Headline News vs Truth in Numbers

If it bleeds, it leads.  This dictum drives much of the modern news media, shapes what is regarded as news, and slants the coverage of everything you see, hear, and read.  Going by the news, you would think that many US cities resemble the towns depicted in old-time westerns, and that hospitals are slaughter houses. ‘Medical accidents kill hundreds of people every day!’ is the inescapable conclusion of anyone who reads the hype in the media that followed, and continues to follow, the fabled IOM report.  ‘It’s like a 747 crashing every day’ is a common refrain in these discussions. 

A colleague pointed me to this article recently:

http://www.dailymail.co.uk/news/article-2090332/Four-patients-die-thirsty-starving-EVERY-DAY-hospital-wards-damning-new-statistics.html?ito=feeds-newsxml

Once again, we have a news item that casts the NHS in a very poor light.  But is this a reasonable interpretation of the data?  The short answer is no.  Almost every hospital in the US admits patients with dehydration every day.  Dehydration is a common feature among elderly patients who fall ill in a nursing home, and is almost invariably accompanied by urinary tract infections/urosepsis, mental status changes from their previous baseline, and bedsores.  The mortality rate of patients admitted with these problems is high.  Attribution of cause of death is a social construct, even when an autopsy has been performed (the certainty of television shows not withstanding).  Most practitioners would represent that these patients died of urosepsis and their underlying condition. Most would list the dehydration as a contributing cause, but not the major cause.  If you read the above item from the Mail Online carefully, you see that a much smaller number of patients are alleged to have died from dehydration while in hospital, the majority (>70%) seem to have it listed as a contributing cause.  My guess is that there is a coding error here, and that at least a few, and likely a majority of the patients who are alleged to have died of dehydration died of some other cause. The database itself is somewhat suspect, because I would contend that admission with dehydration occurs more commonly than this report represents.  The British numbers, are too low to represent an accurate accounting of the patients admitted to their hospitals with dehydration, and simultaneously overestimate the number that die with dehydration as the primary cause. More than anything, the Mail Online story is a lesson in the perils associated with using database information for making any kind of inference.

The commentary that follows this article is informative.  There is clearly a lot of frustration with the NHS, or there are at least a lot of people willing to make comments in the Mail Online about this problem.  I do wonder how the comments would read about a similar story in the USA.