Posner does make the mistake of comparing the status quo with utter anarchy; since no one proposes to completely abolish medical malpractice liability, for example, he of all people should know that this is a question of how well the tort system performs at the margin. Does the US tort system produce safer products and better medical care than European systems that cost half as much? I haven't seen any evidence of that; indeed, the Rubin-Shepherd study implies that at the margin, the tort system's effects on deterring safety innovation and medical practice adversely affects public health and safety. (That said, one can fairly quibble with the way the PRI study incorporated the Rubin-Shepherd results into their final number.)
I'm not completely persuaded that one can extrapolate Kessler and McClellan's defensive medicine findings in the cardiology context to all medicine, but Posner makes a mistake in thinking that all defensive medicine has at least some offsetting benefits at the margin. Not so. The cardiology procedure Kessler and McClellan measured was chosen by the two economists precisely because it was a procedure that had no measurable effect on mortality, and other defensive medicine procedures such as CAT scans and biopsies may well do more harm than good at the margin even aside from the wasted resources.
Posner is correct that the social loss of products from the marketplace is a loss of consumer surplus, and that it is imprecise to simply measure the value of the lost sales: but one cannot definitively say that total consumer surplus is more or less than the value of that loss of products. If one starts to consider the third-order effects of jobs lost because of the lost sales, the number from lost innovation can easily be larger than the number PRI came up with. For better or worse, we measure the size of our economy by measuring gross national product rather than consumer surplus, and PRI did nothing wrong in using a similar metric.
Update, May 2: PRI responds.