Sixteen studies were included (n=1,740): 2 RCTs (n=223), 2 prospective (n=225) and 7 retrospective (n=446) cohort studies, and 5 subgroup analyses comprising 2 RCTs (n=788) and 3 cohort studies (one prospective, n=16 and two retrospective, n=42).
The RCTs used intention-to-treat analysis. One failed to achieve the planned power and was terminated early. One reported assessor blinding. Many studies had questionable applicability: both RCTs excluded many high-risk patients and 6 of the 16 studies included participants who did not have definite APS. Most studies classified patients by the target INR rather than the achieved INR. Only one study reported venous and arterial events separately.
There was marked heterogeneity between the studies with respect to the participants, outcomes, interventions, analysis and findings.
All patients in the 2 RCTs had definite APS. Neither RCT found any benefit from high-intensity anticoagulation (target INR 3.0 to 4.0) compared with standard anticoagulation (target INR 2.0 to 3.0) in preventing recurrent thrombosis. However, patients randomised to receive high-intensity therapy frequently failed to achieve the target INR (43% of the time in one RCT).
In 7 of the 9 cohort studies all patients had definite APS. With the exception of one prospective cohort study (n=67), thrombosis recurrence rates were high among untreated patients (19 to 29% per year). Among treated patients there were generally fewer events among those treated with high-intensity coagulation (target INR 3.0 to 4.0) than among those treated with low-intensity coagulation (target INR 2.0 to 3.0) or aspirin. Patients with arterial events were at higher risk of recurrence than those with venous events in all therapeutic groups, in the single study that reported this outcome (n=61).
In the 5 subgroup analyses, only one analysis was restricted to participants with definite APS. The frequency of events in these studies was lower than in the cohort studies in both treated and untreated groups. In one RCT subgroup (n=720), among patients presenting with stroke there was no significant difference in the risk of recurrent thrombosis between those with a single positive aPL test who were treated with aspirin and those treated with warfarin (median INR 1.9). Rates of recurrent events among both treated and untreated patients were lower than in the cohort studies.
Bleeding (8 studies).
Rates of major bleeding varied widely, ranging from 0.57 to 10% per year. Seventy-four per cent of bleeding episodes occurred in patients with an INR ≥3.0 (where recorded). The proportion of bleeds that were major was 20% in patients receiving standard therapy and 29% in those receiving high-intensity therapy. The relative proportions of major and minor bleeding were similar for patients treated with standard- or high-intensity anticoagulation.
Mortality due to thrombosis and bleeding (4 studies).
Eighteen deaths were reported to be directly related to recurrent thromboses (12 arterial, 5 venous, one multiple) and one due to bleeding. Ten patients in one study died as a result of the presenting thrombosis.
Treatment and INR at time of thrombosis (13 studies).
Among patients receiving warfarin who experienced recurrent thrombosis (n=49), 86% of total events (n=42) occurred while the INR was <3.0. Fifteen per cent of events (n=27), mostly arterial, occurred in patients taking aspirin and 57% (n=104) occurred when patients were not taking any anticoagulant or anti-aggregant drug.