Diabetes increases the risk of deep-vein thrombosis and pulmonary embolism. A population-based cohort study.

Diabetes increases the risk of deep-vein

thrombosis and pulmonary

embolism

A population-based cohort study

Wei-Sheng Chung1,2; Cheng-Li Lin3,4; Chia-Hung Kao5,6

Introduction

A global diabetes epidemic is underway. The world prevalence of diabetes among adults is approximately 6.4 %, affecting 285 million people in 2010, and will increase to 7.7 %, and 439 million

adults by 2030 (1). In accordance with the global trend for diabetes, Taiwan sustained an increase of more than 70 % in the total

diabetic population and a 35 % increase in the standardised prevalence rate from 2000 to 2009 (2).

Most studies have focused on the complications of diabetes, such as coronary artery and peripheral vascular disease (3-6), cerebrovascular accidents (CVA), diabetic neuropathy, amputations (7), renal failure, and blindness, because they result in increased disability, reduced life expectancy, and enormous health

costs for societies worldwide.

Deep-vein thrombosis (DVT) is the formation of blood clots in the deep veins, with thrombi predominantly occurring in the legs. Pulmonary thromboembolism (PE) is a potentially life-threatening disorder involving embolic or thrombotic occlusion of the pulmonary arterial system. Together, DVT and PE constitute venous

thromboembolism (VTE). VTE is not a benign disease and has a 30-day case fatality rate of 11 % to 30 % (8-10).

According to the Virchow triad, VTE is a condition that results from altered blood coagulation, stasis, or abnormalities in the vessel wall (11, 12). Previous studies have identified multiple risk factors for VTE, including age, atrial fibrillation, and CVA (11, 12). Congestive heart failure, leg fractures, and major surgery are also correlated with an increased risk of VTE (13, 14). Numerous cancers have also been shown to correlate with VTE (15, 16).

However, studies on the association between diabetes mellitus (DM) and DVT/PE have yielded inconsistent findings (17-19).

Therefore, we conducted a population-based cohort study to investigate whether DM increases the risks of DVT and PE in

Taiwan.

_Methods

Data source

The National Health Insurance (NHI) program in Taiwan has operated since 1995 and covers more than 99 % of 23.74 million Taiwan residents (20). We retrieved our study sample from the Longitudinal Health Insurance Database 2000 (LHID2000), which consists

of claims data of one million people randomly sampled from

the registry of beneficiaries of the NHI program in 2000. The random samples released by the National Health Research Institutes

(NHRI) have been confirmed to be representative of the Taiwanese population. The NHRI followed all of the aforementioned randomly sampled patients to the year 2011. The LHID2000 contains

medical data identified by encrypted patient numbers, including sex, birthdate, admission dates, discharge dates, International Classification of Diseases, Ninth Revision, Clinical Modification

(ICD-9-CM) diagnostic and procedure codes, and prescription details. Numerous studies have indicated a high accuracy and validity

of the ICD-9-CM diagnoses in the LHID (21, 22). This study

was approved by the Institutional Review Board of China Medical University (CMU-REC-101–012).

Sampled participants

From the LHID2000, we identified patients aged 20 years and older with newly diagnosed type 2 DM (T2DM, ICD-9-CM code 250.x0 and 250.x2) from 2000 to 2011 as the DM cohort. We used the date of T2DM diagnosis as the patient’s index date. Patients with DVT (ICD-9-CM code 453.8) or PE (ICD-9-CM code 415.1; not including iatrogenic PE [ICD-9-CM code 415.11]) at baseline and those with missing information on age or sex were excluded. For each T2DM case, three insured people who did not have a history of DM or DVT and PE were identified as the comparison cohort, and were frequency matched by age (in five-year bands), sex, and index year.

Outcome

new diagnosis of DVT or PE during the follow-up period. Each participant was monitored from the index date to diagnosis of

DVT or PE, or to censor because of loss of follow-up, death, withdrawal from insurance, or to the end of December 31, 2011. In Taiwan,

nearly all patients with DVT and PE underwent comprehensive examinations before receiving anticoagulants and intensive care unless these patients were asymptomatic or undiagnosed. Therefore, the patients with DVT or PE can be extracted in the LHID. Furthermore, the National Health Insurance Administration (NHIA) scrutinised the medical reimbursements and discharge notes of the patients by using a peer-review process.

Definition of unprovoked VTE

Unprovoked VTE was defined as the VTE that occurs in the absence of triggering circumstances as follows: a lower leg fracture

and/or plaster cast immobilisation, surgery under general anesthesia, or bed rest for more than three days, CVA, spinal cord injury,

and active cancer, pregnancy or post–partum.16

Comorbidities

Comorbidities and surgery included in our study were atrial fibrillation (ICD-9-CM code 427.3), hypertension (ICD-9-CM codes

401–405), hyperlipidaemia 9-CM code 272), CVA (ICD-9-CM codes 430–438), heart failure (ICD-(ICD-9-CM code 428), lower leg fracture or surgery (ICD-9-CM codes 820–823 and procedure codes 81.51, 81.52, 81.53, and 81.54, respectively), cancer (ICD-9-CM codes 140–208), pregnancy (ICD-(ICD-9-CM codes

640.x1–676.x1, 640.x2–676.x2, 650–659 and ICD-9-CM procedure codes 72–74, respectively), coronary heart disease (ICD-9-CM codes 410–414), other peripheral vascular shunt or bypass (ICD-9-CM CP 39.29), amputation of lower limb (ICD-(ICD-9-CM CP 84.1), and angioplasty or atherectomy of non-coronary vessel (ICD-9-CM CP 39.50). Charlson-comorbidity index (CCI) score was also counted for each participant from claims data for outpatient visits or hospitalisations at baseline.

Statistical analysis

Distributions in demographic variables, including age, sex, CCI score, comorbidities and surgery were compared between the two cohorts. Baseline characteristics of the two cohorts were compared using standardised mean differences, calculated as the difference in

means or proportions of a variable divided by a pooled estimate of

the standard deviation of variable. A value of standardised mean differences equals 0.05 or less, which indicates a negligible difference in

means between two cohorts. Incidence densities of DVT and PE by demographic variables and comorbidity were calculated. The

T2DM to comparison cohort hazard ratios (HRs) and the 95 % confidence interval (CIs) were calculated to assess the risk of DVT and

PE using univariable and multivariable Cox proportion hazards regression models. The multivariable models were adjusted for age,

sex, and the following comorbidities: atrial fibrillation, hypertension, hyperlipidaemia, CVA, heart failure, lower leg fracture or surgery, cancer, and pregnancy. The cumulative incidence of DVT and PE between the T2DM and comparison cohorts was assessed using the Kaplan–Meier method, and the differences between the curves

were evaluated using a log-rank test. All statistical analyses were performed using SAS statistical software (Version 9.3 for Windows; SAS

Institute, Inc., Cary, NC, USA). A two-tailed p-value < 0.05 was considered statistically significant.

Results

Demographic characteristics and comorbidities in the T2DM and comparison cohorts

Eligible study participants consisted of 56, 158 patients in the

T2DM cohort and 168, 474 persons in the comparison cohort, with similar sex and age distributions (

_

in the T2DM cohort and comparison cohort were 57.1 (SD = 13.2) and 56.4 (SD = 13.7) years, respectively. Patients with T2DM were more prevalent with high CCI scores. At baseline, comorbidities of hypertension, hyperlipidaemia, and coronary heart disease were

more prevalent among the T2DM cohort compared with the comparison cohort.

Incidence densities of DVT and PE hazard ratio between the T2DM and comparison cohorts by demographic characteristics and comorbidity

During the mean follow-up period of 6.02 years for the T2DM cohort and 6.10 years for the comparison cohort, the overall incidence density of VTE was higher in the T2DM cohort than that in the

comparison cohort (12.0 and 7.51 per 10, 000 person-years, respectively). Compared with the comparison cohort, the adjusted

HR (aHR) of VTE was 1.44 (95 % CI = 1.27–1.63) for the T2DM

cohort. T2DM patients exhibited a 1.59-fold greater aHR of unprovoked VTE than did the controls. The T2DM cohort had higher

incidences of both DVT (8.96 vs 5.61 per 10, 000 person-years) and PE (3.92 vs 2.33 per 10, 000 person-years ) than did the comparison cohort. The aHR of DVT were 1.43 (95 % CI = 1.23–1.65) for the T2DM cohort compared with the comparison cohort. Compared with the comparison cohort, the aHR of PE in the T2DM cohort was 1.52 (95 % CI = 1.22–1.90). The sex-specific incidence of DVT was higher in women than in men in both cohorts, with an aHR of 2.08 (95 % CI = 1.56–2.78) for women and an aHR of 1.66 (95 % CI = 1.24–2.22) for men. The DVT incidence increased with age in

both cohorts, but, the age-specific T2DM to non-DM relative risk was the greatest for the young group aged 49 years and younger

(aHR = 5.10; 95 % CI = 3.12–8.32). Similar results were observed when we switched the outcome to PE. The T2DM cohort had a 1.62-fold (95 % CI = 1.21–2.16) risk of developing PE for women, a 1.44-fold (95 % CI = 1.02–2.04) risk of developing PE for men, and a 7.50-fold (95 % CI = 3.29–17.1) greatest risk of developing PE in

the group aged 49 years and younger. When stratified based on patients with or without comorbidities, the T2DM cohort exhibited a

significantly higher risk of DVT than did the comparison cohort for those who did not have any comorbidity (aHR = 3.16; 95 % CI = 2.33–4.28). The incidence of PE increased in those with comorbidities in both cohorts (

_

Cumulative incidence of DVT and PE in the T2DM and comparison cohorts

The results of the Kaplan–Meier analysis, shown in

_

did the comparison cohort. The difference was significant for VTE

(log-rank test p < 0.001). Cox model with HRs and 95 % CIs of DVT and PE

associated with T2DM and covariates

The results of multivariable Cox proportion hazards regression models for the risk of related variables contributing to DVT and PE are shown in

_

The risk factors contributing to DVT included increasing age (in one-year bands; aHR = 1.05, 95 % CI = 1.04–1.05), higher CCI

score (aHR = 1.76, 95 % CI = 1.47–2.10 for CCI = 1; aHR = 2.31, 95 % CI = 1.75–3.04 for CCI = 2; aHR = 3.06, 95 % CI = 2.32–4.03 for CCI _ 3), and hypertension (aHR = 1.28, 95 % CI = 1.09–1.50). Increasing age per year (aHR = 1.05, 95 % CI = 1.04–1.06), higher CCI score (aHR = 1.92, 95 % CI = 1.48–2.50 for CCI = 1; aHR = 2.76, 95 % CI = 1.89–4.04 for CCI = 2; aHR = 2.24, 95 % CI = 1.41–3.57 for CCI _ 3), and coronary heart disease (aHR = 1.70, 95 % CI = 1.34–2.16) were associated with an increased risk of PE.

Discussion

We found a 1.44-fold increased risk of developing VTE in T2DM

patients compared with the comparison cohort. Heit et al. conducted a case-control study using the Rochester Epidemiology

Project to assess the relationship between DM and VTE, but they did not find that DM increased the risk of VTE after adjusting for covariates (19). Stein et al. used a National hospital discharge survey in the United States and indicated that DM carries a 1.05-fold

increased risk of VTE in adults aged 20 years and older (17). Furthermore, Piazza et al. used data from 11 medical centres in the

Massachusetts metropolitan area and indicated that DM is an independent predictor of recurrent DVT (18).

Although T2DM patients exhibited greater prevalence of medical comorbid disorders than did the comparison cohort in our

study, T2DM remained an independent risk factor for developing DVT and PE after adjusting for age, sex, and comorbidities. This is the first study to evaluate the risk of DVT and PE in T2DM patients in Asian populations. The discrepancy of the risk for developing DVT and PE in diabetic patients between this study and

studies on western countries might be associated with various sampling methods and racial differences.

The actual mechanism of DVT and PE development among

T2DM patients is complex. The underlying hypercoagulable status involves impaired fibrinolysis, as well as platelet and endothelial cell dysfunction (23-25). Numerous coagulation factors including fibrinogen, factor V, factor VII, factor VIII, factor X, factor XI, factor XII,

kallikrein, and the von Willebrand factor are elevated in DM patients (26, 27). Activated protein C (APC) was originally described

as an anticoagulant, but has recently been found to exert potent cytoprotective

properties, including inhibiting inflammation and

apoptosis and maintaining endothelial and epithelial barriers (28, 29). The plasma level of APC is reduced in DM patients (30). The higher levels and activity of the circulating plasminogen activator inhibitor type I, possibly related to the oxidation of LDL and insulin resistance, were also reported in DM patients (31).

Patients with T2DM carried a greater risk of developing DVT

and PE than did the comparison cohort in both sexes. The incidence of DVT and PE increased as age increased for both cohorts.

However, after we adjusted for sex and comorbidities, the risk of DVT and PE was substantially higher for young adults with DM, suggesting that DM exerted the greatest effect on the health of young adults.

The strength of this study is that it provides a nationwide population-based cohort longitudinal study on the risk of DVT and PE

development for T2DM patients in an Asian population. The NHI program is universal and mandatory in Taiwan, and NHI beneficiaries are assigned a personal identification number that enables

tracing beneficiaries throughout the follow-up period.

Certain limitations must be considered when interpreting these findings. The NHIRD does not provide detailed lifestyle information, such as smoking, body mass index (BMI), and physical activity, which are all potential confounding factors for this study. However,

we adjusted for potential covariates such as hypertension, hyperlipidaemia, and CVA, which are correlated with BMI to minimise

the effect of BMI (32-34). Second, the study cases were selected according to ICD-9-CM codes, which might have caused a

misclassification bias despite the NHIA using an auditing mechanism to minimise diagnostic uncertainty and misclassification. The

lack of drug data, such as hormone replacement therapy or anticonceptive and antiplatelet drugs, to adjust for the outcomes of interest

might be another limitation.

In conclusion, this nationwide study of approximately 56, 158 T2DM patients with 338, 340 follow-up person-years shows that T2DM patients have a 1.44-fold increased risk of developing VTE compared with the comparison cohort. These findings elucidate the importance of a multidisciplinary team adopting an integrated approach to the intervention of potential risk factors among