Ian Morrison

We know there are enormous variations in medical care across the United States and even within small geographic areas. Dr. Jack Wennberg and his colleagues at Dartmouth have painstakingly documented these variations over the last 30 or more years. Most of the variations can be explained by two factors: 1) larger supply of providers (facilities and specialists) driving overall higher utilization, and 2) provider preferences in use of diagnostic tests and procedures, where doctors simply do more for a given patient with a specific condition. These two factors combined can result in three- to fourfold variations in utilization and costs without any apparent differences in quality and outcome.

Fresh off the boat as an immigrant in Canada, I got my first health care job as an analyst for a consulting group that served the Vancouver teaching hospitals. One of my first projects (which turned into a seven-year gig as a researcher) was focused on clinical laboratory utilization. My boss was a pathologist interested in a few simple questions: Why do doctors order lab tests, and do they really need them? In a global budgeted world, these were prudent questions. We found a threefold variation in use of lab tests across a whole range of diagnoses (DRGs before they existed).

We presented our data to the clinical chiefs of Vancouver General Hospital: “Ya, but did you adjust for severity?” No, like dolts we hadn’t. Off we went, hat in hand, and spent six months developing a case-mix adjustment index. Same answer: a threefold variation and a 5 to 10 percent per annum escalation in lab use for the same diagnosis. I have called this the “Ya, but” defense ever since: When doctors are challenged about their patterns of utilization they always balk.

From Variation to Tiers

Fast forward to today’s threefold variation in use and costs that is not tied to quality. Tiered networks of hospitals, specialists, primary care, reference laboratories, you name it–are all becoming the rage. From Aetna to Wellpoint, health plans are defining skinny networks where as few as 25 percent of the providers in a particular area are included. The goal of the proponents is to drive patients to these efficient, high-value providers through incentives. However, most doctors and hospitals are busy enough; they may be short of payment, but they aren’t short of patients. If the high-value providers don’t have the capacity to absorb more patients, at least the poor performers might be embarrassed into shaping up and emulating the high-performance folk in their clinical behavior. That’s the theory.

But, like my colleagues in Vancouver 25 years ago, the delivery system will not go quietly. Hospitals and doctors who get placed in an economically disadvantaged tier (where the patient has to pay significantly more out of pocket for care) will bitch and complain. If the tiering is not based on transparent, scientific evidence, the provider system has a legitimate beef. This was the case in St. Louis recently, where United Healthcare hastily constructed a narrow network for General Motors employees that omitted a large proportion of specialist and hospitals, including the large, prestigious and enormously ecumenical Barnes Christian Jewish Health System. Specialists in St. Louis went nuts.

In this case, United screwed up. It doesn’t normally do that; it is usually pretty smart. But under pressure from a desperate GM–which had no other health care options, given the generosity of the UAW contract and the prohibition on cost sharing–network design was the only choice GM had, and United listened to the customer. (Advice: Don’t listen to your customers if they ask you to do something stupid.)

From Tiers to Transformation

Tiered networks can shake up the game in a positive way. But we must be very careful that the tiers are based on evidence, that the evidence is transparent and that it passes the laugh test. You can’t exclude the provider that everyone agrees is the best in town and say that your network is based on quality.

The greatest problem with tiered networks and all the variation research is that the punch line is always the same: We need to re-engineer all clinical care to be higher quality and lower cost. Why? Because you can’t actually move all the patients to the 25 percent best providers. They don’t have the capacity, and the other 75 percent will be pissed.

Similarly, you can’t take all the Medicare enrollees from Florida–where medical care utilization is apparently profligate, excessive, ineffectual and larcenous–and move them to Minnesota for the nice, decent, conservative and effective care that the variation folk tell us exists there. The elderly retire to Florida and other warm places with high utilizing doctors, not to cold, northern Midwest states with conservative physicians. (Remember, before you “Ya, but” me, all these sweeping generalizations are backed by data that is on a per capita, risk-adjusted and quality-adjusted basis.)

The real question is, Who will do the clinical transformation? Dr. Don Berwick and his valiant colleagues at the Institute for Healthcare Improvement are teaching hundreds to make these changes. But we have hundreds of thousands to go.

We are seriously deluded if we expect doctors in ones and twos to spontaneously redesign their clinical practice as a result of exposure to tiered networks. They are more likely to complain their way to inclusion (or actively bamboozle their patients that the tiered networks are meaningless rubbish) without embracing the transformation required.

The clinical redesign requires a set of actors with the organizational, financial and clinical scale as well as the resources to pull this off. Integrated delivery systems, large-scale medical groups and community hospitals that are constructively engaged with their medical staffs are the most likely candidates to lead appropriate clinical transformation that will reduce variation, improve quality and safety, and reduce the rate of cost increase.

Ian Morrison is an author, consultant and futurist. He is also a regular contributor to H&HN OnLine.