I’ve blogged before about compliance. It’s a big thing for me. If I had a pound (actually after the last week, if I had a dollar) for every time that I think I’ve implemented some intervention to find after a while that it has not been embedded I’d be on a yacht in the Med. But I’m not, instead I’m reading the very nice meta-analysis of the effectiveness of bundles in preventing CLABSI recently published in the Lancet ID. Ok, so the conclusion is that bundles work, but it’s not that which interested me, as a glance at the figures made me consider whether we should move on from effectiveness to implementation.

When I look at the forest plot a quick calculation shows that roughly 37% of the included studies did not find a significant difference. This could mean that confidence in the bundle approach could be less than 100% and give wriggle room to the naysayers. My question is, having experienced ‘implementation’ of a surgical site infection prevention bundle (with a subsequent rise in infection rate!) whether the bundles were well implemented (in our case, not, as only one in five got the bundle).

The review authors described compliance with the bundles in their analysis, reporting that this was determined (or reported) in only 24% of pre-and post-implementation and in a further 11% post-implementation studies only. That leaves 65% of papers not reporting on compliance at all and in the ones that did report, it was suboptimal in every one. To me, this means that the positive effect that we see is the absolute minimum possible positive effect and I wonder what better compliance would have shown. The studies that did not report a significant effect of a bundle may have been those that did not look at (or report) compliance. Is it time for reviewers of bundle intervention papers to ask authors to include compliance and a discussion of the results of compliance monitoring in the reports?

The authors finish with a call for studies on protocol compliance instead of yet more effectiveness studies. I totally concur, since we already have the 79 included in this review! The evidence is clear; that bundles work. We should now concentrate on how best to implement and ensure that the vulnerable in our care get optimal, evidence-based interventions on every occasion. I think that sometimes we (and almost always our clinical colleagues) think that we already do this. This review demonstrates just how far we are from being in that happy state.