The Not-So-Straightforward Path Between Medicaid and Cancer Health Disparities

November 20, 2023

Information

Yale Cancer Center Grand Rounds | November 17, 2023

Presented by: Dr. Cathy Bradley

ID10998

To CiteDCA Citation Guide

00:03OK. Good morning, everyone.
00:04We're going to go ahead and get started.
00:07Thank you all for being here.
00:08We have a really special treat for you today.
00:10So we are being joined
00:12by Doctor Kathy Bradley,
00:13who is Professor and Dean of the
00:16Colorado School of Public Health as
00:17well as the Deputy Director of the
00:20University of Colorado Cancer Center,
00:22where she also holds the Paul
00:24Bunn Chair in Cancer Research.
00:26So Prior to joining the
00:28University of Colorado Dean,
00:30Bradley was the founding Chair
00:32of the Department of Healthcare
00:33Policy and Research for VCU.
00:35She's a health economist and
00:37received her PhD and MPA from the
00:40University of North Carolina,
00:42Chapel Hill.
00:42She has served on the National
00:44Academies of Science,
00:45Engineering and Medicine's National Cancer
00:48Policy Forum and formerly served on
00:51the National Advisory Committee to HRQ.
00:53She currently serves on the
00:55Methodology Committee for Pacori.
00:57Doctor Bradley has received numerous awards
00:59and honors including the Women in Science,
01:01Dentistry and Medicine Professional
01:03Achievement Award and Leadership,
01:05and she maintains an active resource
01:07portfolio of NIH and Foundation
01:09funded grants where she leads research
01:12related to cancer disparities and outcomes,
01:15financial burden,
01:16and labor market outcomes of
01:18cancer survivors.
01:19Please join me in welcoming Dean Bradley.
01:27Good morning, everyone and thank
01:29you for that introduction and thank
01:31you for the opportunity to be here.
01:33I'm so fortunate to have had
01:35two beautiful days in New Haven.
01:37It's just been fantastic.
01:38I'm happy to report that I
01:40did get to try the pizza.
01:44Everyone, when I tell them
01:46I was given a talk here,
01:47they were saying make sure you
01:49get out and try the pizza.
01:50So yesterday morning,
01:51that's what I had for breakfast.
01:55So and this is not my first time being in
01:58New Haven and giving grand rounds at Yale.
02:02The first time I did it
02:03was probably 20 years ago.
02:05I was an assistant professor at Michigan
02:07State University and Doctor Ruth Mccorkle,
02:10who was a professor in nursing,
02:12invited me out and the room
02:14was entirely different.
02:15And I was telling Michaela about it,
02:16that it was in the this room that was like a,
02:20a well that looked down where they
02:22used to do the old grand rounds
02:24with the patient down below.
02:26And it had to be probably one of the
02:28most intimidating things I've ever done,
02:31being there as an assistant professor and
02:33then sort of in that particular setting.
02:35But it is fantastic to be back.
02:39I titled my presentation The
02:41Winding Path between Medicaid and
02:42Cancer Health Disparities.
02:44And that's because nothing with Medicaid
02:46is straightforward and certainly
02:48nothing with cancer is straightforward.
02:50That too tends to be quite circuitous.
02:53So before launching into the presentation,
02:55I want to sort of go ahead and do the
02:58spoiler alert and talk about the three
03:00things that I think are are takeaways.
03:02The 1st is just the complexity of the
03:05problem, and I think as researchers,
03:08we like a good problem, you know.
03:10So this one is a particularly good problem.
03:13That is something with lots of facets.
03:16And as you think about your own research,
03:17especially if you're doing disparities,
03:20it's hard not to talk about
03:21Medicaid and Medicaid coverage.
03:22It's probably our best hope for
03:25narrowing disparities that just seem
03:28to be widening no matter what we do.
03:30But it does provide coverage to a
03:33population that so desperately needs it.
03:36And so understanding the program
03:38is pretty critical.
03:40And in places that haven't expanded
03:43Medicaid or really maligned
03:45against the Affordable Care Act,
03:48one of the arguments you frequently
03:51hear is substandard care.
03:52It's just terrible healthcare.
03:55And we whether that's true or not,
03:59it's something that is said frequently
04:01and it's up to us to figure that out.
04:03And if it is substandard in some places,
04:05where are those places and how do
04:07we improve it in our policies.
04:09So that's the first area is to
04:11really think about this problem.
04:13The second thing is around data.
04:15And we all need data to do our research.
04:19But but you know,
04:20secretly we hate data because
04:22it's so hard to acquire.
04:24And once we finally got it,
04:26you know,
04:27we get it from these agencies that are
04:29holding data that aren't met for research.
04:31We take it and we're clever.
04:33We're going to merge it with
04:34different data sets and stuff.
04:35But now what do we have in front of us?
04:37And it's understanding that data
04:40and the importance of doing so.
04:42And then the final take away is really
04:44all about the data infrastructure.
04:46And by not having a good infrastructure
04:49because our health system in the United
04:52States is so incredibly fragmented
04:54that we don't have a comprehensive
04:57data infrastructure that we can just
05:00pull down and understand what's
05:02happening with the quality of care,
05:04especially with those who most need it.
05:07So at the end of the presentation,
05:10I leave time for questions.
05:11So please be thinking about those questions.
05:14I really love that part.
05:15And being able to have a discussion,
05:18I think that's the most interesting thing.
05:20So going forward, first I'd like to
05:23just acknowledge the wonderful people
05:25I have the pleasure to work with.
05:26Marcelo, Rich,
05:27Sarah and Faye All from Colorado.
05:31Lindsay Sabick,
05:32University of Pittsburgh always love
05:35working with her and then our colleagues
05:37at the cancer registry and civic
05:39who holds the all payer claims data.
05:42Colorado is one of the few states that
05:45have an all payer claims database that's
05:48available for research and they have
05:51been wonderful about working with us,
05:54but they are not cheap.
05:57The agenda for the talk,
05:59just an overview of Medicaid
06:01getting us all on the same page,
06:03factors that affect cancer outcomes,
06:07specific factors about Medicaid and
06:09that enrollment that can affect what
06:12happens to you once you're diagnosed,
06:14treated and become a cancer survivor.
06:17Then the factors that affect research,
06:19the incomplete data we might have,
06:21the nuances of our data and the
06:24importance of understanding it.
06:26And then just to get a sense from all of you,
06:28the disparities between Medicaid
06:30and other forms of insurance
06:33are what is going on there?
06:35Is it inadequate care?
06:37What do we do about it?
06:40And then wrapping up with
06:42some future directions,
06:44Nothing really to disclose except
06:46that this grant was funded by the
06:48National Cancer Institute and Marcelo
06:51and I are Co principal investigators.
06:56So why this is such an interesting problem?
06:58And I have the puzzle pieces there
07:00because I'm guessing like me,
07:02many of you enjoy a good puzzle, right?
07:04Figure things out.
07:06So what we want to understand is Medicaid,
07:10is it a safety net savior or this
07:13malign purveyor of inadequate care?
07:15Which one is it?
07:16So many years ago when I was here presenting,
07:20I was actually presenting
07:22about Medicaid and cancer,
07:23sort of the very first research I
07:25did in this area and was trying
07:28to really understand things.
07:29And you know,
07:30I was young and and and stupid
07:33and many employees and working
07:35with the state health department
07:38around getting their Medicaid
07:39data and merging it with cancer.
07:42And it was really a complicated process and
07:45they didn't want to let go of their data.
07:48And I have found that these individuals
07:51insured by Medicaid were had worse survival.
07:54And I said something along the lines of
07:57it's a safety net just above the grave,
08:00which did not make the Medicaid people
08:02want to give me their data anymore.
08:04And that wasn't a way to form
08:07the relationship.
08:07So an important lesson learned at that time,
08:10but it formed kind of the basis really
08:13of understanding why is it that people
08:16who are insured by Medicaid did so
08:18much worse and was it the insurance,
08:20Was it something about them?
08:22Was it,
08:23you know,
08:23that they had tons of comorbidities,
08:25got in late but kind of led
08:27to this circuitous journey.
08:29So the next point of enroll too late
08:32and lack of continuous coverage.
08:34If you come in once you're diagnosed,
08:35you probably have later stage
08:37disease and probably have other
08:39problems that aren't being cared for.
08:42So what's the right comparison group?
08:44Is it people have had insurance
08:46all along or people have Medicaid
08:48insurance all along.
08:49So anyway,
08:50thinking about that reimbursement,
08:52So over time,
08:54we know Medicaid reimbursed at a
08:57much lower rate even for people
09:00who are diagnosed through the CDCS
09:04National Cancer or breast and cervical
09:06cancer early detection program that
09:08once they're enrolled in Medicaid
09:10for treatment of their cancer,
09:12that care is reimbursed in
09:14some states at an even
09:16lower rate than a normal Medicaid patient.
09:19So there's all kinds of things that we
09:22do around reimbursement that prohibits
09:25access and then data complexity.
09:27Medicaid data is a mess.
09:29I mean, it is just a complete mess.
09:32People come into Medicaid,
09:33they drop off the next month.
09:35We don't know what happens to them.
09:37And it's trying to understand
09:39how they got into Medicaid.
09:41And because Medicaid is the
09:44pair of absolute last resort,
09:47you may be missing claims if they have
09:49any other type of health insurance. So
09:54Medicaid is the largest insurance
09:56program in the United States and
09:59for most states it is larger now
10:01than their education program.
10:03So it is just a huge program across
10:07the country and in every state
10:09it is ministered differently.
10:12Large provider of people of color protects
10:17against major financial consequences,
10:20which is what insurance is supposed to do,
10:22is to give you that insurance,
10:24that insurance against losing everything.
10:26And then this huge variability.
10:28And the graph that I show is the proportion
10:31of people by different racial ethnic
10:34groups that are covered by Medicaid
10:39under the ACA.
10:40We were supposed to expand Medicaid,
10:42but then it was left up to the
10:45states to do so. At this point,
10:4710 states have still not expanded Medicaid.
10:50North Carolina is set to begin was set
10:54to begin in the beginning of December
10:57if they are able to launch the program.
11:00So still some important holdouts
11:04with Medicaid expansion and
11:07in those particular states,
11:10the threshold for Medicaid to
11:12qualify for Medicaid is quite low.
11:14So one of the things we used to say in
11:17Virginia was you really could not cut
11:19grass and still qualify for Medicaid.
11:21The the level was 13% of
11:24the federal poverty line,
11:26pretty astounding right
11:33As Medicaid expand, we do know that it
11:36provided a lot of good to a lot of people.
11:39It did increase access to care.
11:41We observed improvements in some health
11:44outcomes and it contributed to reductions
11:47in racial dispar racial and ethnic
11:50disparities in healthcare coverage.
11:52So by and large it seemed to be
11:54doing some of the things that
11:56we had hoped that it would.
11:57We we've start to see improvements both
12:00in access and in some health outcomes.
12:04In expansion states,
12:05cancer survivors had greater
12:07access to doctors and non compared
12:10to non expansion states,
12:12women had a lower odds of receiving
12:15recommended mammograms or Pap smears,
12:18and expansion was associated with
12:20earlier detection and appropriate
12:22cancer treatment and in reduced
12:25mortality for those who were able
12:27to receive Medicaid coverage.
12:29So this is some of the work
12:32that Lindsay Sabek and I and
12:34her colleagues were able to do,
12:36showing pre and post expansion and
12:38these differences that it made.
12:41So it does seem to be better than not
12:45having insurance and having to go
12:47through the traditional safety net of
12:49showing up at a safety net hospital.
12:52Despite these improvements though,
12:54we know that there's some variability
12:57that the same benefit was not
12:59seen across the board.
13:01So we have here some evidence that
13:04newly diagnosed patients there
13:06was improved 2 years survival.
13:08But in cancer sites such as
13:11urologic malignancies,
13:12there was no change in stage at
13:15presentation And that the thyroid
13:17cancer showed that Medicaid patients
13:19were still likely to be diagnosed at
13:22an advanced stage and less likely to
13:24receive a guideline from coordinate care.
13:26So the pictures not really complete.
13:28We have evidence and that's a lot
13:30of what we do as researchers.
13:32We we build a body of evidence and
13:35doesn't always agree with each other.
13:37It's the body of evidence and we try
13:39to make the most out of the studies we
13:41have and to understand the validity,
13:44the credibility that they do everything
13:46right and how does this body of
13:48evidence build in One Direction or other.
13:50And what we see with cancer and Medicaid,
13:53it's not initially a clear story,
13:56but there's a signal and I think
13:58it's a reasonably strong signal
14:00that Medicaid is beneficial.
14:04So this is where I coming back to
14:07when I gave the talk here long ago,
14:10I looked at Medicaid merged
14:13with our state cancer registry.
14:15We also merged in Medicare data as well,
14:19one of the first states to do that
14:21and my long term colleague who is
14:25still around had lots and lots of
14:27experience in working with data sets.
14:30Very patient person,
14:32good contrast to me,
14:34especially at that time in my life.
14:36And he said the reason we're doing
14:39this is because we don't have a
14:41meat grinder to put our hand in.
14:43And so it was kind of an interesting
14:47way to think about having to go out
14:49and get this data from the state agency
14:52who had never used it for research
14:54purposes and was in a completely
14:56different part of the state agency
14:58where the cancer registry was held.
15:00And we were just very,
15:01very fortunate that they were all
15:03willing to work together and do
15:05this and create this resource.
15:06And what we found there is that among
15:09people who were insured by Medicaid.
15:11So we have only Medicaid.
15:13The differences between black and white
15:16women and mortality disappeared when
15:18they received the same kind of treatment.
15:20And that at the time,
15:21that's my most cited paper,
15:23interestingly enough,
15:24and it was when I first published it.
15:27Most of the papers that cited it was
15:30pointing out that's not the case,
15:32that there are important racial differences.
15:34And now in recent years the citations are,
15:37you know that is probably the case that
15:39if you do treat everybody the same,
15:41you're probably going to get the
15:43same outcome.
15:43The differences aren't that great.
15:46So it was an interesting study and a
15:48place that where having Medicaid data
15:51and being able to look at people that
15:54are about the same socioeconomic status,
15:57being able to see if there are differences.
16:06So sorry, it looks like something has
16:10got out of order and apologies for that.
16:12So we're going to start
16:14with the enroll too late.
16:15So why is it that despite being able
16:19to show that there are promising,
16:23there's promising evidence towards
16:26Medicaid being a beneficial
16:28expansion for individuals?
16:30Why is it that some of the
16:32disparities continue to persist?
16:34And this is a study that I did
16:36with the National Cancer Institute
16:39colleagues where we actually
16:41took national RCR Medicare group,
16:44was interested in expanding to
16:46CR Medicaid and we merged the two
16:48and started looking at the data.
16:50And what you find that many people
16:54don't get into Medicaid until after
16:56they've been diagnosed with cancer.
16:58And by many people,
16:59I mean more than 1/3 or so really
17:02don't show up into the system
17:04until they're diagnosed.
17:06So they go to an emergency
17:08department somewhere,
17:08sometimes for something else,
17:11sometimes it's for symptoms.
17:13Some tests get done,
17:14find out there's cancer and there's
17:16a social worker financial person
17:18at with associated the hospital
17:20who really whose job is to make
17:22sure they get paid.
17:23They figure out that the person's eligible
17:26for Medicaid and they get them enrolled.
17:28Medicaid then becomes a retrospective
17:31coverage going back and picks up the
17:35claims that occurred from diagnosis forward.
17:38When they come in at that point,
17:41it's because they're symptomatic
17:43and they're having problems.
17:44So of course,
17:46Medicaid doesn't have much of a
17:48chance to really provide them the
17:51kind of care where you're going
17:53to see the same mortality outcome
17:56even if they have screening.
17:59So the breast and cervical cancer
18:01program is an interesting one that
18:03we were able to look at 'cause
18:05we could see how people came into
18:08the Medicaid program.
18:09So the CD CS program has been around a
18:12long time and it provides site specific care.
18:15So free screening to women who do not
18:19have insurance coverage or who are
18:22underinsured but they don't qualify
18:23for Medicaid can get free screening.
18:26So they have a little bit more
18:28money income resources than your
18:31typical Medicaid insured person.
18:34So if they go through, get the screening,
18:36they are then enrolled in Medicaid
18:38for their care.
18:40And you might ask, well,
18:41Gee, you know,
18:42they have a higher income status.
18:44They don't qualify for Medicaid.
18:46They might be better off and we'd
18:49expect them to do better than say,
18:52the person who's been enrolled in
18:54Medicaid all along.
18:55What we did in this study is we looked at
18:58women who came in through the CD CS program.
19:01We looked at women who've been
19:03insured by Medicaid all along and
19:06those who came in after diagnosis.
19:08And we're able to show that the across
19:12the board that those who enrolled
19:15in Medicaid all along did better,
19:18did better than those that
19:20came in through the CDC.
19:22Those who came in after
19:24diagnosis and Medicaid,
19:26while not the same as privately
19:29insured and we showed that
19:31here that they are still doing
19:35much, much better.
19:36And in cervical cancer,
19:37those who were continuously enrolled
19:39in Medicaid actually did better than
19:42women who were privately insured.
19:44So we are seeing a difference that
19:47doesn't the outcomes are not the
19:49same in terms of detection and
19:52mortality as private insurance.
19:54They're just not.
19:55I mean, these are individuals with other
19:57kinds of problems and other challenges,
19:59but if they have continuous coverage,
20:02they do better.
20:03It's the fact that Medicaid is
20:06picking them up when it's already
20:09fairly late in their disease process.
20:12And if we think of it as a
20:14public insurance program,
20:14is that the best way to spend our money?
20:17We're not going to have the best outcomes.
20:20It is expensive care.
20:22At this point,
20:23isn't it better to have them in
20:25a program continuously covered,
20:27getting screening,
20:28getting less expensive care and having much,
20:32much better outcomes over time.
20:37We also looked at Medicaid and and
20:41found across different cancer sites
20:43in Michigan that people enrolled
20:46in Medicaid after diagnosis had
20:48an 8 year lower survival rate.
20:50So big big difference that compared
20:53to Medicaid enrolled continuously
20:56and non Medicaid patients.
20:59There are other studies that have
21:00been done both in North Carolina and
21:02Missouri that has similar findings
21:04and that they attributed also to
21:06the timing of enrollment and we're
21:09able to see this survival gap.
21:14The next question about the
21:16problem is can they see a doctor,
21:18Is it these low reimbursement
21:20rates that hinder accessing care,
21:23So you give them care and they can't get in.
21:25And this was a study by
21:27one of your colleagues,
21:28Victoria Marks here that did a
21:31fascinating study of calling and
21:34trying to get paid an appointment
21:36and and found that they could not
21:39get in that many people just simply
21:42didn't accept Medicaid or in some
21:45safety net institutions what they
21:47do and they they did this at VCU,
21:49which was a large safety net institution.
21:52They booked four people who had Medicaid
21:55insurance for the same slot that
21:57would come in because they anticipated
22:00no shows difficulty getting there,
22:02not coming in And so incredible wait times.
22:06So really fascinating problem that if you
22:10don't at least get reimbursement up to
22:13the point of Medicare may be difficult
22:17getting in managed care programs.
22:19A lot of states have Medicaid managed
22:22care as their approach to Medicaid
22:24delivery to try to offset some of that
22:27to bring in a more managed program.
22:32Savick and colleagues dug a little
22:35bit deeper in this and found that
22:38a mostly positive impact on breast
22:40and cervical cancer screening with
22:42increased physician payments and under
22:44a fee for service managed care plan
22:47reimbursement had less of an impact.
22:49Says that those kind of delivery
22:51plans had already was doing some
22:53things to manage and get people in.
22:56They had agreed to take on Medicaid
22:58patients to begin with and so
23:00the reimbursement did not matter
23:02as much as you would expect.
23:05So I'll take a little breather at this point.
23:08And are there disparities,
23:11what do you think it are there
23:13disparities in the way that people
23:15are treated on Medicaid insurance
23:17compared to other forms of insurance?
23:20So disparities there not there,
23:24seen a lot of nods.
23:25Yep, there's still disparities.
23:27Do you think it's mostly because
23:29of the timing of enrollment,
23:35OK reimbursement, there are lots
23:40of nods on the reimbursement or
23:42is it just they do provide a poor
23:45quality of care and this is a
23:47difficult to treat population.
23:48Is it something endogenous?
23:50In other words there yeah,
23:53not a, not a lot of people buying
23:54that particular argument.
23:56Lot of times
23:59providers, clinicians,
23:59they don't know what kind of insurance
24:02their patient has when they get in front
24:03of them by the time they're there and
24:05the same kind of treatment is provided,
24:07it's getting in the door is the problem
24:11and it's and it's once they are there,
24:14I don't think that those who
24:16actually treat them and lay hands
24:18on them really at that point know
24:20what kind of health insurance.
24:22They may know at some point the
24:24treatment trajectory as they go forward
24:27around reimbursement rates and things.
24:28But initially that cares that no. OK.
24:37So let's understand the data that
24:39we're working with the research so far.
24:42To just recap,
24:43Medicaid is an important safety net,
24:45but it does appear to have some holes.
24:48There's a problem with enrollment
24:49and continuous care.
24:51People who qualify for Medicaid
24:52aren't enrolled in the program.
24:54They just don't realize they
24:56are certainly in Colorado.
24:58We see a lot of people coming
25:02in who qualify for Medicaid,
25:04but they're worried about their citizenship
25:07status and Colorado has a don't ask
25:09policy and we just bring them in.
25:12It tends to be more general,
25:14but they don't want to approach
25:16the health system because of that.
25:18Various other reasons that we
25:20see that people are shying away,
25:22but they qualify none the less.
25:24So we have a problem there
25:26with continuous coverage,
25:28and then we need to understand once
25:30again and what is really happening.
25:33And our team then began to wonder,
25:36well,
25:36what if the data are not telling
25:38the complete story?
25:39What if there's something inherently
25:41wrong with being able to look
25:43because most of the research,
25:45because as my colleague said about
25:46having a meat get grinder to put
25:48our hands in to get all this data.
25:49It's complicated and it's costly
25:51and it takes years of forming those
25:54relationships and being incredibly
25:56patient to get all of those pieces in place.
25:59As a result,
26:00many people use cancer registry.
26:02They use the CR data,
26:03which they can be able to pull easy or NCDB,
26:06other kinds of cancer registry data
26:08that they can get their hands on.
26:11And the question we began to ask well,
26:13what if those data are not right?
26:16So
26:19turning to this part of the study
26:21is from a paper that just recently
26:24got published in JAMA Health Forum.
26:26And there we started to,
26:28we took the all pair claims data,
26:31merged it with our state cancer registry.
26:34And for the first time I was able
26:36to actually compare to private
26:38insurance and to be able to do
26:40lots of controls in the data to
26:43get an equivalent control group.
26:44So it's pretty exciting to
26:46be able to do this.
26:48We started off with the question
26:50of are there treatment disparities
26:52and radiation and hormonal therapy
26:54among women covered by Medicaid
26:56compared to private insurance.
26:58And we compared what was in a cancer
27:01registry versus insurance claims.
27:02And to be able to do this and this
27:04step of our research project,
27:06this wasn't what we intended to start to,
27:08was really our validation.
27:09We were trying to figure out where the
27:11data good and where might some holes be.
27:16And this is the step we all do in our data.
27:18And we think, OK,
27:19we're done with that.
27:20Nobody's going to be interesting,
27:21but that ended up being the story,
27:26our research question,
27:27we knew that there the literature was
27:30filled with papers that women insured by
27:32Medicaid did not get radiation therapy.
27:35They were not put on hormonal
27:37therapy relative to women of
27:39other forms of health insurance.
27:41So we started there and we thought,
27:43OK,
27:43we're going to compare to private
27:45insurance because this is a
27:47group that were picked women
27:49who were younger than age 65.
27:51Then we were going to go through and
27:53just do this toughest comparison
27:55private insurance where they should
27:57be getting the best care compared
27:59to to a public insurance program.
28:02And there's some nuances about Colorado's
28:05Medicaid that I'll get back to,
28:06but this was the setup for our
28:09study and here are some of the
28:11other studies that showed under
28:13use of adjuvant radiation therapy
28:15and post breast conserving surgery
28:17in North Carolina and in Georgia
28:19we see the same sort of thing.
28:21A Missouri study showed a delay
28:23in treatment and increased risk
28:25of death and related it all
28:26to differences in treatment.
28:30We link the cancer registry
28:33with all payer claims data.
28:35Did not take long to do the linkage.
28:37It took about a year and
28:38a half to get the data,
28:41getting everybody to agree,
28:44yes, you can have the data.
28:46And just as we were about to get it,
28:48the privacy officer at the state decided,
28:52you know what,
28:53we're only going to give you year of death,
28:57not and year of diagnosis,
28:59not month and year.
29:01And we're saying how exactly are we
29:04going to do survival analysis if
29:06we only have the year and ended up
29:09in another big discussion of trying
29:12to convince the privacy officer
29:14that we could indeed have the both
29:17the month and the year and that
29:20delayed our project by another,
29:22I don't know eight months or so.
29:24And we had to get everybody at every
29:27level involved and eventually they
29:29ended up changing the regulation for
29:31the state because we had one privacy
29:34officer after everybody agreed
29:36after we'd received the funding,
29:38the letter of support everything decide no.
29:40So I'm going to be really cautious today.
29:43So all of these things just to make
29:46it happen and with secondary data
29:48you think it's going to be easier
29:51but it can be quite difficult.
29:53We this is our five year linkage.
29:55We've actually updated it now and
29:57we have it through 2021 incredibly
30:00high quality and with Medicaid
30:03this was 93% overall,
30:05but Medicaid it was 98%.
30:07They were our best data and then we
30:10found that the APCD was reliable
30:12with treatment and insurance status.
30:15When we went through and really tried
30:16to look at the quality of the APCD data,
30:19we were new to this.
30:22If we had used the cancer registry alone,
30:25we know that there are going to be
30:28problems and all of you know as well
30:30that they collect data of individuals
30:32diagnosed with cancer including
30:34patient and tumor level diagnosis
30:36at date at both date and stage.
30:40The outpatient treatment includes
30:42oral agents we know are under
30:44reported in cancer registries.
30:46It's just tough to get that data.
30:48Registries record the first course
30:51of cancer directed treatment,
30:53and Medicaid and rural residence
30:55treatment data appear to be incomplete.
30:59And it's funny.
31:00Our beautiful state of Colorado,
31:02most of the populations kind of in
31:04Denver through what's called the Front Range,
31:06Denver up through Fort Collins.
31:08And then there's the Rocky Mountains
31:10and the rest of the state,
31:12which is pretty far-flung.
31:13And so the state is mostly rural
31:16and frontier.
31:17And when we think about Colorado,
31:19we think Aspen and Vail.
31:20And by the way,
31:22they're rural counties as well,
31:23really different outcomes than
31:25your typical rural county,
31:26as you might imagine.
31:28And those particular places.
31:30And then the rest of the state
31:32being very rural except for Denver,
31:34and we're the only comprehensive
31:36Cancer Center and getting to us
31:38can be quite complicated.
31:39And you have to sort of think
31:41through all of those things.
31:42When you use our particular cancer registry,
31:46we know insurance data are incomplete
31:48and in fact it's overwritten in
31:51the hospitals that record it.
31:53So you get the insurance at the time of
31:56when it's reported, which can change.
31:59You can come in uninsured,
32:00get Medicaid, pick it up or privately
32:03insured and lose your insurance.
32:05And we know that it's more
32:06than two years old,
32:07whereas APC data is getting
32:09pretty real time claims data in
32:13it's able to overcome some
32:15of these limitations because
32:17you get all medical claims,
32:19pharmacy claims, dental claims,
32:21eligibility and provider
32:22files and you can link them.
32:25You get you get a unique identifier.
32:28So I know when someone moves from
32:31private to Medicaid or the other
32:33way around and you'll be able to
32:35tell all payer claims data is
32:38really some claims of some payers.
32:40To be completely honest,
32:42not all payers are in there.
32:45Payers covered under ARISA are not
32:48required to submit claims and that's
32:51about 30% of payers oddly enough in
32:53Colorado most of them voluntarily do so.
32:56So we having somewhat neat
32:58near complete data,
33:00we can look at a cross and in our state
33:03it includes 36 commercial payers.
33:06Our main managed care payer happens
33:08be Kaiser and Medicaid and Medicare.
33:12Our cohort or women ages 21 to 63,
33:16we wanted to get them before they aged into
33:19Medicare and to see this cleanest Co group,
33:22the cleanest sample we could find,
33:25the CR summary stage of local
33:27or regional breast cancer,
33:29enrolled in Medicaid or private
33:30insurance at the time of diagnosis,
33:33had continual coverage within
33:35three months of diagnosis and
33:38continuously enrolled in nine months.
33:40So I intentionally wanted to get those who've
33:42been in Medicaid sometime prior to diagnosis.
33:45We already know there's a problem
33:47with those who come in late.
33:48Let's look at the continuous
33:50coverage people now and compare
33:51them to our gold standard,
33:53hopefully of privately insured
33:55individuals and see what happens.
33:57So able to control for this and
34:00then for those who were supposed
34:02to receive radiation therapy,
34:04they had breast conserving surgery
34:07and for hormonal therapy it was
34:10women who had surgery and also
34:12had estrogen receptor positive or
34:15progesterone receptor positive cancer.
34:21Our methods, what is descriptive statistics
34:23the standard of what you would expect.
34:26We used a follow up time of nine
34:28months following the month of last
34:31surgery as our observation period.
34:33In this data set,
34:3593% of all surgeries regardless of
34:37insurance occurred within three
34:39months of diagnosis and that gave
34:41us a total follow up time of 12
34:43months to look at whether or not
34:46they received these therapies,
34:47estimated logistic regression and reported
34:51marginals for ease of interpretation.
34:54And then we compared what we saw
34:56if we used registry alone,
34:58if we used APCD or if we use them
35:01both what kinds of treatments they
35:03got and did a sensitivity analysis
35:05because one of the arguments is that
35:07those insured by Medicaid takes longer
35:09for them to get their surgeries,
35:11they can't get in complicated lives,
35:13all those things.
35:14So we increased our follow up
35:16time to make sure,
35:17but we still saw no statistically
35:20significant differences.
35:21And then we looked at poverty quartile
35:24and variables for clinician of whether
35:26the clinician was in a rural area,
35:29whether they practice there.
35:30And that ended up being really
35:32an important variable because
35:34if you're in Aspen or Vail,
35:35chances are you're going to figure out
35:37how to get to Denver and get your healthcare.
35:40But if the clinician treating you is
35:42in a rural area means that you are,
35:45you are a person who can't get to Denver
35:47and is your care going to be different.
35:50So that ended up being a really
35:52interesting part of our analysis as well.
35:57So descriptively just starting to look
35:59at our data, we see that there are,
36:01there are big differences now and
36:04the reporting between Medicaid
36:07and private to the registry,
36:09the registry actually doesn't
36:11pick up nearly the amount of
36:13data that you see with the APCD.
36:16The APCD is adding a big chunk
36:19of claims that the registry never
36:22sees on treatment that's coming in.
36:25So people who are Medicaid providers
36:28aren't reporting as much of the registry.
36:31They're either in places that don't
36:33have systems in place or that they
36:35don't have the resources to be
36:37able to get it to the registry.
36:39But there's not the kind of support
36:42that you get in the Denver and our
36:45University Hospital to the registry
36:47so huge under reporting that that
36:50we initially see that could lead
36:52you to a very different conclusion.
36:56And in fact it did.
36:57If we used our cancer registry alone,
37:00we saw that women insured by Medicaid
37:03were four percentage points less
37:05likely to receive radiation therapy
37:07than privately insured women.
37:09When we add APCD data in,
37:13there are no differences.
37:16So an important part of just
37:18trying to take the problem apart.
37:21And now I've got this group of
37:23people who are continuously insured.
37:25I've got a state with some geographical
37:28challenges to say the least,
37:30and I'm not seeing differences
37:33when I'm using claims data.
37:35Hormonal therapy would do the same thing,
37:3810 percentage point difference in Medicaid.
37:41Insured women less likely
37:43to receive hormonal therapy.
37:45But when we bring in our claims data and
37:48can look at the actual pharmacy claims,
37:50there's no difference.
37:52They're still getting the their
37:56same treatment as our privately
37:58insured cohort once they get in.
38:01So now this gives us a different
38:03look and a different view about these
38:05disparities of and when we can get this
38:08data and have a true control group.
38:10And these comparisons even after beating up
38:12on the data with our sensitivity analysis,
38:15we still find the same kind of results.
38:19At the end of the day,
38:20we end up seeing that despite the fact
38:24that there are differences at disease,
38:27at the stage of disease at diagnosis,
38:29we really are seeing under
38:32reportment or reporting of treatment.
38:34And we tried to figure out whether
38:36that was just the provider,
38:37whether it was the location they were in
38:40and in a far-flung part of the state.
38:42But there is under reporting and some
38:45of the when we cared and compared
38:48to the cancer registry,
38:49APCD has some under reporting as well,
38:52but it was so much less and they were
38:55able to pick up these Medicaid claims.
38:58So disparities were only observed
39:00when using the cancer registry alone.
39:03This has serious implications for
39:05if you rely on, if you go out.
39:08And SEAR is no different.
39:09When we did the SEAR Medicaid linkage,
39:11we the agreement between whatever SEAR
39:14had as the insurance was entirely different.
39:17And as some of you may know,
39:19SEAR no longer reports insurance data
39:21because it's so terribly unreliable.
39:24But if those are the kind of data that
39:26you're using to do disparities research,
39:29there's there's both incorrect data
39:31about what the actual insurance
39:33carrier is and the data they have
39:35are greatly under reported.
39:37If it's like what we observed in Colorado,
39:43there are limitations.
39:44Colorado is 1 state and as
39:45I said at the beginning,
39:47there are 50 different Medicaid programs.
39:49There is something unique
39:51about our Medicaid program.
39:52We are a fee for service state,
39:54not a managed care,
39:55which is unusual across the state.
39:57That made us though feel even more
40:00comfortable with our claims data
40:01because it is mostly fee for service.
40:04The sample and omitted women
40:06who did not receive surgery,
40:07although 93% of the women in
40:09our sample received surgery,
40:11so there probably wasn't
40:12a disparity there either.
40:14We didn't look at treatment completion.
40:17And didn't measure the amount of
40:19treatment that would be a next step.
40:21And then as I also mentioned ERISA,
40:23cover plans are not required,
40:25but about half of them do voluntarily
40:28in Colorado for whatever reason.
40:30So here's where we ended up.
40:32Medicaid does a better job than we think.
40:35The disparities are not quite as great.
40:37The evidence does suggest the need
40:40for continuous coverage and I think
40:42this last point is pretty important,
40:44need to support the data infrastructure.
40:46We are providing the data
40:48that policy makers use.
40:49And in some States and I've heard
40:52this state stated in Texas,
40:54the reason they haven't
40:55expanded Medicaid as well.
40:56It's just crappy coverage.
40:57We want to do something else,
40:59but they don't really have a good
41:01alternative or any alternative
41:03to Medicaid and the data don't
41:05really support that conclusion.
41:07It's we don't provide the
41:10continuous coverage.
41:11So next steps really is replicate
41:13somebody else to do the similar kind
41:15of things somewhere else and for us
41:18to look at other sites of cancer.
41:20If we continue to do this.
41:21None the less we have built a body
41:24of evidence that I think supports
41:27the policy form of both Medicaid
41:29expansion and in fact to have
41:31continuous coverage and to increase
41:33our data infrastructure so that
41:35we provide the right evidence
41:37for policy makers to use.
41:41Thank you all.
41:41Thank you for your time,
41:42attention
41:47and I think we're at the stage of let's talk,
42:02thank you so much for this really
42:04important talk, especially the,
42:06the conclusion that being covered by
42:08Medicaid is associated with similar
42:10outcomes as private insurance.
42:12And I'd like to hear you discuss
42:14a little bit more how to inform
42:17policy changes with Medicaid
42:19expansion in some of those states.
42:21Like is this data enough or if you show
42:24that has to occur in other states as well,
42:26how can we get the states that don't
42:29have Medicaid expansion to expand?
42:31Yeah. I mean it's it's interesting I
42:35how there can be an argument at this
42:38point against expansion and not being
42:41having some care and being able to
42:43get into the system is so critically
42:46important and to be able to show this.
42:48And our Lieutenant governor and both
42:51our Governor and Lieutenant Governor
42:53are very much about healthcare
42:55and making it affordable.
42:57And the Lieutenant Governor has
42:59the awkwardly named office of
43:01saving people money in healthcare.
43:03Literally.
43:03And I quote,
43:04I mean it's just like really anyway.
43:07But she has this office and and really
43:10pays attention to this kind of evidence.
43:13And she herself is a four time cancer
43:15survivor that she says all the time
43:17And she visits our Cancer Center,
43:19she is on our Advisory Board,
43:21comes in and she is always talking
43:24about the affordability of healthcare
43:26and access and for us to be able
43:28to show this data,
43:29she was completely on board
43:31and resonating with it.
43:32And they support the APCD,
43:34the civic they organization that
43:37manages it and wants it to be used.
43:41If you're in a state where
43:42that's just not your philosophy,
43:44you know where you don't believe data,
43:46where you don't trust the data,
43:48where you're looking for ways
43:51to reduce the public safety net,
43:54it sometimes feel like there's
43:55just not enough evidence.
43:56But I think we have to keep
43:58trying and that's our job,
44:00to be able to keep putting this out in front.
44:03When we started this part of the project,
44:06it really was that tedious
44:08validation component that we all do.
44:10And then it became the story like,
44:12wait a minute,
44:13we're not seeing any differences
44:15we expected to, but we're not.
44:18And then even when I mentioned this to
44:20true believers at the National Cancer
44:22Institute that runs the SEER registry,
44:24they said,
44:25well,
44:26are you just getting the claims later
44:29or do they eventually show up in the
44:32Medicaid or in the cancer registry?
44:34No, they never showed up.
44:37Even when we expanded our linkage
44:39out to 2021,
44:40the people we saw being diagnosed
44:42in the earlier part of our cohort,
44:44their claims never made it to the registry.
44:48It just doesn't come in.
44:49And providers who are doing care
44:53for large Medicaid populations,
44:55We don't have the data infrastructure
44:58that's being reported up.
44:59And when I showed this to our cancer
45:02registrar in the state, he said,
45:04yeah, that sounds about right.
45:05We're, you know,
45:06wasn't actually a surprising finding to him.
45:09He says, yeah, we're trying to
45:11provide more support to these other
45:12providers that we know that need it.
45:14So the infrastructure is pretty important.
45:19I have a question as a clinician,
45:22slightly different perspective.
45:23When we, when our patients get Medicaid
45:26or free care where we call it here
45:30we're our team is just ecstatic because
45:32now we can actually get reimbursed.
45:34We can do the care as we
45:35would normally have it.
45:38So I think that that that delay
45:41to enrollment certainly resonates,
45:43but I think that once they get into
45:45our system and then we can start to
45:47hook them up with primary care and
45:48all the things that they haven't had.
45:50So that that's my perspective
45:52in terms of that piece of it is
45:55that once we get that coverage,
45:57we're trying to provide the the exact same
46:00care as we do it as our other patients.
46:04Yeah, I agree with you completely
46:06and I think that is the case in
46:09institutions like ours, right.
46:10You know, if you're a private
46:12provider out in the community,
46:15especially way out in the community,
46:16you might be more sensitive to how many
46:19Medicaid patients you put on your panel.
46:21But I think what you described
46:23is very much the case.
46:24And you know,
46:25the key is being able to get them
46:27here and get them into these kind
46:28of centers where they're going
46:30to get really good care.
46:31And they're and we've actually
46:34done studies to show that if you
46:37get to an NCI designated center
46:39or even a COC designated center,
46:43you're going to get the same care.
46:48Yes. So I first was going to follow up
46:51Melinda's comment about policy changes.
46:54So I mean I I think what we're all
46:56probably saying and this kind of
46:59agrees with our clinician perspective
47:01is is once the patient has Medicaid,
47:04their treatment is similar at
47:06least at a place like this.
47:09So what kind of policy changes do can be
47:12done to deal with that very compelling
47:14data you have that the people who
47:18the pre-existing enrollees do well,
47:20the people who get diagnosed at time
47:22who get insurance at Medicaid at
47:25the time of diagnosis do less Well.
47:27What can you do to to to fix that?
47:32You know is it are are states trying
47:34not to are not enrolling people as
47:37as proactively as they can because
47:39obviously that increased short term
47:41costs or is it is this something
47:44that we haven't figured out how
47:46to enroll those patients?
47:47Yeah, it varies a lot by
47:49state like everything else.
47:50So take Massachusetts,
47:51it has a very low uninsured baseline
47:54on insurance rate and in pre ACA
47:56they had a very low baseline on
47:59insurance rate and they were one of
48:02the first states to expand their
48:05Medicaid and offer a way to have
48:07insurance if you don't qualify for
48:09Medicaid to be able to get into it.
48:11And ACA was modeled after it.
48:13So they tended to do a really good job,
48:15but they had a low baseline
48:17on insurance rates,
48:17so it didn't cost them as much to begin with.
48:20If you're in Alabama where it's a state
48:22you don't have a lot of resources and
48:25much of your population is uninsured,
48:28there's not as aggressive approach
48:30to go out and get insurance.
48:33And Alabama is one of the states
48:34that have an expanded Medicaid,
48:36not unsurprising.
48:37So it's it's more than I,
48:40I it goes beyond the political philosophy,
48:43but what's the burden on the state
48:45budget then to go out If you have a
48:48large uninsured population and you're
48:49not a particularly wealthy state to
48:51begin with and this is a state-run program,
48:55those states are not as willing to
48:57go out and be aggressive about it.
48:59So Virginia just expanded not long ago and I,
49:03they are really worried about the
49:05out of the woodwork phenomenon that
49:07if you offer Medicaid now all these
49:10people who now know about the program
49:13are going to seek it and really
49:15increase it beyond what they thought.
49:18I don't know that states have really
49:20seen a huge bump in that way.
49:22Just depends,
49:23A lot of it is going to be to and
49:27politically this is so hard to do,
49:29but it's the nationalize these programs
49:32and standardize them across the board.
49:37Carrie, oh, sorry.
49:40Thank you so much for your.
49:43Thank you so much for your for your visit,
49:46talking for your body of work be assuring
49:49with regards to the the value of data
49:52and the importance of Medicaid question.
49:54I just wanted to ask you to take
49:55a step back as someone who's been
49:57working in this field typically
49:59Medicaid for quite a while now.
50:01It's just a troubling trend nationwide,
50:0450% of all Medicaid beneficiary nationwide
50:07are now covered by a privately insured plan.
50:12One of there's five companies,
50:14five 14100 companies or now basically
50:20managing their 50% of our many benefits.
50:23Yeah. Their revenues of those five
50:25companies range from 30 billion in Molinas,
50:28over 300 billion for United Healthcare.
50:32So just wanted to ask your thoughts
50:35about privatization of Medicaid
50:36and what what's driving it?
50:39Yeah, I mean, so this comes back to
50:42that last slide around Replicate,
50:44right, and try to get those differences
50:47to see what privatization has
50:49actually done in these,
50:51in these companies.
50:53And that's a great question to be
50:56able to do it.
50:57And you know we can start if you get a,
51:01you know we're still one state if
51:03we were able to get APC DS and
51:06registries across several States
51:08and be able to make exactly those
51:11comparisons because we can identify
51:13what insurance company it is and find
51:16all of that information out around.
51:18I can look at whether they have a
51:21high deductible plan or not and be
51:23able to make these kind of comparisons
51:25and to be able to look at what's
51:27happening in the Medicaid population.
51:28Great question and I'm sorry,
51:31I didn't see that you haven't.
51:32No, fine.
51:33Thank you so much for all for all of this.
51:36My question comes sort of to
51:39as we've seen some really,
51:42really impactful advances in,
51:44you know, cancer surgery,
51:47immunotherapy, targeted therapy.
51:49What are sort of the methodologic
51:53challenges to taking the same approach
51:56to something that maybe actually
51:58has a bigger impact on outcomes,
52:00but that is not as simple as did
52:03you get referred for radiation,
52:05but are these things going to be
52:08able to be approached from large
52:11databases or are you going to need
52:14you know more granular work in in
52:16single counties or something to
52:18address Great question I think and
52:20the reason we looked at hormonal
52:22therapy is because it's oral
52:24outpatient therapy and we actually
52:26was looking at immunotherapy too.
52:28But the sites of it,
52:30you know,
52:31we're not a huge state in terms
52:34of population.
52:34And so when you get out into our rural
52:37areas and gets really teeny tiny.
52:39But immunotherapies,
52:41therapies in these oral treatments
52:43are really under reported to
52:46registries for obvious reasons.
52:47And you're going to need to get these
52:50claims datas from other kinds of sources.
52:52So it's.
52:52You're right.
52:53As we make these in advances and
52:55they're doing more and more in
52:58the outpatient setting. Yeah.
52:59The data challenges get much greater.
53:02Yeah. And Tim. Oh, looks like
53:04we have one on Zoom as well.
53:06How do you do it?
53:08I don't even know.
53:09Where's the where's the mouse?
53:11There it is.
53:19That's the only reason why a
53:20patient with the same socioeconomic
53:22status would say they will qualify.
53:24Yes, that with a known cancer diagnosis.
53:27The only reason. The question is,
53:30isn't the key issue that in some
53:32states they known cancer diagnosis
53:34is the only reason why the patient
53:36with the same socioeconomic status
53:39was able to qualify for Medicaid?
53:41So, and I don't know for sure if I'm
53:44interpreting your question correctly,
53:45but first and foremost,
53:47cancer is not a qualifying
53:49condition for Medicaid.
53:50Unless you're diagnosed
53:51through the CDC program,
53:53cancer does not get you on Medicaid.
53:54You still have to spend down if
53:56you're above the income requirements
53:58to be able to get into the Medicaid
54:01program or to have qualified all
54:03along just simply not knowing it.
54:05But for many people,
54:06there is a spend down period that
54:08they have to go through and get
54:10on to the program and then they
54:12get the coverage that they need.
54:14So the the SES, it's the same.
54:19If you meet that threshold within a state,
54:23you could be similar socioeconomic
54:25status but still have to spend down some
54:27assets to be able to bring in to the program.
54:30And I'm not sure if I answered
54:31that question exactly,
54:32but I hope so or if not,
54:34there's a follow up.
54:36Tim, great talk.
54:38It's more of a philosophical,
54:39political question.
54:40But with Medicare, you mentioned the,
54:42the importance of maybe
54:43having a nationalized program.
54:44Medicare, we had a nationalized,
54:46but Medicaid we don't.
54:47Do you think there's any fundamental
54:48differences between the programs
54:49that have prevented that?
54:50Or, you know,
54:51is there a path forward to getting
54:53a national approach to Medicaid?
54:54Yeah, I don't know. Yeah.
54:57With it's if you think of all the
54:59challenges to the ACA that's already been,
55:02I was a moderator for a panel with
55:04a National Cancer Policy Forum
55:06where we brought together for the
55:0810 year anniversary of the ACA.
55:10And we were talking to Donna Shalala,
55:12the people who really were at
55:14the table when they crafted the
55:16ACA and brought it forward.
55:18And I the question I asked them was,
55:20was there something you do differently?
55:23And the answer was,
55:24Yep,
55:24we would not have compromised that
55:27when we ended up because we we
55:30compromised on so many places in the
55:33bill in hopes for bipartisan support.
55:36And when it passed,
55:37it went right down party lines,
55:39not a single bipartisan vote.
55:42So their answer was the reason
55:43the ACA isn't what we wanted it
55:45to be is because we compromised.
55:47If we did it again,
55:49we would not have done that because
55:51they were never going to play ball.
55:52So you're going to have to have a different,
55:56you know, so it's it's a heavy lift.
55:59And I think the evidence that we
56:01provide and the care that we put
56:04in our research is so critical
56:06and that we keep just pushing that
56:09we have really valid findings.
56:11We're being more creative with our data.
56:13We're finding this and putting it out
56:16there in hopes that there's an audience.
56:21Well, thank you so much. I'd like to take.