The authors concluded that, using indirect meta-analysis, there was a modest benefit for rescue percutaneous coronary intervention compared to repeat thrombolysis in failed thrombolysis. Evidence appeared to support the authors’ conclusions but, as with all indirect comparisons, caution is require when interpreting the results. Incomplete reporting of review methods made it difficult to comment on reliability of findings.

To compare the risks and benefits of repeat thrombolysis (RT) with rescue percutaneous coronary intervention (PCI) in patients with failed thrombolysis.

BioMed Central, Cochrane Central Register of Controlled Trials, mRCT (megaRegister of Controlled Trials) and PubMed were searched without language restriction to September 2007. Search terms were not reported. In addition, major recent international cardiology meetings and reference lists of reviews and original studies were screened.

Randomised controlled trials (RCTs) that compared rescue PCI or RT with conservative treatment in patients with failed thrombolysis were eligible for inclusion. Included studies had to use unequivocal treatment allocation methods and intention-to-treat (ITT) analysis. They also were required to have complete follow up (80% or more) and no significant differences between treatment groups in major characteristics at baseline. Review outcomes were major adverse events (defined as all cause death and non-fatal re-infarction), overall mortality, re-infarction, stroke, congestive heart failure, major bleeds and minor bleeds.

In the included trials, the duration of follow-up was generally ‘in hospital’ to one month; one trial assessed outcomes at six months.

The authors did not state how papers were selected for the review, or how many reviewers performed the selection.

Validity was assessed using the Cochrane Collaboration criteria (risk of selection, performance, attrition and adjudication bias), allocation concealment and the Jadad score (a 5-point scale, where high quality is classified as 5).

The authors did not state how the validity assessment was performed.

Four reviewers independently extracted data and resolved disagreements by consensus.

Pooled odds ratios (OR) and 95% confidence intervals (CI) were calculated for rescue PCI versus conservative treatment and for RT versus conservative treatment. Both the DerSimonian and Laird random-effects model and fixed-effect model were used. Heterogeneity was assessed using the Cochrane Q and the I² statistics; the level of inconsistency was defined as low (<25%), moderate (25 to 75%) and high (>75%). Using random-effects ORs, an In(OR) (natural logarithm interaction odds ratio) with 95% CI was generated to compare rescue PCI with RT using an adjusted indirect meta-analysis with inverse variance weighting was used. The I² statistic of the interaction was also assessed. The number needed to treat (NNT) and number needed to harm (NNH) were also calculated. Sensitivity analysis was performed by excluding studies one at a time starting from the lowest quality to the highest quality. Publication bias was assessed using a funnel plot and Egger’s test.

Eight RCTs were included (n=1,318 reported in text and n=1,177 according to tables). Five RCTs (n=623) compared RT versus conservative treatment, two RCTs (n=127) compared rescue PCI versus conservative treatment and one RCT (n=427) compared RT versus PCI versus conservative treatment.

Overall, trials were judged to be at low risk of bias according to the Cochrane Collaboration criteria. The average Jadad score was 3.2 (2.8 for rescue PCI studies and 4.0 for RT studies). There was no evidence of publication bias.

RT versus conservative treatment: There was no significant difference between RT and conservative treatment for major adverse events, overall mortality, re-infarction or major bleeds. RT was associated with a statistically significant increase in minor bleeds in fixed-effect but not random-effects models (OR fixed-effect 2.08, 95% CI: 1.09, 3.98, p=0.03, I² 55.5%, NNH 14).

Rescue PCI versus conservative treatment: Using fixed effect models, rescue PCI compared to conservative treatment was associated with a statistically significant reduction in the risk of major adverse events (OR 0.59, 95% CI: 0.41, 0.86, p=0.006, I² 33%, NNH 14) and re-infarction (OR 0.55, 95% CI: 0.31, 0.96, p=0.04, I² 22%, NNH 25) but a statistically significant increase in stroke (OR 5.11, 95% CI: 1.11, 23.49, p=0.04, I² 0%, NNH 33). OR from random-effects models were significant for major adverse events and stroke but not for re-infarction. There was a statistically significant increase in minor bleeds for rescue PCI compared to conservative treatment groups (OR 5.47, 95% CI: 2.78, 10.76, p<0.001, I² 0%, NNH 9). Only one trial assessed major bleeds.

Moderate inconsistency was found for most comparisons.

Adjusted indirect comparison of RT versus rescue PCI: There were no significant differences between RT and rescue PCI for major adverse events, overall mortality or congestive cardiac failure. Rescue PCI was associated with a statistically significant reduction in the risk of re-infarction (OR 0.32, 95% CI: 0.14, 0.74; p=0.008, NNT 17). There was no significant difference between treatment groups in the risk of major bleeds but there was a statistically significant increase in minor bleeds with rescue PCI (OR 2.48, 95% CI: 1.08, 5.7, p=0.04, NNH 50).

There was a modest benefit for PCI compared to RT after failed thrombolysis.

The review question and inclusion criteria were clearly stated. Several relevant sources were searched and attempts were made to minimise publication and language bias. Methods were used to minimise reviewer errors and bias in the extraction of data, but it was not clear whether similar steps were taken in study selection and validity assessment. Only studies meeting minimum quality criteria were included and study validity was assessed using specified criteria, although only composite scores were reported making it difficult to independently comment on the reliability of the evidence presented. No details of participants were reported which made it difficult to comment on the likely general applicability of results. Appropriate methods were used for the meta-analyses, heterogeneity was assessed and sensitivity analysis was conducted. Indirect comparisons were conducted using appropriate methods. As the authors stated, studies generally had a short duration of follow-up. Evidence appeared to support the authors’ conclusions but, as with all indirect comparisons, caution is require when interpreting results. Incomplete reporting of review methods made it difficult to comment on reliability of the findings.

Practice: The authors stated that for mandated rescue PCI to be effective and timely, a network similar to that for primary PCI needs to be set up.

Research: The authors stated that there is a need for an improved definition of failed thrombolysis. Estimates of the risk benefit/ratio of rescue PCI could be improved by stratifying patients with failed thrombolysis on the risk of stroke and haemorrhage.

Department of Health NIHR BMRC funding scheme, Oxford Partnership Comprehensive Biomedical Research Centre.

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.