Editor’s blog Sunday 30 May 2010: Weekly publication of hospital infection rates ahoy!

Kingsley Amis once famously wrote in July 1960's Encounter magazine about higher education, "more will mean worse".

If you read this website regularly, you already know that such a belief about healthcare is not at home here.

And yet, and yet ... nevertheless it sprung briefly, heretically to mind when reading this report that the government intends to publish hospital infection rates.

So what could possibly be wrong with putting performance information into the public domain?

This world - not best of all possible ones
The article suggests that "Andrew Lansley, the Health Secretary, will announce on Wednesday that hospitals are publishing weekly data on superbug rates on Government website www.data.gov.uk".

What could possibly go wrong with weekly publication of data?

Well ... some things could. Especially if the many data quality and sampling caveats that can be taken for granted among real healthcare statistical experts (such as I do not pretend to be) cannot be taken as read.

For instance: let us supp[ose that Hospital A has a local nursing / residential care home in the not-for-profit or (more probably) independent sector. Perhaps because of poor staff practice (or maybe because of cost-cutting), there is a major outbreak of Clostridium Difficil among residents. Many of the sick are admitted to Hospital A - perhaps in half of the cases, the infection cannot definitely be diagnosed as occuring prior to infection.

What have we learned that is useful or value-adding about Hospital A's HAI figures from that week?

For instance: the Great British Public remain charmingly surprised to learn that half of all hospitals are below-average performers. Many if not most journalists (with certain honourable exceptions, whose names often grace these pages) are statistical ignoramuses.

I am (more or less) one, happy only in that I know how much I don't know as a result.

With these caveats stated, I still probably tend towards thinking that publishing this information  is a good idea. Data only gets better if it is used regularly, and in much closer to real-time than is currently the NHS tradition.