On 2014 Feb 19, james brophy commented:

Prompt revascularization of the culprit lesion with percutaneous coronary interventions (PCI) is a cornerstone in the treatment of acute myocardial infarction (STEMI). Numerous previous studies have suggested that simultaneous additional interventions of other non-culprit is at best unnecessary and at worst potentially dangerous. However this study that compared simultaneous ‘preventive angioplasty’ of non-infarct-related coronary arteries with ≥50% stenosis with primary PCI alone (with optimal medical treatment) found the composite endpoint of cardiac death, nonfatal myocardial infarction, or refractory angina was reduced by a staggering 65% (HR=0.35; 95% CI: 0.21-0.58, p<0.001). Is this result believable?

First, despite a very slow recruitment period of several years, the trial was prematurely stopped. Premature trial discontinuations for benefits are well known to overestimate treatment effects. This is especially problematic in this small unblinded study with only 16 excess 'hard' events and no formal statistical stopping rule described in their study protocol. The strongly positive results in prematurely stopped small trials are often not confirmed in bigger follow-up studies underlining the important role of chance that arises with repeated looks at the data. Moreover it has been shown that these small prematurely stopped trials have an increased likelihood of being published in high profile journals reinforcing this tendency to early stopping, as in the present case. Second, can medical treatment really be considered optimal in the control group when, despite STEMI and multi-vessel disease patients were discharged in less than 2 days without ischemic testing, the current standard of care (if additional angioplasty has not been performed during the original hospitalization). It seems only 1/3 of patients ever got any ischemic testing and this occurred only several weeks after the acute event, perhaps to late to influence potentially avoidable outcomes. Consequently was the “preventive” up front PCI group not being compared to an inferior "straw man" comparator? Third given the unblinded nature, could not the knowledge of untreated lesions have encouraged a detection bias in the search for clinical outcomes.

In an accompanying editorial, it was proposed that these additional interventions helped to pacify the inflamed coronary vasculature. However this argument seems specious as the coronary vasculature is diffusely and not locally inflamed. The biases discussed above seem a more likely explanation for this large and unexpected effect size.

On 2014 Feb 19, james brophy commented:

Prompt revascularization of the culprit lesion with percutaneous coronary interventions (PCI) is a cornerstone in the treatment of acute myocardial infarction (STEMI). Numerous previous studies have suggested that simultaneous additional interventions of other non-culprit is at best unnecessary and at worst potentially dangerous. However this study that compared simultaneous ‘preventive angioplasty’ of non-infarct-related coronary arteries with ≥50% stenosis with primary PCI alone (with optimal medical treatment) found the composite endpoint of cardiac death, nonfatal myocardial infarction, or refractory angina was reduced by a staggering 65% (HR=0.35; 95% CI: 0.21-0.58, p<0.001). Is this result believable?

First, despite a very slow recruitment period of several years, the trial was prematurely stopped. Premature trial discontinuations for benefits are well known to overestimate treatment effects. This is especially problematic in this small unblinded study with only 16 excess 'hard' events and no formal statistical stopping rule described in their study protocol. The strongly positive results in prematurely stopped small trials are often not confirmed in bigger follow-up studies underlining the important role of chance that arises with repeated looks at the data. Moreover it has been shown that these small prematurely stopped trials have an increased likelihood of being published in high profile journals reinforcing this tendency to early stopping, as in the present case. Second, can medical treatment really be considered optimal in the control group when, despite STEMI and multi-vessel disease patients were discharged in less than 2 days without ischemic testing, the current standard of care (if additional angioplasty has not been performed during the original hospitalization). It seems only 1/3 of patients ever got any ischemic testing and this occurred only several weeks after the acute event, perhaps to late to influence potentially avoidable outcomes. Consequently was the “preventive” up front PCI group not being compared to an inferior "straw man" comparator? Third given the unblinded nature, could not the knowledge of untreated lesions have encouraged a detection bias in the search for clinical outcomes.

In an accompanying editorial, it was proposed that these additional interventions helped to pacify the inflamed coronary vasculature. However this argument seems specious as the coronary vasculature is diffusely and not locally inflamed. The biases discussed above seem a more likely explanation for this large and unexpected effect size.