As anyone with chronic disease knows, access to health care doesn’t always equate with equitable health care outcomes, says guest Alyce Adams, an expert in innovations in health policy.
Too often, care delivery breaks down along racial and socioeconomic lines. Our focus should be on better outcomes for all people, she says. Adams now develops interventions to help communities and health systems improve care delivery — and health equity — as she tells host Russ Altman in this episode of Stanford Engineering’s The Future of Everything podcast.
Transcript
[00:00:00] Alyce Adams: If you live in certain neighborhoods, you have less access to certain things, and that sort of compounded the issues related to covid. In similar ways, those mechanisms have been going on in our society for a long time, and they perpetuate this higher rate of chronic conditions in some of these populations. And so what we try to do is better understand what the drivers, these drivers. Try to identify those are actually modifiable through health system and policy interventions, and then work collectively with communities and health systems and others to see if we can [00:00:30] design better interventions that have the potential to address and alleviate this burden for many of our patients in our populations.
[00:00:42] Russ Altman: This is Stanford Engineering’s The Future of Everything podcast, and I’m your host, Russ Altman. If you enjoy The Future of Everything, please follow or subscribe wherever you listen to your podcast, that will guarantee that you never miss an episode and it will help our listenership grow.
Today, Alyce Adams will tell us why people with [00:01:00] chronic diseases can have vastly different health outcomes. Just because you have access to healthcare doesn’t mean that it serves you well. You still need to get to your appointment, communicate effectively with your provider, get your medications. None of this is a given and we need to work to ensure that folks turn access to into good health.
It’s The Future of Health Outcomes.
Before we jump into this episode, a reminder that if you enjoy the podcast, please take a moment to rate and review it. It will help fellow [00:01:30] listeners discover and understand what this show and the future of everything is all about.
You know, chronic diseases like diabetes and cancer require long-term medical care. But if you are poor or live in a rural environment, access to medical care can be a real challenge. Even if you have access, you need to be able to use that access and there are many ways the system can fail you. And if you don’t get the care you need, [00:02:00] then you, your family, friends, and society will suffer.
How can we create policies and interventions to make sure that the dollars we spend on healthcare are effectively used to help patients? Well, first we need to understand where are the bottlenecks in the delivery of care, and then we need to remove them. Alyce Adams is a professor of epidemiology and population health, health policy and pediatrics at Stanford University.
She studies the racial and socioeconomic disparities in [00:02:30] chronic disease outcomes. She will tell us that although funding is of course important, we must also listen to the patients about what is important for them and their health and where the barriers are for them to getting the best health. This will help us understand how we can best intervene to make sure that they get good care.
Alyce, you study differences in outcomes for people with chronic diseases. Can you just start out telling us what are the key drivers for these [00:03:00] differences, and then we can get into how we should study them.
[00:03:03] Alyce Adams: Absolutely. So first of all, I would say that the most common chronic disease is multiple chronic conditions, right?
So most people don’t just have diabetes or just have cancer. They have multiple chronic conditions that affect their lives. And so what we see is that for certain communities, particularly communities of color, lower income communities, lower communities with lower educational attainment, they tend to get these conditions earlier in the life course. And they get more of them, um, earlier on. So that’s the piling on, if you [00:03:30] will, of chronic conditions, happens very early in some of our populations. And the drivers of those, really the social determinants of health, all of the factors that we see. Certainly with Covid, we saw this, right? Where people you live in certain neighborhoods, you have less access to certain things, and that sort of compounded the issues related to covid.
In similar ways, those mechanisms have been going on in our, in our society for a long time, and they perpetuate this higher rate of chronic conditions in some of these populations. And so what we try to do is better understand, but the drivers, these drivers [00:04:00] try to identify those are actually modifiable through health system and policy interventions, and then work collectively with communities and health systems and others to see if we can design better interventions. They have the potential to address and really alleviate this burden for many of our patients and our populations.
[00:04:16] Russ Altman: Great. Thank you. And I have looked at a lot of the papers you’ve written over the last few years, and of course I think many people understand. Um, even intuitively that if you’re poor and if life has been challenging, your health might not be as good, [00:04:30] but I know it’s much more complicated than that and it’s not a question.
So, you know, you might say, okay, they don’t have enough resources, let’s throw some money at the problem.
[00:04:38] Alyce Adams: Mm-hmm.
[00:04:38] Russ Altman: And of course, people always like throwing money but I notice in your work that you’ve looked at this and there can be mixed results. So tell me about the complexities of these issues and why it’s not just poor people have bad health?
[00:04:52] Alyce Adams: That’s a really great question, and so I would start with some of our work that we’ve done around improving access to healthcare, to high quality [00:05:00] healthcare, right? So often what we talk about is one of the core drivers of the disparities is differences in access to high quality care. So what we’ve tried to do is to understand, well, when the government or even health systems make healthcare more affordable for people, what does that do to these disparities?
And what we see is one, is this a good thing? Yes. Overall, when you improve access to high quality care, what you see is an increase in the utilization of clinically essential medications for treatment of chronic conditions like hypertension, [00:05:30] cancer, diabetes. People are actually using this sort of resource, this benefit in a way that helps them out. The challenge here is that those types of initiatives have only had a modest effect on the overall disparities that we see in terms of clinical healthcare outcomes. So you kind of scratch your head and say, well, why is it we thought if we turned on this spigot, then all of a sudden all of these people would’ve get access to this valuable resource?
Well, the challenge is that if you think of it as a spigot and a hose where we’re trying to sort of address this particular [00:06:00] need, what happens is there are a number of other factors that cause kinks in that hose. So for example, a kink in the hose might be that while you have access to healthcare in terms of having health insurance, you may not be able to get there very easily either because you don’t have great public transportation where you are, or because you just don’t feel comfortable within the healthcare sitting where you’re at because you don’t feel like people treat you with respect and kindness.
[00:06:23] Russ Altman: Right.Right.
[00:06:24] Alyce Adams: Another type of kink in the hose is that when you get into the healthcare system, no one speaks the language that you do. [00:06:30] And so it makes it very difficult for you to be able to communicate your needs and they go on and on and on. And so if you imagine that hose again and all those different kinks, it’s easy to see how very easily, even if you turn on that water, you’re not seeing, you’re only seeing a sort of a trickle by the time you get to the end. And so our job as scientists is really to understand those kinks and how do we un undo them?
[00:06:49] Russ Altman: Yeah.
[00:06:49] Alyce Adams: Systematically and sort of figure out how do we keep that water flowing?
[00:06:53] Russ Altman: So, okay. So how does this research happen? This is great. You’ve given us a great setup and I can see the problems.
[00:06:58] Alyce Adams: Mm-hmm.
[00:06:58] Russ Altman: And I love the garden hose [00:07:00] and the kink analogy ’cause it’s just a very, anybody who has dealt with that knows that this is an issue. And it’s also, I’ve learned in the last 30 seconds that access doesn’t mean that, it doesn’t mean full access. It doesn’t mean that you’re showing up and you’re communicating effectively…
[00:07:16] Alyce Adams: Right.
[00:07:16] Russ Altman: …with your provider. Access has kind of a technical meaning of you’re allowed to come to the clinic.
[00:07:21] Alyce Adams: Right.
[00:07:21] Russ Altman: But it doesn’t really mean that you have come and the clinic is set up to accept you, welcome you, and engage with you. Okay. So that’s very important.
[00:07:28] Alyce Adams: Mm-hmm.
[00:07:28] Russ Altman: How do we study this? Is the [00:07:30] data good?
[00:07:30] Alyce Adams: Yeah.
[00:07:30] Russ Altman: Do elect electronic medical records help? Is it way more than that?
[00:07:35] Alyce Adams: Yeah. So yes to all of the above. I think why I get excited about this work, in particular in the area of prescription medication access is that we monitor prescription medications on a routine basis. We know when you get a prescription, we know if you’ve picked it up, we know if you come back to get it. These are all transactions, right?
[00:07:52] Russ Altman: Mmm-mm
[00:07:52] Alyce Adams: And we can follow transactions using electronic health record data as well as claims data, sometimes linking those two things together. As [00:08:00] well as linking them to other types of data, like what’s going on in your neighborhood? Do you even have a pharmacy close to your house? How far is your house to the medical center? Right? All of those things we can pull together in sort of these big data sort of ways and really look observationally over time at the same people and their experience over time with different types of medications to really start to again, sort of unravel that hose.
So one of the areas that we look at in terms of policy are federal and state policies that impact large swaths of the population. As [00:08:30] scientists, you know, we talk about experiments a lot, but the reality is policies are also experiments.
[00:08:34] Russ Altman: Mm-hmm.
[00:08:35] Alyce Adams: They’re natural experiments, right? We don’t necessarily control everything that happens, but the reality is there are these huge things that happened to large numbers of people, and often we never even ask what happened and whether they were positive or negative on the other end. And so what’s great about those natural experiments is we can identify the populations affected and then really study over time what happens, which is what we’ve done with Medicare Part D, for example. We’ve also done this with Medicaid state policies in order to really [00:09:00] evaluate did that policy have the effect that we intended it to have?
Often what you find is policies, even when well intended to increase access to a given service, what ends up happening is when the policy gets implemented, all of those factors that influence implementation, whether it’s adopted, whether people even understand it, all of those things actually have, may have a different effect and take the policy in a different direction in terms of its effect on different populations.
So one clear example is prior authorization. So that’s where [00:09:30] State medicaid programs, for example, will say for certain drugs, because they are more expensive and there are alternatives available. We are going to require the physician to ask somebody for permission before prescribing. And what’s we’ve shown in the past is that depending on how these things are implemented, while good intent there, right? We wanna save money, we don’t want peoples wasting money.
At the same time sometimes what ends up happening is if the policies are too complex. When these policies go into effect, a clinician may not understand that certain groups are exempt from the policy, nor do [00:10:00] they necessarily have time to look up which groups are exempt from the policy. So what they end up doing is sort of avoiding the hassle.
[00:10:06] Russ Altman: Yes.
[00:10:06] Alyce Adams: So if the, so the prior authorization creates hassle, it’s also an administrative burden on the health system, they might just avoid it altogether. And so you can actually dramatically overnight shift people away from specific classic classes of medication regardless of whether or not they could benefit from those medications, simply because we’ve put up a barrier that clinicians can’t navigate very well.
[00:10:25] Russ Altman: The prior authorization, you know, in my previous life as a practicing physician [00:10:30] though, I looked at those with dread. Because,
[00:10:33] Alyce Adams: Yes.
[00:10:33] Russ Altman: Because either I was in a clinic where I was the one who had to fill it out, or I was in a, you know, in a posh clinic where there was administrative assistance, and just as you described it totally determined how quickly and how well the form was filled out.
If there was an administrator who knew how to do this and who I could say, could you please fill out this form for Mrs. Jones, and this is the reason why we need the drug. Versus, oh my God, I’m 12 minutes behind, I’m 20 minutes behind and now I have to fill out this form and [00:11:00] so it totally makes sense.
[00:11:01] Alyce Adams: Yeah.
[00:11:01] Russ Altman: And I’m guessing that it’s funny ’cause as you describe these access problems, um, it sounds to me like there’s at least two many groups, but you talked about, you know, the poor and the poor in cities, uh, who might not have transportation and access. And I’m also thinking that this must be a very similar set of issues for rural folks who are far away, who have to tend to whatever they do for a living. Um, uh, do we see big differences in those two populations [00:11:30] or do they wind up looking very similar?
[00:11:32] Alyce Adams: Uh, we do see big differences, particularly in rural communities when you add on the fact they often don’t have public health infrastructures, and we know that there are these deserts out there in terms of healthcare access, right?
So in some places there are fewer pharmacies available. There may be fewer, um, uh, uh, clinics where people can actually get, uh, treatment for. Sorry, I’m blanking out right now on the sort of specific treatment I was gonna say, but really sort of there are specific types of [00:12:00] specialty care, I should say, that are harder to get in rural areas, right?
And so when that happens, um, it sort of, it compounds already existing inequalities in terms of income, in terms of education, in terms of language, in terms of access. I’ll give an example right here in Northern California while I sit here in Silicon Valley. And we have an image of what Silicon Valley is, the reality is a county in which we sit actually has some of the greatest, uh, divisions between the rich and the poor and the entire [00:12:30] world in the world. And that’s because if I drive down to Monterey Bay, when I get to hour or so outside of Palo Alto, I see people out there working in the fields exposed to all kinds of pesticides, exposed to the sun, dealing with poor water, um, quality control, all sorts of things on top of what we’re already dealing with in inner cities.
[00:12:52] Russ Altman: Yes.
[00:12:53] Alyce Adams: Right. So that po particular population is underrepresented. Right. They don’t necessarily have a collective voice. Often they’re [00:13:00] often forget they may be undocumented, so you have additional legal concerns from their behalf in terms of seeking out care. So absolutely, it can be much worse in those areas and we really have to think globally when we think about in quality and not just in terms of a specific place or a specific subgroup, but really what is inequality doing for all of us or do to all of us? And how do we work together to address it.
[00:13:20] Russ Altman: Yeah, that really good and really clear, uh, set, set of challenges. So you’ve written a bunch of stuff, uh, a bunch of stuff, period. But with some of the [00:13:30] things that really caught my, uh, interest is you wrote about disparities and kind of patient knowledge and understanding of their disease. And I think at least the paper that I saw was about tobacco use.
[00:13:39] Alyce Adams: Mm-hmm.
[00:13:39] Russ Altman: But I imagine it applies to many other areas. How do we study that and what are the that’s a hard one, right? Because if you’re not even signed up for the whole mission of health, because of the, just because of your life experience, you can imagine how frustrating and difficult it would be for providers who wanna help to get you on board with the plan. [00:14:00] So where are we with that? And what are the options there?
[00:14:04] Alyce Adams: That’s a great question, and it’s actually that kind of question that has really been driving our more recent efforts to integrate patient, caregiver, and other voices into the research we do as research partners, not just research subjects from the very beginning.
[00:14:18] Russ Altman: Mm-hmm.
[00:14:18] Alyce Adams: And sort of learn this early on, even working with policymakers is that I can come up with a gazillion questions as a resource. I’m naturally curious, right? So I love asking questions and I love questioning things. So [00:14:30] it’s no problem for me to come up with a question whether I’m asking the right questions or the questions that really address the priorities of those most affected is a different issue.
[00:14:38] Russ Altman: Yeah.
[00:14:38] Alyce Adams: Right. The only way I can get that is by talking to those people. So early on in my career, I started to talk with policymakers, whether it was tribal leaders, if I was working on Native American health issues or members of Congress or state legislators, and now, Um, a lot of my work involves patients from the very beginning. What that does is it reorients our thinking, instead of me thinking about maybe clinical outcomes that [00:15:00] NIH National Institutes of Healthcare is about necessarily, they’ve got me thinking about, well, what about my functional health? Right? You’re talking about my A one C, I wanna be able to mow my lawn, or I wanna be able to take care of my grandkids.
[00:15:10] Russ Altman: Yeah.
[00:15:10] Alyce Adams: Their orientation and their priorities are different. So they’ve really shaped how we think about the questions that we ask, the outcomes that we’re interested in, and then how we pursue them. And it’s only by doing that, I believe that we can fully understand how to integrate. Sort of what patients are thinking about and what communities are thinking about into the work that we do. [00:15:30] And so a lot of what we do end up being mixed methods. So while I may be a big data person, that’s sort of how I was raised for me in terms of my background. But I think for me, what’s been really enlightening and wonderful to learn about is qualitative methods and how we can integrate those more seamlessly into the work that we do, so that we can actually understand the data that we see. Numbers are just numbers.
[00:15:51] Russ Altman: Right.
[00:15:51] Alyce Adams: To understand people, you have to talk to people.
[00:15:53] Russ Altman: So that sounds great. And the idea of in, uh, involving the patients and their providers really is very compelling. [00:16:00] However, you and I know very well that there are populations that don’t feel well served by medical research, so I’m guessing it is not a walk in the park to say, Hey, we’re gonna do this great research with you, come join us because some people might say, well, wait a minute. I’ve heard a lot of stories that are true about my group, uh, my, uh, cohort not being treated well by the medical research, um, establishment in the past. Does that come up in these kinds of, uh, projects?
[00:16:29] Alyce Adams: All the time. [00:16:30] All the time. And I think, you know, part of it is as researchers, we first have to walk in with this, a great deal of humility. I may know my field of research really well, right? I may understand how to run natural experiments incredibly well. However, I don’t know what it’s like to live in some of these communities. Multiple chronic conditions, sort of having the weight of his historic, um, um, oppression on top of it. I don’t have that experience. I come from a different place [00:17:00] and so having the humility to understand that I don’t know everything before I walk in the room is incredibly important. So I think researchers first have to do their own homework Yeah. To understand what’s the history there.
The second thing we have to do is acknowledge it. We can’t pretend like I sit in an ivory tower. I’m here at Stanford, right? I can’t pretend that Stanford has not had a very negative impact on the communities that live here. You know, so I, so for me to sort of pretend that doesn’t exist, doesn’t help anybody.
[00:17:26] Russ Altman: So you have to get it out on the table.
[00:17:28] Alyce Adams: We have to get it out on the [00:17:30] table. The third thing we have to do is listen and truly listen in order to understand. So rather than sort of go in to say, I need a clinical trial run. How do I get there? How do I get people sign onto my study? You have to go with a completely different mindset, which is how can I be a good partner to these communities? And that’s what we’re trying to build here at Stanford. We have an Office of Cancer Health Equity where we’re really working directly with communities. We have a wonderful community advisory board. We were out in East Bay just about a week and a half ago with the whole sort cancer service [00:18:00] line team, right?
We all went out to the East Bay and it was remarkable to be out there to talk with people about their experiences. Many of them had never seen Stanford in the community before, so that was humbling for us. But it was also incredibly important for us to hear that, to hear what they think about Stanford in order to have that conversation and Stanford research in particular.
So I think that engagement and showing up and showing up consistently and maintaining that humility. And the last thing that I’ve learned, mostly I’ve learned is one of my community partners, Ms. Libby Hoy, who has her own patient group that [00:18:30] she PF, uh, PFA network and uh, PFCC Partners, which is a group of patients and caregivers and others, really focused on integrating. How do we, sort of systematically integrate patients and caregivers into health systems and into research. And she tells me it’s about the infrastructure.
[00:18:45] Russ Altman: Hmm.
[00:18:46] Alyce Adams: You have to set up an infrastructure, create a charter, which delineates what the goals are and what we’re gonna do. You have to have a north star, right? Something that we’re all working towards together. You have to agree ahead of time. What does engagement mean? What does it look like? And you have to constantly [00:19:00] evaluate the process. How are we doing? Are we, is, how is it working for all the partners? And those kinds of things. It’s only through that iterative process. It doesn’t mean you’re not gonna make mistakes. But the chances of you being able to come back from those mistakes is much higher if you show that consistency and that willingness to work in sort of bipartisan relationships with community partners.
[00:19:18] Russ Altman: This is The Future of Everything with Russ Altman. More with Alyce Adams next.[00:19:30]
Welcome back to The Future of Everything. I’m Russ Altman. I’m speaking with Professor Alyce Adams from Stanford University.
In the last segment, Alyce told us about the challenges of ensuring that people with chronic disease from poorly served communities get the healthcare they need. She told us that policies with good intentions can sometimes lead to not improved outcomes.
In this segment, she will tell us how research is becoming broader and how engaging patients [00:20:00] and their providers is giving us information that’s critical for removing the barriers to care. She will also tell us that policymakers are listening and are open to using the data to make new policies and interventions that maximize the chance that people with chronic disease will get good care.
Alyce, I wanted to ask you that I think this falls generally under the field of health services research, health policy. How is that field doing and what does the future hold? Uh, we, uh, we hear about AI. Is [00:20:30] AI making an impact in this field? Um, Where are we going?
[00:20:33] Alyce Adams: Absolutely. So health services research has been around for quite some time, even though most people probably don’t know what that is. And I think the reason is that it’s inherently multidisciplinary. We sort of bring together sort of the strengths of fields like economics, statistics, uh, political science, you know, a lot of different types of fields and tried to use that information or try to use those theories to fundamentally [00:21:00] understand how individuals, um, utilize healthcare services and what are the influences of various factors that influence their access to those healthcare services and what can be done to really drive us towards greater efficiency. Sort of lower costs, but sort of higher bang for the buck in terms of what we get out of it, but also really trying to understand inequities in the system and how those can be adjusted, right? So if you think about healthcare as sort of a, you know, a public good, if you [00:21:30] will, something that we all strive for form that we think everyone has the right to. The question then becomes, well, we know historically some groups have underutilized. We’ve all under-utilized things that we know are beneficial to us, and we know that that’s complicated. That’s not just about individual human behavior, right? That really also in sort of encapsulates the interactions between people and their providers, the context and of care is delivered in terms of the healthcare system. What cardiac system is it? Who owns it? How was it run? What is sort of, how [00:22:00] is it organized? Who’s on board to actually provide care in those systems and how are they paid? Um, how do we evaluate the quality of care all the way through to policy makers who decide how they get paid, how much, um, what’s appropriate sort of nursing ratios.
There’s like so many different financial and organizational aspects to that. And really AI is helping us to do, um, to think about this in a number of different ways. So one of the things is that we know that health is complicated. Even any given disease is very [00:22:30] complicated. There are a lot of different variables that go into understanding sort of whether someone’s at risk for a particular disease and what their prognosis will be and what their trajectories will be. It’s nearly impossible for a given individual to hold all of those different factors. In their heads as they’re making a decision is what should we ask of clinicians? So a lot of the work that we do in AI, particularly in the healthcare context, is to build algorithms to augment decision making by providing additional information and or really trying to understand and bring together diverse sets of [00:23:00] information. For the provider in a cogent way, such that when they’re sitting there with a patient or when they’re thinking about an overall patient panel or the health system’s thinking about implementing some sort of intervention, we can use AI to help inform that practice. So for example, we developed an algorithm at Kaiser Permanente California when we had been talking with patients about diabetic neuropathy, which is nerve damage, it’s caused by diabetes. It’s a leading cause, a risk for lower extremity amputation in the world. So we noticed that a lot of patients were talking about [00:23:30] neuropathy, but really did not necessarily have, um, the conversation with our patient, the doc, with our physicians, about that condition. So we said, well, there’s something going on here.
We worked with them very closely to better understand what was happening. And one of the things we realized was for a clinician, they could not tell for the patient in front of them what the likelihood was that patient in particular was at particularly high risk for bad outcome versus another patient. And when you have so many up to 50% of diabetes patients who are adults who’ve had diabetes for 10 years or more who have [00:24:00] neuropathy, that’s problematic. We develop,
[00:24:01] Russ Altman: And one of the things you said is that this is not just about the diabetes disease. There’s a zillion other things that physician might not know about and this could be helpful in that way.
[00:24:12] Alyce Adams: Absolutely. So taking all of the different, so these large electronic health records, taking all of this diverse information and pulling it in a, using machine learning to pull all that information together to provide a simple risk score, if you will, for a given patient that enables the physician or the clinician who was ever working [00:24:30] with a risk score to sort of make different decisions about a particular patient’s risk.
[00:24:34] Russ Altman: Okay.
[00:24:34] Alyce Adams: And so we look to see like adverse events over the near term and so you see this ad. So, um, algorithms are often used in that way in healthcare settings is really to help us assess risk, either for individuals or for groups of patients in ways that help doctors make better decisions.
[00:24:47] Russ Altman: Great. So I’m glad you gave that example ’cause it is so rich and I’m exploding now with questions because, uh, there are many perspectives, as you said, this is a multidisciplinary field. There’s [00:25:00] many players that have different or only slightly aligned interests. You mentioned payers who have a certain perspective, and then clinicians, clinicians, patients. Um, then when you implement these algorithms, how do you think about the perspective like who’s priorities should be encoded in these algorithms. And then how do you make sure that those priorities are actually being encoded and not some priorities of somebody who’s saying, my only goal is to spend the least amount of [00:25:30] money. In which case there’s gonna be a certain number of decisions that are gonna be obvious, like, no, no, no. Uh uh. So tell me about that.
[00:25:37] Alyce Adams: Oh, it’s incredibly rich and exciting to be working in this area. I would say one of the things we thought about from the very beginning was, who’s gonna have access to this information and how will they use it? Right?
Because that’s gonna dictate who, how comfortable physicians feel about it, how comfortable patients feel about it. So one of the first things we decided is this was not be something that sits in the hands of the health plan. So someone who’s deciding who gets [00:26:00] covered or not.
[00:26:00] Russ Altman: Right.
[00:26:00] Alyce Adams: It can’t sit there. It has to sit with the clinician who is the trusted provider of the patient.
[00:26:05] Russ Altman: Right.
[00:26:06] Alyce Adams: We also talked about out sort of which clinician, right? Who actually is going to use this? Is it the primary care physician? Is it a nurse practitioner or is it someone else? And that really depends on the healthcare system in which you’re working.
[00:26:17] Russ Altman: Yes.
[00:26:17] Alyce Adams: So at Kaiser Permanente, which is a physician, run an organization, for the most part, we definitely have a lot of nurses and others, often as a physician making a decision. So we worked with physicians to then design the algorithm, essentially, not the algorithm [00:26:30] itself, but how the algorithm might be used. But here’s the rub. The challenge here is it’s almost impossible to anticipate all of the different ways in which these algorithms could, um, be used or misused, right? We also have to think about subgroups who are inherently harmed by these algorithms because of biases in the data itself.
[00:26:50] Russ Altman: Yeah.
[00:26:50] Alyce Adams: That then get perpetuated. We also have to think about again, how others might interpret the results of the algorithm and how those results might be interpreted differently [00:27:00] depending on who’s sitting in front of them. Right. So none of those things go away. So more and more we’re pulling from the bioethical literature.
And from our bioethics experts to help us sort of delineate what are some of these pitfalls? So what are the benefits and what are the pitfalls? And then how do we think about those things from the very beginning when we’re initially designing the algorithm. So we’re really taking more of a systems level approach. I think when we started in healthcare, it really was we want a tool, we wanna deploy that tool to address a problem.
[00:27:28] Russ Altman: Right .
[00:27:28] Alyce Adams: Now we’re realizing it’s not [00:27:30] enough to just focus on the tool. You really have to think about the whole system in which the tool is being implemented because that’s really what’s going to dictate whether this algorithm on the whole is gonna be helpful or harmful to individual patients or to heal individual communities.
[00:27:43] Russ Altman: That’s great, and that leads me into, in the last couple of minutes I wanted to ask you about, we. Policies. So you said at the very beginning of our conversation that policies, we can think of them as little experiments where we make a perturbation and then we are hoping that the world is a better place because of the policy. Uh, and then you said part of your goal is [00:28:00] to actually do the follow-up so we understand the implications. So what is the future of these policies like? Where are, are we starting to see that because of the work that you and your colleagues do that the policies are now coming in with a little bit more information, um, or is that a Pollyanna-ish thing to say? And we’re gonna continue to have kind of random policies that we then always study in arrears, but not prospectively.
[00:28:24] Alyce Adams: Gosh, you know, I wish I could tell you yes, I really wish I could. I think some of the [00:28:30] challenges are that even when policies are well-meaning by well-meaning, I think they’re being written by thoughtful people who are trying to make things better for everyone as opposed to making things worse for a specific group, right? So even in the best circumstances when policy’s done that way, often is driven by budgets. Not by the research, right?
So we’ve been talking a lot about how do we integrate research early on. Some policy makers will be proactive about that, but on the whole, they’re not. So I think we as researchers have to get more savvy about [00:29:00] communicating about our research in ways that anyone can understand.
[00:29:03] Russ Altman: Yeah.
[00:29:03] Alyce Adams: Right? It’s not about showing how smart we are, it’s really about making sure that we disseminate that information.
[00:29:08] Russ Altman: You should be on a podcast.
[00:29:09] Alyce Adams: and we can talk about it’s limitation.
[00:29:11] Russ Altman: I’m sorry.
[00:29:13] Alyce Adams: Great idea.
[00:29:14] Russ Altman: Please continue. I couldn’t help. Couldn’t help myself.
[00:29:19] Alyce Adams: No, that’s fantastic. But I think that’s true. I think that’s absolutely true, right? I mean, honestly, I started working in this field, what, almost 25 years ago now. And podcasts were just, didn’t, wasn’t even on [00:29:30] anybody’s radar. Uh, but we talked a lot about the central problem. And the reality is policy makers, that’s how they’re finding out about things. They’re listening to podcasts. They were, they’re, their staffers are looking at Twitter, they’re looking at LinkedIn. They’re trying to find out. Who’s saying what about what’s happening? And then they dig deeper. So we also have to adapt again, sort of get out of our ivory towers and really think about, go to where people are, talk with them about these policies and their implications, but also help them to understand the implications of our research and hopefully in the end also [00:30:00] make our research more responsible.
[00:30:01] Russ Altman: So that’s great and my guess my final question, and I really don’t know what you’re gonna say is, Have the policy makers shown an interest? Are they taking your calls? Should we be optimistic about them listening?
[00:30:14] Alyce Adams: Yes, so I think they definitely are certainly here at Stanford, the human, um, centered artificial intelligence.
Um, institute has an annual, uh, meeting where we have congressional staffers come in, is actually coming up. Um, and so I spoke in that last year. I’ll do it [00:30:30] again. This year about the ethics of AI in healthcare, and I’ll give some of those specific…
[00:30:34] Russ Altman: Right.
[00:30:34] Alyce Adams: ….examples around neuropathy. We also have associations, so we have a national Association, academy Health of Health Services researchers, and we talk to policy makers all the time. We make ourselves available to a bipartisan, in a bipartisan way. It’s not about the politics. It really is about elevating the science in a way that hopefully can be useful for real world decision making.
[00:30:55] Russ Altman: Thanks to Alyce Adams. That was the future of health outcomes.
“Stanford University, officially Leland Stanford Junior University, is a private research university in Stanford, California. The campus occupies 8,180 acres, among the largest in the United States, and enrols over 17,000 students.”
