Abstract and Introduction

Abstract

Purpose: The 90-day risk of venous thromboembolism (VTE) among medically ill patients admitted to a hospital was estimated and is discussed.
Summary: Patients aged ≥40 years who were hospitalized between January 1, 1998, and June 30, 2002, for reasons other than traumatic injury, labor and delivery, mental disorder, or VTE and who did not undergo surgery were identified in a large U.S. healthcare claims database. Patients receiving anticoagulants in the 90-day period preceding hospital admission were excluded. We estimated the percentage of study subjects who developed clinical deep-vein thrombosis (DVT) or pulmonary embolism (PE) within 90 days of hospital admission using Kaplan-Meier methods. We also estimated hazard ratios (HRs) for potential risk factors for VTE using univariate and stepwise multivariate Cox proportional hazards regression models.
Among 92,162 study subjects, 1468 (1.59%) developed clinical DVT or PE within 90 days of hospital admission; 18% of these events occurred postdischarge. In multivariate analyses, significant risk factors for clinical VTE included: 1) history of cancer (HR, 1.67; 95% confidence interval [CI], 1.45-1.93); 2) history of VTE within six months of index admission (HR, 6.14; 95% CI, 4.74-7.96); 3) operating room procedure within 30 days of index admission (HR, 1.81; 95% CI, 1.47-2.24); 4) peripheral artery disease during index admission (HR, 1.68; 95% CI, 1.28-2.21); and 5) heart failure during index admission (HR, 1.72; 95% CI, 1.52-1.95).
Conclusion: The risk of clinical VTE among medically ill patients admitted to a hospital, although less than that of patients undergoing major surgery, is not negligible. Patients with a history of recent VTE or surgery, those who are admitted to the intensive care unit, those with an admitting diagnosis of heart failure, and those with active cancer are at especially high risk of VTE and deserve increased consideration for prophylaxis.

Introduction

Deep-vein thrombosis (DVT) and pulmonary embolism (PE)-together referred to as venous thromboembolism (VTE)-are important causes of disability and death. Most cases of VTE occur among persons dwelling in the community (and not in nursing homes), but VTE risk is more than 260-fold higher among those who are hospitalized (3.6 versus 961 cases per 10,000 person-years).^[1] Although the greatest risk factor for VTE is recent major surgery, the number of cases among medical and surgical patients is roughly equal, as there are far more medical than surgical admissions.^[2,3]

Among medical diseases, those reported to confer the highest independent risk of VTE in a population-based, case-control study with multivariate analyses were malignant neoplasms with and without chemotherapy (odds ratio [OR], 6.5 and 4.1, respectively), prior superficial vein thrombosis (OR, 4.3), and neurologic disease with extremity paresis (OR, 3.0).^[2] Among hospitalized nonsurgical patients, VTE risk is especially high among the critically ill. In studies utilizing periodic screening (e.g., venography) to detect asymptomatic thrombosis, DVT has been estimated to develop within the first week in more than 30% of patients who are admitted to intensive care units (ICUs) and do not receive prophylaxis.^[4] High risks of DVT in the absence of antithrombotic therapy have also been documented among patients with acute myocardial infarction (about 24%) and those with ischemic stroke and a paralyzed lower extremity (pooled risk of 55%).^[5] It is important to note that these high rates of DVT are based on screening and that the great majority of thromboses so identified would remain undiagnosed in clinical practice. For example, in a randomized placebo-controlled trial of enoxaparin prophylaxis among acutely ill medical patients at increased risk of VTE, the rate of DVT among those receiving placebo was 14%, but of 42 cases documented by screening (almost all by venography), only 2 (5%) were symptomatic.^[6]

On the basis of well-established risks and the demonstrated efficacy of prophylaxis with anticoagulants, VTE prophylaxis is recommended and widely prescribed for many types of surgery.^[5] The risks of VTE among medical patients have been less well studied and there have been many fewer trials of antithrombotic therapy for VTE prophylaxis in this patient population. Consequently, VTE prophylaxis is believed to be underutilized for medical patients compared with those undergoing surgery. (Patients with acute myocardial infarction, ischemic cerebrovascular disease, and atrial fibrillation are exceptions to this statement, as they usually receive antithrombotic therapy; the primary motivation for such therapy is prevention of arterial thrombosis, however. Prevention of VTE is a secondary benefit.) Information on the risks of clinical VTE in hospitalized medically ill patients undoubtedly would highlight the benefits of prophylaxis in this patient population.

While such information could be collected via a large randomized trial or prospective observational study, these approaches would be both expensive and time-consuming. We therefore address this question using a large U.S. health care claims database. We examined the overall risk of VTE among medical patients, as well as the relative risk of this complication among subgroups of patients with known risk factors, especially those listed as an indication for prophylaxis in the guidelines of the American College of Chest Physicians (ACCP).^[5] In addition to examining the risk of VTE during hospitalization, we also explored the risk during the postdischarge period, as a heightened risk of VTE following hospital discharge has been documented for certain surgical patients and may exist for medical patients as well.^[6-8]

Methods

Overview

We conducted a retrospective cohort study to estimate the risk of VTE in hospitalized medically ill patients. Using data from the Constella Health Strategies Managed Care Organization Database (CHS-MCOD), we identified all persons aged ≥40 years who were admitted to a hospital between January 1, 1998, and June 30, 2002, for reasons other than traumatic injury and who did not undergo surgery. Patients were followed for 90 days from the date of their earliest such hospital admission for the occurrence of clinical VTE. The incidence of VTE was estimated using Kaplan-Meier methods, and Cox proportional hazards regression models were used to characterize the relationship between selected demographic and clinical variables and time to VTE.

Data Source

The CHS-MCOD is comprised of facility, professional, and retail pharmacy claims that have been submitted for reimbursement to various private health plans. The plans provide health care services to approximately three million members annually residing in 20 states, mostly in the South Atlantic (28%), South Central (37%), and North Central (31%) regions of the U.S.

The database includes member demographic and plan eligibility information, inpatient and outpatient diagnoses (in International Classification of Diseases, 9th Edition [ICD-9-CM] format), inpatient and outpatient procedures (in ICD-9-CM, Physician's Current Procedural Terminology, 4th Edition [CPT-4] and Health Care Financing Administration Common Procedure Coding System [HCPCS] formats), and retail (i.e., outpatient) prescription drugs dispensed (in National Drug Code [NDC] format), as well as dates of care for all medical and pharmacy services.

Sample Selection

We identified all persons in the CHS-MCOD with one or more hospital admissions between January 1, 1998, and June 30, 2002. For each such patient, the first (if any) admission that was not for traumatic injury or VTE and that did not involve surgery was identified; the date of this admission was designated the "index date." Admissions for trauma were defined as those with principal ICD-9-CM diagnosis codes in the range 800-959. Admissions for VTE were identified based on principal diagnoses of DVT or thrombophlebitis (451.XX, 452, 453.XX, 996.74, 997.2, 999.2), PE (415.1X), or post-thrombotic syndrome (454.0, 454.2, 459.1, 459.81, 707.1). Hospital admissions involving surgery were defined as those with ICD-9-CM procedure codes in the range 01-86.99, excluding codes for common medical procedures (e.g., bronchoscopy, cardiac catheterization, spinal tap; see appendix for full list). Admissions were excluded from the study sample if: 1) the patient was <40 years of age as of the index date; 2) the patient was not continuously eligible for health benefits during the six-month period preceding the index date; 3) the admission record contained one or more ICD-9-CM diagnosis codes for labor and delivery (630-677) or any hypercoaguable state (286.3, 286.6, 289.8, 289.9); 4) the principal diagnosis was a mental disorder (290-319); 5) length of stay in hospital was >60 days; or 6) the patient had evidence of receipt of an anticoagulant in the 90-day period preceding the index date.

Follow-up

All study subjects were followed from their index date until 90 days thereafter or until hospital readmission or disenrollment from the health plan, whichever occurred first.

Patient Characteristics

We characterized the study population in terms of selected demographic and clinical characteristics at index admission, including age, gender, principal diagnosis group (based on a classification scheme utilized in the U.S. National Hospital Discharge Survey), geographical region, and payer type. The presence of several established risk factors for VTE during the index admission was also noted, including: 1) acute coronary syndrome; 2) stroke; 3) chronic obstructive pulmonary disease; 4) heart failure; and 5) admission to an ICU. We further characterized study subjects according to whether they had any of the following additional established VTE risk factors at the index admission or during the immediately preceding six-month ("history") period: 1) cancer; 2) post-thrombotic syndrome (but not as a principal diagnosis on the index admission); and 3) neurological disorders with plegia, paresis, or paralysis. Finally, any diagnosis of DVT or PE during the history period was also ascertained.

Patients with acute coronary syndrome, heart failure, stroke, or chronic obstructive pulmonary disease were identified based on relevant diagnosis codes during the index admission. Patients with cancer or neurological disorders involving plegia, paresis, or paralysis were identified based on the presence of at least one inpatient claim with an associated ICD-9-CM diagnosis code, or at least two outpatient claims separated by a minimum of 30 days with any of these codes. Patients with post-thrombotic syndrome were identified based on the presence of any relevant ICD-9-CM code during the history period, or by a secondary diagnosis code for this condition during the index admission. Patients with a history of DVT or PE were identified using methods similar to those employed to identify patients who developed VTE in the 90-day period subsequent to their index date, as described below.

Outcome Measures

The primary measure of interest was the occurrence of clinical VTE between the index admission and the end of follow-up. Patients who developed clinical VTE were identified on the basis of: 1) a secondary diagnosis code for DVT (ICD-9-CM 451.1X, 451.2, 451.81, 451.9, 453.8, 453.9, 997.2, 999.2) or PE (415.1X) on the facility claim for the index admission; 2) a principal diagnosis code of DVT or PE on a patient's first readmission within 90 days of the index date; or 3) the presence of a diagnosis code for DVT or PE on any subsequent facility or provider claim within 90 days of the index date plus receipt of an anticoagulant (i.e., unfractionated heparin, low-molecular-weight heparin, and/or warfarin) within 15 days of the date of this subsequent claim. (DVT or PE is often the diagnosis recorded on the outpatient claims of patients who are worked up for, but not found to have, VTE; the stipulation regarding receipt of an anticoagulant helps ensure that "rule-outs" are not erroneously designated as having postdischarge VTE.) The date of diagnosis of VTE was noted for all patients.

Analyses

Demographic and clinical characteristics of patients were summarized using descriptive statistics. The incidence of VTE during the 90-day follow-up period was estimated using Kaplan-Meier methods; patients who disenrolled from the health plan or who were readmitted to a hospital (for reasons other than VTE) were censored as of the date of their disenrollment or readmission.

We estimated univariate and multivariate Cox proportional hazards regression models. The latter included all of the selected risk factors for VTE (as listed above), as well as age (in 10-year intervals) and gender; the model was estimated in stepwise fashion, in which each of the variables of interest was entered into the model one at a time. A significance value of p < 0.10 was used as the threshold for inclusion of variables in the model. The independent contribution of each risk factor is reported in terms of its estimated hazard ratio (HR) and corresponding 95% confidence interval (CI).

Results

A total of 92,162 patients met all inclusion criteria. Mean age was 71 years; 54% of study subjects were women ( Table 1 ). About one-half had a principal diagnosis involving either the circulatory or respiratory system. During the index admission or the preceding six months, 2.25% of study subjects had a recorded diagnosis (principal or secondary) of post-thrombotic syndrome. During the index admission, 10.5% of patients had a diagnosis of acute coronary syndrome; 9.7%, stroke; and 15.0%, heart failure. One-half of all patients were admitted to an ICU or a critical care unit (CCU).

The cumulative incidence of clinical VTE at 90 days was 1.59% (Figure 1); 18% of these cases occurred postdischarge. Inpatient mortality (unadjusted) was 42% higher for patients who developed VTE than those who did not (2.79% versus 1.96%). In univariate analyses, significant risk factors for clinical VTE included: 1) peripheral arterial disease during index admission; 2) chronic obstructive pulmonary disease during index admission; 3) any diagnosis of post-thrombotic syndrome, neurological disease with paresis or paralysis, cancer, or heart failure; 4) a diagnosis of VTE during the six-month history period; 5) admission to ICU or CCU; and 6) an operating room procedure during the 30-day period prior to index admission ( Table 2 ).

In the stepwise multivariate Cox proportional hazards regression, 10 variables were found to be significant independent predictors of clinical VTE: 1) female sex (HR, 1.13; 95% CI, 1.02-1.26); 2) peripheral artery disease during index admission (HR, 1.68; 95% CI, 1.28-2.21); 3) chronic obstructive pulmonary disease during index admission (HR, 1.33; 95% CI, 1.17-1.51); 4) a history of cancer (HR, 1.67; 95% CI, 1.45-1.93); 5) heart failure during index admission (HR, 1.72; 95% CI, 1.52-1.95); 6) neurological disease with paresis or paralysis (HR, 1.35; 95% CI, 1.05-1.75); 7) post-thrombotic syndrome (HR, 2.00; 95% CI, 1.59-2.52); 8) history of VTE in the six months prior to index admission (HR, 6.14; 95% CI, 4.74-7.96); 9) ICU admission (HR, 1.35; 95% CI, 1.21-1.50); and 10) an operating room procedure within 30 days prior to index admission (HR, 1.81; 95% CI, 1.47-2.24). Two variables were found to be associated with a reduced risk of VTE: 1) acute coronary syndrome (HR, 0.80; 95% CI, 0.66-0.97); and 2) other coronary heart disease during index admission (HR, 0.83; 95% CI, 0.72-0.95).

Discussion

Using a large U.S. health care claims database, we found that 1.58% of hospitalized medically ill patients-or about one in every 63 such admissions-develop clinical VTE over 90 days. The population of interest in our study included a very broad range of medical patients, and not just those with high-risk conditions. The risk among patients with known risk factors for VTE, such as those listed as appropriate for prophylaxis in the ACCP guidelines, is substantially greater.

While most cases of clinical VTE among hospitalized medically ill patients occur during initial hospitalization, almost one in five events occur postdischarge. To the best of our knowledge, our study is the first to examine VTE risk following hospital discharge among these patients. In studies of hip and knee arthroplasty patients, in contrast, 49-81% of all cases of clinical VTE have been reported to occur during the three months following hospital discharge.^[9-12] While the risk of postdischarge VTE appears to be lower in medical patients than in those who have undergone major orthopedic surgery, it nonetheless should be of concern.

Risk factors that we found to be independent predictors of VTE in multivariate analysis-such as history of cancer, history of VTE within six months of index admission, an operating room procedure within 30 days of index admission, and a diagnosis of heart failure or peripheral artery disease during index admission-are generally well known. The only exception is peripheral artery disease during the index admission; this finding, however, may reflect a diagnostic bias based on increased scrutiny of the lower extremities in this disease. The large HR (6.1) estimated for prior history of VTE probably reflects the recentness of diagnosis (i.e., within 180 days of index admission) and our exclusion of patients who received anticoagulants during the 90-day period prior to index admission.

We did not find, however, that other variables previously reported to be important predictors (e.g., stroke, acute coronary syndrome) were related to VTE risk. In fact, a diagnosis of coronary heart disease was found to be associated with a decreased risk of VTE. This finding probably reflects the widespread use of antithrombotic agents for prevention and treatment of arterial thrombosis.

It is important to acknowledge limitations of our study. For one, our analyses were based on data from a large U.S. health insurance claims database. Such databases have a number of shortcomings, for example, they often contain errors in coding and have limited information on medical history. They nonetheless have been used in prior studies of VTE-related outcomes in other patient populations (e.g., major orthopedic surgery).^[12] In addition, the study database did not contain any information on the medications patients received while in the hospital (information was limited to medications dispensed through retail pharmacies). Thus, we could not examine, for each medical condition or group of conditions of interest, the proportion of patients that received VTE prophylaxis and its relationship to the rate of clinical VTE.

Conclusion

This study confirms findings of earlier studies that the risk of clinical VTE in hospitalized medically ill patients, while less than that among patients undergoing major surgery, is not negligible. Our findings also suggest that VTE risk persists following hospital discharge. Patients with a history of recent VTE or surgery, those who are admitted to the ICU, those with an admitting diagnosis of heart failure, and those with active cancer are at especially high risk of VTE and deserve increased consideration for prophylaxis.

Characteristic	Value
Age, mean ± S.D., years	70.53 ± 12.73
Sex, n (%)
Male	42,055 (45.63)
Female	50,107 (54.37)
Principal Diagnosis During Index Admission, n (%)
Infectious diseases	2456 (2.66)
Neoplasms	2569 (2.79)
Endocrine, nutritional and metabolic diseases, and immunity disorders	7000 (7.60)
Nervous system and sense organs	2732 (2.96)
Diseases of the circulatory system	29,790 (32.32)
Diseases of the respiratory system	15,501 (16.82)
Diseases of the digestive system	8743 (9.49)
Diseases of the genitourinary system	5615 (6.09)
Diseases of the skin and subcutaneous tissue	1642 (1.78)
Diseases of the musculoskeletal system and connective tissue	2805 (3.04)
Congenital anomalies	103 (0.11)
Certain conditions originating in the perinatal period	9 (0.01)
Symptoms, signs, and ill-defined conditions	18,575 (20.15)
Injury and poisoning	1450 (1.57)
Other	3776 (4.10)
Potential Risk Factors, n (%)
Acute coronary syndromes	9354 (10.15)
Other coronary heart disease during index admission	18,951 (20.56)
Stroke during index admission	8910 (9.67)
Peripheral artery disease during index admission	1779 (1.93)
COPD during index admission	14,979 (16.25)
Cancer	9161 (9.94)
Heart failure during index admission	13,796 (14.97)
Heart failure prior to index date	4089 (4.44)
Neurological disease with paresis or paralysis	2892 (3.14)
Post-thrombotic syndrome	2072 (2.25)
Prior VTE	512 (0.56)
ICU admission	34,174 (37.08)
CCU admission	11,457 (12.43)
OR procedure in 30 days prior to index date	2962 (3.21)
OR procedure in 60 days prior to index date	4450 (4.83)
OR procedure in 90 days prior to index date	5419 (5.88)
Geographic Region, n (%)
Northeast	637 (0.69)
South	61,601 (66.84)
Midwest	24,943 (27.06)
West	4963 (5.39)
Unknown	18 (0.02)
Payer Type, n (%)
Indemnity	23 (0.02)
HMO	17,703 (19.21)
Medicare HMO	67,998 (73.78)
PPO	5303 (5.75)
Other	1135 (1.23)
Discharge Destination, n (%)
Home	74,835 (81.20)
Skilled nursing facility	7384 (8.01)
Death	1815 (1.97)
Other	8128 (8.82)

^aCCU = Critical Care Unit; COPD = Chronic Obstructive Pulmonary Disease; HMO = Health Maintenance Organization; ICU = Intensive Care Unit; OR = Operating Room; PPO = Preferred Provider Organization; VTE = Venous Thromboembolism; S.D. = Standard Deviation.

 

Characteristic	Univariate Hazard Ratio (CI)	Multivariate Hazard Ratio (CI)
Age	1.00 (0.99-1.00)	-
Female	1.06 (0.96-1.18)	1.13 (1.02-1.26)
Other Risk Factors
Acute coronary syndromes	0.84 (0.70-1.01)	0.80 (0.66-0.97)
Other coronary heart disease during index admission	0.89 (0.78-1.01)	0.83 (0.72-0.95)
Stroke during index admission	0.91 (0.76-1.09)	-
Peripheral artery disease during index admission	2.04 (1.56-2.66)	1.68 (1.28-2.21)
COPD during index admission	1.45 (1.28-1.64)	1.33 (1.17-1.51)
Cancer	1.74 (1.52-2.01)	1.67 (1.45-1.93)
Heart failure during index admission	1.83 (1.63-2.07)	1.72 (1.52-1.95)
Heart failure prior to index date	1.82 (1.50-2.21)	-
Neurological disease with paresis paralysis	1.40 (1.09-1.80)	1.35 (1.05-1.75)
Post-thrombotic syndrome	2.57 (2.05-3.22)	2.00 (1.59-2.52)
Prior VTE	9.30 (7.27-11.91)	6.14 (4.74-7.96)
ICU admission	1.33 (1.20-1.47)	1.35 (1.21-1.50)
CCU admission	1.20 (1.04-1.38)	-
OR procedure in 30 days prior to index date	2.29 (1.87-2.80)	1.81 (1.47-2.24)

^aCI = Confidence Interval; CCU = Critical Care Unit; COPD = Chronic Obstructive Pulmonary Disease; ICU = Intensive Care Unit; OR = Operating Room; VTE = Venous Thromboembolism.

 

Appendix

Editorial support for this article has been provided by sanofi-aventis U.S., Inc.

Financial support for this article has been provided by sanofi-aventis U.S., Inc.

Dr. Oster at Policy Analysis Inc., Four Davis Court, Brookline, MA 02445. Email: goster@pai2.com