The review concluded that treatment with intravitreal bevacizumab pre-treatment before vitrectomy eye surgery could achieve very good clinical outcomes for patients with severe diabetic retinopathy. Given unclear reporting of study quality and unexplained variability between studies, the authors' conclusions should be interpreted with caution.

To investigate potential benefits of intravitreal bevacizumab treatment before vitrectomy in severe or complicated diabetic retinopathy.

PubMed, EMBASE, and the Cochrane Central Register of Controlled Trials (CENTRAL) were searched for relevant papers up to April 2010 . Some search terms were reported. Bibliographies of retrieved papers were searched manually. No language restrictions were applied to the search.

Randomised controlled trials (RCTs) and high-quality non-randomised comparative studies that compared vitrectomy alone versus vitrectomy plus intravitreal bevacizumab pre-treatment for severe diabetic retinopathy were eligible for inclusion. Participants had to be older than 18 years. Studies had to report at least one intra-operative and/or postoperative clinical outcome.

In included studies, the mean age of patients ranged from 44.4 to 56.7 years (where stated). Types of proliferative diabetic retinopathy included active-severe, severe, active and complications of proliferative diabetic retinopathy. The dose of intravitreal bevacizumab ranged from 1.25mg to 2.5mg. Included studies were conducted in Taiwan, Italy, Iran and Brazil; the publication year ranged from 2008 to 2010.

Two reviewers assessed study eligibility. Disagreements were resolved in discussions until consensus was reached.

The authors appeared to evaluate study quality using a modified Jadad scale by assessing allocation concealment, method of allocation, blinding of outcome assessment, and completeness of follow-up.

Quality assessment was carried out independently by two reviewers.

Intention-to-treat data were extracted from each study to calculate the odds ratio (OR) or weighted mean difference (WMD); these were estimated with 95% confidence intervals (CIs).

The authors did not state how many reviewers extracted the data.

The studies were synthesised by meta-analysis using a fixed-effect or random-effects model. Heterogeneity was assessed using Ι² and the Q statistic of the Χ² test. If the Ι² value indicated heterogeneity (Ι²>50%), a random-effects meta-analysis was used to synthesise studies using the DerSimonian-Laird method. Intra-operative bleeding, mean surgical time, re-absorption time of blood, and recurrent vitreous haemorrhage were assessed using random-effects models. A sensitivity analysis was carried out by excluding non-randomised studies.

Publication bias was assessed using a funnel plot.

Six RCTs and one non-randomised comparative study were included (278 patients, 281 eyes). Sample sizes ranged from 20 to 68 patients. Follow-up time ranged from one day to around seven months. Scores on the modified Jadad scale ranged from 1 to 7 points.

Intra-operative outcomes

There was a significant difference favouring intravitreal bevacizumab pre-treatment for intra-operative bleeding (OR 8.85, 95% CI 2.08 to 37.61; five studies), endodiathermy (OR 15.06, 95% CI 4.01 to 56.53; four studies), and mean surgical time (WMD 14.13, 95% CI 1.17 to 27.09, units not reported; five studies).

There was no significant difference in the number of iatrogenic tears between the treatment and control groups (OR 3.72, 95% CI 1.00 to 13.79; three studies).

Significant heterogeneity was found for intra-operative bleeding (Ι²=74%) and mean surgical time (Ι²=76%).

Sensitivity analyses excluding the non-randomised study did not change the treatment effect for intra-operative bleeding or mean surgical time.

Postoperative outcomes

There was a significant difference favouring intravitreal bevacizumab pre-treatment for the re-absorption time of blood after surgery (WMD 15.82, 95% CI 0.43 to 31.20, units not reported; two studies) and the mean best-corrected visual acuity (WMD 0.31, 95% CI 0.10 to 0.52, units not reported; six studies).

There was no significant difference in the proportion of recurrent vitreous haemorrhage between the treatment and control groups (OR 5.48, 95% CI 0.97 to 31.02; five studies). Postoperative complications did not differ significantly between the groups for early elevation of intra-ocular pressure (three studies), late elevation of intra-ocular pressure (three studies), final retinal detachment (three studies) and repeat vitrectomy (two studies).

Significant heterogeneity was found for re-absorption time of blood after surgery (Ι²=94%) and proportion of recurrent vitreous haemorrhage (Ι²=73%).

Sensitivity analyses excluding the non-randomised study did not change the treatment effect for the reabsorption time of blood after surgery or the proportion of recurrent vitreous haemorrhage.

There was no evidence of publication bias.

Intravitreal bevacizumab pre-treatment in vitrectomy could achieve very good clinical outcomes for patients with severe diabetic retinopathy. It had the potential to facilitate surgeons' manoeuvres and to reduce intra-operative and postoperative complications.

The review question was clear. Inclusion criteria for participants were less clear; severe proliferative diabetic retinopathy was specified as the population inclusion criterion, but studies that contained patients with complications from proliferative diabetic retinopathy were also included. Relevant databases and bibliographies were searched with no language restrictions. However, the authors search for unpublished studies, so the possibility of missing studies or publication bias could not be ruled out. The risk of reviewer bias was reduced by the use of independent, duplicate review processes for study selection and quality assessment, but it was unclear whether the same processes were applied to data extraction.

It appeared that a modified Jadad scale was used to assess study quality. Scores were reported for individual studies without further result detail or explanation of how scores were derived, so the reliability of the included studies remained unclear. Appropriate trial details were reported. Data were synthesised appropriately. Differences in study design were addressed in sensitivity analyses. Suitable measures were used to assess heterogeneity; the potential effects of the observed heterogeneity on the reliability of results were reported appropriately. However, sources of heterogeneity were not fully explored.

In light of the unexplored heterogeneity and uncertainty in the reliability of the results because of unclear reporting of study quality, the authors' conclusions should be interpreted with caution.

Practice: The authors stated that intravitreal bevacizumab pre-treatment may present a new strategy to improve the safety and effectiveness of vitrectomy for severe diabetic retinopathy.

Research: The authors recommended that larger RCTs or systematic studies should be conducted to allow a better evaluation of the long-term benefits and safety of intravitreal bevacizumab pre-treatment for severe diabetic retinopathy. They also stated that larger RCTs were needed to clarify the optimal interval time and dosage of intravitreal bevacizumab.

Shanghai Leading Academic Discipline Project, China; General Program of Bio-medical Division of Shanghai Science and Technology Division, China.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.