Discussions of health care reform tend to focus on who pays for care. This book is a much-needed reminder to address an even more important question: Why are we struggling to pay for treatments and drugs that may not be accomplishing anything?
The authors walk us through a discussion of risk with detailed examples and illustrations. Sure, it's a little simple, but not everyone has studied statistics. I've had graduate-level stats courses and I found the discussions helpful and enlightening.
What's really scary is that we're exposed to hype in news reports, which often seem to come directly from press releases of the pharmaceutical companies. I wonder how many MDs read these statistics without understanding what's going on.
Even worse, we're getting propaganda from medical institutions. The authors show a misleading flyer from the prestigious M.D. Anderson Health Center in Houston.
My favorite part of the book is the discussion on survival rates. If you're diagnosed early you may not get an extra day of life. You just live with the knowledge longer.
I can't help wondering if the millions of dollars we're spending on drugs claiming to lower cholesterol and reduce hypertension might not be better spent on healthy food, exercise and stress reduction. As the authors point out, we need evidence that people with better "numbers" really live longer and experience less suffering. We also need evidence that these drugs really contribute to meaningful outcomes, not just lower numbers.
Just this morning the Wall Street Journal solemnly reported a drug that promised to lower "prostate cancer risk" by 23% among a large sample of high-risk men. Following the guidelines of this book, it was easy to spot flaws. The difference between the placebo group and the drug group was just 6.5%, not 23%. In other words, out of 1000 men, 65 seem to have been spared the diagnosis - not 230. Is that a big number? The authors advise, "It's up to you."
The authors warn us to look with skepticism at promised outcomes. For instance, "shrink the tumor" doesn't always mean "reduce risk death by cancer." "Increase bone density" doesn't mean "avoid hip-fracturing falls."
In this article, the outcome was "diagnosed with prostate cancer," not "death from prostate cancer." If many of the men were 70 or over, it's possible that they would end up dying *with* prostate cancer, as opposed to dying *from* prostate cancer.
The only point I'd add (and I may have missed it in the book) is that extremely large samples can lead to misleading results. When you have huge samples, you can get significant correlations by chance. A study of hundreds of thousands sounds impressive but you need to look more closely.
Everyone needs to read this book, especially consumers of the medical system, legislators and regulators.