Much ado about bundles

Following yesterdays post about bundles, a couple of interesting papers from this weeks NEJM. Two studies here and here, and an editorial about reducing infection in hospitals. The two studies came to different conclusions; the first examined the efficacy of screening for MRSA in ICUs and implementing barrier precautions on all patients until their MRSA status was known. The interventions were not shown to be effective in reducing transmission of MRSA or VRE. However, and this is the really big take home message for me, in cases where hand hygeine or glovers were indicated, the compliance with these measures was as low as 62%!
In contrast, the VA study instituted a bundle approach; a MRSA bundle was instituted in 2007 and over 2 million hospital discharges were studied. The rate of health care associated MRSA infections fell by between 45% (outside ICU) and 62% (inside ICU) over a two year period.
For some expert comment, see this CDC post.

Bundles of joy or of sorrow?

The idea of bundles makes intuitive sense, gather together some evidence based interventions and ensure they are implemented reliably all the time, sit back and watch the magic. Pronovost showed the power of bundles in his NEJM paper some years ago. An observational study by Jarman and others in the BMJ last year suggested that using various bundles in a large London hospital was associated with a large reduction in hospital mortality.
So along come the surgeons to upset the consensus. A study in Archives of Surgery compared standard institutional practice with an extended bundle of interventions, all of which had been previously demonstrated in isolation to be effective. These interventions included:

1. No pre-operative bowel preparation
2. Pre-operative and intra-operative patient warming
3. Supplemental oxygen intra and post operatively
4. Intra-operative fluid restriction
5. Use of a surgical wound protector 

Over 200 patients were studied; main outcome measure was surgical site infection rate at 30 days as assessed by blinded infection control experts. The rate of SSI in the control group was 24%, vs. 45% in the bundle arm. Most of these SSIs were superficial.

Any explanation? Its possible that the interventions are ineffective, or less likely that they are counter synergistic, i.e. although singly beneficial, when combined they counteract one another. Alternatively even positive studies may due to chance be negative. The main concern though apart from giving ammunition to those who wish to reject the benefits of standardisation is that there will be a call for more RCTs in this and every other areas. While we have traditionally believed that RCTs are the gold standard, Ioanniddis work is even questioning this dictum. The other big problem in conducting RCTs is that we assume that apart from the intervention being studied, all other factors are uniform. This is clearly  not the case, and this has proven to be a difficult obstacle to surmount in trials of equipment such as new ventilators.