Why don't we measure health care outcomes?

Reihan Salam offers some clear thinking on the fallacy of judging how well we are caring for peoples' health by looking at how much we are spending on their health care. Instead, we should look at outcomes, that is, whether our health care makes them healthier.

Caring for sick people, and indeed caring for people who aren’t necessarily sick, is much cheaper in some contexts than others. It is not obvious that we should pay attention to expenditures as the metric for how much or how well we care for the sick. Rather, it might make sense to consider hard outcomes... It makes perfect sense that providers would want us to think that higher spending levels mean better care. That doesn’t mean we should embrace this narrative ourselves.

Engrave this in stone. I don't know an analyst who disagrees with this.

So why don't we routinely hold providers accountable for better outcomes? Well, usually we do not even have outcomes: providers do not routinely report, and in many cases do not routinely measure the outcomes of care. This fact astounds laypeople.

One problem is that many of the outcomes we care about are not hard, like death. They are subjective, like pain or fatigue, or intangible, like your ability to think clearly and recall the events of your life. These things correlate with objectively physiological parameters, but they don't correlate as closely as you may imagine. This isn't necessarily because subjective outcomes are less reproducible than objective measures. It's really the case that minds and bodies have a certain independence. Anyway, if care has multiple outcomes, what measure should we use to benchmark providers?

Second, death aside, there usually isn't a definitive time at which the outcome of a medical intervention is clear. You have to follow people over the course of recovery, which someone has to pay for. Who? The patient doesn't want to, unless the measurement directly contributes to some additional benefit for the patient, as tracking hemoglobin A1c can for a diabetic. Measuring the outcome doesn't benefit the doctor and competes with actually delivering care to other patients.

So, outcomes measurement will not happen routinely until the purchasers of health care require submission of  outcomes data -- and data that can be audited -- as a condition of receiving payment. And that will be resisted. Many providers do very well under the current system and would face considerable downside if they were paid on outcomes rather than services. 

Comments

[Jonathan]

I haven't read the Reihan article, but isn't this essentially the purpose of hospital report cards. Now I don't know the entire literature, but it seems that consumers find it difficult to interpret these measures when they need to choose a physician. Also for a lot of these report cards we use an outcome measure different from mortality (readmissions). Anyone know more about this?

[Brad F]

I think you missed the biggest factor of all.

Attribution.

If you are speaking of providers, see Table 1
http://www.nejm.org/doi/full/10.1056/NEJMsa063979#t=article

Just look at a chronically ill beneficiary. Who is responsible in the sea of providers?

If you are envisioning a future where patients subscribe to integrated systems--a long ways off for nationwide uptake, the system owns the outcomes. Easier to oversee..

However, if mortality, SMRs for inpatients are woefully inadequate, and as outpatients, who knows. Thats tough, probably too high a hurdle to use as meaningful outcome.
http://www.bmj.com/content/340/bmj.c2016

I am not saying impossible, just real, real hard. Lots of risk adjustment, lots of measurement tools we dont yet have, and registries and databases that are far from complete.

Brad

[Bill Gardner]

Great point, Brad. As Reihan said in his post, we have to shift to systems with salaried providers. That solves the attribution problem as well.