How many people do those doctors kill?

"14,000 preventable medical deaths"

The proponents of quackery and medical fraud love to tell us about how many people are killed by doctors each year. In the opposition to the NSW anti-quackery committee the number of iatrogenic deaths in Australia has been mentioned several times to make the point that doctors should clean up their own act before trying to do anything about charlatans and pretend doctors. Not only has it been mentioned several times, but it has several values. The ones quoted so far are 19,000, 18,000, 14,000 and 10,000. I have done some investigating to find out where these widely-varying numbers have come from. You will have to pay attention carefully here, but it will be worth the effort.

The figure of 19,000 could possibly be a mistake, as it was only mentioned once and the same person also said 18,000 somewhere else, so we don't have to worry about that one any more. The 18,000 (with a 95% confidence interval of 12,000 to 23,000!) comes from a study published in 1995 (using data from 1992) of hospital adverse events in two states of Australia. (1) The author of that report published another report in 2001 which says that deaths may be as high as 10,000, so it looks like he has rethought his previous research. (2) The 2001 report (which says 10,000 maximum) is cited by several people who mention the number 14,000, although they never provide an actual reference for the paper so anyone can check. (The characters "1" and "4" do not appear together anywhere in the document.) They also say "14,000 preventable deaths" when the paper talks about adverse events and quite clearly says that not all of them were preventable. I am never surprised by quacks lying or acting on the assumption that their readers have no ability to check facts.

To unequivocally illustrate the lie, I will give you the words which are actually used to cite the 2001 report on several web sites:

Iatrogenic Injury in Australia – This is the executive summary of a 150 page official report revealing 14,000 preventable medical error deaths (only in hospitals – not private practice). (Full report on file).

The "Executive Summary" can be seen here.

The real mystery is the 14,000 number. Where did it come from? I first heard it before the 2001 report was published, when someone cited that report as if it existed (people knew that it was coming). The answer is that the original 1995 study, which came up with an estimate of 18,000 iatrogenic deaths per year, involved an examination of 14,000 patient records. (14,000 records examined, 2302 adverse events, 111 deaths, 80% of which were of people aged over 65.) So, in the minds of the quacks, a sample of 14,000 medical records became 14,000 preventable deaths, despite the fact that the author said 18,000 on one occasion and 10,000 on another. Simple, isn't it? But wait, there's more! In June 1995, five months before the original 18,000 (plus or minus 6,000) number was published, a politician issued a press release which said that the study would show a rate between 10,000 and 14,000, and a newspaper reported the release. This apparently makes it legitimate to use the number 14,000 when citing a paper that says 10,000.

By the way, there were 133,707 deaths from all causes combined in Australia in 2002. The likelihood that doctors are killing, through negligence or error, half as many people as die from all cancers combined or 150% as many as die of stroke is ludicrous. Whatever it is it is too high, but it certainly isn't 18,000.

References:

1. Wilson RM, Runciman WB, Gibberd RW, Harrison BT, Newby L, Hamilton JD. The Quality in Australian Health Care Study. Med J Aust 1995; 163(9):458-71.

2. Runciman WB, Moller J. Iatrogenic Injury in Australia. Adelaide: Australian Patient Safety Foundation Inc. 2001 (Read the report here)

Update 29/4/2012 – Australian Bureau of Statistics figures run a year or so behind the calendar. The number of deaths in the country in 2009 (the latest year available in April, 2012, was 141,070. This larger number has no significant effect on the comments made in my 2002 article.

You can see more about this here.