Back in 2005, Phillip Longman wrote an article in the Washington Monthly, touting the strides the Veterans Administration had made in improving quality.  Since then, it has become the model for a fair number of reformers, who frequently cite its ability to control costs and coordinate care as proof that we should move towards such a system nationwide.

I was thus interested to learn that the Congressional Budget Office had issued a report on the VA's quality initiatives.  I've been meaning to get around to reading it for weeks, but I've been sidetracked by other things.  (To judge by an internet search, I'm not alone; few people seem to have blogged or linked to it.)  But I finally have, and I think I can say two things definitively:  the report suggests that the VA really has made outstanding improvements in the quality of care they deliver.  And the report also suggests that the VA does not represent any sort of workable model for the US health care system.

The advocates of a VA-style system usually have three core arguments:

  1. The VHA provides better care because it has centralized records (VistA) to coordinate treatment and reduce errors
  2. The VHA provides better care because it will care for a patient their entire life, and therefore has incentives to manage diseases rather than mindlessly do procedures
  3. The VHA can reduce costs through administrative efficiency, central procurement, integrated care, and salaried doctors

For example, Longman says:

So what's left? Consider why, ultimately, the veterans health system is such an outlier in its commitment to quality. Partly it's because of timely, charismatic leadership. A quasi-military culture may also facilitate acceptance of new technologies and protocols. But there are also other important, underlying factors.

First, unlike virtually all other health-care systems in the United States, VHA has a near lifetime relationship with its patients. Its customers don't jump from one health plan to the next every few years. They start a relationship with the VHA as early as their teens, and it endures. That means that the VHA actually has an incentive to invest in prevention and more effective disease management. When it does so, it isn't just saving money for somebody else. It's maximizing its own resources.

The system's doctors are salaried, which also makes a difference. Most could make more money doing something else, so their commitment to their profession most often derives from a higher-than-usual dose of idealism. Moreover, because they are not profit maximizers, they have no need to be fearful of new technologies or new protocols that keep people well. Nor do they have an incentive to clamor for high-tech devices that don't improve the system's quality or effectiveness of care.

And, because it is a well-defined system, the VHA can act like one. It can systematically attack patient safety issues. It can systematically manage information using standard platforms and interfaces. It can systematically develop and implement evidence-based standards of care. It can systematically discover where its care needs improvement and take corrective measures. In short, it can do what the rest of the health-care sector can't seem to, which is to pursue quality systematically without threatening its own financial viability.

The analysis is theoretically appealing, but there are a couple of problems for it.  First, on the theory side:  it's not clear what incentives the VA has to become more cost efficient. Finding ways to save money in a government bureaucracy is not generally rewarded.  Finding ways to save money usually means you get your budget cut, while cost growth can provide an argument for a bigger appropriation. And when the VA runs out of money, it can reallocate services between priority levels, cutting some of its customers off in order to cover higher priority patients.

Which brings me to the empirics:  the VA does not, in fact, provide lifetime care for its patients practically from kribbe to grav.  The VHA works on Priority Groups, which categorizes who is eligible for what treatments.  Some of the categories are rather fun:

Veterans in priority group 6 served in World War I or the Mexican Border War, are seeking care solely for disorders associated with exposure in the line of duty to chemical, nuclear, or biological agents (including, for example, Agent Orange), have compensable SCDs [Service Connected Disabilities] rated zero percent disabling1, or are within a five-year period of special eligibility for recent combat veterans.

The footnote goes on to explain that this covers a handful of vets receiving payments for tuberculosis, "special monthly compensation under 38 U.S.C. 1114(k)" or other disabilities.  And here is 38 USC. 1114(k):

if the veteran, as the result of service-connected disability, has suffered the anatomical loss or loss of use of one or more creative organs, or one foot, or one hand, or both buttocks, or blindness of one eye, having only light perception, has suffered complete organic aphonia with constant inability to communicate by speech, or deafness of both ears, having absence of air and bone conduction, or, in the case of a woman veteran, has suffered the anatomical loss of 25 percent or more of tissue from a single breast or both breasts in combination (including loss by mastectomy or partial mastectomy) or has received radiation treatment of breast tissue, the rate of compensation therefor shall be $89 per month for each such loss or loss of use independent of any other compensation provided in subsections (a) through (j) or subsection (s) of this section but in no event to exceed $3,075 per month; and in the event the veteran has suffered one or more of the disabilities heretofore specified in this subsection, in addition to the requirement for any of the rates specified in subsections (l) through (n) of this section, the rate of compensation shall be increased by $89 per month for each such loss or loss of use, but in no event to exceed $4,313 per month;

Pardon me for a moment while I meditate on the notion that we must have more government involvement in our health care in order to bring some rational, efficient order to our broken system.

But I digress.  The reason I bring all this up is that none of these groups, according to the CBO, gets as much as 50% of its healthcare through the VA:


This makes any sort of statements about its relative quality deeply problematic.  If your patients are only coming to you for the things that you happen to be better at than their alternative sources of healthcare (private insurance, Medicare, Medicaid, etc), then your results will be deeply skewed.  Comparisons were already made difficult enough by the fact that veterans are probably substantially different from the population at large--no matter how relaxed the recruiting standards were in the 1970s, the very fact of making it into the military rules out a number of serious underlying health problems.  And their prospective patient population got a great deal more educated and healthy since then.

So the problem you have to contend with is that if your patient population is different from the general population, and your patients almost all use a mix of your services and other providers, most measurements of your quality may well be picking up the quality of the other helath care they have access to.  Or the quality of the patients themselves.

In fact the CBO specifically says that this is a problem with Longman's analysis:

Advocates have claimed that because VHA is free from concerns about generating profits from medical services and faces at least part of the long-term costs associated with chronic diseases, the agency has an incentive to invest in preventive care, coordination of services, and quality improvement26.  However, data on the way in which veterans use the system make it clear that most enrollees also rely on other sources of care for a significant portion of their health care needs.

Footnote 26 goes to Longman's book, which was based on the 2005 article.

This is also a problem with analysis of cost control.  The VHA controls its costs in a number of ways, but one of them is managing its priority lists--making it harder for those who are lower priority to get care.  For example, thanks to the influx of war veterans, the VA in 2003 closed its lists to many higher-income people who didn't have service-connected disabilities--AKA Priority Group 8, which as you can see, isn't getting much of its health care from the VHA.   

So in some sense, you are trivially controlling cost per enrollee, but you're not necessarily doing it by delivering better care at a better price; you may just be providing less care for some enrollees.  But that doesnt mean that total national healthcare spending on those people goes down.  By contrast, most people with private insurance do most of that spending through their insurance.

Furthermore, the changing mix of the priority groups can dramatically effect how much you need to spend.  Much has been made of the VA's ability to control costs, but according to the CBO:

Some proponents of the veterans' health system have suggested that VistA has helped the Veterans Health Administration hold down cost growth when compared with other federal health programs, such as Medicare.  But such comparisons are difficult to make.  The substantial changes in the VHA's structure and in eligibility for care make it particularly difficult to interpret usch metrics as cost per enrollee when enrollment was rising dramatically from 1999 through 2002.  In this assessment, the Congressional Budget Office (CBO) adjusted enrollment data to account for changes in the mix of enrollees adn found that VHA's spending per enrollee was relatively flat from 1999 through 2002, but since that date it has risen about as rapidly as spending per enrollee in the Medicare program.

This cuts both ways, of course--I imagine the influx of new vets is quite expensive.  The point is, you can't compare a system that manages its costs by allocating a set budget among a shifting group of people, few of whom actually depend on the VA for all their healthcare, with any other system.  

We can't do anything like what the VA does on a national scale--not to manage our costs, not to deliver our services.  A system in which people got 20-50% of their care from the government, and the rest from some other provider, would not be a good system--especially because while applauding VistA, the CBO also points out that it can make data sharing between systems very difficult.  

That's not to say there are not lessons we can learn from the VHA.  Just for starters, the conversion to electronic medical records is long overdue, and we ought to be using the Medicare payment system to herd doctors and hospitals faster towards adopting systems that are interoperable and meet certain broad standards.  But whatever the merits of the system, it simply doesn't tell us much about how to design some sort of comprehensive government health care system.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.