Susan R. Kahn, Center for Clinical Epidemiology and Community Studies, Sir Mortimer B. Davis Jewish General Hospital, 3755 Cote Ste. Catherine Rm. A-127, Montreal, Quebec H3T 1E2, Canada. Tel.: +1 514 340 8222 #4667; fax +1 514 340 7564. E-mail: email@example.com
Objectives: In this multinational multicenter study, we evaluated whether subtherapeutic warfarin anticoagulation was associated with the development of PTS.
Methods: Patients with a first unprovoked deep venous thrombosis (DVT) received standard anticoagulation for 5–7 months and were then assessed for PTS. The time in the therapeutic range was calculated from the international normalized ratio (INR) data. An INR below 2, more than 20% of the time, was considered as subtherapeutic anticoagulation.
Results: Of the 349 patients enrolled, 97 (28%) developed PTS. The overall frequency of PTS in patients with subtherapeutic anticoagulation was 33.5%, compared with 21.6% in those with an INR below two for ≤ 20% of the time (P = 0.01). During the first 3 months of therapy, the odds ratio (OR) for developing PTS if a patient had subtherapeutic anticoagulation was 1.78 (95% confidence interval [CI] 1.10–2.87). After adjusting for confounding variables, the OR was 1.84 (95% CI 1.13–3.01). Corresponding ORs for the full period of anticoagulation were 1.83 (95% CI 1.14–3.00) [crude] and 1.88 (95% CI 1.15–3.07) [adjusted].
Conclusion: Subtherapeutic warfarin anticoagulation after a first unprovoked DVT was significantly associated with the development of PTS.
Post-thrombotic syndrome (PTS) is the most frequent complication of a deep vein thrombosis (DVT). In spite of varied estimates, it occurs in one-third to one half of patients with a DVT . PTS is associated with important morbidity. In a study assessing quality of life in patients with a DVT, the development of PTS was the strongest predictor of quality of life within the first 2 years of the diagnosis of a DVT . PTS also markedly increases health care costs [3,4].
Risk factors for PTS include an iliofemoral DVT, an ipsilateral recurrence of DVT, increased body mass index (BMI) and advanced age [1,5–7]. However, predictors of PTS are still poorly understood. The pathophysiology of PTS involves a combination of venous obstruction and valvular damage . Two previous single center studies reported a relationship between subtherapeutic vitamin-K antagonist (VKA) therapy and the development of PTS [9,10]. The mechanism for the potential association between subtherapeutic anticoagulation and PTS could be that subtherapeutic anticoagulation might be responsible for inadequate clot dissolution and thus might lead to persistent venous obstruction and valvular damage  or that subtherapeutic anticoagulation may lead to ongoing, subclinical formation of new thrombi resulting in chronic inflammation and valvular damage. The implication of this finding, if confirmed, is that optimizing the time in the therapeutic range (TTR) while on VKA therapy could reduce the incidence of PTS.
Mean TTRs reported in clinical studies involving VKA therapy are in the range of 60% [11–13]. The remainder of the time spent outside the therapeutic range is more often subtherapeutic than supratherapeutic . The occurrence of subtherapeutic anticoagulant therapy is therefore common. Indeed, the documented average time spent with a subtherapeutic international normalized ratio (INR) (below 2.0) in well-executed randomized clinical trials is approximately 20% [13,14].
Considering that there are no effective strategies to treat PTS, prevention of DVT is ideal. However, once a DVT has occurred, optimizing the intensity of oral anticoagulation may prove to be an additional strategy in reducing PTS incidence. We therefore sought to analyze data from a prospective multinational multicenter cohort study of patients with a first episode of an unprovoked DVT treated with the VKA warfarin, to assess whether subtherapeutic anticoagulation during the 6 months after a DVT diagnosis is associated with the development of PTS.
Material and methods
The REVERSE study (Recurrent Venous thromboembolism Risk Stratification Evaluation) was a prospective cohort study of patients with an unprovoked venous thromboembolism (VTE) that was designed to develop a clinical prediction rule to identify patients with a low risk of developing a recurrent VTE after stopping anticoagulation .
Patients were included in the REVERSE study if they had an objectively proven, first episode of an unprovoked proximal DVT or a PE (index VTE) 5–7 months previously. They would then have received heparin or low-molecular-weight heparin (LMWH) overlapped with warfarin for an initial period of at least 5 days, followed by continuation of warfarin for 5–7 months with a target INR of 2.0–3.0. In addition, patients were only included if they did not develop a recurrent VTE during the course of their anticoagulant therapy.
A proximal DVT was defined as a DVT in a lower limb vein proximal to and including the popliteal vein. A DVT was objectively confirmed by the lack of compressibility of a segment of the vein on compression ultrasonography. A PE was defined as a filling defect within a segmental or more proximal pulmonary artery demonstrated on a spiral computed tomography (CT) scan or a high probability ventilation-perfusion scan. An unprovoked VTE was defined as a DVT or PE that occurred in the absence of a lower limb fracture or cast immobilization, surgery requiring a general anesthetic within the previous 3 months, immobilization for over 3 days and without the diagnosis of cancer within the previous 5 years.
Patients were excluded if they were 17 years old or younger, were not willing to participate in the study, had discontinued their anticoagulants, required anticoagulant therapy for reasons other than VTE, had a previous unprovoked DVT, or had known major thrombophilia (protein C deficiency, protein S deficiency, antithrombin deficiency, persistently positive lupus anticoagulant, persistently elevated anticardiolipin antibody titers [> 30 U mL−1]), or had homozygous or double heterozygous thrombophilic mutations (factor V Leiden and the prothrombin gene G20210A mutation). Thrombophilia testing was performed prior to enrollment but not mandatorily.
Ethics approval was obtained from the institutional research ethics board at all participating centers (Ottawa Hospital Research Ethics Board for the lead institution). All patients provided written consent before study participation.
Baseline characteristics and collection of INR data
Data were collected at the baseline study visit on demographic and clinical characteristics, including features of PTS (Fig. 1). The use of elastic compression stockings since the time of VTE diagnosis was also recorded. Anticoagulation was stopped once the patient was enrolled into the study. All available results of INR testing since the index VTE were obtained for each patient for the antecedent period of anticoagulation (Fig. 1).
Assessment for PTS
Patients were evaluated for PTS at the baseline visit using the validated Villalta scale [16,17]. The Villalta scale assesses the presence and severity of PTS using five patient-reported symptoms and six clinician-observed clinical signs. PTS was defined by a score of > 4. Mild PTS was defined by a score of 5–9, moderate PTS by a score of 10–14 and severe PTS was defined by a score of ≥ 15 or the presence of an ipsilateral leg ulcer [16,18].
Data were summarized as means and standard deviations (SDs) for continuous variables, and proportions for categorical variables. Using INR data from the period of warfarin anticoagulation for the index VTE, TTR was calculated using the Rosendaal method . This is a method of determining the duration of therapeutic anticoagulation by interpretation of a series of INR values measured from each patient using the principle of linear interpolation in order to generate the TTR value. The assumption of this method is that INR values vary linearly between two recordings. TTR data were analyzed to evaluate whether there was an association between sub-therapeutic INR values during various time windows since the index DVT and the development of PTS. Based on published trials of warfarin anticoagulation for VTE, an INR < 2 for more than 20% of the time was considered to represent subtherapeutic anticoagulation [13,14]. Univariate analysis was performed to determine the odds ratio (OR) for the development of PTS if anticoagulation was subtherapeutic during the first 3 months, and during the full period of anticoagulation. Multivariate analysis adjusted for known and potential confounding variables was performed to assess the independent contribution of subtherapeutic anticoagulation to the development of PTS. A P-value < 0.05 was considered statistically significant. All analyzes were done with SAS 9.1 Statistical software (SAS Institute, Cary, NC, USA).
Between October 2001 and March 2006, 646 patients were enrolled into the REVERSE study, of whom 410 had a DVT (with or without PE) as their index event. Of these, 61 were excluded because of insufficient INR or PTS data. Hence, the study population of this PTS sub-study comprised 349 patients (Fig. 2).
Table 1 shows the baseline characteristics of the study population. The mean age (SD) was 54.2 years and 55.6% were males. The mean (SD) period of anticoagulation was 199 days (16.96) with a range of 160–280 days.
Table 1. Baseline characteristics
Study population n = 349
PTS n = 97 (28%)
No PTS n = 252 (72%)
DVT, deep vein thrombosis; VTE, venous thromboembolism; PTS, post-thrombotic syndrome.
*P-value pertains to PTS vs. no PTS groups.
≥ 65 years; n (%)
Female; n (%)
Caucasian; n (%)
Body mass index
kg m−2; mean (SD)
Previous secondary VTE
Family history of DVT
DVT with PE
Any use of graduated compression stockings
Total duration of anticoagulation
Mean (SD) (days)
Time in therapeutic range, entire period of anticoagulation
Mean % (SD)
Severity of PTS
n (%) Mild
Ninety-seven patients (27.8%) developed PTS; of these, 77 (79.4%) had mild PTS, 16 (16.5%) had moderate PTS and 4 (4.1%) had severe PTS. There were no significant differences in the general baseline characteristics in the group with PTS vs. the group without PTS.
The overall mean (SD) percent of time spent in the therapeutic range over the study period was 64.3% (19.4%). Among patients who developed PTS, the overall mean (SD) TTR during oral anticoagulation was 62.3% (20.0%), vs. 65.1% (19.1%) in patients without PTS (P = 0.23). However, the mean (SD) percentage time spent with an INR below two differed in patients with vs. without PTS. During the first 3 months of anticoagulation, patients who developed PTS had an INR of < 2 for a mean (SD) of 30% (24%) of the time, compared with 24% (22%) of the time in patients without PTS (P = 0.023). During the entire period of anticoagulation, patients who developed PTS had an INR of < 2 for 27% (19%) of the time compared with 23% (18%) of the time in patients without PTS (P = 0.08) (Table 2).
Table 2. Relationship between per cent of time spent at various INR levels during the first 3 months and the full period of anticoagulation, and development of PTS
PTS % time mean (SD)
No PTS % Time mean (SD)
INR, International Normalized Ratio.
*Subgroup of the group ‘% time INR < 2.0 for first 3 months’; †subgroup of the group ‘% time INR < 2.0 for full period of anticoagulation (5–7 months)’.
First 3 months
% time INR 2.0–3.0
% time INR < 2.0
% time INR 1.5–2.0*
% Time INR < 1.5*
Full period of anticoagulation (5–7 months)
% time INR 2.0–3.0
% time INR < 2.0
% time INR 1.5–2.0†
% Time INR < 1.5†
The overall frequency of PTS in patients with subtherapeutic anticoagulation (as per Methods, defined as INR < 2 for more than 20% of the time) was 33.5%, compared with 21.6% in those with an INR below two for ≤ 20% of the time (P = 0.01). During the first 3 months of anticoagulation, 62.9% (n = 61) of the patients with PTS received subtherapeutic anticoagulation compared with 48.8% (n = 123) of patients without PTS (P = 0.02). The same was also true during the entire period of anticoagulation, where 62.9% (n = 61) of those with PTS compared with 48.0% (n = 121) of those without PTS received subtherapeutic anticoagulation (P = 0.01). Interestingly, subtherapeutic anticoagulation within the first month of VKA treatment was not associated with PTS (INR was < 2 for more than 20% of the time in 32.0% [n = 31] of the patients with PTS vs. 29.4% [n = 74] of the patients without PTS; P = 0.64).
In univariate analysis, the OR for development of PTS in those with subtherapeutic anticoagulation during the first 3 months of anticoagulation was 1.78 (95% CI 1.10–2.87). In multivariate analysis adjusting for age, gender, BMI, concurrent PE and previous secondary VTE, the association between subtherapeutic anticoagulation and PTS remained robust (OR 1.84; 95% CI 1.13–3.01). ORs for the development of PTS in those with subtherapeutic anticoagulation during the entire period of anticoagulation were 1.83 (95% CI 1.14–3.00) [crude] and 1.88 (95% CI 1.15–3.07) [adjusted].
Among patients with PTS, the mean (SD) Villalta score in those exposed to subtherapeutic anticoagulation was 7.3 (2.57) whereas the mean (SD) Villalta score in those who were not exposed to subtherapeutic anticoagulation was 7.7 (2.78) (P = 0.56).
While subtherapeutic anticoagulation is known to be associated with recurrent VTE , few studies have looked at the association between subtherapeutic anticoagulation and other DVT-related outcomes such as PTS. In this study, we found that subtherapeutic anticoagulation was an independent predictor of a two-fold increased risk of PTS.
This study had a number of strengths. It was a relatively large study and to our knowledge it was the first multicenter study to address this issue. The study used the validated Villalta scale to measure PTS, as is recommended by the International Society of Thrombosis and Haemostasis subcommittee on Control of Anticoagulation . This study also used a standardized method of determining TTRs using the Rosendaal method. Furthermore, the average TTR among participants in this study was similar to that noted in published randomized controlled trials involving warfarin anticoagulation [11,12]. The study was not prone to interpretation bias based on the individual patient’s quality of anticoagulation as TTRs were determined after patients had been assessed for PTS. This study also examined the association between subtherapeutic anticoagulation and the severity of PTS, not just the presence or absence of PTS. However, we were not able to identify an association between subtherapeutic anticoagulation and PTS severity as reflected by Villalta score values. A possible explanation for this could be that among patients with PTS, the distribution of scores is heterogeneous such that a distinct gradient between the Villalta score and subtherapeutic anticoagulation could not be detected in our patient sample.
There were some limitations of this study which require consideration. The REVERSE Study from which our patients were drawn excluded patients with previously known severe thrombophilia and those who had previously had an unprovoked DVT (or PE), hence the results may not be generalizable to these groups. Our results are obtained in a population with an unprovoked DVT. However, two other previous studies, which are discussed below, have reported similar results derived from populations including both provoked and unprovoked DVT [9,10]. Patients who developed recurrent DVT during anticoagulant treatment after the index DVT were not eligible to be enrolled in the REVERSE study and therefore this sub-study. This group of patients may have had even more pronounced subtherapeutic anticoagulation than our study population. Additionally, ipsilateral recurrence is in itself a risk factor for PTS and if included it could have been a potential confounder in this study, or along the causal path between poor INR control and PTS. No information on the extent of DVT, a known predictor of PTS, was collected for the REVERSE study, so we were unable to include this variable in our multivariate analyzes. Forty-five patients with a DVT were enrolled into the REVERSE study before this PTS study was conceived, hence could not be included in our analysis as no baseline PTS data were collected in these patients. It is not possible to predict the impact that data from these patients may have had on the study. We cannot be sure that all patients enrolled into this study were free of PTS at the time of the index DVT as 18 patients (5.2%) had a previous provoked DVT at enrollment. However, adjusting for a previous provoked DVT in our multivariate analysis did not affect the overall result. The use of graduated compression stockings, which have been reported to reduce the risk of PTS , was not standardized in our study. However, any use vs. non-use of compression stockings after a DVT was diagnosed did not alter the risk of PTS in our study subjects (data not shown), and in multivariate analyzes adjustment was made for any use of graduated compression stockings. Although there did not appear to be a correlation between subtherapeutic anticoagulation during the first month of therapy and the development of PTS, this period includes INR data while patients were still on therapeutic dose LMWH (for the first 5–15 days). Therefore, patients may not have been exposed to suboptimal anticoagulation during this time in spite of having subtherapeutic INRs.
Few studies have previously investigated whether there is a relationship between subtherapeutic anticoagulation and the development of PTS. One such study was performed by van Dongen et al. . This was a retrospective study of 244 patients, based at a single center. The mean duration of anticoagulation was about 3 months, about half that of our study. Their definition of PTS differed slightly from ours in that a Villalta score of > 4 was required on two consecutive assessments, whereas our patients were all assessed for PTS at a single assessment; however, recent consensus guidelines from the International Society of Thrombosis and Haemostasis state that the diagnosis of PTS can be made at a single assessment . Overall, the incidence of PTS observed in the van Dongen study was 33%, in a similar range to that of our study (28%) as well as other studies [1,7,22]. The average TTR of 60% on oral anticoagulant therapy in the van Dongen study is also comparable to that of our study (64%) and other previously published studies [11–13]. The van Dongen study, however, showed a proportionately greater amount of time that patients were exposed to subtherapeutic anticoagulation than in to our study. Over the complete duration of treatment, patients in the van Dongen study spent a mean of 30% of the time with an INR of < 2 compared with 23% in our study. The adjusted OR for the development of PTS if the INR was < 2 for > 30% of the time over 3 months in the van Dongen study was 1.89, similar to our adjusted OR of 1.84 if the INR was < 2 for > 20% of the time at 3 months. Therefore, in spite of some modest differences, the findings of our study essentially are consistent with those of the van Dongen study.
A study by Ziegler et al.  also reported a link between subtherapeutic anticoagulant therapy and the development of PTS. This was a retrospective single center study of 161 patients followed for 6.6 years. Unlike our study, their definition of PTS was not based on the Villalta scale and analysis of TTR did not employ the Rosendaal method of linear interpolation. These factors make it difficult to perform a direct comparison with our study.
The main clinical implication of our findings is that optimizing warfarin anticoagulant therapy by limiting periods of subtherapeutic anticoagulation within the first 5–7 months of therapy post DVT may be helpful in reducing the incidence of PTS. Consideration should also be given to whether alternative anticoagulants, either oral or parenteral, that offer more predictable anticoagulation than warfarin could have an impact on the incidence of PTS. One study showed a significantly lower incidence of PTS-like symptoms associated with the long-term use of therapeutic LMWH compared with standard oral anticoagulant therapy to treat an acute DVT . In spite of this being a multicenter prospective randomized controlled trial, this study described patient-reported PTS-like symptoms obtained by questionnaire without objective physical assessment by a physician or trained nurse. It is therefore difficult to conclude that this reflects more effective therapeutic anticoagulation resulting in a lower incidence of PTS. Further trials comparing the risk of PTS on warfarin therapy compared with alternative anticoagulant therapy may be of value.
In conclusion, we found that subtherapeutic anticoagulation as defined by an INR of < 2 for > 20% of the time during the first 5–7 months of VKA treatment for a first unprovoked DVT was significantly associated with the development of PTS.
M.A. Rodger, M.J. Kovacs, P.S. Wells, D.R. Anderson and S.R. Kahn designed the original REVERSE study, performed the research and collected the data. R.S. Chitsike and S.R. Kahn designed this REVERSE sub-analysis, analyzed the data and drafted the manuscript. G. Le Gal, A. Perrier, R.H. White, I. Chagnon, S. Solymoss, M.A. Crowther and L.M. Vickars performed the research and collected the data. M.T. Betancourt and T. Ramsay analyzed the data. All authors revised the manuscript for important intellectual content and approved the final version of the manuscript.
This study was funded by the Canadian Institutes of Health Research (grant no. MOP 64319) and bioMérieux (through an unrestricted research grant). Calculation of the time in the therapeutic range by linear interpolation was performed by Marc Carpentier. Dr Kahn is supported by a National Investigator (chercheur national) award of the Fonds de la recherche en santé du Québec (FRSQ). Dr Marc Rodger received career investigators support from the Heart and Stroke Foundation of Canada and the University of Ottawa Faculty of Medicine Clinical Research Chair in Venous Thrombosis and Thrombophilia.
Disclosure of Conflicts of Interest
The authors state that they have no conflict of interest.