It’s not only individual hospitals, medical groups, and health systems whose leaders are working to leverage data to improve outcomes and lower costs, in the current shift away from pure fee-for-service payment, and towards value, in the U.S. healthcare system; the leaders of associations of provider organizations are doing so as well.
One of the provider associations whose leaders are moving forward quickly in this area is the Austin-based Texas Hospital Association (THA). The THA’s leaders have steadily been developing a strategic data management framework for unifying patient records and exchanging information across the state. Based in Austin, THA is one of the largest hospital associations in the country, representing more than 85 percent of the state's acute-care hospitals and health care systems, which in turn employ some 365,000 health care professionals statewide.
Driven by its vision of a unified healthcare delivery system for Texas, THA is leveraging the EMPI solution from the Monrovia, California-based NextGate, in order to assess and analyze data from its member hospitals, in order to help them improve patient outcomes and reduce costs. The EMPI gives THA the ability to accurately match and link patient information in the absence of a common identifier, allowing providers to achieve a unified view of any patient across all facilities and health systems with the ability to share information, regardless of system, for improved care coordination.
Fernando Martinez, Ph.D., who is both senior vice president and chief digital officer of the Texas Hospital Association, and president and CEO of the Texas Hospital Association Foundation, the organization’s research and educational arm, spoke recently with Healthcare Informatics Editor-in-Chief Mark Hagland regarding THA’s bold data analytics initiative. Below are excerpts from that interview.
Please tell me a bit about yourself and your role at the association and the foundation.
I’ve been with THA for three years, as of December 1. My background is as a turnaround CIO in healthcare; I’ve worked in healthcare since I was 16. I spent the first half of my career working in finance, including revenue cycle, and then transitioned from finance into running IT, and the last 15 years, I’ve worked as a turnaround CIO, primarily in public healthcare. The CEO at THA had been my former boss at Parkland Health System in Dallas, and also at Jackson Health System in Miami, before that.
You’re the first chief digital officer at THA, correct? And what is your mission at the association?
Yes, that’s correct. And my core mission is not only to provide the technical platform from which to run the organization, but also to create, from a digital strategy point of view, a platform of data that will accurately allow our policy, advocacy and legislative folks to provide folks in the Legislature with recommendations that impact Texas hospitals. We’re very dependent on data. And one of the strategic objectives I’ve been asked to execute on has been building a statewide database that provides a longitudinal view of the data, including elements around costs, utilization, and quality, in order to help us support and advocate for, Texas hospitals.
What have been the challenges in creating that statewide database with a longitudinal view?
The most significant challenge has been the diversity of healthcare organizations in the state of Texas. I frequently interact with my peers at other state hospital associations; and it turns out that some associations don’t have the level of diversity around data, with regard to all the critical-access hospitals, that we have here in Texas. In Texas, we have a very large number of critical-access hospitals, a very large group of rural hospitals, a significant number of independent community hospitals; as well as some large for-profit hospitals, like Tenet, which is based here, and HCA, which has a strong presence; as well as many large not-for-profit hospitals. So the biggest challenge is representing the reality of all these diverse hospitals. And that’s challenging, because if getting 10 hospitals to agree to anything is challenging, getting 500 to do so is really challenging.
Can you share what the timeframe around this initiative has been?
Yes, it started a few months after I came, so we’ve been working on it for nearly two years.
And how long has it been live?
For between six months and a year.
What types of data does the database encompass?
It’s primarily claims data. This is a statewide, all-payer, longitudinal database, founded on claims data.
How is it being used?
The association is aggregating the data, and in terms of how it’s being used in the policy and advocacy arena, our policy folks, legislative team, and analysts working with state legislators, are the ones who consume the data and look at the trends, for policy people in the legislature. Meanwhile, some of our hospitals participate in the data program, which uses this longitudinal database as well.
In that regard, the database also serves as a tool for our hospitals—roughly 250 hospitals that participate in the data program, and are able to look at claims data across the state, and to model data. The data program itself has been in place for many years, well before the database was established; we just now have the additional capability of a broader, all-payer view of the state, which wasn’t available before. We’re not alone in the U.S. in building an all-payer, longitudinal database; probably 35 of the state hospital associations have built longitudinal or all-payer databases for their states.
What led you and your colleagues to partner with NextGate?
NextGate came into the equation when we were looking at creating the potential for additional analysis around patients, patient behaviors, clinical trends, etc. The industry at large is going from predictive analytics to cognitive analytics. Having said that, our members wanted greater insight, and want to reconcile social determinant data to clinical outcomes, to any number of diagnostic and post-care metrics that you can pick up from claims. But to do that, you need to normalize a patient’s identity. To see the social determinants associated with a particular individual, how that’s reconciled against a pattern of care and treatment, you have to be able to identify that person across a variety of hospitals and care venues. So having the master-patient index capability is inherent to evolving a data program forward to look at more advanced analytics, and model your data in meaningful ways. So NextGate happened to be the organization that we felt most confident about in terms of their capabilities.
So some of your hospitals are looking to analyze patients across venues for population health and care management?
I wouldn’t necessarily make that leap, but what hospitals are doing is identifying individual patients, in order to see what a particular patient is doing across multiple organizations. And in that context, de-identified information, in terms of trends and behaviors, is valuable. And we’ve found that our members do find our members are interested in that. Reconciling care patterns and social determinants allows them to think about these issues. It’s not so much that we have Mark’s information—it’s that we have a certain type of patient, and we notice that their behavior patterns and outcomes differed from someone with a different set of social determinant characteristics, as shown by their data. And then, having looked at the larger body of data and identified trends and characteristics, they can go into their databases and look for patients who meet particular types of criteria, and can work on adjusting their strategies as a result.
The ability to identify trends and reconcile certain outcomes and patient characteristics and attribute them to social determinant or geographic data, is helpful. For instance, rural versus urban differences—data on those differences is actionable. Because they have much more than just claims data, they also have financial and clinical data. And once they’ve identified a pattern that relates to behavior around a particular service line—some hospitals have good revenue cycle information, as well as other types of information, and they can correlate trends they see statewide, and can go into their data set and create better treatment protocols, develop a new service line, and can anticipate care or treatment options.
In the shift from volume to value in U.S. healthcare, hospital and health system leaders are trying to figure out, using data, how to track and improve their clinical and operational performance, correct?
Yes. I will tell you that ten years ago, we had a problem in that we couldn’t get enough data. We were constantly looking for sources of data in order to build big enough databases in order to do enough meaningful analysis. In the past ten years, the pendulum has swung to the other side, where organizations are struggling with the large volume of data they have. So the challenge going forward is not whether you have enough data to do analysis, but rather, how to transform data into information, and then into analysis. So, industrywide, I think there is a significant push towards deriving value from data.
The era of, we have to collect more data, has given way to a new era of trying to derive value from the data. And certainly, claims data is a subset of the overall range of data available. But when you take certain types of data and you can then normalize certain types of patients, all sorts of interesting trends come out of it, which allow you to focus more and mine your own data to analyze trends. And everyone’s facing similar challenges: all associations want to provide greater value to our members; we want to help our hospital members provide more efficient care—the margins are diminishing all the time—and ultimately, we want to provide better care for patients. And to the extent that we can use analytics to support that, we’re meeting our goals in helping our member hospitals. I think that all of us at hospital associations nationwide are trying to help our members to succeed in that shift.
Does the state of Texas require hospitals to publicly post outcomes data?
Yes, in Texas, the hospitals are required to report inpatient, outpatient, and ED data to the state; it is compulsory.
Does that mean reporting specific data on conditions and treatments?
No, it’s claims data reporting. At the end of each quarter, each hospital has to report to the state that quarter’s claims data. All payers they work with, all visits—the typical 835 code set—each hospital provides a quarterly file to the state, with all its claims. And all the states in turn consolidate that data, and report it to the federal government, and that’s how the federal government develops national data sets around conditions. But it’s all claims data, using the normal 835 claims data set.
What will happen over the next few years at THA?
That really is the “64,000-dollar question.” I’ve been a professor of graduate education for over a decade, and I’ve found that more and more, the industry is evolving forward away from having its data isolated in vertical silos, and towards sharing data. A lot of the data is not yet publicly available, simply because the mechanism for doing so is not available, or because of policy; and in that regard, I think the barriers will be coming down. HHS [the federal Department of Health and Human Services] and other agencies, have been seeing net improvements—quality scores overall have dramatically improved—and the reality is that a lot of the improvement has come from data analysis, and from reconciling different data sources.
Once the mechanisms for data-sharing have been developed, it will be logical to evolve forward and say, maybe there are more ways to reduce cost and improve outcomes. So there will be a growing trend of transparency around aggregate bodies of data, with the hopes that organizations will continue to use that data to drive down waste and inefficiency and drive up outcomes quality and drive up care models.
Is there anything else that you’d like to add?
I certainly think it’s important to recognize that the transition away from traditional fee-for-service healthcare and towards value-centric payment models is built on a foundation of providing care in the most efficient way. And that really requires a lot of integration of data. And we’re finding that it’s convenient to be able to normalize patient identity. And if all the participants in that patient care continuum have the ability to normalize patient identity, it enables a more seamless and elegant way for patients to flow from one venue on the continuum to another. As long as we can’t normalize individuals across disparate organizations, sources, or venues of care, there’s an inherently organically built-in barrier to that transition. So in spite of what we may be doing around normalizing patient identity, there’s a much greater objective we should be seeking as an industry. To build these value-based models, we need to get better at identifying our patients uniquely; doing so will accelerate this work, and improve the value of care.