Monday, 3 October 2011

Reckoning with risk

Having studied statistics for years and always struggled, this book makes the point over and over again that most other people have the same problem. As a clinician wanting to explain risk to patients, this book has been a very useful nudge and Gerd Gigerenzer repeatedly demonstrates that explaining risk using frequencies rather than probabilities means more people understand what's going on - both doctors and patients.

A simple thing I picked up was indicating to patients the risk of failure of a procedure. For example, one could say that a molar treated tooth has a chance of success of 80-90% over, say, 10 years (this is just an example figure!). A patient, though, may well understand the following better: Out of ten molar teeth treated with endodontics, 8 or 9 will be successful after 10 years.

More complex ideas are dealt with using frequencies too. For example, if a male heterosexual in a particular area gets tested for HIV and it turns out it's positive, what is the likelihood that he is indeed HIV positive? Here, the prior probability comes into play - something I think we pay little attention to most of the time (susbstitute oral cancer if you're a dentist). The book explains how, despite the HIV test having a sensitivity of 99.9%, the chance of that man actually having HIV is 1 in 2. i.e. there's an equal chance he isn't HIV positive despite having the positive test (sorry for not explaining how this occurs but I won't explain it nearly as well as it is explained in the book!).

If you've read about prior probability, absolute risk reduction, risk ratios and similar and wondered what's going on then this book guides you in a gentle manner to better understand the risks and benefits of doing things and not doing things.

No comments:

Post a Comment