This Article Abstract Full Text (PDF) Free All Versions of this Article: chest.06-2628v1 131/5/1508    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Article Archive Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by van Walraven, C. Articles by Forster, A. J. Search for Related Content PubMed PubMed Citation Articles by van Walraven, C. Articles by Forster, A. J. Related Content Related Editorial

Burden of Potentially Avoidable Anticoagulant-Associated Hemorrhagic and Thromobembolic Events in the Elderly^*

^* From the Ottawa Health Research Institute, Ottawa, ON, Canada.

Correspondence to: Carl van Walraven, MD, MSc, Clinical Epidemiology Program, Ottawa Health Research Institute, C404, Ottawa Hospital, Civic Campus, 1053 Carling Ave, Ottawa ON, Canada K1Y 4E9; e-mail: carlv{at}ohri.ca

Abstract

Background: On average, patients receiving therapy with oral anticoagulants (OACs) in the community are in the therapeutic range only 55% of the time. Anticoagulation control strongly influences the risk of hemorrhagic and thromboembolic events in such patients. However, not all anticoagulation-associated events are attributable to poor anticoagulation control, nor do all hemorrhagic or thromboembolic events occur in anticoagulated patients.

Objective: Measure the proportion of serious hemorrhagic and thromboembolic events that would be avoided if anticoagulation control was perfect.

Methods: A retrospective cohort study of eastern Ontario using population-based administrative databases. Anticoagulation control was determined for each day of OAC exposure using linear interpolation. Incident hemorrhagic or thromboembolic hospitalizations for control and OAC patients were identified. Hemorrhages and thromboemboli in OAC patients were deemed to be avoidable if they occurred at international normalized ratios of > 3 and < 2, respectively.

Results: The study included > 183,000 patient-years of observation with 6,400 patient-years of OAC exposure. Anticoagulation control could be determined for 51.5% of OAC exposure time. Control patients had hemorrhagic and thromboembolic event rates of 1.8% and 1.5% per year, respectively. A total of 10,020 people were exposed to OACs, and spent 14.2% and 26.7% of the time, respectively, with excessively high and low anticoagulation intensity. Excessively high anticoagulation intensity explained 25.6% (95% confidence interval [CI], 19.4 to 31.7) and 2.0% (95% CI, 1.5 to 2.5) , respectively, of all serious hemorrhages in the anticoagulated and entire population. Excessively low anticoagulation intensity explained 11.1% (95% CI, 4.4 to 17.7) and 1.1% (95% CI, 0.7 to 1.6) of all thromboemboli, respectively.

Conclusions: Our study showed that extreme anticoagulation intensity significantly impacted the health of the population. Improving anticoagulation control will have significant effects on the incidence of serious hemorrhagic and thromboembolic events in the both the anticoagulated and entire populations.

Key Words: anticoagulation • biostatistics • bleeding • deep venous thrombosis • gastroenterology • pulmonary embolism

On average, patients receiving long-term anticoagulation therapy are in the therapeutic range 55% of the time.¹ The risk of hemorrhagic and thromboembolic events in anticoagulated patients is strongly associated with anticoagulation control.^234567891011 Improved anticoagulation control, possibly with interventions including anticoagulation clinics¹²¹³¹⁴ and patient self-monitoring,¹⁵ should be a therapeutic goal for all anticoagulated patients.

However, the potential population-level benefit of improved anticoagulation control is unclear for three reasons. First, not all hemorrhagic or thromboembolic events occur in patients receiving anticoagulant therapy. Obviously, the risk of these events would not change with improved anticoagulation control. Second, not all anticoagulation-associated events are due to poor anticoagulation control. With a target international normalized ratio (INR) range of 2 to 3, hemorrhagic events at INRs of < 3 and thromboembolic events at INRs of > 2 would not be avoided with better anticoagulation control. Finally, not all anticoagulant-related clinical events occurring at extreme INRs are avoidable with perfect anticoagulation control. This is due to an unavoidable event risk that all people have regardless of anticoagulation levels.¹⁶

To our knowledge, no study has previously measured the potential effect of improved anticoagulation control at a population level. While the proportion of GI bleeding with exposure to anticoagulants has been measured,¹⁷ three factors need to be considered to truly quantify the health benefit of improved anticoagulation control. First, both thromboembolic and hemorrhagic events must be considered. Second, only events that occur outside of therapeutic INRs should be counted. Finally, the population effect of eradicating all extreme INRs should be quantified using accepted formulas for population-attributable risk (PAR).¹⁸ The PAR quantifies the expected effect of eradicating the exposure of interest.¹⁹ In this study, we estimated the proportion of serious hemorrhagic and thromboembolic events that would be avoided if extreme anticoagulation levels were replaced by therapeutic INRs.

Materials and Methods

Study Setting
The study occurred between September 1, 1999, and September 1, 2000, in eastern Ontario (^{Appendix A}). Costs of all hospitalizations, physician visits, and laboratory tests were covered by the publicly administered health-care system. To estimate the elderly population in the study area at the midpoint (ie, February 1, 2000), we used exponential interpolation²⁰ between census counts in 1996 and 2001 for people 65 years of age in the census divisions that comprise the study area.

Databases Used in the Study
This study used five administrative databases. The Ontario Drug Benefit Database (ODBD) records the medication, amount dispensed, and date of all prescriptions for Ontarians > 65 years of age. Because the ODBD does not capture all data for people < 65 years of age, we limited our study to the elderly. The Registered Persons Database (RPD) records the location, sex, date of birth, date of death, and date of last known contact with the Ontario health-care system. The Database of Laboratory Tests in Eastern Ontario (DOLTEON) contains the date and result of 98.5% of INRs from both private and hospital-based medical laboratories in eastern Ontario between September 1, 1999, and September 1, 2000.²¹ The Discharge Abstract Database (DAD) records all admissions to Ontario hospitals and documents diagnoses in a standardized fashion. Finally, the Ontario Myocardial Infarction Database records data on all people admitted to the hospital for acute myocardial infarction.²² All databases are anonymous and were linked by common, scrambled, unique patient identifiers. The study was approved by the Research Ethics Board of the Sunnybrook and Women’s College Hospital.

Determining Exposure to Anticoagulants
People were defined as having been exposed to oral anticoagulants (OACs) if at least two OAC prescriptions had been dispensed to them within 100 days of each other during the study period (from ODBD), and if they had undergone at least one INR test at a community or hospital laboratory (from DOLTEON). The 100-day time frame was used because it is the maximum prescription duration for OBDB dispensations.²³ OAC exposure started on the date of the first OAC prescription during the study period. If an OAC prescription was filled prior to the study within 100 days of the dispensation of the first study prescription, OAC exposure started on September 1, 1999. OAC exposure continued as long as the prescription was renewed within 100 days of the previous dispensation.

Because of dosing adjustments, OAC prescriptions can extend beyond 100 days (Fig 1 ). If a person’s OAC prescription was not renewed within 100 days, we classified them as remaining exposed to OAC as long as they had undergone INR testing at least every 8 weeks (from DOLTEON) and their INR did not remain < 1.5 for > 8 weeks (also from DOLTEON). Otherwise, OAC exposure ended on the last date that these criteria were met. Observation also ended on the date of death (from the RPD) or the date of the last known contact with the health-care system (also from the RPD).

View larger version (7K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Schematic outline of how observation time was classified for patients exposed to OACs. This figure illustrates how the observation time of OAC patients was classified. Observations were classified as "not exposed to OAC" (blue), "unmonitored or discontinued OAC exposure" (green), or "monitored OAC exposure" (red). This classification was based on the timing of OAC prescriptions (top row) and INR testing (bottom row). Two OAC prescriptions < 100 days apart were required to be classified as having been exposed to OACs. After the last prescription, patients remained exposed to OACs as long as the result of INR test of > 1.5 was recorded within 8 weeks of the last prescription. Two INR tests < 8 weeks apart were required for monitoring. See text for details.

Classification of Observation Time for OAC People
The observation time for people receiving therapy with OACs was classified into three mutually exclusive groups (Fig 1). The time when people were not receiving OACs was classified as "not exposed to OAC." The time when people were exposed to OACs but had no INR value was classified as "unmonitored or discontinued OAC exposure." All other time was classified as "monitored OAC exposure."

Outcomes
We considered both hemorrhagic and thromboembolic events requiring hospitalization. These events were identified in the DAD using the prespecified diagnostic codes listed in ^{Appendix B}. All acute myocardial infarctions were identified by linking to the Ontario Myocardial Infarction Database. We determined whether people experiencing events were from the study area using the first three alphanumerics of the postal code listed in the DAD (^{Appendix A}). For each person, we only considered the first hemorrhagic or thromboembolic event occurring during the study period. All observation time following the first event was censored. For people receiving therapy with OACs, the event was assigned to the exposure classification (outlined above) in which the person resided when the event had occurred (Fig 1).

For OAC patients, the INR at the time of the event was determined by linking to the DOLTEON. The INR that was closest to the hospital admission date was assigned to each outcome. If a patient had undergone more than one INR determination on a given day, we used the highest INR to account for the administration of vitamin K or plasma. Outcomes that had no INR recorded within 3 days of the event were categorized as "INR unknown."

Analysis
The primary outcome was the incident hemorrhagic or thromboembolic rate expressed as the number of events per 100 patient-years of observation. All monitored OAC exposure was categorized into clinically relevant INR ranges. The 95% confidence intervals (CIs) for incidence rates were calculated using exact methods if the number of events was 20.²⁵ Otherwise, the normal approximation to the Poisson distribution was used.

To quantify how extreme anticoagulation intensity contributed to population-based event rates, we calculated the PAR. The PAR is the proportion by which the incident rate of the outcome in the entire population would be reduced if the exposure was eliminated (^{Appendix C}).¹⁹ We calculated the PAR of extreme anticoagulation intensity in the following two populations: all patients receiving therapy with anticoagulants; and all people in the study population. In the base analysis, we included the unmonitored OAC time and events in the nonextreme anticoagulation-intensity group. As a sensitivity analysis, we recalculated the PAR after excluding such unmonitored observation time.

To estimate the annual number of events that would be avoided if extreme anticoagulation intensity was avoided, we multiplied the PAR (and its 95% CI) by the annual number of events in the study population. To estimate the number of serious hemorrhagic and thromboembolic events that would be avoided if anticoagulation control was optimized, we calculated the PAR due to "extreme anticoagulation intensity." Since an INR range of 2 to 3 is recommended for most anticoagulated patients,²⁶ extreme anticoagulation control was defined as an INR of < 2 for thromboembolic events and an INR of > 3 for hemorrhagic events. Patients with previous valvular replacements have an INR range of 2.5 to 3.5 recommended for mitral tilting disk and bileaflet designs as well as all caged ball-and-disk valves.²⁷ People who had undergone previous valvular repair were identified by linking to the DAD back to 1988, but valve type was not specified. Therefore, all valvular heart disease patients had a target INR of 2.0 to 3.5, and hemorrhagic events of INR > 3.5 were attributed to extreme anticoagulation intensity.

Results

During the study period, the study area (^{Appendix A}) included 188,740 seniors. After excluding time following incident events, these people contributed a total of 183,570 years of observation for hemorrhagic events and 185,142 years of observation for thromboembolic events. During the study period, 10,020 people (5.3%) were prescribed an OAC, totaling 6,422 years of exposure time (3.5% of all population observation time). People who had received anticoagulation therapy had an average age of 77 years, and 50% were men. They spent 26.7% of the time with an INR of < 2 and 14.2% of the time with an INR of > 3.

Hemorrhagic Events
Table 1 describes hemorrhagic events. Control patients were admitted to the hospital for hemorrhagic events at a rate of 1.8% per year (95% CI, 1.7 to 1.8). Hemorrhagic risk was significantly higher in people receiving therapy with OACs (4.0% per year [95% CI, 3.6 to 4.6]; relative risk [RR], 2.3 [95% CI, 2.0 to 2.6]). Hemorrhagic rates during monitoring periods for people who had undergone anticoagulation therapy were the same as those for the control population. During monitoring, people who had undergone anticoagulation therapy had an overall hemorrhagic rate of 6.0% per year (RR vs control patients, 3.4; 95% CI, 3.0 to 3.9). The hemorrhagic risk was strongly associated with anticoagulation intensity. Bleeding rates increased significantly when INRs exceeded 3 (RR vs INR < 3, 19.4; 95% CI, 14.4 to 26.0). Overall, we saw similar patterns for each of the hemorrhagic subtypes.

PARs
Critically high anticoagulation intensity contributed significantly to hemorrhagic events (Fig 2 ). The PAR of critically high anticoagulation intensity for serious hemorrhagic events was 25.6% (95% CI, 19.4 to 31.7%) in patients who had received anticoagulation therapy and 2.0% (95% CI, 1.5 to 2.5%) in the entire population. This translates to an annual decrease of 67 serious hemorrhagic events in eastern Ontario (95% CI, 50 to 82) if all of the time that elderly patients who had received anticoagulation therapy spent with critically high INRs was avoided. When unmonitored OAC time was excluded, the PAR for hemorrhagic events in the anticoagulated and entire population increased insignificantly to 31.5% (95% CI, 23.4 to 39.5) and 2.0% (95% CI, 1.5 to 2.5), respectively. PAR estimates were similar for the hemorrhagic subtypes. The PAR of critically high INRs for lethal hemorrhages was 28.1% (95% CI, 9.0 to 47.2%) and 1.8% (95% CI, 0.5 to 3.0), respectively, for the anticoagulated and entire population.

View larger version (19K): [in this window] [in a new window] [Download PPT slide]   Figure 2. RRs and PARs of critical INRs for hemorrhagic and thromboembolic events. Critical INRs for hemorrhagic and thromboembolic events were those > 3 and < 2, respectively. For each outcome (horizontal axis), the RR (circles, left axis) and PAR (squares, right axis) of critical INRs is presented for anticoagulation patients (gray) and the entire population (black). The PAR is the proportion of events that would be avoided by eradicating critical INRs.

To our knowledge, this is the first study to quantify how extreme anticoagulation intensity contributes to important outcomes. We found that critically high anticoagulation intensity contributed significant numbers of hemorrhages to the population, explaining 25.6% and 1.9%, respectively, of all serious events in the anticoagulated and entire elderly populations. This means that eradicating INRs of > 3 would avoid one of every four serious anticoagulation-associated hemorrhages. We also found that critically low anticoagulation intensity was responsible for 11.1% and 1.1%, respectively, of all serious thromboembolic events in the anticoagulated and entire elderly populations. This means that eradicating subtherapeutic INRs would avoid 1 of every 10 anticoagulation-associated thromboemboli. Increased patient education²⁸ and the use of technologies that have been shown to significantly improved anticoagulation control^12131415 should show important benefits to the health of the population. Although achieving therapeutic INRs 100% of the time is impossible, our results quantify the maximal benefit of improving anticoagulation control in a population.

We believe that our estimates for the population burden of extreme anticoagulation intensity are valid and meaningful. First, we used a population-based study to estimate each component of PAR, resulting in PAR estimates that are precise and not biased from sampling. Second, PAR estimates are biased by risk persistence and confounding.¹⁸ The term risk persistence refers to the presence of an increased event probability even after the removal of a risk factor. This does not occur with extreme anticoagulation intensity. The association between extreme INRs and events is likely causal and not completely explained by confounders.²⁹

We were surprised by the increased risk of thromboembolic events at high INR levels (Table 1). Increased thromboembolic risk at high INR levels has been seen in several other studies.^8930313233 The cause of this observation is unclear. For our study, in which outcomes were counted using codes in hospitalization databases, we wonder whether patients with very high INRs were admitted to the hospital as a precautionary measure and were coded for the thromboembolic event that required anticoagulation therapy. Comorbidities might also explain this observation. For example, renal failure is associated with both an increased risk of high INR levels³⁴ and an increased risk of thrombotic events.³⁵³⁶ Finally, two patient groups with a higher risk of thromboembolic events, namely, those using an older generation of prosthetic valves and patients who had previously experienced thromboembolic events, commonly have a higher target INR. Such patients could increase the thromboembolic rate at high INR levels.

Another notable finding was the similarity in hemorrhagic rates recorded outside of monitoring periods between patients who had received anticoagulation therapy and the control patients (Table 1). This observation suggests to us that unmonitored OAC time could largely be due to patients who had temporarily discontinued OAC use.

Our study has several limitations that must be considered when interpreting its results. Our study included only elderly people. Since administrative databases were used to count events, some misclassification likely occurred from miscoding. Some of the outcomes used in the study, such as stroke, had relatively poor coding accuracy (^{Appendix B}). We did not capture events that proved lethal prior to hospitalization. We missed some OAC exposure time if patients had received prescriptions > 100 days apart and had prolonged time with an INR of < 1.5. We missed some OAC monitoring time if patients had no INR monitoring, were monitored at laboratories outside of the study area, or were self-monitoring. We did not measure the prevalence of comorbidities that could influence the hemorrhagic or thromboembolic risk, including thrombophilias and neoplasia. Our PAR estimates represent the benefit that could be derived from perfect anticoagulation control, which is unlikely to be achieved with OACs.

Our study shows that extreme anticoagulation intensity significantly impacted the health of the population. Improving anticoagulation control should significantly decrease the incidence of serious hemorrhagic and thromboembolic events. We believe that these results justify further implementation of interventions that have been shown to improve anticoagulation control as well as the continued search for other interventions that do the same.

Appendix A

Figure 3 presents the study area defined as all of Ontario east of the lines presented above. For census estimates, the census division/census subdivision numbers presented above defined the western boundary of the study area. For outcomes, patient location is indicated by their postal code. The first three alphanumerics of the postal codes that define the western border of the study area are presented.

View larger version (52K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Study area defined by census and postal code geography boundaries.

International Classification of Diseases, Ninth Revision, Clinical Modification codes indicating study hemorrhagic and thromboembolic events. Codes are shown in Table 2.

Appendix C

Formula for Calculating PAR

where pe is the proportion of individuals (or time) in the population exposed to the factor of interest and RR is the RR of the event in the exposed vs the nonexposed population. 95% CIs for the PAR were calculated using formulas from Walter.²⁹

Footnotes

Abbreviations: CI = confidence interval; DAD = Discharge Abstract Database; DOLTEON = Database of Laboratory Tests in Eastern Ontario; INR = international normalized ratio; OAC = oral anticoagulant; ODBD = Ontario Drug Benefit Database; PAR = population-attributable risk; RPD = Registered Persons Database; RR = relative risk

Dr. van Walraven had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

This study was supported in part by the Institute for Safe Medication Practices-Canada.

Dr. Forster was an Ontario Ministry of Health Career Scientist when this study was conducted.

The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.

Received for publication October 27, 2006. Accepted for publication January 4, 2007.

References

1. van Walraven, C, Jennings, A, Oake, N, et al (2006) Effect of study setting on anticoagulation control: a systematic review and metaregression. Chest 129,1155-1166[Abstract/Free Full Text]
2. Tangelder, MJD, Algra, A, Lawson, JA, et al Optimal oral anticoagulant intensity to prevent secondary ischemic and hemorrhagic events in patients after infrainguinal bypass graft surgery. J Vasc Surg 2001;33,522-527[CrossRef][ISI][Medline]
3. Cannegieter, SC, Rosendaal, FR, Wintzen, AR, et al Optimal oral anticoagulant therapy in patients with mechanical heart valves. N Engl J Med 1995;333,11-17[Abstract/Free Full Text]
4. Veeger, NJ, Piersma-Wichers, M, Tijssen, JG, et al Individual time within target range in patients treated with vitamin K antagonists: main determinant of quality of anticoagulation and predictor of clinical outcome: a retrospective study of 2300 consecutive patients with venous thromboembolism. Br J Haematol 2005;128,513-519[CrossRef][ISI][Medline]
5. Palareti, G, Manotti, C, DAngelo, A, et al Thrombotic events during oral anticoagulant treatment: results of the inception-cohort, prospective, collaborative ISCOAT study: iSCOAT study group (Italian Study on Complications of Oral Anticoagulant Therapy). Thromb Haemost 1997;78,1438-1443[ISI][Medline]
6. Oden, A, Fahlen, M Oral anticoagulation and risk of death: a medical record linkage study. BMJ 2002;325,1073-1075[Abstract/Free Full Text]
7. Stroke Prevention in Reversible Ischemia Trial (SPIRIT) Study Group.. A randomized trial of anticoagulants versus aspirin after cerebral ischemia of presumed arterial origin. Ann Neurol 1997;42,857-865[CrossRef][ISI][Medline]
8. European Atrial Fibrillation Trial Study Group.. Optimal oral anticoagulant therapy in patients with nonrheumatic atrial fibrillation and recent cerebral ischemia. N Engl J Med 1995;333,5-10[Abstract/Free Full Text]
9. Azar, AJ, Cannegieter, SC, Deckers, JW, et al Optimal intensity of oral anticoagulant therapy after myocardial infarction. J Am Coll Cardiol 1996;27,1349-1355[Abstract]
10. Casais, P, Luceros, AS, Meschengieser, S, et al Bleeding risk factors in chronic oral anticoagulation with acenocoumarol. Am J Hematol 2000;63,192-196[CrossRef][ISI][Medline]
11. Palareti, G, Leali, N, Coccheri, S, et al Bleeding complications of oral anticoagulant treatment: an inception-cohort, prospective collaborative study (ISCOAT); Italian Study on Complications of Oral Anticoagulant Therapy. Lancet 1996;348,423-428[CrossRef][ISI][Medline]
12. Wilson, SJ-A, Wells, PS, Kovacs, MJ, et al Comparing the quality of oral anticoagulant management by anticoagulation clinics and by family physicians: a randomized controlled trial. Can Med Assoc J 2003;169,293-298[Abstract/Free Full Text]
13. Samsa, GP, Matchar, DB, Goldstein, LB, et al Quality of anticoagulation management among patients with atrial fibrillation: results of a review of medical records from 2 communities. Arch Intern Med 2000;160,967-973[Abstract/Free Full Text]
14. Witt, DM, Sadler, MA, Shanahan, RL, et al Effect of a centralized clinical pharmacy anticoagulation service on the outcomes of anticoagulation therapy. Chest 2005;127,1515-1522[Abstract/Free Full Text]
15. Heneghan, C, Alonso-Coello, P, Garcia-Alamino, JM, et al Self-monitoring of oral anticoagulation: a systematic review and meta-analysis. Lancet 2006;367,404-411[CrossRef][ISI][Medline]
16. Hanley, JA A heuristic approach to the formulas for population attributable fraction. J Epidemiol Community Health 2001;55,508-514[Abstract/Free Full Text]
17. Johnsen, SP, Sorensen, HT, Mellemkjoer, L, et al Hospitalisation for upper gastrointestinal bleeding associated with use of oral anticoagulants. Thromb Haemost 2001;86,563-568[ISI][Medline]
18. Rockhill, B, Newman, B, Weinberg, C, et al Use and misuse of population attributable fractions. Am J Public Health 1998;88,15-19[Free Full Text]
19. Last, JM A dictionary of epidemiology. 1995 Oxford University Press. New York, NY:
20. Rowland, D Population growth and decline: demographic methods and concepts. 2003,45-75 Oxford University Press. London, UK:
21. van Walraven, C, Raymond, M A population-based study of repeat laboratory testing. Clin Chem 2003;49,1997-2005[Abstract/Free Full Text]
22. Tu, JV, Naylor, CD, Austin, P Temporal changes in the outcomes of acute myocardial infarction in Ontario, 1992–1996. Can Med Assoc J 1999;161,1257-1261[Abstract/Free Full Text]
23. Ontario Ministry of Health and Long Term Care.. Ontario Drug Benefit Formulary: Part 1. Introduction. 2003;I,14
24. Rosendaal, FR, Cannegieter, SC, van der Meer, FJ, et al A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost 1993;69,236-239[ISI][Medline]
25. Fleiss, JL, Levin, B, Paik, MC Poisson regression. Statistical methods for rates and proportions. 2003,340-372 John Wiley & Sons. Hoboken, NJ:
26. Ansell, J, Hirsh, J, Poller, L, et al The pharmacology and management of the vitamin K antagonists: the Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy. Chest 2004;126(suppl),204S-233S[Abstract/Free Full Text]
27. Salem, DN, Stein, PD, Al Ahmad, A, et al Antithrombotic therapy in valvular heart disease–native and prosthetic: the Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy. Chest 2004;126(suppl),457S-482S[Abstract/Free Full Text]
28. Lip, GY, Kamath, S, Jafri, M, et al Ethnic differences in patient perceptions of atrial fibrillation and anticoagulation therapy: the West Birmingham Atrial Fibrillation Project. Stroke 2002;33,238-242[Abstract/Free Full Text]
29. Walter, SD, Walter, SD Calculation of attributable risks from epidemiological data. Int J Epidemiol 1978;7,175-182[Abstract/Free Full Text]
30. Torn, M, Algra, A, Rosendaal, FR Oral anticoagulation for cerebral ischemia of arterial origin: high initial bleeding risk. Neurology 2001;57,1993-1999[Abstract/Free Full Text]
31. Tangelder, MJ, Algra, A, Lawson, JA, et al Optimal oral anticoagulant intensity to prevent secondary ischemic and hemorrhagic events in patients after infrainguinal bypass graft surgery: Dutch BOA Study Group. J Vasc Surg 2001;33,522-527[CrossRef][ISI][Medline]
32. Hylek, EM, Go, AS, Chang, Y, et al Effect of intensity of oral anticoagulation on stroke severity and mortality in atrial fibrillation. N Engl J Med 2003;349,1019-1026[Abstract/Free Full Text]
33. Chimowitz, MI, Lynn, MJ, Howlett-Smith, H, et al Comparison of warfarin and aspirin for symptomatic intracranial arterial stenosis. N Engl J Med 2005;352,1305-1316[Abstract/Free Full Text]
34. van Walraven C, Austin PC, Oake N, et al. The effect of hospitalization on oral anticoagulation control: a population-based study. Thromb Res 2006 (July 14, E-pub ahead of print)
35. Anavekar, NS, McMurray, JJ, Velazquez, EJ, et al Relation between renal dysfunction and cardiovascular outcomes after myocardial infarction. N Engl J Med 2004;351,1285-1295[Abstract/Free Full Text]
36. Tonelli, M, Wiebe, N, Culleton, B, et al Chronic kidney disease and mortality risk: a systematic review. J Am Soc Nephrol 2006;17,2034-2047[Abstract/Free Full Text]
37. Arnason, T, Wells, PS, van Walraven, C, et al Accuracy of coding for possible warfarin complications in hospital discharge abstracts. Thromb Res 2006;118,253-262[CrossRef][ISI][Medline]
38. van Walraven, C, Mamdani, M, Cohn, A, et al Risk of subsequent thromboembolism for patients with pre-eclampsia. BMJ 2003;326,791-792[Free Full Text]

This Article Abstract Full Text (PDF) Free All Versions of this Article: chest.06-2628v1 131/5/1508    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Article Archive Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by van Walraven, C. Articles by Forster, A. J. Search for Related Content PubMed PubMed Citation Articles by van Walraven, C. Articles by Forster, A. J. Related Content Related Editorial