusive disease, manifest as diabetic retinopathy, nephropathy or neuropathy. The same process that causes such arterial vascular disease among persons with diabetes has been suggested to cause VTE.3 Consequently, we also tested microvascular complications of diabetes (i.e., retinopathy, nephropathy or neuropathy) for an association with VTE. Finally, persons with diabetes who are susceptible to ketoacidosis may have yet a different process that could cause VTE. Therefore, we also tested for an association between diabetic ketoacidosis and VTE.
Using the population-based resources of the Rochester Epidemiology Project,13 we identified all Olmsted County, MN residents with incident DVT and/or PE over the 35-year period, 1966-2000, as previously described.14,15 We then performed a matched case-control study nested within the Olmsted County population. All Olmsted County residents with a first lifetime objectively-diagnosed DVT or PE during the 25-year period, 1976 to 2000, were included in the present study as VTE cases. The Rochester Epidemiology Project also provides an enumeration of the entire Olmsted County population from which controls can be sampled.13 Using this system, one to two age- (± 5 years) and sex-matched Olmsted County residents who had a episode of medical care within ± 1 year of the case event date and whose medical record number was closest to the case's medical record number were selected as controls as previously described.8 Since a patient's medical record number is assigned sequentially and in perpetuity, matching on medical record number assures a similar length of prior medical history among cases and matched controls. The study was approved by the Mayo Clinic Institutional Review Board.
Using explicit criteria, trained and experienced nurse abstractors reviewed all medical records (inpatient and outpatient) in the community for cases and controls who provided consent to review of their medical records for research purposes. All records were reviewed from date first seen by a REP healthcare provider until the earliest of death, date of last medical record follow-up, or 200016 as previously performed.8,14 For cases, data were recorded on the method of diagnosis and type of incident VTE event (DVT, PE or both; chronic thromboembolic pulmonary hypertension). A DVT was categorized as objectively-diagnosed when symptoms or signs of acute DVT were present and the diagnosis was confirmed by venography, compression venous duplex ultrasonography, impedance plethysmography, computed tomographic venography, magnetic resonance imaging or pathology examination of thrombus removed at surgery or autopsy. A PE was categorized as objectively-diagnosed when symptoms and/or signs of acute PE were present and the diagnosed was confirmed by pulmonary angiography, a ventilation/perfusion lung scan interpreted as high probability for PE, computed tomographic pulmonary angiography, magnetic resonance imaging or pathology examination of thrombus removed at surgery or autopsy. Mayo Clinic pathologists performed all autopsy examinations and completed the death certificates of persons dying within Olmsted County during the study period.
For Olmsted County residents meeting our criteria for objectively-diagnosed DVT or PE and matched controls, the study nurses also collected data from the medical record on year of incident event (cases) or index episode of care (controls); patient age at incident event (cases) or index episode of care (controls); gender; patient location at incident event onset (cases) or index episode of care (controls) (three categories, defined as community-dwelling, confined to a hospital or community-dwelling but hospitalized in the previous 90 days, or confined to a nursing home [including chronic rehabilitation facility]); body mass index (BMI; kg/m2); active cancer (excluding non-melanoma skin cancer); serious neurologic disease (stroke or other disease affecting the nervous system with associated leg paresis, or acute stroke with leg paresis requiring hospitalization within the previous three months); surgery requiring anesthesia; trauma requiring hospital admission (major fracture or severe soft tissue injury); prior superficial vein thrombosis; varicose veins (varicose veins, or treated varicose veins [injection sclerotherapy or stripping]); hormone therapy (estrogen or progesterone); and for women: pregnancy or postpartum at the time of the incident event; oral contraceptive use; and gynecologic surgery. We also collected data on ambulatory blood glucose (value, date of assay, and fasting status); diabetes mellitus therapy (drug[s], date of therapy); and a physician diagnosis of diabetes and diabetes complications (retinopathy, neuropathy, nephropathy, and ketoacidosis). Serious neurologic disease with leg paresis, all surgery variables, anesthesia, trauma, hormone therapy, and oral contraceptive use had to have been documented in the three months before the incident VTE event for cases or the index episode of medical care for controls. Active cancer had to have been documented in the three months before or after the incident VTE event (cases) or the index episode of care (controls). Location at VTE event date for cases, or index event date for controls, was determined as follows. If a patient had the VTE in hospital, the patient was designated as “hospitalized” even if also a nursing home resident. If the patient had the VTE in the nursing home (NH) the patient was designated as a NH patient even if hospitalized in the previous 90 days. If the patient had the VTE in the community but had been hospitalized in the preceding 90 days, the patient was designated as community, hospitalized in the preceding 90 days. For the purposes of this analysis, in-hospital and hospitalized within the previous 90 days were combined; nursing home was left separate. Finally, if a hospitalized patient had a surgery in the previous 90 days, the patient was designated as having a surgery. For this analysis, only those in hospital without surgery in the previous 90 days were considered hospitalized with acute medical illness. Diabetes mellitus and diabetes complications, prior superficial vein thrombosis, and varicose veins could be documented any time before the incident VTE event (cases) or index episode of medical care (controls). BMI was based on the most recent height and weight measurements before the incident VTE event (cases) or index episode of care (controls).
For the present study, diabetes mellitus was defined and analyzed in multiple ways. First, persons with any physician diagnosis of diabetes recorded in the medical record at any time before the incident VTE event (cases) or index episode of care (controls) were categorized as having diabetes by clinical diagnosis. Secondly, a more strict definition of diabetes was based on review of all medical records (including all laboratory blood glucose values) before the incident event (cases) or index episode of care (controls) for any ambulatory blood glucose value ≥140 mg/dL (excluding gestational diabetes),10-12 and/or use of anti-diabetic drug therapy; subjects meeting these stricter criteria were categorized as having diabetes by strict blood glucose criteria. Finally, subjects were categorized as having a diagnosis of diabetic microvascular complications (retinopathy, neuropathy or nephropathy) and, separately as having a diagnosis of diabetic ketoacidosis.
Using conditional logistic regression we tested the association of diabetes with VTE, where diabetes was defined both by clinical diagnosis and by the stricter blood glucose criteria. For each definition of diabetes, we also tested the association of VTE and the three degree of freedom variable of diabetes with microvascular complications, diabetes with ketoacidosis and diabetes with no complications, versus no diabetes. All of the above definitions of diabetes were modeled using conditional logistic regression unadjusted for other risk factors, after individually adjusting for each baseline characteristic, and after adjusting for all important covariates jointly in a multivariable model. All but the first were also adjusted for age at event, a matching variable. Specifically, for each definition of diabetes and diabetes complications, we performed a series of bivariate analyses in order to estimate the effect of each previously-identified independent VTE risk factor (i.e., patient age, BMI, current or recent hospitalization within the past 90 days with and without surgery, nursing home confinement, trauma or fracture, active cancer, neurologic disease with leg paresis, prior superficial vein thrombosis, varicose veins)8 on the odds of VTE for those with diabetes and with diabetes complications. Because patients initially hospitalized for acute medical illness often undergo major surgery and/or are confined to a nursing home or chronic care facility, these three baseline characteristics often are not mutually exclusive. We previously showed that the combination of hospitalization for acute medical illness, major surgery or nursing home confinement is a potent and independent VTE risk factor.8 Consequently, for each definition of diabetes and diabetes complications, we also performed a bivariate analysis to estimate the effect of the combination of hospitalization for acute medical illness, major surgery or nursing home confinement on the odds of VTE for those with diabetes and with diabetes complications. Finally, for each definition of diabetes and diabetes complications we included diabetes in a multivariate model, adjusting for all of the above VTE risk factors. Separately, we limited our analysis to the subset of incident VTE cases with idiopathic VTE and their matched controls, and used conditional logistic regression to test the association of diabetes and diabetic complications with idiopathic VTE. Idiopathic VTE was defined as objectively-diagnosed DVT or PE occurring in the absence of hospitalization for surgery or for acute medical illness, nursing home confinement, trauma, fracture, active cancer, neurological disease with leg paresis, pregnancy or the postpartum period, oral contraceptive use or hormone therapy (estrogen and/or progesterone) in the three months preceding the VTE event date. Each definition of diabetes was examined.
In order to insure that we tested diabetes for an association with the full spectrum of VTE event types occurring in the community, we also repeated our analyses after restricting VTE cases and matched controls to the 13-year period, 1988-2000, a timeframe when over 90% of all incident VTE events occurring in Olmsted County were objectively diagnosed.
Over the 25-year period, 1976-2000, 2,367 residents of Olmsted County developed a first lifetime DVT or PE. Of these, 1922 (81%) incident VTE events were objectively-diagnosed and made up the incident VTE cases for this study. The distribution of VTE events by event type was 980 (51%) DVT alone, 937 (48.8%) PE with or without DVT, and 5 (0.3%) chronic thromboembolic pulmonary hypertension; 329 of 1992 (17.1%) were diagnosed at autopsy (310 [16.1%] of which were PE), and 1035 (53.9%) were women. The mean ± SD age of the cases and matched controls (n=2115) was 64.8 ± 19.3 and 64.6 ± 19.2 years, respectively, and did not differ significantly (Table 1). The overall mean ± SD duration of prior medical record documentation was 35.4 ± 20.6 and 35.3 ± 20.7 years, respectively, for cases and controls.
Among all 1922 cases and 2115 controls, respectively, 231 (12.0%) and 199 (9.4%) had a clinical diagnosis of diabetes mellitus; 208 (10.8%) and 163 (7.7%) met strict blood glucose criteria for diabetes; 64 (3.3%) and 52 (2.5%) had a diagnosis of diabetic retinopathy, nephropathy or neuropathy; and 7 (0.4%) and 3 (0.1%) had a diagnosis of diabetic ketoacidosis (Table 1). Univariately (unadjusted for other VTE risk factors), a clinical diagnosis of diabetes was associated with overall incident VTE (odds ratio [OR]=1.32; 95% CI: (1.07, 1.63; p=0.009) as was diabetes based on strict blood glucose criteria (OR=1.47; 95% CI: 1.18, 1.84; p=0.001) (Tables 2 and and3).3). In addition, clinically-diagnosed diabetes with microvascular complications and with ketoacidosis were marginally associated with VTE (OR=1.39, 95% CI: 0.95, 2.03; p=0.09; and OR=3.19, 95% CI: 0.81, 12.59; p=0.10; respectively; OR=1.26, 95% CI: 0.99, 1.60 for diabetes without complications, p=0.06, model p-value=0.03). Likewise, diabetes by strict blood glucose criteria with microvascular complications and with ketoacidosis were marginally associated with VTE (OR=1.39, 95% CI: 0.95, 2.04; p=0.09; and OR=3.23, 95% CI: 0.82, 12.74; p=0.09; respectively; OR=1.46, 95% CI: 1.12, 1.90 for diabetes without complications, p=0.005, model p-value=0.004).
In the bivariate analyses, diabetes by clinical diagnosis remained associated with VTE after controlling individually for age (matching variable), and for BMI, hospitalization for surgery, nursing home confinement, active cancer, trauma or fracture, neurologic disease with leg paresis, prior superficial vein thrombosis and varicose veins (Table 2). Diabetes remained marginally associated with VTE after controlling for hospitalization for acute medical illness. However, the proportions of patients with recent hospitalization for surgery or for acute medical illness, or nursing home confinement were markedly higher among VTE cases with diabetes compared to VTE controls with diabetes (Table 4). In a bivariate analysis of diabetes after controlling for hospitalization for major surgery or for acute medical illness or nursing home confinement, the association of diabetes with VTE was lost (p=0.68; Table 2). In similar bivariate analyses, diabetes by strict blood glucose criteria remained significantly associated with VTE after controlling individually for each VTE risk factor (Table 3). Again, however, after controlling for hospitalization for major surgery or for acute medical illness or nursing home confinement, diabetes by strict blood glucose criteria also was not associated with VTE (p=0.82; Table 3). Diabetic complications behaved similarly for both definitions of diabetes; after adjusting for hospitalization for major surgery or for acute medical illness or nursing home confinement, the association with VTE was lost (p=0.59 and 0.70 for clinically diagnosed diabetes and more strictly defined diabetes, respectively, data not shown).
In the multivariate models that included diabetes along with all previously identified VTE risk factors, we did not find an association between VTE and diabetes, defined either by clinical diagnosis or by strict blood glucose criteria (p=0.59 and p=0.87; Tables 2 and and3,3, respectively). For both cases and controls, a larger percentage of those cared for in hospital or nursing home in the previous three months had clinically diagnosed diabetes than those dwelling in the community who had no recent hospitalization: 15% of the former compared to 8% of the latter. Cases and controls had nearly identical percentages of diabetes once this was accounted for (14.9% of cases and 16.2% of controls for those recently hospitalized or confined to a nursing home; 7.7% of cases and 8.2% of controls for those not; Table 5). Nearly identical percentages were observed using strict blood glucose criteria as well (Table 5). Diabetes appeared to be associated with VTE status only because so many more VTE cases were cared for in hospital or nursing home in the previous three months than controls (60% of cases compared to 15% of controls).
Over the 25-year period, 1976-2000, 458 of the 1922 objectively diagnosed incident VTE were idiopathic VTE and were matched to 518 controls as part of the larger study. Using conditional logistic regression, neither clinically diagnosed diabetes nor diabetes by strict blood glucose criteria were associated with idiopathic incident VTE (OR=1.07, 95% CI: 0.69, 1.67; p=0.76; and OR=1.22, 95% CI: 0.76, 1.96; p=0.40, respectively). Diabetic complications were not associated with idiopathic VTE using either definition (all p-values ≥ 0.20).
Over the more recent period, 1988-2000, 1374 Olmsted Count residents developed an incident DVT or PE; of these, 1277 (92.9%) were objectively diagnosed. Analyses restricted to 1988-2000 objectively diagnosed incident VTE cases and matched controls (n=1470) yielded similar findings. Univariately, diabetes by clinical diagnosis and by strict blood glucose criteria remained associated with VTE (OR=1.31, 95% CI: 1.02, 1.69, p=0.04; and OR=1.42, 95% CI, 1.09, 1.85, p=0.009, respectively). However, diabetes by clinical diagnosis and by strict blood glucose criteria was not associated with VTE after controlling for institutionalization or all VTE risk factors (data not shown).
Similar to previous studies,1,2 we univariately identified diabetes mellitus as a risk factor for incident VTE. Also similar to previous studies,1 we found a slight attenuation of the apparent risk of VTE with diabetes after controlling for BMI, suggesting that at least some of association with diabetes is related to the higher prevalence of obesity among persons with diabetes. However, the VTE risk with diabetes was further attenuated after controlling for hospitalization for acute medical illness and for nursing home confinement, and completely lost after controlling for the combination of hospitalization for surgery or for acute medical illness, or nursing home confinement. We confirmed our hypothesis that persons with diabetes are more frequently hospitalized or confined to a nursing home, and that it is these latter exposures that place diabetic individuals at risk for VTE.
In support of our findings, diabetes was not a risk factor among outpatients with acute deep vein thrombosis (DVT),17 and the seven-day incidence of DVT did not differ significantly among hospitalized diabetic individuals compared to hospitalized non-diabetic controls.18 Furthermore, analyses of data from the Physicians Health Study19 and a case-control study of patients referred to three Italian tertiary care hospitals 20 failed to find diabetes as a risk factor for unprovoked or idiopathic VTE.
Microvascular occlusive disease, manifest as retinopathy, nephropathy or neuropathy, is a major complication of diabetes mellitus. We hypothesized that the same pathologic mechanisms that cause diabetic microvascular occlusive disease might also cause VTE. However, we were unable to find an association between diabetic microvascular disease (as reflected by diabetic retinopathy, nephropathy or neuropathy) and VTE. Consequently, while we cannot completely exclude this possibility, we believe that diabetes mellitus is not a pathologic mechanism for VTE.
We were unable to identify diabetic ketoacidosis as a risk factor for VTE, but cannot exclude such an association since our sample size and resulting power were quite low. However, in a retrospective cohort study using discharge diagnosis codes from the California Patient Discharge Data Set, hyperosmolarity (but not ketoacidosis) among diabetics admitted with marked hyperglycemia was a risk factor for in-hospital VTE or readmission for VTE in the subsequent three months.21 We hypothesize that the pathologic mechanism for VTE in these diabetic individuals is the hyperosmolarity rather than other vascular or hemostatic changes related to diabetes. In support of this hypothesis, increased serum osmolality also is a risk factor for VTE among patients admitted with acute ischemic stroke.22
The metabolic syndrome, defined as abdominal obesity, elevated blood pressure, high triglycerides, reduced levels of high-density lipoprotein cholesterol and elevated fasting plasma glucose,23,24 has been associated with idiopathic DVT20 and with idiopathic VTE.25 However, in the first study, only abdominal obesity and hypertriglyceridemia were univariately associated with idiopathic DVT while hyperglycemia (blood glucose >110 mg/dL) was not associated.20 In the second study, hypertriglyceridemia again was univariately associated with idiopathic VTE, while hyperglycemia (blood glucose >100 mg/dL) was only marginally associated.25 Finally, in a third study reporting an association between the metabolic syndrome and DVT among patients admitted to a cardiac rehabilitation facility, only hypertriglyceridemia (but not hyperglycemia) was univariately associated with DVT.26 While we cannot exclude the metabolic syndrome as a risk factor for VTE, we believe any such risk is primarily conveyed by hypertriglyceridemia and not hyperglycemia.
Our results are likely to be valid. Since our study sample size is several fold larger than previous studies which identified diabetes as a VTE risk factor, it is unlikely that we failed to identify an association of VTE with diabetes due to inadequate power. We also avoided the potential distortions associated with referral bias by performing a population-based study with both cases and controls selected from residents in the community. All cases met strict criteria for acute DVT and/or PE, and we performed analyses using two different criteria for diabetes mellitus (diabetes by clinical diagnosis and by rigorous blood glucose criteria) which are likely to encompass all diabetes mellitus occurring in the community. Moreover, we were unable to identify diabetic complications (i.e., clear evidence of other diabetic vascular disease such as retinopathy, nephropathy or neuropathy) as a risk factor for VTE. Finally, we tested diabetes as a risk factor for incident VTE over a timeframe (1988-2000) where essentially the entire spectrum of VTE events occurring in the community was included since over 90% of events in this timeframe were objectively diagnosed, and we failed to find an independent association of diabetes with incident VTE.
It is also important to address potential limitations of our study. Since the racial and ethnic demography of Olmsted County is predominantly white of non-Hispanic ancestry, our findings may not be generalizeable to populations of other races or ethnicities. Of note, however, diabetes was not a risk factor for VTE among whites or African-Americans in a single previous study.27 We also required that all risk factors be documented in the medical record prior to the onset of the VTE event and data on such characteristics as immobilization or prolonged travel could not be reliably ascertained from the medical record for controls.
In conclusion, diabetes mellitus and diabetes microvascular complications are not associated with overall or idiopathic incident VTE. Some have argued that reducing diabetes prevalence should be a priority as a means of reducing VTE incidence.1,3,5 However, our findings suggest that the more direct path to reducing VTE incidence is to reduce the need for surgery, hospitalization for acute medical illness or nursing home confinement among persons with diabetes.
Sources of Funding. This study was funded, in part, by grants from the National Institutes of Health (HL66216, AR30582), U.S. Public Health Service; the American Diabetes Association, and by Mayo Foundation
Disclosures. Dr. Heit has received a research grant from Sanofi-Aventis.