Flatiron Health Leverages Expertise With Real-World Data to Examine Cancer Care Disparities

Evidence-Based Oncology, December 2021, Volume 27, Issue 9
Pages: SP397

During the COVID-19 pandemic, the challenge of reducing disparities in health care delivery and outcomes soared to the top of the policy agenda. It’s a topic at most scientific meetings, and health plans and pharmaceutical manufacturers are building teams to address this issue.

For investigators who work with real-world evidence, health care disparities have been on the radar for some time. However, the current focus on disparities is shaping research priorities.

Flatiron Health is well known for its digital products that help community oncology practices collect and analyze data, provide insights for patient care, and promote clinical trial participation. Today, Flatiron is expanding its footprint in health equity and disparities research, with both community and academic partners.

A team from Flatiron Health recently published data first presented at the American Society of Clinical Oncology (ASCO),1 which highlighted how Medicaid expansion under the Affordable Care Act (ACA) has narrowed the gap between Black and White patients in receiving timely care for cancer. Results were published in the July issue of The American Journal of Managed Care® (AJMC®).2

The study covered 30,310 patients, including 12.3% who were documented as Black. Investigators found that, in states without Medicaid expansion, Black patients were less likely to receive timely treatment than White patients—43.7% of the Black patients started systemic first-line therapy within 30 days of diagnosis, compared with 48.4% of White patients (adjusted difference, –4.8 percentage points (PP); P < .001). In states with Medicaid expansion, this gap closed to the point where there was no significant difference (49.7% vs 50.5%; adjusted difference, –0.8 PP; P = .605).2

In an interview with Evidence-Based Oncology™ (EBO), Flatiron Health’s Kathleen Maignan, MSN, NP; and Blythe Adamson, PhD, MPH, discussed the AJMC® study, what the data have continued to show, and Flatiron’s research plans. Maignan is an oncology nurse practitioner who served as senior clinical director of Research Oncology at the time of this interview. She is now the Medical Affairs executive director at Genentech. Adamson is a principal quantitative scientist at Flatiron Health and the lead author of the study that appeared in AJMC®.

EBO: Your study covered a period from 2011 through January of 2019. Have you continued to look at these data in period since the study ended? Have the patterns you observed continued?

Kathleen Maignan: Absolutely. We continue to track our data longitudinally, so there’s never going to be a pause in investigating. But what we’re dealing with now—in the treatment landscape in 2021 versus what we encountered prepandemic—is very different. It’s difficult to draw any early conclusions.

For example, we collaborated on a study led by investigators at the University of Pennsylvania—one of our academic partners—and this study was presented at ASCO during the 2021 meeting.3 It looked at the impact of COVID-19 on time-to-treatment initiation of systemic therapy for patients with advanced cancers. Surprisingly, at a national level, we did not find evidence to suggest that the COVID-19 pandemic was associated with longer time to starting systemic treatment. That might seem counterintuitive, but one explanation could be that fewer patients were diagnosed with advanced cancer during that period and that some patients potentially delayed surveillance follow-up visits. The smaller volume of patients may have been able to access timely care despite pandemic-related health care disruptions. Compared with previous years, it’s hard to compare this using the metric of time-to-treatment initiation, but that’s something we’re going to look at very closely in the years to come—I would even say the months to come, as this is all in near real time.

EBO: Blythe, did you want to add anything?

Blythe Adamson: I’m really proud of the work evolving and all the research questions being responsive to the policy dilemmas we’re seeing today.

EBO: The phenomenon of fewer patients seeking cancer care has occurred alongside huge drops in screening, so I presume you’re looking at that as well?

Maignan: Absolutely. The drop in screening [documented by others] was quite alarming in the beginning of the pandemic. We’re starting to see those numbers pick up, but it’s something we’re keeping a very close eye on throughout the progression of the pandemic, as it’s still ongoing.

EBO: As Medicaid expansion has unfolded, I’ve been interested in differences between states that adopted expansion at different points in time. In addition to the overall natural experiment, I’m wondering if you’ve seen differences in the data between the states that expanded right away vs those, such as Louisiana, that expanded later in the process. Has this phenomenon had any effect on the data that is noteworthy?

Adamson: That’s such an interesting research question. It’s not one we were able to answer with this study. We would probably want to design an entirely new study. You would have to create a whole new hypothesis to be able to test and answer that one, but it brings up an interesting point in thinking about regional differences and how treatment patterns differ between states. As I brainstorm, how [I would] design a research study to test the hypothesis you’re describing, and try to unpack and explore what the confounders might be, I think about representativeness. One thing I’ve been learning a lot lately is that, even across the United States, treatment patterns can vary from the West Coast to the East Coast. Even sequences of treatments can differ within the United States and outside the United States. So, While I wasn’t able to look specifically into the early vs late adopters [of Medicaid expansion], considerations I would think an academic partner who might be interested in doing a study like that would be adding in attributes of individual states and even using methods, [such as] instrumental variables, that might be able to help guide based on what the types of policy decisions or treatment patterns those individual states have made in the past and how that might have influenced the time at which they adopted [expansion] early vs late. But it’s an interesting research question.

EBO: Kathleen, did you want to add anything?

Maignan: Blythe hit it on the head. It’s one of the things we were curious about when developing the study. But again, over time, we could start to think of new methods to do that.

Adamson: A parallel of why I think it’s important for researchers to figure out how to untangle policy effects for early vs late adopting states [involves] what we’ve seen was important at the beginning of COVID-19. With lockdown policies, masking policies, or closures of bars and restaurants, [there were] states that were early adopting states vs late adopting states. Untangling what the characteristics of those individual states were that led to those decisions—and the timing of those decisions—is important to be able to understand the causal effect of that policy on the outcomes that were achieved.

EBO: You’ve brought up the issue of regional differences. Obviously, that’s been a standout phenomenon of the whole Medicaid expansion experience. As you know, most of the states that have not expanded are in the South. Unfortunately, [these states] have a lot of the population that is most in need. Knowing that this area is where Medicaid expansion has not happened, I’m wondering whether you’ve had any engagement with state policymakers or any legislators in the states that have not expanded, and what those discussions have been like?

Adamson: It’s been one of my original intentions in designing this study that it would be useful for policymakers, not just to describe health disparities, but also to evaluate the effectiveness of [Medicaid expansion]. I’m delighted to report that these research findings have been cited in policy briefs for legislators across the country. It was cited in ASCO’s position statement on block grants in Medicaid and their impact on cancer care. This is exactly why this type of research is critical for informing policy.

EBO: So, your conclusion is that the use of clinical data during routine care can allow for real-time evaluation of health care policy. Since your study appeared, we’ve had CMS announce a “refresh” of the work on payment models undertaken by the Center for Medicare and Medicaid Innovation (CMMI).4 If you’ve had time to review the CMMI proposal, what can we expect? Can we anticipate that the use of real-world evidence on disparities will be used by CMMI? To create the successor to the Oncology Care Model (OCM)?

Maignan: We absolutely hope so. We’re early adopters, in that Flatiron is an early adopter of unlocking the use of real-world data and generating real-world evidence for cancer care and research. If you look at everything we know about how drugs work, we’re getting most of that information from clinical trials. We understand that is the gold standard for medical evidence about efficacy, but there are relatively few patients in trials that represent the totality of the patients we see in our network. In clinical trials, patients tend to be younger, they are better educated, they’re less diverse, [and] they have fewer comorbidities than the overall cancer patient population. So, when you look at a field, such as oncology, where it’s becoming increasingly specialized, we need the ability to use real-world data and real-world evidence to investigate things that are critically important to life after OCM.

The OCM expires in 2022, and we know there will probably be a gap. So there is a sense of urgency, especially given the treatment landscape. Novel therapies tend to be incredibly expensive. We have the opportunity to look at outcomes, using patients in our database who are adopting some of these early novel therapies to ask, are they even more effective than what’s available? Are they more effective than what’s been traditionally available in the market? Can we compare them with traditional chemotherapy or immunotherapies? Is it worth the extra expense of those novel therapies? We don’t know that until we can look at those outcomes across a broader population.

We are also thinking about different care settings. As we have seen with COVID-19 and the explosion of the use of telehealth, I don’t think that’s a genie that’s ever going to go back in the bottle. Although we may not see telehealth continue at the level we saw during lockdown, we know this is something we’re going to start to integrate into everyday care and management of oncology patients. So what is the impact? That’s something we can see in real time or near real time, and that will be critically important to building that next model beyond OCM.

EBO: Since the start of the ACA, given the decisions by different states whether to adopt Medicaid expansion, I’ve wondered when we would see results that show the differences in the health of different state populations who are entering the Medicare population. Would we see differences in population health from expansion vs nonexpansion states? For example, would we see different percentages of patients who have had regular cancer screenings vs those that did not? Are these the kind of studies you’ll be doing in the future? How can real-world data help in this regard?

Adamson: This is such an important research question. It’s not one my team is currently answering right now. There could be an opportunity for an academic research partner to answer this with Flatiron data. One of the specific challenges of designing a study like that and understanding, “Is the population entering Medicare healthier?” is that we would hope that, among populations with cancer, who are close to aging into Medicare, if they were receiving poor-quality care and had limited access to treatments, they might die before they even are able to age into Medicare. If they received high-quality care and had great access to life-extending treatments, then we would expect more people to be alive to age into Medicare, which is good.

It would require a thoughtful study because, if you just simply looked at the health of people entering Medicare, you might think a sicker population over time meant people were less and less healthy, when really, your pool of people was actually getting larger because you’ve kept more people alive to be able to enter Medicare. I would invite any of our academic partners to think through the best way to answer that because all of us have an investment in a healthy Medicare population and want everyone to be able to reach that point, with all the investments made in preventing severe illness.

EBO: We talked earlier about the ongoing issues with cancer screening. Are there any other issues related to the deferral of cancer screening that you’ll be examining?

Adamson: There are studies going on right now, within Flatiron and with our partnerships with academic researchers, particularly related to the impact of COVID-19—including the impact it’s made on cancer screening visits. We have more work coming out soon on the substitution of in-person office visits with telemedicine visits, which has been an incredible gift that I hope we continue to see going forward.

Maignan: When we think about ongoing work in this area, obviously, the pandemic has highlighted alarming disparities in health outcomes among all types of underrepresented groups. What we know is that race alone can’t tell the full story. Among the things we’ve been working on in this past year is not just our ability to capture race data, but look beyond that—to investigate what are better ways to look at socioeconomic status, understand relevant social determinants of health, and build an index using Flatiron data? How do we improve capture of our insurance variables? How do we increase the representation of patients on Medicare or Medicaid? How can we better standardize our race data? Again, this is one of the great advantages of working with a very recent data set where we can be flexible and improve on those variables and our ability to tell those stories in almost real time.

EBO: Finally, we wanted to discuss whether there’s a growing appreciation for the importance of natural experiments. And is there greater acceptance of natural experiments in health care since you published your study?

Adamson: It was very affirming that, just recently, the Nobel [Memorial] Prize in Economic Sciences has been awarded to 3 professors for the development of the methods we were able to use and deploy in our ACA study.5When we first presented these results in 2019, there was a lot of pushback from oncologists who had never seen methods like this used before. In medicine, we have the privilege of designing randomized controlled trials that are excellent at helping us understand the cause and effect to evaluate things, such as [therapeutic] treatments. But when we think about designing effective health care policy, it would be very difficult to run a randomized experiment to test the effect. Although the approach we used has been used for decades within the field of economics, it was much more unfamiliar to oncologists [at the time]. I was grateful for the patients, providers, and cancer community members who took the time to understand the methods and approach used to understand the effect on racial disparities.

EBO: You’re referring, in part, to the work of David Card and Alan Krueger, and the work on the minimum wage they did at Princeton University.

(Editor’s Note: Card, now at the University of California, Berkeley, and Krueger, who remained at Princeton University and died in 2019, showed that a minimum wage increase in New Jersey in 1992 did not lead to job losses. Until then, it was assumed that raising the minimum wage would cause job losses.6)

Adamson: Yes! In economics, natural experiments have been an incredible tool that’s been essential for us to understand things, [such as] the effect of minimum wage, immigration, or adding an extra year of education for someone. Those are things that would be very difficult to do a randomized control trial for. I’m really grateful for all the oncology community [for having] the patience, curiosity, and willingness to learn of a study design that they had not been familiar with. I’m happy to see there is recognition of the importance of this type of work and the opportunity we have to better understand policies and our institutions—the way that even cancer clinics vary from each other, and the ways we can modify the system to get more equitable health outcomes for all these patients.

EBO: Is there anything we didn’t cover that you’d like to add? I always ask that at the end, if there’s anything we didn’t touch on that you would like to talk about.

Maignan: One thing I always come back to when we start thinking about disparities work—there’s a James Baldwin quote we anchor to: “Not everything that is faced can be changed, but nothing can be changed until it is faced.” And [although] we’re not in the hospital daily, [and I’m] not at the frontlines the same way I used to be, as a nurse practitioner. Just being able to tell the story is so vitally important to making real change.

Adamson: There’s one more thing I would add. One of the spillover gifts of the research we’re discussing today is that the response from Flatiron as a company has been incredibly encouraging. The investments they’re making now in health equity and health disparities research is just astounding [to me]. We are currently recruiting for a head of health equity and disparities research to help us build a productive and transformative scientific and data-driven agenda, and Kathleen has been one of the essential leaders within Flatiron for this type of research and building this program. So I’m really grateful [for] her and hope that we are able to quickly identify a talented [investigator] to join us.

Maignan: In the meantime, we are actively supporting investigators who are working on disparities research. We have a joint grant-making program with the American Cancer Society. We’re now in our third year, which is exciting. We also awarded 2 health equity and health disparity grants to [investigators] within our academic network, so we’re excited to bring new talent on board so we can keep this energy going.

References

1. Adamson BJS, Cohen AB, Estevez M, et al. Affordable Care Act (ACA) Medicaid expansion impact on racial disparities in time to cancer treatment. J Clin Oncol. 2019;37(suppl 18). doi:10.1200/JCO.2019.37.18_suppl.LBA1

2. Adamson BJS, Cohen AB, Gross CP, et al. ACA Medicaid expansion association with racial disparity reductions in timely cancer treatment. Am J Manag Care. 2021;27(7):274-281. doi:10.37765/ajmc.2021.88700

3. Takvorian SU, Parikh RB, Vader D, et al. Impact of COVID-19 pandemic on time to treatment initiation for patients with advanced cancer. J Clin Oncol. 2021;39:(suppl 15). doi:10.1200/JCO.2021.39.15_suppl.1528

4. Innovation center strategy refresh. Center for Medicare and Medicaid Innovation. October 20, 2021. Accessed November 10, 2021. https://innovation.cms.gov/strategic-direction-whitepaper

5. The Sveriges Riksbank Prize in Economic Sciences in memory of Alfred Nobel 2021. The Nobel Prize. October 11, 2021. Accessed November 10, 2021. https://www.nobelprize.org/prizes/economic-sciences/2021/summary/

6. Card D, Krueger AB. Minimum wages and employment: a case study of the fast food industry in New Jersey and Pennsylvania. National Bureau of Economic Research. October 1993. Accessed November 10, 2021. https://www.nber.org/papers/w4509