I worked in health insurance, and had a graduate degree in health policy, before I second-careered into medicine. I helped implement some of the programs that Guwande promotes in this article; I was and am a true believer in some of these programs, and have built my medical career around implementing them properly.
I say all that to provide an important caveat: Guwande's a Harvard-associated surgeon largely insulated from what any of this looks like on the ground, and basically collects anecdotes that match his views until he can tell a nice just-so story. The reality is far more complicated than he relates, and shared savings programs are far from some sort of medical panacea that addresses all the imbalanced and chaotic incentives in medicine.
A few big nitpicks:
(1) The ACA Shared Savings program is and was bullshit for providers. Because it was built on top of "traditional Medicare", meaning a non-HMO arrangement that did not infringe on pt's right to seek as much care as they wanted, from whomever they wanted, there was no actual assignment of responsibility. If Doctor X is to share in the profits of saving the system D dollars, you have to be able to measure the savings he generates. However, no patients are "his" - we just measure which PCP patients go to the most in a year and give the doc a benchmark based on those. This was based on an assumption that most people saw their PCP at least once/yr. Turns out that not only do people not see their PCP once/yr, but plenty are snowbirds, or have a rotating list of multiple PCPs. Medicare has not fixed the attribution issue.
(2) The most effective way to create savings remains to deny care. There's a reason there was an HMO backlash. Creating that incentive reduces costs/spending, but ...
(3) The vast majority of physicians do not work in pure capitated programs. They work in a mixture of shared downside, shared upside, and FFS. You can't manage patients with a mix of payors like this, because you still have to build your daily operations around the largest volume - the FFS folks. This is what I saw the most of in the rollout of the ACA: docs were "enrolled," sure, but ACO's still pay out on top of FFS, and the rest of the patients are normal FFS + shared risk, the result is daily operations built around volume, not around "spending 45 minutes to review how to take insulin."
4) Docs over-test and overprescribe for medicolegal issues that no amount of incentive-shifting can fix. Hell, my hospital antibiogram doesn't even have pseudomonas sensitivity to pip, just pip/tazo, despite the fact that pseudomonas pretty much never needs pip/tazo (the /tazo is a spectrum extender). But prescribing pip/tazo over pip is something akin to "no one ever got fired for buying IBM." Hell, we do baseline kidney function tests before giving contrast in emergency radiology, despite the fact that the biggest studies to date show that baseline kidney function doesn't have any predictive value for contrast-induced nephropathy-related outcomes 6 mos out. It's a pointless delay that actually hurts patients and costs money. But you know what? No one sues a doc for getting worse in an ED while awaiting a lab result - that " just happens". They do sue a doc for contrast-induced nephropathy, because that looks like something a doc actively caused (well, it is) - and a malpractice attorney will slam your balls to the wall for "why didn't you look at his kidney function before throwing kidney toxins at him?"
The malpractice thing is honestly more insidious than that, even. It gets docs out of the mindset of thoughtfully asking "well, WHY would I order that test? How does it advance the diagnosis or management meaningfully?" to a "might as well order it to be on the safe side" mindset that infects the rest of their practice. The latter creates sloppy, over-broad testing regimens.
Heck, we can't even utilize the latest research unless we can convince multiple departments worth of docs to agree on it. Because the standard for malpractice is what your peers would do, "I did X as part of our department standard, as agreed upon by all the docs in X and related specialty Y" is a very strong malpractice defense. If you can't convince all of those people to get on board and change department policy and just try to practice based on the most up to date data yourself, a malpractice attorney's question becomes: "Where did you get your PhD in statistics and clinical trial design? Oh, you don't have one? So what made you qualified to take in this study and overrule a panel of national experts on the topic that decided what the standard practice is?"
We really need an iatrogenesis compensation fund built at the national level, and to get rid of malpractice suits. That will bring down bullshit spending by a fair degree right off the bat.
A related issue that people overlook is that about 70% of the growth in HC costs is attributable to technological advance: new drugs, or hiked up drugs, new tech, etc. We want all these things, but refuse to take into account that they demand a premium (even when they're not actually more effective.)
I could rant for ages.