Sage Journals: Discover world-class research

Abstract

Background:

Hospitalized medically ill patients receiving antithrombotic medications experience increased risk of bleeding. We examined antithrombotic use, bleeding rates, and associated risk factors at 30 days post discharge.

Methods:

This retrospective database analysis included nonsurgical patients aged ≥40 years hospitalized for ≥2 days during 2005 to 2009. Previously cited, validated International Classification of Diseases, Ninth Revision, Clinical Modification codes for major bleeding were used to define clinically relevant bleeding.

Results:

Of the 327,578 patients, 9.1% received antithrombotic medications, of which 3.7% were anticoagulants. Rates of major and minor bleeding were 1.8% and 7.1%, respectively. Preindex gastroduodenal ulcer, thromboembolic stroke, blood dyscrasias, liver disease, and rehospitalization were the strongest predictors of major bleeding. Other risk factors included increasing age, male gender, and hospital stay of ≥3 days.

Conclusions:

Careful consideration of these demonstrated bleed-associated comorbidities before initiating anticoagulation or combining antithrombotic medications in medically ill patients may improve strategies for prevention of postdischarge thromboembolism.

Keywords

bleeding venous thromboembolism thrombosis prophylaxis antiplatelet drugs anticoagulants

Introduction

Venous thromboembolism (VTE) frequently complicates hospitalization for medical conditions and has historically been considered the most common cause of preventable hospital mortality.^1,2 Although pharmacologic prophylaxis effectively reduces the risk of VTE, only 13% to 34% of eligible medically ill patients receive guideline-recommended anticoagulation at the suggested dose and duration.^3,4 Key elements impacting effective use of thromboprophylaxis in the US health care settings include difficulty in recognizing risk factors for both VTE and bleeding, clinicians’ concerns about the risk of bleeding, and the absence of institutional protocols for VTE prophylaxis.^{5
–7} Postdischarge bleeding may be related to the use of antithrombotic agents. Although several recent studies have investigated adverse bleeding events in hospitalized medically ill patients,^{8

–13} few have examined risk factors for postdischarge bleeding in this unique patient population.^{14
–16}

The incidence of postdischarge major bleeding ranges from 0.2%⁸ to 1.7%⁹ in VTE prevention studies and from 2.8% to 4% in VTE treatment studies.^10,11 A number of factors increase the risk of a bleeding event after hospitalization. Advanced age, impaired renal function, concurrent use of antithrombotic medications, recent trauma or surgery, and comorbid conditions such as heart failure and active cancer have all been associated with an increased risk of inpatient and postdischarge bleeding in patients receiving anticoagulant agents for VTE prophylaxis or treatment.^{12

–16} An increased bleeding risk of concurrent use of aspirin with warfarin has been convincingly demonstrated in several clinical trials, meta-analyses, and observational studies.¹⁷ ^–21

In a bleeding risk assessment model (RAM) from a large registry in hospitalized medical patients, bleeding events were attributed to hypertension, gastroduodenal ulcer, bleeding 3 months prior to hospitalization, low platelet count, admission to intensive care or cardiac care unit, central venous catheterization, and rheumatic disease when these risk factors were present at hospital admission.¹⁵ Although numerous risk factors have been reported, no independent predictors of major bleeding have been consistently identified across the studies. This may be attributed to the use of different data sources and inconsistent definitions of bleeding in this patient population.

Many RAMs have been used to quantify bleeding risk in patients with VTE.^14,16,22,23 Of those, only 1 model is fully applicable to the medically ill patient population,¹⁵ but it is used only in the patient data evident at admission to the hospital. This study has been designed to address this evidence gap by examining the rates and identifying associated risk factors for clinically relevant bleeding at 14 and 30 days post discharge, along with the use of antithrombotic medications in hospitalized medical patients.

Methods

Data Source

This was a retrospective analysis of an administrative claims data set from the HealthCore Integrated Research Database (HIRD), consisting of eligibility, medical claims, and pharmacy claims of approximately 32 million patients from 14 geographically dispersed health plans in the Northeastern, Southern, Midwestern, and Western regions of the United States. The database contains claims information from the largest commercially insured population in the United States and includes health plan organizations such as health maintenance organizations, point of service plans, preferred provider organizations, consumer-directed health plans, and indemnity plans. Overall, the HIRD is representative of the US population in terms of age and gender, with the HIRD being slightly younger, consistent with the fact that the majority of patients are commercially insured (thus, the largest age and gender discrepancies are in the cohorts aged 65 years or older). Validity of the HIRD has been previously documented with refined claims algorithms for severe upper gastrointestinal bleeding, demonstrating up to an 87.8% positive predictive value in true outcomes.²⁴Patient clinical settings included community hospitals (urban and rural), private hospitals, regional medical centers (academic and nonacademic), teaching hospitals, and other health care facilities. This analysis was conducted in compliance with state and federal laws, including the Health Insurance Portability and Accountability Act of 1996.

Patient Population

The study cohort included patients aged ≥40 years who had at least 1 inpatient nonsurgical hospitalization with a length of stay (LOS) of ≥2 days for an acute medical illness between January 1, 2005 and December 31, 2009. Medical conditions used to assess eligibility included heart failure, severe lung disease, ischemic stroke, active cancer, acute infection, rheumatologic inflammatory diseases, and endocrine disorders (diabetes mellitus, pancreatitis, and cholecystitis). Patients with these diagnoses were classified as acutely “medically ill patients,” while the other patients who met study eligibility criteria but were hospitalized for conditions different from those mentioned above were categorized as “other hospitalized patients.” Continuous health plan eligibility for the 12-month period prior to the index hospitalization date was also required. Exclusion criteria included a diagnosis of mental disorder, trauma, or VTE during the index hospitalization; hospitalization for obstetric conditions; a surgical procedure; diagnosis of atrial fibrillation during the 12-month preindex period; and use of warfarin (defined as a first prescription fill) 30 days or more post discharge.

Study Analysis

The key outcomes of interest were major and minor bleeding observed at 14 and 30 days after discharge. The other outcomes including postdischarge VTE, rehospitalization (defined as the first postdischarge hospitalization with the same diagnosis as the index hospitalization), and all-cause mortality (captured through the Social Security Death Index database) were assessed at 90 days postdischarge. Because bleeding events were identified through the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes, the definitions of postdischarge major and minor bleeding (Appendix A) were based on previously published studies with minor modifications.²⁵ In our study, claims code diagnoses, including validated codes for bleeding derived from the literature^{26

–29} were validated in a similar way and achieved similar results.

Patient preindex characteristics included demographics, risk factors for bleeding, comorbid conditions (acute infections; neoplasm; endocrine and metabolic disorders; diseases of the blood and blood-forming organs; immune, neurological, and musculoskeletal disorders; and digestive, genitourinary, circulatory, and respiratory system disorders), Deyo-Charlson Comorbidity Index (DCI) score,³⁰ and concomitant medication use. Evidence of antithrombotic drug use was determined by the presence of one or more pharmacy or medical claims for anticoagulant or antiplatelet agents in 14- and 30-day periods after hospital discharge.

Descriptive statistics were compiled for all the variables. Hazard ratios (HRs) for bleeding risk factors—including age, gender, major comorbidities, length of index hospitalization, postdischarge antithrombotic use, and rehospitalization—were estimated using multivariate survival analysis and Cox proportional hazards models. Statistical software SAS version 9.1 (SAS Institute, Cary, North Carolina,) was used to perform all analyses.

Results

A total of 24,361,751 health care claims records were screened for the study. Of these, 327,578 unique patients (141,628 medically ill patients and 185,950 other hospitalized patients) met the study entry criteria and were included in the analysis (Figure 1). Cohort median age was 68 years; 55.8% of the patients were female. The mean DCI score was 2.0 (standard deviation [SD] =2.2). A significant proportion of the patients had underlying endocrine, metabolic, or immune disorders (46%) and a history of blood dyscrasias (14%) during the preindex hospitalization period. The mean hospital LOS was 4.6 days (SD = 5.2 days; Table 1). Compared to patients who did not receive postdischarge antithrombotic medications at 30 days, those who did tended to be male, older with slightly higher DCI score, and had longer hospital stays.

Figure 1.

Patient selection flowchart. VTE indicates venous thromboembolism.*The most frequent diagnoses in the “other hospitalized patients” category were respiratory diseases, alteration of consciousness, hallucinations, syncope and collapse, and abdominal or pelvic pain or swelling.

Table 1.

Patient Characteristics and Antithrombotics Use After Hospitalization.

Variables	Antithrombotic Use ≤30 Days Postdischarge, N = 29 947		No Antithrombotic Use ≤30 Days Postdischarge, N = 2 97 631		P Value
Variables	N/ Mean/Median	Column %/SD	N/ Mean/Median	Column %/SD	P Value
Male	16 341	54.6%	1 28 467	43.2%
Female	13 606	45.4%	1 69 164	56.8%	<.001
Age	68/68	±12.9	66/64	±15.0	<.001
Deyo-Charlson Comorbidity Index	1.9/1.0	±2.3	1.6/1	±2.2	<.001
Pre-index comorbidities
Endocrine/metabolic/immune disorders	15 515	51.8%	1 34 875	45.3%	<.001
Blood dyscrasias	4445	14.8%	41 144	13.8%	<.001
Neoplasms	3638	12.2%	41 162	13.8%	<.001
Liver disease	733	2.5%	9934	3.3%	<.001
Thrombocytopenia	185	0.6%	2095	0.7%	.087
Active peptic ulcer	117	0.4%	1479	0.5%	.012
Thromboembolic stroke	2258	7.5%	9443	3.2%	<.001
Rheumatoid arthritis^a	376	1.3%	2248	0.8%	<.001
Length of index hospitalization	4.9/4.0	±5.5	4.6/3.0	±5.1	<.001

Abbreviation: SD, standard deviation.

^a Rheumatoid arthritis was captured at index admission; all other preindex risk factors occurred at any time during the preindex period.

Of the study cohort of 327,578 patients, 9.1% received postdischarge antithrombotic medications, of which 3.7% were anticoagulants. Among the 29,947 patients who used antithrombotic medications within 30 days postdischarge (Table 2), 59.1% used at least 1 antiplatelet agent and 39.1% used at least 1 anticoagulant agent. A small proportion (1.8%) of patients used both anticoagulants and antiplatelets. Clopidogrel and aspirin-dipyridamole were the most frequently prescribed antiplatelets (15,745 and 1863 patients, respectively) while warfarin and the low-molecular-weight heparin (LMWH) enoxaparin were the most frequently prescribed anticoagulants (11,330 and 1332 patients, respectively). The highest utilization of warfarin was observed among patients with underlying heart failure (6.3%), stroke (5.4%), and rheumatic inflammatory disease (4.8%), whereas patients with ischemic stroke had the highest rate of clopidogrel use (18.1%; data not shown). The overall rates of antithrombotic use between the disease categories of “acutely ill medical patients” and “other hospitalized patients” were similar (9.4% and 9.0%, respectively).

Table 2.

Utilization of Antithrombotic Medications Within 30 Days After Hospitalization.^a

Type of Antithrombotic Agent	All Patients, N = 3 27 578		Medically Ill Patients,^b N = 1 41 628		Other Hospitalized Pts,^c N = 1 85 950
Type of Antithrombotic Agent	N/Mean	%/SD	N/Mean	%/SD	N/Mean	%/SD
All antithrombotic agents	29 947	9.1%	13 310	9.4%	16 637	9.0%
Anticoagulants, n (%)	12 219	3.7%	5467	3.9%	6752	3.6%
Warfarin	11 330	3.5%	5030	3.6%	6300	3.4%
Enoxaparin	1332	0.4%	590	0.4%	742	0.4%
Heparin	180	0.1%	126	0.1%	54	<0.1%
Fondaparinux	89	<0.1%	34	<0.0%	55	<0.1%
Other^d	70	<0.1%	28	<0.0%	42	<0.1%
Antiplatelets, n (%)	18 270	5.6%	8081	5.7%	10 189	5.5%
Clopidogrel	15 745	4.8%	6567	4.6%	9178	4.9%
Aspirin-dipyridamole	1863	0.6%	1212	0.9%	651	0.4%
Other^e	981	0.3%	470	0.3%	511	0.3%
Anticoagulant + antiplatelet agents	542	0.2%	238	0.2%	304	0.2%

Abbreviations: Pts, patients; SD, standard deviation.

^a Use of therapeutic agents is not mutually exclusive.

^bThe disease groups of the “medically ill patients” included acute infection, cancer, endocrine disorders, heart failure, ischemic stroke, rheumatologic inflammatory diseases, and severe lung disease; some patients had multiple disease diagnoses at admission.

^c The most frequent diagnoses in the “other hospitalized patients” category were respiratory system and other chest conditions, alteration of consciousness, hallucinations, syncope and collapse, and symptoms involving the abdomen and pelvis.

^dDalteparin, tinzaparin.

^eAbciximab, anagrelide, cilostazol, dipyridamole, eptifibatide, prasugrel, ticlopidine, and tirofiban.

At 14 days of discharge 3815 (1.2%) patients had a major bleeding event; this number increased to 5951 (1.8%) patients at 30 days. Of the 29,947 patients who received antithrombotic medications postdischarge, 404 (1.4%) and 658 (2.2%) of the patients experienced major bleeding at 14 and 30 days, respectively. Among the patients who did not use antithrombotics, the rates of major bleeding at 14 and 30 days were 1.2% and 1.8%, respectively. The mean times to occurrence of a major bleeding event at 14 and 30 days were 6 and 10 days, respectively (Table 3). Of those patients who received antithrombotic agents and experienced major bleeding at 30 days postdischarge, two-thirds received antiplatelets, nearly one-third received anticoagulants, and a few patients received both anticoagulants and antiplatelets (data not shown).

Table 3.

Outcomes, Time to Event, and Antithrombotic Use After Hospitalization.

Variables	All Patients (N = 3 27 578)		Antithrombotic Use Within 30 Days Postdischarge (N = 29 947)		No Antithrombotic Use Within 30 Days Postdischarge (N = 2 97 631)		P Value
	N/Mean/Median	Column %/SD	N/Mean/Median	Column%/SD	N/Mean/Median	Column%/SD	P Value
Major bleeding
At 14 days	3815	1.2%	404	1.4%	3411	1.2%	.002
Days to event	6/4	±4.6	7/6	±4.5	5/4	±4.6	<.001
At 30 days	5951	1.8%	658	2.2%	5293	1.8%	<.001
Days to event	10/8	±9	12/10	±9.5	10/7	±9.3	<.001
Minor bleeding
At 14 days	16 099	4.9%	3594	12.0%	12 505	4.2%	<.001
Days to event	6/6	±4.0	6/5	±4.0	6/6	±4.0	<.001
At 30 days	23 313	7.1%	4817	16.1%	18 496	6.2%	<.001
Days to event	11/9	±8.4	10/7	±8.1	11/9	±8.4	<.001
All bleeding
At 14 days	19 914	6.1%	3998	13.4%	15 916	5.4%	<.001
Days to event	6/6	±4.0	6/5	±4.0	6/6	±4.0	<.001
At 30 days	29 264	8.9%	5475	18.3%	23 789	8.0%	<.001
Days to event	11/9	±8.4	10/8	±8.1	11/9	±8.4	<.001
VTE event
At 30 days	2656	0.8%	1081	3.6%	1575	0.5%	<.001
Days to event	13/11	±8.3	12/10	±8.0	13/12	±8.4	<.001
Rehospitalization^a
At 30 days	23 748	7.3%	2806	9.4%	20 942	7.0%	<.001
Days to event	11/9	±8.5	12/10	±8.3	11/9	±8.5	<.001
Death
At 30 days	6574	2.0%	149	0.5%	6425	2.2%	<.001
Days to event	12/10	±9.0	20/22	±7.0	12/10	±9.0	<.001

Abbreviations: SD, standard deviation; VTE, venous thromboembolism. ^a Rehospitalization is defined as an inpatient stay for the same diagnosis as the index hospitalization.

Patients exposed to antithrombotics had a slightly longer mean time to occurrence of a major bleeding event within 30 days postdischarge, compared to those who were not exposed to such therapeutics (12 vs 10 days; Table 3). Overall, at 30 days, the mean time to a postdischarge major bleeding event was 10 days (SD = 9.4 days). Figure 2 shows Kaplan-Meier curves for cumulative survival without major bleeding up to 30 days, stratified by the use of antithrombotic medications.

Figure 2.

Kaplan-Meier estimation of cumulative survival without major bleeding up to 30 days following discharge.

The rates of minor bleeding at 14 and 30 days were 12% and 16% among patients who used antithrombotics and 4.2% and 6.2% among those who did not use these therapeutics postdischarge, respectively. The time to occurrence of minor bleeding events at 14 and 30 days was similar for both the patient groups, 4 and 8 days, respectively.

Approximately 1 of 14 (7.3%) of patients was readmitted to the hospital with the same diagnosis as the index hospitalization (ie, other than bleeding) during the 30 days postdischarge. A higher proportion of the patients who received antithrombotics was rehospitalized than those who did not receive such therapeutics (9.4% vs 7%; P<.001). The opposite pattern was observed with respect to postdischarge short-term (at 30 days) all-cause mortality; a higher death rate occurred in those without post-discharge antithrombotic drug exposure than in those who were exposed to it (2.2% vs 0.5%; P <.001; Table 3). Among patients who received antithrombotic medications compared to those who did not, the VTE rates at 30 days were 3.6% vs 0.5%, a difference that was statistically significant (P < .001).

After adjusting for differences in patient characteristics via multivariate analyses, several independent predictors of major bleeding at 30 days postdischarge were identified (Table 4). At 30 days, the strongest predictor of major bleeding was rehospitalization (with the same diagnosis as the index hospitalization; HR = 2.37, 95% confidence interval [CI] 2.23-2.56), followed by preindex gastroduodenal ulcer (HR = 2.08, 95% CI 1.67–2.59), thromboembolic stroke (HR = 1.7, 95% CI 1.55-1.88), blood dyscrasias (HR = 1.69, 95% CI 1.59-1.8), and liver disease (HR = 1.4, 95% CI 1.24-1.58). Other risk factors included increasing age, male gender, and an index hospital stay of ≥3 days. Interestingly, a preindex cancer diagnosis appeared to have a protective effect (HR = 0.68, 95% CI 0.61-0.75), while postdischarge use of antithrombotic medications did not reach statistical significance as a predictor of major bleeding at 30 days (Table 4).

Table 4.

Multivariate Survival Analysis: Predictors of Major Bleeding Within 30 Days After Hospitalization.

Variable	Hazard Ratio	P Value	95% Confidence Intervals
Demographics
Age 40-54 yrs (reference group)
Age 55-64 yrs	1.34	<.001	(1.21-1.47)
Age 65-74 yrs	1.69	<.001	(1.55-1.84)
Age ≥75 yrs	2.08	<.001	(1.93-2.25)
Male gender	1.22	<.001	(1.16-1.28)
Preindex risk factors
Insufficient renal function	1.23	.173	(0.92-1.64)
Cancer ^a	0.68	<.001	(0.61-0.75)
Rheumatoid arthritis ^a	1.04	<.001	(0.81-1.34)
Gastroduodenal ulcer	2.08	<.001	(1.67-2.59)
Blood dyscrasias	1.69	<.001	(1.59-1.80)
Thrombocytopenia	1.16	.199	(0.93-1.45)
Liver disease	1.40	<.001	(1.24-1.58)
Central venous catheter	1.10	.324	(0.91-1.33)
Thromboembolic stroke	1.71	<.001	(1.55-1.88)
Estrogen use	0.95	.426	(0.82-1.09)
Postindex risk factors
Postdischarge anithrombotic meds use	1.05	.265	(0.97-1.14)
Rehospitalization/s	2.37	<.001	(2.23-2.56)
LOS ^b = 2 days (reference group)
LOS ^b = 3-5 days	1.19	<.001	(1.11-1.27)
LOS ^b ≥6 days	1.34	<.001	(1.24-1.45]

Abbreviations: yrs, years (of age); LOS, length of stay.

^a Rheumatoid arthritis and cancer diagnoses were captured at index admission; all other preindex risk factors occurred at any time during the pre-index period.

^bLength of index hospitalization.

Discussion

The current study extends our understanding of the risk factors for major bleeding and the therapeutic decisions for prevention of postdischarge thromboembolism following hospital discharge in medically ill patients. At 30 days, all and major bleeding rates of 8.9% and 1.8% were observed, with 9.1% of the patients receiving postdischarge antithrombotics in the study cohort. The primary predictors of postdischarge major bleeding identified in this study were rehospitalization for reasons other than bleeding, preindex gastroduodenal ulcer, thromboembolic stroke, blood dyscrasias, and liver disease along with several other risk factors such as increasing age, male gender, and an index hospital stay of ≥3 days. These results add to the findings of previous work in this field.^{14
–16} The postdischarge major bleeding risk factors from this analysis show a great degree of overlap on 5 variables at 30 days with those included in the inhospital bleeding RAM derived from the International Medical Prevention Registry On Venous Thromboembolism (IMPROVE),⁵ and they also coincide with some of the risk factors contained in the IMPACT-ILL⁵ and Padua Prediction Score VTE RAMs.^31,32 Additional risk factors listed in the IMPROVE bleeding RAM include a prior history of bleeding, renal failure rheumatic disease, a stay in the intensive care unit, central venous catheter placement, and cancer.⁵ This study also found that an index hospitalization of ≥3 days, rehospitalization for reasons other than bleeding, blood dyscrasias, and thromboembolic stroke are independent predictors of postdischarge bleeding. Interestingly, postdischarge antithrombotic use was not a predictor of major bleeding at 30 days. This may be explained by the fact that usually the postdischarge VTE prophylaxis with anticoagulants in the US patients diagnosed with an acute medical illness is suboptimal and provided for a rather short duration.³³ In addition, only 0.2% of the study patients used a combined anticoagulant–antiplatelet therapy after hospitalization. It was somewhat surprising to find out that the diagnosis of cancer during the preadmission period was a protective factor against major bleeding at 30 days. This finding is, however, consistent with previous reports, which showed that the use of VTE prophylaxis in patients with cancer is complicated by the fact that although they are at an increased risk of VTE, they are also at an increased risk of bleeding.^34,35 Therefore, patients with cancer are significantly less likely than noncancer patients to receive thromboprophylaxis prior to VTE occurrence.³⁶ In summary, our findings about the risk factors for major bleeding have important implications for existing bleeding RAMs for medically ill patients. After validation by other studies, they can be used to refine the existing RAMs for use in other clinical settings.^15,32

The rates of postdischarge major bleeding reported here are consistent with previously noted clinical studies.^8,9,15 However, overall bleeding rates appear somewhat higher than those described in the literature. This can be explained by the type of therapeutic exposure that, in addition to anticoagulants, included frequent use of antiplatelet agents, known risk factors for gastrointestinal bleeding, especially when used concomitantly with steroids and nonsteroidal anti-inflammatory drugs.^37,38 Additionally, this may be due to the difference in “real-world” rates of bleeding compared to the bleeding events reported by carefully controlled clinical studies.

In this study, many patients who were at risk of bleeding were prescribed antithrombotic medications after hospital discharge. This finding highlights the need for improved bleeding risk assessment approaches and further research to maximize patient care and safety. Use of antithrombotic agents in this patient population, and likely in other patient groups, should be guided by risk stratification for both VTE and bleeding in order to identify the best therapeutic approach to patient care.

Although 9% of the study population received antithrombotic agents, only 3.7% of the patients received an anticoagulant agent alone (predominantly warfarin or enoxaparin). These findings are consistent with previous reports of low rates of anticoagulant use for postdischarge prevention of thromboembolism in this patient setting.^{3
–5,39,40} The American College of Chest Physicians (ACCP) recommends LMWH, unfractionated heparin, or fondaparinux as VTE prophylaxis for hospitalized medically ill patients, with mechanical thromboprophylaxis recommended only for patients in whom anticoagulants are contraindicated. Furthermore, the ACCP does not recommend extending VTE prophylaxis after discharge.⁴¹ In our study, despite the strict exclusion of patients already receiving chronic warfarin therapy, warfarin was still the primary antithrombotic agent used postdischarge.

A key finding was that higher rates of postdischarge adverse outcomes, including bleeding, VTE, and rehospitalization (for reasons other than bleeding), were observed in patients who received antithrombotic agents compared to those who did not. This appears to suggest that clinicians tend to prescribe postdischarge antithrombotics for patients they consider to be at higher risk of postdischarge morbidity. In support of this interpretation, the preindex DCI score clearly showed statistically significant higher comorbidity scores, older age, and longer hospital stay in patients with postdischarge antithrombotic use than in those without such use. Hence, we hypothesized that practitioners prescribe antithrombotic agents, including anticoagulants for VTE prophylaxis, more frequently for patients with more severe medical conditions than for those with less severe medical conditions. On the other hand, the fact that the all-cause mortality was significantly lower among patients with the use of postdischarge antithrombotic agents than among those without such usage may in part reflect reluctance among physicians to administer antithrombotic medications to patients with terminal illnesses.

The strength of this study was the use of an integrated health care claims database that allowed access to a large, diverse patient population. This enabled robust multivariate analyses over multiple disease categories in hospitalized medical patients. Study limitations included the retrospective design, which may have allowed for classification bias. For example, the use of aspirin is likely underrepresented in this study due to the fact that it is an over-the-counter drug in the United States, and therefore is not well captured by pharmacy claims data. Certain comorbid conditions may have not been captured or may have been underreported because health care claims may contain data gaps and coding errors. Reliance on ICD-9-CM codes to identify bleeding events and use grouping antithrombotic drugs all together made the study more inclusive, but it also may have affected the specificity of the findings. Finally, although we adjusted for preindex characteristics in the multivariate analysis, unmeasured confounders, such as the use of over-the-counter medications like aspirin, may still limit statistical inferences from the data. Despite these limitations, we believe the results provide important new evidence on postdischarge utilization of antithrombotics and risk factors for bleeding following hospitalization of medically ill patients which can guide future studies.

Conclusion

To our knowledge, this is the first study to specifically examine postdischarge bleeding rates and clinically relevant bleeding risk in hospitalized medically ill patients. Several identified predictors of postdischarge bleeding overlap with those reported for postdischarge VTE. Use of a risk stratifying approach for both bleeding and VTE would help clinicians guide therapeutic decisions for prevention of postdischarge thromboembolism. It must be recognized, however, that bleeding is multifactorial, and careful consideration of associated comorbidities and concomitant therapies needs to be taken into account prior to either initiating anticoagulation or combining potent antithrombotic drugs in medically ill patients. Physician concerns about increased bleeding risk with the use of antithrombotic medications are not unfounded, as any amount of clinically relevant bleeding is known to prognosticate significant increases in the risk of postdischarge complications, as well as death. Further research aimed to evaluate the benefit–risk ratio of antithrombotics use is warranted.

Footnotes

Appendix A

Acknowledgments

The authors would like to acknowledge Ruth Sussman, PhD, for providing editorial support with funding from Janssen Scientific Affairs, LLC.

Authors’ Note

Dr Mahan contributed to the study design, interpretation of the data and statistical results, and preparation of the manuscript. He will serve as the guarantor of the article, taking responsibility for the integrity of the work as a whole. Dr Spyropoulos contributed to the study design, interpretation of the data and statistical results, and critical review of the manuscript. Dr Fisher contributed to the study design, interpretation of the data collection and processing, and interpretation of the statistical analysis. Ms Stephenson contributed to the study design, protocol development, and interpretation of the data and statistical results. An-Chen Fu performed the data collection, processing, and statistical analyses. Dr Fields contributed to the interpretation of the data and statistical results and critical review of the manuscript. Dr Mills contributed to the study conception, interpretation of the data and statistical results, and critical review of the manuscript. Dr Klaskala contributed to the study conception and design, protocol development, interpretation of the data and statistical results, preparation and critical review of the manuscript, obtaining funding, and supervising the project, before his untimely death in December 2012. Data analyses were performed at HealthCore, Inc. All authors had full unrestricted access to the data. All authors participated in drafting and revising the manuscript.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Maxine Fisher, An-Chen Fu, and Judith Stephenson have no conflicts of interest to declare. Alex Spyropoulos has served as a consultant for Boehringer Ingelheim, Johnson & Johnson (J&J), Eisai, Bayer, and Astellas, and has served in advisory committees for Bristol-Myers Squibb, Boehringer Ingelheim, and J&J. Charles Mahan has received investigator-initiated grants from Sanofi-Aventis and funding from the North American Thrombosis Forum Traveling Fellowship Award, served as a consultant for Sanofi-Aventis, Boehringer Ingelheim, J&J, Polymedix Inc., Leo, and Eisai, and served as a speaker for Sanofi-Aventis, Boehringer Ingelheim, and Eisai.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Janssen Scientific Affairs, LLC.

References

Zhan

Miller

. Excess length of stay, charges, and mortality attributable to medical injuries during hospitalization. JAMA. 2003;290(14):1868–1874.

Kearon

. Natural history of venous thromboembolism. Circulation. 2003;107(23 suppl 1):I22–I30.

Amin

Spyropoulos

Dobesh

. Are hospitals delivering appropriate VTE prevention? The venous thromboembolism study to assess the rate of thromboprophylaxis (VTE start). J Thromb Thrombolysis. 2010;29(3):326–339.

Dylan

Lin

Dubois

. Hospitals’ compliance with prophylaxis guidelines for venous thromboembolism. Am J Health Syst Pharm. 2007;64(1):69–76.

Tapson

Decousus

Pini

. Improve investigators. Venous thromboembolism prophylaxis in acutely ill hospitalized medical patients: findings from the international medical prevention registry on venous thromboembolism. Chest. 2007;132(3):936–945.

Tapson

. Prophylaxis strategies for patients with acute venous thromboembolism. Am J Manag Care. 2001;7(17 suppl):S524–S531.

McKean

Deitelzweig

Sasahara

Michota

Jacobson

. Assessing the risk of thromboembolism and identifying barriers to thromboprophylaxis in the hospitalized patient. J Hosp Med. 2009;4(8 suppl):S1–S7.

Cohen

Davidson

Gallus

. ARTEMIS investigators. Efficacy and safety of fondaparinux for the prevention of venous thromboembolism in older acute medical patients: randomised placebo controlled trial. Brit Med J. 2006;332(7537):325–329.

Samama

Cohen

Darmon

J-Y

. Prophylaxis in medical patients with enoxaparin study group. A comparison of enoxaparin with placebo for the prevention of venous thromboembolism in acutely ill medical patients. N Engl J Med. 1999;341(11):793–800.

10.

Ferretti

Bria

Giannarelli

. Major bleedings in the comparisons between low-molecular-weight heparin versus oral anticoagulant therapy for venous thromboembolism. Thromb Res. 2007;119(4):525–529.

11.

Zidane

Schram

Planken

. Frequency of major hemorrhage in patients treated with unfractionated intravenous heparin for deep venous thrombosis or pulmonary embolism: a study in routine clinical practice. Arch Intern Med. 2000;160(15):2369–2373.

12.

Ellis

Hadari

Tchuvero

. Hemorrhagic complications in patients treated with anticoagulant doses of a low molecular weight heparin (enoxaparin) in routine hospital practice. Clin Appl Thromb Hemost. 2006;12(2):199–204.

13.

Rothschild

Conen

. Characteristics of bleeding complications in patients with anticoagulation treatment. Swiss Med Wkly. 2008;138(47-48):719–724.

14.

Kuijer

Hutten

Prins

Büller

. Prediction of the risk of bleeding during anticoagulant treatment for venous thromboembolism. Arch Intern Med. 1999;159(5):457–460.

15.

Decousus

Tapson

Bergmann

; Improve investigators. Factors at admission associated with bleeding risk in medical patients: findings from the Improve investigators. Chest. 2011;139(1):69–79.

16.

Trujillo-Santos

Herrera

Page

; RIETE investigators. Predicting adverse outcomes in outpatients with acute deep vein thrombosis: findings from the RIETE Registry. J Vasc Surg. 2006;44(4):789–793.

17.

Schalekamp

Klungel

Souverein

de Boer

. Effect of oral antiplatelet agents on major bleeding in users of coumarins. Thromb Haemost. 2008;100(6):1076–1083.

18.

Dentali

Douketis

Lim

Crowther

. Combined aspirin-oral anticoagulant therapy compared with oral anticoagulant therapy alone among patients at risk for cardiovascular disease. Arch Intern Med. 2007;167(2):117–124.

19.

Larson

Fisher

. Should aspirin be continued in patients started on warfarin? J Gen Intern Med. 2004;19(8):879–886.

20.

Penning-van Beest

Erkens

Petersen

Koelz

Herings

. Main co medications associated with major bleeding during anticoagulant therapy with coumarins. Eur J Clin Pharmacol. 2005;61(5-6):439–444.

21.

Hallas

Dall

Andries

. Use of single and combined antithrombotic therapy and risk of serious upper gastrointestinal bleeding: population based case-control study. Br Med J. 2006;333(7571):726.

22.

Ruíz-Giménez

Suárez

González

. RIETE investigators. Predictive variables for major bleeding events in patients presenting with documented acute venous thromboembolism: findings from the RIETE Registry. Thromb Haemost. 2008;100(1):26–31.

23.

Jiménez

Yusen

Otero

. Prognostic models for selecting patients with acute pulmonary embolism for initial outpatient therapy. Chest. 2007;132(1):24–30.

24.

Tamariz

Harkins

Nair

. A systematic review of validated methods for identifying venous thromboembolism using administrative and claims data. Pharmacoepidemiol Drug Saf. 2012;21(suppl 1):154–162.

25.

Vekeman

LaMori

Laliberté

. In-hospital risk of venous thromboembolism and bleeding and associated costs for patients undergoing total hip or knee arthroplasty. J Med Econ. 2012;15(4):644–653.

26.

Birman-Deych

Waterman

Yan

Nilasena

Radford

Gage

. Accuracy of ICD-9-CM codes for identifying cardiovascular and stroke risk factors. Med Care. 2005;43(5):480–485.

27.

Schalekamp

Klungel

Souverein

de Boer

. Effect of oral antiplatelet agents on major bleeding in users of coumarins. Thromb Haemost. 2008;100(6):1076–1083.

28.

White

Beyth

Zhou

Romano

. Major bleeding after hospitalization for deep-venous thrombosis. Is J Med. 1999;107(5):414–424.

29.

Raeford

Perez Guttmann

Garcia Rodriguez

. Positive predictive value of ICD-9 codes in the identification of cases of complicated peptic ulcer disease in the Saskatchewan hospital automated database. Epidemiology. 1996;7(1):101–104.

30.

Deyo

Cherkin

Ciol

. Adapting a clinical co morbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiology. 1992;45(6):613–619.

31.

Spyropoulos

Anderson

Jr Fitzgerald

. Predictive and associative models to identify hospitalized medical patients at risk for VTE. Chest. 2011;140(3):706–714.

32.

Barbar

Noventa

Rossetto

. A risk assessment model for the identification of hospitalized medical patients at risk for venous thromboembolism: the Padua Prediction Score. J Thromb Haemost. 2010;8(11):2450–2457.

33.

Wang

Sengupta

Baser

. Risk of venous thromboembolism and benefits of prophylaxis use in hospitalized medically ill US patients up to 180 days post-hospital discharge. Thrombosis J. 2011;9(1):15.

34.

Monreal

Falgá

Valdés

Suárez . Riete investigators. Fatal pulmonary embolism and fatal bleeding in cancer patients with venous thromboembolism: findings from the RIETE registry. J Thromb Haemost. 2006;4(9):1950–1956.

35.

Imberti

Agnelli

Ageno

. Master investigators. Clinical characteristics and management of cancer-associated acute venous thromboembolism: findings from the master registry. Haematologica. 2008;93(2):273–278.

36.

Seddighzadeh

Shetty

Goldhaber

. Venous thromboembolism in patients with active cancer. Thromb Haemost. 2007;98(3):656–661.

37.

Hashash

Shamseddeen

Skoury

Aoun

Barada

. Gross lower gastrointestinal bleeding in patients on anticoagulant and/or antiplatelet therapy: endoscopic findings, management, and clinical outcomes. J Clin Gastroenterol. 2009;43(1):36–42.

38.

Barada

Abdul-Baki

El Hajj

Hashash

Green

. Gastrointestinal bleeding in the setting of anticoagulation and antiplatelet therapy. J Clin Gastroenterol. 2009;43(1):5–12.

39.

Cohen

Tapson

Bergmann

. Venous thromboembolism risk and prophylaxis in the acute hospital care setting (ENDORSE study): a multinational cross-sectional study. Lancet. 2008;371(9610):387–394.

40.

Amin

Varker

Princic

Lin

Thompson

Johnston

Duration of venous thromboembolism risk across a continuum in medically ill hospitalized patients. J Hosp Med. 2012;7(3):231–238. doi:10.1002/jhm.1002

41.

Kahn

Lim

Dunn

. Prevention of VTE in nonsurgical patients: antithrombotic therapy and prevention of thrombosis, 9th end: American College of Chest Physicians evidence-based clinical practice guidelines. Chest. 2012;141(2 suppl):e195S–e226S. doi:10.1378/chest.11-2296

Antithrombotic Medication Use and Bleeding Risk in Medically Ill Patients After Hospitalization

Abstract

Background:

Methods:

Results:

Conclusions:

Keywords

Introduction

Methods

Data Source

Patient Population

Study Analysis

Results

Discussion

Conclusion

Footnotes

Appendix A

Acknowledgments

Authors’ Note

Declaration of Conflicting Interests

Funding

References