Assessing treatment-time interaction in clinical trials with time to event data: a meta-analysis of hypertension trials

Individual participant data meta-analysis of intervention studies with time-to-event outcomes: A review of the methodology and an applied example

Many randomized trials evaluate an intervention effect on time-to-event outcomes. Individual participant data (IPD) from such trials can be obtained and combined in a so-called IPD meta-analysis (IPD-MA), to summarize the overall intervention effect. We performed a narrative literature review to provide an overview of methods for conducting an IPD-MA of randomized intervention studies with a time-to-event outcome. We focused on identifying good methodological practice for modeling frailty of trial participants across trials, modeling heterogeneity of intervention effects, choosing appropriate association measures, dealing with (trial differences in) censoring and follow-up times, and addressing time-varying intervention effects and effect modification (interactions).We discuss how to achieve this using parametric and semi-parametric methods, and describe how to implement these in a one-stage or two-stage IPD-MA framework. We recommend exploring heterogeneity of the effect(s) through interaction and non-linear effects. Random effects should be applied to account for residual heterogeneity of the intervention effect. We provide further recommendations, many of which specific to IPD-MA of time-to-event data from randomized trials examining an intervention effect.We illustrate several key methods in a real IPD-MA, where IPD of 1225 participants from 5 randomized clinical trials were combined to compare the effects of Carbamazepine and Valproate on the incidence of epileptic seizures.

Effect model law: an approach for the implementation of personalized medicine

The effect model law states that a natural relationship exists between the frequency (observation) or the probability (prediction) of a morbid event without any treatment and the frequency or probability of the same event with a treatment. This relationship is called the effect model. It applies to a single individual, individuals within a population, or groups. In the latter case, frequencies or probabilities are averages of the group. The relationship is specific to a therapy, a disease or an event, and a period of observation. If one single disease is expressed through several distinct events, a treatment will be characterized by as many effect models. Empirical evidence, simulations with models of diseases and therapies and virtual populations, as well as theoretical derivation support the existence of the law. The effect model could be estimated through statistical fitting or mathematical modelling. It enables the prediction of the (absolute) benefit of a treatment for a given patient. It thus constitutes the theoretical basis for the design of practical tools for personalized medicine.

Evidence of subgroup-specific treatment effect in the absence of an overall effect: is there really a contradiction?

PURPOSE

Interaction and subgroup analyses remain controversial topics in epidemiology. A recent theoretical paper suggested that a combination of no overall treatment-outcome association and treatment effect limited to a single subgroup would imply a clinically implausible interaction, with opposite treatment effects in the two subgroups. However, this argument was based entirely on point estimates and ignored sampling error and statistical inference.

METHODS

We simulated hypothetical studies in which treatment truly affected the outcome in only one subgroup, with no effect in the other subgroup. We generated 1000 random samples for three study designs (small clinical study, case-control, and large cohort), and different values of total sample size (N), relative size of the affected subgroup, and treatment effect. We estimated the frequency of significant results for tests of overall and subgroup-specific treatment effects, and treatment-by-subgroup interaction.

RESULTS

Combination of statistically non-significant overall treatment effect and significant treatment-by-subgroup interaction occurred frequently, especially if the affected subgroup was proportionally smaller, even in studies with high power to detect the overall effect (e.g. in 37.1% of samples with N = 20 000, with 600 outcomes, and an effect (odds ratio of 1.5) limited to 30% of subjects). Furthermore, in most samples with a significant interaction, subgroup analyses correctly indicated that the significant effect was limited to one subgroup.

CONCLUSION

In studies where the treatment truly affects the risks in only one subgroup, a non-significant overall effect will often coincide with a statistically significant treatment-by-subgroup interaction. Thus, a non-significant overall effect should not prevent testing plausible interactions.

Investigating clinical heterogeneity in systematic reviews: a methodologic review of guidance in the literature

BACKGROUND

While there is some consensus on methods for investigating statistical and methodological heterogeneity, little attention has been paid to clinical aspects of heterogeneity. The objective of this study is to summarize and collate suggested methods for investigating clinical heterogeneity in systematic reviews.

METHODS

We searched databases (Medline, EMBASE, CINAHL, Cochrane Library, and CONSORT, to December 2010) and reference lists and contacted experts to identify resources providing suggestions for investigating clinical heterogeneity between controlled clinical trials included in systematic reviews. We extracted recommendations, assessed resources for risk of bias, and collated the recommendations.

RESULTS

One hundred and one resources were collected, including narrative reviews, methodological reviews, statistical methods papers, and textbooks. These resources generally had a low risk of bias, but there was minimal consensus among them. Resources suggested that planned investigations of clinical heterogeneity should be made explicit in the protocol of the review; clinical experts should be included on the review team; a set of clinical covariates should be chosen considering variables from the participant level, intervention level, outcome level, research setting, or others unique to the research question; covariates should have a clear scientific rationale; there should be a sufficient number of trials per covariate; and results of any such investigations should be interpreted with caution.

CONCLUSIONS

Though the consensus was minimal, there were many recommendations in the literature for investigating clinical heterogeneity in systematic reviews. Formal recommendations for investigating clinical heterogeneity in systematic reviews of controlled trials are required.

Is blood pressure reduction a valid surrogate endpoint for stroke prevention? An analysis incorporating a systematic review of randomised controlled trials, a by-trial weighted errors-in-variables regression, the surrogate threshold effect (STE) and the Biomarker-Surrogacy (BioSurrogate) Evaluation Schema (BSES)

Background

Blood pressure is considered to be a leading example of a valid surrogate endpoint. The aims of this study were to (i) formally evaluate systolic and diastolic blood pressure reduction as a surrogate endpoint for stroke prevention and (ii) determine what blood pressure reduction would predict a stroke benefit.

Methods

We identified randomised trials of at least six months duration comparing any pharmacologic anti-hypertensive treatment to placebo or no treatment, and reporting baseline blood pressure, on-trial blood pressure, and fatal and non-fatal stroke. Trials with fewer than five strokes in at least one arm were excluded. Errors-in-variables weighted least squares regression modelled the reduction in stroke as a function of systolic blood pressure reduction and diastolic blood pressure reduction respectively. The lower 95% prediction band was used to determine the minimum systolic blood pressure and diastolic blood pressure difference, the surrogate threshold effect (STE), below which there would be no predicted stroke benefit. The STE was used to generate the surrogate threshold effect proportion (STEP), a surrogacy metric, which with the R-squared trial-level association was used to evaluate blood pressure as a surrogate endpoint for stroke using the Biomarker-Surrogacy Evaluation Schema (BSES3).

Results

In 18 qualifying trials representing all pharmacologic drug classes of antihypertensives, assuming a reliability coefficient of 0.9, the surrogate threshold effect for a stroke benefit was 7.1 mmHg for systolic blood pressure and 2.4 mmHg for diastolic blood pressure. The trial-level association was 0.41 and 0.64 and the STEP was 66% and 78% for systolic and diastolic blood pressure respectively. The STE and STEP were more robust to measurement error in the independent variable than R-squared trial-level associations. Using the BSES3, assuming a reliability coefficient of 0.9, systolic blood pressure was a B + grade and diastolic blood pressure was an A grade surrogate endpoint for stroke prevention. In comparison, using the same stroke data sets, no STEs could be estimated for cardiovascular (CV) mortality or all-cause mortality reduction, although the STE for CV mortality approached 25 mmHg for systolic blood pressure.

Conclusions

In this report we provide the first surrogate threshold effect (STE) values for systolic and diastolic blood pressure. We suggest the STEs have face and content validity, evidenced by the inclusivity of trial populations, subject populations and pharmacologic intervention populations in their calculation. We propose that the STE and STEP metrics offer another method of evaluating the evidence supporting surrogate endpoints. We demonstrate how surrogacy evaluations are strengthened if formally evaluated within specific-context evaluation frameworks using the Biomarker- Surrogate Evaluation Schema (BSES3), and we discuss the implications of our evaluation of blood pressure on other biomarkers and patient-reported instruments in relation to surrogacy metrics and trial design.

Effect of smoking relapse on outcome after acute coronary syndromes

The aim of the present study was to evaluate the smoking relapse rate among smokers who had become abstinent during admission for acute coronary syndromes. The association between smoking relapse and mortality was also analyzed. A cohort of 1,294 consecutive active smokers who had interrupted smoking after admission for acute coronary syndromes (1,018 men and 276 women, mean age 59.7 ± 12.3 years) was followed up for 12 months after the index admission. All patients received a brief in-hospital smoking cessation intervention consisting of repeated counseling sessions. During follow-up, 813 patients (62.8%) resumed regular smoking (median interval to relapse 19 days, interquartile range 9 to 76). Increasing age (hazard ratio [HR] 1.034 per year, 95% confidence interval [CI] 1.028 to 1.039, p = 0.001) and female gender (HR 1.23, 95% CI 1.09 to 1.42, p = 0.03) were independent predictors of smoking relapse. Patients enrolled in a cardiac rehabilitation program (HR 0.74, 95% CI 0.51 to 0.91, p = 0.02) and those with diabetes (HR 0.79, 95% CI 0.68 to 0.94, p = 0.03) were more likely to remain abstinent. During follow-up, 97 patients died (1-year probability of death 0.075, 95% CI 0.061 to 0.090). Multivariate analysis with the Cox proportional hazard regression method, including smoking relapse as a time-dependent covariate, demonstrated that, after adjustment for patient demographics, the clinical history features and variables related to the index event, the resumption of smoking was an independent predictor of total mortality (HR 3.1, 95% CI 1.3 to 5.7, p = 0.004). In conclusion, smoking relapse after acute coronary syndromes is associated with increased mortality, and counseling interventions should be integrated into the postdischarge support to reduce the negative effects of smoking resumption.

Rising adiposity curbing decline in the incidence of myocardial infarction: 20-year follow-up of British men and women in the Whitehall II cohort

Aims

To estimate the contribution of risk factor trends to 20-year declines in myocardial infarction (MI) incidence in British men and women.

Methods and results

From 1985 to 2004, 6379 men and 3074 women in the Whitehall II cohort were followed for incident MI and risk factor trends. Over 20 years, the age–sex-adjusted hazard of MI fell by 74% (95% confidence interval 48–87%), corresponding to an average annual decline of 6.5% (3.2–9.7%). Thirty-four per cent (20–76%) of the decline in MI hazard could be statistically explained by declining non-HDL cholesterol levels, followed by increased HDL cholesterol (17%, 10–32%), reduced systolic blood pressure (13%, 7–24%), and reduced cigarette smoking prevalence (6%, 2–14%). Increased fruit and vegetable consumption made a non-significant contribution of 7% (−1–20%). In combination, these five risk factors explained 56% (34–112%). Rising body mass index (BMI) was counterproductive, reducing the scale of the decline by 11% (5–23%) in isolation. The MI decline and the impact of the risk factors appeared similar for men and women.

Conclusion

In men and women, over half of the decline in MI risk could be accounted for by favourable risk factor time trends. The adverse role of BMI emphasizes the importance of addressing the rising population BMI.

One-stage parametric meta-analysis of time-to-event outcomes

Methodology for the meta-analysis of individual patient data with survival end-points is proposed. Motivated by questions about the reliance on hazard ratios as summary measures of treatment effects, a parametric approach is considered and percentile ratios are introduced as an alternative to hazard ratios. The generalized log-gamma model, which includes many common time-to-event distributions as special cases, is discussed in detail. Likelihood inference for percentile ratios is outlined. The proposed methodology is used for a meta-analysis of glioma data that was one of the studies which motivated this work. A simulation study exploring the validity of the proposed methodology is available electronically. Copyright © 2010 John Wiley & Sons, Ltd.

Clinical implications of switching from intensive to moderate statin therapy after acute coronary syndromes

BACKGROUND AND AIMS

Intensive statin therapy represents an effective option after acute coronary syndromes (ACS). Despite evidence, switching to less effective statins frequently occurs in practice. Aim of this observational study was to assess the impact of switching from intensive to moderate statin therapy on clinical outcomes after ACS.

METHODS AND RESULTS

A cohort of 1321 consecutive ACS patients (886 men, age 71 ± .8 years) discharged on atorvastatin 80 mg/d in a 6.5 year period was followed for 12 months after discharge. During follow-up, 557 patients (42%) were switched by primary care physicians to moderate statin therapy, either for side effects (56%) or for safety concerns (44%). No major adverse reaction was reported. Increasing age (HR 1.52 per 10-year increase, 95% CI 1.23-1.78, p=0.01), and female gender (HR 1.11, 95% CI 1.06-1.23, p=0.02) were associated with a higher probability of switch. Patients following a cardiac rehabilitation program (HR 0.64 95% CI 0.49-0.86, p=0.02) and diabetic subjects (HR 0.81, 95% CI 0.67-0.92, p=0.02) were more likely to continue atorvastatin 80 mg/d. During follow-up, a major adverse clinical event occurred in 331 patients (one-year probability 0.25, 95% CI 0.22-0.27). Multivariate analysis with Cox proportional hazards method, including statin switching as a time-dependent covariate, demonstrated that, after adjustment for demographic and clinical variables, reduction from intensive to moderate statin therapy was an independent predictor of adverse clinical outcomes (HR 2.7, 95% CI 1.7-5.1, p=0.004).

CONCLUSION

Switching from intensive to moderate statin therapy after ACS is associated with an increased incidence of adverse clinical events.

The ScanBrit randomised, controlled, single-blind study of a gluten- and casein-free dietary intervention for children with autism spectrum disorders

There is increasing interest in the use of gluten- and casein-free diets for children with autism spectrum disorders (ASDs). We report results from a two-stage, 24-month, randomised, controlled trial incorporating an adaptive 'catch-up' design and interim analysis. Stage 1 of the trial saw 72 Danish children (aged 4 years to 10 years 11 months) assigned to diet (A) or non-diet (B) groups by stratified randomisation. Autism Diagnostic Observation Schedule (ADOS) and the Gilliam Autism Rating Scale (GARS) were used to assess core autism behaviours, Vineland Adaptive Behaviour Scales (VABS) to ascertain developmental level, and Attention-Deficit Hyperactivity Disorder - IV scale (ADHD-IV) to determine inattention and hyperactivity. Participants were tested at baseline, 8, and 12 months. Based on per protocol repeated measures analysis, data for 26 diet children and 29 controls were available at 12 months. At this point, there was a significant improvement to mean diet group scores (time*treatment interaction) on sub-domains of ADOS, GARS and ADHD-IV measures. Surpassing of predefined statistical thresholds as evidence of improvement in group A at 12 months sanctioned the re-assignment of group B participants to active dietary treatment. Stage 2 data for 18 group A and 17 group B participants were available at 24 months. Multiple scenario analysis based on inter- and intra-group comparisons showed some evidence of sustained clinical group improvements although possibly indicative of a plateau effect for intervention. Our results suggest that dietary intervention may positively affect developmental outcome for some children diagnosed with ASD. In the absence of a placebo condition to the current investigation, we are, however, unable to disqualify potential effects derived from intervention outside of dietary changes. Further studies are required to ascertain potential best- and non-responders to intervention. The study was registered with ClincialTrials.gov, number NCT00614198.

A randomized clinical trial of exercise to alleviate postpartum depressed mood

OBJECTIVE

To evaluate whether a 12-week home-based exercise program is more effective than usual care for alleviating depressive symptomology in the postpartum.

METHODS

Eighty-eight women experiencing postpartum depressed mood were randomly assigned to a 12-week home-based exercise program or usual care. Outcomes assessed immediately post-treatment and 3-months post-treatment were the Hamilton Rating Scale for Depression (HAM-D) and Edinburgh Postnatal Depression Scale (EPDS).

RESULTS

In the intention-to-treat analysis, the effect of the intervention on EPDS did not change from 3 to 6 months evaluations, but was modified by the baseline EPDS score, with subjects with greater depression at baseline (EPDS > 13) in the intervention group having a significantly lower postbaseline EPDS score compared with the usual care group (mean difference 4.06 points, 95%CI 1.51-6.61, p < 0.001). After adjusting for baseline HAM-D, subjects in the intervention group had a significantly lower HAM-D score at post-treatment compared with subjects in the usual care group (mean difference 1.83 points, 95%CI 0.24-3.41, p = 0.02). The difference in HAM-D became non-significant at 3-months post-treatment.

CONCLUSIONS

Home-based exercise is a feasible nonpharmacological intervention with the potential to alleviate postpartum depressive symptoms, especially in women with higher initial depressed mood scores as measured by the EPDS. These findings may guide the design of future exercise clinical trials with postpartum depressed women.

How much of the recent decline in the incidence of myocardial infarction in British men can be explained by changes in cardiovascular risk factors? Evidence from a prospective population-based study

BACKGROUND

The incidence of myocardial infarction (MI) in Britain has fallen markedly in recent years. Few studies have investigated the extent to which this decline can be explained by concurrent changes in major cardiovascular risk factors.

METHODS AND RESULTS

The British Regional Heart Study examined changes in cardiovascular risk factors and MI incidence over 25 years from 1978 in a cohort of 7735 men. During this time, the age-adjusted hazard of MI decreased by 3.8% (95% confidence interval 2.6% to 5.0%) per annum, which corresponds to a 62% decline over the 25 years. At the same time, after adjustment for age, cigarette smoking prevalence, mean systolic blood pressure, and mean non-high-density lipoprotein (HDL) cholesterol decreased, whereas mean HDL cholesterol, mean body mass index, and physical activity levels rose. No significant change occurred in alcohol consumption. The fall in cigarette smoking explained the greatest part of the decline in MI incidence (23%), followed by changes in blood pressure (13%), HDL cholesterol (12%), and non-HDL cholesterol (10%). In combination, 46% (approximate 95% confidence interval 23% to 164%) of the decline in MI could be explained by these risk factor changes. Physical activity and alcohol consumption had little influence, whereas the increase in body mass index would have produced a rise in MI risk.

CONCLUSIONS

Modest favorable changes in the major cardiovascular risk factors appear to have contributed to considerable reductions in MI incidence. This highlights the potential value of population-wide measures to reduce exposure to these risk factors in the prevention of coronary heart disease.

Discontinuation of Statin Therapy and Clinical Outcome After Ischemic Stroke

Background and Purpose— The majority of patients with previous ischemic stroke are expected to benefit significantly from long-term statin therapy. However, discontinuation of medication therapy frequently occurs in clinical practice. The aim of this study was to assess the impact of discontinued statin therapy on clinical outcome in patients discharged after an acute ischemic stroke.

Methods— The study population included 631 consecutive stroke survivors (322 men and 309 women; mean±SD age, 70.2±7.6 years) without clinical evidence of coronary heart disease. All patients were discharged on statin therapy and were followed up for 12 months after the acute ischemic stroke.

Results— Within 12 months from discharge, 246 patients (38.9%) discontinued statin therapy; the mean time from discharge to statin discontinuation was 48.6±54.9 days (median time, 30 days; interquartile range, 18 to 55 days). During follow-up, 116 patients died (1-year probability of death=0.18; 95% CI, 0.15 to 0.21). Multivariate analysis demonstrated that after adjustment for all confounders and interactions, statin therapy discontinuation (hazard ratio=2.78; 95% CI, 1.96 to 3.72; P =0.003) was an independent predictor of all-cause 1-year mortality.

Conclusions— A large number of patients discontinue their use of statins early after acute stroke. Moreover, patients discontinuing statins have a significantly increased mortality during the first year after the acute cerebrovascular event. These findings suggest that patient care should be improved during the transition from a hospital setting to outpatient primary care.

Simulations showed that validation of database-derived diagnostic criteria based on a small subsample reduced bias

OBJECTIVE

To evaluate alternative approaches to correct for bias due to inaccurate diagnostic criteria in database studies of associations.

STUDY DESIGN AND SETTINGS

A simulation study of a hypothetical cohort of 10,000 subjects selected based on database-derived diagnostic criteria with positive predictive value (PPV) of either 53% or 80%. Analyses focus on the putative association between a drug and the time to a negative outcome. The association is confounded for "false positive" subjects, where the drug acts as a marker for unobserved frailty. First, we estimate the conventional multivariable Cox's Model 1. We then assume having in-depth evaluation of a fraction of subjects, which permits estimating the probabilities of having the disease for all subjects in the cohort. Alternative correction methods use the estimated probability as a confounder (Model 2), a modifier of the drug effect (Model 3), or an importance weight (Model 4).

RESULTS

With a PPV of 53%, Models 1 and 2 induced about 50% underestimation bias for the drug effect. Interaction-based Model 3 yielded the least biased estimates (25% bias), whereas weighting by probability (Model 4) resulted in slightly more biased (33%), but more stable estimates.

CONCLUSION

Proposed methods help reducing bias due to sample contamination.

Treatment of high blood pressure and gain in event-free life expectancy

Objective

Consideration of absolute risk has been recommended for making decisions concerning preventive treatment in hypertension. We performed simulations to estimate the benefit of antihypertensive therapy over a life-time.

Methodology

The rate of nonfatal and fatal events of untreated hypertensives in the US population were estimated using data from Individual Data ANalysis of Antihypertensive drug intervention trials (INDANA; a meta-analysis on individual data in hypertension) and specific cause of death from national statistics. Disease-free survival curves until all patients have died were built using the “life-table” method. The treatment effect estimated from INDANA was applied to this curve to obtain the disease-free survival curve of the life-long treated population. Gains in event-free life expectancy (GLE) were estimated from survival curves. A sensitivity analysis was performed to assess the impact of possible death misclassifications.

Results

For a 40-year-old man, the gain in life expectancy without stroke and major cardiovascular events were 27 and 32 months, respectively, and were more substantial than those without coronary disease (19 months). The GLE decreased slowly with increasing age at the beginning of treatment, whereas short-term absolute risk reductions increase sharply with age.

Conclusions

Policies based on the selection of patients to treat according to absolute benefit do not maximize the GLE compared with strategies that treat low-risk patients.

Estimating the distribution of lag in the effect of short-term exposures and interventions: adaptation of a non-parametric regression spline model

Information on the distribution of lag duration between exposure or intervention and the subsequent changes in risk can help in assessing the impact of exposure, predicting cost-effectiveness of intervention, and understanding the underlying biological mechanisms. Previous approaches focused more on optimizing the strength of the exposure-disease association than on directly estimating lag duration. We propose an alternative approach applicable to the analysis of the lagged effects of binary exposure variables. The density function of the distribution of lags is estimated based on flexible modelling of changes in hazard ratio of exposed versus unexposed subjects. The methodology is evaluated in a simulation study and is applied to the Framingham data to investigate the lagged effect of smoking cessation on coronary heart disease risk.

Aggregate data meta-analysis with time-to-event outcomes

In a meta-analysis of randomized controlled trials with time-to-event outcomes, an aggregate data approach may be required for some or all included studies. Variation in the reporting of survival analyses in journals suggests that no single method for extracting the log(hazard ratio) estimate will suffice. Methods are described which improve upon a previously proposed method for estimating the log(HR) from survival curves. These methods extend to life-tables. In the situation where the treatment effect varies over time and the trials in the meta-analysis have different lengths of follow-up, heterogeneity may be evident. In order to assess whether the hazard ratio changes with time, several tests are proposed and compared. A cohort study comparing life expectancy of males and females with cerebral palsy and a systematic review of five trials comparing two anti-epileptic drugs, carbamazepine and sodium valproate, are used for illustration.

Survival plots of time-to-event outcomes in clinical trials: good practice and pitfalls

Survival plots of time-to-event data are a key component for reporting results of many clinical trials (and cohort studies). However, mistakes and distortions often arise in the display and interpretation of survival plots. This article aims to highlight such pitfalls and provide recommendations for future practice. Findings are illustrated by topical examples and also based on a survey of recent clinical trial publications in four major journals. Specific issues are: should plots go up or down (we recommend up), how far in time to extend the plot, showing the extent of follow-up, displaying statistical uncertainty by including SEs or CIS, and exercising caution when interpreting the shape of plots and the time-pattern of treatment difference.

[Duration of oral anticoagulant therapy in venous thromboembolism]

PURPOSE

The optimal duration of oral anticoagulant therapy after a first episode of venous thromboembolism is still a matter of debate. It is essential to balance the desired effect of the anticoagulants in reducing recurrences against the risk of major bleeding. The aims of this paper are to describe the current concepts in this field.

CURRENT KNOWLEDGE AND KEY POINTS

Recent data, based on randomised controlled trials, suggest that it is necessary to tailor the duration of anticoagulation individually according to the topography of venous thromboembolism and the presence of risk factors. A 6-week treatment for patients with isolated calf vein thrombosis is sufficient. For proximal thrombosis and/or pulmonary embolism, a short anticoagulant course seems sufficient in patients with temporary risk factors (3 months) and a longer anticoagulant course (6 months at least) is recommended for cases with permanent risk factors or idiopathic venous thromboembolism. The inherited or acquired hypercoagulable states can be divided into those that are common and associated with a modest risk of recurrence (i.e., isolated factor V Leiden or G20210A prothrombin gene) and those that are uncommon but associated with a high risk of recurrence (i.e., antithrombin, protein C or S deficiencies and anticardiolipin antibodies). Thus, the presence of one of these last abnormalities favours more prolonged anticoagulant therapy.

FUTURE PROSPECTS AND PROJECTS

1) For patients at high risk of recurrence, there is a paucity of evidence-based medicine, particularly for patients with biological thrombophilia, and randomised controlled trials in this population are required. An assessment of low- or fixed-dose oral anticoagulation is also necessary in order to reduce the bleeding risk. 2) It is not always possible to precisely determine the optimal duration with the available data. We have performed a meta-analysis on summary data which suggests that a long course of oral anticoagulant therapy is superior to a short course. An individual meta-analytic approach is needed to draw more precise conclusions on an interesting and important clinical topic and we propose to perform this analysis in a international collaborative group.

Optimal duration of anticoagulant therapy after an episode of venous thromboembolism

The optimal duration of oral anticoagulant therapy after a first episode of venous thromboembolism (VTE) is still a matter of debate. It is essential to balance the desired effect of the anticoagulants in reducing recurrences against the risk of major bleeding. The aims of this paper are to describe the current concepts in this field. Recent data, based on randomized controlled trials, suggest that it is necessary to tailor the duration of anticoagulation individually according to the topography of VTE and the presence of risk factors. A 6-week treatment for patients with isolated calf vein thrombosis is sufficient. For proximal thrombosis and/or pulmonary embolism, a short anticoagulant course is sufficient in patients with temporary risk factors (3 months), and a longer anticoagulant course (6 months at least) is recommended for cases with permanent risk factors or idiopathic VTE. For these high-risk of recurrence patients, an assessment of low- or fixed-dose oral anticoagulation is necessary in order to reduce the bleeding risk. It is not possible to precisely determine the optimal duration with the available data. We have already performed a meta-analysis on summary data that suggests a long course of oral anticoagulant therapy is superior to a short course. An individual meta-analytic approach is needed to draw more precise conclusions on an interesting and important clinical topic.