In an article by Peter Boyle et al. in UROLOGY 48 (3) 1996 it is argued that the data on peak flow rates and Q-IPSS improvement scores gives evidence for finasteride providing improvement for urinary flow and Q-IPSS index, increasing with the size of the prostate. Below we have reproduced the two figures from their paper, on which their claim is based.

Fig 1 shows that there is no improvement in flow for the placebo group with prostates larger than 40 ml and a significant improvement for those with smaller prostates. The average improvement for the three low volume group points is about 0.8 +/- .3 and for the larger points show a value of about .1 +/- .25. This is a remarkable new result, showing that a placebo helps those with small prostates. On the other hand the drug users show a dependence of the change in flow rate that is essentially independent of prostate size. If one subtracts the two curves one gets the result that the improvement in rate grows with prostate size. Of course the real answer is that the difference is due to the strange placebo curve not to the larger improvement with the prostate size!

The same effect appears, but not as clearly, in Fig. 2 which shows the plots for the Q-IPSS index vs volume.

This illustrates what appears to be missing in many reports of medical results on drugs. There is always reporting of statistical errors but little attention to unexplained systematic errors which clearly result in incorrect conclusions.