The quality of oral anticoagulant therapy and recurrent venous thrombotic events in the Leiden Thrombophilia Study

Authors

  • A. P. A. GADISSEUR,

    1. Department of Haematology/Haemostasis Unit Antwerp University Hospital (UZA), Edegem, Belgium
    2. Department of Haematology/Haemostasis and Thrombosis Research Center, Leiden
    3. Leiden Anticoagulation Clinic, Leiden
    Search for more papers by this author
  • S. C. CHRISTIANSEN,

    1. Department of Clinical Epidemiology, Leiden University Medical Center (LUMC), Leiden, the Netherlands
    Search for more papers by this author
  • F. J. M. VAN DER MEER,

    1. Department of Haematology/Haemostasis and Thrombosis Research Center, Leiden
    2. Leiden Anticoagulation Clinic, Leiden
    Search for more papers by this author
  • F. R. ROSENDAAL

    1. Department of Haematology/Haemostasis and Thrombosis Research Center, Leiden
    2. Leiden Anticoagulation Clinic, Leiden
    3. Department of Clinical Epidemiology, Leiden University Medical Center (LUMC), Leiden, the Netherlands
    Search for more papers by this author

Alain P. A. Gadisseur, Antwerp University Hospital (UZA), Department of Haematology, Wilrijkstraat 10 2650 Edegem, Belgium.
Tel.: +32 3 821 3198; fax: +32 3 821 4301; e-mail: alain.gadisseur@uza.be

Abstract

Summary. Background: The International Normalized Ratio (INR) target range is a relatively narrow range in which the efficacy of oral anticoagulant treatment, i.e. prevention of extension and recurrence of thrombosis, is balanced with the risk of hemorrhagic complications. Over the years, different INR target ranges have been implemented for individual indications, depending on their thrombotic potential. In most of the studies defining these INR targets, the treatment of the patients was aimed at a certain INR range, but in the analysis no account was taken of the time that the patients spent within this range in reality. Methods: The Leiden Thrombophilia Study (LETS) is a population-based case-control study on risk factors for venous thrombosis, in which many genetic and acquired factors have been investigated. Our aim was to investigate the effect of the quality of the oral anticoagulant therapy for the initial venous thrombosis and its relationship with recurrence of thrombosis. Quality of anticoagulation was defined as the time spent at various INR levels during treatment, and we focused on the effect of sustained intensities above a certain INR in preventing recurrences later on. Results: Two hundred and sixty-six patients with a total follow-up of 2495 patient-years were studied. The mean duration of the initial anticoagulant therapy was 194.5 days (range 48–4671). During follow-up, 58 recurrences were diagnosed (cumulative recurrence rate of 21.8% over 9 years). The mean INR during initial therapy was 2.90, with 90.3% [95% confidence interval (CI) 88.4–92.3%] of the time being spent above an INR of 2.0, and 39.1% (95% CI 35.5–42.7%) above an INR of 3.0. Patients who spent more time below the target range, or who had a shorter duration of anticoagulation, did not experience a higher risk of recurrence after the initial period of anticoagulation had passed. Conclusion: Provided that oral anticoagulant treatment is adequately managed, according to international guidelines, recurrent thrombosis cannot be ascribed to variation in the primary treatment. Further attempts to reduce the risk of recurrence should therefore be aimed at identifying other explanatory factors, and subsequently fine-tuning the target ranges.

Introduction

For decades, the treatment of venous thrombosis (VT) has been based upon heparins (intravenous or subcutaneous) in the acute phase and oral coumarin vitamin K antagonists (warfarin, phenprocoumon or acenocoumarol) for long-term treatment. This treatment has proven effective, and over the years much work has been done on further improving it.

Vitamin K antagonists inhibit the production by the liver of the vitamin K-dependent coagulation factors (factor II, FVII, FIX and FX), and are absorbed in the gut. They are susceptible to many influences, such as diet, comedication and illness, and need frequent monitoring. Their action is measured through the prothrombin time (PT), expressed as the International Normalized Ratio (INR). The INR target range is a relatively narrow range in which efficacy of treatment, i.e. prevention of extension and recurrence, is balanced with the risk of hemorrhagic complications. Over the years, different INR target ranges have been implemented for individual indications, depending on their thrombotic potential [1–5]. In most of the studies defining the INR targets, a fixed target was aimed for, and in the analysis, no account was taken of the time that the patients spent within this range in reality.

The interpretation of the quality of oral anticoagulant therapy (OAT) has received more attention in studies concentrating on the bleeding complications arising from this treatment [6–8], in studies comparing the different coumarins used for oral anticoagulation [9–11], and in studies dealing with the development of patient self-management in this area [12–14], than in studies of efficacy. Several methods have been proposed in the past to determine the quality of coumarin therapy, e.g. fraction of INRs in range, cross-section of the files, and linear interpolation [15,16].

The Leiden Thrombophilia Study (LETS) [17,18] is a population-based case-control study on risk factors for VT, in which many genetic and acquired factors have been investigated [19–25]. Although the patient inclusion was finished in 1993, laboratory analyses have been performed up to the present, in addition to which the records have been kept up to date with regard to follow-up of the patients [26].

Our aim was to investigate the effect of the quality of the OAT for the initial VT and its relationship with recurrence of thrombosis in the LETS patients.

Patients and methods

Study population

The design of the population-based case-control study (LETS) has been described previously [17,18]. Briefly, consecutive patients with a first episode of an objectively diagnosed deep vein thrombosis (DVT) were selected from the files of three anticoagulation clinics in The Netherlands (Leiden, Amsterdam, and Rotterdam) in the period between 1 January 1988 and 31 December 1992. These regional anticoagulation clinics are responsible for the management of OAT of virtually all patients within a well-defined geographic area. All patients were younger than 70 years without evidence of an underlying malignancy. The original LETS study included 474 patients and 474 sex-matched and age-matched controls.

For the purpose of the quality of anticoagulation analysis, only those patients were included who originated at the Leiden anticoagulation clinic (n = 272) and for whom follow-up was complete up to 1 January 2002 (n = 266, 98%). All patients enrolled in the LETS study gave informed consent for long-term follow-up. All patients were followed from the date of the initial thrombosis until 1 January 2002, unless they were lost to follow-up through emigration, had a recurrent thrombosis, or died. For the purpose of the analysis of the initial anticoagulation treatment period in regard to the development of recurrent VT, the follow-up period ended at the time of recurrence (n = 58), death (n = 5), or emigration (n = 10). The diagnosis of recurrent VT had to be confirmed by compression ultrasound echography, venography or impedance plethysmography for DVT, or ventilation/perfusion lung scanning or spiral computerized tomography for pulmonary embolism. All patients were regularly contacted by telephone or mail to ascertain their condition and to update the follow-up. In cases of hospital admissions or visits to physicians possibly related to a recurrence of thrombosis, the treating physicians were contacted for detailed information.

The data concerning INR values and coumarin dosages for the initial period of anticoagulant therapy were retrieved from the computerized files of the Leiden anticoagulation clinic. The Leiden clinic serves an area of around 500 000 inhabitants, and provides anticoagulant monitoring to 10 000 patients per year, with around 5 000 patients being actively on anticoagulant treatment at any given time. Approximately 75% of the patients are treated with the long-acting phenprocoumon (Marcoumar®; Hoffmann-La Roche Ltd, Basel, Switzerland) and 25% with the short-acting acenocoumarol (Sintrom Mitis®; Novartis, Basel, Switzerland). The choice of the anticoagulant is based on personal preferences of physicians, not on clinical features of the patient. Laboratory checks and subsequent adjustments of the dosing schedules occur at intervals of 1–6 weeks.

Determination of OAT dosage schedule

Determination of the anticoagulant dosage is done by physicians in the Leiden Anticoagulation Clinic with the aid of a computerized dosing program (TRODIS, Infotrom, Leiden, the Netherlands). This program evaluates the stability of the PT/INR values and proposes dosing schedules in about 50% of patients that are then checked by the physicians. Approximately 15–20% of these computer-generated proposals are changed by the physicians. In the other 50% of patients, no dosage proposal can be generated, and dosing is done completely by the physicians. Details of the dosing algorithm have been published previously [27].

Therapeutic INR target ranges are defined for all patients on OAT, based on their indication for the treatment. For the indication DVT, the target range was 2.5–3.5 INR.

Treatment quality

In this analysis, the endpoint was recurrence, and we did not look at hemorrhage. Quality of anticoagulation was defined as the time spent at various INR levels during treatment, and we focused on the effect of sustained intensities above a certain INR in preventing recurrences later on. The quality of the OAT was calculated as time above or below a certain INR or as time in range: the estimated time spent by the patient above/below a target INR, or within a certain range, based on the method of linear interpolation. This method of approximation of the time in range has been published previously [8,16].

This analysis was done for the anticoagulation treatment period for the initial DVT, both for the treatment period as a whole, and on a monthly basis.

Study endpoints

The follow-up time started at the end of the initial period of anticoagulation, and lasted until the end of follow-up (1 January 2002), emigration, death or recurrent thrombosis, whichever occurred first. The endpoints were:

  • 1Occurrence of recurrent VT after discontinuation of oral anticoagulant therapy;
  • 2Time to recurrence (disease-free survival) after discontinuation of OAT.

Statistical considerations

The analysis was performed as a cohort with events over patient-time, in which the patients were also grouped into subcohorts based on the INR distribution of the initial treatment period. We used standard survival analysis techniques, such as Kaplan–Meier survival tables and Cox regression. The SPSS for Windows computer program version 11.0.1 (SPSS Inc., Chicago, Illinois, USA) was used for the statistical calculations.

Results

Total follow-up for 266 patients from the end of the initial anticoagulation until the end of follow-up in 2001, or an event, was 2495 patient-years, with a mean follow-up per patient of 3339 days [95% confidence interval (CI) 3128–3551], i.e. 9.1 years. Of the original episodes of DVT, 216/266 were considered as spontaneous (i.e. in the absence of clear provoking factors or circumstances), and 74/266 patients were diagnosed with hereditary thrombophilia [i.e. antithrombin III deficiency, protein C deficiency, protein S deficiency, FV Leiden mutation (R506Q), prothrombin 20210A mutation]. During the follow-up after the end of the anticoagulant therapy, 58 recurrences were diagnosed. Detailed patient characteristics are given in Table 1.

Table 1.   Patient characteristics
 MeanRange
  1. VTE, venous thromboembolism.

Number of patients 266 
Age at first VTE (years) 40.215–69
Sex: male/female 117/149 
Follow-up (days)3906.2116–5114
Recurrences/no recurrences  58/208 
Oral anticoagulant
 Phenprocoumon 226 
 Acenocoumarol  32 
 Phenprocoumon and acenocoumarol   8 
Duration of anticoagulation after first VTE (days) 194.548–4671 (10th–90th percentile = 89–261)

There were 58 recurrences, representing a cumulative incidence of recurrence of 21.8% over 9 years of follow-up. Only 7/58 (12.1%) recurrences occurred within the first year of the end of the anticoagulant therapy, which implies a recurrence incidence rate of only 2.6% for the first year. The yearly incidence rate for the first 5 years after the end of therapy was, at 2.9%/year (95% CI 2.3–3.4), higher than for second 5 years after therapy (1.6%; 95% CI 0.8–2.4). Median time to recurrence was 1417 days (i.e. almost 4 years). Details are given in Table 2.

Table 2.   Recurrent venous thromboembolism (VTE)
 Mean95% CI
Number of recurrent VTEs58/266 (21.8%) 
Recurrences in term of time after end of anticoagulant treatment
 Early recurrences <1 year after end of therapy 7 
 Medium-term recurrences
  1–3 years after end of therapy17 
  3–5 years after end of therapy12 
 Late recurrences
  5–10 years after end of therapy18 
  >10 years after end of therapy 3 
 Under anticoagulant therapy 1 
Median time to recurrence in days (range)1417 (41–4167)1287–1865
Yearly recurrence rate: overall1.0%/year1.4–2.6%
 <5 years after end of therapy2.9%/year2.3–3.4%
 >5 years after end of therapy1.6%/year0.8–2.4%

The mean duration of the OAT for the initial venous thromboembolism (VTE) was 194 days (95% CI 146.0–243.0), with a range of 48–4671 days (10th–90th percentile range =89–261 days). Five of the original 266 patients were put on life-long therapy after the first event, and as such were not included in the analysis. In total, 234 patient-years of OAT were analyzed for treatment quality. In Fig. 1, the cumulative recurrence-free survival is given over time, for various durations of the anticoagulation treatment.

Figure 1.

 Recurrence free survival based on mean INR.

Overall, the mean INR value was 2.90 (range 1.7–4.7), with 90.3% (95% CI 88.4–92.3) of the time spent above an INR of 2, and 39.1% of the time spent (95% CI95 35.5–42.7) above an INR of 3.0. The percentage of time spent above the target range (2.5–3.5) was 17.7% (95% CI 15.1–20.3). Conversely, only 9.7% (95% CI 7.7–11.6) of the time was spent below an INR of 2.0, and only 6.5% (95% CI 4.0–5.5) below an INR of 1.8. In Fig. 2, the cumulative recurrence-free survival is given, based on the mean INR during the preceding period of anticoagulation. In Table 3, the data concerning the quality of the OAT in patients with and without recurrent VTE are summarized. There is no evidence that those who experienced recurrences had been treated for shorter periods or less intensely during the initial period of anticoagulation. In fact, it even appeared that the mean duration of the OAT was somewhat longer and the intensity somewhat higher in the patients with recurrences than in those without. The mean treatment duration for patients without recurrences was 177.1 days (95% CI 134.9–220.6), whereas those with recurrences had a longer initial treatment duration of 250.3 days (95% CI 93.4–407.2). This difference between the two groups was much smaller if stratified for the absence of thrombophilia (159.2 days vs. 168.4 days, P = 0.648). Patients with inherited thrombophilia were, on the whole, anticoagulated for a longer period than those without thrombophilia (292.2 vs. 161.1 days, P = 0.02), both in the group without recurrences (242.2 vs. 159.2 days, P = 0.648) and in the group with recurrent DVT (394.6 vs. 168.4 days, P = 0.447).

Figure 2.

 Cumulative recurrence free survival in function of anticoagulation treatment duration.

Table 3.   Oral anticoagulation therapy (OAT) stratified for the occurrence of recurrent venous thromboembolism (VTE)
 No recurrent VTE (n = 208)Recurrent VTE (n = 58)P
  1. NS, no statistically significant difference.

Mean duration of OAT for first VTE in days (excluding life-long)177.7 (95% CI 134.9–220.6)250.3 (95% CI 93.4–407.2)NS
Mean duration of OAT for first VTE in days (categorized)
 Less than 3 months19 (9.1%)7 (12.1%) 
 3–6 months120 (59.4%)25 (43.1%) 
 6–9 months38 (18.7%)21 (36.2%) 
 9–12 months7 (3.4%)3 (5.2%) 
 More than 1 year14 (6.9%)2 (3.4%) 
 Indefinite5 (2.5%) 
Linear interpolation
 % time >3.0 INR36.8% (95% CI 32.8–40.8)46.8% (95% CI 39.0–54.5)P < 0.05
 % time >2.5 INR68.4% (95% CI 64.9–71.9)74.3% (95% CI 67.8–80.7)NS
 % time >2.0 INR90.0% (95% CI 87.7–92.2)91.5% (95% CI 87.5–95.5)NS
 % time >1.8 INR94.4% (95% CI 92.7–96.1)94.8% (95% CI 91.2–98.4)NS

We performed Cox regression analysis to further analyze aspects of the initial treatment with anticoagulants affecting the risk of recurrence In none of the analyses did a longer time at lower INRs contribute to the occurrence of recurrences. The mean INR did not contribute to the risk for recurrence [relative risk (RR) = 1.02, 95% CI 0.60–1.75], and nor did the number of days below an INR of 1.8 (RR = 1.003, 95% CI 0.997–1.006). In a multivariate analysis model with the inclusion of the initial treatment of 6 months or more, the presence of thrombophilia, the absence of provoking circumstances or factors (i.e. spontaneous DVT), and whether the number of days above an INR of 3.0 was more than 60, we found no effect of a lower risk of recurrence with a higher number of days above an INR of 3.0 (RR = 1.70, 95% CI 0.95–3.05). Details are given in Table 4.

Table 4.   Multivariate analysis (Cox regression)
ModelVariablesRelative risk95% Confidence intervalP
  1. INR, International Normalized Ratio.

>60 days above 3.0 INRUnadjusted2.121.24–3.630.006
Adjusted for sex, age2.121.23–3.680.007
Adjusted for sex, age, treatment duration1.951.09–3.500.025
Adjusted for sex, age, treatment duration, thrombophilia, idiopathic thrombosis1.770.98–3.180.057
>6 months’ treatment durationUnadjusted1.421.06–1.910.020
Adjusted for sex, age1.401.04–1.890.025
Adjusted for sex, age, treatment duration1.480.87–2.510.145
Adjusted for sex, age, treatment duration, thrombophilia, idiopathic thrombosis1.550.91–2.640.105

When we looked at the intensity of OAT during the first month of treatment, no difference was seen for different intensity strata. Patients who had an INR below 2.0 for more than 50% of the time during the first month had a cumulative recurrence risk of 25.7% (9/35) in subsequent years, whereas those who had an INR above 2.0 for all of the first month of OAT showed a cumulative recurrence risk of 23.2% (32/138). The times to recurrence were 1094 days (95% CI 500–1688) and 1432 days (95% CI 1091–1773), respectively. Those who had an INR above 2.5 or 3.0 for the whole of the first month had cumulative recurrence risks of 28.2% (20/71) and 34.5% (10/29), respectively, and, again, similar times to recurrence. The results were similar when we looked at only the last month of treatment, or patients who were treated for only 3–6 months.

Discussion

Coumarins have now been in use for OAT for several decades, and much work has been done on refining the way in which this treatment is handled. Over the years, the target INR ranges have been lowered in an effort to balance clinical efficacy with the occurrence of hemorrhagic complications, and much effort has been made to define the optimum treatment duration. In many of the studies underlying these developments, the conclusions about the ideal intensity have mostly been drawn on the basis of the aimed-for INR target ranges rather than the ‘real’ INR behavior, or the mean INR.

In this present study, we have looked for associations between recurrent VTE and the intensity and duration of the OAT for the first VTE. To determine the quality of the intensity of the OAT, we have made use of the method of linear interpolation [8,16], which has become an accepted tool for such analyses. Surprisingly, there was no evident difference in the recurrence rate with the intensity of the initial treatment. If anything, in some subanalyses, the patients with recurrences showed longer periods at higher INR levels than those who did not have recurrences.

This was not a randomized study, but an observational cohort study in which the determinants of the anticoagulant treatment were left in the hands of the treating physician, with regard to the length and prescribed intensity, i.e. the desired INR range, and in which the actual handling of the INR range delivered was left to the anticoagulation clinic, based upon the information provided by the treating physician. Regression analysis showed that when other variants were studied, such as duration of treatment over 6 months, absence of provoking circumstances (i.e. spontaneous VT) and the presence of thrombophilia, these served as confounding factors in the analysis as to whether longer periods of time at higher INR levels were related to a higher risk of recurrence. Patients considered at a higher risk by the treating physician, probably on the basis of the absence of provoking circumstances or the presence of thrombophilia, were, on the whole, treated for longer periods and at slightly higher INR levels, the latter probably because of a tendency by dosing physicians at the anticoagulation clinic to keep patients who they considered to be ‘high risk’ more at the higher end of the desired INR range.

The overall quality of the OAT in the studied patient cohort was high with a mean treatment duration – excluding life-long OAT patients – of 194.5 days, a mean INR value of 2.90, and an INR above 2.0 for 90.3% of the time. This is well in line with the latest international guidelines [5,28]. This implies that our conclusion of no association between intensity of treatment and risk of recurrence cannot be extrapolated to low intensities, and should be viewed as an experience within a framework of high-quality anticoagulation. It is likely that this quality of OAT management obscured relationships between the intensity and duration of treatment and recurrent thrombosis, and that with treatment such as this, as much has been achieved as can be in the prevention of recurrences, short of opting for life-long anticoagulation. Some studies have indicated that prolonged periods below an INR of 1.5 are associated with a higher recurrence rate [29]. In our patient cohort, very little time was spent at these low levels. There is no evidence in this study that extending the duration of the OAT beyond 6 months or increasing the target INR above 2.5 will result in fewer recurrences after discontinuation of the treatment.

Provided that OAT is adequately managed, according to international guidelines, recurrent thrombosis cannot be ascribed to variation in the primary treatment. Further attempts to reduce the risk of recurrence should therefore be aimed at identifying other explanatory factors and subsequently fine-tuning the target ranges.

Keeping the INR above 2.0 for most of the advised treatment period will protect as much as possible against recurrences. If this goal can be met, further refinement of the management technique should be aimed at lowering the number of hemorrhagic complications through prevention of over-anticoagulation.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.

Ancillary