Agraphia Medical Tragicomedy


Zac Fact #16 (In The Land Of The Blind…)

I'm not a statistician by any means, but I will say an example from High Yield Biostatistics by Dr. Glaser really caught my attention. This stuff is pretty confusing, so I'm going to simplify his numbers a bit further.

There was a study called WOSCOPS that treated obese men with statins. WOSCOPS was a huge study, involving 6,000 men for 5 years. 3,000 of them were given the statin, and 3,000 were given placebo.

Over the course of the study, about 75 men in the placebo group died from heart attacks, while only 50 died in the statin group. We calculate the mortality rates for each group as 50/3000 = 1.6%, and 75/3000 = 2.5%.

Calculating the relative risk reduction is easy:
1 - (risk in the population of interest / risk in the population as a whole)
1 - (1.6%/2.5%) = 33%
Therefore, the WOSCOPS study reported an incredible 33% reduction in cardiovascular mortality from statins. Amazing! Cardiologists everywhere rushed to put their patients on statins, right?

Not so fast.

The absolute risk reduction is calculated by subtracting the 2 percentages. 2.5% - 1.6% = 0.6%. We then calculate the number needed to treat as 1/ARR. This comes out to 166... which means we would have to give 166 men statins for 5 years to prevent one of them dying from a heart attack. This reflects the fact that the heart attacks happened at a relatively low frequency in the study.

Multiply 166 people x $100 a month (the cost of the statin) x 60 months in the study... and you get roughly $1,000,000.

It costs $1,000,000 to buy enough statins to save one life over a five year period. This is why healthcare costs so damn much.

Comments (3) Trackbacks (0)
  1. Zac…I just want you to know that I read this post the night before my test, and I totally had a question about number needed to treat…which neither of our books went over! So, thanks!!

  2. Pretty good biostats analysis buddy!

  3. You typed the numbers into your calculator wrong. 2.5 – 1.6 = 0.9

    NNT = ~ 111

    P.S. Question: was there any decrease in morbitity noted in the study?

Trackbacks are disabled.