Map of life expectancy at birth from Global Education Project.

Monday, August 10, 2015

Strife and woe

I'm working on a project now which has to do with measures a certain state's Medicaid program wants to use to reward primary care providers for quality. For various reasons, which I may get into more deeply in the future, these measures are never really great. They basically reward care processes, in this case giving certain lab tests to people with diabetes, and that can get providers falling all over themselves to do things that shouldn't necessarily be the highest priority for a patient, plus getting dinged for circumstances beyond their control.

But my mission isn't to worry about that so much, it's to worry about the accuracy of the data. So I've been reviewing electronic medical records at various clinical sites to try to understand the difference between data and reality. And no, they are never the same, we just hope for a knowable degree of correspondence. Usually we don't even have that.

So, if you select for a pre-expansion Medicaid population, old enough to have a high prevalence of diabetes, you are filtering for people who are legally disabled. What you get is incredible co-morbidity. A typical person in my sampling frame not only has a diagnosis of diabetes (although, due to the vicissitudes of data, some of them do not, oddly enough), but also serious and persistent mental illness, COPD, heart disease, kidney disease, substance use disorder, liver failure, maybe HIV and/or HepC, arthritis -- sometimes basically all of the above. Many of them are frequently hospitalized, they are constantly on the brink of homelessness or starvation -- did you know that there is a diagnostic code for insufficient material resources? It's just appalling the afflictions with which so many people live.

So no, they can't take personal responsibility. They are unable to work because they are too sick. But the same hypocritical morons who were screaming about the non-existent death panels in the affordable are act want to leave these people to die in the street. Because they're takers.


kathy a. said...

Totally with you on states against medicaid expansion really sucking.

It really makes no sense to just look at medical processes for one part of a complex picture. (And what is supposed to happen when someone gets their blood sugar results? Do they also get meds and support?)

Can totally see how other things might be the immediate priority, and yet the rewards go to this one process. Adequate food, shelter, safety, clothing are fundamental human needs; a person can't watch diet to control diabetes very well if food is beyond their means. Somebody who can't breathe needs to address that first. Somebody who is mentally ill or has dementia probably cannot deal with health management by themselves.

Shouldn't doctors trying to manage multiple conditions be rewarded more amply? Of course, the needs may sprawl beyond traditional medical care, so connections with social services should also be rewarded.

Because patients' conditions are complex and records are not so great, it may not be possible to make great comparisons. But is it possible to sort data in an "orders of magnitude of other things going on" kind of way? Like, diabetes plus one other condition = X% harder. With 2 other conditions = Y% harder. Etc. (I've seen something like this done for trauma -- multiple sources of trauma vastly increase the likelihood of adverse effects.)

Running numbers on those who are impoverished vs. those who are not could also be useful. The lack of basic life necessities is a staggering blow, by itself. The lack of health insurance, regular care, and needed assistance to comply with management of medical conditions means that treatable things do not get taken care of until there is a crisis.

Guess I'm advocating use of the biopsychosocial model in the context of a medicaid reward system. One little piece of process does not an effective approach make.

Cervantes said...

There is a lot of info in the medical records -- counseling sessions, case management, phone outreach -- but most of it is not billed and doesn't show up in claims data. This is a safety net community health center -- they also do a lot of free care and do their best to take care of people who can't pay at all. It probably would be possible to devise richer ways of measuring quality of care from EHRs than it's possible to do with claims data, but there are many difficulties and complexities. This gets very wonky and I don't necessarily want to go there on this blog, but it would be difficult to implement your suggestion, even though it's the right one. I'll think about it some more though, as I write my report.