The DMCB explains one of them.
It recalls a very elderly patient (let's call her Mary) who was desperately ill with a bad infection. Before coming in the hospital, Mary was mildly confused, had multiple chronic conditions and rarely left her apartment. The DMCB thought she was a goner, but thanks to prompt, evidence-based and very expensive therapy, Mary not only avoided the ICU but eventually walked out of the hospital. Mary is important because her admission to the DMCB's hospital service was bracketed by similar patients (let's call them Bob and Alice) who were not so lucky.
Given the prevailing metrics of our evidence-based age, Bob and Alice would be considered to have received wasteful care, while Mary would be considered a success. The problem for the DMCB is that that definition of waste can only be made in retrospect. When it was standing outside Mary's, Bob's and Alice's hospital rooms, it didn't know who was going to survive.
Consider the "crumbling sand pile" thought experiment made famous by Nassim Taleb's The Black Swan. While it's very possible to predict the general size and shape of a pile made by the one-at-a-time addition of sand grains, it's impossible to predict when it will reach a critical mass and where or how big the avalanche will appear. However, once it happens, it is possible to look backward and reconstruct the event.
And so it was with the DMCB's three patients. Survival for many conditions such as infections near the end of life is surprisingly random and unpredictable. Retrospective reconstruction of events (even with an autopsy) isn't of much use for the next patient, even if the hospital is littered with "avalanches" that can be categorized as wasteful failures.
Which is why the DMCB wished Dr. Berwick had long ago stopped repeating the simplistic 30% waste canard, especially when he is smart enough to grasp the shortcomings of retrospectively measuring waste. While there are plenty of opportunities to reduce costs and increase quality, too much of the science is based on flawed and backwards logic. Go into any hospital on any given day and doctors will ask this about their care for their patients:
Which ones won't benefit and which ones will?
Image from Wikipedia