Collecting, sharing and analyzing data can help health care providers with the decision support to deliver more efficient care to beneficiaries. Making use of data in this manner allows commanders of military treatment facilities (MTF) to determine how many primary care providers, how many surgeons and what type of resource planning they need to execute in order to support their beneficiaries, said Captain Jamie Lindly, chief of the analytics division at Defense Health Agency (DHA), the centerpiece of the military health system (MHS).
“We are the foundation for a lot of the central processing of our enterprise data,” said Lindly. “We’re the ones that write the functional specifications, and there are multiple roles in how we use health care analytics. Part of that, from an enterprise responsibility, is making sure those central systems at MTFs have good data to make decisions.” Some examples of this data analysis supported inside MHS include how much utilization a patient has at a facility or how much purchased care a particular service has consumed for their enrollees in a given period.
“So we’re the back end of the central algorithms that go to support decisions across the enterprise,” said Lindly. “Just like the planning at an MTF, scale that up into an enterprise as a whole. One of the important things is that when we develop that number or methodology, it’s the same methodology we would [use] for a facility in Oak Harbor [Washington] that we would apply for any Army facility in the country, or the world. There’s standard methodology and there’s standard processing of our central data.” These standard methodologies are determined by a functional proponency group, which includes a representative from each service, said Lindly. These service reps propose a methodology or processing specification and then have the ability to agree, disagree or come up with amendments to those processes. It’s an evolving activity because things are constantly changing, and MHS is re-releasing data as changes are made to the process.
“It can change,” he said. “We do retrofits of the same data as well. We’ll say, ‘You know what? This variable isn’t populating the way it should. Let’s make this change and then reprocess a lot of the old data.’”
Being an analyst and looking at data is one thing, but the actual processing and ownership of the hardware, the actual day-to-day processing, is handled by the health information technology (HIT) people in the program office, said Lindly.
“We say here’s how you should process this data,” he said. “Then we write that specification and give it to HIT, who then build it into the weekly and monthly processes as we get more data from the CHCS [Composite Health Care System] host in these markets, or from other sources. One of the frustrations is that an analyst will always have a bigger appetite than the HIT folks can either execute or are budgeted for. That prioritization of analytic approaches or projects is probably the biggest challenge.”
Lindly added that he feels health care analytics will become timelier over the next five years.
“There are significant lags from a central point of view,” he said. “I basically have a 90-day lag before I can tell you with confidence what the enterprise looks like. I think that window is going to shorten.”
The lag is due to data maturity. For example, when a CHCS has an inpatient, they will send Lindly their inpatient data records once a month when they’re completed. If someone was in the hospital on the 15th of the month, but their records weren’t coded until the fifth of the following month, Lindly wouldn’t get that completed data until 45 to 50 days later. After it’s coded, it will be sent for central processing, causing the 90-day delay.
DHA is fed data from multiple sources and needs to standardize the analysis of this data.
“We take that data and look at all of the elements, and we try to process that in a way that’s common with the other feeds that we’re getting,” said Lindly. “We get feeds from the DMDC [Defense Manpower Data Center] for who’s eligible for care, we get feeds from Aurora [Health Care] that tell us what claims we paid, we get feeds from facilities from PDTS [Pharmacy Data Transaction Service]; we have all of these different feeder systems routinely giving us data on a scheduled basis. We process it in a way that’s similar across all [systems].”
One of the leading providers of health analytics to the military is SAS, which specializes in having a standardized process for receiving and storing data, as well as making data available to users. Their solutions and technologies encompass applicable analytic capacities available to the DHA, Department of Veterans Affairs and other military health units. SAS’ expertise in health analytics has been developed over decades in both the private and public health sectors, said Rick Ingraham, health intelligence officer, SAS Federal. He added that the objectives of military and veteran health are the same. Analytics should help improve care quality, care effectiveness, care readiness and managing care costs. SAS directly impacts clinical performance and patient outcomes across a wide spectrum of focus areas—from enabling increased insight into clinical factors (and non-factors) driving readmissions to analyzing huge volumes of structured and unstructured clinical and operational data to uncover variables impacting the delivery of care, as well as zeroing in on patient safety signals.
SAS has been involved in several health care analytic projects that are changing the nature of health quality and outcomes, care delivery, risk and incentives, and potentially avoidable care and costs, expediting the operationalizing research findings and enabling care financial sustainability via episodic care analytics.
SAS health analytics utilize previous investments in data to capture both internal military and external data sources, said Ingraham. SAS draws upon transactional care records within existing electronic health records, patient data, physician (both staff and purchased care) data and any other external data sources and metrics that would increase opportunities to view the efficacy and efficiency of current care operations in a different and more informed light.
SAS analytic software can be installed on site, or can be used via cloud-based and software-as-a-service approaches to gain the most analytic strength from SAS’ Center for Health Analytic Insight.
Advanced analytics include modeling and forecasting capabilities to go beyond simple statistical summarizations of past activity to better anticipate the future and take actions in anticipation of expected health care performance. Their analytics shifts analysis perspective from hindsight to foresight. This entails an analytic maturation process that moves an organization from addressing the “what?” and “so what?” of health care benchmarks and trends through statistical analysis to now addressing “what if?” via forecasting and predictive modeling and “what now?” via optimization techniques.
The discipline of health analytics covers every facet of the health ecosystem’s operational processes, said Ingraham.
“The opportunities for tangible benefits are now being realized in ways heretofore unimagined. Certainly, the HITECH [Health Information Technology for Economic and Clinical Health] Act funding of electronic medical records has triggered more rapid conversions to digital records,” he said. “These investments have been significant, but can be at-risk if the health community expects outcomes and quality to improve solely from digitizing the data surrounding care transactions. Rather, the use of advanced analytics can aid in quicker ROI via insights that will both impact costs and quality.”
SAS has been focusing on the gap these products raise as clinical and operational staff struggle to convert volumes of data into understandable insights and changes to care delivery. A recent collaboration between SAS and a national integrated health provider used an analytics-as-a-service approach to identify opportunities to reduce readmissions, target (initially) congestive heart disease and sepsis for best-practice determination, manage pharmacy costs and outcomes and create tools to improve each patient’s experience. “As both the public and private health sectors are exploring avoidable care costs and more value-based payment approaches, we have been aggressively developing an episodic analytic solution focused on better understanding efficiency, a patient-centered focus, cost management and increased quality of care,” said Ingraham.
He added that some of those specific areas include episodes of care measured by condition signals, the entirety of care across an episode of care, potentially avoidable complications relevant to a condition, clinical associations between episode identification, patient-severity/risk-adjusted cost comparisons and accurate calculation of true episode costs.
There are challenges associated with analyzing this magnitude of data. The challenge is less of development and more of adoption, said Ingraham.
“Moving advanced health analytics into the mainstream of the military and veteran health care organizations has followed a course similar to other private and public health organizations,” he said. “In fact, the health industry has lagged behind others as far as adoption of advanced analytics as a cornerstone for operational excellence. It is not uncommon to hear health management leaders discuss electronic medical records without mentioning methods to analyze care data. This would be a tremendous missed opportunity.” Ingraham said SAS has observed two significant barriers to accelerating analytic adoption rate. The first barrier is that organizations tend to have one of two cultural mindsets around data and technology: scarcity or abundance. A scarcity environment tends to be constrained by data and technology, be process-oriented, be focused primarily on cost control and operate with an “everything is forbidden unless permitted” standard that hinders innovation.
An abundance environment, conversely, leans toward empowerment from data and technology, being discovery-centric, having a focus on value and operating with an “everything is permitted unless forbidden” standard to drive innovation, said Ingraham. He added that patient privacy and data security need not be a victim of this standard.
“Identifying where an organization stands in the analytic maturation process is a critical first step to making changes,” he said. “SAS has developed very precise business analytic management assessments that can aid in the transition from a culture of scarcity to abundance and assisted health management.”
The second challenge Ingraham sees is that some organizations are better at interpreting analytics results than others.
“Advanced analytics has often been met with hesitancy among management to adopt the full scope of techniques,” he said. “Why? It’s primarily because of difficulty in interpreting results. Health organizations have addressed this via new data scientist positions as well as investment in clinical informatics and economics units. However, it remains critical that managers responsible for using analytic health care insights to change care procedures can easily use the insights surfaced via analytics.”
SAS has worked toward empowering these “change managers” through the use of SAS Visual Analytics and Visual Statistics and via delivery to common personal devices such as the iPad.
Ingraham feels health analytics will evolve in several ways over the next several years.
“While it may not be restricted to five years, I anticipate the discipline of health analytics to include an increasing number of data sources and observations,” he said. “Partly driven by the Affordable Care Act’s focus on accountable care organizations and value-based reimbursement (moving away from fee for service), there must be even further exploration of the factors that either directly, indirectly or subtly impact care outcomes, care quality, patient engagement and costs. This cannot be done without advanced analytics.”
He added that he expects increasing diligence in addressing the identification and reaction to gaps in patient care as they initially happen (signal detection), identification of in-patient/out-patient care variables impacting readmission, reoccurrence and errors, and further understanding of patient and family-caregiver acuity to develop more individual engagement methods to impact medical directive adherence.
“If we do this right, health providers will come to utilize analytic results on a daily, if not minute-by-minute (and patient-by-patient) basis,” said Ingraham. ♦
- Issue: 5
- Volume: 18