• warfarin management;
  • computer-assisted warfarin management;
  • elderly;
  • quality of anticoagulation;
  • anticoagulation clinics


  1. Top of page
  2. Abstract

OBJECTIVES: To determine the effect of patient characteristics and of specific guidelines that were developed for managing warfarin therapy in older adults and included in an in-house computer program on anticoagulation quality.

DESIGN: Thirteen-month observational study.

SETTING: Acute care, extended care, and rehabilitation geriatric wards of a teaching hospital in Paris, France.

PARTICIPANTS: Hospitalized patients (N=307, mean age 86.1 ± 6.1) treated with warfarin with a therapeutic international normalized ratio range of 2.0 to 3.0.

INTERVENTION: Patients were assigned according to care unit to the computer-generated dosing group (CGD) or the standard management group (SM; usual physician-based care).

MEASUREMENTS: Relationships between anticoagulation quality criteria and covariates (age, sex, warfarin indication, treatment phase, follow-up duration, model of care).

RESULTS: According to multivariate analysis, only model of care and follow-up duration independently influenced anticoagulation control; the proportion of time within therapeutic INR range 2.0 to 3.0 was significantly greater in the CGD group than in the SM group (59% vs 48%, P=.004). When a wider INR range was analyzed (1.8–3.2), the proportion of time within range was 73% versus 64% (P=.006). Use of the computer was associated with fewer days with INRs greater than 3, a smaller percentage of INRs of 4 or greater, a longer time to the first INR of 4.0 or greater, and a smaller mean number of INRs per month than SM (all P<.01).

CONCLUSION: Initiation regimen and long-term rules that have specifically been developed and included in a computerized dosage program improve quality of anticoagulation in elderly inpatients, allowing them to benefit from a quality of care as high as that of younger ambulatory patients.

The number of elderly patients who require warfarin therapy is rising steadily, because the aging of the population is causing increases in the prevalences of atrial fibrillation and venous thromboembolism. Warfarin has a narrow therapeutic index, so achieving effective yet safe anticoagulation is challenging,1 especially in frail elderly patients, who are at high risk of bleeding.2 To minimize the risks associated with anticoagulant therapy, anticoagulation clinics have been developed and shown to improve outcomes more than standard care.2 Computer programs are generally used in anticoagulation clinics to determine warfarin dosages,2–5 but no computer programs specifically designed for elderly patients are available, although the use of specific algorithms for treatment initiation and closer monitoring are recommended in elderly patients to maximize time spent within the therapeutic range.2,6,7 In a previous study, a warfarin induction algorithm specifically designed for elderly inpatients and based on the international normalized ratio (INR) measured after three daily intakes of 4 mg was validated.8 Rules were then developed for adjusting warfarin dosages after Day 3. These induction and adjustment guidelines were included in a computer program for assisting in warfarin dosage selection in elderly inpatients. Various considerations were taken into account to design the elderly-specific computer-generated dosing software. Elderly patients require lower dosages than younger patients, although the mechanism for this greater sensitivity to warfarin with aging is not well understood; they are more prone to instability, because they often have comorbidities and use drugs that interact with warfarin, and their response to warfarin dosage adjustments is slower.2,6,7

The main aim of this observational study conducted in elderly inpatients treated with warfarin with a target INR range of 2.0 to 3.0 was to assess the effect of the computer program on anticoagulation control, especially on the time spent in the therapeutic range. The effect of patient characteristics on anticoagulation quality was also evaluated.


  1. Top of page
  2. Abstract

Warfarin Management

This observational study was conducted at the Charles Foix teaching hospital in Paris, France. The hospital has 800 beds in acute care and extended care geriatric wards and rehabilitation wards.

Management of anticoagulant therapy in older adults has been a concern in the hospital for many years, with a specific training of all new arriving physicians organized twice a year. The training consisted of one information session on the management of anticoagulant therapy, especially warfarin in older people. During this class, all physicians were advised how to use the specific recommendations for warfarin initiation in older people that are listed on a pocket-sized card. These recommendations included the low-dose warfarin induction algorithm based on Day 3 INR values that was previously validated in elderly inpatients and that is recommended in The Eighth American College of Chest Physicians (ACCP) Consensus Conference on Antithrombotic and Thrombolytic Therapy, completed by rules based on the Day 6 INR value (Table 1).2,8 Induction was defined as warfarin therapy from initiation to achievement of the maintenance dose (the dose that produces INR values in the 2.0–3.0 range on two consecutive samples). These guidelines were included in the “induction module” of a personal computer–based system developed to monitor oral anticoagulant treatments and that was implemented in the hematology laboratory.9 The program has a second main module, the “maintenance module,” that is used after achievement of the maintenance dose in the initiation period or, in the case of long-term warfarin treatment, for finely adjusting the dose to the therapeutic range and sustaining it. In this maintenance module, the software calculates whether the dosage should be adjusted based on several variables including current and previous INR values, frequency of INR monitoring, and warfarin dosage history. Briefly, the rules are as follows. If needed, warfarin dosage is adjusted in 1-mg steps at least 4 days apart. The INR is determined every few days until the maintenance dose is achieved and then less often, with the maximum interval being 3 weeks. If dosage adjustments are required, the INR is again determined at shorter intervals. In patients with INR values slightly above (3.0–3.2) or below (1.8–2.0) the therapeutic range, INRs are closely monitored before a dosage adjustment is performed. Supratherapeutic INRs (>5) are managed according to expert recommendations published in the seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy.10 A trained hematologist systematically reviews INR value, warfarin dosage, and time to the next INR determination indicated by the computer program before they are transmitted to the attending physicians. In summary, compared to younger patients, in elderly-specific guidelines, the starting dose is lower (4 mg) without daily sampling,8 and monitoring is closer, with a maximum interval of 3 weeks; the intervals between two dosage adjustments are longer, at least 4 days apart, especially in case of higher dosage, and if needed, the warfarin dosage is adjusted by 1-mg steps.

Table 1. Rules for Determining Warfarin Dosages
DayINR ValueWarfarin Dosage
  • *

    This algorithm does not apply to patients who have a pretreatment international normalized ratio (INR) >1.3.

  • Supratherapeutic INRs are managed according to expert recommendations.10

  • INRs were determined in the morning, and warfarin doses were given in the evening. The target INR was 2.5.

0Do not measure4 mg*
1Do not measure4 mg
2Do not measure4 mg
3<1.305 mg
1.30–1.494 mg
1.50–1.693 mg
1.70–1.892 mg
1.90–2.491 mg
≥2.50Measure INR daily and omit doses until INR <2.5, then give 1 mg
6≤1.6Increase by 1 mg/d
1.61–2.50Maintain dosage
If warfarin dosage ≥2 mgReduce by 1 mg/d
If warfarin dosage=1 mgMaintain dosage
>3.50Management of supratherapeutic INR

Warfarin was administered in the evening, and INRs were measured in the morning. Testing was performed at the local licensed clinical laboratory of the Charles Foix teaching hospital, using STA-Neoplastine CI (Diagnostica Stago, Asnières sur Seine, France) with an international sensitivity index of approximately 1.7 on a STAR coagulation analyzer (Diagnostica Stago).

Patients and Study Design

Over a 13-month period (June 1, 2006–June 30, 2007), consecutive inpatients aged 70 and older who were undergoing warfarin therapy or scheduled to be started on warfarin therapy with a therapeutic INR range of 2.0 to 3.0 were enrolled. Exclusion criteria were oral anticoagulant therapy other than warfarin within the previous 10 days. All patients were followed until discharged from the hospital or until warfarin therapy was stopped while still hospitalized. Patients hospitalized in two geriatric departments were assigned to computer-generated dosing (CGD) using the software developed for this study and those hospitalized in the other four geriatric departments of the hospital or to standard management (SM). The two departments of the CGD group and the four departments of the SM group were general geriatric wards that admitted the same types of patients. Specialized departments were excluded from the study (geriatric psychiatry and rehabilitation). The assignment of departments to the CDG or SM group was made to have the same type of patients, with a similar proportion of acute care, intermediate care, and chronic care patients in the two groups. Moreover, the proportion of attending senior and junior physicians was similar in the two groups. The senior physicians were trained geriatricians, and the junior physicians were general physicians (residents). All physicians worked in the designated wards and did not work in more than one ward. Physicians of patients in the CGD group were asked to manage warfarin therapy using suggestions made by the computer and transmitted by an experienced hematologist. The management of patients receiving SM care was at the discretion of attending senior and junior (residents) physicians.

A standardized computerized data collection form was used for each patient to record age, sex, indication for warfarin therapy, period of treatment (whether warfarin treatment was at the induction or maintenance phase), and follow-up duration (defined as the time between the first and the last INR measured during the study period). Major bleeding events, as previously defined,11 were also recorded. INRs measured during the study period were taken from the laboratory database.

The Medical Ethics Committee of the Charles Foix teaching hospital approved the study protocol.

Quality of Anticoagulation

Eight criteria were used to evaluate quality of anticoagulation: the percentages of time within the therapeutic INR range (2.5 ± 0.5) and within the INR range 2.5 ± 0.7 (the range for which no dosage adjustments are systematically performed), the percentage of time with INR values less than 2.0 and greater than 3.0; the percentage of INR values of 4.0 and greater, reflecting over anticoagulation; the percentage of INR values of 1.5 and less, reflecting subtherapeutic levels; the time to the first INR value of 4.0 or greater; and the number of INR determinations per month. Times spent within, above, and below the target INR range were calculated using the Rosendaal method.12 For patients with testing gaps longer than 30 days, only the period up to the last INR before the 30-day gap was analyzed. For all criteria, the INR values obtained within the first 6 days after study inclusion were excluded to ensure that patients started on warfarin were near their maintenance dose and that patients already on warfarin no longer had INR variations related to dosage adjustments made before the study.

Sample Size Calculation

The sample size was calculated as follows. Assuming that the proportion of time spent in the therapeutic range is 10 percentage points higher in patients included in the computer intervention group (60%) than in those in the standard management group (50%), with a standard deviation of approximately 28%,5 it was calculated with a ratio 0.5=2 that 186 patients were necessary in the SM group and 93 patients in the CGD group (two-sided Student t-test, significance level .05, power 80%).

Statistical Analysis

Of the eligible patients, those who had at least two INR determinations and a follow-up of at least 7 days during the study period were included. Patient and treatment characteristics in the two groups were compared using Student t-tests for quantitative variables and chi-square tests for qualitative variables.

Of the eight criteria used to evaluate anticoagulation, the time to the first INR of 4.0 or higher was considered censored at completion of follow-up in patients with no INR of 4.0 or higher, and the other seven criteria were considered quantitative variables. Univariate analyses were performed, followed by multivariable analyses to test relationships between the eight anticoagulation criteria and six covariables: age, sex, indication for warfarin therapy, treatment phase (initiation or maintenance), follow-up duration, and model of care (CGD or SM). Covariates with P<.10 according to univariate analysis were entered into stepwise regression models. For the seven quantitative criteria, the univariate tests were the Spearman rank correlation coefficient test for quantitative covariates and the two-sample Wilcoxon test for qualitative covariates; stepwise multiple linear regression was used for the final model. For time to the first INR of 4.0 or higher, the univariate analysis involved log-rank tests for qualitative covariates and univariate Cox models for quantitative variables; a stepwise multivariable Cox regression model was built. P<.05 was considered significant. Computations were performed using the SAS V8 statistical package (SAS Institute, Inc., Cary, NC).


  1. Top of page
  2. Abstract

Patient and Treatment Characteristics

Of the 355 eligible patients, 307 were included in the final analysis; the remaining 48 patients had fewer than two INR determinations or less than 7 days of follow-up. The 307 included patients had a mean age of 86.1 ± 6.1. Their main characteristics are shown in Table 2.

Table 2. Patient Characteristics in Each Model of Care Group
CharacteristicTotal (N=307)Standard Management (n=199)Computer-Generated Dosing (n=108)
  1. SD=standard deviation.

Age, mean ± SD86.1 ± 6.185.8 ± 6.386.6 ± 5.9
Female, %727369
Patients hospitalized in acute care unit, %454446
Indication for warfarin, %
 Atrial fibrillation697067
 Venous thromboembolism282729
 Arterial thrombosis334
Patients with warfarin treatment initiated during the study, %454154
Number of concomitant medications per patient, mean ± SD5.8 ± 25.7 ± 1.85.8 ± 2
Follow-up duration, days, mean ± SD83.2 ± 94.778.7 ± 96.791.4 ± 90.7

Of the 307 patients, 108 were assigned to the CGD group and 199 to the SM group. There were 16 physicians managing the patients in the CGD group and 30 in the SM group. Over the 13-month study period, 6,009 INR determinations were performed in the 307 study patients. Adherence of CGD group physicians to the dose suggestions made by the software was 91%.

Overall INR Outcomes

Overall, the 307 study patients spent 51.7 ± 28.0% of the time within the therapeutic INR range 2.5 ± 0.5, 66.9 ± 27.2% within the wider INR range 2.5 ± 0.7, 31.2 ± 28.2% with INR values less than 2.0, and 16.8 ± 20.7% with INR values greater than 3.0. The mean percentage of INRs of 4.0 or higher was 5.5 ± 12.5%, and the mean percentage of INRs of 1.5 or less was 11.5 ± 20.8% (Table 3).

Table 3. Effect of the Model of Care on Anticoagulation Control: Results of the Multivariable Analysis
Quality of Anticoagulation CriteriaMean ± Standard Deviation
Total (N=307)Model of Care
Standard Management (n=199)Computer-Generated Dosing (n=108) P-Value
  1. INR=international normalized ratio.

Percentage of time with INR 2.0–3.051.7 ± 28.048.3 ± 28.458.6 ± 26.2.004
Percentage of time with INR>3.016.8 ± 20.719.6 ± 23.012.0 ± 14.4.004
Percentage of time with INR<2.031.2 ± 28.232.2 ± 29.229.4 ± 26.2.39
Percentage of time with INR 1.8–3.266.9 ± 27.263.6 ± 28.373.2 ± 23.9.006
Percentage of INRs≥4.05.5 ± 12.57.0 ± 14.82.5 ± 4.4.004
Percentage of INRs≤1.511.5 ± 20.912.2 ± 22.510.0 ± 16.9.34
Monthly number of INR determinations10.7 ± 6.611.6 ± 7.39.1 ± 4.8.002

Factors Influencing Anticoagulation Control

Of the six covariates studied, only model of care (CGD or SM) and follow-up duration (defined as the time between the first and the last INR measured during the study period) were independently associated with anticoagulation control in the multivariable analysis. Adjusting for potential confounders did not change the strength of the associations with these two covariates (Table 3). Patients in the CGD group spent a significantly larger percentage of time within the therapeutic INR range 2.5 ± 0.5 (59%) than did patients in the SM group (48%, P=.004). When a wider INR range is analyzed (2.5 ± 0.7), the proportions of time within range are 73% in the CGD group and 64% in the SM group (P=.006) (∼10 percentage points better). The percentages of time spent with INR values higher than 3.0 and INR values of 4.0 or higher were smaller in the CGD group than in the SM group (Table 3). Finally, the median time to the first INR of 4.0 or higher was significantly longer in the CGD group than in the SM group (88 vs 50 days, P=.009) (Figure 1). Longer follow-up was associated with greater percentages of time within the therapeutic INR ranges 2.5 ± 0.5 (P<.001) and 2.5 ± 0.7 (P=.003). The mean number of INR determinations per month was smaller in the CGD group than in the SM group (9.1 ± 4.8 vs 11.6 ± 7.3, P=.002).


Figure 1.  Kaplan-Meier analysis of the time to the first international normalized ratio (INR) of 4.0 or higher according to the treatment group.

Download figure to PowerPoint

Bleeding Events

During the study period, seven of the 307 patients experienced a nonfatal major bleeding event: three in the CGD group and four in the SM group. The characteristics of the seven patients are shown in Table 4. In four patients, warfarin therapy had been initiated during the study period (Patients 1, 2, 6, and 7). In all patients, warfarin was discontinued; one patient received vitamin K (Patient 5), and four patients required blood transfusions (Patients 1, 2, 3, and 7).

Table 4. Characteristics of the Seven Patients Who Experienced a Major Bleeding Event During the Study Period
Patient NumberModel of CareMale/ FemaleAgeIndication for WarfarinWarfarin Initiated During the Study PeriodType of BleedingInternational Normalized Ratio the Day of BleedingTime Since the Beginning of the Study (Days)
  1. SM=standard management; CGD=computer-generated dosing; AF=atrial fibrillation; VTE=venous thromboembolism.

4SMF80AFNoIntraperitoneal bleeding3.0106
6SMF85AFYesIntraocular bleeding2.19
7SMM73VTEYesPsoas hematoma2.18


  1. Top of page
  2. Abstract

This study showed that software specifically designed to assist in warfarin dose determination in elderly inpatients resulted in better anticoagulation quality in hospitalized elderly patients than dose determination by the physicians. The 10 percentage points greater time spent within the therapeutic range is of similar magnitude to improvements reported earlier in studies comparing usual care with different models of intervention.5,13,14 A greater improvement would probably have been found had the hospital physicians not received specific training in warfarin dose management in elderly patients.

Despite evidence that warfarin is highly effective in preventing thromboembolism, it remains underused in elderly patients. Concern about inducing bleeding events and anticipated difficulties with anticoagulation control have been suggested as indications for not prescribing warfarin to elderly patients.6 Few data are available to guide clinicians in determining warfarin dosages for older adults. In most warfarin regimens, the initial dose of 10 mg often causes overanticoagulation during treatment induction in older patients, and the subsequent doses are determined using a nomogram based on the results of the daily INR value.6,8 Moreover, no protocols concerning the dose adjustments during the maintenance phase in older adults are available. The rules presented here and that were used in the computer program to manage the care of patients receiving warfarin treatment consisted of the algorithm that predicts the maintenance dosage from an INR value measured after the third daily intake of a 4-mg dose.8 This algorithm was extended beyond Day 3 using rules to manage long-term anticoagulation therapy in older adults. With the software, elderly inpatients spent 59% of the time within the therapeutic INR range 2.5 ± 0.5 and 73% within the INR range of 2.5 ± 0.7. This second, wider range is the range for which no dosage adjustments are systematically performed.

Previous studies comparing standard management with other approaches show considerable heterogeneity, particularly regarding design (retrospective or prospective), setting (hospital or community), and patient characteristics.2 In a systematic review of 67 studies representing more than 50,000 patients from anticoagulation clinics (68%), clinical trials (7%), or community practices (24%), the overall percentage of time spent in the therapeutic INR range was 63.5% (60.6% in patients treated with warfarin).13 Practice setting had the greatest effect on anticoagulation control, with time within the therapeutic range varying from 56.7% in community practices to 66.4% in randomized trials.13 Of the 67 studies, only two included all INR measurements, as opposed to only INRs performed as part of the anticoagulation management service.13 INR values obtained during acute illnesses are probably less often within the therapeutic range than those obtained as part of oral anticoagulant monitoring. In the current study, all INR values were included. In a retrospective population-based study of elderly patients (mean age 77 ± 7), the percentage of time spent within the therapeutic range (2.0–3.0) was 59.2% overall and was 15% smaller in hospitalized patients than in outpatients.15 In a study conducted in five countries (Canada, France, Italy, Spain, and the United States) to assess anticoagulation control in ambulatory patients with chronic nonvalvular atrial fibrillation requiring vitamin K antagonist therapy for stroke prophylaxis, the percentage of time with INR values between 2.0 and 3.0 in the 1,511 included patients ranged from 58% to 69.5% (59.3% in France).16 In an observational study of interventions to improve anticoagulation management by general practitioners in Belgium, including a CGD program, the percentage of time spent within 0.75 INR units from the chosen target of 2.5 or 3.5 in the study population with a mean age of 70 ranged across interventions from 73% to 80%.17 Finally, in a recent study in a population similar to that of the current study (110 hospitalized patients with a mean age of 85) managed without a dose-adjustment protocol, the time within the therapeutic range (2.0–3.0) was only 31%.18 The high rate of INRs within the therapeutic range obtained using the software is particularly remarkable, because the patients in the current study were hospitalized and older than those in previous studies and therefore more prone to anticoagulation instability due to polypharmacy and comorbid conditions. The software also significantly decreased overanticoagulation (percentage of days spent with INR values ≥4.0 and time to the first INR ≥4.0). Concern about overanticoagulation may make physicians reluctant to use anticoagulants in elderly patients. The results of the current study suggest that specific management may overcome this obstacle to appropriate anticoagulant therapy. Finally, using the software decreased the number of required INR determinations. Because the frequency of INR testing differed between the two management groups, it was determined that the better anticoagulation control in the computer group than in the standard management group was not related to the frequency of INR testing; there was a statistical relationship between the time spent within the two INR ranges and the frequency of INR testing, but after adjustment for INR frequency, the model of care and the time spent within the two INR ranges were independently associated.

The sample size was too small for an assessment of the effects of CGD on patient outcomes, but there is widespread agreement that anticoagulation control influences anticoagulation-related outcomes.1,2,14,19,20 Thus, improving anticoagulation control and minimizing INR fluctuations using a computer-assisted warfarin management system specifically devoted to elderly patients would probably result in better clinical outcomes in this population.

One limitation of this study is the absence of random allocation to the treatment groups. It is unlikely that this accounted for the difference in the quality of anticoagulation between the two management systems, because the two groups were similar in age, sex, indication for anticoagulation, and duration of follow-up. Because patients participating in population-based intervention studies cannot easily be randomized or blinded to treatment condition, and because investigators cannot be blinded to treatment condition, even randomized studies can be biased.

In conclusion, the present study, in which rules of warfarin management are proposed, provides evidence that initiation regimen and long-term rules specifically devoted to elderly patients and included in a computerized dosage program, allow hospitalized elderly patients to benefit from a quality of care as high as that of younger ambulatory patients.


  1. Top of page
  2. Abstract

We thank Vincent Vieillefond and Claire Visseaux for collecting the study data and all the physicians of the Charles Foix Hospital for participating in the study.

Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper.

Author Contributions: All authors contributed significantly to various aspects of this manuscript.

Sponsor's Role: None.


  1. Top of page
  2. Abstract
  • 1
    Nichol MB, Knight TK, Dow T et al. Quality of anticoagulation monitoring in nonvalvular atrial fibrillation patients: Comparison of anticoagulation clinic versus usual care. Ann Pharmacother 2008;42:6270.
  • 2
    Ansell J, Hirsh J, Hylek E et al. Pharmacology and management of the vitamin K antagonists: American College of Chest Physicians evidence-based clinical practice guidelines (8th Edition). Chest 2008;133 (6 Suppl):160S198S.
  • 3
    Ageno W, Johnson J, Nowacki B et al. A computer generated induction system for hospitalized patients starting on oral anticoagulant therapy. Thromb Haemost 2000;83:849852.
  • 4
    Chiquette E, Amato MG, Bussey HI. Comparison of an anticoagulation clinic with usual medical care: Anticoagulation control, patient outcomes, and health care costs. Arch Intern Med 1998;158:16411647.
  • 5
    Poller L, Shiach CR, MacCallum PK et al. Multicentre randomised study of computerised anticoagulant dosage. European concerted action on anticoagulation. Lancet 1998;352:15051509.
  • 6
    Garcia D, Regan S, Crowther M et al. Warfarin maintenance dosing patterns in clinical practice: Implications for safer anticoagulation in the elderly population. Chest 2005;127:20492056.
  • 7
    Merli GJ, Tzanis G. Warfarin: What are the clinical implications of an out-of-range-therapeutic international normalized ratio? J Thromb Thrombolysis 2009;27:293299.
  • 8
    Siguret V, Gouin I, Debray M et al. Initiation of warfarin therapy in elderly medical inpatients: A safe accurate regimen. Am J Med 2005;118:137142.
  • 9
    Bal dit Sollier C, Cambus C, Mahé I et al. A multi-rule expert system to manage OAT in most real-life clinical situations. J Thromb Haemost 2005;3 (Supp 1):P1623.
  • 10
    Ansell J. Tthe seventh ACCP conference on antithrombotic and thrombolytic therapy. Chest 2004;126 (3 Suppl):204S233S.
  • 11
    Schulman S, Kearon C. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients. J Thromb Haemost 2005;3:692694.
  • 12
    Rosendaal FR, Cannegieter SC, van der Meer FJ et al. A method to determine the optimal intensity of oral anticoagulant therapy. Thromb Haemost 1993;69:236239.
  • 13
    van Walraven C, Jennings A, Oake N et al. Effect of study setting on anticoagulation control: A systematic review and metaregression. Chest 2006;129:11551166.
  • 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;127:15151522.
  • 15
    van Walraven C, Austin PC, Oake N et al. The effect of hospitalization on oral anticoagulation control: A population-based study. Thromb Res 2007;119:705714.
  • 16
    Ansell J, Hollowell J, Pengo V et al. Descriptive analysis of the process and quality of oral anticoagulation management in real-life practice in patients with chronic non-valvular atrial fibrillation: The International Study of Anticoagulation Management (ISAM). J Thromb Thrombolysis 2007;23:8391.
  • 17
    Claes N, Buntinx F, Vijgen J et al. The Belgian Improvement Study on Oral Anticoagulation Therapy: A randomized clinical trial. Eur Heart J 2005;26:21592165.
  • 18
    Vogel T, Coriol V, Kaltenbach G et al. Prospective study of oral anticoagulation control in 110 very elderly hospitalized patients and of risk factors for poor control. Presse Med 2008;37:17231730.
  • 19
    Kaatz S. Determinants and measures of quality in oral anticoagulation therapy. J Thromb Thrombolysis 2008;25:6166.
  • 20
    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;128:513519.