Sunday, 15 April 2012

Do the numbers lie?

I hope you all had a nice Easter. I hadn't meant to be offline for so long, but you know how it is when you go away for a week and put your feet up: getting anything at all done can feel like it needs a superhuman effort.

Who knows, perhaps the rest will have recharged my creative juices, and leave you lot in line for some stellar commentary. Perhaps, but don't get your hopes up.

One of the stories that caught my attention last week was the news report that a high number of elderly patients are discharged from A & E in the middle of the night. This seemed to raise a few heckles, but the problem with this data was that it failed to inform us of whether there is actually a problem at all. The news report I heard acknowledged that with the way the data is collected, the category 'discharged patients' encompassed a range of scenarios, including patients who had died.

Sending a frail, elderly patient out of an Emergency Department to get a taxi home, without ensuring appropriate care arrangements is unacceptable, but we didn't need this data to highlight that. What was revealing about this particular news story was what it tells us about how we use information, and the lack of care that we take in interpreting it.

Let me explain - numbers tell a story, but the nature of that story is heavily influenced by the context that it is used. For example, hearing that 10,000 elderly patients (these are made up numbers) are discharged from Emergency Departments after midnight without appropriate care arrangements will be interpreted differently if one later finds out that of those 10000 'discharges', 7500 were patients who presented to A&E and then died.In this fictional scenario, without knowing more detail, one cannot know whether patients are being sent home inappropriately, or whether patients are experiencing excess mortality late at night in Emergency departments.

My suggestion is that we tend to use these raw figures as they are presented to us as a means of drawing our conclusions. However, without further work, such figures are rarely a suitable basis on which to base your opinions. The alternative, and more circumspect, use of data is to use it to form your further enquiry, and not to draw conclusions.

This issue is dealt with in a different guise in this week's New England Journal, where one of the topics of discussion is the current focus in American hospitals on readmission rates. On face value, it could seem as if a hospital with a high readmission rate is offering low quality care, as patients must be representing because their primary medical problem is not being dealt with properly the first time round.

This automatic conclusion needs to be challenged. For example, what if a hospital experiences really high readmission rates because patients' social care needs are not being met after they leave hospital? Or what if another hospital has a much higher survival rate that other hospitals for some conditions? The patients who survive are liable to add to the readmission rate, because where other patients didn't make it, they survived and went home, and are therefore in a position to be readmitted, whereas the patients who went to other hospitals are not.

So when someone gives you some figures that paint a dramatic picture, pause for a minute, challenge your instincts, and ask yourself if there is another way that this information can be interpreted.

No comments:

Post a Comment