Occurrence of Deep Vein Thrombosis among Hospitalized Non-Surgical Japanese Patients

Abstract

Objective: To estimate the frequency of deep vein thrombosis (DVT) among non-surgical inpatients, and to evaluate the D-dimer assay as a screening tool for DVT.

Methods: Subjects were non-surgical inpatients aged 20 years or older who had been bedridden for at least 24 hours and had moderate-to-high risk factors for DVT. We assessed the presence of DVT by venous ultrasonography. Patients who received a diagnosis of venous thromboembolism (VTE) before admission, who had symptoms or findings of VTE at admission, or who had surgery or trauma within the past 3 months before admission were excluded.

Results: DVT was confirmed in 96 of 525 patients (18.3%). In a logistic regression analysis, longer duration of hospitalization, higher D-dimer value, and history of cancer surgery were significantly associated with the occurrence of DVT. The D-dimer assay showed high sensitivity (96.1%) and high negative predictive value (97.6%).

Conclusion: Non-surgical inpatients with a long-term hospitalization or history of cancer surgery have a risk for DVT, and need to be considered for added DVT preventive measures as recommended in the prevention guidelines. In addition, the D-dimer assay is beneficial for the screening of DVT in medical practice.

Introduction

Recent studies have demonstrated that the occurrence of deep vein thrombosis (DVT) is comparable among Japanese and Caucasian patients. For example, DVT after total hip replacement and after total knee replacement was observed in 23% and 49% of Japanese patients, respectively,¹⁾ and in 42%–57% and 41%–85% of Caucasian patients, respectively.²⁾

DVT occurs in hospitalized non-surgical patients as well as surgical patients, and the risk of DVT is as high in some non-surgical patients as in surgical patients. The frequency of DVT was reported to be 24% in Caucasian patients with myocardial infarction and 55% in Caucasian patients with ischemic stroke.³⁾ Several other reports of DVT among Caucasian non-surgical patients have demonstrated that DVT occurrence can vary (9%–55%).^3–6) In the United States and Europe, the risk of DVT among hospitalized non-surgical patients is well established⁷⁾ and has been incorporated into their DVT prevention guidelines for such patients. On the other hand, in Japan, the focus of most studies assessing the risk of DVT has been surgical patients, not non-surgical patients.

This study aimed to estimate the frequency of DVT among hospitalized non-surgical Japanese patients, and identify risk factors for the occurrence of DVT. We also evaluated the clinical usefulness of the D-dimer assay as a screening tool for DVT.

Materials and Methods

Subjects

Subjects in this study were hospitalized non-surgical patients aged 20 years or older who had been bedridden (except for trips to the bathroom) for at least 24 h and had moderate-to-high risk factors for venous thromboembolism (VTE) as proposed by Haas⁷⁾ (Table 1). Patients who had been diagnosed with VTE before admission, who had symptoms or findings of VTE at admission, who had a history of VTE, or who had surgery or trauma within the past 3 months before admission were excluded from the study.

Study design and method

Patients who met the criteria described in the previous section were prospectively recruited at Mie University Hospital and Niigata University Medical and Dental Hospital from April 2006 to April 2008 after getting approval for the conduct of this study from the Ethics Committee of each institution. The presence or absence of DVT in the lower legs (in the common femoral vein, femoral vein, popliteal vein, posterior tibial vein, peroneal vein, soleal vein, gastrocnemius vein, and other veins) was assessed by venous ultrasonography during hospitalization. We used Aplio (Toshiba Medical Systems Corporation) and SSD-5500 (Hitachi Aloka Medical, Ltd.) diagnostic ultrasound systems for our study. Assessment of DVT was performed independently in each institution and not by a central committee set up for evaluating ultrasonography readings. D-dimer and soluble fibrin levels were determined by the latex photometric immunoassay (LPIA) at a cut-off point of 1.0 µg/mL for D-dimer and 7.0 µg/mL for soluble fibrin. D-dimer and soluble fibrin levels were measured independently in each institution. The results from the analysis of blood samples collected within three days prior or after the ultrasonography were included in the data analysis.

Statistical analysis

The cumulative incidence of DVT over the mean observation time during hospitalization was tabulated for each subject. Each variable (sex, smoking, obesity, etc.) was compared between the DVT positive and DVT negative groups by the Fisher exact test for categorical parameters and by the t-test for continuous variables. Variables found to be significantly different between the two groups were used in a multivariate logistic regression model where occurrence of DVT was set as a dependent variable. Sensitivity, specificity, positive predictive value, and negative predictive value of D-dimer and soluble fibrin assays for detecting DVT were calculated. Statistical analyses were performed in SAS (Version 8.2) at a significance level of p <0.05.

Results

The data of 525 patients (292 males and 233 females; average age 64 ± 14 years) were analyzed. Background characteristics of the patients are presented in Table 2. The mean number of days between admission and ultrasonography was 27 ± 78 days (median: 12 days). The clinical departments with many subjects were pulmonology/ nephrology/diabetology (33.3%), cardiology (25.5%), neurology (16.0%), and gastroenterology/hepatology (10.5%).

Out of the 525 patients, 96 patients (18.3%) had confirmed DVT (Fig. 1A). DVT occurrence in prevention group was 22.2%, while that in non-prevention group was 16.7%. In the prevention group, the proportion of elderly patients aged 70 years or older was high (45.8% vs. 39.2%), the proportion of patients who had catheterization within 3 months was high (7.8% vs. 3.0%), length of hospital stay was long (average of 44.5 days vs. 31.0 days), and there were few patients with liver disease (6.5% vs. 8.3%). After carrying out the exploratory subgroup analysis among patients aged less than 70 years, within 4 weeks in length of hospital stay, and without liver disease, the DVT occurrence was the lowest in the pharmacological prevention group (0.0% (0/12)). The DVT occurrences were 5.9% (1/17) in mechanical prevention group and 10.2% (9/88) in non-prevention group (Fig. 1B). However, no statistical significance was observed by Cochran-Armitage trend test (p = 0.21). The number of the patients with DVT at the right extremity only, left extremity only, and at both extremities were 33 (34.4%), 28 (29.2%), and 35 (36.5%), respectively. The most frequent site of DVT development was the soleal vein (71.9%) followed by femoral vein (26.0%), peroneal vein (14.6%), popliteal vein (13.5%), posterior tibial vein (9.4%), and common femoral vein (7.3%). The cumulative incidence of DVT was calculated for each diagnosis at admission grouped according to the ICD-10 (International Statistical Classification of Diseases and Related Health Problems 10th Revision). The cumulative incidence was the highest in patients with cerebrovascular diseases (25.7%) followed by malignant neoplasms (24.8%), renal failure (17.0%), diabetes mellitus (15.2%), glomerular diseases (14.0%), other forms of heart disease (11.1%), and ischaemic heart diseases (9.3%) (Fig. 1C).

Fig. 1.

(A) The occurrence of DVT in the prevention group was higher than non-prevention group. The occurrence of DVT in the pharmacological prevention group was higher than mechanical prevention group and non-prevention group. (B) Among patients aged less than 70 years, within four weeks in length of hospital stay, and without liver disease, the DVT occurrence was the lowest in the pharmacological prevention group. (C) The occurrence of DVT was the highest in patients with cerebrovascular diseases followed by malignant neoplasms, renal failure, diabetes mellitus, glomerular diseases, other forms of heart disease, and ischaemic heart diseases. DVT: deep vein thrombosis

Table 3 shows the comparison of each variable between the DVT positive and negative groups. Higher age, more frequent catheterization history, longer hospitalization, higher D-dimer and fibrin levels, lower platelet count, higher CRP level, lower frequency of liver disease comorbidity, and more frequent cancer surgery history were observed in the DVT positive group than the DVT negative group. A multivariate logistic regression model including all variables that differed significantly between the DVT positive and negative groups identified longer hospitalization, higher D-dimer level, and history of cancer surgery as significantly associated with the presence of DVT (Table 4). Forty one patients had cancer surgery history (more than 3 months ago) and the DVT occurrences exceeded 30% except among those with liver cancer. The DVT occurrences by site were breast 66.6% (2/3), stomach 50.0% (2/4), colon/rectum 50.0% (2/4), lung 42.9% (3/7), kidney/urinary tract 40.0% (2/5), gynecology 33.3% (1/3), liver 0.0% (0/7), and others 37.5% (3/8). The DVT occurrence of liver was significantly lower than others (other than liver) (p = 0.04 by Fisher’s exact test (two-sided)).

Of the 96 patients found to have DVT, 68 patients (70.8%) were asymptomatic, demonstrating the poor sensitivity of symptoms as a diagnostic tool for detecting DVT (only 29.2%) (Table 5A). On the other hand, sensitivities of D-dimer and soluble fibrin levels were 96.1% (Table 5B) and 61.5% (Table 5C), respectively. D-dimer level also showed a high negative predictive value of 97.6% (Table 5B).

Discussion

The present study was conducted in hospitalized non-surgical patients having a wide variety of medical conditions including cerebral and cardiovascular diseases, malignant neoplasms, urogenital diseases, and diabetes. A total of 18.3% of the studied patients were found to have DVT. The number of studies on DVT in non-surgical patients in Japan is limited. Ota et al. reported that the cumulative incidences of DVT in patients with heart failure of NYHA (New York Heart Association) class II, III, and IV were 4.4%, 4.8%, and 25.5%, respectively.⁸⁾ Matsuo et al. reported detecting DVT in 34% of patients with congestive heart failure (NYHA class III or IV).⁹⁾ Ohmori et al. reported detecting DVT in 34.8% of patients with severe motor and intellectual disabilities.¹⁰⁾ Ueda et al. reported that reduced blood flow in the calf veins caused by immobility increased the risk of DVT.¹¹⁾ In Western countries, Alikahn et al. reported the cumulative incidence of DVT in non-surgical patients to be 14.9% in general, including congestive heart failure (14.6%), respiratory disease (13.1%), infection (15.5%), and inflammatory disease (20.7%).¹²⁾ Cohen et al. reported the cumulative incidence of DVT in patients with congestive heart failure, respiratory disease, infection, and inflammatory disease (all-inclusive) to be 10.5%.¹³⁾ Taken together, findings in the Japanese as well as Caucasian population indicate that a significant proportion of hospitalized non-surgical patients have a risk for DVT.

Japanese DVT guideline including prevention strategies were issued in 2004 and this recommend that patients at risk receive preventive measures based on the magnitude of their risk.¹⁴⁾ In this study, the DVT occurrence in prevention group was not lower than that in non-prevention group. This study was not a randomized clinical trial, and the physician judged the preventive need depending on the risk of each patient. Therefore we may have to consider a potential bias of the study subjects, in which relatively high-risk patients could have been included more in the prevention group. The DVT occurrence in prevention group was not higher than non-prevention group among patients whose DVT risk was not so high and had low bleeding risk in anticoagulant administration as shown in Fig. 1B. In the patient group who the physician did not judge the prevention to be necessary, DVT occurred in 16.7% of the patients. It suggests that the preventive measures (as recommended by the VTE prevention guidelines) will need to be considered for the patients whose preventive measure was not performed in the judgment of physicians.

In this study, longer hospitalization, higher D-dimer level, and cancer surgery history were found to be significantly associated with the DVT presence in the multivariate analysis. Barbar et al defined a risk assessment model including reduced mobility and evaluated the risk for VTE.¹⁵⁾ From the study results reported by Barbar et al.,¹⁵⁾ reduced mobility was shown as a risk for VTE in American College of Chest Physicians (ACCP) evidence-based clinical practice guideline.¹⁶⁾ Gibbs reported that VTE occurred with high frequency when the duration of bed rest was more than one week.¹⁷⁾ Flordal et al. reported that long hospital stay was independent predictor for major thromboembolism.¹⁸⁾ From the study results reported by Gibbs and Flordal et al.,^17,18) long term bed rest was shown as a moderate risk for VTE in Japanese prevention guideline.¹⁴⁾ ACCP guideline showed that favored strategy for diagnosis of first DVT was combine use of pretest probability assessment, D-dimer, and ultrasonography.¹⁹⁾ A couple of reports in Japan have shown the association between the presence of DVT and D-dimer level in the blood collected postoperatively over several days.^20,21) However, data in non-surgical patients are limited. Kuwashiro et al. has recently demonstrated that D-dimer level is a factor significantly associated with DVT occurrence in hospitalized patients with stroke.²²⁾ Active cancer was also included in the risk assessment model,¹⁵⁾ and shown as a risk for VTE in the ACCP guideline.¹⁶⁾ Malignancy was shown as a moderate risk for VTE also in Japanese prevention guideline.¹⁴⁾ Because the number of subjects of each cancer site was limited, it was difficult to mention what sites of cancers had higher risk of DVT. From three retrospective studies among patients with cancers, pancreatic, stomach, and renal/other abdominal cancer were reported as high risks for VTE in these three studies.^23–25) Brain tumor and lung cancer were reported as high risks for VTE in two of these three studies.^23–25) In case-control study among patients with VTE and general population, hematological, lung, and gastrointestinal cancer were reported as high risks for VTE.²⁶⁾ From the study results reported by Khorana et al.,²⁷⁾ stomach, pancreas, and primary brain tumor were shown as very high risks, and lung, lymphoma, gynecologic, bladder, testicular, and renal tumor were shown as high risks for chemotherapy-associated VTE in ambulatory setting in American Society of Clinical Oncology clinical practice guideline.²⁸⁾ Patients with cancer often have multiple risks for VTE and following factors other than cancer site were also shown as risks: cancer stage (advanced stage), cancer histology (adenocarcinoma), time after initial diagnosis (highest in first 3 to 6 months), chemotherapy, antiangiogenic agents (eg, thalidomide, lenalidomide), hormonal therapy, erythropoiesis-stimulating agents, transfusions, indwelling venous access devices, radiation therapy, surgery >60 min, older age, race (African Americans), medical comorbidities (infection, renal disease, pulmonary disease, arterial thromboembolism), obesity, history of VTE, diminished performance status, and inherited prothrombotic mutations.²⁸⁾

Of patients diagnosed as having DVT, 70.8% had no symptoms. Thus, it would be beneficial for the most patients at high risk to be monitored closely even if they have no symptoms of DVT. The sensitivities of D-dimer and soluble fibrin levels for detecting DVT were 96.1% and 61.5%, respectively, and D-dimer level showed a high negative predictive value of 97.6%. Considering such high sensitivity and high negative predictive value, the D-dimer assay could be a useful tool for DVT screening in hospitalized non-surgical patients as well as surgical patients.

The two sites of this study are general hospitals and we carried out the investigation in various clinical departments. When investigation is carried out in trauma unit or stroke unit, results will be different. Patients who had trauma within the past 3 months before admission were excluded from the study. The two sites of this study are advanced treatment hospitals and the wards provide acute medical care mainly. Unlike the medical institution where long term care ward accounts for most of beds, the patient has treatment and aims for leaving the hospital after being improved. When the subjects are bedridden patients for long term or patients in terminal stage, results will be different.

This study has collected data from more than 500 hospitalized non-surgical patients to accurately estimate the frequency of DVT. A strength of this study is the large size of the study population, which is one of the largest in DVT/VTE studies on Japanese patients. This study has three limitations. The first limitation is that data was collected at only two hospitals. The second is that the diagnosis by ultrasonography was made without consulting members of a central committee (for diagnostic standardization purposes); this may have resulted in differences in judgment between the two hospitals to some extent. However, both hospitals used ultrasonography and not venography for DVT diagnosis in daily practice and had a skilled examination staff for diagnosis of DVT. Thus, there is sufficient rationale for inclusion of the ultrasonography data in our investigation. And the third is that activities of daily living (ADL) and bed rest level were not assessed. Thus, generalization of the results of this study should be viewed with caution. ADL and bed rest level may indeed affect DVT frequency, and so this point may need clarification in future studies.

Conclusion

The results of this study suggest that hospitalized non-surgical patients, especially those with a long-term hospitalization or a history of cancer surgery have a risk for DVT, and that additional strategies for preventing DVT, which is recommended by the DVT prevention guidelines, need to be considered for such patients. This study also suggests the D-dimer level screening is beneficial in patients with suspected DVT.

Acknowledgment

We express our gratitude to all the patients who participated in the study, and to the medical technologists and other staff members who contributed to the study.

Disclosure Statement (Conflict of Interest)

This study was conducted as a sponsored study of GlaxoSmithKline K.K. with its full financial support. Norikazu Yamada, Kazuhiko Hanzawa, Satoshi Ota, Mashio Nakamura, Koichi Sato, Maiko Ikura, and Masaaki Ito received a research grant from GlaxoSmithKline K.K. Takeo Suzuki, Toshihiko Kaise, and Hiromu Nakajima are employees and Toshihiko Kaise and Hiromu Nakajima are stockholders of GlaxoSmithKline K.K.