I am speaking at the U.K. Parliament next week about how to get better outcomes from addiction treatment. Like virtually all other forms of health care in the U.K. and U.S., addiction treatment is under pressure to deliver better outcomes without an increase in budget.
A number of projects (such as this one) have successfully improved aspects of the process of addiction treatment, e.g., reduced waiting times, increased use of evidence-based counseling methods and incentivized staff to retain patients longer in care. However, these process improvements have rarely translated into significant product improvements. That is, patients are seen more promptly, their treatment is better planned and organized, they stick around longer and they are more satisfied with their care, but their likelihood of recovering from addiction stays roughly the same.
In response to my recent post on hospital readmissions, some commenters suggested that psychiatric and addiction medicine are different than the rest of health care in that factors outside of treatment (e.g., housing, social class, family stress) rather than care quality explain most of the variance in how well patients do over time. But “hard medicine” is largely in the same boat: A study of Medicare’s quality of care measures for how hospitals dealt with heart attacks, chronic heart failure and pneumonia found that even dramatic differences in the quality of care during hospitalization relate only weakly to post-discharge death rates.
Many scholars committed to health care quality improvement would argue that it is hard to measure quality of care precisely and reliably, which leads to underestimates of the strength of the relationship between process and outcomes of care. They would also point out that the relationship between quality of care and outcome might be non-linear and hence go undetected in studies that employ conventional statistical approaches (e.g., correlations). For example, maybe genuinely lousy care damages health but given some baseline level of care adequacy, further improvements make little difference to outcome, creating the illusion that quality doesn’t matter in simple correlational studies. These are fair points and it would be foolhardy to give up on quality improvement just because the work is difficult.
But it would be equally foolhardy to not simultaneously try to improve the outcomes of health care in ways other than manipulating the process of care. The U.K. government has answered the challenge by leapfrogging questions about the process of addiction treatment to directly reward providers based on patient outcomes. “Payment by results” has been used in the NHS in a number of areas, but this is its first application to addiction treatment. In a small group of experimental areas around the country, addiction treatment providers will be paid based on their patients’ outcomes (e.g., drug use, employment, overall health and well-being) with the nature and process of care left up to the providers’ best judgment. There are a bevy of details to work out, including how to set the payment such that treatment programs will not shun hard-to-treat patients, but the basic concept has real promise.
Will it work? I don’t know, which is exactly why I am glad the experiment is being conducted. Bagehot is correct that boffins are ascendant in Whitehall these days, as two parties long out of power bruit the ideas they developed when they were in the political wilderness. The payment by results experiment in addiction treatment is among a number of demonstration projects that will put these new ideas to empirical test (The brainy Minister Oliver Letwin is a key player). Being experiments, some will generate negative results, but the spirit of innovation is encouraging given the pressing need to wring more health benefit out of every penny we invest in health care.
The Dutch seem to use payment by results successfully for public employment services. I read somewhere that the Dutch agencies are now prepared to try new things, like giving the unemployed bicycles. Of course the metric of success is far simpler and more objective: has the client (a) found a job and (b) held on to it? Health has lots more dimensions overall; but drill down to a sub-aspect like addiction, and measurement becomes more tractable.
There’s one surefire way of making sure your recovery rates are good: take people whose problems were minor in the first place, or ideally who had no substance abuse disorder at all. Unscrupulous drug-treatment operators in the U.S. have been manipulating their waiting lists for this purpose time out of mind.
To work at all, such a system would require third-party prognoses as a baseline. Has the clever Minister been clever enough to game this out?
What Mark said. It reminds me of the ed reform leading to an increase? in cheating (or maybe just the usual amount). There are similar issues about the fairness of blaming the provider if someone relapses.
But it’s still good to try the experiments. You never know where the next good idea will come from. I assume someone already tried just paying the addict to stay clean?
Hi Mark — Yes, there is a third party assessment team that is not involved in treatment. It does a baseline assessment, sets the amount of money for attaining good outcome and does an independent follow-up to assess outcome at the conclusion of care. The sicker the patient the higher the payment, so if you treated people who didn’t really need treatment, you’d go broke.
NCG: Steve Higgins and Nancy Petry are among a number of scientists who have shown that contingency management does indeed promote abstinence in addicted people. Sometimes it’s money, sometimes small prizes, sometimes privileges (e.g., take home does in a methadone clinic). Such programs often face political headwinds however, and haven’t been as widely adopted as might be expected given the evidence.
It seems to me that the third-party team is going to have a heck of a conundrum on its hands. The success payments have to be high enough to cover the (presumable) increased up-front costs of the more complex out-of-office interventions, but not high enough that it makes sense to game the system in a broad way. If the payments aren’t high enough, you’re setting providers up to fail in really unpleasant ways, because they’ll see that even if they succeed they’ll lose money, so then the incentive is to fail in the cheapest possible way.
It may also be plausible to encourage the tougher patients to move to another jurisdiction.
(Which is not to say I’m opposed to doing this, but for a pilot project at least they should be seriously willing to let the market set the prices for successful interventions, rather than imposing caps that yield distorted information at best.)
It is reassuring that the systems are designed to be at least somewhat resistant to attempts to game them thru lay-up patients.
Out of the discussion last week on readmissions, we ended up at a point where we agreed targeting resources is the optimal strategy.
What about targeting patients who according to the baseline assessment are in the highest outcome variance groups? Ex ante some patients are very likely to relapse, another set are unlikely to relapse, and a third set could go either way. What about focusing resources on the patients that could go either way, in terms of getting over their addiction?
This would be conceptually similar to focusing on lowering the cost of providing care to the individuals in an insurance pool who are the most expensive to treat, which has the desirable incidental effect of improving care for chronic users of the health care system whose needs are not getting met.