Objectives To examine the difference in outcome between percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG), to see if it has changed over the years in diabetics deemed eligible for both treatments; and to contrast the long-term mortality findings with those in non-diabetics.

Design Meta-analyses using data from randomised controlled trials found by searches on MEDLINE, EMBASE and the Cochrane Controlled Trials Register, from their inception until March 2015.

Setting Studies had to be randomised controlled trials comparing PCI with CABG.

Participants Those taking part in the studies had to have multivessel cardiac or left main artery cardiac disease and be deemed eligible for both treatments.

Interventions PCI or CABG.

Primary and secondary outcomes The primary outcome was all cause mortality. Secondary outcomes were a composite of mortality, stroke and myocardial infarction; cardiovascular death; and MACCE (Major Adverse Cardiac or Cerebrovascular Event). The longest follow-up was used in the analysis.

Results Among 14 studies (4868 diabetics) reported over three decades, meta-regression shows no relationship between the year of publication and the difference in long term all cause mortality between PCI and CABG. CABG has maintained an approximately 30% mortality advantage compared to PCI. The other outcomes used showed the same lack of change over the years. These findings held true among insulin-requiring and non-insulin-requiring diabetics. However, among non-diabetics included in the 14 studies, there was no difference in mortality outcome between PCI and CABG.

Conclusions The difference in outcome between PCI and CABG in diabetics has not narrowed from the beginning—with balloon angioplasty to current PCI—with the second generation of drug eluting stents. In contrast to the non-diabetics, there is a persistent 30% benefit in all cause mortality favouring CABG in diabetics, and this should be a major factor in treatment recommendation.