Associations and Clinical Significance in Meta-Analysis: When are the Deductions Too Presumptive?

This meta-analysis of publications over an approximate ten-year period from 2002-12 looked at mortality after coronary presentations. A total of 487 reports were studied and 48 were assessed for the impact of nominal ‘off hours’ (night and weekends) presentation for catheter-based revascularisation for ST elevation myocardial infarction (STEMI). Smaller studies tended to show increased mortality off hours, larger samples little or no adverse effect. The net impact, using a random effect meta-analysis model, was a statistically significant 2-4% mortality excess. This was linked by the authors to an increase in hospital ‘door to balloon’ times.

Data on time to recognition and hospital presentation were not available but were said to be shorter off hours as seen in other publications. More adverse effects were seen off hours in non-North American centres and was associated with a reduced door to balloon time in North American registries.

The authors go on to suggest these observations have implications for the assessment of quality of care by hospitals (and staff) and should guide ‘value based purchasing’ on the basis of off hours performance. The limitations - lack of randomisation, inconsistent definition of ‘off hours', high heterogeneity, publication bias, overlapping patients - were all felt to be offset by the statistical power of analysis (more than one million patients from the 48 publications included). The authors considered that previous failed attempts to confirm the adverse off hours effect on mortality were due to a lack of statistical power.