NIMH Director’s Innovation Speaker Series - Dr. Alyce S. Adams, PhD

Transcript

Phil Wang:
Good afternoon, and welcome everyone to our innovation lecture.  And it's a pleasure to -- well, I'm Phil Wang, the Deputy Director here at NIMH -- it's a pleasure to introduce our speaker for today's innovation talk, Dr. Alyce Adams.  She received her Ph.D. in health policy from Harvard Medical School and then stayed on as faculty there for about a decade, I think.  And in 2008 made the cross-country trek to Kaiser Permanente's Division of Research in Oakland where she's a research scientist and the Director of Health Care Delivery and Policy, did I get that right?  Close enough?  All right.  But her research has really spanned a pretty impressive breadth, I think.  From everything from -- I was looking at publications that you had published recently, looking at things from prior approvals and some prescription benefit designs that might be of help to disadvantaged populations, including populations of concern at least to the folks here at NIMH.  People with schizophrenia, bipolar illness, depression.  And also looking at modifiable factors that could be intervened upon, things like patient-doctor communication or churn rates in health plans, that are amenable potentially to intervention that could then benefit patients in terms of the treatments they use, the outcomes they experience, and also with a goal of also potentially mitigating disparities.  So, all of importance and interest to us here.  And her funding is equally broad and impressive.  It comes in part from us, NIMH, NIA, AHRQ, PCORI foundations, so she's well-funded, and a lot of people are interested in having you do research.  So with that, let me just say that her talk today is When Access Isn't Enough: Persistent Disparities Among Our Nation's Insured.  Dr. Adams?

[applause]

Alyce Adams:
Thank you so much.  I want to thank everyone here at NIMH, particularly Drs. Insel and Armstrong and Dr. Schulte and everyone for inviting me here and for making me feel at home.  Thank you so much.  I've had a wonderful day, had great conversations with people about how people are thinking about, not just health care disparities, but in general health care delivery research, and how we move from the sort of, the research paradigm that we're used to one that's more interactive, which health care systems and policymakers and others, in a way that we actually get the impacts we're looking for, the outcomes that we're interested in, as opposed to sort of writing the papers and hoping someone reads it.  So it's been a really exciting day for me here, thank you so much.  I'm going to tell you a little bit about Kaiser Permanente's Division of Research, because many people say what?  Where?  And, so basically, the Division of Research is essentially a research arm of Kaiser, however we acted pretty much like a medical school department.  So we're soft money for the most part, we primarily rely on funds from NIH, as well as AHRQ and PCORI and all the usual places as well as foundations, et cetera.  

And we do a lot of -- a variety of work from epidemiology to, not so much bench research, but a lot of health services research, health policy research.  And in the last couple of years when Joe Selby left the Division of Research to go head up PCORI, Dr. Tracy Lieu came on as our director.  And one of the first things she said was we don't really have an organization that really defines who we are and where we're going as an organization.  And she saw this vision of health care delivery science being something that we were uniquely positioned to do as researchers who were embedded within a health care system.  So we reorganized everything, and I got lucky enough to be chief of the Health Care Delivery and Policy section, which is a group of about 10 investigators, overall about 50 people, including staff.  And a lot of our work really is motivated by real-world clinical issues and real-world policy issues that we're hoping our research can influence.  

And as part of that we've had to think really creatively about how do we get the right people in the room at the right time in order to generate those research questions that are most important, not only to Kaiser, but to other health care systems.  And then how do we move from knowing there's a problem to actually doing something about it in a systematic way while retaining our scientific integrity?  So, we don't have all the answers yet, but I feel like we're getting a little bit closer.  

In addition to my role as Chief, as Dr. Wang kindly mentioned, I'm also a research scientist myself.  We have a very similar trajectory as you would have in a medical school department, for example.  So we have assistant associate in full in the associate level, so the two on the posters referring to associate level position with respect to research.  And the poster also mentioned we have a number of different administrative roles, including outreach and collaborations core for the Cancer Research Network.  We're also part of the MHRN, as it were, the Mental Health Research Network, the CBRN, and a lot of other types of network activities that a lot of you have been very supportive of over the years.  So what I'm going to talk about today is this issue that even when patients have access to care, it's really not enough.  

Now when I started this role as a researcher, it was back in 1999, and at that time, we were still having this conversation about well, if you just build the right health care system, if you just give people access to the right services, would you get better outcomes, so that all of these disparities that we're all worried about go away.  Because at the time there was a lot of concern about this issue between health -- having health insurance and not, or having underinsured and what that meant for disparities.  So, a lot of conversation about that.  And over the years what we found is that access alone isn't enough.  And it's pretty logical, I mean one, we know that there's a lot of big stuff besides health care that affects your health outcomes.  But also, we also know that when patients are in the same system and have equal access to care, you can get some differential results.  

So a lot of the stuff I'm going to talk about today is some of the work that we've done to sort of show sort of why does that happen.  And when it does happen what can we do about it?  And so I'm also going to talk specifically about modifiable determinants.  These are things that affect how people experience the health care system and the outcomes of that health care system that are essentially intervenable by the health care system itself.  Things that we can do today.  And this really is driven by clinicians who've said to us you know, that paper you wrote is great, it's really great that you got it in JAMA or Health Affairs or wherever that is, what does it mean for me?  What am I supposed to do tomorrow with my patients or tomorrow with my quality improvement program to help me move forward?  They didn't care about co-variants, they didn't care about p-values, they didn't care about any of that.  What they wanted to know was what can you tell me about what I can do to make a difference in terms of these disparities, as opposed to just telling me that they exist.  

So, I've spent a lot of my last 15 years sort of focusing on this idea of what are those modifiable determinants?  What are those things that we could do that if we turn them on they would have a chance of actually having an impact on the things that we care about?  In particular I focus a lot on medication adherence, which I'll talk about.  And finally I'm going to talk about what I'm really excited about, which is the job that I have now in addition to being a researcher, but really getting people engaged and involved in this idea of stakeholder-engaged research.  How do we create an environment where we can generate new evidence, where we can test out innovative strategies for dealing with these disparities together as a group?  It's us, it's the patients, it's the physicians, it's the people developing quality improvement programs, getting all of us together in the same places to hash out what do we need to be doing, where do we need to be going, and how could researchers help us get there?  So, as you all know, you know, in the pre-ACA world at least, the fact that we had so many people that had lacked health insurance was a concern for disparities because it tended to affect disparities populations.  We know that at that time at least 35 percent of Latinos and 30 percent of American Indians and Alaska Natives had no health insurance coverage at all, which made them vulnerable to all sorts of things in terms -- including chronic disease outcomes which were potentially avoidable if they had just had access to a usual source of care.  A lot of that is because they did not necessarily have employer-based health insurance, which as you know in the United States is very heavily reliant on that.  

We also have this additional factor of underinsurance.  So, not all insurance is created equal, right?  There's different generosity of coverage.  And the fact is if you happen to have coverage that doesn't cover a whole lot, that leaves you sort of vulnerable and -- sorry, at important times, you're going to get worse outcomes than somebody who was better coverage.  And certainly one of the things that came up over and over again was this issue of higher out-of-pocket cost burden.  Patients saying I didn't get that service because quite frankly, it was too expensive.  Or I cut my medications in half because I simply couldn't afford it.  Or I started sharing, you know, my medications with my brother-in-law who also has hypertension, because quite frankly, we both couldn't afford that medication.  Making those decisions, we've done a lot of interviews, for example, with patients who were saying: "When I'm faced with yet another medication to add to the regimen, I have to ask myself how important this this?"  We've had patients tell us okay, this is going to help me from going blind, this is going to help me prevent maybe a stroke sometime in the future, I think I'm going to take the first one.  So they're making these calculations regardless of how important we're saying the medications are for their health, regardless of all the evidence saying this is cost-effective, you should take these things.  This is their reality.  And so we're seeing these things even among patients who have some insurance, but also patients who have underinsurance or get insurance only part of the year.  

So, over time there's been this push, as you know, to expand coverage.  The idea if we expand coverage, we're going to do a lot to reduce disparities.  And absolutely, that's certainly the case.  We have Medicaid and CHIP and Medicare Part D and the Mental Health Parity Law and Affordable Care Act, and you can go on and on about all the ways in which we've tried to, you know, sort of fill the holes in this patchwork system that we have of health insurance in the United States so that those people who are most vulnerable don't fall through those cracks.  And the impact that coverage, while undeniably has increased people's access to health care services, the impact on disparities has been more mixed.  And in fact, we find that even when patients have insurance, for example, privately insured populations, we still see disparities, right?  And again, when I started this work, that wasn't so clear.  People were saying oh, well, if you're in a managed care situation, you shouldn't have disparities, right?  Everyone has access to the same stuff, it's all good quality care, what are we worried about?  But we do see that.  And we also see that coverage expansion often improves access, but disparities often persist, and in some cases, disparities actually can get worse when we expand coverage because the very people that we're most concerned about giving the coverage to are often the last to sign up for it, right.  So you get all of these bizarre sort of outcomes we didn't intend, but these policies were actually intended to help people, don't necessarily end up where we wanted to go.  I just have a few examples of that.  

This is some of our earlier work in diabetes, and this was in a health plan where patients had more or less equal access to a really great disease management program.  Very progressive at the time, a lot of outreach, a lot of support, RN calls on a periodic basis to help them manage their diabetes, and yet when we looked overall at the population -- bottom is the year, so between 1994 and 2001, and the A1C control is on the left hand side.  We saw that for both men and women, blacks had higher A1C than whites, and that kind of persisted over time.  So even though here we were just getting the results of some of this, you know, sort of groundbreaking diabetes work and understanding what makes a difference and we started changing things in terms of quality improvement, all the way to 2001 where we felt like we know what we need to do to address diabetes control in terms of, you know, medications that people need, self-support, the chronic disease management program, et cetera, the disparity persists across all of that time.  So we said well maybe it's because we just sort of randomly select these people with known diabetes.  What about those people who are new to diabetes when they were in a system like this?  We should see the disparity go away then, right?  Because they're all new, none of them have been exposed to it.  As soon as they get exposed, we'll see the disparity go away.  Unfortunately, that's not what happened.  We saw a couple of things.  

One is when they were diagnosed with diabetes, blacks were still doing worse than whites with that first initial diagnosis.  Which wasn't supposed to happen, again, in this wonderful managed care world where everyone gets the same care.  Furthermore, the disparity persisted over time and for men, it actually got bigger.  So this was not the answer I think people were looking for from managed care at the time.  We really thought this was going to be it, if we could just get everything right so that patients had access to a usual source of care, primary care as well as disease management, we would get rid of these things, but they did persist.  Furthermore -- this is not our work, it's the work of a former fellow who looked at the Massachusetts health reform.  She asked the question well, what happens when you do health reform on a global scale?  You say we're going to give people insurance coverage, we're going to do all these things, what happens?  Well, she found that in the early years after the Massachusetts health care reform, certainly there was a decrease in the likelihood of not having insurance, right.  Which is what you would expect.  That's what it was designed to do, have everybody, it does that at the minimum.  So it did that.  But she also looked at cost as a barrier, as well as having no personal doctor, again, all very positive.  So we saw a decrease in these bad things that we, like, don't want see when we actually offered people health insurance coverage.  

The bad news is that black/white differences in these things persisted despite the fact that we'd launched this health care reform.  Right?  So we had black race and Hispanic race still being significant predictors of not having health insurance, of having cost as a barrier, and not having a personal doctor.  So despite the fact that we have this policy that's designed to sort of raise all boats, we're still having this persistent issue.  Similarly, we recently released some work that we've been doing on the Medicare Part D law, where we looked at patients who were living in states where there was a cap on the number of medications that they could actually -- that would actually be reimbursed by Medicaid in any given month.  So say there might be four medications would be covered by Medicaid, after that you're on your own.  So, what Medicare Part D was it lifted that cap.  Because before that we had this variation by state in terms of whether or not you had a cap.  Medicare Part D came in and said no caps.  So all of a sudden this financial barrier was lifted.  So we thought okay, when Part D goes in, we should see everyone going up, which we did.  Everyone increased their utilization.  In this case it was patients with diabetes with lipid-lowering drugs, we're seeing a similar pattern regardless of what we looked at.  We're looking at schizophrenia, bipolar disorder, cancer, antidepressants.  No matter what we sort of put on here, what we're seeing is when you lift that cap, you see an increase in use.  

However -- and this is where the red line comes in -- we found a couple of things.  For this condition in particular, for lipid-lowering drugs among diabetes patients, for this particular medication, what we found was that the response under those 65 and older, the response to the policy among whites was stronger than it was for blacks.  So when you lifted the cap, whites and blacks both started using more, but whites used more at a higher rate, such that there became this difference, this statistically significant difference between whites and blacks in terms of who was using.  So the disparities got broader as opposed to getting narrower.  

Similarly, when you look at those age 65 or older, this was a non-statistically significant difference before Part D, even though it was trending upwards, in the difference between whites and blacks.  So basically, the lines are on top of each other.  Post-Part D, the whites are suddenly using more than blacks.  And again, it had to do with this differential response to the policy, which was not what I anticipated, but ultimately what it meant was where before we were concerned about these differences, particularly in this group with respect to utilization, they seemed to be getting worse despite the fact that we're doing what we're supposed to do, which is to provide health care insurance.  

So, why is it that despite all of these efforts, particularly at the federal and state level, to improve access to care would either one, have no impact on disparities, in other words, make everyone a little bit better off, but we saw this disparities gap, but then also two, in some cases exacerbate it?  Well, we already know that there's lots of things that matter in terms of health care disparities and they go way beyond the health care system.  There's a lot of stuff that we don't have a whole lot of control over within the context of the health care delivery system.  And certainly those factors all come into play with respect to these findings.  We also find, however, that there's variation within and across health care systems, even when you have coverage that matter, differences in quality and delivery care, true affordability, and true accessibility.  So does it really mean you have access if your doctor doesn't speak your language?  Those types of things.  Further, there are issues in the patient/provider and patient-health system dynamic that may also play a part in terms of why it is that when we provide health insurance coverage, people don't necessarily just show up.  

So, this next sort of part I'm going to talk to you about is really this issue of quality improvement.  A big question that we were asked when we first started this work is that well, if you have this great disease management program, et cetera, and you implement it, by definition if we're treating everybody the same, everyone should get better and all the boats will rise at the same level and we'll have sort of this even playing field where everyone will have equitable access to care.  And they kept, you know, likening it to the tide rising all boats.  And so we started on this question well, what systematic -- can systematic improvements associated with integrated care delivery systems reduce or eliminate disparities just by their very nature of being systematic?  One of the studies that we had done looked at changes over time pre-EHR, pre-concerted disease management programs in 1997 in a single HMO relative to 2001, which is after all this stuff got turned on, they got their EHR, their Electronic Health Record, they got their quality improvement, they got all of these things.  

So the good news here is that yes, when you look at things like cholesterol testing and cholesterol control, we did see a huge bump up.  You know, so this told us, well, there's something the health systems, especially these sort of integrated care systems can do that can improve control for chronic diseases for these patient populations, and that it seems to be true across racial and ethnic groups.  Furthermore, there is -- a lot of variation that we saw before across race, ethnicity, and gender, didn't completely go away, but it certainly narrowed.  A sort of narrowing of this variation in the later years versus the earlier years.  So that was sort of the good news.  However, when we looked a little closer and broke down various components, we sort of found this pattern where when we were looking at things that had to do with processes of care, whether or not someone gets tested, whether or not someone is initiated on a particular medication, whether or not someone is titrated in terms of a medication that they need to go up a dose, that's when we saw a lot of synergy and a lot of same delivery of care across different racial and ethnic groups.  Where we saw the variation was actually sort of on the patient's side of things.  The patient actually had to maintain that.  It was a follow-up, it was adherence and persistence with things where we saw variation.  And in particular when we looked at newly-diagnosed patients to say okay, what's going on with them, we found that they were just as likely to start self-monitoring, they were just as likely to have dose augmentation or clinical intensification of treatment when they needed it, unfortunately a lot of people weren't getting that, but in general, there weren't racial differences by it.  But where we saw the racial and ethnic differences and people started spreading apart was when we looked at things like medication adherence.  

Now, why is this important?  It's important because we also found that medication adherence is a strong predictor, a good predictor of control.  This is again, diabetes patients.  And furthermore, we found that when we controlled for medication adherence, much of the black/white difference in A1C was attenuated.  When we said okay, let's have everyone have the same medication adherence, now what do we see, we actually saw rates that were more comparable between the two groups.  So that told us that maybe something about these areas where it's more self-management, self-support, those types of things where we're likely to see more variation within these health care systems where people have equal access, more than it is these processes of care and things that they do automatically and do well, and they're sort of taken out of the realm of individual-level decision-making and more at a systematic level.  We sort of implement these things at a system-wide level, they get implemented and it works fine.  Anything that seems to involve individual patient decisions or the dynamic between the physician and the provider is where we started to see people separating apart.  And this is just sort of more of the same.  

When we looked at initiation of medications for diabetes, we found that for men, whites and blacks looked very similar.  However, when we looked at whites and blacks in terms of other oral medications, we found that blacks were actually more likely to be initiated on these medications.  Again, something systematic that happens within the system that says okay, this is the result you get, here's a recommendation, put them on this number of milligrams, starting now, twice a day.  It's a very automated process.  And through that automated process, not only were we getting similar rates, but blacks were actually faring better.  But after they initiated therapy, again, the lines start to separate.  Blacks had higher rates of discontinuation than whites, and it happened very quickly.  Right?  I just showed you data showing they're no different here, right?  But come three months out, five months, six months, we start to see the separation of the two lines.  And so we started feeling there's something about this follow-up process, there's something about self-management, support, what have you, that's just not happening with these patients, at least not at a level that is allowing them to maintain that equitable health care outcome that we're so interested in.  

Again, what we found in hypertension is similar to what we found in diabetes, and this is actually in the Kaiser system, where we looked at adherence to anti-hypertensive therapy.  Where we said okay, what is the likelihood of you picking up a prescription if you're prescribed it?  Almost 100 percent.  Almost everybody picks up that first prescription, no problem.  It's -- the Kaiser pharmacy's right downstairs.  You know, there's no hassle, there's no logistic issue, you know, you just go downstairs, you pick it up, and you take it home.  Are you taking it?  Some are and some aren't.  And in particular, we found -- and this is one of the first studies I think to find this for Asians as well as blacks and Hispanics, where this early-on persistence, what happens for the first refill that you're supposed to fill, how many people fall out, being black, being Asian, or being Hispanic was statistically significant predictor of not picking up that first refill.  Similarly we looked at non-adherence which is over a longer period of time, six to 12 months of adherence following the initial diagnosis and initiation on medication.  

Again, we found that for blacks and Hispanics there was this persistent non-adherence to these medications.  Even though they started off, again, in the same place.  We found the same thing with anti-depressants among patients with hypertension.  We found the same thing among patients using anti-depressants for their -- regardless of whether or not they had some other comorbidity, where we found that consistently, there was this difference in non-persistence, higher rates of non-persistence among blacks, Asians, and Latinos in this period of time prior to their initial hypertension diagnosis, at six months and at 12 months.  And that non-persistence with anti-hypertensive therapy, and this is true for everyone, was predictive of non-persistence with anti-hypertensives later on.  
So if you were not adherent to your anti-hypertensive therapy, you were more likely to be not adherent to your anti-hypertensive therapy later on.  And that was true of all racial and ethnic groups.  So, I would say this led us to a lot of work focusing on medication adherence.  

And I want to pause for a minute just to say that I don't think medication adherence is a panacea.  I don't think if we fix medication adherence that all the other patient problems with disparities will go away.  However, I like it.  And the reason I think adherence as something to focus on is that it's a common potentially modifiable determinant across chronic conditions.  So instead of saying okay, we're going to work on this disease silo and see what's affecting disparities here, and then we're going to go over there and do something else, it's something that's pervasive across the system.  It doesn't matter which disease or problems you have, it seems to bother everybody.  And so it gives us a nice starting place to say looking at the whole patient, and many of these patients do have chronic conditions, more than one, what are some of the things that we can focus on, and medication adherence might be one of those targets.  It's also very easily monitored and targeted.  We know how much they're taking.  We can look, we can see, right?  It's a very observable thing.  

And finally, there's a strong evidence base for telling us whether or not these medications are likely to work for them and who should be on what and at what dose.  Right?  There's a lot of strong evidence around these issues.  So we should be able to utilize that.  Within that context, there are a lot of different factors that might influence why we're seeing these differences in medication adherence across racial and ethnic groups and across conditions, and we think they include patient as well as patient/provider dynamics and health systems factors.  And these might be the opportunities for intervention here with respect to improving and moving us towards equitable health care outcomes.  Some of the things I'm particularly excited about on the patient level that I think are interesting and that we're trying to test out in our own health care system is -- some of it is like motivational interviewing.  And it may be most effective when it's in a clinic system by professional trained -- professionally trained and motivational interviewing.  

However, one of the issues with motivational interviewing is that it's not clear who needs it and where it's going to have the most bang for the buck, right?  When we talk with the health care system about motivation interviewing for medication adherence, they're like oh, gosh, hey now, who do we choose for this?  Because this is a lot, you know, this is a lot for us to take on.  So, to the extent that we as researchers could help them do a better job of segmenting the population and targeting who would be most appropriate for motivational interviewing we might be better off.  Furthermore, we can look at things like values affirmation.  So this is something that is coming from the educational literature and research where they found that a simple values affirmation exercise, then someone goes through that prior to a test for example, they tend to test better.  

Similarly, the idea here is that before you go into your provider, if you have trouble communicating with your provider, do this values affirmation exercise with them beforehand, which is basically asking about what do you value, what do you care about, et cetera?  It has nothing to do with your health necessarily, or your health care, but it has more to do with your own personal values.  And then having them go see the physician may actually increase the lines of communication around medication adherence and around control.  And so we're doing -- we're trying to do an intervention in this area but there's really strong evidence from the education literature about this.  

One of the interesting things about motivational interviewing versus values affirmation is motivational interviewing is a patient-centric intervention, but that's not race or ethnicity specific.  Okay it's not about cultural tailoring, necessarily.  It really is about the individual patient, what's going to work best for them.  Values affirmation speaks directly to racial and ethnic differences in experience that may result in differences in how they experience the world, how they experienced their conversations with their provider.  And so I was trying to directly address those stereotypes and those issues that are underlying subconscious bias -- whatever you want to call it, that may be influencing the patient/provider dynamic.  These are very different approaches but they both have some promise in terms of helping us to understand better what might be going on in terms of medication adherence.  

Another thing that I'm particularly excited about is the idea of conjoint analysis, which really combines these two topics to say we want to find out from patients themselves what are their preferences with regard to treatments?  How do they feel about side effects?  How do they feel about their different medication choices?  And asking these things overtly because we know patients are making choices.  In the past I think we sort of keep giving them medication and sort of hoping that they'll take it.  This really says you know what?  Patients are making decisions with or without us.  How about we try to find a way to solicit those preferences early on and then as much we can, tailor treatment, tailor regiments to those preferences?  And we also have the opportunity to potentially uncover differences and beliefs across racial ethnic groups that would be underlying some of these adherence behaviors.  

So for example, John Pate [phonetic sp] and others have done a lot of work showing that, for example, African-Americans tend to have more negative attitudes towards medicines.  More likely they think that they might be addictive.  More likely they think that they're not going to be helpful.  And so to the extent that those things may be true, culturally there may be something going on.  This is an effort to sort of drive -- to understand if it's one of those factors that are driving the behavior and then have a conversation about it and/or target therapies to more better fit what the patient needs.  At the provider level, there's less evidence about things we could do at the provider level that are in disparities reduction per se.  The cultural competency literature isn't as strong as we'd like to see in terms of providing that really strong evidence of what works.  We do know that in general, feedback mechanisms are probably more effective than simple reminders.  So having something just pop up on the EMR for them to remind them to do something is not only an annoyance, but often it doesn't really generate changes in behavior.  

However, feeding back information to providers about what's going on with their individual patients and how that relates to the broader practice, does seem to be quite effective.  So to the extent that we can actually leverage that to look at racial and ethnic disparities in how patients are experiencing their care, we may be able to get closer to behavior change that would lead us to more equitable care.  One important thing, strategies that are aimed at only at the provider and only at the patient may be less effective than strategies that were aimed at both.  So if we're interviewing at the patient level through motivational interviewing or values affirmation what have you, also interviewing at the provider level; providing the feedback about information about their patient panel, how they're doing and what's going on in sort of, this academic detailing cycle, doing both together may be more effective than doing either in isolation.  

Finally, there are things at the health system level that may make a difference.  Things like copayments.  There's a lot of talk right now about co-pay reduction.  If we just reduce co-pays will that do something?  And so we think that there's hope that that co-pay reduction will likely improve or increase medication use, but that we might need to do something else to reduce disparities because of the data that I showed you earlier.  Simply reducing the co-pay may not be enough.  We may have to have concurrent strategies that are aimed at addressing the stigma or what have you that are associated with lower rates of usage among those populations for whom cost is not the primary barrier, right?  It's like yes, okay, cost might be a convenient reason why I'm not taking this, but the truth is I don't want to be on this medication in the first place.  So how do we combine strategies so that we reduce the costs but also talk with the patients about what their concerns are so that we can hopefully work with them together to get to where we need to be.  Furthermore, opt-out strategies; when we know something works.  With our setting we found that mail-order pharmacy is highly associated with medication adherence and it's partly because they often get a financial benefit of doing it instead of paying for three months of medication they only pay for two months of medication.  But it also automates the process to an extent where you're getting multiple medications at once so you don't have to come back as frequently for refills.  But African-Americans, for example, in our setting are less likely to sign up for it even when it's financially beneficial to them.  And when you talk to them they start talking about their concerns about it.  Well, I don't know.  You have to pay up front.  You have to use a credit card to pay up front.  What if someone wants to drop something off at my house?  How can I be sure that it's going to be secure on my porch?  Those kinds of things [unintelligible] are talking about and they haven't quite found [unintelligible] even know about it.

So thinking about are there ways that we can have opt-out strategies where we automatically sign people up.  Oh you have diabetes, here's your new medication.  By the way, I've signed you up for mail-order pharmacy, rather than relying on the patient to find out about it and then sign up for it on their own.  Those might be some strategies we should start looking into.  Furthermore, workforce diversity may also be important.  This is not my area of research, but certainly some others in our field have looked at race and language concordance, where they found that for African-Americans race concordance might be important.  They've also found that language concordance, might be particularly important for Latinos in terms of improving adherence, in this case to cardiovascular medications.  

So here's the rub.  So I've talked about all of these things that we should be doing et cetera, and that health system should be doing based on the literature today, but in the reality, race and ethnicity are actually really poor predictors of future adherence behaviors.  What do I mean by that?  

So we know that there's a strong association between the two, but if you were to take your patient population and isolate all the patients of particular race ethnicity and say okay, they're going to be non-adherent, we need to focus on them, you won't be so successful.  I mean, the reason is there's a lot more variation within than there is across these racial ethnic groups in terms of medication adherence, right?  So I would vote that we should probably move away from a little bit of this idea of race/ethnicity focused approaches, and really think about patient-centric approaches.  The patient is a whole person where race ethnicity is part of it, but certainly not the whole story and that I think has potential to address both individual and sub-cultural factors that are going on.  We also need more sophisticated risk segmentation strategies.  So how are health plans supposed to even identify these people?  How are health systems supposed to be able to identify them?  Again, identifying them based on race ethnicity is probably not a good idea.  However, to the extent that we can tap into those things that are correlated with race ethnicity, health literacy, low English language proficiency, patient/provider communication as reported by patients; those types of things might be really good harbingers of poor adherence behavior into future and to the extent that we can target those things we might be able to find patients much quicker than if we just rely on race/ethnicity as our indicators for who needs help in terms of medication adherence.  

So finally, I want to open this up for broader discussion, but just to close out: so we've been working on these issues since I started at Kaiser Permanente in 2008.  We've been trying to figure out how do we get from knowing we have disparities to do something about them given all of this information that we've just shown you.  And we've shown this to the health plan in the health care system and so they're also aware of all of these research findings.  And we thought well, you know, we're in the perfect place to do this.  We have data, we have scientists who are willing to do the work.  We have these partnerships between patients and clinicians and researchers where we can really sort of kind of use our EMR and use the email and use kp.org and all of these different mechanisms for communicating with one another.  We can mess with the incentives.  So we can mess with the incentives providers.  We can also reduce or increase co-pays for patients and then we also have leadership support, you know, so how could we possibly go wrong?  We're at the prime to do this.  We're at the perfect place to do this work, and this is what we saw in front of us know we started this process.  We literally got into a room much like this one and we said here's the problem.  This is what we're going to do and this is what we saw, right?  

We saw sort of the summit here as equitable care outcomes.  That's what we wanted to get to, but we didn't know where we were on this mountain.  We didn't know what the best route was to get from point A to where we wanted to go and so it was a very intimidating initial process and people we're like -- the clinicians and the health plan leaders we're telling us we are overwhelmed.  You're telling us to address something that is historical; something that has very even little even to do with us.  How are we supposed to make an impact?  And so, what we started was sort of with this sort of what I'm calling sort of, global positioning for equitable health care outcomes.  We said that first of all we need to know where we are.  Where are we on that edifice?  Where do we start?  And so the first thing is identify where we're at.  And so we used the EHR to identify where there were these racial and ethnic differences in the metrics that we commonly use every day?  Where are they?  Where are the differences?  Then we decided okay we need a common destination.  We need to decide where on this mountain are we comfortable with landing and saying okay we'll be successful if we get where?  

So we identified, you know, this is made up right now, but we're sort of for two areas for this colorectal cancer and hypertension control, we said we want to reduce the gap by X percent in a certain number of months and this is how we'll know we're making progress towards those goals.  And all of this again, was based on the evidence that we'd gathered today.  And then finally we created a map.  We said okay, how do we get there?  How do we make sure the way we started this process?  And we literally just started meeting with one another with the idea that the first few meetings we're know that get anywhere, but eventually we will.  We literally started meeting on a monthly basis, every couple of months, between physicians, quality leaders and researchers sitting in a room hashing out how do we get to where we want to be?  And through this process, I think we're succeeding in terms of our general strategy, which is that researchers -- we have to eat, right?  So we can't just stop doing research and all lead to quality improvement, right?  So we're still doing our RO1 applications, but we're doing them starting with the clinical question, right?  So instead of before we were sitting at our desks again, what I think is interesting?  What's based on my last research grant?  What should I write next?  We went to the clinicians and we said okay here's all the stuff that we've done to date.  What about this is useful to you?  What is not useful and what's the next place that we should go together?  

So they helped us define what they thought was interesting research -- a clinical question.  We helped them turn that into a research question and then those things go to NIMH and NIDK an PCORI and all these other places to try to get our core funding base.  Simultaneously, because the clinicians are telling us, look you guys have five years of funding that's great, I can't wait five years to find the answers to some of these problems.  So we also started smaller RFAs internally; smaller research grants that are $50,000, $100,000, but they all have an end date.  They're all usually no more than two years, but usually just one year.  Some are even six months long, where we say in six months here are the deliverables.  Here's what we want to be able to say and they're mostly about the feasibility.  How is it that we can get some of these sort of, very innovative, very interesting interventions to work in practice?  What are the barriers?  What do the pharmacists need?  What do the MAs need?  What do the physicians need from us?  Figuring out the strategies and figuring out sort of what's feasible to do in practice, and then testing them out in small places.  

We have one right now in Richmond where we're basically saying here's data, here's what we think you should do; working with pharmacists to say okay, I'm willing to do that and then they're implementing it.  And the idea is not to change the world with that, but to see is there an appetite for this?  Is there something that's potentially sustainable so that if our grant on values affirmation in that same area hits, then they'll have more to buy in for; we'll have the tools ready for them to use to say okay, that went well, let's add this component of this values affirmation to get the patient side of things if you're going to focus on the physicians side, for example.  So there's a lot of different things that we can do together that we could not do before when we were working on our own.  And this is just basically some of the various strategies that we were playing with; these large pragmatic trials were new for us but they're very exciting.  

Of course natural experiment is another way to get at this.  If we know what's going on in the health care system we can evaluate it in real time and get the information back to them sooner.  Mixed methods approaches, of course; and then of course validation studies to say okay how can we use some of these new data sources for research purposes in a way that feeds back into the health care system?  So with all of that, sort of what I wanted to close with is again, that health care access is necessary but not sufficient for what we want to do and that we do need to focus on factors that are potentially modifiable health system intervention.  There's promising strategies that are emerging and a very exciting and sometimes would get them from non-health-related like this educational values affirmation stuff that really does come from education has not been applied as much in the health care arena.  And then finally, and I can't emphasize this enough, these collaborations between researchers and health care systems are desperately needed, and they do provide this opportunity for us to come together and say how can we facilitate the translation of these research findings into practice in a clinically-meaningful way but they still allow us as investigators from [unintelligible] our scientific integrity and survive in the world that we live in?  And it's been really exciting and I think one of the best things about it is that I think, as researchers we have more accountability than we've ever had with the work that we're doing because it's very hard to not produce when a clinician is calling you and saying I'm meeting with the board tomorrow.  They want to know how it's going, you know, where are we at?  It's a very different level of intensity than I think you could have with our normal grant process.  And so I think both are necessary, but it certainly does sort of elevate the bar for us, and it's just an exciting field to work in.  So thank you so much for your time and I'd like to open it up for questions okay?

[applause]

Male Speaker:
So we are very invested in how learning health care systems can completely change the paradigm around research where we go from research to practice and we have hand offs and this sort of thing, to what I think you traced.  You start with the stakeholders questions, their problems you get those worked on by researchers; you're actually going to get their buy-in and hopefully they'll take up whatever it is.  So you and I were talking about what does it concretely mean, you know, because we're in alignment on it, but does it mean we solicit ideas from your executives, your people who, you know, re-run your clinical operations, we limit their researcher's questions to only those that -- I mean, how do we concretely do this?

Alyce Adams:
[unintelligible] more successful than others.  So when I talked about earlier when we tried to ask the health plan leaders, the CEOs, in other words and the chiefs, you know, the pie in the sky what would you want to know, it didn't work so well.  In part because [unintelligible] they were [unintelligible] researchable.  They might be important to them but they weren't really sort of in the form that was readily researchable.  The other issue was it just wasn't on top of their priority.  I think it felt a little bit too blue sky for them.  And so we didn't get the participation that I think we would've liked.  We also did these -- created these RFA mechanisms that I told you about.  Internally we said here are some areas that are important to the clinician group and they came up with a list and they were very general areas and literally it might be improve depression care; so very generic.  But it's a specific pot of money and in that description of the proposal you have to answer questions like at the end of this, what will we know that we can implement tomorrow?  You know, how is this going to help the problems faced by the health system today?  Who are your clinical partners?  It's a very specifically oriented towards entering the clinical question.  And so that's one way we've gotten it going.  

The other way quite frankly, is just relationship-building.  It's nothing new.  It's literally just being in the room together trusting one another, getting to know each other so that the clinicians feel comfortable coming in and saying well, I had a really bad day yesterday, here's what happened.  I'm frustrated because the evidence doesn't support what I want to do but I just don't know how to get there from here.  We don't -- there is no obvious clinical answer to this question.  And so from there we start flushing it out and saying okay is this a researchable question?  Is this one patient?  Is this all of your patients?  What's the pattern?  Then we take the data and we look at the data that say, you know, we were looking at your question and here's some of the patterns that were starting to see.  Is this consistent with what you're feeling?  And we literally work through it together, but it's a relationship and it's an ongoing one.  

So it's not enough to for us to sit together in a room once every six months and talk to each other.  We literally meet every other month and then we're on the phone with each other all the time and sometimes quite frankly, from the researcher's perspective, it does take a researcher who is okay with community service.  And by that I mean sometimes the clinicians just need you to show up somewhere and give a talk, right?  And I do it because it ultimately I'm reliant on them to help me find those important questions.  So I show up when they ask me to show up and they're very respectful of my time, but I do it.  But through that they became to trust me and say okay, I can tell her my woes, I can work through this with her and she's going to help me fix it or at least move us closer to where we want to be.  So it's a combination of things.  I don't think there's a single strategy, but I do think ultimately it comes down to relationship-building and then finding those clinicians that see the spark, right, of research and they appreciate it.  You start talking about research and their eyes light up.  Not because they want to be researchers, but because they see an opportunity.  They see oh this is something that I'm missing, right?  And that's what you want to see.  That's sort of, the energy that you need to get from that collaborator in order to sense that this is worth investing in and this is worth us spending some time figuring out some real-life research questions that we can actually develop and answer over time.

Female Speaker:
Thank you.  That was a great talk.  So I want to follow on a little bit of what Phil was saying and just a practicality.  So for example, what's really exciting is that you have the self-contained system, right?  You can do the individual level; you can do the provider level.  So let's say that you decided, you know, there is not as much evidence around the provider level interventions for example, and you decided, you know, this is where we need to try to do something.  How does that work and then what happens afterwards?  I mean, how do you make it happen?  And does it happen in one section of the organization?  Or does everyone then have to -- can you just talk a bit more about -

Alyce Adams:
Yes, we actually have tried to encourage our organization to think a little bit more in terms of pilots.  You know often in the past whenever I've worked with health plans there seems to be this tendency to say okay we belong -- so the "New York Times" had an article on X, we're going to employ that right now, whether, it's, you know, [unintelligible] or whatever it is.  Whatever it is sort of the thing of the moment is, and so we're just going to implement it system-wide, right?  And I don't mean to sound so pejorative about it but that's part of it.  And what we've said is you know what, you don't know yet whether or not this is going to work for you, so start small.   

So for example, the one that we did, this project that we have right now which is about physician prescribing of diuretics is really physician-focused, to say on the on their right medication?  Are they taking the right dose of it?  It's very simplistic, right?  But it's something that we haven't looked at before and it's not something that they've looked at before.  So we said okay when you decide how you're going to intervene with these different patients and what they need, you have the screen, it's called a prompt, and it has in specific HEDA scores basically that you're looking for each patient, and it redlines those that they need to focus on.  We said okay, what color are you not using?  And they said yellow.  I said okay, from now on you're going to have yellow is the color that means they need diuretics.  They need to be on thiazide diuretics.  We're going to create a new column in Prompt so that the data is there for -- we're going to give you the algorithm.  We're going to show you how to run it.  But from now on it's going to show up in yellow.  So if they need to increase the dose, what have you, that's what's going to happen, and that's what they look at now and it's sort of part of it their own automated process and we did it in one place, Richmond, all right?  We're not even done evaluating it yet but already again, they're implementing it potentially across the system.  Why?  Not because it's successful in terms of improving patient outcomes and reducing disparities, but because it was so easy to do, right?  

It was easy for the clinicians to do and I think that's part of it, a lot, particularly if you're working in collaboration with the health care system that you're not really completely fully a part of, it's convincing them this is something that you already do, we're tweaking it a little bit.  We're just adding to it.  We're given you something that's when to help you make better decisions as opposed to we're going to disrupt your life and change everything you do and the way that you do it.  So anything that you can do that doesn't mess with their workflow at all, but it really sort of uses existing strategies and existing workflow, and fit that in, you're going to have better success than if you're doing something outside.  But yes, starting small; starting with one; getting one facility and they didn't want to do it at first, but we kept on them, and eventually they agreed and now they are talking about what a great thing it is and they feel like special because they have it, right?  So a part of it is that just that working on small tests of change.

Female Speaker:
Thank you very much for your talk.  I wanted to hear more about the vision that you had on reducing the gaps by 5 percent, and the question that I have is for whom and using what kinds of interventions?  And I mean it's something that we think a lot about in our office where disparities in global mental health, and does identifying those people, especially, you know, since you mentioned that you're moving away from race/ethnicity to more patient-centered care, and that could be a combination of various things.  So I wanted to hear your thoughts on that.

Alyce Adams:
Yes, I mean there's a lot in the question.  So in terms of the way that we had done it is we started off with the literature and what we know and based on our own research as well as research of others with respect to where are the gaps, as well as what are people finding to be useful in terms of interventions?  We then also had the clinical guidelines that we use to say okay, we know what sort of, the evidence stands, here's what the clinical guidelines says that sort of goes along with that and then we said where are the modifiable banks?  What are the things that we can do something about?  So with this issue with race/ethnicity, like I said, this was way too broad and it doesn't really tell you a lot about what's going on with any particular individual.  So we said okay, what are those things that sort of explain those racial/ethnic differences?  And things that we came up with were things like health literacy; [unintelligible] were a lot of that.  

Often health literacy sometimes eliminates the disparity when you control for it.  It's a huge component.  Things like trouble paying for medication.  People who say that they have really poor communication with their providers often have poor medication adherence.  And those were all things that we measure all the time.  We're constantly asking our patients whether communication was like.  We ask it all the time, right?  But we don't necessarily look at it in this systematic way and use it for segmentations, that was part of it, but it was a very long process.  I don't mean to make it sound like it to one meeting and we were done.  It took some of time that I there in 2008, it was 2010 before we would get all the regions in the same room and to agree to that goal.  It took a good two years and it took a lot of dog and pony shows, and it took taking that they took around to different places to all of the different quality leaders around the different regions and saying here's what we're seeing.  

This is why we think no the data aren't bad.  You know, know that there's -- the data are really true.  It's not just you, it's everybody, but we need to work together on this and yes, there's something we can do about it.  So having that big meeting in 2010 and from there we didn't dictate specific interventions that they were supposed to do.  We said really that's up to the local groups.  That's at the physician office level.  That's at a medical office level, the facility level.  That's where that clinical question needs to happen.  That's where they need to start designing sort of what is the question that we want to be able to answer.  We gave them a lot of tools.  We gave them a lot of information about what had been tested before.  It was really up to them to figure out the context of their own situation and what best fit them and then they started again, these small tests of change.  

Can I say that this is like a big success yet?  No, it's still a work in progress.  We're still working through it.  But I have to tell you I was thrilled that we were able to sit in a meeting and have all of our quality leaders across the regions sit there and commit to doing something about this over the next however many years we said we would do it.  I mean, that was -- just getting them together in a room was a huge step, and getting them to commit at that was equally exciting.  So it takes time sometimes and sometimes there will be delays.  There were some regions that said we're going to let you guys go ahead and then we'll catch up later, but they did.  Eventually they caught up later, like our own region at first sort of, eschewed the whole process and said we're going to do our own thing, just last year came back to me and said, you know, could you do what you did with the whole program nationwide with us?  Just with us?  And then we did.  We held another meeting.  We had another symposium.  We showed them the data again and then we had like an all-day meeting where they brought their -- they paid their clinicians to come to this meeting over lunch, over the rest of the afternoon to sort of hash out where we thought the rubber hit the road.  And that's what started these frequent meetings.  That's what started this -- sort of these ongoing communications with our clinicians and operation leader partners.  But it's taken some time.  It's not something that could be done overnight I don't think.  Thank you so much for your time.  I just -- obviously NIMH has been really helpful with us.  These are just a smattering of some of the grants that myself and my colleagues have received that have supported this work.  

So I just wanted to acknowledge that and the great support we received from NIMH and others in this work, as well as a number of investors -- see there's not everybody, but these are just some of the people I work with most closely on these issues.  On the right-hand side you can see the long list of health systems leaders that have been in it with us from the very beginning and really leading the fight, and it's just been a joy to work with them as well as our scientific colleagues.  So thank you so much.

[applause]