Safety of continuing warfarin therapy during cataract surgery: A systematic review and meta-analysis

Methodological quality of the review: Low confidence

Author: Jamula E, Anderson J, Douketis JD.

Region: Details not provided

Sector: Cataract surgery

Sub-sector: Warfarin therapy

Type of cataract: Age-related cataract

Equity focus: None specified

Review type: Effectiveness review

Quantitative synthesis method: Meta-analysis

Qualitative synthesis methods: Not applicable

Background

Warfarin is a frequently used anticoagulant, commonly used by elderly patients, patients who are also more likely to undergo cataract surgery, since the incidence of cataract is very high amongst the elderly. To date the safety around using Warfarin in the perioperative period has not been systematically evaluated.

Research objectives

‘To determine the safety (bleeding risk) of continuing warfarin therapy during cataract surgery. We also assessed the efficacy (thromboembolism risk) of perioperative warfarin continuation.’

Main findings

The authors reviewed 11 studies including one randomized control trial, four cohort and six case series. The geographical location of the studies was not addressed by the authors. All studies assessed bleeding risk associated with warfarin continuation during cataract surgery. Authors reported the following results:

Patients who continued warfarin had an increased risk of bleeding (odds ratio; 3,26; 95% confidence interval [CI]: 1.73-6.16). The overall incidence of bleeding (95% CI) was 10% (5-19). In the two studies looking at thromboembolic events, rates of thromboembolism were low at (< 1%).

Based on findings reported, authors concluded that patients who received warfarin therapy and underwent cataract surgery without interruption of the treatment had approximately a three-fold greater risk for bleeding. However, bleeds were ‘self-limiting and without clinical consequences’. They also noted that only two studies reviewed reported on the perioperative incidence of thromboembolic events and reported low rates of thromboembolism. However, the lack of evidence precluded firm conclusions around rates of thromboembolism. Further research was needed to determine the safest perioperative strategy for anticoagulated patients who need cataract surgery.

Methodology

Authors included studies which assessed bleeding risk associated with warfarin continuation during cataract surgery and the outcome evaluated in the review was the rate of any bleeding events during surgery.

Two authors independently searched for articles in the following databases: MEDLINE, EMBASE and PUBMED with no language or date restrictions for any search applied. Reference lists of relevant articles were manually searched and a content expert was contacted to attempt to identify unpublished studies. Eligibility and quality assessment was performed by two independent reviewers.

The authors identified 11 studies for inclusion in the review, including one randomized control trial, four cohort studies and six case series. The four cohort studies were of good quality, the randomized trial was of high quality and the six case series were of poor quality. Authors conducted three different meta-analyses: one included all studies; the second consisted of cohort studies and randomized trial, and the third included case series only.

Applicability/external validity

Authors did not discuss the applicability/external validity of the results.

Geographic focus

Geographical location of the include studies was not addressed by the authors.

Quality assessment

This review was based on comprehensive search of the literature, avoiding language bias. Authors noted that studies with different risk of bias assessment are combined in the meta-analysis as were different study designs (cohort, case series and RCT studies), although they did not discuss the appropriateness of this methodology in the review.

The case series were generally descriptive studies, tracking patients derived by a health care setting with a known exposure or receiving the same treatment and examine their outcome. These were not epidemiological studies and may not have been appropriate for a pooled analysis. In total, six case studies were included in the meta-analysis; three were retrospective reviews and three prospectively followed patients. The quality of these was very poor as they lacked a comparator group. Most of these had flaws in terms of exposure or ascertainment reporting. On the other hand, the cohort studies were of good quality and the randomized trial was of high quality. Therefore, overall, this review was awarded low confidence in conclusions about the effects of this study.