Skip to Main Content

"Leveraging Real-World Data through Pragmatic Clinical Trials" and "Insurance Coverage Mandates and the Adoption of Digital Breast Tomosynthesis"

 .

"Leveraging Real-World Data through Pragmatic Clinical Trials" and "Insurance Coverage Mandates and the Adoption of Digital Breast Tomosynthesis"

March 16, 2022

Yale Cancer Center Grand Rounds | March 15, 2022

Presentations by: Dr. Joseph Ross and Dr. Susan Busch

ID
7544

Transcript

  • 00:00I'm delighted to introduce our first
  • 00:02speaker today, Doctor Joel Ross.
  • 00:04He's a professor of medicine in general
  • 00:06medicine and professor of public health
  • 00:09and health policy and management.
  • 00:11After receiving his medical degree at
  • 00:13the Albert Einstein College of Medicine,
  • 00:15Doctor Ross came to Yale.
  • 00:17Follow in the Robert Wood Johnson
  • 00:20Clinical Scholars Program in 2004.
  • 00:22He has had a very distinguished career since,
  • 00:26with a focus on examining factors that
  • 00:28affect use or delivery of recommended
  • 00:31hospital and ambulatory care,
  • 00:33as well as clinical outcomes of such care.
  • 00:37Today his topic is leveraging real-world
  • 00:40data through pragmatic clinical trials.
  • 00:43Doctor Ross the floor is yours.
  • 00:46Thank you chairman.
  • 00:47Thank you for inviting me to
  • 00:49speak before the Cancer Center and
  • 00:50part of the grand rounds today.
  • 00:51I'm delighted to share some of the
  • 00:54work that I've been working on
  • 00:56over the past several years and to
  • 00:58identify potential opportunities for
  • 00:59collaboration with investigators
  • 01:01throughout the Cancer Center.
  • 01:02I also could not be happier to
  • 01:04be sharing the stage with Susan
  • 01:06Bush today because, you know,
  • 01:07when I was a clinical scholar,
  • 01:09kind of lost looking for a mentor.
  • 01:11Way back almost 20 years ago now,
  • 01:13Susan was the only person to
  • 01:15open her door to me.
  • 01:16When she got she helped me get my
  • 01:18career started so I couldn't be more
  • 01:20grateful for everything she's done
  • 01:22to help me get started in my career.
  • 01:23So I'm just going to get started
  • 01:25and talk about this work.
  • 01:26Please you know,
  • 01:27jump in with questions through the chat.
  • 01:29I'll try to keep an eye on it just to note,
  • 01:32some of the potential competing
  • 01:33interests that inform the work that
  • 01:35I'm going to be presenting today.
  • 01:36I do get research grant funding
  • 01:38through Yale from the FDA as part
  • 01:41of the Yale Mayo Clinic Center for
  • 01:43Excellence in Regulatory Science
  • 01:44and Innovation.
  • 01:45I'll talk a little bit about that work.
  • 01:47As well as from the medical Devices
  • 01:49Innovation Consortium to run something
  • 01:51called Nest along with some funds
  • 01:52from Johnson and Johnson for clinical
  • 01:54trial data sharing initiatives
  • 01:56at federal government awards,
  • 01:57as well as the Laura and John
  • 02:00Arnold Foundation.
  • 02:01So this is just,
  • 02:02you know,
  • 02:03to get us started you know here
  • 02:04you know we see some pictures of
  • 02:06individuals you know searching
  • 02:08for evidence to,
  • 02:09you know as they have a clinical question,
  • 02:10they're trying to make a decision
  • 02:11about what to do for their patients.
  • 02:12Or to then, you know,
  • 02:13sit down with their patient and
  • 02:15make a suggestion or recommendation
  • 02:17around a drug to use and you know,
  • 02:19typically you know when we think
  • 02:20about this sort of the hierarchy
  • 02:22of evidence and you know what
  • 02:24we want to guide our decisions.
  • 02:25You know,
  • 02:26we look for evidence that you
  • 02:27know is at this level or higher.
  • 02:29You know randomized control trials.
  • 02:31You know to guide our decisions or
  • 02:34perhaps systematic reviews that
  • 02:35are aggregating RCT evidence,
  • 02:37and ideally when it's been meta
  • 02:39analyzed to put it all together.
  • 02:41But there's been a lot of changes
  • 02:43in the way we understand evidence,
  • 02:45in part because of the advancement
  • 02:48and methods to use a large data sources,
  • 02:51but also because of other challenges
  • 02:53that have faced both the FDA and others.
  • 02:56But what you'll have noticed over
  • 02:58the past decade is, you know,
  • 03:00increasingly thinking about.
  • 03:02A new and novel ways to evaluate
  • 03:05medical products and the the.
  • 03:08With failings of the past to
  • 03:10identify a safety issues earlier,
  • 03:12you began to see ways to think
  • 03:14about what's being called a
  • 03:16lifecycle approach to evaluation.
  • 03:18So it's not just about that
  • 03:20first RCT evidence,
  • 03:21it's going to inform use and in part
  • 03:24that was because premarket studies that
  • 03:26inform FDA approval are often limited,
  • 03:28limited in size, limited in scope,
  • 03:30limited in the end points
  • 03:31on which they're focused on.
  • 03:32They're not looking at the kind of
  • 03:33they're not big enough studies to
  • 03:35identify important safety concerns,
  • 03:36and sometimes they're not
  • 03:37even studies that are.
  • 03:38Guaranteed to confirm the
  • 03:40efficacy of a product.
  • 03:42They're they're focusing on surrogate
  • 03:43markers as endpoints in order to project,
  • 03:45benefit, predict,
  • 03:47benefit through these markers,
  • 03:49and then those are supposed to be done
  • 03:52in tandem with postmarket studies.
  • 03:54You know,
  • 03:54trials that are going to happen
  • 03:56after the the approval and but the
  • 03:58problem has been that those trials
  • 04:01frequently are delayed and they're
  • 04:03just not even consistently completed.
  • 04:05This in combination with the fact
  • 04:07that we were never ever going to
  • 04:09be able to address each remaining
  • 04:11uncertainty through clinical trials,
  • 04:12has led to, you know,
  • 04:14opportunities for you know
  • 04:15what you're hearing now.
  • 04:16Kind of real world data.
  • 04:17Real-world data as a the way forward,
  • 04:21and regulatory science and evaluation.
  • 04:23And you know,
  • 04:24this is this quote from a high level
  • 04:27official at the FDA is illustrative.
  • 04:28You know,
  • 04:29using RWE to begin to address
  • 04:32these questions as preferable to
  • 04:34having no evidence whatsoever.
  • 04:36And you know, with the advent of,
  • 04:37you know industry and FDA talking
  • 04:39more about real-world data.
  • 04:40You're starting to see you know
  • 04:42more and more companies popping
  • 04:43up that you know promising.
  • 04:44We're world analytics to
  • 04:46deliver real-world evidence.
  • 04:47And you know, I'll just sort of say,
  • 04:49you know, this is, you know,
  • 04:50buzzword alert, right?
  • 04:51This is a big problem that where the
  • 04:54the sort of the promise is getting
  • 04:56way ahead of what is actually,
  • 04:58you know what we're capable of,
  • 04:59and what we're capable of,
  • 05:01sort of understanding reliably.
  • 05:03Really,
  • 05:03what we're talking about now are the use of.
  • 05:06You know cohort studies case control studies.
  • 05:08You know,
  • 05:09leveraging observational data resources
  • 05:10and and in part this is not only
  • 05:12a recognition of the limitations
  • 05:14of premarket regulatory approval,
  • 05:15but also you know a major advocacy
  • 05:17push that's happening towards
  • 05:18real-world data that has led to,
  • 05:20you know,
  • 05:21new legislation the 21st Century
  • 05:23Cures Act that passed at the tail
  • 05:25end of the Obama administration,
  • 05:27you know,
  • 05:28had very clear goals that to push
  • 05:31towards a real world data use,
  • 05:32including requiring the FDA
  • 05:34to establish a program.
  • 05:36To evaluate real-world evidence which
  • 05:38that was defined in the legislation
  • 05:40as data regarding the usage or the
  • 05:43potential benefits or risks of a
  • 05:45drug or device derived from sources
  • 05:46other than randomized control trials.
  • 05:49Now, this isn't to say that you
  • 05:51know all real world data are bad.
  • 05:53The typical or traditional PWB of today
  • 05:56is work that you know many investigators,
  • 05:59including my group at Yale,
  • 06:01do right.
  • 06:01So it's advanced observation.
  • 06:03ULL research,
  • 06:04including clinical epidemiology,
  • 06:05to inform product development.
  • 06:07You know issues around disease prevalence,
  • 06:09prognosis and treatment adherence.
  • 06:12This type of evidence is generally used
  • 06:15for secondary indication approvals
  • 06:16for rare diseases or for you know,
  • 06:19diseases that are with well understood
  • 06:21pathophysiology and progression,
  • 06:23and it's very limited and it's used for
  • 06:26initial regulatory approval decisions,
  • 06:28mostly because those products are
  • 06:30not used in such widespread way
  • 06:32that you can actually leverage
  • 06:34existing data sources to study there,
  • 06:36that the effectiveness and safety
  • 06:38of the product.
  • 06:39And of course,
  • 06:40most commonly of these types of studies
  • 06:43are used for safety surveillance
  • 06:45or registry registry based medical
  • 06:47device studies and just to bring
  • 06:49your attention to some of the work
  • 06:51that we've done as part of our group.
  • 06:53And I did want to just sort of flag
  • 06:55this because there's individuals
  • 06:56here attending the grand rounds who
  • 06:58may be interested in collaborating.
  • 07:00I lead a couple of efforts that
  • 07:02essentially work closely with FDA to
  • 07:05generate evidence to address kind of
  • 07:07unmet needs at the at the Agency this often.
  • 07:09This is through our Searcy.
  • 07:11We are one of four that are funded
  • 07:13by the FDA to do collaborative
  • 07:15regulatory science research,
  • 07:16but it's also through nest,
  • 07:17which is a a network of health systems
  • 07:20that are working with real world data.
  • 07:22Or, you know,
  • 07:23essentially,
  • 07:23we're working with our health system
  • 07:25data to try to evaluate medical
  • 07:27devices in practice and these types
  • 07:29of studies you know tend to look
  • 07:32like this project where we look,
  • 07:34try to better understand the
  • 07:36safety and efficacy of individuals
  • 07:38who are switching from branded.
  • 07:39We both are rocks into generic
  • 07:42looking at its impact and effect.
  • 07:45Thyroid stimulation hormone levels
  • 07:47and other markers of efficacy.
  • 07:49This project,
  • 07:50where we're where we're aggregating
  • 07:52data across the state of Connecticut,
  • 07:54including hospital data and mortality
  • 07:57data and other vital statistic data,
  • 07:59then even EMS data to try to better
  • 08:02understand opioid use disorder and
  • 08:04overdose including, uh, you know,
  • 08:07throughout the state.
  • 08:09Work like this where we're trying to
  • 08:11understand the comparative effectiveness
  • 08:12of safety of oral anticoagulants in
  • 08:14patients with atrial fibrillation
  • 08:16who have poor kidney function.
  • 08:18These types of patients are often
  • 08:20excluded from clinical trials,
  • 08:21but FDA is often tasked with trying
  • 08:23to understand and give direction
  • 08:24on their safety and benefits for
  • 08:26use of this kind of research,
  • 08:28as well as this.
  • 08:29This registry based study where you
  • 08:32know we looked at different types of cardio.
  • 08:35Cardiac pump devices and looking
  • 08:37at their safety,
  • 08:38especially for patients who are
  • 08:39having acute heart attack and
  • 08:41are in cardiogenic shock,
  • 08:43so lots of individuals are doing
  • 08:44work like this that are leveraging
  • 08:46existing data sources to try to bring
  • 08:49greater insights into the safety
  • 08:51and benefit of various products.
  • 08:54But I think you know,
  • 08:54as the sort of the call for real
  • 08:56world evidence gets louder.
  • 08:57You know one caution to keep in mind
  • 08:59is that observation ULL data sources
  • 09:01should not be expected to answer the
  • 09:03same clinical questions that are
  • 09:04being answered by traditional clinical.
  • 09:06Clinical trials and we have
  • 09:07to think about ways
  • 09:08to make sure that the evidence is being used.
  • 09:10Compliment you know,
  • 09:12to complement the existing RCT evidence.
  • 09:15This is an example of a project that
  • 09:16a student working with me did a couple
  • 09:18of years ago trying to understand
  • 09:20the feasibility of using real-world
  • 09:22data to replicate clinical trial
  • 09:23evidence and what she did is she
  • 09:26identified among all the clinical
  • 09:28trials that had been published in
  • 09:30high impact medical journals in 2017.
  • 09:33She determined what proportion
  • 09:35had and in clinical intervention.
  • 09:37The clinical indication of the of the
  • 09:40patients who were studied enrollment
  • 09:41criteria as well as a primary
  • 09:43endpoint that could be successfully
  • 09:45in routinely ascertained from
  • 09:47either electronic health records.
  • 09:49Structured electronic health records,
  • 09:50data or claims data,
  • 09:52and what we found is that only 15%
  • 09:54of these trials could feasibly have
  • 09:57been replicated using this kind
  • 10:00of real world data resource.
  • 10:02When the 21st Century Cures Act passed,
  • 10:05the FDA was actually pretty quick to say,
  • 10:07listen,
  • 10:07real-world data should be defined by
  • 10:10the context in which the evidence
  • 10:12is gathered in clinical care or
  • 10:14home and community settings,
  • 10:16as opposed to necessarily in
  • 10:18research or academic environments,
  • 10:20and the distinction is not based
  • 10:22necessarily on the presence or
  • 10:24absence of a planned intervention or
  • 10:26use of randomization randomization.
  • 10:28Essentially, they're saying,
  • 10:29you know,
  • 10:29continue to seek out opportunities
  • 10:31to conduct.
  • 10:32Randomized evaluations using
  • 10:33pragmatic trials that better leverage
  • 10:36kind of the existing data resource
  • 10:38infrastructure to make them perhaps
  • 10:40cheaper or easier to conduct.
  • 10:42But it's not just about substituting
  • 10:45observation,
  • 10:45ULL data analysis for randomized
  • 10:48control trials,
  • 10:49and I'm always reminded of this quote.
  • 10:51You know,
  • 10:52if you want more evidence based practice,
  • 10:53you need more practice based evidence.
  • 10:56So in in the next 10 minutes I'm
  • 10:58going to talk a little bit about
  • 11:00some of the work that we've been
  • 11:01doing to try to better leverage.
  • 11:03Kind of pragmatic clinical trials in
  • 11:06the hopes of showing you what I think is,
  • 11:09I think,
  • 11:09the future of real world data investigations.
  • 11:13It's not just about leveraging
  • 11:15observational data resources.
  • 11:16This this is a slide from Cuba.
  • 11:20Take a data warehouse company that
  • 11:22aggregates information across of you,
  • 11:24know multiple multiple sources and
  • 11:25you know they talk about kind of
  • 11:28all the real world data that are out
  • 11:30there for for an individual from pharmacy,
  • 11:32data,
  • 11:33lab and biomarker data to mortality data,
  • 11:37hospital data claims data survey
  • 11:39data disease registry data.
  • 11:41All these things could ideally
  • 11:42be linked together,
  • 11:43including even potentially social
  • 11:45media data or wearables data or
  • 11:47or even you know something like
  • 11:49credit card data.
  • 11:50And this kind of is like the optimal
  • 11:52environment when you talk to people
  • 11:54like the future of clinical trials,
  • 11:56it's going to pull all this
  • 11:57information together.
  • 11:57Putting the patient at the center
  • 11:59and mostly people talk about that as
  • 12:02being idealistic and not really achievable.
  • 12:04But we've been working with a group
  • 12:07called Hugo that actually does just this.
  • 12:10It aggregates multiple data platforms
  • 12:12into a patient centered medical
  • 12:14record that the patient can then share
  • 12:16out with the research team as part of a,
  • 12:19you know, our research project.
  • 12:20And so we this is the first study
  • 12:23we did at leveraging this platform.
  • 12:26It was done as part of our city.
  • 12:28Our FT had funded center where
  • 12:30we aggregated data for just 60
  • 12:33patients who were getting care
  • 12:35at Yale and at the Mayo Clinic.
  • 12:37We recruited 15 patients at each
  • 12:40site who are undergoing bariatric
  • 12:42surgery or A-fib ablation procedures.
  • 12:44A 59 patients under actually
  • 12:46underwent the procedure and completed
  • 12:48our eight week follow up.
  • 12:50And what's?
  • 12:51The beauty of this platform for research
  • 12:53purposes is you sit down with a patient.
  • 12:55You enroll them in the platform
  • 12:57you link their electronic health
  • 12:59record data from any health system
  • 13:02from which they're gaining care
  • 13:04or as well as their pharmacy data
  • 13:06and and and other information.
  • 13:08And that takes time.
  • 13:09It took a little over an hour
  • 13:11for all of our patients,
  • 13:13but once you do that,
  • 13:14everything that happens next over
  • 13:16the 88 week follow up for the
  • 13:19patients is all passive patient
  • 13:21their patients data aggregates.
  • 13:23Automatically into the the the system
  • 13:26being shared with the research
  • 13:27team for research purposes and the
  • 13:29patient never has to come back,
  • 13:30and so you know,
  • 13:31this shows you that we were able to do this.
  • 13:34You know,
  • 13:34with 60 patients you know we've had a
  • 13:37nice sort of broad spectrum of age ranges.
  • 13:39You know,
  • 13:39including a number of patients
  • 13:40over the age of 65 who were
  • 13:42able to do this successfully.
  • 13:43And here are the data we aggregated
  • 13:45and I'll start at the bottom left.
  • 13:47The electronic health record data.
  • 13:49So everyone was getting care
  • 13:51at either the Yale at.
  • 13:53Ill or the Mayo Clinic for their
  • 13:55specialty care for this procedure,
  • 13:56but also and so everyone you know
  • 13:58their care is managed through Epic
  • 13:59and they have access to their
  • 14:01my chart and they connect their
  • 14:02my chart to their Hugo account,
  • 14:04but also individuals who have
  • 14:06primary care elsewhere were able
  • 14:08to link their my charts either
  • 14:09through Epic or Cerner based systems
  • 14:11from any health system.
  • 14:13So if we were taking care of a
  • 14:15patient who was getting there,
  • 14:16a FIB ablation here at Yale,
  • 14:17but they're there, their primary care,
  • 14:19perhaps was at Hartford Hospital.
  • 14:21For whatever reason they could
  • 14:22link that system too.
  • 14:24Also,
  • 14:24we linked their pharmacy data,
  • 14:26so that's not the upper right and so
  • 14:28this was individuals were getting care.
  • 14:30Their pharmacies met their
  • 14:31medications through CVS or Walgreens.
  • 14:34They also use a mark.
  • 14:35My chart based system that allows this.
  • 14:37They're essentially their health
  • 14:39record to get linked right into Hugo.
  • 14:42We also then used Hugo to send out surveys.
  • 14:45Patient reported outcome measures.
  • 14:47Both short questions post procedure
  • 14:50along with longer questions at 148
  • 14:52weeks and patients get a link.
  • 14:54Right to their phone they they.
  • 14:56They signify their preference.
  • 14:57If they want a text message
  • 14:58or email, they click the link and they
  • 15:00fill it all out right on their phone
  • 15:02and and and it's all kind of easy peasy.
  • 15:05They don't have to come back to go through,
  • 15:07you know a structured questionnaire with
  • 15:08a nurse or any other study coordinator.
  • 15:11They can just do it on their own,
  • 15:12fill it out and that allows
  • 15:14you to ask more questions.
  • 15:16And then we also gave every patient
  • 15:18some two different digital devices.
  • 15:20Everyone got a Fitbit in order to track
  • 15:23activity and patients who underwent
  • 15:24bariatric surgery got a Withings scale.
  • 15:26Digital scale and people.
  • 15:28Patients who underwent the 8th
  • 15:30ablation procedure got us a two finger,
  • 15:32a single lead EKG that you
  • 15:35measured through Kardia mobile.
  • 15:36And this is just some quick results
  • 15:38to show you kind of what we could do.
  • 15:40Again, this was really just figuring out
  • 15:42the feasibility of doing work like this,
  • 15:45but we were able to link health records for
  • 15:47100% of patients who underwent procedures.
  • 15:49A 55% of patients also had a primary
  • 15:51care that was based at Yale or Mayo,
  • 15:53so all of their electronic health records
  • 15:56get pulled in for purposes of the study.
  • 15:5810 patients, LinkedIn,
  • 16:00additional 13 portals and then we had
  • 16:0340% of patients who are getting their
  • 16:06prescriptions through CVS or Walgreens.
  • 16:08Now, Walmart also has a my chart like
  • 16:11function that allows you to pull in
  • 16:14information like medication names,
  • 16:16dosages,
  • 16:17start and end dates along with refills,
  • 16:20and again,
  • 16:20all these data were passively
  • 16:22aggregated after our initial enrollment,
  • 16:24allowing for Neil near real time,
  • 16:26streaming data aggregation and this
  • 16:28just kind of shows you kind of how it
  • 16:30worked at the time when we did the study,
  • 16:32people had to actually sync their Fitbits.
  • 16:36Now that happens automatically,
  • 16:37but this shows you of course.
  • 16:39Of things tail off over time,
  • 16:40but even over the eight weeks we had,
  • 16:42well more than half of patients syncing
  • 16:45their Fitbits their their cardio mobile
  • 16:48devices and their withing scale which
  • 16:50allows you to kind of project you know.
  • 16:53Scraf,
  • 16:54the sort of trajectories of recovery.
  • 16:56So on the top you can see kind of
  • 16:58average steps per day for patients
  • 17:00who underwent bariatric surgery,
  • 17:01you know,
  • 17:02kind of visually demonstrating the
  • 17:04how patients recovered over time.
  • 17:06The bottom half on the left is the the
  • 17:08steps per day for patient patients who
  • 17:11underwent a fibrillation on the right.
  • 17:13Is that the cumulative weight change for
  • 17:15patients undergoing bariatric surgery
  • 17:17on the lower right is the patient.
  • 17:19The average heart rate and again,
  • 17:21this is more just to determine,
  • 17:22you know, the accuracy.
  • 17:24That the integrity of the data
  • 17:26that was being aggregated here.
  • 17:28Our response rate to the patient reported
  • 17:31outcome measures consistently above 80%
  • 17:33for all the patients for all the surveys,
  • 17:36and it allows you also to to to determine
  • 17:39how patients are doing so you know,
  • 17:41we're over time graphing estimates of pain,
  • 17:44appetite and palpitations
  • 17:46in the two patient groups,
  • 17:49but this is really just more
  • 17:52for illustrative purposes.
  • 17:53And this has led to a lot of future
  • 17:55work that I'm really proud of,
  • 17:56and I'm really excited.
  • 17:57It's all kind of coming soon,
  • 17:59but I did want a sort of flag
  • 18:01for people in case it prompts
  • 18:03potential collaborations,
  • 18:04but this is the biggest of the
  • 18:06studies that we're working on now.
  • 18:08Also funded through the Searcy,
  • 18:09it's a where aggregating sensually
  • 18:13a large cohort study of more than
  • 18:161500 patients who are receiving
  • 18:18a new opioid prescription for
  • 18:20acute pain recruiting from sites
  • 18:21across the United States and Yale
  • 18:23at the University of Alabama.
  • 18:24Birmingham,
  • 18:25including from their network of
  • 18:27dental practices that run up the
  • 18:29Appalachian Mountains from the
  • 18:31Mayo Clinic from Monument Health,
  • 18:32which is basically South Dakota
  • 18:34and Cedar Sinai in Los Angeles.
  • 18:36Patients are being recruited for
  • 18:38in the urgent care settings,
  • 18:40emergency departments,
  • 18:41dental care and patients post
  • 18:43cesarean section.
  • 18:44We started recruitment in about
  • 18:46in September 2020.
  • 18:47We now have more than 1000
  • 18:49patients recruited.
  • 18:49Even with all the challenges from COVID.
  • 18:52Our primary endpoint is the
  • 18:53number of days using.
  • 18:54Opioids and we're following
  • 18:56up patients over six months,
  • 18:57including additional measures
  • 18:58for patient or outcome measures,
  • 19:00from pain and anxiety.
  • 19:02Other measures of health care
  • 19:04utilization activity measured
  • 19:05using Fitbits and opioid disposal,
  • 19:07and just to give you a sense of
  • 19:09the kind of data that this allows
  • 19:11us to aggregate on patients.
  • 19:12This is mean daily pain,
  • 19:14reportings among those reporting
  • 19:16they are in pain,
  • 19:17and you can just see how
  • 19:20pain essentially persists.
  • 19:21This is over 180 days.
  • 19:23The average pain dots are in blue.
  • 19:25Worst pain or in red?
  • 19:27Here's the median days elapsed
  • 19:28to 1st report of no pain among
  • 19:30patients with pain fully resolved
  • 19:32and you can see the difference in
  • 19:35pain experienced by patients in
  • 19:37different settings with patients
  • 19:38who are recruited either from the
  • 19:41inpatient setting or a primary
  • 19:43care having longer median days
  • 19:44until the first report of no pain.
  • 19:46Whereas patients for the dentist
  • 19:49heading having slightly shorter
  • 19:51durations and then this shows
  • 19:52you the mean daily pain ratings
  • 19:54among those taking.
  • 19:55A treatment for pain and
  • 19:56this could be any treatment.
  • 19:57It could be tylanol it could be an opioid,
  • 19:59it could be anything,
  • 20:00but you can see here the blue dots
  • 20:02are patients who are not using
  • 20:03an opioid for treatment and the
  • 20:05yellow dots are patients who are
  • 20:07using an opioid for treatment and
  • 20:08you can see how the on average the
  • 20:11patients who are taking an opioid
  • 20:13are having higher rates of pain.
  • 20:15All of this is being done in
  • 20:18collaboration with the FDA as
  • 20:19part of their efforts to better
  • 20:21address and understand the risks
  • 20:24associated with opioid use.
  • 20:25Couple of other things,
  • 20:27just to mention briefly one is these
  • 20:29are projects that are funded by Nest.
  • 20:31This is what we call the sleep I study.
  • 20:33It's a prospective RCT of 100
  • 20:36patients with depression receiving
  • 20:37outpatient treatment for insomnia,
  • 20:39comparing usual care of a prescription
  • 20:41digital therapeutic device that's essentially
  • 20:43cognitive behavioral therapy for insomnia,
  • 20:46following patient treating
  • 20:47patients over 9 weeks.
  • 20:49With the primary endpoint of insomnia
  • 20:51severity index and we're following them up
  • 20:52over a year and again just to emphasize.
  • 20:55All of this is done using the Hugo platform,
  • 20:57so we enroll patients at baseline.
  • 20:59They're randomized to one treatment or
  • 21:01another they undergo, you know, they they.
  • 21:04They undergo the treatment associated
  • 21:06with that arm, and they get, you know,
  • 21:07serving questions out, you know,
  • 21:09through their phone or via email,
  • 21:11and all of their data that the health care,
  • 21:13utilization data,
  • 21:14and other information
  • 21:15otherwise passively aggregates.
  • 21:17It's you know,
  • 21:18a pragmatic RCT that's leveraging.
  • 21:19Real world data for all of our endpoints,
  • 21:22we're doing another study that
  • 21:23we call the Heart Watch study,
  • 21:24which is essentially an RCT of the
  • 21:26Apple Watch where we're perspective
  • 21:29prospectively enrolling 150 patients
  • 21:31undergoing cardioversion for AFIB.
  • 21:33They either get an Apple Watch or they
  • 21:36get a Withings watch without any activity.
  • 21:38That's just an activity tracker without an
  • 21:41EKG and abnormal rhythm notification feature.
  • 21:45We're enrolling patients at
  • 21:46Yale Duke in the Mayo Clinic.
  • 21:48We have about 40 patients enrolled thus far.
  • 21:50Our primary endpoint is the the
  • 21:52effect Global Score questionnaire is
  • 21:54essentially at A-fib quality of life prom,
  • 21:56and again,
  • 21:57we're following up patients over a year,
  • 21:59including additional prompts for anxiety.
  • 22:01Other measures of health care utilization,
  • 22:02as well as cagey accuracy.
  • 22:05And then last,
  • 22:06I just want to note this one this
  • 22:08project we're doing in collaboration
  • 22:10with numerous investigators
  • 22:11associated with copper,
  • 22:12the cancer outcomes public
  • 22:14policy and effectiveness Research
  • 22:16Center led by Carrie Gross,
  • 22:18Sarah McLachlan and Scott Huntington.
  • 22:20Where we're quantifying a physical
  • 22:22function in cancer patients undergoing
  • 22:24chemotherapy using a clinician,
  • 22:26reported patient reported and
  • 22:28wearable device data sources.
  • 22:29This is done being done through our Searcy.
  • 22:32The FDA funded center.
  • 22:33We're doing it directly with collaborators
  • 22:35at the oncology Center of Excellence,
  • 22:38a prospective study of 200 cancer patients
  • 22:41undergoing frontline cytotoxic therapy.
  • 22:43Rolling patients at Yale and Mayo Clinic,
  • 22:46100 solid tumor patients.
  • 22:47Breast cancer patients stage one
  • 22:49through three, as well as a hunt.
  • 22:50100 high grade B cell lymphoma
  • 22:53patients and our primary endpoint is
  • 22:55physical function over nine months
  • 22:57that's being measured weekly for
  • 22:58two months and then monthly again,
  • 23:01all leveraging the Hugo platform
  • 23:03for measurements with patient
  • 23:05reported outcome measures.
  • 23:07Clinician reported outcome measures
  • 23:09to the E COG performance measurement.
  • 23:11The six minute walk test at baseline
  • 23:13and at the at the end of two
  • 23:15months and then again later on,
  • 23:17as well as activity measured
  • 23:19as every patient,
  • 23:20has a daily Fitbit to measure daily
  • 23:23steps and again part of the purpose
  • 23:25of this is to work with the FDA to to
  • 23:28better understand a physical function
  • 23:30as a surrogate measure of recovery.
  • 23:32Compare these data sources identifying
  • 23:34change thresholds and inform
  • 23:36the way the FDA thinks about.
  • 23:38Of these measures,
  • 23:39as part of clinical trials,
  • 23:40so I will stop there and I hope that if
  • 23:44anyone has questions you can follow up.
  • 23:46But thanks for the time, show me.
  • 23:50I'll stop sharing.
  • 24:02Thank you very much Doctor Ross for
  • 24:05this very informative presentation.
  • 24:07Renee, I was wondering,
  • 24:08do we ask people to raise hands?
  • 24:16I'm not sure how this is really handled.
  • 24:18Oh, post your questions in the chat.
  • 24:25I do have a question as we're
  • 24:28waiting for others to pitch in.
  • 24:30I wonder when you submit
  • 24:33work for publication.
  • 24:34Is it subject to more scrutiny
  • 24:36because it's not the traditional
  • 24:38trial that people are used to?
  • 24:43Yes, there's a lot of explaining going
  • 24:45on when we you know when we're putting
  • 24:48these papers together and and even just
  • 24:50proposing them for funding right now,
  • 24:52as people kind of question like,
  • 24:54well, how is this done?
  • 24:55I don't get it. You know,
  • 24:56how are you pulling in these data sources?
  • 24:58But when you talk to people who are
  • 25:01clinical trialists and explain the
  • 25:03difference in the approach and the
  • 25:05efficiency that comes with it and the
  • 25:08the you know the kind of trade offs that
  • 25:09are always happening in any clinical trial,
  • 25:11but the you know how much more information.
  • 25:13You can aggregate passively and not
  • 25:15requiring patients to come back,
  • 25:17minimizing the burden on patients
  • 25:20in terms of participation.
  • 25:21People see. Ah, I get it now.
  • 25:23There's a there's there's,
  • 25:24there's great promise to this,
  • 25:25and it's not to say that that
  • 25:27we've worked everything out,
  • 25:28but I feel like we're kind of pilot
  • 25:30testing new ways to do trials like this,
  • 25:34which I hope are going to,
  • 25:36you know,
  • 25:36be useful and informative and and
  • 25:37and set the stage for the future
  • 25:39so it doesn't need to be kind
  • 25:40of an all or nothing either.
  • 25:42Do a kind of a traditional clinical trial.
  • 25:44Bringing patient back every couple of
  • 25:45weeks for kind of standardized assessment,
  • 25:47or we're doing observation,
  • 25:48ULL data source and data analysis.
  • 25:51There's there's kind of a middle Rd.
  • 25:54I do see there was one question
  • 25:56from Doctor Boffa on how to handle
  • 25:58contradicted data from different sources,
  • 25:59and that's an interesting challenge,
  • 26:02right in the sense of you know,
  • 26:04how do you if you see,
  • 26:06you know essentially prescription data
  • 26:09in the electronic health record at Yale,
  • 26:12but not in the pharmacy data
  • 26:14and how to understand that.
  • 26:15And some of it is about understanding
  • 26:17the various functions that
  • 26:18are used for the data sources.
  • 26:20Right?
  • 26:20Prescription is ordered by a
  • 26:22physician at Yale and it's filled
  • 26:24at a pharmacy so that it's at.
  • 26:26It actually gives you a sense of you know,
  • 26:28adherence,
  • 26:28like our patients going and filling
  • 26:30their their their their prescriptions.
  • 26:32But other times you know if there's you know,
  • 26:35particularly for the the physical
  • 26:36function we're going to have to
  • 26:38decide exactly what does it mean.
  • 26:40If if different you know,
  • 26:42patient reported outcome measures
  • 26:43or clinician reported outcome
  • 26:45measures do not align.
  • 26:52I see Kerry Gross asked a question
  • 26:55around thinking about ways to
  • 26:57adapt the EHR and its interface
  • 26:59in order to be more proactive
  • 27:01in terms of making information
  • 27:03like this more readily available.
  • 27:05And I, I couldn't agree more.
  • 27:06Some of the challenges and part of the
  • 27:09reason why we're using survey questions
  • 27:12out to patients is because it's not,
  • 27:14you know, you know,
  • 27:15uniformly collected as part of
  • 27:17the HR and then extracted and
  • 27:19available to investigators who are
  • 27:20leveraging health system data for.
  • 27:22For research or for you know,
  • 27:24to inform clinical practice.
  • 27:25The more structured data we think
  • 27:27about embedding within our reach are
  • 27:29the better the data are going to be,
  • 27:31the more it's going to allow us to
  • 27:34use kind of actually more typical
  • 27:36observational data resources for research.
  • 27:39One of the things when and when I
  • 27:41presented that project done by the
  • 27:43medical student who identify that only
  • 27:4515% of clinical trials could actually
  • 27:47be routinely or fees abli done today
  • 27:49using routinely ascertainable information.
  • 27:51Part of it is because.
  • 27:53Like patient reported outcome measures
  • 27:55are not routinely included as part
  • 27:57of structured data elements,
  • 27:58so there's a real opportunity there.
  • 28:10And then I'll the last question
  • 28:12I see is about addressing self
  • 28:14selection bias in our data.
  • 28:17I think what Doctor Hooley is referring
  • 28:19to is the participation bias that
  • 28:22individuals are going to be more
  • 28:24likely to participate in the study.
  • 28:26And that raises an issue of bias.
  • 28:29I don't think that the selection into
  • 28:31our studies is different any different
  • 28:34than the selection of any individual
  • 28:36individual into a clinical trial,
  • 28:38but hopefully ideally by lowering the
  • 28:41barriers to participation and and making
  • 28:44it easier on patients to participate by.
  • 28:47By diminishing that burden of kind
  • 28:49of haven't come in our trials.
  • 28:51Using this this type of approach may be
  • 28:54more representative of clinical practice,
  • 28:57although that's that remains to be seen,
  • 28:59and it's an important issue
  • 29:01for us to address so.
  • 29:02I'll stop there so that Susan
  • 29:04Bush has plenty of time to
  • 29:05go through her presentation.
  • 29:08Thank you Joe.
  • 29:11It is my pleasure to introduce our
  • 29:13next speaker, doctor Susan Bush,
  • 29:15who is Professor Public House in House
  • 29:18Policy and professor in the Institution
  • 29:21for social and Policy Studies.
  • 29:23She received a master degree in House
  • 29:25policy in a PhD in House Economics.
  • 29:27Those from Harvard University.
  • 29:29A number of us have been lobbying
  • 29:32for her to join the Cancer Center
  • 29:34and very happy when she did recently.
  • 29:37Doctor Bush's research examines the effects
  • 29:40of policies and regulations on health care,
  • 29:42cost and quality,
  • 29:44and she's a renowned and highly
  • 29:47respected expert in this field today.
  • 29:50Her topic is insurance coverage,
  • 29:52mandates and the adoption of
  • 29:54digital breast Tomo synthesis.
  • 29:58Doctor Bush for as yours.
  • 30:00OK, thank you. Thank you so
  • 30:02much Johnny for inviting me.
  • 30:04I just wanna make sure.
  • 30:05Can you see my slide show it's working?
  • 30:12Some show me you can see my slideshow.
  • 30:14Yes, OK perfect. So first,
  • 30:17for those of you who don't know me,
  • 30:19I'm a health service researcher and
  • 30:21health economist, and I teach at
  • 30:23the Yale School of Public Health.
  • 30:24I teach advanced health economics here,
  • 30:26and most of my work is really around
  • 30:29mental health and substance use disorder
  • 30:31with a focus on access to care and how we
  • 30:34can optimize benefit design to increase
  • 30:36the value of the healthcare system.
  • 30:38So I sort of took my knowledge about those
  • 30:41issues and and now I'm applying it to cancer.
  • 30:44So generally I'm interested when you think
  • 30:46about is as we change payment mechanisms.
  • 30:48What are the impacts on access to care,
  • 30:50cost of care and value?
  • 30:52And it's always really tough to get at that.
  • 30:53You know idea of value,
  • 30:55but I I really do strive in
  • 30:57my work to do that.
  • 30:59So if anybody has any problem
  • 31:01projects related to that,
  • 31:02I would love to meet with them.
  • 31:04I also have several projects related
  • 31:06to tobacco control that people that
  • 31:08might be of interest to people.
  • 31:10I'm not going to go into
  • 31:12detail here about those,
  • 31:12but if you're interested I would love to
  • 31:15meet with you and talk about that about that.
  • 31:18So over the past several years wanna
  • 31:21say it's really been a delight to get
  • 31:24to know the faculty at the Cancer Center
  • 31:28both at the medical school and also
  • 31:30here at the School of Public Health?
  • 31:32So in particular,
  • 31:33I want to mention Carrie Gross,
  • 31:35who invited me to work with his
  • 31:38team a couple of years ago and has
  • 31:41really taught me a lot about both
  • 31:44breast cancer screening and about
  • 31:46how to use health care claims.
  • 31:48Related to some of the issues that
  • 31:50we're going to talk about today and
  • 31:52also I want to give a big shout out.
  • 31:55I hope she's on the call to Alana
  • 31:57Richmond and this is very specifically
  • 31:59related to the work of presenting today.
  • 32:01Elan is an internal medicine and she
  • 32:03is the first author on this paper,
  • 32:06and I can't emphasize enough how
  • 32:08much I've learned from having
  • 32:09the opportunity to work with her
  • 32:11over the past couple of years.
  • 32:13So the paper that I'm going to
  • 32:15talk about today is the latest
  • 32:17in a series of papers related to
  • 32:18breast cancer screening related to
  • 32:20issues around patient preferences,
  • 32:22diffusion of new technologies and cost.
  • 32:24And you know,
  • 32:25this paper is not really focused on value,
  • 32:27but also a lot of our papers
  • 32:29are focused on that.
  • 32:31So this paper has not yet it's been accepted
  • 32:33for publication at that not out yet,
  • 32:35but we're thinking it's going to be out
  • 32:36even in just the next couple of days,
  • 32:38so.
  • 32:42OK, so these are some our collaborators,
  • 32:45my collaborators, on this paper.
  • 32:46Alana, as I mentioned,
  • 32:47Jessica Long Kelly Kenco,
  • 32:49who is at NYU. She's a primary
  • 32:52care physician at NYU and Xiaoju,
  • 32:54who is also here at Yale,
  • 32:56and of course Kerry.
  • 33:00So you know, over the past decade,
  • 33:03cancer screening has undergone substantial
  • 33:05technological shift in the US in
  • 33:08which digital breast tone was insist.
  • 33:10DBT has supplanted standard 2D2 dimensional
  • 33:13Mogra fi alone as the standard of care.
  • 33:17Advantages of DBT are that
  • 33:19DBT may reduce recall.
  • 33:21That is that fewer women are called back
  • 33:24for additional testing after screening,
  • 33:26and also that it may improve sensitivity
  • 33:29that we may identify more breast cancers
  • 33:32using DBT compared to 2D mammography.
  • 33:35Yet DBT is still not rated A
  • 33:37or B by the US United Service.
  • 33:40United States Preventive Services Task Force.
  • 33:46This map is from an earlier paper,
  • 33:50so just to get a sense of the
  • 33:52variation in DBT adoption, this paper
  • 33:55looks at hospital referral regions,
  • 33:58so the different geographic regions you can
  • 34:00see here are hospital referral regions,
  • 34:02and we look at three years of data
  • 34:04from 2015 to 2017 and over this time
  • 34:08period over the US in the US over the
  • 34:11whole USDBT increase from 13 to 43%.
  • 34:15Of screenings, so this looks very
  • 34:19specifically at trajectories,
  • 34:21and we know by the end of 2017
  • 34:23the lowest use HRR's hospital for
  • 34:26regions are about only about 4%
  • 34:29of screenings where DBT, well,
  • 34:31the highest where it was at 68% of screening.
  • 34:34So there really is significant variation.
  • 34:46So related to insurance coverage
  • 34:48really today we're talking about
  • 34:50private insurance coverage.
  • 34:52And private insurers are
  • 34:54not required to cover DBT,
  • 34:56and that's because it doesn't have the
  • 34:58A or B recommendation by the USPSTF.
  • 35:01So absent a federal mandate,
  • 35:02many private insurers didn't immediately
  • 35:04cover DBT characterizing it as elective,
  • 35:07or citing that there might not be long
  • 35:10term data and states got involved.
  • 35:12To date, 17 states.
  • 35:14I think it's actually maybe 19 now.
  • 35:17It's 17 states where the paper was written,
  • 35:18have enacted laws that require
  • 35:20private health insurance cover DBT.
  • 35:22Without any cost sharing,
  • 35:24so of course self insured plans are not
  • 35:27covered due to the ERISA exemption,
  • 35:29but generally privately insured individuals
  • 35:31and women in these states do not have to.
  • 35:35Pay any out of pocket payments
  • 35:38when they receive DVT screening.
  • 35:40So this figure just gives you a sense of
  • 35:44the variation in timing of these laws,
  • 35:48so we're going to study laws that
  • 35:50occurred from 2016 to 2019, and you can.
  • 35:53You can see it really is like a
  • 35:55staggered implementation that's really
  • 35:57important for identification strategy.
  • 35:59Connecticut was the 3rd state
  • 36:01to adopt in 2017.
  • 36:12So insurance benefit mandates such as
  • 36:14these have been widely used in other
  • 36:17contexts as a policy tool to protect
  • 36:19consumers against high out of pocket cost,
  • 36:22so it reduces their financial
  • 36:24burden and also to facilitate
  • 36:26access to important health services.
  • 36:28However, you know these types of
  • 36:31benefit mandates have have been
  • 36:33criticized by some because they
  • 36:36may have some complex effects.
  • 36:39Potentially, if you mandate may
  • 36:41contribute to higher insurance
  • 36:44premiums and thereby this may increase
  • 36:46uninsurance rates as if insurance
  • 36:48premiums get prohibitively expensive.
  • 36:50There's been some criticism that
  • 36:52they may reduce plan design,
  • 36:54plan benefit, design, flexibility.
  • 36:56And also that increasing the price of
  • 36:59the specific of a specific mandated
  • 37:01service may reduce negotiating power
  • 37:03and this is going to be particularly
  • 37:06problematic for a service or a drug,
  • 37:08potentially where they have the
  • 37:10supplier has some monopoly power.
  • 37:15So our goal in this paper was
  • 37:17to evaluate the relationship
  • 37:19between DBT coverage laws.
  • 37:22The 17 laws that I noted in
  • 37:24the last slide and DBT use
  • 37:27DBT out of pocket payments,
  • 37:28and also DBT price.
  • 37:34So to study this, we use data from
  • 37:37Blue Cross Blue Shield access data set,
  • 37:41which is a a deidentified database
  • 37:43of health insurance claims.
  • 37:45There are claims from all 50 states,
  • 37:47so the geographic diversity of this sample,
  • 37:50along with the fact that has a
  • 37:51longitudinal structure so you
  • 37:52can follow patients over time.
  • 37:53It makes it really well suited to
  • 37:56evaluate policies that vary by state.
  • 37:59Within this data set we identified screening.
  • 38:01Mammography is performed among women
  • 38:03ages 40 to 64 between January 2015 and
  • 38:08July 1st up through June 30th, 2019,
  • 38:10and we have a a standard validated
  • 38:13algorithm that we've been using
  • 38:14to identify a screen mammography.
  • 38:16I won't get into details on that
  • 38:18in this talk,
  • 38:19so we did exclude women 65 and over,
  • 38:21and the reason we did that is because
  • 38:23Medicare is not really represented in
  • 38:25these data or Medicare Advantage as well,
  • 38:28and we felt that.
  • 38:29Older women that were then included in
  • 38:31the BCBS data might be highly selected.
  • 38:35So we use the patient level data
  • 38:36to describe the characteristics
  • 38:37of the women and mammograms,
  • 38:39including the study.
  • 38:40But when we do our additional analysis,
  • 38:43our event study design,
  • 38:44we perform it at the state level.
  • 38:46That is, we collapse cells to the state.
  • 38:49We aggregated data to the state
  • 38:51and six month period and use use
  • 38:53the data in that way.
  • 39:02So the exposure that we're interested
  • 39:04in this study was a legislative
  • 39:06mandate requiring a whether the
  • 39:07patient lived in a state that had a
  • 39:10legislative mandate requiring coverage
  • 39:11of DVT during the study period.
  • 39:14All states included as mandate
  • 39:17states in this analysis.
  • 39:19Also, a limited cost sharing with
  • 39:21the exception of Connecticut,
  • 39:22which eliminated cost sharing
  • 39:24one year after passage of
  • 39:26the general coverage mandate.
  • 39:28So in these laws when we say cost sharing,
  • 39:31these are including out of pocket
  • 39:33payments towards deductibles,
  • 39:34coinsurance or coherence similar
  • 39:36to what the ACA law would have is
  • 39:39for services that are rated A or B.
  • 39:42The control states were states that did
  • 39:44not pass a mandate during the study period.
  • 39:46And we assigned mammograms were
  • 39:47assigned to a state based on
  • 39:49location of the billing provider.
  • 39:53So, as I noted,
  • 39:54the outcomes we looked at were DBT.
  • 39:56Use the proportion of screening
  • 39:59mammograms performed with DVT
  • 40:00among all screening mammograms
  • 40:02for estate in a six month period.
  • 40:05So DBT is many people,
  • 40:06probably on the call,
  • 40:08probably know is typically read and built
  • 40:10in conjunction with standard 2D imaging,
  • 40:12so we consider DBT to have been
  • 40:14performed when there was a claim
  • 40:16for DBT in conjunction with a
  • 40:17claim for screening mammography.
  • 40:19We looked at the proportion of women
  • 40:21with any out of pocket payment.
  • 40:23We did also look at the mean
  • 40:24out of pocket payment,
  • 40:25but it became not that relevant.
  • 40:26So today I'm just going to present
  • 40:28results on the proportion that
  • 40:30had any out of pocket payment.
  • 40:33This is people,
  • 40:34women that had out of pocket payment.
  • 40:36We only looked at those with
  • 40:38DVT because women screened with
  • 40:392D mammography already had no
  • 40:41cost sharing which is mandated
  • 40:42by the Affordable Care Act.
  • 40:48So we used an event study design
  • 40:51which estimates changes in an
  • 40:53outcome among states that pass a
  • 40:55law relative to states that did not.
  • 40:57At each six month interval
  • 41:00after law implementation.
  • 41:01So this specification allows for
  • 41:03the effective laws to vary by
  • 41:05the time since implementation.
  • 41:07So basically what you do in event
  • 41:08study design is you line up the
  • 41:10implementation dates and look at
  • 41:11whether there are changes in our
  • 41:13outcomes in the first six months
  • 41:14post implementation that in the next
  • 41:16six months post implementation.
  • 41:18And this also has the advantage of.
  • 41:21It allows you to see if there were
  • 41:23changes in the six months prior
  • 41:36Which you would not necessarily expect us,
  • 41:39as is typical in any
  • 41:40different difficulty level,
  • 41:41and models were weighted by the
  • 41:43screened population in each state.
  • 41:46So this table represents our patient
  • 41:48characteristics and outcomes at baseline,
  • 41:51so we also are not so for outcomes,
  • 41:54it is the outcomes at baseline, so.
  • 41:59Right, OK at the start of the study period,
  • 42:02women and in mandate and non
  • 42:04mandate states had similar age.
  • 42:05Mean age was 53 in both among women in
  • 42:09mandate states 42% lived in the northeast
  • 42:12versus 12% in the non mandate states.
  • 42:14In early 2015, women living in states that
  • 42:18eventually pass a DBT coverage mandate 16%.
  • 42:21Of women who underwent mammography
  • 42:24were screened with DVT.
  • 42:26Versus among women living in states
  • 42:28that never passed a mandate 11% so
  • 42:31the screen was a little bit lower in
  • 42:32states that never passed a mandate.
  • 42:34Note, this is before the mandate,
  • 42:36though important to our study.
  • 42:38Really very few women in 2015 had any
  • 42:43out of pocket payment for DVT only 7% in
  • 42:47both mandate and eventual mandate and
  • 42:50eventual non and and non mandate states.
  • 42:53You can see that the DBT price was
  • 42:55higher than the mean 2D price.
  • 42:57For example, in mandate states,
  • 42:58the man demean DPT price was
  • 43:01$311.00 versus for two D $266.
  • 43:09Pre mandate.
  • 43:17So next we look at DBT, use and here's
  • 43:20our first outcome that we look at.
  • 43:22So let me Orient you 'cause the next
  • 43:23couple of slides all have the same
  • 43:25sort of framework as this slide.
  • 43:27So we lined up implementation dates
  • 43:29with the period labeled here years
  • 43:32from LA negative .5 being the period
  • 43:34in which the law was implemented.
  • 43:37So this shows the percentage point
  • 43:39change in DBT use in the period before,
  • 43:42and the period after the law was implemented
  • 43:44relative to states with no law implemented,
  • 43:46which is our comparison group.
  • 43:48So by construction,
  • 43:49the value for the period in which
  • 43:51the law is enacted is basically 0%,
  • 43:53because you're sort of normalizing
  • 43:55everything to be to for them to be the
  • 43:57same in that in that moment of enactment.
  • 44:00So first thing to look at is if you look
  • 44:02at the three periods that we can measure
  • 44:04here in the period prior to the law,
  • 44:06you see that there was no significant
  • 44:09effects of eventually passing a law.
  • 44:14We find no significant changes in DB use, D.
  • 44:17Use relative to the comparison test.
  • 44:19So this is really equivalent to
  • 44:21the standard parallel trends,
  • 44:22test apparel pre trans test that you
  • 44:23see in a different different analysis
  • 44:26in the periods after the law we do see
  • 44:28you can see a steady increases in.
  • 44:30Mandate states relative to other states.
  • 44:32So by one year post law these differences
  • 44:35are statistically significant.
  • 44:37One year after enactment of
  • 44:38a coverage mandate,
  • 44:39DBT use increased 7.6 percentage points.
  • 44:44Relative to other states.
  • 44:48Compared to states without a mandate,
  • 44:50I'm sorry 7.6% greater than states
  • 44:52without a mandate, and after two years,
  • 44:55D BTU's had risen 9 percentage points
  • 44:57more in mandate states compared to
  • 44:59states that did not pass mandates.
  • 45:06Next we look at DBT price.
  • 45:10And so it's the same format I noted before
  • 45:13from our patient characteristics table,
  • 45:15the average cost of a DVT was $311.00
  • 45:18among maintenance performed in states
  • 45:19that eventually passed a mandate and
  • 45:22347 states that did not pass a mandate.
  • 45:24And here we find that two years post
  • 45:26mandate and this was a surprise to us.
  • 45:28DBT Price had declined in mandate states
  • 45:30compared to the change in price in nine
  • 45:33non mandate states about $38.00 and I
  • 45:35don't have a graph here to show it.
  • 45:38'cause we have limited time,
  • 45:39but we also did not observe.
  • 45:41A significant change in the price
  • 45:43of 2D mammography.
  • 45:46Next we look at weather.
  • 45:50Here it is. At the percent of DBT DBT
  • 45:54exams with any added pocket payment.
  • 45:59Among women's group with CBT and we found
  • 46:01that even at the start of the study,
  • 46:03as I said earlier, only a minority of women
  • 46:06had any out of pocket payments with DVT.
  • 46:08We did not observe a statistically
  • 46:11significant change in the proportion of
  • 46:13women who had any out of pocket payments for
  • 46:15DVT even as we go to two years post mandate.
  • 46:18We did also look among those that
  • 46:20did have an out of pocket payment.
  • 46:22The mean out of pocket payment and we
  • 46:24did not find a statistic statistically
  • 46:26significant change there either.
  • 46:34So. A central policy objective of the
  • 46:38coverage, mandates or any coverage mandate
  • 46:40is to ensure access to a particular
  • 46:43medical technology or service by protecting
  • 46:46patients against financial liability,
  • 46:49and indeed, our results suggest that
  • 46:51women in states with coverage mandates
  • 46:52were more likely to begin to use DBT
  • 46:54for breast cancer screening,
  • 46:56which probably is one of
  • 46:57the intents of the law.
  • 46:58And this finding is consistent with other
  • 47:01studies across other types of services
  • 47:04that found that expanding coverage,
  • 47:07and in particular eliminating cost sharing,
  • 47:09can increase the use of
  • 47:10specific cell health services.
  • 47:11I'll say it's very difficult in many cases
  • 47:14to get patients to change their behavior,
  • 47:16but changing even by very small
  • 47:18amounts the amount they have to
  • 47:19pay is one way you can get them.
  • 47:21Generally,
  • 47:21the literature is found to
  • 47:23change their behavior,
  • 47:24but in our study this really
  • 47:26raises some new questions about the
  • 47:28mechanism by which mandates.
  • 47:30May increase use of an emerging
  • 47:33technology because we didn't find
  • 47:35changes in out of pocket payments
  • 47:37and even before these mandates,
  • 47:39the out of pocket payment was low,
  • 47:42so it it's unlikely that a change
  • 47:44in what the patient had to pay
  • 47:47is what led to these changes.
  • 47:50So one explanation for these findings
  • 47:53is that by ensuring payment coverage,
  • 47:56mandates may have encouraged
  • 47:57radiologists and other health care
  • 47:59institutions to enter the market.
  • 48:01And offer DBT and this may have led to
  • 48:03a relative price in at least two ways.
  • 48:06One when more radiologists offer DBT,
  • 48:09insurers really may have greater
  • 48:11ability to negotiate lower prices
  • 48:13and this could lead to lower prices
  • 48:15or at least slower growth in prices
  • 48:17over all providers.
  • 48:18Second,
  • 48:19it could be the case that early
  • 48:20entrance we would expect the
  • 48:22early entrance in this market.
  • 48:23When DBT first started to be
  • 48:25providers that have higher prices.
  • 48:27So for example, academic medical centers,
  • 48:29if mandates incentivize new entrants who
  • 48:32tend to offer services at lower prices
  • 48:35compared to established providers,
  • 48:37the average market,
  • 48:38the average price in the market
  • 48:39will decrease mechanically,
  • 48:40so you have a high price pipe,
  • 48:42high price providers, low,
  • 48:43lower price providers,
  • 48:44the average is going to go down.
  • 48:47But in that scenario,
  • 48:48no provider has actually changed their price,
  • 48:50right?
  • 48:50But the the price that is paid
  • 48:52in the market will decline,
  • 48:54so other explanations are possible.
  • 48:56For example,
  • 48:57it's possible that coverage mandates
  • 49:00might be perceived by patients or others
  • 49:03as an endorsement of this service.
  • 49:05And this could increase interest
  • 49:07in this new technology,
  • 49:08so we can't say for certain that this is
  • 49:10one of the two things that is happening.
  • 49:12Unfortunately we don't have a provider
  • 49:14identifier in our data that would
  • 49:17allow us to say whether it is.
  • 49:19Different lower price providers
  • 49:21entering the market.
  • 49:24Hey, I think we need to note
  • 49:26some limitations to the paper,
  • 49:28so there definitely could be some
  • 49:30issues with generalizability.
  • 49:31Since all data was from Blue Cross,
  • 49:33it is really really good data to look
  • 49:36at these this study because it is from
  • 49:38all 50 states in a very large data set.
  • 49:41Also, there are important known
  • 49:43limitations to using claims data.
  • 49:45Claims could be subjected to error
  • 49:48misclassification problems or bias.
  • 49:50Another issue with this very particular
  • 49:52setting is our approach focused.
  • 49:54We chose to look at the price of the initial
  • 49:56test rather than the screening episode.
  • 49:58In some of our papers we have
  • 49:59looked at the screening episode,
  • 50:01but you know that could be
  • 50:02very very different here.
  • 50:03If DBT does reduce recall and that could
  • 50:06lead to additional cost savings from for
  • 50:09DBT relative to to to 2D mammography.
  • 50:13Also this was an an observation ULL study.
  • 50:18Although we believe we used to study
  • 50:20design that intended to limit confounding,
  • 50:22unmeasured confounding is always
  • 50:24a possibility and could explain
  • 50:26some of our findings.
  • 50:28Could be you know other concurrent
  • 50:30legislative policies or other
  • 50:31things going on in the market.
  • 50:35Finally, although our event study plots
  • 50:38didn't show significant differences in DBT
  • 50:41user price prior to the law being enacted,
  • 50:44it's important to acknowledge that pre
  • 50:46period trends in DBT use or cost and mandate
  • 50:48states may may influence our results.
  • 50:50So there could be some pre-existing trends.
  • 50:55Hey, just to conclude,
  • 50:57although DVD mandates were associated
  • 50:59with an increase in DBT use,
  • 51:01they were not associated with any
  • 51:03change in out of pocket payments and
  • 51:06this suggests that mandates and this
  • 51:08has implications for other services,
  • 51:10well, may influence DBT adoption through
  • 51:12mechanisms other than by reducing
  • 51:14financial liability for patients.
  • 51:23Thank you Susan for a great presentation
  • 51:26that clearly damn straight close link
  • 51:29between policy and clinical practice.
  • 51:31I was wondering whether there are
  • 51:35studies being planned by you or others
  • 51:39to potentially look at the impact of
  • 51:42DBT of identifying more patients.
  • 51:45I was thinking that eventually,
  • 51:48if there's evidence that DBT
  • 51:51would identify more patients
  • 51:53because increased sensitivity,
  • 51:55that more B might be more
  • 51:57incentive for more states to
  • 51:59have similar laws mandating it.
  • 52:02When you see identify more patients,
  • 52:04are you saying that some people that
  • 52:06previously didn't get a mammography
  • 52:08would get a mammography because
  • 52:10the the DBT is available? Right,
  • 52:13I just thinking like like on what
  • 52:15basis would this states that occur,
  • 52:18like not mandating it like what?
  • 52:20Why would they be encouraged to do so?
  • 52:22Why would they be mandating so the
  • 52:241st that is really interesting?
  • 52:25Especially because we didn't
  • 52:27find it like where's the problem?
  • 52:28Out of pocket payments were
  • 52:31not particularly high.
  • 52:33Sort of before these are mandated.
  • 52:35Well, you know there might be some insurers,
  • 52:36but there might be some fear from
  • 52:39suppliers that insurers may stop covering
  • 52:40it or may start implement, you know?
  • 52:43Start putting in some out of pocket payments.
  • 52:46Yep. You know, I think,
  • 52:48why would a state not pass a mandate?
  • 52:52You know they may be looking to
  • 52:53the evidence and maybe looking
  • 52:55to the USPS TF if they're thought
  • 52:56of as an independent body,
  • 52:58they still have not gone up
  • 53:00to the ARDA or B rating,
  • 53:02suggesting there probably there
  • 53:03may be still some uncertainty.
  • 53:05Right in in the studies,
  • 53:07so that's why you might not mandate the
  • 53:10reason that you you know or also because.
  • 53:14It's not really clear that there's a
  • 53:16problem since people are not paying large
  • 53:18out of pocket payments for this service.
  • 53:22Sure. Yeah, because the laws are
  • 53:26at the state level and and the
  • 53:30US preventive taskforce hasn't
  • 53:32made a ARB recommendation.
  • 53:34I think that could be where
  • 53:36states are looking for.
  • 53:40Yes.
  • 53:43Am I supposed to look for questions?
  • 53:48Right? Please feel free to.
  • 53:52Type your question through chat.
  • 53:58Oh, here's Regina. Thanks so much, Susan.
  • 54:06OK, so Regina Hooley just has a comment
  • 54:09that Yale they first started using
  • 54:11DBT in 2011 and they didn't charge
  • 54:13patients for insurance for many years.
  • 54:15Probably not until 2018.
  • 54:16So I think Medicare did start
  • 54:19charging to 2:15 till 2015,
  • 54:21so I think few private insurers.
  • 54:24Maybe we're charging before that.
  • 54:26I think a lot of I think there
  • 54:29wasn't even a code until 2015
  • 54:31to 2 allow people to charge.
  • 54:34But that is great that Yale
  • 54:36was able to do that.
  • 54:59So what are the follow up?
  • 55:02Studies that that you are carrying
  • 55:05your team is planning. I know Alana
  • 55:08has a huge interest in this too.
  • 55:12Yeah, so so that's great and I you know,
  • 55:15I think we're talking about that
  • 55:17right now because this is one of the
  • 55:19this is a project I have two minutes.
  • 55:21I'll just describe how this project started.
  • 55:24And interestingly,
  • 55:24Joe Ross was also on this train ride.
  • 55:27I had a personal experience with.
  • 55:31Breast ultrasound,
  • 55:32which is what we really the technology
  • 55:34we were really interested in studying,
  • 55:36and Joe Ross and Carrie Gross and I
  • 55:40were on the same metro North train down
  • 55:42to New York for the same meeting and
  • 55:44we were just chatting on the train and
  • 55:46I said to Kerry, what's up with this?
  • 55:47You know what's going on this is,
  • 55:48you know, many years ago and I
  • 55:50said this is so interesting that
  • 55:51they're doing this mandate.
  • 55:52Let's write a grant and we ended
  • 55:54up writing a an ACS grant that
  • 55:57was funded to do this work.
  • 55:59And then Alana gotten bored and
  • 56:00we sort of extended it.
  • 56:01It's a DBT,
  • 56:02so it really did start out as this
  • 56:05just sort of kind of very random thing
  • 56:08that people just sort of talking.
  • 56:09It's funny that Joe is here about
  • 56:12this and ended up being this project
  • 56:13so that project has ended now
  • 56:15so that ACS project has ended.
  • 56:17So we're really thinking about what
  • 56:18would be the the best next steps and
  • 56:20what are the most interesting questions.
  • 56:22So I think Regina probably has some
  • 56:24good ideas so she's sort of been
  • 56:26involved in this so we we haven't
  • 56:27sort of gotten to the next project.
  • 56:29We're sort of finishing up the
  • 56:30the old project right now.
  • 56:31The older project.
  • 56:35There's a comment from Carrie
  • 56:37if you could address briefly.
  • 56:42I'm not sure that state
  • 56:44legislatures would look at that,
  • 56:45but I do think like that.
  • 56:46The advocates when you,
  • 56:47if you say you publish something
  • 56:49that suggests that they're really
  • 56:50good benefits to passing a law,
  • 56:52I think that the the advocates may
  • 56:54bring that to state legislatures
  • 56:56and that that can be very helpful.
  • 56:58Especially, I do think like some of the
  • 57:00state laws were the earlier studies
  • 57:02showing that state laws around mammography.
  • 57:04This is pre ACA and cost sharing that
  • 57:07those actually you know led to more
  • 57:10movies and potentially had some interest,
  • 57:11some effect.
  • 57:13On breast cancer identification?
  • 57:15Not necessarily.
  • 57:15I don't know if they ever got to mortality,
  • 57:17but I think those did have an impact.
  • 57:19Those studies.
  • 57:21And there's one more from Alana.
  • 57:25Yes, so a lot of notes and I that
  • 57:28these implications have other firm
  • 57:30or other emerging technologies.
  • 57:32So to thinking about how that
  • 57:34will adopt that, how that those
  • 57:36influence adoption and price.
  • 57:39Thank you so much Susan for taking
  • 57:42the time to share with us your
  • 57:44important work. Also thanks to Joe.
  • 57:48Help you both have a nice day. But by.