Duración del tratamiento con antagonistas de la vitamina K en el tromboembolismo venoso sintomático
Resumen
Antecedentes
Actualmente, el tratamiento secundario utilizado con más frecuencia en los pacientes con tromboembolismo venoso son los antagonistas de la vitamina K, dirigidos a obtener una razón internacional normalizada (RIN) de 2,5 (rango: 2,0 a 3,0). Sin embargo, debido al riesgo continuo de hemorragias y a la incertidumbre con respecto al riesgo de TEV recurrente, todavía existe discusión acerca de la duración apropiada del tratamiento con AVK en estos pacientes. Varios estudios han comparado los riesgos y los efectos beneficiosos de las diferentes duraciones del tratamiento con AVK en pacientes con TEV. Esta es la tercera actualización de una revisión publicada por primera vez en 2000.
Objetivos
Evaluar la eficacia y seguridad de diferentes duraciones del tratamiento con antagonistas de la vitamina K en pacientes con tromboembolismo venoso sintomático.
Métodos de búsqueda
Para esta actualización, el Coordinador de Búsqueda de Ensayos del Grupo Cochrane de Enfermedades Vasculares Periféricas (Cochrane Peripheral Vascular Diseases Group) realizó búsquedas en el Registro especializado (última búsqueda en octubre de 2013) y en el Registro Cochrane Central de Ensayos Controlados (Cochrane Central Register of Controlled Trials, CENTRAL) 2013, Número 9.
Criterios de selección
Ensayos clínicos controlados aleatorizados que compararan diferentes duraciones del tratamiento con antagonistas de la vitamina K en pacientes con tromboembolismo venoso sintomático.
Obtención y análisis de los datos
Tres autores de la revisión (SM, MP y BH) extrajeron los datos y evaluaron la calidad de los ensayos de forma independiente.
Resultados principales
Se incluyeron 11 estudios con un total de 3716 participantes. Se observó una fuerte y consistente reducción del riesgo de eventos tromboembólicos venosos recurrentes durante el tratamiento prolongado con AVK (riesgo relativo [RR] 0,20, intervalo de confianza [IC] del 95%: 0,11 a 0,38) independientemente del período transcurrido desde el evento trombótico índice. No se observó un fenómeno 'de rebote', es decir, un exceso de recurrencias poco después de la suspensión del tratamiento prolongado (RR 1,28; IC del 95%: 0,97 a 1,70). Además, se observó un aumento significativo de las complicaciones hemorrágicas en los pacientes que recibieron un tratamiento prolongado durante todo el período posterior a la asignación al azar (RR 2,60; IC del 95%: 1,51 a 4,49). Durante todo el período de estudio no se observó una reducción de la mortalidad (RR 0,89; IC del 95%: 0,66 a 1,21; p = 0,46).
Conclusiones de los autores
En conclusión, esta revisión muestra que el tratamiento con AVK reduce en gran medida el riesgo de TEV recurrente durante el tiempo en que se utilizan. Sin embargo, el riesgo absoluto de tromboembolismo venoso recurrente declina con el tiempo, mientras el riesgo de hemorragias graves permanece inalterado. Por lo tanto, la eficacia de la administración de AVK disminuye con el transcurso del tiempo desde el evento índice.
PICO
Resumen en términos sencillos
Duración del tratamiento con antagonistas de la vitamina K y prevención de la recurrencia en pacientes con tromboembolismo venoso
El tromboembolismo venoso (TEV) se produce cuando se forma un coágulo de sangre en una vena profunda, o cuando se desprende y se aloja en los vasos pulmonares. Estos coágulos pueden ser mortales si el flujo de sangre al corazón está bloqueado. Los antagonistas de la vitamina K (AVK) se administran a los pacientes que han presentado un TEV, para prevenir la recurrencia. La principal complicación de este tratamiento es la hemorragia. El riesgo continuo de hemorragia con el uso de los fármacos y la incertidumbre sobre el alcance del riesgo de recurrencia hacen que sea importante considerar la duración adecuada del tratamiento con AVK en estos pacientes. Los autores de la revisión buscaron en la literatura y pudieron combinar los datos de 11 ensayos clínicos controlados aleatorizados (3716 participantes) que compararon diferentes duraciones del tratamiento con AVK en pacientes con un TEV sintomático. Los participantes que recibieron un tratamiento prolongado tuvieron un riesgo cinco veces menor de recurrencia del TEV. Por otro lado, tuvieron un riesgo tres veces mayor de complicaciones hemorrágicas. El tratamiento prolongado no redujo el riesgo de muerte. La administración prolongada de AVK redujo de forma consistente el riesgo de coágulos recurrentes durante su uso, pero el efecto beneficioso disminuyó con el transcurso del tiempo y el riesgo de hemorragias graves permaneció.
Authors' conclusions
Summary of findings
| Long‐term or short‐term treatment with vitamin K antagonists for patients with venous thromboembolism | ||||||
| Patient or population: patients with venous thromboembolism | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Short‐term treatment with VKA | Long‐term treatment with VKA | |||||
| Incidence of recurrent VTE | 88 per 1000 | 18 per 1000 | RR 0.2 | 3536 | ⊕⊕⊕⊕ | |
| Incidence of major bleeding | 4 per 1000 | 15 per 1000 | RR 3.44 | 1350 | ⊕⊕⊕⊝ | |
| Mortality | 38 per 1000 | 26 per 1000 | RR 0.69 | 1049 | ⊕⊕⊕⊝ | |
| *The basis for the assumed risk (eg, the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence. | ||||||
| 1Relatively wide 95% confidence interval around the estimate. | ||||||
Background
Description of the condition
Venous thromboembolism (VTE), the collective term for deep venous thrombosis (DVT) and pulmonary embolism (PE), is a disorder that is frequently encountered in medical practice, affecting two to three of every 1000 persons (general population) per year. Venous thromboembolism may occur after surgery, after trauma and immobilization, in cancer patients, during hormonal contraceptive use or pregnancy, and after delivery (provoked), but it also occurs in the absence of such clinical risk factors (unprovoked). Hereditary thrombophilic conditions such as antithrombin, protein C, and protein S deficiencies, as well as Factor V Leiden and prothrombin 20210A mutations, increase the risk for both provoked and unprovoked venous thromboembolic events.
Description of the intervention
The most important aim of treatment for patients with VTE is to prevent recurrence, including potentially fatal PE. Patients are usually treated with an initial course of heparin or low‐molecular‐weight heparin (for approximately six days) associated with vitamin K antagonists (VKA) started simultaneously and continued for a period thereafter. This prolonged use of VKA has proven efficacy in comparison with placebo and low‐dose heparin (Hull 1979; Lagerstedt 1985). The general consensus is that VKA should be targeted to prolongation of prothrombin time, compatible with an international normalized ratio (INR) of 2.0 to 3.0.
Why it is important to do this review
Based on the continuing risk of bleeding and uncertainty regarding the risk of recurrent VTE, discussion on the proper duration of oral anticoagulant treatment in patients with VTE is ongoing. Several studies have compared the risks and benefits of different durations of VKA treatment in patients with VTE.
Therefore we evaluated in this review the reduction in the incidence of recurrent VTE and the excess of major bleeding associated with different durations of VKA in patients with VTE.
Objectives
To evaluate the efficacy and safety of different durations of treatment with vitamin K antagonists in patients with symptomatic venous thromboembolism.
Methods
Criteria for considering studies for this review
Types of studies
Studies in which participants were randomly allocated to different durations of VKA. Studies were excluded if they were duplicate reports or preliminary reports of data later presented in full.
Types of participants
Studies were included if participants had symptomatic VTE. Studies were excluded if the trialists had not used accepted objective tests to confirm the diagnosis of DVT (eg, venography, ultrasonography) or PE (eg, high‐probability ventilation‐perfusion lung scan, pulmonary angiography).
Types of interventions
Studies were included if different durations of treatment with a VKA, such as warfarin and acenocoumarol, were compared. Studies were excluded if different target INR ranges were used in the treatment arms, or if continuous use of another anticoagulant or antiplatelet drug was reported.
Types of outcome measures
Primary outcomes
-
Incidence of recurrent venous thromboembolism (DVT or PE).
Secondary outcomes
-
Incidence of major bleeding.
-
Mortality.
Studies were excluded if no data for thromboembolic events and bleeding were available, if outcome assessment was performed by assessors who were aware of study allocation, or if no breakdown was provided for minor and major bleeding.
The following criteria were accepted for the diagnosis of recurrent symptomatic DVT: an extension of an intraluminal filling defect on a venogram; a new intraluminal filling defect or an extension of nonvisualization of proximal veins in the presence of a sudden cutoff defect on a venogram that was seen on at least two projections; if no previous venogram was available for comparison, an intraluminal filling defect; if no venogram was available, abnormal results of compression ultrasonography in an area where compression had been normal, or a substantial increase in the diameter of the thrombus during full compression at the popliteal or femoral vein (Koopman 1996; Levine 1996); or, if neither a venogram nor an ultrasonographic study was available, a change in the results of impedance plethysmography from normal to abnormal.
The following criteria were accepted for the diagnosis of (recurrent) PE: a new intraluminal filling defect, an extension of an existing defect, or the sudden cutoff of vessels larger than 2.5 mm in diameter on a pulmonary angiogram; if no prior angiogram was available, an intraluminal filling defect or sudden cutoff of vessels larger than 2.5 mm in diameter on a pulmonary angiogram; or if no pulmonary angiogram was available, a defect of at least 75% of a segment on the perfusion scan, with normal ventilation. If the ventilation‐perfusion scan was nondiagnostic (and no pulmonary angiogram was available), satisfaction on the criteria for DVT was acceptable, or PE could be demonstrated at autopsy.
Hemorrhages were classified as major if they were intracranial or retroperitoneal, led directly to death, necessitated transfusion, or led to interruption of antithrombotic treatment or (re)operation. All other hemorrhages were classified as nonmajor.
Search methods for identification of studies
Electronic searches
For this update the Cochrane Peripheral Vascular Diseases Group Trials Search Co‐ordinator (TSC) searched the Specialised Register (last searched October 2013) and the Cochrane Central Register of Controlled Trials (CENTRAL) 2013, Issue 9, part of The Cochrane Library, (www.thecochranelibrary.com). See Appendix 1 for details of the search strategy used to search CENTRAL. The Specialised Register is maintained by the TSC and is constructed from weekly electronic searches of MEDLINE, EMBASE, CINAHL, AMED, and through handsearching relevant journals. The full list of the databases, journals and conference proceedings which have been searched, as well as the search strategies used are described in the Specialised Register section of the Cochrane Peripheral Vascular Diseases Group module in The Cochrane Library (www.thecochranelibrary.com).
Searching other resources
Additional studies were sought by a manual search through reference lists of relevant studies and through personal communication with experts in the field.
Data collection and analysis
Selection of studies
Evaluation of potentially eligible studies to confirm eligibility and to assess methodological quality was performed independently by the three review authors (SM, MP, BH). Disagreements were resolved by discussion, and consensus was reached.
Data extraction and management
Eligible articles were reviewed and summary information extracted. The following information was sought: participant characteristics (age, gender, comorbidity); number of participants in each treatment arm; duration, type, and intensity of VKA; incidence and timing of symptomatic recurrent VTE and major bleeding episodes; and mortality.
Data were extracted independently by the three review authors (SM, MP, BH), using a standard form. Disagreements were resolved by discussion, and consensus was reached.
Assessment of risk of bias in included studies
For this update, two review authors (SM, BH) independently assessed the risk of bias of each trial according to Higgins 2011 and based on the following domains: random sequence generation, allocation concealment, blinding (participants, care providers, or outcome assessors), incomplete outcome data, and other bias. For each of the domains, we assessed whether the study was at high risk of bias, low risk of bias, or unclear risk of bias by using the guidance provided by Higgins 2011. Disagreements were resolved by discussion, and consensus was reached.
Measures of treatment effect
The incidence of recurrent VTE, major bleeding, and mortality for the different treatment arms was used to calculate a risk ratio (RR) separately for each trial. Our outcomes were dichotomous, and we expressed results as RRs with 95% confidence intervals (CIs).
Unit of analysis issues
The participant was the individual unit of analysis for all comparisons.
Dealing with missing data
We analyzed available data (ie, while ignoring missing data).
Assessment of heterogeneity
Heterogeneity of study results was evaluated using the Chi2 test and I2 statistic for each outcome separately. When the probability value of the Chi2 test was < 0.10 and/or the I2 statistic was > 40%, heterogeneity was considered significant.
Assessment of reporting biases
To assess the risk of bias from selective reporting of outcomes, we searched in clinicaltrials.gov and controlled‐trials.com for a study protocol of each trial. If a study protocol was available, we evaluated whether all of the study's prespecified outcomes of interest in our review had been reported in the prespecified way in the final publication.
Data synthesis
All data were analyzed by using the Review Manager software of The Cochrane Collaboration (RevMan 2012). The incidence of recurrent VTE, major bleeding, and mortality for the different treatment arms was used to calculate an RR separately for each trial. These RRs were combined across studies, giving weight to the number of events in each of the two treatment groups in each separate study, using the Mantel‐Haenszel procedure, which assumes a fixed treatment effect (Collins 1987; Mantel 1959; Yusuf 1985).
The advisability of combining the trials was addressed by performing a statistical test of heterogeneity, which considers whether differences in treatment effect over individual trials are consistent with natural variation around a constant effect (Collins 1987). In addition, qualitative assessment of heterogeneity was performed if indicated. When the probability value of the Chi2 test was < 0.10 and/or I2 was > 40%, heterogeneity was considered significant. In cases of significant heterogeneity, data from the studies were combined using a random‐effects model according to the method of DerSimonian and Laird. A Z‐test was performed to test the overall effect. If no significant heterogeneity was observed, studies were combined by using a fixed‐effect model.
Analyses were performed separately for:
-
the period from VKA cessation in the shorter duration arm until VKA cessation in the longer duration arm;
-
the period after cessation of study medication until the end of follow‐up;
-
the entire period after randomization reported in the publication;
-
if available in more than one study, comparisons of two specific durations (eg, six weeks vs six months) of VKA use;
-
studies with adequately concealed randomization; and
-
studies without missing values.
Subgroup analysis and investigation of heterogeneity
If we identified substantial heterogeneity (ie, I2 > 60%), we performed subgroup analyses to explore heterogeneity.
Sensitivity analysis
To determine whether conclusions were robust to decisions made during the review process, we performed analyses separately for studies with adequate randomization and for studies in which none of the participants dropped out or were lost‐to‐follow up, for the period from cessation of treatment with VKA in the short duration arm until cessation of treatment in the long duration arm.
Results
Description of studies
Results of the search
See Figure 1.
Study flow diagram.
Included studies
Three additional studies were included in this update (Eischer 2009; Ridker 2003; Siragusa 2008), making a total of 11 included studies (Agnelli 2001; Agnelli 2003; Eischer 2009; Kearon 1999; Kearon 2004; Levine 1995; Pinede 2001; Ridker 2003; Schulman 1995; Schulman 1997; Siragusa 2008), which were published between 1995 and 2009 with a total of 3716 participants. In seven studies (Agnelli 2001, Agnelli 2003, Eischer 2009, Kearon 1999, Kearon 2004, Schulman 1995, Siragusa 2008), participants with a first episode of VTE (ie, DVT,PE, or both) were included. Of these studies, Eischer 2009 included only participants with levels of FVIII above 230 IU/dL, and Siragusa 2008 included participants with residual vein thrombosis. In the study of Schulman 1997, participants with a second episode of VTE were included, and in the other two studies (Levine 1995; Pinede 2001), participants with acute proximal DVT were included. For the study of Ridker 2003, it was unclear whether participants with a first or second episode of VTE were included. In all studies, objective diagnostic tests were used to confirm the diagnosis.
The 11 studies compared the following different periods of treatment with VKA: four weeks versus three months (Kearon 2004; Levine 1995), six weeks versus 12 weeks (Pinede 2001), six weeks versus six months (Schulman 1995), three months versus six months (Agnelli 2003, Pinede 2001), three months versus one year (Agnelli 2001; Siragusa 2008), three months versus 27 months (Kearon 1999), four months versus 27 months (Ridker 2003), six months versus 30 months (Eischer 2009), and six months versus four years (Schulman 1997). For details of these studies, see the Characteristics of included studies section.
Excluded studies
For this update, seven additional studies were excluded (Agrawal 2011; Ascani 1999; Campbell 2007; Farraj 2004; Ferrara 2006; Palareti 2006; Prandoni 2009), making a total of 14 excluded studies (Agrawal 2011; Ascani 1999; Campbell 2007; Drouet 2003; Farraj 2004; Fennerty 1987; Ferrara 2006; Holmgren 1985; Lagerstedt 1985; O'Sullivan 1972; Palareti 2006; Prandoni 2009; Schulman 1985; Sudlow 1992). Studies were excluded for the following reasons (some studies were excluded for more than one reason): no objective tests used to confirm VTE for all participants (Campbell 2007; Fennerty 1987; Holmgren 1985; O'Sullivan 1972; Sudlow 1992); no blinded or partly blinded outcome assessment or unclear whether blinded outcome measurement was used (Agrawal 2011; Campbell 2007; Drouet 2003; Farraj 2004; Fennerty 1987; Ferrara 2006; Holmgren 1985; Lagerstedt 1985; O'Sullivan 1972; Schulman 1985; Sudlow 1992); INR target range not the same in the treatment arms (Ascani 1999); duration of VKA in one arm tailored on the basis of ultrasonography findings (flexible duration) (Prandoni 2009); and discontinuation of treatment by all participants for one month before randomization (Palareti 2006).
Risk of bias in included studies
See also the Risk of bias in included studies summary (Figure 2 and Figure 3).
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Allocation
Nine studies (82%) were deemed to have adequate sequence generation and therefore were classified as being at low risk of bias for this domain (Agnelli 2001; Agnelli 2003; Kearon 1999; Kearon 2004; Levine 1995; Pinede 2001; Schulman 1995; Schulman 1997; Siragusa 2008). The risk of bias was unclear for two studies because they did not provide information about the randomization process (Eischer 2009; Ridker 2003). In eight studies, the assigned treatment was adequately concealed before allocation (low risk of selection bias) (Agnelli 2001; Agnelli 2003; Kearon 1999; Kearon 2004; Levine 1995; Pinede 2001; Schulman 1995; Schulman 1997) and was unclear for the remaining three studies (Eischer 2009; Ridker 2003; Siragusa 2008).
Blinding
In four of the included studies, participants were randomly assigned to the sham duration of treatment with VKA and received placebo and sham monitoring (Kearon 1999; Kearon 2004; Levine 1995; Ridker 2003). These studies were classified as being at low risk for performance bias. In the other seven included studies, treatment was not blinded (Agnelli 2001; Agnelli 2003; Eischer 2009; Pinede 2001; Schulman 1995; Schulman 1997; Siragusa 2008). Outcome assessment was performed by a committee unaware of treatment allocation in all studies. In two of these studies, blinded outcome assessment was performed only for recurrent VTE, not for bleeding events (Schulman 1995; Schulman 1997). For these studies, the risk for detection bias was classified as unclear.
Incomplete outcome data
One trial reported that 0.6% of participants were excluded after randomization (Schulman 1995), and another study reported that 2.7% of participants withdrew shortly after randomization (Levine 1995). In two studies, 3% of participants were lost to follow‐up (Levine 1995; Pinede 2001). Schulman 1995 and Schulman 1997 mentioned that 4.9% and 6.2% of participants dropped out; however, the study authors were able to collect information about outcome events among these participants from computer registries.
Selective reporting
Only one study was registered at clinicaltrial.gov (Siragusa 2008), and all prespecified outcomes of interest in the review were reported in the prespecified way in the final publication. Therefore the risk for reporting bias was considered low in this study. For the remaining studies, the risk of reporting bias was classified as unclear.
Other potential sources of bias
Upon review of the studies, no other potential sources of bias were identified.
Effects of interventions
Incidence of recurrent VTE
Ten of the 11 studies reported on the occurrence of symptomatic VTE during the period from cessation of treatment with VKA in the short duration arm until cessation of treatment in the long duration arm (Agnelli 2001; Agnelli 2003; Eischer 2009; Kearon 1999; Kearon 2004; Levine 1995; Pinede 2001; Ridker 2003; Schulman 1995; Schulman 1997).
Five of the 10 studies showed statistically significant protection from recurrent venous thromboembolic complications during prolonged treatment with VKA (Kearon 1999; Levine 1995; Ridker 2003; Schulman 1995; Schulman 1997). Four studies showed a clear trend (Agnelli 2001; Agnelli 2003; Kearon 2004; Pinede 2001). The study of Eischer 2009 showed no difference (Analysis 1.1). Combining the ten studies revealed that 155 (8.8%) of 1765 participants had thromboembolic complications in the short arm, and only 30 (1.6%) of 1771 participants had thromboembolic complications in the long arm. Analysis of pooled data from these studies showed a statistically significant reduction in thromboembolic events during this period (RR 0.20, 95% CI 0.11 to 0.38, P < 0.00001).
Seven of the 11 studies evaluated the incidence of recurrent VTE in the period after cessation of study medication until the end of follow‐up (Agnelli 2001; Agnelli 2003; Eischer 2009; Kearon 2004; Levine 1995; Pinede 2001; Schulman 1995). These individual studies did not show a statistically significant increase in venous thromboembolic events among participants in the long arm after cessation of treatment. Combining these studies showed that 101 (7.6%) of 1321 participants who were treated for the longer period with VKA and 78 (5.9%) of 1318 participants who were treated for a shorter period experienced a recurrence (Analysis 2.1). Analysis of the pooled data showed a non–statistically significant difference in the incidence of recurrence during this period (RR 1.28, 95% CI 0.97 to 1.70, P = 0.09). Although a somewhat higher risk was found for participants in the long duration arm after cessation of treatment compared with those in the short duration arm, a rebound effect could not be clearly demonstrated.
When the entire period after randomization reported in the publication was considered, four of the 11 studies showed statistically significant protection from recurrent thromboembolic complications (Kearon 1999; Ridker 2003; Schulman 1995; Schulman 1997). See Analysis 3.1. Substantial heterogeneity could be observed between the studies (P = 0.0005, I2 = 68%). Graphically, two groups can be identified: those with extended follow‐up after cessation of treatment with VKA in the long arm (Agnelli 2001; Agnelli 2003; Eischer 2009; Kearon 2004; Levine 1995; Pinede 2001; Ridker 2003; Schulman 1995; Siragusa 2008) (pooled data: RR 0.74, 95% CI 0.54 to 1.01) and those without extended follow‐up after cessation of treatment with VKA in the long arm (Kearon 1999; Schulman 1997) (pooled data: RR 0.10, 95% CI 0.04 to 0.29). Therefore, we decided to refrain from pooling all studies.
Comparisons of two specific durations
It was possible to extract data from more than one study for the following durations.
One month versus three months
Pooling the data from Kearon 2004 and Levine 1995 showed a significant reduction in recurrent VTE in participants who had received prolonged VKA treatment (RR 0.18, 95% CI 0.04 to 0.79, P = 0.02). See Analysis 4.1.
Three months versus six months
Pooling the data from Agnelli 2001, Agnelli 2003, Kearon 1999, and Pinede 2001 showed a significant reduction in recurrent VTE among participants who had received prolonged VKA treatment (RR 0.10, 95% CI 0.02 to 0.43, P = 0.002). See Analysis 5.1.
Three months versus 12 months
Pooling the data from Agnelli 2001, Agnelli 2003, and Kearon 1999 revealed a significant reduction in recurrent VTE among participants who had received prolonged VKA treatment (RR 0.18, 95% CI 0.071 to 0.45, P = 0.0002). See Analysis 6.1.
Incidence of major bleeding
Six studies reported on the incidence of major bleeding during the period from cessation of treatment with VKA in the short duration arm until cessation of treatment in the long duration arm (Agnelli 2001; Eischer 2009; Kearon 1999; Kearon 2004; Levine 1995; Ridker 2003). None of the individual studies showed a statistically significant increase in bleeding complications during prolonged treatment with VKA. Combining these studies revealed that three (0.4%) of 675 participants had major bleeding in the short treatment arm versus 14 (2.1%) of 675 participants in the long treatment arm (Analysis 1.2). Analysis of pooled data from these studies showed a statistically significant increase in major bleeding complications during this period (RR 3.44, 95% CI 1.22 to 9.74, P = 0.02).
Two studies reported on the incidence of major bleeding in the period after cessation of study medication in the long duration arm until end of follow‐up (Eischer 2009; Kearon 2004). However, both studies reported no major bleeding events during this period.
All included trials reported on the occurrence of major bleeding complications for the entire period after randomization until end of follow‐up (Agnelli 2001; Agnelli 2003; Eischer 2009; Kearon 1999; Kearon 2004; Levine 1995; Pinede 2001; Ridker 2003; Schulman 1995; Schulman 1997; Siragusa 2008). None of the individual studies showed a statistically significant increase in bleeding complications during prolonged treatment with VKA. Combining these studies revealed that 44 (2.4%) of 1859 participants with prolonged treatment and 16 (0.9%) of 1857 participants with short treatment had major bleeding (Analysis 3.2). Analysis of pooled data showed an increase in major bleeding during the entire study period (RR 2.609, 95% CI 1.51 to 4.49, P = 0.0006).
Mortality
Four studies reported mortality during the period from cessation of treatment with VKA in the short duration arm until cessation of treatment in the long duration arm (Kearon 1999; Kearon 2004; Levine 1995; Ridker 2003). Two of these studies showed a non–statistically significant reduction in mortality during prolonged treatment with vitamin K antagonists (Kearon 1999; Ridker 2003). No trend was observed in Levine 1995. Combining these studies showed that 20 (3.8%) of 525 participants in the short arm died, as did 14 (2.7%) of 524 participants in the long arm (Analysis 1.3). Analysis of pooled data from these studies showed that prolonged treatment was associated with a non–statistically significant reduction in mortality (RR 0.69, 95% CI 0.35 to 1.34, P = 0.27).
Only one study reported specifically on the number of participants who died during the period after cessation of study medication in the long arm until end of follow‐up (Kearon 2004). Nine studies reported on mortality for the entire period after randomization (Agnelli 2001; Agnelli 2003; Kearon 1999; Kearon 2004; Levine 1995; Pinede 2001; Ridker 2003; Schulman 1995; Schulman 1997). None showed a statistically significant reduction in mortality. Combining these studies revealed that 75 (4.3%) of 1753 participants died with prolonged treatment and 83 (4.7%) of 1749 participants died without prolonged treatment with VKA (Analysis 3.3). Analysis of pooled data showed a non–statistically significant reduction in mortality during the entire study period (RR 0.89, 95% CI 0.66 to 1.21, P = 0.46).
Sensitivity analysis
Analysis of studies with adequate concealment of allocation before randomization
Separate analyses for studies with adequate allocation concealment did not change the results significantly.
Analysis of studies without missing values
Separate analyses for studies in which none of the participants dropped out or were lost to follow‐up did not change the results significantly.
Discussion
Summary of main results
In this review, data from included studies were pooled to evaluate the efficacy and safety of different durations of treatment with VKA among participants with symptomatic VTE.
We found a statistically significant reduction in recurrent VTE during the period in which treatment with VKA was prolonged, which was independent of the period elapsed since the index thromboembolic event (RR 0.20, 95% CI 0.11 to 0.38, P < 0.00001). Although the periods of treatment differed greatly between studies, these periods could be combined because the relative effect of oral anticoagulant treatment was considered and was homogeneous during the period of continuation. Reduction in recurrent VTE during the period that treatment with VKA was prolonged may be somewhat counterbalanced by an excess of recurrences shortly after cessation of prolonged treatment, but this finding did not reach statistical significance (RR 1.28, 95% CI 0.97 to 1.70, P = 0.09). In addition, a substantial increase in bleeding complications was observed among participants during the period in which VKA was prolonged (RR 3.44, 95% CI 1.22 to 9.74), and a non–statistically significant reduction in mortality was shown (RR 0.69, 95% CI 0.35 to 1.34).
Overall completeness and applicability of evidence
Only four studies reported mortality during the period from cessation of treatment with VKA in the short duration arm until cessation of treatment in the long duration arm (Kearon 1999; Kearon 2004; Levine 1995; Ridker 2003). Furthermore, as in all randomized controlled trials, stringent inclusion and exclusion criteria were applied, which means that the current evidence is applicable to patients without increased risk of bleeding. It is likely that absolute bleeding risk is higher in real‐world patients.
Quality of the evidence
See summary of findings Table for the main comparison. In general, high‐quality evidence suggests that prolonged treatment with VKA reduces the risk for recurrent VTE. Moderate‐quality evidence indicates that prolonged treatment with VKA increases the risk for major bleeding events. We considered the quality of evidence as moderate because of relatively wide confidence intervals around the point estimate; however the point estimates were very similar across all included trials. Moderate‐quality evidence suggests that prolonged treatment with VKA does not reduce mortality significantly, although this was measured in only four studies, one of which reported no deaths (Kearon 1999; Kearon 2004; Levine 1995; Ridker 2003).
Potential biases in the review process
As two review authors selected and extracted the data independently, the risk of potential bias will be low. Furthermore, we consider it unlikely that we have missed important trials in our search for data because of the extensive literature searches that we conducted. As all studies were investigator‐initiated, we consider it unlikely that trials with a less favourable outcome have not been published.
Agreements and disagreements with other studies or reviews
Although the relative risk reductions remain stable over time elapsed since the index event (RRs varying around 80%), the absolute risk reduction is decreased over time. This can be illustrated by comparing the studies of Levine 1995 and Schulman 1995 with that of Schulman 1997. The studies of Levine 1995 and Schulman 1995 showed an absolute risk reduction of 8% to 9% achieved with only two to 4.5 months of prolonged treatment in the early phases after the thrombotic event. This is far more efficient than the absolute risk reduction of 18% achieved with 42 months of additional treatment in the study of Schulman 1997, six months after the index event. The decline in the incidence of recurrent VTE over time was also observed in the cohort studies of Prandoni 1996 and in the meta‐analysis of cohort studies and randomized controlled trials performed by van Dongen 2003. All of this indicates that a greater amount of effort (ie, years of treatment) will be needed to prevent one recurrent event when the time since the index event is increased. This fact is further complicated by a statistically significant and clinically important increase in bleeding complications, which continued during prolongation. A meta‐analysis of 33 trials and prospective cohort studies showed that absolute bleeding risk among participants with VTE treated for longer than three months with VKA was 2.7 per 100 patient‐years (Linkins 2003). The case fatality rate of major bleeding was 13.4% (9.4% to 17.4%), and the rate of intracranial bleeding was 1.15 (1.14 to 16) per 100 patient‐years (Linkins 2003). For participants who received anticoagulants for longer than three months, the case fatality rate of major bleeding remained high at 9.1% (2.5% to 21.7%), and the rate of intracranial bleeding was 0.65 (0.63 to 0.68) per 100 patient‐years. This adds to a further decrease in net clinical benefit of prolonged treatment with VKA after VTE.
The decrease in efficacy and the remaining risk for bleeding during continuing treatment indicate that at some point in time, further continuation is not cost‐effective, nor is it harmful. Given that the case fatality rate of recurrent VTE decreases over time (Carrier 2010), whereas the risk of major bleeding increases with age, the optimum length of anticoagulant treatment after an episode of VTE remains uncertain (De Jong 2012; Middeldorp 2011). As patients have different risk profiles, it is likely that this optimal duration will vary. However, on the basis of our results alone, the optimal duration cannot be defined. For this purpose, a decision analytic approach could be used by balancing benefit and risk on the basis of individual risk profiles (Prins 1999a; Prins 1999b).
In conclusion, this meta‐analysis shows that treatment with VKA reduces the risk of recurrent VTE for as long as they are used. However, the absolute risk of recurrent VTE declines over time, although the risk for major bleeding remains. Thus, the efficacy of VKA administration decreases over time from occurrence of the index event.
Study flow diagram.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Comparison 1 Long vs short, period from cessation of VKA in short arm until VKA cessation in long arm, Outcome 1 Incidence of recurrent VTE.
Comparison 1 Long vs short, period from cessation of VKA in short arm until VKA cessation in long arm, Outcome 2 Incidence of major bleeding.
Comparison 1 Long vs short, period from cessation of VKA in short arm until VKA cessation in long arm, Outcome 3 Mortality.
Comparison 2 Long vs short, period after cessation of study medication until end of follow‐up, Outcome 1 Incidence of recurrent VTE.
Comparison 2 Long vs short, period after cessation of study medication until end of follow‐up, Outcome 2 Incidence of major bleeding.
Comparison 2 Long vs short, period after cessation of study medication until end of follow‐up, Outcome 3 Mortality.
Comparison 3 Long vs short, entire period after randomization reported in publication, Outcome 1 Incidence of recurrent VTE.
Comparison 3 Long vs short, entire period after randomization reported in publication, Outcome 2 Incidence of major bleeding.
Comparison 3 Long vs short, entire period after randomization reported in publication, Outcome 3 Mortality.
Comparison 4 One month vs 3 months, period from cessation of VKA in short arm until VKA cessation in long arm, Outcome 1 Incidence of recurrent VTE.
Comparison 5 Three months vs 6 months, period from cessation of VKA in short arm until VKA cessation in long arm, Outcome 1 Incidence of recurrent VTE.
Comparison 6 Three months vs 12 months, period from cessation of VKA in short arm until VKA cessation in long arm, Outcome 1 Incidence of recurrent VTE.
| Long‐term or short‐term treatment with vitamin K antagonists for patients with venous thromboembolism | ||||||
| Patient or population: patients with venous thromboembolism | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No. of participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Short‐term treatment with VKA | Long‐term treatment with VKA | |||||
| Incidence of recurrent VTE | 88 per 1000 | 18 per 1000 | RR 0.2 | 3536 | ⊕⊕⊕⊕ | |
| Incidence of major bleeding | 4 per 1000 | 15 per 1000 | RR 3.44 | 1350 | ⊕⊕⊕⊝ | |
| Mortality | 38 per 1000 | 26 per 1000 | RR 0.69 | 1049 | ⊕⊕⊕⊝ | |
| *The basis for the assumed risk (eg, the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence. | ||||||
| 1Relatively wide 95% confidence interval around the estimate. | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Incidence of recurrent VTE Show forest plot | 10 | 3536 | Risk Ratio (M‐H, Random, 95% CI) | 0.20 [0.11, 0.38] |
| 2 Incidence of major bleeding Show forest plot | 6 | 1350 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.44 [1.22, 9.74] |
| 3 Mortality Show forest plot | 4 | 1049 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.69 [0.35, 1.34] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Incidence of recurrent VTE Show forest plot | 7 | 2639 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.28 [0.97, 1.70] |
| 2 Incidence of major bleeding Show forest plot | 2 | 179 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] |
| 3 Mortality Show forest plot | 1 | 165 | Risk Ratio (M‐H, Fixed, 95% CI) | 3.11 [0.13, 75.24] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Incidence of recurrent VTE Show forest plot | 11 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 1.1 With extended follow‐up after VKA cessation | 9 | 3327 | Risk Ratio (M‐H, Random, 95% CI) | 0.74 [0.54, 1.01] |
| 1.2 Without extended follow‐up after VKA cessation | 2 | 389 | Risk Ratio (M‐H, Random, 95% CI) | 0.10 [0.04, 0.29] |
| 2 Incidence of major bleeding Show forest plot | 11 | 3716 | Risk Ratio (M‐H, Fixed, 95% CI) | 2.60 [1.51, 4.49] |
| 3 Mortality Show forest plot | 9 | 3502 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.89 [0.66, 1.21] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Incidence of recurrent VTE Show forest plot | 2 | 379 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.18 [0.04, 0.79] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Incidence of recurrent VTE Show forest plot | 4 | 1113 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.10 [0.02, 0.43] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Incidence of recurrent VTE Show forest plot | 3 | 610 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.18 [0.07, 0.45] |
