Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Review - Intervention

Antiplatelet and anticoagulant agents for secondary prevention of stroke and other thromboembolic events in people with antiphospholipid syndrome

Authors' declarations of interest

Version published: 12 October 2020 Version history

https://doi.org/10.1002/14651858.CD012169.pub3
Collapse all Expand all

Background

Antiphospholipid syndrome (APS) is a systemic autoimmune disease characterized by arterial or venous thrombosis (or both), and/or pregnancy morbidity in association with the presence of antiphospholipid antibodies. The prevalence of APS is estimated at 40 to 50 cases per 100,000 people. The most common sites of thrombosis are cerebral arteries and deep veins of the lower limbs. People with a definite APS diagnosis have an increased lifetime risk of recurrent thrombotic events.

Objectives

To assess the effects of antiplatelet (AP) or anticoagulant agents, or both, for the secondary prevention of recurrent thrombosis, particularly ischemic stroke, in people with APS.

Search methods

We last searched the MEDLINE, Embase, CENTRAL, Cochrane Stroke Group Trials Register, and ongoing trials registers on 22 November 2019. We checked reference lists of included studies, systematic reviews, and practice guidelines. We also contacted experts in the field.

Selection criteria

We included randomized controlled trials (RCTs) that evaluated any anticoagulant or AP agent, or both, in the secondary prevention of thrombosis in people with APS, according to the criteria valid when the study took place. We did not include studies specifically addressing women with obstetrical APS.

Data collection and analysis

Pairs of review authors independently worked on each step of the review, following Cochrane methods. We summarized the evidence using the GRADE approach.

Main results

We identified eight studies including 811 participants that compared different AP or anticoagulant agents.

NOAC (non‐VKA oral anticoagulant: rivaroxaban 15 or 20 mg/d) versus standard‐dose VKA (vitamin K antagonist: warfarin at moderate International Normalized Ratio [INR] ‐ 2.5) or adjusted [INR 2.0‐3.0] dose):

In three studies there were no differences in any thromboembolic event (including death) and major bleeding (moderate‐certainty evidence), but an increased risk of stroke (risk ratio [RR] 14.13, 95% confidence interval [CI] 1.87 to 106.8; moderate‐certainty evidence). One of the studies reported a small benefit of rivaroxaban in terms of quality of life at 180 days measured as health state on Visual Analogue Scale (mean difference [MD] 7 mm, 95% CI 2.01 to 11.99; low‐certainty evidence), but not measured as health utility on a scale from 0 to 1 (MD 0.04, 95% CI ‐0.02 to 0.10; low‐certainty evidence).

High‐dose VKA (warfarin with a target INR of 3.1 to 4.0 [mean 3.3] or 3.5 [mean 3.2]) versus standard‐dose VKA (warfarin with a target INR of 2.0 to 3.0 [mean 2.3] or 2.5 [mean 2.5]):

In two studies there were no differences in the rates of thrombotic events and major bleeding (RR 2.22, 95% CI 0.79 to 6.23, low‐certainty evidence), but an increased risk of minor bleeding in one study during a mean of 3.4 years (standard deviation [SD] 1.2) of follow‐up (RR 2.55, 95% CI 1.07 to 6.07). In both trials there was evidence of a higher risk of any bleeding (hazard ratio [HR] 2.03 95% CI 1.12 to 3.68; low‐certainty evidence) in the high‐dose VKA group, and for this outcome (any bleeding) the incidence is not different, only the time to event is showing an effect.

Standard‐dose VKA plus a single AP agent (warfarin at a target INR of 2.0 to 3.0 plus aspirin 100 mg/d) versus standard‐dose VKA (warfarin at a target INR of 2.0 to 3.0):

One high‐risk‐of‐bias study showed an increased risk of any thromboembolic event with combined treatment (RR 2.14, 95% CI 1.04 to 4.43; low‐certainty evidence) and reported on major bleeding with five cases in the combined treatment group and one case in the standard‐dose VKA treatment group, resulting in RR 7.42 (95% CI 0.91 to 60.7; low‐certainty evidence) and no differences for secondary outcomes (very low‐ to low‐certainty evidence).

Single/dual AP agent and standard‐dose VKA (pooled results):

Two high‐risk‐of‐bias studies compared a combination of AP and VKA (aspirin 100 mg/d plus warfarin or unspecified VKA at a target INR of 2.0 to 3.0 or 2.0 to 2.5) with a single AP agent (aspirin 100 mg/d), but did not provide any conclusive evidence regarding the effects of those drugs in people with APS (very low‐certainty evidence).

One of the above‐mentioned studies was a three‐armed study that compared a combination of AP and VKA (aspirin 100 mg/d plus warfarin at a target INR of 2.0 to 2.5) with dual AP therapy (aspirin 100 mg/d plus cilostazol 200 mg/d) and dual AP therapy (aspirin 100 mg/d plus cilostazol 200 mg/d) versus a single AP treatment (aspirin 100 mg/d). This study reported on stroke (very low‐certainty evidence) but did not report on any thromboembolic events, major bleeding, or any secondary outcomes.

We identified two ongoing studies and three studies are awaiting classification.

Authors' conclusions

The evidence identified indicates that NOACs compared with standard‐dose VKAs may increase the risk of stroke and do not appear to alter the risk of other outcomes (moderate‐certainty evidence). Using high‐dose VKA versus standard‐dose VKA did not alter the risk of any thromboembolic event or major bleeding but may increase the risk of any form of bleeding (low‐certainty evidence). Standard‐dose VKA combined with an AP agent compared with standard‐dose VKA alone may increase the risk of any thromboembolic event and does not appear to alter the risk of major bleeding or other outcomes (low‐certainty evidence).

The evidence is very uncertain about the benefit or harm of using standard‐dose VKA plus AP agents versus single or dual AP therapy, or dual versus single AP therapy, for the secondary prevention of recurrent thrombosis in people with APS (very low‐certainty evidence).

PICOs

Population
Intervention
Comparison
Outcome

The PICO model is widely used and taught in evidence-based health care as a strategy for formulating questions and search strategies and for characterizing clinical studies or meta-analyses. PICO stands for four different potential components of a clinical question: Patient, Population or Problem; Intervention; Comparison; Outcome.

See more on using PICO in the Cochrane Handbook.

Anticoagulant drugs or antiplatelet drugs, or both, for reducing the risk of blood clots and stroke in people with antiphospholipid syndrome

Review question

This review aimed to find out which type of treatment works best for preventing stroke and other blood clotting (thrombotic) events in people with antiphospholipid syndrome (APS).

Background

APS is a disease where the immune system produces antibodies against the person's own cells. The presence of such antibodies increases the risk of developing blood clots (thrombosis) in blood vessels (veins or arteries, or both) and can cause pregnancy‐related complications such as miscarriage. Blood clots in arteries can cause strokes. Two types of drugs are commonly used to prevent blood clots in people with APS: anticoagulants such as warfarin (which interfere with proteins that form blood clots) and antiplatelet drugs such as aspirin (which prevent cells called platelets from sticking together).

Study characteristics

The evidence is current to 22 November 2019. We looked for studies that randomly allocated people with APS to different treatments, including anticoagulants, antiplatelet drugs, or both. We identified eight studies involving 811 participants, with the average age between 36 and 50 years. Three studies compared rivaroxaban (a new oral anticoagulant) with standard‐dose warfarin (a vitamin K antagonist – VKA). Two studies compared two intensities of VKA anticoagulation: high versus standard. One study compared standard‐dose VKA plus aspirin versus standard‐dose VKA alone. The remaining studies compared different combinations of anticoagulants or antiplatelet drugs (e.g. aspirin). One study compared dual antiplatelet drugs versus single antiplatelet drugs. Interventions lasted from 180 days to a median of 58.4 months.

Key results

In three studies comparing rivaroxaban (NOAC) with standard‐dose VKA there were similar proportions of participants with blood clots, bleeding, or death (moderate‐certainty evidence), but there were more participants with stroke in the NOAC group than in the VKA group (moderate‐certainty evidence). In the two studies comparing high‐dose versus a standard‐dose VKA, similar proportions of participants had blood clotting and major bleeding problems (low‐certainty evidence), but participants in the high‐dose group had a greater risk of minor and any bleeding problems (low‐certainty evidence). A study comparing standard VKA plus an antiplatelet drug with standard‐dose VKA observed more participants with any thromboembolic event with combined treatment (low‐certainty evidence) and similar proportions of participants for other outcomes (very low‐ to low‐certainty evidence).

The remaining studies comparing different combinations of antiplatelet drug and/or VKA were inconclusive regarding benefits or harms (low‐ or very low‐certainty evidence).

We identified two ongoing studies and three studies are awaiting classification.

Certainty of the evidence

Most of the results were imprecise and did not clearly indicate benefit or harm.

Authors' conclusions

Implications for practice

The evidence identified in this systematic review indicates that NOAC compared with standard‐dose VKA probably increases risk of stroke and likely does not alter the risk of other adverse outcomes, such as any thromboembolic events, major bleeding, mortality, clinically‐relevant non‐major bleeding (moderate‐certainty evidence). The evidence suggests that NOAC compared with standard‐dose VKA may result in little to no difference in quality of life (low‐certainty evidence).

The evidence suggests that using high‐dose versus standard‐dose VKA may not alter the risk of any thromboembolic events, major bleeding, death, and stroke and may increase the risk of any bleeding (low‐certainty evidence).

The evidence suggests that standard‐dose VKA combined with antiplatelet agent compared with standard‐dose VKA only may increase the risk of any thromboembolic event and may not alter the risk of major bleeding, stroke, and minor bleeding (low‐certainty evidence). The evidence is very uncertain about the effect of combined VKA and antiplatelet on the risk of death (very low‐certainty evidence).

The evidence is very uncertain about the benefit or harm of using standard‐dose VKA plus antiplatelet versus single or dual antiplatelet therapy, or dual versus single antiplatelet therapy, for the secondary prevention of recurrent thrombosis in people with APS (very low‐certainty evidence).

Implications for research

Future research should be adequately powered and ensure proper adherence to treatment to assess the effects of the intervention on clinically‐important outcomes in people with APS, enabling meaningful conclusions regarding the effects of antiplatelet and anticoagulant agents and their intensity. The thrombotic risk in APS depends on many factors, including the type and number of positive aPL antibodies. Therefore, there is a need for studies with adequate patient stratification as the response to treatment might be different in people with high‐ and low‐risk antibody profile or people with previous arterial or venous thrombotic events. There is a special need to evaluate the efficacy and safety of other NOAC agents (i.e. dabigatran, apixaban, edoxaban) versus standard care for treating APS.

Summary of findings

Open in table viewer
Summary of findings 1. NOAC (rivaroxaban) versus standard‐dose VKA

NOAC (rivaroxaban) versus standard‐dose VKA

Patient or population: people with antiphospholipid syndrome and a history of stroke and or thromboembolic events
Setting: specialists centres
Intervention: NOAC (rivaroxaban)
Comparison: standard‐dose VKA

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with standard‐dose VKA

Risk with NOAC

Any thromboembolic event
Follow‐up: range 7 months to 35.4 months

Study population

RR 4.08
(0.48 to 34.79)

425
(3 RCTs)

⊕⊕⊕⊝
Moderateb

28 per 1000

115 per 1000
(14 to 975)

Major bleeding

Follow‐up: range 7 months to 35.4 months

Study population

RR 1.10
(0.45 to 2.68)

425
(3 RCTs)

⊕⊕⊕⊝
Moderateb

42 per 1000

47 per 1000
(19 to 113)

All‐cause mortalitya
Follow‐up: range 7 months to 35.4 months

Study population

RR 1.45
(0.44 to 4.78)

425
(3 RCTs)

⊕⊕⊕⊝
Moderateb

19 per 1000

28 per 1000
(9 to 91)

Strokea
Follow‐up: range 7 months to 35.4 months

Study population

RR 14.13
(1.87 to 106.81)

425
(3 RCTs)

⊕⊕⊕⊝
Moderatec

0 per 1000

0 per 1000

Mean quality of life ‐ health utility

Follow‐up: 180 days

MD 0.04 higher
(0.02 lower to 0.1 higher)

‐

111
(1 RCT)

⊕⊕⊝⊝
Lowd,e

Mean quality of life ‐ health state

Follow‐up: 180 days

MD 7 higher
(2.01 higher to 11.99 higher)

‐

112
(1 RCT)

⊕⊕⊝⊝
Lowd,e

Clinically relevant non‐major bleeding
Follow‐up: range 7 months to 35.4 months

Study population

RR 1.70
(0.69 to 4.19)

302
(2 RCTs)

⊕⊕⊕⊝
Moderatea

47 per 1000

80 per 1000
(33 to 197)

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference; RCT: randomized clinical trials; RR: risk ratio; VKA: vitamin K antagonists

GRADE Working Group grades of evidence.
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

aAll cause mortality and stroke are shown in the table, however other types of thromboembolic events (TIA, venous thromboembolism, MI) were also considered and are presented in text.
bDowngraded by one level due to imprecision: wide 95% CI including both benefit and harm, low number of events.
cDowngraded by one level due to imprecision: wide 95% CI, low number of events.
dDowngraded by one level due to within‐study risk of bias: patients, personnel and outcome assessors were not blinded.
eDowngraded by one level due to imprecision: wide 95% CI.

Open in table viewer
Summary of findings 2. High‐dose VKA versus standard‐dose VKA

High‐dose VKA versus standard‐dose VKA

Patient or population: people with antiphospholipid syndrome and a history of stroke or thromboembolic events
Setting: specialist centres
Intervention: high‐dose VKA
Comparison: standard‐dose VKA

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Assumed risk

Risk with standard‐dose VKA

Risk with‐high dose VKA

Any thromboembolic event

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

Study population

RR 2.22
(0.79 to 6.23)

223
(2 RCTs)

⊕⊕⊝⊝
Lowc,d

44 per 1000

98 per 1000
(35 to 275)

Major bleeding

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

Study population

RR 0.74
(0.24 to 2.25)

223
(2 RCTs)

⊕⊕⊝⊝
Lowc,d

62 per 1000

46 per 1000
(15 to 140)

All‐cause mortalitya

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

Study population

RR 1.53
(0.27 to 8.79)

223
(2 RCTs)

⊕⊕⊝⊝
Lowc,d

18 per 1000

28 per 1000
(5 to 159)

Strokea

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

Study population

RR 1.37
(0.26 to 7.12)

223
(2 RCTs)

⊕⊕⊝⊝
Lowc,d

18 per 1000

25 per 1000
(5 to 129)

Any bleedingb

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

Study population

RR 1.56

(0.93 to 2.62)

HR 2.03e
(1.12 to 3.68)

223

(2 RCTs)

⊕⊕⊝⊝
Lowc,d

168 per 1000

263 per 1000
(157 to 441)

Adverse events

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

See footnotef

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; HR: hazard ratio; RCT: randomized clinical trials; RR: risk ratio; SD: standard deviation; VKA: vitamin K antagonists

GRADE Working Group grades of evidence.
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

aAll cause mortality and stroke are shown in the table, however other types of thromboembolic events (TIA, venous thromboembolism, MI) were also considered and are presented in text.
bOur review has an outcome 'any bleeding that does not meet criteria for major bleeding', however both studies reported any bleeding, therefore we decided to present it in the 'Summary of findings' table for the information on harm.
cDowngraded by one level due to within‐study risk of bias: some issues concerning incomplete outcome reporting and selective outcome reporting; WAPS 2005 was seriously underpowered as was terminated early due to poor recruitment.
dDowngraded by one level due to imprecision: wide 95% CI including both benefit and harm, low number of events.
eThe results were not statistically significant when analyzed by RR; however when the time to event (HR) was taken into account there was a statistically significant difference between treatment groups.
fOnly one of the two included studies reported adverse events other than bleeding as outcomes and these adverse events lead to withdrawal from the study (WAPS 2005); these were essential thrombocythemia in one participant and headache in one participant, but the study did not indicate the group in which those participants were included.

Open in table viewer
Summary of findings 3. Standard‐dose VKA plus single antiplatelet agent versus standard‐dose VKA

Standard‐dose VKA plus single antiplatelet agent versus standard‐dose VKA

Patient or population: people with antiphospholipid syndrome and a history of stroke or thromboembolic events
Setting: specialist centres
Intervention: standard‐dose VKA plus single antiplatelet agent
Comparison: standard‐dose VKA

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Assumed risk

Risk with standard‐dose VKA

Risk with standard‐dose VKA plus single antiplatelet agent

Any thromboembolic event

Follow‐up: median 58.4 months (IQR 0.9; 38.9), in VKA+AP group follow‐up and 51.6 (IQR 0.9; 48.8) in VKA group

Study population

RR 2.14
(1.04 to 4.43)

82
(1 RCTs)

⊕⊕⊝⊝
Lowc,d

184 per 1000

394 per 1000
(192 to 816)

Major bleeding

Follow‐up: median 58.4 months (IQR 0.9; 38.9), in VKA+AP group follow‐up and 51.6 (IQR 0.9; 48.8) in VKA group

Study population

RR 7.42
(0.91 to 60.70)

82
(1 RCTs)

⊕⊕⊝⊝
Lowc,e

20 per 1000

149 per 1000
(19 to 1214)

All‐cause mortalitya

Follow‐up: median 58.4 months (IQR 0.9; 38.9), in VKA+AP group follow‐up and 51.6 (IQR 0.9; 48.8) in VKA group

Study population

RR 0.49
(0.02 to 11.68)

82
(1 RCTs)

⊕⊝⊝⊝
Very lowc,f

20 per 1000

10 per 1000
(1 to 234)

Strokea

Follow‐up: median 58.4 months (IQR 0.9; 38.9), in VKA+AP group follow‐up and 51.6 (IQR 0.9; 48.8) in VKA group

Study population

RR 4.45
(0.48 to 41.00)

82
(1 RCTs)

⊕⊕⊝⊝
Lowc,e

20 per 1000

89 per 1000
(10 to 820)

Minor bleedingb

Follow‐up: median 58.4 months (IQR 0.9; 38.9), in VKA+AP group follow‐up and 51.6 (IQR 0.9; 48.8) in VKA group

Study population

RR 1.29

(0.86 to 1.94)

82
(1 RCTs)

⊕⊕⊝⊝
Lowc,e

469 per 1000

606 per 1000
(404 to 910)

Adverse events

Not reported

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; IQR: interquartile range; RCT: randomized clinical trials; RR: risk ratio; VKA: vitamin K antagonists

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect,
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different,
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.

aAll cause mortality and stroke are shown in the table, however other types of thromboembolic events (TIA, venous thromboembolism, MI) were also considered and are presented in text.
bOur review has an outcome 'any bleeding that does not meet criteria for major bleeding', however the study reported only major and minor bleeding, therefore we decided to present both outcomes in the 'Summary of findings' table for the information on harm.
cDowngraded by one level due to within‐study risk of bias: no allocation concealment.
dDowngraded by one level due to imprecision: wide 95% CI, low number of events.
eDowngraded by one level due to imprecision: wide 95% CI including both benefit and harm, low number of events.
fDowngraded by two levels due to imprecision: very wide 95% CI including both benefit and harm, single event in one group only.

Open in table viewer
Summary of findings 4. Standard‐dose VKA plus single antiplatelet agent versus single antiplatelet agent

Standard‐dose VKA plus single antiplatelet agent versus single antiplatelet agent

Patient or population: people with antiphospholipid syndrome, with previous stroke
Setting: Japan, 1 centre or unknown number of centres
Intervention: standard‐dose VKA plus single antiplatelet agent
Comparison: single antiplatelet agent

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with single antiplatelet agent

Risk with
standard‐dose VKA plus single antiplatelet agent

Major bleeding (minor cerebral hemorrhage)

Follow‐up: mean 3.9 years (SD 2.0)

Study population

RR 0.40
(0.02 to 8.78)

20
(1 RCT)

⊕⊝⊝⊝
Very lowa,b

—

91 per 1000

37 per 1000
(2 to 799)

Stroke

Follow‐up: 1 year

Study population

RR 0.14 (0.01 to 2.60)

40
(1 RCT)

⊕⊝⊝⊝
Very lowa,c

1 small study published only as conference abstracts; single antiplatelet drug group discontinued after 1 year for humanitarian considerations

150 per 1000

21 per 1000
(2 to 390)

Any bleeding that does not meet criteria for major bleeding ‐ Gastrointestinal bleeding (no definition)

Follow‐up: mean 3.9 years (SD 2.0)

Study population

RR 3.60 (0.16 to 79.01)

20
(1 RCT)

⊕⊝⊝⊝
Very lowa,b

—

0 per 1000d

0 per 1000d

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; RCT: randomized clinical trials; RR: risk ratio; SD: standard deviation; VKA: vitamin K antagonists

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect,
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different,
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.

aDowngraded by two levels due to imprecision: very low number of events and sample size, very wide 95% CI including both benefit and harm.
bDowngraded by one level due to within‐study risk of bias: insufficient information regarding randomization, concealment, blinding, selective outcome reporting.
cDowngraded by one level due to within‐study risk of bias: insufficient information regarding all aspects of study design, no clear sequence generation, allocation concealment, blinding, completeness of outcome data and selective outcome reporting.
dNumbers could not be calculated as number of events in control group was 0.

Open in table viewer
Summary of findings 5. Standard‐dose VKA plus single antiplatelet agent versus dual antiplatelet agent

Standard‐dose VKA plus single antiplatelet agent versus dual antiplatelet agent

Patient or population: people with antiphospholipid syndrome, with previous stroke
Setting: Japan, 1 centre
Intervention: standard‐dose VKA plus single antiplatelet agent
Comparison: dual antiplatelet agent

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with dual antiplatelet agent

Risk with
standard‐dose VKA plus single antiplatelet agent

Stroke

Follow‐up: 3 years

Study population

RR 5.00 (0.26 to 98.00)

40
(1 RCT)

⊕⊝⊝⊝
Very lowa,b

1 small study published only as conference abstracts; single antiplatelet drug group discontinued after 1 year for humanitarian considerations

0 per 1000c

0 per 1000c

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; RCT: randomized clinical trials; RR: risk ratio; VKA: vitamin K antagonists

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect,
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different,
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.

aDowngraded by two levels due to imprecision: very low number of events and sample size, very wide 95% CI including both benefit and harm.
bDowngraded by one level due to within‐study risk of bias: insufficient information regarding all aspects of study design, no clear sequence generation, allocation concealment, blinding, completeness of outcome data and selective outcome reporting.
cNumbers could not be calculated as number of events in control group was 0.

Open in table viewer
Summary of findings 6. Dual antiplatelet agent versus single antiplatelet agent

Dual antiplatelet agent versus single antiplatelet agent

Patient or population: people with antiphospholipid syndrome, with previous stroke
Setting: Japan, 1 centre
Intervention: dual antiplatelet agent
Comparison: single antiplatelet agent

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with dual antiplatelet agent

Risk with
single antiplatelet agent

Stroke

Follow‐up: 1 year

Study population

RR 0.14 (0.01 to 2.6)

40
(1 RCT)

⊕⊝⊝⊝
Very lowa,b

1 small study published only as conference abstracts; single antiplatelet drug group discontinued after 1 year for humanitarian considerations

150 per 1000

21 per 1000
(2 to 390)

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; RCT: randomized clinical trials; RR: risk ratio

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect,
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different,
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.

aDowngraded by two levels due to imprecision: very low number of events and sample size, very wide 95% CI including both benefit and harm.
bDowngraded by one level due to within‐study risk of bias: insufficient information regarding all aspects of study design, no clear sequence generation, allocation concealment, blinding, completeness of outcome data and selective outcome reporting.

Background

Description of the condition

Antiphospholipid syndrome (APS) is an autoimmune condition where the presence of antiphospholipid (aPL) antibodies is associated with recurrent thrombosis (both arterial and venous), pregnancy morbidity, or both. The pathogenesis of APS involves the activation of monocytes, platelets, endothelial cells and complements, which induce thrombosis (Chighizola 2015; Giannakopoulos 2007; Giannakopoulos 2013). Primary APS is diagnosed in 53.1% of cases, while 36.2% of cases are secondary APS (associated with other autoimmune diseases, especially with systemic lupus erythematosus, or SLE) (Cervera 2002). In the general population, estimates of the prevalence of aPL antibodies range from 1% to 5% of otherwise healthy people (Petri 2000), up to 10% (George 2009). The prevalence is higher in people with rheumatoid arthritis (16%) and SLE (30% to 40%) (George 2009). According to the AntiPhospholipid Syndrome Alliance for Clinical Trials and InternatiOnal Networking (APS ACTION) Group data, aPL prevalence in people with thrombotic events was 6% in women with pregnancy morbidity and 13% in people with stroke (Andreoli 2013).

The prevalence of APS is estimated at 40 to 50 cases per 100,000 people, and the incidence is about five new cases per 100,000 people per year (Gomez‐Puerta 2014). The estimated association between aPL positivity and annual risk of thrombosis in people with no previous thrombosis is 0% to 4% (Erkan 2007), while in people with SLE the annual risk for thrombotic events is 2.5% to 3.8%. However, 4% to 21% of people with thrombosis are positive for aPL antibodies (Lim 2006).

The diagnosis of APS is based on the 2006 modified classification criteria, which include relatively specific and the most common clinical and laboratory findings: that is, vascular thrombosis, pregnancy morbidity, or both, with the presence of lupus anticoagulant (LA) and/or anticardiolipin antibodies (aCL) and/or anti‐beta2 glycoprotein‐I antibodies (anti‐β2GPI) in plasma in medium to high titers. Antibodies must be detected at least twice in a 12‐week period. To confirm the diagnosis of APS, one clinical and one laboratory criterion must be fulfilled (Miyakis 2006). The previous classification criteria for APS, established in Sapporo in 1999, did not include anti‐β2GPI antibodies and set the minimum time between two measurements at six weeks (Wilson 1999).

Associated with thrombosis, aPL antibodies are a heterogeneous group of antibodies found in people with APS. The presence of LA in plasma is the strongest risk factor for both venous and arterial thrombosis (Galli 2003).

Thrombosis in APS may affect both venous and arterial vessels, with the most common sites being deep veins of the lower limbs and cerebral arteries (Keeling 2012). In people diagnosed with APS, about 13% have had a stroke and 7% a transient ischemic attack (TIA) (Panichpisal 2012), whereas aPL antibodies were found in about 20% of people under 50 years of age diagnosed with stroke (Bushnell 2000). It is well established that the simultaneous presence of all three types of antibodies (LA, aCL, and anti‐β2GPI), the so called 'triple‐antibody positivity,' is associated with a significantly higher thrombotic risk than the combination of two antibodies ('double‐antibody positivity') or the presence of just one type of antibody ('single‐antibody positivity') (Iwaniec 2016; Pengo 2011; Pengo 2015). In a large cohort of unselected APS cases, the most frequently occurring clinical manifestations of APS were deep vein thrombosis, thrombocytopenia, livedo reticularis, and stroke, followed by pulmonary embolism, pregnancy loss, and transient ischemic attack (TIA) (Cervera 2002; Cervera 2009). TIA is defined as a condition with similar symptoms to a stroke, usually caused by a clot. However, the main difference between a stroke and TIA is that with TIA the blockage of the vessel is temporary and causes no permanent injury to the brain. According to the original definition, TIA symptoms should have resolved within 24 hours (Chatzikonstantinou 2013). According to the more recent definition ‐ a tissue‐based definition adopted by the American Heart Association and American Stroke Association (AHA/ASA) ‐ TIA is defined as a transient episode of neurologic dysfunction, which is caused by focal ischemia of the nervous system tissue (brain, spinal cord, retinal), but with no acute infarction (Easton 2009).

Other clinical manifestations comprise heart valve disease, pre‐eclampsia or eclampsia, premature birth, pulmonary hypertension, and leg ulcers (Ruiz‐Irastorza 2010). The most severe form of APS is catastrophic APS (CAPS), which occurs in less than 1% of people with APS and has a mortality rate of 30% (Cervera 2010).

Some studies have reported that people with arterial thrombosis are at higher risk of developing recurrence than those with venous thrombosis (Chighizola 2015). In a large European cohort study in which most participants had index venous thrombosis, recurrent arterial thromboses were the most common events (Cervera 2009). However, another study in people at high risk did not show that recurrent events depend on the index event (Chighizola 2015; Pengo 2010).

In a large cohort of unselected APS cases, the five‐year survival rate in people with APS was approximately 90% to 94% (Cervera 2002; Cervera 2009). The median age at disease diagnosis was 31 years, and most people were diagnosed between the age of 15 and 50 years. Taking into account the clinical manifestations of the disease, which include cerebrovascular events (13% of people with stroke and 7% of people with TIA) and the young age of disease diagnosis onset, APS may exert a strong impact with high socioeconomic costs (Chighizola 2015).

Several studies have linked increased risk for thrombosis in people with aPL antibodies with the presence of cardiovascular risk factors such as hypercholesterolemia, smoking, and hypertension (Erkan 2007; Matyja‐Bednarczyk 2014; Saraiva 2015). Therefore, careful assessment of the cardiovascular risk factors and their management in people with aPL is advised (Chighizola 2015).

Description of the intervention

There are no unconfounded randomized trials confirming the efficacy of either anticoagulants or antiplatelets compared with placebo or no treatment in the secondary prevention of thromboembolic events in APS. The thrombotic risk in definite APS is too high to allow using a placebo. Secondary thromboprophylaxis consists of antiplatelets (most commonly aspirin), anticoagulants (most commonly vitamin K antagonists (VKA): warfarin or acenocoumarol), or both (Espinosa 2015). People may receive aspirin or warfarin with the target international normalized ratio (INR) of 2.0 to 3.0, or heparin when VKAs are contraindicated. Non‐VKA antagonist oral anticoagulants (rivaroxaban, dabigatran, edoxaban, or apixaban) and antiplatelets (clopidogrel or prasugrel) can also be used alone or with aspirin. Where there are thrombotic complications at INR of 2.0 to 3.0, clinicians may modify the therapy to a higher target INR, combine two drugs, or, in high‐risk patients, even prescribe triple antithrombotic therapy (one anticoagulant such as warfarin plus two antiplatelet agents, such as aspirin and clopidogrel). However, this is associated with a higher risk of hemorrhage or major bleeding (Nalli 2014). Selecting the appropriate intensity of anticoagulation to balance the risk of recurrent thrombosis and the risk of bleeding in APS is a real challenge for clinicians. Retrospective studies have suggested high‐intensity (high‐dose) anticoagulation with warfarin (INR set > 3) to prevent recurrent thrombosis in people with APS (Lim 2006). However, randomized trials suggested adopting moderate‐intensity anticoagulation with an INR targeted at 2.5 (range 2.0 to 3.0) as the best choice available for secondary thrombosis prevention in people with APS (Crowther 2003; Finazzi 2005). On the other hand, a randomized, double‐blind study testing the risk of subsequent thromboembolic events associated with positive aPL in patients with non‐cardioembolic stroke failed to demonstrate the superiority of warfarin over aspirin therapy. Therefore, the study authors concluded that testing for aPL might not be warranted in all patients with ischemic stroke (APASS Investigators 2004).

Another important issue is the value of the INR at the time of recurrent thrombosis—studies have reported that in people who received VKA, most of the recurrent thrombotic events occurred in those with an INR below 3, while people in high‐dose anticoagulation groups were below an INR range for over 40% of time, which may have biased the results (Chighizola 2015; Ruiz‐Irastorza 2011). This advice of moderate‐intensity anticoagulation applied mainly to people with APS and venous thromboembolism; therefore, clinicians should exercise caution when adapting it to arterial thrombotic events, which are less frequent (Finazzi 2005; Ruiz‐Irastorza 2011). It would be important to look at these two anticoagulant regimens as separate interventions. There are also studies suggesting that hydroxychloroquine can have an antithrombotic effect due to its antiplatelet properties (Erkan 2014; Ruiz‐Irastorza 2011).

Intake of foods rich in vitamin K, as well as drugs or other substances that may enhance or inhibit the metabolism of the anticoagulant agent used, can influence the effectiveness and safety of VKA (Chighizola 2014). VKA treatment often requires lifestyle modification and regular dose adjustment based on measured INR values, with monitoring for bleeding. These measures are necessary because they help to assure treatment effectiveness and safety. However, they may have a strong impact on the quality of life of the patient, since they would require regular visits to a healthcare facility: they can be time‐consuming, may incur additional costs, and may interfere with daily activity to an extent that they are a considerable burden for some patients (Hasan 2015).

How the intervention might work

Aspirin is an antiplatelet agent, inhibiting cyclooxygenase 1 (COX‐1) in platelets, which in turn inhibits the production of thromboxane A2 (TXA2) (Warner 2011). Clopidogrel is also an antiplatelet drug, but it works by inhibiting P2Y receptors and impairs the activation of the glycoprotein (GP) IIb/III complex by fibrinogen (Wijeyeratne 2011). The mechanism of action of oral anticoagulant agents such as warfarin or acenocoumarol is antagonizing vitamin K and thus inhibiting the production of coagulation factors II, VII, IX, X, and C and S proteins (Ageno 2012). Heparin binds to antithrombin and then makes a complex with an activated factor X, inactivating it, which leads to the inhibition of blood coagulation (Hirsh 2001). Non‐vitamin K antagonist oral anticoagulants (NOACs) are the direct inhibitors of either factor IIa (thrombin) or factor Xa (Weitz 2016). Dabigatran etexilate strongly and reversibly inhibits thrombin and thus restrains the conversion of fibrinogen into fibrin, while rivaroxaban and apixaban directly inhibit active factor X (both free and thrombus‐bound), which breaks the coagulation cascade (Ageno 2012).

The thrombotic risk in people with APS is high. Therefore, treatment guidelines currently recommend providing people with an APS diagnosis and thrombosis with life‐long anticoagulants, antiplatelet therapy, or both, to prevent future arterial or venous thrombotic events (Ruiz‐Irastorza 2011). However, these treatments increase the risk of bleeding, especially in people receiving the combination therapy (Nalli 2014), and they can also be associated with other adverse events (Raschi 2016). In the new guidelines for the diagnosis and management of acute pulmonary embolism developed by the European Society of Cardiology (ESC) in collaboration with the European Respiratory Society (ERC) the treatment with NOACs in patients with APS is not recommended (Konstantinides 2020). Similarly in the new recommendations for the management of APS in adults developed by the European League Against Rheumatism (EULAR) treatment with NOACs in high‐risk APS patients and arterial events is not recommended due to higher risk of recurrent thrombosis (Tektonidou 2019a; Tektonidou 2019b).

Why it is important to do this review

No Cochrane Review has fully addressed the prevention of recurrent thrombosis in people with diagnosed APS. One recently published Cochrane Review focused on the effects of use of aspirin and/or heparin in women with persistent aPL antibodies and recurrent miscarriage (Hamulyak 2020), and a protocol registered by Cochrane Vascular considered using antiplatelet or anticoagulant agents to prevent recurrent peripheral vascular thrombosis in such patients (Islam 2016). A separate Cochrane Review focuses on primary prevention of thrombosis in people with aPL antibodies since they are different from those already diagnosed with APS (Bala 2018). However, none of these reviews addresses the issue of prevention of other types of thrombosis, such as stroke, in people with APS. Several randomized trials have examined the efficacy of using an antiplatelet drug (aspirin) or anticoagulant agents (such as warfarin) in people diagnosed with APS. A recent RCT of rivaroxaban versus warfarin in high‐risk APS patients was prematurely terminated due to an increased risk of recurrent arterial thromboembolic events in the rivaroxaban arm (Pengo 2018). Some clinical trials using other NOACs in such patients are ongoing with a modified protocol (ASTRO‐APS). Therefore, it is important to summarize the effects of those therapies in people with APS while awaiting the results of the ongoing studies (ASTRO‐APS; RISAPS).

Objectives

To assess the effects of antiplatelet or anticoagulant agents, or both, for the secondary prevention of recurrent thrombosis, particularly ischemic stroke, in people with APS.

Methods

Criteria for considering studies for this review

Types of studies

We included randomized controlled trials (RCTs) comparing participants allocated to one of two or more different treatment regimens. We also included RCTs comparing anticoagulant or antiplatelet treatments with placebo, usual care, or alternative treatment approaches.

Types of participants

People with APS, diagnosed according to the criteria valid when the study was carried out, such as the Sapporo or Sydney criteria (Miyakis 2006; Wilson 1999). We did not include studies specifically addressing women with recurrent miscarriages, as a separate Cochrane Review covers this topic (Hamulyak 2020).

Types of interventions

We included trials comparing any antiplatelet agents (e.g. aspirin, cilostazol, clopidogrel or prasugrel), any anticoagulant agents (e.g. VKA: acenocoumarol, warfarin; or NOAC: rivaroxaban, dabigatran, edoxaban, apixaban), or their combinations in any dose and mode of delivery versus no intervention/placebo or another antiplatelet/anticoagulant regimen.

Types of outcome measures

Primary outcomes

  • Any thromboembolic event, including death or any arterial or venous thrombosis

  • Major bleeding

We defined major bleeding according to the International Society on Thrombosis and Haemostasis (ISTH) criteria, as clinically overt bleeding with a confirmed decrease in the hemoglobin level of at least 2 g/dL; or transfusion, due to the occurrence of clinical symptoms, of at least two units of packed red cells, occurring at a critical site (intracranial, intraocular, intraspinal, intra‐articular, intramuscular with compartment syndrome, pericardial, retroperitoneal); or resulting in death (Schulman 2005). This definition does not consider any time restrictions.

Secondary outcomes

  • Each type of thromboembolic event analyzed separately (i.e. all‐cause mortality, stroke, TIA, venous thromboembolism, etc)

  • Quality of life measured with a validated questionnaire

  • Any bleeding that does not meet the criteria for major bleeding

  • Adverse event other than bleeding

We analyzed thromboembolic events as defined by the authors of the primary studies, especially with regard to TIA; this definition was originally time‐based (Advisory Council 1975), but updates later based it on tissues (Easton 2009). If possible, we planned to take this into account in the subgroup analysis.

We assessed all outcomes at the end of follow‐up.

Search methods for identification of studies

See the methods for the Cochrane Stroke Group Specialised register. We searched for trials in all languages and arranged for the translation or extraction of data of relevant articles where necessary.

Electronic searches

In addition to the Cochrane Stroke Group's trials register, we searched the following electronic databases (last search on 22 November 2019).

  • Cochrane Central Register of Controlled Trials (CENTRAL; 2017, Issue 2)

  • MEDLINE Ovid (from 1948)

  • Embase Ovid (from 1980)

The search strategies for these electronic databases are presented in Appendix 1, Appendix 2, and Appendix 3. The subject strategies used for the included databases were based on the search strategy for MEDLINE designed by the Cochrane Stroke Group’s Information Specialist (Appendix 2). These search strategies were combined with subject strategy adaptations of the highly‐sensitive search strategy designed by Cochrane for identifying RCTs and controlled clinical trials, as described in the Cochrane Handbook for Systematic Reviews of Interventions Chapter 6.4.11 (Higgins 2011).

We also searched the following ongoing trials registries (last search on 22 November 2019).

The search strategies for the trials registries are presented in Appendix 4.

Searching other resources

We checked reference lists of all included studies, systematic reviews, and practice guidelines relevant to the topic of the review. We contacted experts in the field and manufacturers of the original drugs to inquire about additional studies.

Data collection and analysis

Selection of studies

We used Covidence in the process of study selection. This software allows importing of search results, and facilitates independent screening by two review authors as well as comparing the results of screening, and extracting data from eligible studies. Pairs of review authors (MMB, MC‐L, WS, AP, MK, MJS, and trainee reviewers named in the Acknowledgements) independently screened titles and abstracts of the references obtained as a result of our searching activities, and excluded obviously irrelevant reports. We retrieved the full‐text articles for the remaining references, and pairs of review authors (MMB, MC‐L, WS, AP, MK, MJS and trainee reviewers) independently screened them and identified studies for inclusion, as well as identifying and recording reasons for exclusion of the ineligible studies. We resolved any disagreements through discussion or, if required, we consulted a third review author (AU, MMB, WS, MC‐L). We collated multiple reports of the same study so that each study, not each reference, was the unit of interest in the review. We recorded the selection process and completed a PRISMA flow diagram (Moher 2009).

Data extraction and management

We planned to use Covidence for data extraction, but the forms available in Covidence were not flexible enough to fit our extraction purposes; therefore, we decided to use Microsoft Excel 2013 spreadsheets. We extracted data on study settings, time frame and methods, population inclusion and exclusion criteria as well as population characteristics, details of interventions and co‐interventions, and details of outcomes and their definitions. Pairs of review authors (MMB, MC‐L, WS, AP, MK, MJS) independently extracted data from the included studies. We compared the extracted results and resolved any discrepancies by discussion. One review author (MMB) additionally checked all the data extracted.

Assessment of risk of bias in included studies

Pairs of review authors (MMB, MC‐L, WS, AP, MK, MJS) independently assessed risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions chapter 8.5 (Higgins 2011). We resolved any disagreements by discussion or by involving another review author (MMB, MC‐L, WS). We assessed the risk of bias according to the following domains.

  • Random sequence generation

  • Allocation concealment

  • Blinding of participants and personnel

  • Blinding of outcome assessment

  • Incomplete outcome data

  • Selective outcome reporting

  • Other bias

We graded the risk of bias for each domain as high, low, or unclear and provided information from the study report together with a justification for our judgment in the 'Risk of bias' tables. We judged trials as being at low risk of bias if they were at low risk of bias in all of the domains; we judged other cases as being at high risk of bias.

Measures of treatment effect

For binary outcomes we calculated the risk ratio (RR) with 95% confidence intervals (CIs); for continuous outcomes (quality of life) we planned to calculate the mean difference (MD) or standardized mean difference (SMD) (when different scales were used) with 95% CIs. Since only one study reported quality of life data, we presented the results as MD with 95% CI. For survival outcomes, such as hazard ratio (HR) for death, we used a generic inverse variance method for the meta‐analysis. In all analyses we planned to calculate pooled estimates using the random‐effects model (Der Simonian 1986), and we planned to conduct sensitivity analyses using the fixed‐effect model meta‐analyses (Greenland 1985; Mantel 1959). As we detected no heterogeneity in analyses where there was more than one study, we decided not to pursue the sensitivity analyses with the fixed‐effect model.

Unit of analysis issues

Regarding unit of analysis issues, we planned to follow the advice of the Cochrane Handbook for Systematic Reviews of Interventions chapter 9.3 (Higgins 2011). We anticipated that in most trials the unit of analysis would be individual participants. However, if there were cluster‐randomized trials, the unit of analysis would be the cluster. For cross‐over trials, we would include the first phase only in the analysis. We did not include any cluster‐randomized trials or cross‐over trials in the review.

Dealing with missing data

If data were missing, we attempted to contact the study authors to request them. If unsuccessful, after examining the data set, we had two strategies: first, we assumed the data to be missing at random and we planned to analyze the data as reported. Second, if missing at random was not adequate, we assumed the data to be missing in a systematic way and we planned to assign all missing participants as treatment failures. Finally, we planned to conduct sensitivity analyses on missing data to test the impact of these approaches.

Assessment of heterogeneity

We used the I2 statistic to measure heterogeneity among the trials in each analysis according to the Cochrane Handbook for Systematic Reviews of Interventions chapter 9.5.2 (Higgins 2011). We followed the interpretation in the Cochrane Handbook chapter 9.5.2 and considered I2 over 50% as representing substantial heterogeneity (Higgins 2011). In case of I2 over 50%, we examined the characteristics of individual trials contributing to the comparison to identify potential sources of heterogeneity.

Assessment of reporting biases

We planned to analyze reporting bias using funnel plots for all primary outcomes if there were a sufficient number of studies (at least 10 studies as recommended in the Cochrane Handbook chapter 10.4 (Higgins 2011).

Data synthesis

Where we considered studies to be clinically, methodologically, and statistically similar, we conducted a meta‐analysis by pooling the appropriate data using Review Manager (RevMan 2014). For binary outcome data we calculated RR or HR (if such data were available); for continuous outcomes (quality of life) we used MD. If necessary, we used the methods described by Parmar 1998 and Thierney 2007 to calculate data relevant to pool HR from the data available in the study (hazard rates, log rank P values, events, ratios, curve data, follow‐up information), using a spreadsheet in Microsoft Office Excel 2003 (freely available with the Thierney 2007 publication).

If pooling was not possible, we planned to summarize the results narratively, using text, figures, and tables.

Subgroup analysis and investigation of heterogeneity

If possible, we planned to explore heterogeneity by subgroup analyses taking into account single‐, double‐ and triple‐antibody positivity; lupus anticoagulant positivity versus other antibodies; presence versus absence of traditional cardiovascular risk factors; and type of index event (arterial versus venous).

We planned subgroup analyses for all outcomes only if there was a sufficient number of studies, i.e. for a single analysis at least six included studies with the information relevant for subgroup analysis (Deeks 2001).

Sensitivity analysis

We planned to conduct sensitivity analyses for missing data for each primary and secondary outcome for which they were missing, using worst‐best, best‐best, best‐worst and worst‐worst case scenarios. In addition, we planned to conduct sensitivity analyses according to low and high risk of bias and the amount of missing data (trials with no missing data versus trials with missing data).

GRADE and 'Summary of findings' table

We summarized the evidence in three 'Summary of findings' tables using GRADEpro (GRADEpro). In the development process, we followed the GRADE approach as outlined in the Cochrane Handbook for Systematic Reviews of Interventions chapter 11.5 (Higgins 2011). We planned to include the following outcomes in our 'Summary of findings' tables: any thromboembolic event, major bleeding, each type of thromboembolic event analyzed separately, quality of life, any bleeding that does not meet the criteria for major bleeding, and adverse events other than bleeding.

Results

Description of studies

See Characteristics of included studies; Characteristics of excluded studies; Characteristics of studies awaiting classification; Characteristics of ongoing studies.

None of the included studies compared antiplatelet or anticoagulant agent with placebo or no intervention; therefore, the effect of those agents compared to placebo or no intervention in secondary prevention of thromboembolic events in people with APS cannot be determined. The comparisons presented in the current review remain confounded because the interventions of unknown effect are compared with another intervention of unknown effect. No unequivocal evidence exists from randomized unconfounded studies that prove the efficacy of antiplatelets or anticoagulants in the secondary prevention of thromboembolic events in APS, but due to weaker evidence, these treatments are generally accepted and even included in recent clinical guidelines (Tektonidou 2019a; Tektonidou 2019b). Therefore, it is justified to perform the comparisons done in the current review.

Results of the search

We searched the electronic databases on 27 February 2017 and identified 12,180 references. We obtained an additional 13 records through searching other sources. Altogether, we screened 12,190 unique records by title and abstract after duplicates were removed (Figure 1). We downloaded full texts for 305 references. We disregarded 268 records because the study design (n = 219), population (n = 48), or intervention (n = 1) did not meet inclusion criteria. The remaining 37 references constituted five studies reported in 24 records (Crowther 2003; Okuma 2010; RAPS 2016; WAPS 2005; Yamazaki 2009); four ongoing trials, reported in eight references (ASTRO‐APS; JASPRES; Ordi‐Ros 2019; TRAPS 2016); and three studies awaiting classification, reported in five records (Kondratyeva 2010; Okuma 2014; Yamazaki 2007).

Figure 1
Primary study flow diagram.

Primary study flow diagram.

We updated our searches of the electronic databases and clinical trials registers on 22 November 2019, resulting in 459 records to screen. An additional two references were identified through searching other sources (Figure 2). Of 461 references, we excluded 448 based on title and abstract. Subsequently 13 records were checked in full text. We excluded five of them because of irrelevant study design (n = 4) and intervention (n = 1).

Figure 2
Flow chart of identification of randomized trials for inclusion in the 22 November 2019 update.RCT: randomized clinical trial

Flow chart of identification of randomized trials for inclusion in the 22 November 2019 update.

RCT: randomized clinical trial

Eight new references were included in the review. Furthermore, we received additional information regarding the Kondratyeva 2010, TRAPS 2016, and Ordi‐Ros 2019 studies, which resulted in the inclusion of these studies. We have not been able to obtain additional information on two of the awaiting assessment studies (Okuma 2014; Yamazaki 2007). As a result, in this update we included a total of eight studies reported in 37 records (Crowther 2003; Kondratyeva 2010; Okuma 2010; Ordi‐Ros 2019; RAPS 2016; TRAPS 2016; WAPS 2005; Yamazaki 2009), two studies (five records) were listed as ongoing studies (ASTRO‐APS; RISAPS), and three studies (three records) are awaiting classification (JASPRES; Okuma 2014; Yamazaki 2007).

Included studies

The study authors described the eight included studies as RCTs. Three were non‐inferiority trials (Ordi‐Ros 2019; RAPS 2016; TRAPS 2016), two were described as double blind (Crowther 2003; Okuma 2014), five were open label, but three had blinded endpoint adjudication for all outcomes (Ordi‐Ros 2019; TRAPS 2016; WAPS 2005) or for safety events (RAPS 2016), and one study was published only in conference abstracts (Yamazaki 2009). We present detailed information on each study in Characteristics of included studies.

Participants

All of the studies included people with diagnosed with APS. In total, 811 participants were randomized and 805 participants were analyzed for the outcomes relevant for this review.

The mean age of the participants in the study groups was 36.4 to 50 years; their median age was from 47 to 51 years. The criteria for inclusion of studies differed: Okuma 2010 specified and cited criteria for participants' diagnosis; Crowther 2003, Kondratyeva 2010 (information from authors), RAPS 2016, TRAPS 2016, and Ordi‐Ros 2019 followed the criteria published in Miyakis 2006; WAPS 2005 used criteria published in Wilson 1999; and Yamazaki 2009 did not report specific inclusion criteria.

Yamazaki 2009 and Okuma 2010 included only people with previous stroke, while in RAPS 2016 a previous arterial event was an exclusion criterion, as was recurrent venous thromboembolism (VTE) while on warfarin with INR in the therapeutic range 2.0 to 3.0. Crowther 2003, Kondratyeva 2010, WAPS 2005, TRAPS 2016, and Ordi‐Ros 2019 included people with both previous arterial and VTE. However in those five studies the majority of participants had previous venous events (from 64% to 75%).

Six studies reported the prevalence of SLE in study participants, which ranged from 13% to 36% (Crowther 2003; Okuma 2010; Ordi‐Ros 2019; RAPS 2016; TRAPS 2016; WAPS 2005).

Five studies reported details about antibodies present (Crowther 2003; Ordi‐Ros 2019; RAPS 2016; TRAPS 2016; WAPS 2005). Lupus anticoagulant was the only type of antibody detected for 43% of participants in Crowther 2003, 26% in WAPS 2005, 46% in RAPS 2016, and 35% in Ordi‐Ros 2019. Anticardiolipin antibodies were the sole type of antibody for 39% of participants in Crowther 2003, 18% in WAPS 2005, and for 3% of participants in RAPS 2016. Both types of antibodies were present for 18% of participants in Crowther 2003, and 56% of the participants in WAPS 2005. RAPS 2016 also reported the percentage of participants with beta2glycoprotein I antibodies only (4%) and more than one type of antibodies, both without (30%) and including triple positivity (16%). TRAPS 2016 included only participants with triple positivity for anticardiolipin, beta2glycoprotein I and lupus anticoagulant antibodies, while Ordi‐Ros 2019 reported 61% of participants with IgG anticardiolipin and beta2glycoprotein I antibodies, while 67% of participants with IgG/IgM anticardiolipin antibodies, 64% with IgG/IgM beta2glycoprotein I antibodies and 34% with IgG/IgM antiphosphatidylserine/prothrombin antibodies.

Three studies reported on cardiovascular risk factors: Okuma 2010 reported hypertension (59.6%), diabetes (20.2%), hyperlipidemia (20.2%), and atrial fibrillation (10.1%); TRAPS 2016 reported smoking (50%), hypertension (31%), diabetes (3%), dyslipidemia (23%), and other conditions with increased tendency for coagulation (15%); and Ordi‐Ros 2019 reported smoking (35%), dyslipidemia (39%), diabetes mellitus (6%), and hypertension (41%).

Location

Yamazaki 2009 and Okuma 2010 took place in Japan, Crowther 2003 in Canada, Kondratyeva 2010 in Russia, Ordi‐Ros 2019 in Spain, TRAPS 2016 in Italy, RAPS 2016 in the UK, and WAPS 2005 in Italy, Norway, Poland, Argentina, Czech Republic, and Slovak Republic.

Setting

Crowther 2003, WAPS 2005, Kondratyeva 2010, RAPS 2016, and TRAPS 2016 took place in specialist centres or clinics, Okuma 2010 in neurology departments of university hospitals, Ordi‐Ros 2019 at university hospitals, and Yamazaki 2009 did not specify the setting.

Interventions

Three studies compared NOACs with standard‐dose warfarin treatment (Ordi‐Ros 2019; RAPS 2016; TRAPS 2016). RAPS 2016 compared a standard‐dose warfarin treatment (mean INR 2.5) versus a NOAC: rivaroxaban 20 mg/d. The mean INR in the warfarin group was 2.7, and the mean time in therapeutic range at day 180 was 55%. TRAPS 2016 compared rivaroxaban 20 or 15 mg/d (depending on renal function) with standard adjusted‐dose warfarin with a target INR between 2 and 3. Ordi‐Ros 2019 compared rivaroxaban 20 or 15 mg/d (depending on renal function) with standard dose‐adjusted VKA with INR 2 to 3 or 3.1 to 4 in case of a history of recurrent thrombosis. In the VKA group INR values were within the therapeutic range for a mean of 56% (median 58% [IQR 46 to 70]).

Two studies compared treatment with two doses of warfarin: Crowther 2003 assessed high‐dose warfarin with a target INR of 3.1 to 4.0 and an average value of 3.3 versus a standard dose with a target INR of 2.0 to 3.0 and an average value of 2.3, and WAPS 2005 evaluated warfarin with a target of 3.5 and mean of 3.2 (range 3.0 to 4.5) versus standard antithrombotic therapy. Standard antithrombotic therapy included warfarin at a target 2.5 (range 2.0 to 3.0; mean 2.5) in participants with previous VTE, cardioembolic cerebral or peripheral ischemias, atrial fibrillation or rheumatic valve disease (95% of participants), or aspirin 100 mg/d in participants with non‐embolic arterial thrombosis (5% of participants). In addition, in Crowther 2003, 14% of participants in the high‐dose group and 10% participants in the standard‐dose group received aspirin, while in WAPS 2005 7% of participants in the high‐dose group and 5% of participants in the standard‐dose group received anticoagulation and aspirin according to the criteria of the treating physician. In Crowther 2003, participants in the high‐dose group were within the target INR for 40% of the time and below it for 43% of the time (but 86% of the time between 2.0 and 3.1), while in the standard‐dose groups those values were 71% and 19%, respectively.

Kondratyeva 2010 compared combined treatment with warfarin at a dosing regimen adjusted to a target INR of 2.0 to 3.0 and low‐dose aspirin (100 mg/d) with treatment with warfarin alone.

Okuma 2010 and Yamazaki 2009 compared a single antiplatelet agent with combinations of antiplatelet and anticoagulant agents (VKA) or dual antiplatelet therapy. They included a comparison of aspirin 100 mg/d with a combination of aspirin and anticoagulant agents (a non‐specified vitamin K antagonist) with a target INR of 2.0 to 3.0 (Okuma 2010), or a three‐arm comparison of aspirin 100 mg/d, aspirin 100 mg/d plus cilostazol 200 mg/d, and aspirin 100 mg/d plus warfarin (with a target INR of 2.0 to 2.5) (Yamazaki 2009). The mean INR in the combined treatment group in Okuma 2010 was 2.4, while Yamazaki 2009 did not report these data.

The duration of intervention varied among the studies and ranged from 180 days to a median of 58.4 months (IQR 0.9 to 38.9). Only one study reported an additional 30 days follow‐up without intervention (RAPS 2016). In Yamazaki 2009, one of the arms was stopped for "humanitarian" considerations (strokes revealed on MRI in three participants taking aspirin only). The TRAPS 2016 study was terminated prematurely due to excess thrombotic events in the rivaroxaban group.

Outcomes

The primary outcome in Yamazaki 2009 and Okuma 2010 was recurrent stroke. In Crowther 2003, recurrent thrombosis was the primary outcome. WAPS 2005 reported two co‐primary outcomes: vascular death or major thrombosis (arterial or venous) and vascular death or major thrombosis or major hemorrhage. The primary outcome in RAPS 2016 was a surrogate outcome: percent change in endogenous thrombin potential from randomization to day 42 of study, plus reported thromboembolism (VTE or any other thrombotic events) up to day 210 as a secondary outcome. Kondratyeva 2010 studied frequency of recurrent thromboses and TIA. The primary outcome in Ordi‐Ros 2019 was new thrombotic event, while TRAPS 2016 specified composite primary outcome including thromboembolic events, major bleeding, and vascular death.

In five trials, safety outcomes included major, minor, or any bleeding (Crowther 2003; Okuma 2010; Ordi‐Ros 2019; RAPS 2016; WAPS 2005), one study reported only on major bleeding (TRAPS 2016), and one study on major and minor bleeding (Kondratyeva 2010).

Study authors did not specify the secondary outcomes in Crowther 2003, Kondratyeva 2010, Yamazaki 2009 and Okuma 2010. Ordi‐Ros 2019 specified as secondary outcomes the time to thrombosis, type of thrombotic event, death from cardiovascular causes, and changes in selected biomarkers' levels. The secondary outcomes for efficacy in RAPS 2016 included other coagulation measures and quality of life (measured using the EQ‐5D‐5L questionnaire). TRAPS 2016 specified as secondary outcomes type of thrombotic events and mortality from all causes. The secondary endpoints in WAPS 2005 included combinations of different thrombotic events. Two trials specifically reported adverse events as outcomes (RAPS 2016; WAPS 2005).

Excluded studies

We excluded one study because it was irrelevant for the review population; see Characteristics of excluded studies.

Risk of bias in included studies

We presented details for each study in the Characteristics of included studies table. Figure 3 shows the overall risk of bias in each domain for studies in this review; Figure 4 shows risk of bias by trial.

Figure 3
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.Duplicate blinding domains are due to separate assessments for objective or objectively verified outcomes (Obj.) and subjective and self‐reported outcomes (Subj.). Blank sections are left because those studies did not assess subjective outcomes or patient self‐reported outcomes not objectively verified.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Duplicate blinding domains are due to separate assessments for objective or objectively verified outcomes (Obj.) and subjective and self‐reported outcomes (Subj.). Blank sections are left because those studies did not assess subjective outcomes or patient self‐reported outcomes not objectively verified.

Figure 4
Risk of bias summary: review authors' judgements about each risk of bias item for each included study. Duplicate blinding domains are due to separate assessments for objective or objectively verified outcomes (Obj.) and subjective and self‐reported outcomes (Subj.). Blank sections are left because those studies did not assess subjective outcomes or patient self‐reported outcomes not objectively verified.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study. Duplicate blinding domains are due to separate assessments for objective or objectively verified outcomes (Obj.) and subjective and self‐reported outcomes (Subj.). Blank sections are left because those studies did not assess subjective outcomes or patient self‐reported outcomes not objectively verified.

Seven of the included studies were published as full articles. One consisted of several conference abstracts, so data extracted from this study were very limited (Yamazaki 2009).

Only one study was at low risk of bias in all domains (Crowther 2003). One study was at low risk of bias in all domains for objective outcomes, but not for subjective ones (RAPS 2016).

Allocation

In two studies, the risk of bias in random sequence generation and allocation concealment domains was unclear, as no detailed information was provided (Okuma 2010; Yamazaki 2009).

Kondratyeva 2010 used coin tossing for treatment assignment without allocation concealment; we therefore judged it to be low risk of bias for sequence generation and at high risk of bias for allocation concealment.

The five other studies presented clear information, and we judged them to be at low risk of bias in these domains (Crowther 2003; Ordi‐Ros 2019; RAPS 2016; TRAPS 2016; WAPS 2005). Studies used central randomization or sequentially‐numbered sealed envelopes: sequence generation was ensured by means of a random numbers table (Crowther 2003), computer‐generated random number sequences in blocks of 10 (Ordi‐Ros 2019), random permuted blocks of various length with stratification by centre and patient type (RAPS 2016), with the use of web‐based management system with random blocks and stratified (TRAPS 2016), and a program based on the biased‐coin algorithm (WAPS 2005).

Blinding

We assessed blinding of participants, personnel, and outcome assessors for two groups of outcomes: objective outcomes, for example stroke, bleeding, and mortality; and subjective outcomes, such as quality of life or self‐reported outcomes (adverse events, minor bleeding).

Yamazaki 2009 did not provide any information about blinding, so the risk of bias in this study was unclear. Okuma 2010, although described as double‐blinded, did not provide clear information about blinding; it also did not provide clear information about outcome definition or verification, so we judged the risk of bias as unclear.

Crowther 2003 explicitly stated that the trial was double‐blinded and that the participants, treating physicians, auxiliary personnel, and a panel of outcomes assessors were all unaware of the treatment assignments; we considered the risk of bias in this study to be low. WAPS 2005, Kondratyeva 2010, RAPS 2016, and Ordi‐Ros 2019 did not blind the participants or personnel. In our judgement, due to the objective definition or verification of outcomes in those studies, the lack of blinding likely did not influence objective outcomes, so we judged them to be at low risk of bias for those outcomes, but the lack of blinding could influence subjective outcomes, such as quality of life in RAPS 2016 or only self‐reported outcomes (Kondratyeva 2010; Ordi‐Ros 2019), so we judged it to be at a high risk of bias for subjective or not objectively verified outcomes. Ordi‐Ros 2019 also reported the independent committee which reviewed all adverse events (also those not specified as objectively verified) therefore we judged blinding of outcome assessors to be at low risk of bias for this study. WAPS 2005 explicitly stated that outcome assessors were blinded, so we judged the risk of bias to be low. TRAPS 2016 did not blind participants and personnel, but reported blinded adjudication for primary outcomes and its composites, so it was judged to be low risk of bias for those outcomes but it did not report information about outcome assessors for adverse events so it was judged to be unclear for subjective outcomes.

Incomplete outcome data

We assessed Crowther 2003, Kondratyeva 2010, Okuma 2010, RAPS 2016, TRAPS 2016, and Ordi‐Ros 2019 to be at low risk of bias for incomplete outcome data. Reasons included little or no missing data, trials applied the intention‐to‐treat (ITT) principle, missing data were balanced between the groups, or the reasons for missing data were reported and were unlikely to be related to study outcomes. Although WAPS 2005 reported results for all randomized participants, it did not follow up 6/109 participants (information confirmed with the study authors; it is not clear from which group), and although the number of participants in the analysis equaled the number of participants randomized, it is not clear how the participants who were not followed up were included in the analysis. In Yamazaki 2009, there was insufficient information available, so we judged the risk of bias as unclear.

Selective reporting

We assessed the risk of bias for selective reporting in Crowther 2003, Kondratyeva 2010, RAPS 2016, and Ordi‐Ros 2019 as low. RAPS 2016 provided a protocol and reported outcomes as specified in the protocol. We did not identify the protocol for Crowther 2003 or for Kondratyeva 2010 but judging from the texts of the articles it was clear that the published reports include all expected outcomes, including those that were pre‐specified in the methods section.

TRAPS 2016 had published a protocol in which it also indicated minor bleeding and adverse events but in the short report from the study it did not report those findings. As a result, we judged this trial to be at high risk of bias for selective reporting.

We obtained a protocol for WAPS 2005, but there were several discrepancies between the outcomes listed there and in the study publication, so we judged the risk of bias to be high.

We did not identify the protocol for Okuma 2010, and the primary outcomes were partially reported as indicated in the methods section; but the numbers of events were not reported, so we judged the risk of bias to be unclear.

We did not identify the protocol for Yamazaki 2009, and the information provided was insufficient to judge the risk of bias.

Other potential sources of bias

We did not identify any source of other potential bias in three studies (Crowther 2003; Okuma 2010; RAPS 2016). WAPS 2005 was seriously underpowered: the planned sample size was 500 participants per arm, while the number of participants recruited to the study was 109 in total and the study was terminated early due to poor recruitment. With respect to the early termination of the trial, we assessed risk of bias as high for this domain. TRAPS 2016 planned to enroll 536 participants; however, the trial was terminated early after including 120 participants. Therefore, we judged this study to be at high risk of bias for this domain. In Yamazaki 2009 all three groups were planned to be followed up for three years; however, in the group treated with aspirin alone the intervention was discontinued after a year for "humanitarian" reasons, thus we judged this study to be at high risk of bias for this domain. Kondratyeva 2010 and Ordi‐Ros 2019 did not provide sufficient information to judge any other potential sources of bias.

Effects of interventions

See: Summary of findings 1 NOAC (rivaroxaban) versus standard‐dose VKA; Summary of findings 2 High‐dose VKA versus standard‐dose VKA; Summary of findings 3 Standard‐dose VKA plus single antiplatelet agent versus standard‐dose VKA; Summary of findings 4 Standard‐dose VKA plus single antiplatelet agent versus single antiplatelet agent; Summary of findings 5 Standard‐dose VKA plus single antiplatelet agent versus dual antiplatelet agent; Summary of findings 6 Dual antiplatelet agent versus single antiplatelet agent

The studies contributed data regarding six separate comparisons:

  • NOAC (rivaroxaban) versus standard‐dose VKA;

  • high‐dose VKA versus standard‐dose VKA;

  • standard‐dose VKA plus single antiplatelet agent versus standard‐dose VKA;

  • standard‐dose VKA plus single antiplatelet agent versus single antiplatelet agent;

  • standard‐dose VKA plus single antiplatelet agent versus dual antiplatelet agent;

  • dual antiplatelet agent versus single antiplatelet agent

We reported the results for each outcome within these comparisons, and provided separate 'Summary of findings' tables (summary of findings Table 1; summary of findings Table 2; summary of findings Table 3; summary of findings Table 4; summary of findings Table 5; summary of findings Table 6).

The comparisons presented in this review are based on the assumption (and not on unconfounded randomized evidence) that usual treatment with VKA (INR 2.0‐3.0) is effective in reducing the recurrence of thromboembolic events in patients with APS. For the last comparison (dual antiplatelet versus single antiplatelet) the assumption is, that a single antiplatelet is effective in reducing recurrence of thromboembolic events in patients with APS.

With respect to missing data, RAPS 2016 excluded participants' data from the analysis only in the control group (one participant for the analysis of thrombotic events and four participants for the analysis of bleeding events) but not in the experimental group, so we tested worst‐case (assuming that all participants receiving warfarin who had missing data had the best possible outcome) and best‐case (assuming that all participants receiving warfarin who had missing data had the worst possible outcome) scenarios for the missing data of participants in the control group. WAPS 2005 did not provide information about the treatment arm of participants who were not followed up, therefore we could not attempt any sensitivity analysis.

Due to low number of studies for each comparison and outcome, we did not produce funnel plots.

We did not detect significant heterogeneity in any of the comparisons; however, we report results separately for different comparisons separately for anticoagulants, antiplatelets, and different sets of their combinations. We did not attempt any subgroup analyses because of the limited number of studies included in the review.

In the 'Summary of findings' tables, due to the limitations in the number of outcomes reported, we only included stroke and death for 'each type of thromboembolic event analyzed separately.'

1 NOAC (rivaroxaban) versus standard‐dose VKA

1.1 Any thromboembolic event

Three studies were included in the analysis (Ordi‐Ros 2019; RAPS 2016; TRAPS 2016); however, RAPS 2016 did not report any events of recurrent VTE or other thrombotic event at 210 days of follow‐up; it was also not powered to detect differences in the occurrence of clinical events. TRAPS 2016 did not report occurrence of any events in the standard dose VKA (warfarin) group. Overall total number of events was 20 in the NOAC (rivaroxaban) group and six in the standard‐dose VKA group. Meta‐analysis of the results provided an RR of 4.08 (95% CI 0.48 to 34.79; 425 participants; three studies; I2 = 57%; moderate‐certainty evidence; Analysis 1.1). The heterogeneity observed in this comparison may be partially explained by the difference in the population, i.e. in the TRAPS 2016 study only patients with high risk (triple‐positivity), while in Ordi‐Ros 2019 such high‐risk patients comprised 60% of the whole population.

The sensitivity analysis did not change the statistical significance of the results (Analysis 7.1; Analysis 7.2).

1.2 Major bleeding

Three studies were included in the analysis (Ordi‐Ros 2019; RAPS 2016; TRAPS 2016); however, RAPS 2016 did not report any events of major bleeding at 210 days of follow‐up; it was also not powered to detect differences in the occurrence of clinical events. Overall total number of events was 10 in the NOAC (rivaroxaban) group and nine in the standard‐dose VKA group. Meta‐analysis of the results provided an RR of 1.10 (95% CI 0.45 to 2.68; 425 participants; three studies; I2 = 0%; moderate‐certainty evidence; Analysis 1.2). Statistical significance of the results did not change when we pooled the log HR, calculated on the basis of data reported in the studies (Analysis 1.3).

The sensitivity analysis did not change the statistical significance of the results (Analysis 7.3; Analysis 7.4).

1.3 All‐cause mortality

Three studies were included in this analysis (Ordi‐Ros 2019; RAPS 2016; TRAPS 2016). Total number of deaths was six in the NOAC (rivaroxaban) group and four in the standard‐dose VKA group. RAPS 2016 reported one death due to non‐Hodgkin's lymphoma in a participant taking warfarin during 210 days of follow‐up, TRAPS 2016 reported one death due to cardiovascular causes in the VKA group, while Ordi‐Ros 2019 reported eight deaths due to cancer, septic shock, pulmonary hemorrhage, pulmonary hypertension, and cardiac failure. Meta‐analysis of the results provided an RR of 1.45 (95% CI 0.44 to 4.78; 425 participants; three studies; I2 = 0%; moderate‐certainty evidence; Analysis 1.4).

The sensitivity analysis did not change the statistical significance of the results (Analysis 7.5; Analysis 7.6).

1.4 Stroke

Three studies were included in this analysis (Ordi‐Ros 2019; RAPS 2016; TRAPS 2016). The total number of events was 14 in the NOAC (rivaroxaban) group and no cases in the standard dose VKA group. RAPS 2016 did not report any events in the NOAC (rivaroxaban) or the warfarin standard dose groups at 210 days of follow‐up, but the trial was not powered to detect differences in the occurrence of clinical events. TRAPS 2016 reported four participants with stroke in the NOAC (rivaroxaban) group, while Ordi‐Ros 2019 reported 10 participants with stroke in the NOAC (rivaroxaban) group. Meta‐analysis of the results indicate an increased risk of stroke with NOAC (rivaroxaban) treatment with an RR of 14.13 (95% CI 1.87 to 106.81; 425 participants; three studies; I2 = 0%; moderate‐certainty evidence; Analysis 1.5).

1.5 Transient ischemic attack

Three studies were included in this analysis (Ordi‐Ros 2019; RAPS 2016; TRAPS 2016). None of them reported any TIA events during follow‐up (Analysis 1.6).

1.6 Venous thromboembolism (VTE)

Three studies were included in this analysis (Ordi‐Ros 2019; RAPS 2016; TRAPS 2016). RAPS 2016 did not report any events at 210 days of follow‐up, but it was not powered to detect differences in the occurrence of clinical events. TRAPS 2016 reported one participant with VTE in the NOAC (rivaroxaban) group, and Ordi‐Ros 2019 reported two cases of VTE in the NOAC (rivaroxaban) group and three in the standard dose VKA group. Meta‐analysis of the results provided an RR of 0.96 (95% CI 0.20 to 4.49; 425 participants; three studies; I2 = 0%; Analysis 1.7).

1.7 Myocardial infarction (MI)

Three studies were included in this analysis (Ordi‐Ros 2019; RAPS 2016; TRAPS 2016). Similar to other endpoints, RAPS 2016 did not report any events at 210 days of follow‐up. TRAPS 2016 reported three participants with MI in the NOAC (rivaroxaban) group, and Ordi‐Ros 2019 reported no cases of MI in both groups.

Meta‐analysis of the results provided an RR of 7.23 (95% CI 0.38 to 137.08; 425 participants; three studies; I2 = 0%; Analysis 1.8).

1.8 Other thrombotic events

Three studies were included in this analysis (Ordi‐Ros 2019; RAPS 2016; TRAPS 2016). RAPS 2016 reported no cases of microvascular thrombosis at 210 days of follow‐up (Analysis 1.9). TRAPS 2016 reported no cases of other arterial thrombotic events, and Ordi‐Ros 2019 reported two arterial events other than MI and stroke in the NOAC (rivaroxaban) group and three in the standard dose VKA group (Analysis 1.9). The RR for the other arterial thrombotic events, based on Ordi‐Ros 2019 and no events in TRAPS 2016, was 0.67 (95% CI 0.11 to 3.90; 310 participants; two studies; I2 = not applicable; Analysis 1.9)

1.9 Quality of life measured with a validated questionnaire

Only one study reported on this outcome (RAPS 2016). The study reported the mean results of the EQ‐5D‐5L questionnaire at day 180 in terms of health utility. The calculated MD was 0.04 (95% CI ‐0.02 to 0.10 on a scale from 0 to 1; 111 participants; one study; low‐certainty evidence; Analysis 1.10).

Regarding the Health State Visual Analogue Scale, the results showed a statistically significant difference in favor of NOAC (rivaroxaban) with MD 7 mm (95% CI 2.01 to 11.99; 112 participants; one study; low‐certainty evidence; Analysis 1.10).

1.10 Any bleeding that does not meet the criteria for major bleeding

Two studies were included in this analysis (Ordi‐Ros 2019; RAPS 2016). RAPS 2016 reported three participants with clinically relevant non‐major bleeding in the NOAC (rivaroxaban) group and two participants in the standard dose VKA group. Ordi‐Ros 2019 reported nine participants with clinically relevant non‐major bleeding in the NOAC (rivaroxaban) group and five in the standard dose VKA group. Meta‐analysis of the results provided an RR of 1.70 (95% CI 0.69 to 4.19; 302 participants; two studies; I2 = 0%; moderate‐certainty evidence; Analysis 1.11).

RAPS 2016 reported 10 participants with minor bleeding in the NOAC (rivaroxaban) group and eight participants in the standard dose VKA group. Ordi‐Ros 2019 reported 16 participants with minor bleeding in the NOAC (rivaroxaban) group and 14 in the standard dose VKA group. Meta‐analysis of the results provided an RR of 1.17, 95% CI 0.69 to 1.96; 302 participants; two studies; I2 = 0%; Analysis 1.12).

As four participants from the warfarin group were excluded from the analysis in the RAPS 2016 study, we performed a sensitivity analysis. It did not change the statistical significance of the results (Analysis 7.7; Analysis 7.8; Analysis 7.9; Analysis 7.10).

1.11 Adverse events other than bleeding

Ordi‐Ros 2019 reported similar numbers of participants with any adverse events during treatment (62.1% versus 58.9%; RR 1.05, 95% CI 0.84 to 1.3; 190 participants; calculated using RevMan calculator). Eleven participants in the NOAC (rivaroxaban) group and six participants in the standard dose VKA group experienced adverse events which led to withdrawal from the study. There were eight cases of serious adverse events in each group, which occurred between the date when first dose was administered up to 30 days after last visit. In the NOAC (rivaroxaban) group they included events leading to death, such as malignant conditions (three cases), urinary septic shock and pulmonary hemorrhage due to capillaritis (one case each), and not leading to death, such as flare of SLE (two cases) and pneumonia (one case). In the VKA group serious adverse events included one case of malignant condition leading to death, and single cases of cardiac failure, pulmonary hypertension, endocarditis, soft tissue infection, SLE flare, cholecystitis, and intestinal perforation.

RAPS 2016 reported the occurrence of serious adverse events in four participants receiving NOAC (rivaroxaban) and four participants receiving standard dose VKA (warfarin). In the NOAC (rivaroxaban) group, investigators judged two of those events to be unrelated to the study drug: one was a previous intracranial hemorrhage, incidentally detected on brain imaging without any new or recurrent symptoms, and the other was a grade 1 event (grade 2 abdominal pain, vomiting, arthralgia, and myalgia). Additionally, two events were judged unlikely to be related to the study drug (grade 4 intestinal perforation; grade 2 suspected deep vein thrombosis on the basis of a Doppler ultrasound judged to be related to previous femoral vein deep vein thrombosis and without any new thrombosis). In the standard dose VKA (warfarin) group, investigators judged three events to be unrelated to the study drug (grade 3 asthma exacerbation; grade 4 sepsis; high‐grade non‐Hodgkin's lymphoma stage IV B, which resulted in death), and they classified one event as a grade 3 serious adverse reaction, probably related to standard dose VKA (warfarin) treatment (hemorrhoidal hemorrhage).

TRAPS 2016 did not report on adverse events other than bleeding.

High dose VKA versus standard dose VKA

2.1 Any thromboembolic event

Two studies compared the effects of high and standard doses of VKA (warfarin): Crowther 2003 assessed the recurrence of any thrombosis, and WAPS 2005 vascular death or major thrombosis. Together, they reported a total of 11 versus five events during a mean of 2.7 years (SD not reported) and a mean of 3.4 years (SD 1.2) of follow‐up respectively. Meta‐analysis of the results provided an RR of 2.22 (95% CI 0.79 to 6.23; 223 participants; two studies; I2 = 0%; low‐certainty evidence; Analysis 2.1). Statistical significance of the results did not change when we pooled the log HR, calculated on the basis of data reported in the studies (Analysis 2.2).

Crowther 2003 reported INR values in the two participants with thrombotic events in the standard dose group (INR 1.6 and INR 2.8), while the INR values were 3.1, 1.0, 0.9, 1.9, and 3.9 in five out of six participants with events in the high‐dose group (one participant discontinued treatment).

2.2 Major bleeding

Two studies comparing high and standard dose VKA (warfarin) examined major bleeding (Crowther 2003; WAPS 2005), reporting five versus seven cases in total during a mean of 2.7 years (SD not reported) and a mean of 3.4 years (SD 1.2) of follow‐up, respectively. Meta‐analysis of the results provided an RR of 0.74 (95% CI 0.24 to 2.25; 223 participants; two studies; I2 = 0%; low‐certainty evidence; Analysis 2.3). Statistical significance of the results did not change when we pooled the log HR, calculated on the basis of data reported in the studies (Analysis 2.4).

2.3 All‐cause mortality

Two studies comparing high and standard dose VKA (warfarin) reported either no death during a mean of 2.7 years (SD not reported) of follow‐up (Crowther 2003), or three versus two deaths during a mean of 3.4 years (SD 1.2) years of follow‐up (WAPS 2005). Meta‐analysis of the results provided an RR of 1.53 (95% CI 0.27 to 8.79; 223 participants; two studies; I2 = 0%; low‐certainty evidence; Analysis 2.5).

2.4 Stroke

Two studies assessed the effect of high‐dose versus standard‐dose VKA (warfarin) (Crowther 2003; WAPS 2005). The number of events was low (three versus two events) during a mean of 2.7 years (SD not reported) and a mean of 3.4 years (SD 1.2) of follow‐up, respectively. Meta‐analysis of the results provided an RR of 1.37 (95% CI 0.26 to 7.12; 223 participants; two studies; I2 = 0%; low‐certainty evidence; Analysis 2.6).

2.5 Transient ischemic attack

Crowther 2003 and WAPS 2005 both used the TIA definition based on the time of symptoms occurrence. Crowther 2003 reported no TIA events over a mean of 2.7 years (SD not reported) of follow‐up, while in WAPS 2005 the number of events was low in both the high‐ and standard dose warfarin groups (two versus one event) during a mean of 3.4 years (SD 1.2) of follow‐up. Meta‐analysis of the results provided an RR of 2.04 (95% CI 0.19 to 21.81; 223 participants; two studies; I2 = 0%; Analysis 2.7).

2.6 Venous thromboembolism (VTE)

In two studies comparing the effect of high‐dose versus standard‐dose VKA (warfarin), the number of events was low (six versus one event) during a mean of 2.7 years (SD not reported) and a mean of 3.4 years (SD 1.2) of follow‐up, respectively (Crowther 2003; WAPS 2005). Meta‐analysis of the results provided an RR of 4.44 (95% CI 0.77 to 25.72; 223 participants; two studies; I2 = 0%; Analysis 2.8).

2.7 Myocardial infarction

From two studies comparing high‐dose warfarin with standard‐dose VKA (warfarin), WAPS 2005 did not report any event at a mean of 3.4 years (SD 1.2) of follow‐up, while Crowther 2003 reported a single event in each of the treatment groups during a mean of 2.7 years (SD not reported) of follow‐up. Meta‐analysis of the results provided an RR of 1.04 (95% CI 0.07 to 16.16; 223 participants; two studies; I2 = 0%; Analysis 2.9).

2.8 Other thrombotic events

WAPS 2005 reported a single case of superficial vein thrombosis in the high‐dose group over a mean of 3.4 years (SD 1.2) of follow‐up, resulting in the calculated RR of 3.05 (95% CI 0.13 to 73.37; 109 participants; one study; Analysis 2.10).

2.9 Any bleeding that does not meet the criteria for major bleeding

In WAPS 2005 the occurrence of minor bleeding was more frequent in the high‐dose warfarin group compared with the standard‐dose warfarin group during a mean of 3.4 years (SD 1.2) of follow‐up (RR 2.55, 95% CI 1.07 to 6.07; 109 participants; one study; Analysis 2.11). The authors of the WAPS 2005 study also reported minor bleeding as a HR, showing a higher rate of those events with a higher dose (HR 2.92, 95% CI 1.13 to 7.52; 109 participants; one study).

Two studies comparing high‐dose and standard‐dose VKA (warfarin) reported on any bleeding during a mean of 2.7 years (SD not reported) and a mean of 3.4 years (SD 1.2) of follow‐up, respectively (Crowther 2003; WAPS 2005). Meta‐analysis of the results provided an RR of 1.56 (95% CI 0.93 to 2.62; 223 participants; two studies; I2 = 0%; low‐certainty evidence; Analysis 2.12). However, when we pooled log HRs, calculated using the data reported in the published studies, the difference became significant, indicating a higher risk of any bleeding in the high‐dose warfarin group (HR 2.03, 95% CI 1.12 to 3.68; Analysis 2.13).

2.10 Adverse events other than bleeding

WAPS 2005 reported on any adverse events leading to treatment withdrawal and reported two withdrawals associated with reported events, such as essential thrombocythemia in one participant and headache in one participant, but the study authors did not indicate the group in which those participants were included.

3 Standard dose VKA plus single antiplatelet agent versus standard dose VKA

3.1 Any thromboembolic event

Kondratyeva 2010 reported on recurrent thrombosis and recurrent TIA and the combined numbers of participants with those events were 13 in the standard‐dose VKA plus antiplatelet agent group and nine in the standard‐dose VKA group, resulting in an RR of 2.14 (95% CI 1.04 to 4.43; 82 participants; one study; low‐certainty evidence; Analysis 3.1).

3.2 Major bleeding

Kondratyeva 2010 reported on major bleeding with five cases in the combined treatment group and one case in the standard dose VKA treatment group, resulting in an RR of 7.42 (95% CI 0.91 to 60.70; 82 participants; one study; low‐certainty evidence; Analysis 3.2).

3.3 All‐cause mortality

Kondratyeva 2010 reported one death in the standard‐dose VKA‐only group during a median of 51.6 months follow‐up for this group, resulting in an RR of 0.49 (95% CI 0.02 to 11.68; 82 participants; one study; very low‐certainty evidence; Analysis 3.3).

3.4 Stroke

Kondratyeva 2010 reported three participants with stroke in the combined‐treatment group and one participant with stroke in the standard‐dose VKA‐only group, resulting in an RR of 4.45 (95% CI 0.48 to 41.00; 82 participants, one study; low‐certainty evidence; Analysis 3.4).

3.5 Transient ischemic attack

Kondratyeva 2010 reported five participants with TIA in the combined‐treatment group and three participants with TIA in the standard‐dose VKA‐only treatment group, resulting in an RR of 2.47, 95% CI 0.63 to 9.66; 82 participants; one study; Analysis 3.5).

3.6 Venous thromboembolism (VTE)

Kondratyeva 2010 reported one participant with VTE in the combined‐treatment group and three participants with VTE in the standard‐dose VKA group, resulting in an RR of 0.49 (95% CI 0.05 to 4.56; 82 participants; one study; Analysis 3.6).

3.7 Myocardial infarction

Kondratyeva 2010 reported one participant with myocardial infarction in the combined‐treatment group and two participants in the standard‐dose VKA‐only treatment group, resulting in an RR of 0.74 (95% CI 0.07 to 7.86; 82 participants; one study; Analysis 3.7).

3.8 Other thrombotic events

Kondratyeva 2010 reported on three participants with other thrombotic events (hearing loss and retinal artery thrombosis) in the combined‐treatment group and no other thrombotic events in the standard‐dose VKA‐only treatment group, resulting in an RR of 10.29 (95% CI 0.55 to 192.97; 82 participants; 1 study; Analysis 3.8).

3.9 Any bleeding that does not meet the criteria for major bleeding

Kondratyeva 2010 reported on 20 participants with minor bleeding in the combined‐treatment group and 23 in the VKA‐only group, resulting in an RR of 1.29 (95% CI 0.86 to 1.94; 82 participants; one study; low‐certainty evidence; Analysis 3.9).

4 Standard dose VKA plus single antiplatelet agent versus single antiplatelet agent

The studies included in this comparison did not report on any thromboembolic events (only on stroke), all‐cause death, TIA, VTE, MI, other thrombotic events, quality of life, adverse events other than bleeding.

4.1 Major bleeding

Yamazaki 2009 did not provide any information regarding occurrence of major bleeding in participants included in the study, while Okuma 2010 reported a single case of minor cerebral hemorrhage (which we defined as major bleeding) in the single antiplatelet group at a mean of 3.9 years (SD 2.0) of follow‐up, which resulted in RR of 0.40 (95% CI 0.02 to 8.78; 20 participants; one study; very low‐certainty evidence; Analysis 4.1).

4.2 Stroke

Although two studies comparing standard dose VKA plus single antiplatelet agents versus single antiplatelet drug (aspirin) reported stroke as an outcome (Okuma 2010; Yamazaki 2009), only one small study (reported only in conference abstracts) provided results that could be shown on a forest plot (Yamazaki 2009). It did not show significant differences between the treatment groups at one‐year follow‐up, but the aspirin group was discontinued for "humanitarian" reasons, as all three events took place in this group, while no events occurred in the combined treatment group (RR 0.14, 95% CI 0.01 to 2.60; 40 participants; one study; very low‐certainty evidence; Analysis 4.2). Okuma 2010 reported significant differences in the cumulative incidence of stroke in 3.9 years mean follow‐up (SD 2.0) in favor of the combination group (log‐rank test, P = 0.026) but did not report the numbers of participants with an event or HR.

4.3 Any bleeding that does not meet the criteria for major bleeding

Yamazaki 2009 did not provide any information regarding the occurrence of any bleeding in participants included in the study, while Okuma 2010 reported a single case of subcutaneous hemorrhage (which we defined as minor bleeding) in the combined treatment group resulting in an RR of 3.60 (95% CI 0.16 to 79.01; 20 participants; one study; Analysis 4.3), and no cases of gastrointestinal bleeding (no definition or classification provided; very low‐certainty evidence) (Analysis 4.4).

5 Standard dose VKA plus single antiplatelet agent versus dual antiplatelet agent

The study included in this comparison did not report on any thromboembolic events (only on stroke), major bleeding, all‐cause death, TIA, VTE, MI, other thrombotic events, quality of life, any bleeding that does not meet the criteria for major bleeding, or adverse events other than bleeding.

5.1 Stroke

Only Yamazaki 2009 (reported in conference abstracts) provided results at three‐year follow‐up, but there were only two events in the VKA plus antiplatelet group and no events in the dual antiplatelet group, resulting in an RR of 5.0 (95% CI 0.26 to 98.00; 40 participants; one study; very low‐certainty evidence; Analysis 5.1).

6 Dual antiplatelet agent versus single antiplatelet agent

The study included in this comparison did not report on any thromboembolic events (only on stroke), major bleeding, all‐cause death, TIA, VTE, MI, other thrombotic events, quality of life, any bleeding that does not meet the criteria for major bleeding, or adverse events other than bleeding.

6.1 Stroke

Only Yamazaki 2009 (reported in conference abstracts) provided results at one‐year follow‐up. However, the aspirin group was discontinued for "humanitarian" reasons, as all three events occurred in this group, while there were no events in the combined treatment group (RR 0.14, 95% CI 0.01 to 2.60; 40 participants; one study; very low‐certainty evidence; Analysis 6.1).

Discussion

Summary of main results

Of six comparisons reported in this review, three showed some evidence of difference between treatments for several outcomes.

We identified three studies comparing NOAC (rivaroxaban) with standard‐dose VKA, which did not show differences between those groups for any thromboembolic event, major and clinically relevant non‐major bleeding, and death (moderate‐certainty evidence). We found increased risk of stroke with NOAC compared with standard dose VKA (moderate‐certainty evidence).

We identified two studies comparing high‐dose VKA with standard‐dose VKA in the secondary prevention of recurrent thrombosis in people with APS. We found that the differences in the rates of thrombotic events or major bleeding between the treatment groups were not statistically significant, but there was some evidence of an increased risk of minor and any bleeding in the high‐dose group (low‐certainty evidence). However, one of those studies was underpowered, and in the other one the rate of thrombosis was lower than expected.

We identified only one small study, at high risk of bias because of lack of allocation concealment, comparing standard‐dose VKA plus single antiplatelet agents versus standard‐dose VKA. It showed increased risk of any thromboembolic event with combined treatment (low‐certainty evidence), and no differences for other outcomes (very low‐ to low‐certainty evidence).

For three comparisons (standard‐dose VKA plus single antiplatelet agent versus single antiplatelet agent and versus dual antiplatelet agent and dual antiplatelet agent versus single antiplatelet agent) small, poorly‐reported studies at high risk of bias did not provide any conclusive evidence regarding benefits or harms of those drugs in the secondary prevention of stroke in people with APS.

Overall completeness and applicability of evidence

The studies included in the review were small with very small numbers of events. For the comparison of NOAC with standard‐dose VKA, the number of events for primary outcomes (thrombotic events and major bleeding) was low and one of the included studies did not report any events. The number of thrombotic events and major bleedings reported for high‐dose VKA compared with standard‐dose VKA was also low, as was the number of events in a single study comparing standard‐dose VKA and antiplatelet therapy with standard‐dose VKA. The evidence for antiplatelet agents or a combination of antiplatelet and anticoagulant agents was even poorer, due to very small studies or very limited reporting, or both.

The completeness of data, therefore, is a concern in this review with regard to the effects of antiplatelets or anticoagulants, or both, as several studies did not report the data required for meta‐analysis in the assessment of either benefit or harm, and one study was terminated without results published. In addition, we have not been able to finally assess two additional studies due to unclear reporting of methods or data.

All eight included studies reported including people with APS, but qualifying clinical events differed between the studies. Five studies included people with both previous arterial and venous events, two studies included only people with previous stroke, while one study comparing NOAC to the standard‐dose VKA included only people with previous venous thromboembolism while taking no or sub‐therapeutic doses of anticoagulation treatment and without VTE while on warfarin at INR 2.0 to 3.0, so its results may not be applicable to people with a previous arterial event related to APS and with the recurrent event despite standard anticoagulation. Although five studies (two comparing high‐dose VKA with standard‐dose VKA, two comparing NOAC with standard‐dose VKA and one comparing standard‐dose VKA and antiplatelet with standard‐dose VKA) included people with both arterial and venous thrombosis, most participants had prior venous thrombosis (64% to 75%). Therefore, those results may not be fully applicable to people with previous arterial thrombosis.

Similarly, two studies comparing the use of anticoagulant plus antiplatelet agents versus a single antiplatelet agent or dual antiplatelet therapy included only people with previous stroke; results may therefore not be applicable to people with previous venous thromboembolism.

The proportions of participants with each type of antibody, and participants positive for two or three types of antibodies, differed among studies. In addition, one study included only people who were positive for three types of antibodies (high‐risk patients). This would seem to increase the generalizability of evidence in this review, but due to the small number of studies we could not explore the influence of those factors on the effects of our review's eligible interventions. Therefore, properly designed RCTs, including homogenous thrombotic APS patient populations, are needed. People with various constellations of antibodies may have different outcomes and different responses to treatment.

Additional ongoing studies of NOACs compared with standard anticoagulants, in particular the ASTRO‐APS trial (ASTRO‐APS), may add to the body of evidence and help to provide higher‐certainty evidence on the benefits and harms of using NOACs in secondary prevention of thrombotic events in people with thrombotic APS.

Quality of the evidence

We analyzed data from eight trials involving 811 participants with APS. All eight trials took place over the previous 20 years. We judged only one study to be at low risk of bias in all domains (Crowther 2003), while another was at low risk of bias in all domains for one group of outcomes (RAPS 2016). We judged the other six studies to be at low, high, or unclear risk of selection bias; at low, high, or unclear risk for performance bias; at low, high, or unclear risk for detection bias; low or unclear for attrition bias, and/or reporting bias; or at low, high, or unclear risk of other bias. We did not detect important heterogeneity among the results of the studies. However, the studies differed in duration of follow‐up. For the NOAC versus standard‐dose VKA comparison, the follow‐up ranged between 210 days (RAPS 2016), to 35.4 months (Ordi‐Ros 2019), while for high‐dose VKA versus standard‐dose VKA it ranged from 2.7 years (Crowther 2003), to 3.6 years (WAPS 2005). From a clinical perspective, such differences are acceptable as recurrent events usually occur early in the observation. The RAPS 2016 study differed from the other two studies in the NOAC versus standard‐dose VKA comparison in terms of the follow‐up, but also due to a surrogate end‐point and no clinical events during the follow‐up.

All of the analyses provided imprecise results with wide confidence intervals. We could not assess publication bias due to the small number of studies.

We judged the certainty of evidence to be low to moderate for the outcomes in the comparison of NOACs versus standard‐dose VKA; low for the outcomes in the comparison of high‐dose VKA and standard‐dose VKA; low to very low for the comparison of standard‐dose VKA plus antiplatelets with standard‐dose VKA. For all outcomes in other comparisons, we considered the evidence certainty to be very low. We downgraded the evidence mostly for risk of bias, imprecision, or both.

Potential biases in the review process

In our comprehensive searches, supplemented by seeking additional information from experts, unpublished sources, and manufacturers, we attempted to identify all RCTs of potential relevance to the review. We did not apply any limitations to our searches, and for studies published in a language in which none of the review authors was fluent, we sought help with translation. In fact, when we compared the number of studies identified in other recent reviews, meta‐analyses, or practice guidelines, our review identified more studies (published, ongoing, awaiting classification) than previous published reviews (Alegria 2010; Da Silva 2015; Danowski 2013; Dufrost 2016; Erkan 2014; Keeling 2012; Kim 2016; Ruiz‐Irastorza 2011).

Due to the very small number of included studies in each comparison, we did not produce funnel plots.

Agreements and disagreements with other studies or reviews

We identified several recent reviews that covered topics similar to our review. Da Silva 2015 compared high‐ and moderate‐intensity warfarin treatment on the basis of two studies that were also included in our review (Crowther 2003; WAPS 2005), concluding that moderate‐intensity anticoagulation is more suitable for people with APS. However, they based their conclusion on the findings of higher rates of thrombotic events in the high‐intensity group, which was probably an error, as the number of events on the forest plot does not match the number of events reported by the study; and on a higher risk of minor bleeding on the basis of pooled results for both studies. In fact Crowther 2003 did not report minor bleeding events.

Dufrost 2016 examined effects of NOACs in APS and, in addition to RAPS 2016 (which we also identified), this review included case reports and case series. They concluded that NOACs should be used with caution in people with APS and called for additional RCTs with clinical primary endpoints. Authors of patient‐level data meta‐analysis based on all types of studies (including case series, case‐reports, and post‐hoc subgroup analyses of RCTs) observed more thrombotic events in APS patients taking NOACs, especially in those with more criteria for definite APS, with previous arterial thrombosis, and with triple antibody positivity. This might suggest that a high‐risk aPL antibody profile is associated with lower effectiveness of NOACs (Dufrost 2018). We should, however, await the result of still‐ongoing RCTs comparing NOACs to standard care in APS to be able to draw conclusions applying to the whole group of these drugs.

Kim 2016 focused on the intensity of warfarin anticoagulation in people with APS and included several retrospective studies and two RCTs that we also included. That review concluded that more evidence is required with larger sample sizes and better adherence to treatment. We agree that weak adherence to treatment, expressed as INR outside the therapeutic range, is a significant flaw of studies evaluating effects of VKA treatment in APS.

Tektonidou 2019b published systematic reviews which supported EULAR recommendations on the management of adults with APS. They reviewed evidence up to January 2018 on both primary and secondary prevention and included a total of 188 studies of various designs. For comparisons of high‐dose VKA versus standard‐dose VKA, they included RCTs also identified in our review. while for the analysis of the effect of NOAC they included both RCTs published up to 2018, which we also identified. They concluded that there is a need for properly designed studies with homogenous APS populations including patients with single‐ or double‐positive APS.

Primary study flow diagram.

Figures and Tables -
Figure 1

Primary study flow diagram.

Navigate to figure in ReviewOpen in new tab

Flow chart of identification of randomized trials for inclusion in the 22 November 2019 update.RCT: randomized clinical trial

Figures and Tables -
Figure 2

Flow chart of identification of randomized trials for inclusion in the 22 November 2019 update.

RCT: randomized clinical trial

Navigate to figure in ReviewOpen in new tab

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.Duplicate blinding domains are due to separate assessments for objective or objectively verified outcomes (Obj.) and subjective and self‐reported outcomes (Subj.). Blank sections are left because those studies did not assess subjective outcomes or patient self‐reported outcomes not objectively verified.

Figures and Tables -
Figure 3

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Duplicate blinding domains are due to separate assessments for objective or objectively verified outcomes (Obj.) and subjective and self‐reported outcomes (Subj.). Blank sections are left because those studies did not assess subjective outcomes or patient self‐reported outcomes not objectively verified.

Navigate to figure in ReviewOpen in new tab

Risk of bias summary: review authors' judgements about each risk of bias item for each included study. Duplicate blinding domains are due to separate assessments for objective or objectively verified outcomes (Obj.) and subjective and self‐reported outcomes (Subj.). Blank sections are left because those studies did not assess subjective outcomes or patient self‐reported outcomes not objectively verified.

Figures and Tables -
Figure 4

Risk of bias summary: review authors' judgements about each risk of bias item for each included study. Duplicate blinding domains are due to separate assessments for objective or objectively verified outcomes (Obj.) and subjective and self‐reported outcomes (Subj.). Blank sections are left because those studies did not assess subjective outcomes or patient self‐reported outcomes not objectively verified.

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 1: Any thromboembolic event at the longest follow‐up

Figures and Tables -
Analysis 1.1

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 1: Any thromboembolic event at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 2: Major bleeding at the longest follow‐up

Figures and Tables -
Analysis 1.2

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 2: Major bleeding at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 3: Major bleeding at the longest follow‐up ‐ calculated from log hazard ratio

Figures and Tables -
Analysis 1.3

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 3: Major bleeding at the longest follow‐up ‐ calculated from log hazard ratio

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 4: All‐cause mortality at the longest follow‐up

Figures and Tables -
Analysis 1.4

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 4: All‐cause mortality at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 5: Stroke at the longest follow‐up

Figures and Tables -
Analysis 1.5

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 5: Stroke at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 6: TIA at the longest follow‐up

Figures and Tables -
Analysis 1.6

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 6: TIA at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 7: VTE at the longest follow‐up

Figures and Tables -
Analysis 1.7

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 7: VTE at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 8: Myocardial infarction at the longest follow‐up

Figures and Tables -
Analysis 1.8

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 8: Myocardial infarction at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 9: Other thrombotic events at the longest follow‐up

Figures and Tables -
Analysis 1.9

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 9: Other thrombotic events at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 10: Mean Quality of life at day 180

Figures and Tables -
Analysis 1.10

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 10: Mean Quality of life at day 180

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 11: Clinically relevant non‐major bleeding at the longest follow up

Figures and Tables -
Analysis 1.11

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 11: Clinically relevant non‐major bleeding at the longest follow up

Navigate to figure in ReviewOpen in new tab

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 12: Minor bleeding at the longest follow‐up

Figures and Tables -
Analysis 1.12

Comparison 1: NOAC (rivaroxaban) versus standard dose VKA, Outcome 12: Minor bleeding at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 1: Any thromboembolic event at the longest follow‐up

Figures and Tables -
Analysis 2.1

Comparison 2: High dose VKA versus standard dose VKA, Outcome 1: Any thromboembolic event at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 2: Any thromboembolic event at the longest follow‐up ‐ calculated from log hazard ratio

Figures and Tables -
Analysis 2.2

Comparison 2: High dose VKA versus standard dose VKA, Outcome 2: Any thromboembolic event at the longest follow‐up ‐ calculated from log hazard ratio

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 3: Major bleeding at the longest follow‐up

Figures and Tables -
Analysis 2.3

Comparison 2: High dose VKA versus standard dose VKA, Outcome 3: Major bleeding at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 4: Major bleeding at the longest follow‐up ‐ calculated from log hazard ratio

Figures and Tables -
Analysis 2.4

Comparison 2: High dose VKA versus standard dose VKA, Outcome 4: Major bleeding at the longest follow‐up ‐ calculated from log hazard ratio

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 5: All‐cause mortality at the longest follow‐up

Figures and Tables -
Analysis 2.5

Comparison 2: High dose VKA versus standard dose VKA, Outcome 5: All‐cause mortality at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 6: Stroke at the longest follow‐up

Figures and Tables -
Analysis 2.6

Comparison 2: High dose VKA versus standard dose VKA, Outcome 6: Stroke at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 7: TIA at the longest follow‐up

Figures and Tables -
Analysis 2.7

Comparison 2: High dose VKA versus standard dose VKA, Outcome 7: TIA at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 8: VTE at the longest follow‐up

Figures and Tables -
Analysis 2.8

Comparison 2: High dose VKA versus standard dose VKA, Outcome 8: VTE at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 9: Myocardial infarction at the longest follow‐up

Figures and Tables -
Analysis 2.9

Comparison 2: High dose VKA versus standard dose VKA, Outcome 9: Myocardial infarction at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 10: Other thrombotic events at the longest follow‐up (superficial vein thrombosis)

Figures and Tables -
Analysis 2.10

Comparison 2: High dose VKA versus standard dose VKA, Outcome 10: Other thrombotic events at the longest follow‐up (superficial vein thrombosis)

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 11: Minor bleeding at the longest follow‐up

Figures and Tables -
Analysis 2.11

Comparison 2: High dose VKA versus standard dose VKA, Outcome 11: Minor bleeding at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 12: Any bleeding at the longest follow‐up

Figures and Tables -
Analysis 2.12

Comparison 2: High dose VKA versus standard dose VKA, Outcome 12: Any bleeding at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 2: High dose VKA versus standard dose VKA, Outcome 13: Any bleeding at the longest follow‐up ‐ calculated from log hazard ratio

Figures and Tables -
Analysis 2.13

Comparison 2: High dose VKA versus standard dose VKA, Outcome 13: Any bleeding at the longest follow‐up ‐ calculated from log hazard ratio

Navigate to figure in ReviewOpen in new tab

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 1: Any thromboembolic event at the longest follow up

Figures and Tables -
Analysis 3.1

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 1: Any thromboembolic event at the longest follow up

Navigate to figure in ReviewOpen in new tab

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 2: Major bleeding at the longest follow up

Figures and Tables -
Analysis 3.2

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 2: Major bleeding at the longest follow up

Navigate to figure in ReviewOpen in new tab

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 3: All‐cause mortality at the longest follow up

Figures and Tables -
Analysis 3.3

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 3: All‐cause mortality at the longest follow up

Navigate to figure in ReviewOpen in new tab

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 4: Stroke at the longest follow‐up

Figures and Tables -
Analysis 3.4

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 4: Stroke at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 5: TIA at the longest follow‐up

Figures and Tables -
Analysis 3.5

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 5: TIA at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 6: VTE at the longest follow‐up

Figures and Tables -
Analysis 3.6

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 6: VTE at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 7: Myocardial infarction at the longest follow‐up

Figures and Tables -
Analysis 3.7

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 7: Myocardial infarction at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 8: Other thrombotic events at the longest follow‐up (hearing loss, retinal artery thrombosis)

Figures and Tables -
Analysis 3.8

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 8: Other thrombotic events at the longest follow‐up (hearing loss, retinal artery thrombosis)

Navigate to figure in ReviewOpen in new tab

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 9: Minor bleeding at the longest follow‐up

Figures and Tables -
Analysis 3.9

Comparison 3: Standard dose VKA plus single antiplatelet agent versus standard dose VKA, Outcome 9: Minor bleeding at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 4: Standard dose VKA plus single antiplatelet agent versus single antiplatelet agent, Outcome 1: Major bleeding (minor cerebral haemorrhage) at a mean of 3.9 years

Figures and Tables -
Analysis 4.1

Comparison 4: Standard dose VKA plus single antiplatelet agent versus single antiplatelet agent, Outcome 1: Major bleeding (minor cerebral haemorrhage) at a mean of 3.9 years

Navigate to figure in ReviewOpen in new tab

Comparison 4: Standard dose VKA plus single antiplatelet agent versus single antiplatelet agent, Outcome 2: Stroke at 1‐year follow‐up

Figures and Tables -
Analysis 4.2

Comparison 4: Standard dose VKA plus single antiplatelet agent versus single antiplatelet agent, Outcome 2: Stroke at 1‐year follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 4: Standard dose VKA plus single antiplatelet agent versus single antiplatelet agent, Outcome 3: Minor bleeding (subcutaneous haemorrhage) at a mean of 3.9 years

Figures and Tables -
Analysis 4.3

Comparison 4: Standard dose VKA plus single antiplatelet agent versus single antiplatelet agent, Outcome 3: Minor bleeding (subcutaneous haemorrhage) at a mean of 3.9 years

Navigate to figure in ReviewOpen in new tab

Comparison 4: Standard dose VKA plus single antiplatelet agent versus single antiplatelet agent, Outcome 4: GI bleeding (no definition) at a mean of 3.9 years

Figures and Tables -
Analysis 4.4

Comparison 4: Standard dose VKA plus single antiplatelet agent versus single antiplatelet agent, Outcome 4: GI bleeding (no definition) at a mean of 3.9 years

Navigate to figure in ReviewOpen in new tab

Comparison 5: Standard dose VKA plus antiplatelet agent vs dual antiplatelet therapy, Outcome 1: Stroke at 3 years

Figures and Tables -
Analysis 5.1

Comparison 5: Standard dose VKA plus antiplatelet agent vs dual antiplatelet therapy, Outcome 1: Stroke at 3 years

Navigate to figure in ReviewOpen in new tab

Comparison 6: Dual antiplatelet therapy vs single antiplatelet agent, Outcome 1: Stroke at 1 year

Figures and Tables -
Analysis 6.1

Comparison 6: Dual antiplatelet therapy vs single antiplatelet agent, Outcome 1: Stroke at 1 year

Navigate to figure in ReviewOpen in new tab

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 1: Best case any thromboembolic event at the longest follow‐up

Figures and Tables -
Analysis 7.1

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 1: Best case any thromboembolic event at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 2: Worst case any thromboembolic event at the longest follow‐up

Figures and Tables -
Analysis 7.2

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 2: Worst case any thromboembolic event at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 3: Best case major bleeding at the longest follow‐up

Figures and Tables -
Analysis 7.3

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 3: Best case major bleeding at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 4: Worst case major bleeding at the longest follow‐up

Figures and Tables -
Analysis 7.4

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 4: Worst case major bleeding at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 5: Best case all‐cause death at the longest follow‐up

Figures and Tables -
Analysis 7.5

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 5: Best case all‐cause death at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 6: Worst case all‐cause death at the longest follow‐up

Figures and Tables -
Analysis 7.6

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 6: Worst case all‐cause death at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 7: Best case clinically relevant non‐major bleeding at the longest follow up

Figures and Tables -
Analysis 7.7

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 7: Best case clinically relevant non‐major bleeding at the longest follow up

Navigate to figure in ReviewOpen in new tab

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 8: Worst case clinically relevant non‐major bleeding at the longest follow up

Figures and Tables -
Analysis 7.8

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 8: Worst case clinically relevant non‐major bleeding at the longest follow up

Navigate to figure in ReviewOpen in new tab

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 9: Best case minor bleeding at the longest follow‐up

Figures and Tables -
Analysis 7.9

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 9: Best case minor bleeding at the longest follow‐up

Navigate to figure in ReviewOpen in new tab

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 10: Worst case minor bleeding at the longest follow‐up

Figures and Tables -
Analysis 7.10

Comparison 7: Sensitivity analysis NOAC vs VKA, Outcome 10: Worst case minor bleeding at the longest follow‐up

Navigate to figure in ReviewOpen in new tab
Summary of findings 1. NOAC (rivaroxaban) versus standard‐dose VKA

NOAC (rivaroxaban) versus standard‐dose VKA

Patient or population: people with antiphospholipid syndrome and a history of stroke and or thromboembolic events
Setting: specialists centres
Intervention: NOAC (rivaroxaban)
Comparison: standard‐dose VKA

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with standard‐dose VKA

Risk with NOAC

Any thromboembolic event
Follow‐up: range 7 months to 35.4 months

Study population

RR 4.08
(0.48 to 34.79)

425
(3 RCTs)

⊕⊕⊕⊝
Moderateb

28 per 1000

115 per 1000
(14 to 975)

Major bleeding

Follow‐up: range 7 months to 35.4 months

Study population

RR 1.10
(0.45 to 2.68)

425
(3 RCTs)

⊕⊕⊕⊝
Moderateb

42 per 1000

47 per 1000
(19 to 113)

All‐cause mortalitya
Follow‐up: range 7 months to 35.4 months

Study population

RR 1.45
(0.44 to 4.78)

425
(3 RCTs)

⊕⊕⊕⊝
Moderateb

19 per 1000

28 per 1000
(9 to 91)

Strokea
Follow‐up: range 7 months to 35.4 months

Study population

RR 14.13
(1.87 to 106.81)

425
(3 RCTs)

⊕⊕⊕⊝
Moderatec

0 per 1000

0 per 1000

Mean quality of life ‐ health utility

Follow‐up: 180 days

MD 0.04 higher
(0.02 lower to 0.1 higher)

‐

111
(1 RCT)

⊕⊕⊝⊝
Lowd,e

Mean quality of life ‐ health state

Follow‐up: 180 days

MD 7 higher
(2.01 higher to 11.99 higher)

‐

112
(1 RCT)

⊕⊕⊝⊝
Lowd,e

Clinically relevant non‐major bleeding
Follow‐up: range 7 months to 35.4 months

Study population

RR 1.70
(0.69 to 4.19)

302
(2 RCTs)

⊕⊕⊕⊝
Moderatea

47 per 1000

80 per 1000
(33 to 197)

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; MD: mean difference; RCT: randomized clinical trials; RR: risk ratio; VKA: vitamin K antagonists

GRADE Working Group grades of evidence.
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

aAll cause mortality and stroke are shown in the table, however other types of thromboembolic events (TIA, venous thromboembolism, MI) were also considered and are presented in text.
bDowngraded by one level due to imprecision: wide 95% CI including both benefit and harm, low number of events.
cDowngraded by one level due to imprecision: wide 95% CI, low number of events.
dDowngraded by one level due to within‐study risk of bias: patients, personnel and outcome assessors were not blinded.
eDowngraded by one level due to imprecision: wide 95% CI.

Figures and Tables -
Summary of findings 1. NOAC (rivaroxaban) versus standard‐dose VKA
Navigate to table in Review
Summary of findings 2. High‐dose VKA versus standard‐dose VKA

High‐dose VKA versus standard‐dose VKA

Patient or population: people with antiphospholipid syndrome and a history of stroke or thromboembolic events
Setting: specialist centres
Intervention: high‐dose VKA
Comparison: standard‐dose VKA

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Assumed risk

Risk with standard‐dose VKA

Risk with‐high dose VKA

Any thromboembolic event

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

Study population

RR 2.22
(0.79 to 6.23)

223
(2 RCTs)

⊕⊕⊝⊝
Lowc,d

44 per 1000

98 per 1000
(35 to 275)

Major bleeding

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

Study population

RR 0.74
(0.24 to 2.25)

223
(2 RCTs)

⊕⊕⊝⊝
Lowc,d

62 per 1000

46 per 1000
(15 to 140)

All‐cause mortalitya

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

Study population

RR 1.53
(0.27 to 8.79)

223
(2 RCTs)

⊕⊕⊝⊝
Lowc,d

18 per 1000

28 per 1000
(5 to 159)

Strokea

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

Study population

RR 1.37
(0.26 to 7.12)

223
(2 RCTs)

⊕⊕⊝⊝
Lowc,d

18 per 1000

25 per 1000
(5 to 129)

Any bleedingb

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

Study population

RR 1.56

(0.93 to 2.62)

HR 2.03e
(1.12 to 3.68)

223

(2 RCTs)

⊕⊕⊝⊝
Lowc,d

168 per 1000

263 per 1000
(157 to 441)

Adverse events

Follow‐up: mean 2.7 years (SD not reported) and 3.4 years (SD 1.2)

See footnotef

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; HR: hazard ratio; RCT: randomized clinical trials; RR: risk ratio; SD: standard deviation; VKA: vitamin K antagonists

GRADE Working Group grades of evidence.
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

aAll cause mortality and stroke are shown in the table, however other types of thromboembolic events (TIA, venous thromboembolism, MI) were also considered and are presented in text.
bOur review has an outcome 'any bleeding that does not meet criteria for major bleeding', however both studies reported any bleeding, therefore we decided to present it in the 'Summary of findings' table for the information on harm.
cDowngraded by one level due to within‐study risk of bias: some issues concerning incomplete outcome reporting and selective outcome reporting; WAPS 2005 was seriously underpowered as was terminated early due to poor recruitment.
dDowngraded by one level due to imprecision: wide 95% CI including both benefit and harm, low number of events.
eThe results were not statistically significant when analyzed by RR; however when the time to event (HR) was taken into account there was a statistically significant difference between treatment groups.
fOnly one of the two included studies reported adverse events other than bleeding as outcomes and these adverse events lead to withdrawal from the study (WAPS 2005); these were essential thrombocythemia in one participant and headache in one participant, but the study did not indicate the group in which those participants were included.

Figures and Tables -
Summary of findings 2. High‐dose VKA versus standard‐dose VKA
Navigate to table in Review
Summary of findings 3. Standard‐dose VKA plus single antiplatelet agent versus standard‐dose VKA

Standard‐dose VKA plus single antiplatelet agent versus standard‐dose VKA

Patient or population: people with antiphospholipid syndrome and a history of stroke or thromboembolic events
Setting: specialist centres
Intervention: standard‐dose VKA plus single antiplatelet agent
Comparison: standard‐dose VKA

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Assumed risk

Risk with standard‐dose VKA

Risk with standard‐dose VKA plus single antiplatelet agent

Any thromboembolic event

Follow‐up: median 58.4 months (IQR 0.9; 38.9), in VKA+AP group follow‐up and 51.6 (IQR 0.9; 48.8) in VKA group

Study population

RR 2.14
(1.04 to 4.43)

82
(1 RCTs)

⊕⊕⊝⊝
Lowc,d

184 per 1000

394 per 1000
(192 to 816)

Major bleeding

Follow‐up: median 58.4 months (IQR 0.9; 38.9), in VKA+AP group follow‐up and 51.6 (IQR 0.9; 48.8) in VKA group

Study population

RR 7.42
(0.91 to 60.70)

82
(1 RCTs)

⊕⊕⊝⊝
Lowc,e

20 per 1000

149 per 1000
(19 to 1214)

All‐cause mortalitya

Follow‐up: median 58.4 months (IQR 0.9; 38.9), in VKA+AP group follow‐up and 51.6 (IQR 0.9; 48.8) in VKA group

Study population

RR 0.49
(0.02 to 11.68)

82
(1 RCTs)

⊕⊝⊝⊝
Very lowc,f

20 per 1000

10 per 1000
(1 to 234)

Strokea

Follow‐up: median 58.4 months (IQR 0.9; 38.9), in VKA+AP group follow‐up and 51.6 (IQR 0.9; 48.8) in VKA group

Study population

RR 4.45
(0.48 to 41.00)

82
(1 RCTs)

⊕⊕⊝⊝
Lowc,e

20 per 1000

89 per 1000
(10 to 820)

Minor bleedingb

Follow‐up: median 58.4 months (IQR 0.9; 38.9), in VKA+AP group follow‐up and 51.6 (IQR 0.9; 48.8) in VKA group

Study population

RR 1.29

(0.86 to 1.94)

82
(1 RCTs)

⊕⊕⊝⊝
Lowc,e

469 per 1000

606 per 1000
(404 to 910)

Adverse events

Not reported

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; IQR: interquartile range; RCT: randomized clinical trials; RR: risk ratio; VKA: vitamin K antagonists

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect,
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different,
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.

aAll cause mortality and stroke are shown in the table, however other types of thromboembolic events (TIA, venous thromboembolism, MI) were also considered and are presented in text.
bOur review has an outcome 'any bleeding that does not meet criteria for major bleeding', however the study reported only major and minor bleeding, therefore we decided to present both outcomes in the 'Summary of findings' table for the information on harm.
cDowngraded by one level due to within‐study risk of bias: no allocation concealment.
dDowngraded by one level due to imprecision: wide 95% CI, low number of events.
eDowngraded by one level due to imprecision: wide 95% CI including both benefit and harm, low number of events.
fDowngraded by two levels due to imprecision: very wide 95% CI including both benefit and harm, single event in one group only.

Figures and Tables -
Summary of findings 3. Standard‐dose VKA plus single antiplatelet agent versus standard‐dose VKA
Navigate to table in Review
Summary of findings 4. Standard‐dose VKA plus single antiplatelet agent versus single antiplatelet agent

Standard‐dose VKA plus single antiplatelet agent versus single antiplatelet agent

Patient or population: people with antiphospholipid syndrome, with previous stroke
Setting: Japan, 1 centre or unknown number of centres
Intervention: standard‐dose VKA plus single antiplatelet agent
Comparison: single antiplatelet agent

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with single antiplatelet agent

Risk with
standard‐dose VKA plus single antiplatelet agent

Major bleeding (minor cerebral hemorrhage)

Follow‐up: mean 3.9 years (SD 2.0)

Study population

RR 0.40
(0.02 to 8.78)

20
(1 RCT)

⊕⊝⊝⊝
Very lowa,b

—

91 per 1000

37 per 1000
(2 to 799)

Stroke

Follow‐up: 1 year

Study population

RR 0.14 (0.01 to 2.60)

40
(1 RCT)

⊕⊝⊝⊝
Very lowa,c

1 small study published only as conference abstracts; single antiplatelet drug group discontinued after 1 year for humanitarian considerations

150 per 1000

21 per 1000
(2 to 390)

Any bleeding that does not meet criteria for major bleeding ‐ Gastrointestinal bleeding (no definition)

Follow‐up: mean 3.9 years (SD 2.0)

Study population

RR 3.60 (0.16 to 79.01)

20
(1 RCT)

⊕⊝⊝⊝
Very lowa,b

—

0 per 1000d

0 per 1000d

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; RCT: randomized clinical trials; RR: risk ratio; SD: standard deviation; VKA: vitamin K antagonists

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect,
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different,
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.

aDowngraded by two levels due to imprecision: very low number of events and sample size, very wide 95% CI including both benefit and harm.
bDowngraded by one level due to within‐study risk of bias: insufficient information regarding randomization, concealment, blinding, selective outcome reporting.
cDowngraded by one level due to within‐study risk of bias: insufficient information regarding all aspects of study design, no clear sequence generation, allocation concealment, blinding, completeness of outcome data and selective outcome reporting.
dNumbers could not be calculated as number of events in control group was 0.

Figures and Tables -
Summary of findings 4. Standard‐dose VKA plus single antiplatelet agent versus single antiplatelet agent
Navigate to table in Review
Summary of findings 5. Standard‐dose VKA plus single antiplatelet agent versus dual antiplatelet agent

Standard‐dose VKA plus single antiplatelet agent versus dual antiplatelet agent

Patient or population: people with antiphospholipid syndrome, with previous stroke
Setting: Japan, 1 centre
Intervention: standard‐dose VKA plus single antiplatelet agent
Comparison: dual antiplatelet agent

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with dual antiplatelet agent

Risk with
standard‐dose VKA plus single antiplatelet agent

Stroke

Follow‐up: 3 years

Study population

RR 5.00 (0.26 to 98.00)

40
(1 RCT)

⊕⊝⊝⊝
Very lowa,b

1 small study published only as conference abstracts; single antiplatelet drug group discontinued after 1 year for humanitarian considerations

0 per 1000c

0 per 1000c

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; RCT: randomized clinical trials; RR: risk ratio; VKA: vitamin K antagonists

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect,
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different,
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.

aDowngraded by two levels due to imprecision: very low number of events and sample size, very wide 95% CI including both benefit and harm.
bDowngraded by one level due to within‐study risk of bias: insufficient information regarding all aspects of study design, no clear sequence generation, allocation concealment, blinding, completeness of outcome data and selective outcome reporting.
cNumbers could not be calculated as number of events in control group was 0.

Figures and Tables -
Summary of findings 5. Standard‐dose VKA plus single antiplatelet agent versus dual antiplatelet agent
Navigate to table in Review
Summary of findings 6. Dual antiplatelet agent versus single antiplatelet agent

Dual antiplatelet agent versus single antiplatelet agent

Patient or population: people with antiphospholipid syndrome, with previous stroke
Setting: Japan, 1 centre
Intervention: dual antiplatelet agent
Comparison: single antiplatelet agent

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with dual antiplatelet agent

Risk with
single antiplatelet agent

Stroke

Follow‐up: 1 year

Study population

RR 0.14 (0.01 to 2.6)

40
(1 RCT)

⊕⊝⊝⊝
Very lowa,b

1 small study published only as conference abstracts; single antiplatelet drug group discontinued after 1 year for humanitarian considerations

150 per 1000

21 per 1000
(2 to 390)

*The risk in the intervention group (and its 95% confidence interval) is based on the average risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: confidence interval; RCT: randomized clinical trials; RR: risk ratio

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect,
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different,
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.

aDowngraded by two levels due to imprecision: very low number of events and sample size, very wide 95% CI including both benefit and harm.
bDowngraded by one level due to within‐study risk of bias: insufficient information regarding all aspects of study design, no clear sequence generation, allocation concealment, blinding, completeness of outcome data and selective outcome reporting.

Figures and Tables -
Summary of findings 6. Dual antiplatelet agent versus single antiplatelet agent
Navigate to table in Review
Comparison 1. NOAC (rivaroxaban) versus standard dose VKA

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1.1 Any thromboembolic event at the longest follow‐up Show forest plot

3

425

Risk Ratio (M‐H, Random, 95% CI)

4.08 [0.48, 34.79]

1.2 Major bleeding at the longest follow‐up Show forest plot

3

425

Risk Ratio (M‐H, Random, 95% CI)

1.10 [0.45, 2.68]

1.3 Major bleeding at the longest follow‐up ‐ calculated from log hazard ratio Show forest plot

2

Hazard Ratio (IV, Random, 95% CI)

1.15 [0.46, 2.87]

1.4 All‐cause mortality at the longest follow‐up Show forest plot

3

425

Risk Ratio (M‐H, Random, 95% CI)

1.45 [0.44, 4.78]

1.5 Stroke at the longest follow‐up Show forest plot

3

425

Risk Ratio (M‐H, Random, 95% CI)

14.13 [1.87, 106.81]

1.6 TIA at the longest follow‐up Show forest plot

3

Risk Ratio (M‐H, Random, 95% CI)

Subtotals only

1.7 VTE at the longest follow‐up Show forest plot

3

425

Risk Ratio (M‐H, Random, 95% CI)

0.96 [0.20, 4.49]

1.8 Myocardial infarction at the longest follow‐up Show forest plot

3

425

Risk Ratio (M‐H, Random, 95% CI)

7.23 [0.38, 137.08]

1.9 Other thrombotic events at the longest follow‐up Show forest plot

3

Risk Ratio (M‐H, Random, 95% CI)

Subtotals only

1.9.1 microvascular thrombosis

1

115

Risk Ratio (M‐H, Random, 95% CI)

Not estimable

1.9.2 arterial events other than stroke and MI

2

310

Risk Ratio (M‐H, Random, 95% CI)

0.67 [0.11, 3.90]

1.10 Mean Quality of life at day 180 Show forest plot

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

1.10.1 Health utility

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

1.10.2 Health state

1

Mean Difference (IV, Random, 95% CI)

Totals not selected

1.11 Clinically relevant non‐major bleeding at the longest follow up Show forest plot

2

302

Risk Ratio (M‐H, Random, 95% CI)

1.70 [0.69, 4.19]

1.12 Minor bleeding at the longest follow‐up Show forest plot

2

302

Risk Ratio (M‐H, Random, 95% CI)

1.17 [0.69, 1.96]
Figures and Tables -
Comparison 1. NOAC (rivaroxaban) versus standard dose VKA
Navigate to table in Review
Comparison 2. High dose VKA versus standard dose VKA

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

2.1 Any thromboembolic event at the longest follow‐up Show forest plot

2

223

Risk Ratio (M‐H, Random, 95% CI)

2.22 [0.79, 6.23]

2.2 Any thromboembolic event at the longest follow‐up ‐ calculated from log hazard ratio Show forest plot

2

Hazard Ratio (IV, Random, 95% CI)

2.17 [0.74, 6.31]

2.3 Major bleeding at the longest follow‐up Show forest plot

2

223

Risk Ratio (M‐H, Random, 95% CI)

0.74 [0.24, 2.25]

2.4 Major bleeding at the longest follow‐up ‐ calculated from log hazard ratio Show forest plot

2

Hazard Ratio (IV, Random, 95% CI)

0.83 [0.25, 2.72]

2.5 All‐cause mortality at the longest follow‐up Show forest plot

2

223

Risk Ratio (M‐H, Random, 95% CI)

1.53 [0.27, 8.79]

2.6 Stroke at the longest follow‐up Show forest plot

2

223

Risk Ratio (M‐H, Random, 95% CI)

1.37 [0.26, 7.12]

2.7 TIA at the longest follow‐up Show forest plot

2

223

Risk Ratio (M‐H, Random, 95% CI)

2.04 [0.19, 21.81]

2.8 VTE at the longest follow‐up Show forest plot

2

223

Risk Ratio (M‐H, Random, 95% CI)

4.44 [0.77, 25.72]

2.9 Myocardial infarction at the longest follow‐up Show forest plot

2

223

Risk Ratio (M‐H, Random, 95% CI)

1.04 [0.07, 16.16]

2.10 Other thrombotic events at the longest follow‐up (superficial vein thrombosis) Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Subtotals only

2.11 Minor bleeding at the longest follow‐up Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Subtotals only

2.12 Any bleeding at the longest follow‐up Show forest plot

2

223

Risk Ratio (M‐H, Random, 95% CI)

1.56 [0.93, 2.62]

2.13 Any bleeding at the longest follow‐up ‐ calculated from log hazard ratio Show forest plot

2

Hazard Ratio (IV, Random, 95% CI)

2.03 [1.12, 3.68]
Figures and Tables -
Comparison 2. High dose VKA versus standard dose VKA
Navigate to table in Review
Comparison 3. Standard dose VKA plus single antiplatelet agent versus standard dose VKA

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

3.1 Any thromboembolic event at the longest follow up Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Totals not selected

3.2 Major bleeding at the longest follow up Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Totals not selected

3.3 All‐cause mortality at the longest follow up Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Totals not selected

3.4 Stroke at the longest follow‐up Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Totals not selected

3.5 TIA at the longest follow‐up Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Subtotals only

3.6 VTE at the longest follow‐up Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Subtotals only

3.7 Myocardial infarction at the longest follow‐up Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Subtotals only

3.8 Other thrombotic events at the longest follow‐up (hearing loss, retinal artery thrombosis) Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Totals not selected

3.9 Minor bleeding at the longest follow‐up Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Totals not selected
Figures and Tables -
Comparison 3. Standard dose VKA plus single antiplatelet agent versus standard dose VKA
Navigate to table in Review
Comparison 4. Standard dose VKA plus single antiplatelet agent versus single antiplatelet agent

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

4.1 Major bleeding (minor cerebral haemorrhage) at a mean of 3.9 years Show forest plot

1

20

Risk Ratio (M‐H, Random, 95% CI)

0.40 [0.02, 8.78]

4.2 Stroke at 1‐year follow‐up Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Totals not selected

4.3 Minor bleeding (subcutaneous haemorrhage) at a mean of 3.9 years Show forest plot

1

20

Risk Ratio (M‐H, Random, 95% CI)

3.60 [0.16, 79.01]

4.4 GI bleeding (no definition) at a mean of 3.9 years Show forest plot

1

20

Risk Ratio (M‐H, Random, 95% CI)

Not estimable
Figures and Tables -
Comparison 4. Standard dose VKA plus single antiplatelet agent versus single antiplatelet agent
Navigate to table in Review
Comparison 5. Standard dose VKA plus antiplatelet agent vs dual antiplatelet therapy

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

5.1 Stroke at 3 years Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Totals not selected
Figures and Tables -
Comparison 5. Standard dose VKA plus antiplatelet agent vs dual antiplatelet therapy
Navigate to table in Review
Comparison 6. Dual antiplatelet therapy vs single antiplatelet agent

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

6.1 Stroke at 1 year Show forest plot

1

Risk Ratio (M‐H, Random, 95% CI)

Totals not selected
Figures and Tables -
Comparison 6. Dual antiplatelet therapy vs single antiplatelet agent
Navigate to table in Review
Comparison 7. Sensitivity analysis NOAC vs VKA

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

7.1 Best case any thromboembolic event at the longest follow‐up Show forest plot

3

426

Risk Ratio (M‐H, Random, 95% CI)

2.34 [0.43, 12.74]

7.2 Worst case any thromboembolic event at the longest follow‐up Show forest plot

3

426

Risk Ratio (M‐H, Random, 95% CI)

4.08 [0.48, 34.79]

7.3 Best case major bleeding at the longest follow‐up Show forest plot

3

429

Risk Ratio (M‐H, Random, 95% CI)

0.87 [0.28, 2.75]

7.4 Worst case major bleeding at the longest follow‐up Show forest plot

3

429

Risk Ratio (M‐H, Random, 95% CI)

1.10 [0.45, 2.68]

7.5 Best case all‐cause death at the longest follow‐up Show forest plot

3

426

Risk Ratio (M‐H, Random, 95% CI)

1.32 [0.40, 4.29]

7.6 Worst case all‐cause death at the longest follow‐up Show forest plot

3

426

Risk Ratio (M‐H, Random, 95% CI)

1.46 [0.44, 4.79]

7.7 Best case clinically relevant non‐major bleeding at the longest follow up Show forest plot

2

306

Risk Ratio (M‐H, Random, 95% CI)

1.04 [0.31, 3.49]

7.8 Worst case clinically relevant non‐major bleeding at the longest follow up Show forest plot

2

306

Risk Ratio (M‐H, Random, 95% CI)

1.73 [0.70, 4.27]

7.9 Best case minor bleeding at the longest follow‐up Show forest plot

2

306

Risk Ratio (M‐H, Random, 95% CI)

1.01 [0.62, 1.66]

7.10 Worst case minor bleeding at the longest follow‐up Show forest plot

2

306

Risk Ratio (M‐H, Random, 95% CI)

1.20 [0.71, 2.02]
Figures and Tables -
Comparison 7. Sensitivity analysis NOAC vs VKA
Navigate to table in Review