HMT: Will you define in consumer/general/layman’s terms “leverage clinical data analytics to predict outcomes, measure trends and establish correlations that drive quality care at lower costs?”
Todd Rothenhaus, Chief Medical Information Officer, athenahealth
This assumes that leveraging clinical data is the first step to improving care and lowering costs, but it’s not. In order to drive quality care at lower costs, provider organizations must first gain insight into the activity happening across their care team and patient populations, and then work to identify where big savings can be found — by reducing overutilization of services, maximizing contractual gains or improving coding and billing practices. Fundamental operational explorations and changes like this don’t require “big clinical data” analysis, but can contribute much higher immediate value. However, for large health systems that have already addressed operational pain points for financial gain and have budgets large enough to invest in tools that leverage big clinical data, predictive analytics is worth it because in most cases these organizations have bigger budgets to explore with and to mine data for clinical advantage. But for many, this is not the case.
In short, to address cost effectiveness of care delivery an organization needs more than clinical data, it needs claims data and a clear view on things like provider productivity, scheduling trends, payer mix, coding patterns, etc. – things that weigh heavily on financial performance in today’s healthcare landscape. It’s most important for practices to understand areas for financial and workflow improvement, and then bring these learnings back to the point of care.
Patricia Birch, Vice President and Healthcare Consulting Practice Head, Cognizant
Leveraging clinical data analytics for quality care at lower costs can be accomplished in many ways.
- Predictive analytics can be applied to identify those individuals within a population that are most likely to respond to care management interventions.
- Clinical data analytics can be leveraged to measure trends in disease prevalence and population risk profiles.
- Analytics can be utilized to evaluate the relative effectiveness of specific care management programs on engaging patients in better self-care, improving overall population health status and reducing acute and emergent care events. In addition, assessing and confirming the correlations between perceived data patterns/trends – and revising care management strategies and interventions accordingly – has proven to be an effective component of industry-leading population health management programs.
Eric Mueller, Director, Product Management, Lumeris
Leveraging clinical data analytics to predict outcomes, measure trends and establish correlations that drive quality care at lower costs means that a patient’s primary care physician, specialist and ancillary service providers have a complete view of a patient’s health status informed by claims, EMR, lab, pharmacy and other data. When we are able to use data to make better-informed decisions, we can impact quality, cost and utilization for patients and populations, which is critical to achieving success in value-based care.
Bonnie Cassidy, Senior Director of HIM Innovation, Nuance
Clinical data analytics, if used to its fullest potential, can enable the healthcare industry to create a true partnership between patients, providers and payers. Personal health information supplies patients with the knowledge they need to make informed health decisions. Additionally, it serves as a set of metrics that physicians can use to discuss lifestyle improvements that can help their patients better manage their own health and well-being. A knowledgeable and well-informed patient population will make better health decisions, which leads to improved health outcomes. This means that payers will also see the impact of analytics through reduced claims information, as engaged patients take a more proactive approach to managing their health.
Analyzing accurate clinical data also offers a clear snapshot of the larger patient population that a particular healthcare organization or facility is serving. Armed with such insights, providers can tailor their health strategies to their community, offering health education courses or other health services that will benefit their patient population.
Anil Jain, M.D., FACP, Senior Vice President & Chief Medical Information Officer, Explorys Inc.
Simply put, it means that we ought to look at all the information that we already collect and record about our patients over a period of time such as their habits, vital signs, diagnoses, laboratory test results and medication use to identify patterns that tell us which patients are going to do well and those that will not do well. We can use similar information to show what procedures and medications physicians prescribe appear to be necessary for the patient’s well being and those procedures or medications that don’t appear to benefit the patient or even harm them. Avoiding unnecessary services reduces the overall cost while maintaining the highest levels of quality and patient satisfaction.
Dan Riskin, M.D., CEO, Health Fidelity
To date, the U.S. has aimed to capture healthcare information electronically. The goal in doing so was to improve outcomes and reduce costs. Thus far, we have created the foundation for accomplishing these goals, but not yet created value.
Leveraging clinical data requires efforts at the point of care and at the population level. At the point of care, we must provide clinical decision and interoperable patient information to reduce errors. At the population level, we must find areas of low performance and ensure these areas and instances are addressed. We must perform predictive analytics to find patients at risk and drive resources their way in a preventative approach. We must also leverage technology to assure the standard of care is followed not just for sickness, but also for wellness.
Tony Jones, M.D., CMO, Philips Healthcare’s Patient Care and Clinical Informatics’ Business Unit
Most healthcare organizations today are using two sets of data:
1. Retrospective – basic event-based information collected from medical records or insurance claims; and
2. Real-time clinical – the information captured and presented at the point of care (imaging, blood pressure, oxygen saturation, heart rate, etc.).
Predictive analytics is combining these two data sources so that clinicians can access the relevant information they need to identify trends that will impact the decisions they make at the point of care.
Much of what’s needed for predictive analytics is already available – labs, images, patient history and vital signs. What’s changing is the ability to capture and store this data, and then apply new tools and algorithms to analyze the data. These analyses can detect patterns, and if the patterns repeat over and over, this provides confidence that it may be useful in predicting a future event rather than waiting for that event to occur before initiating treatment. In many cases, a life-threatening, expensive medical event can be avoided or easily managed if the signs and symptoms (i.e., the pattern) can be detected earlier than we often do today.
For example, if a diabetic patient enters the hospital with numbness in their toes, instead of immediately assuming the cause is their diabetes, the clinician might monitor their blood flow and oxygen saturation, and potentially determine if there’s something more threatening, like an aneurism or stroke, around the corner. That data is entered into the EMR as part of the retrospective data for that specific patient.
If more diabetic patients enter the hospital and present a similar trend of numbness in their toes, the coupling of real-time data (these new patients entering the hospital) and retrospective data (which was collected and stored when the first patient came in) can potentially help doctors reach a treatment decision more quickly (potentially saving costs), and over time can allow them to analyze how certain treatments will work on other diabetic patients.
Karen Handmaker, MPP, Vice President, Population Health Strategies, Phytel
Most analytics solutions use medical, ancillary, lab and pharmacy claims as the primary sources of data for the reports and tools they offer health systems. Claims data provide utilization, cost and diagnosis information that can be several months old by the time they are adjudicated by payers and available in reports. Clinical data, on the other hand, has the twin advantages of being current and more specific.
For example, claims data can tell us that a patient had an office visit associated with a diabetes diagnosis and an HbA1c test, but only clinical data in the health system’s electronic medical record (EMR) can tell us that this patient’s HbA1c result was 10.2. With this information “teed up” for the doctor or care manager, they can intervene readily to help the patient improve. Therefore, “leveraging clinical data” means incorporating current and pertinent information from the EMR to make the patient’s profile more “actionable” for care teams charged with “driving quality care.”
Going further, analytics that incorporate clinical data offer health systems longitudinal and comparative performance reports on key quality measures they are required to track across their entire organization as well as by practice and physician. These reports give providers and care teams valuable information on an ongoing basis about how they are doing on key quality indicators across the board and for each individual patient, enabling them to implement improvement strategies readily.
HMT: Are healthcare organizations collecting too much data to make meaningful decisions, thereby generating waste? Why?
Todd Rothenhaus, Chief Medical Information Officer, athenahealth
Medical leadership and technology leadership should be listening to the CFO. Getting into the data and analyzing risk should never be at the expense of the bottom line. Leadership needs to always be asking itself, “What business problems are we trying to solve, what clinical goals are we trying to achieve, and are they aligned?” If clinical goals and the bottom line aren’t aligned, risk analysis falls off the ROI chain.
I deeply believe that if an organization is actively managing patients in a risk-based contract that the next dollar they spend should not be on anything but understanding the total cost and the total picture of the population health through the claims-based analysis. If they don’t have that, they can’t succeed. No amount of other technology will help them succeed. I believe that the addition of clinical data to the claims data assets is an essential next step, but I think that sucking data out of antiquated EHR systems that are under the desk of a doctor’s office or in a data center is a really nasty business. I am hopeful that there will be some standards-based integration solutions that will help them do this without breaking the bank.
We at athenahealth, of course, support standards-based interfaces for free, but asking us to suck data out of the back end of another EHR isn’t necessarily a best use of everybody’s money. So the first dollar must go to claims, the second dollar to clinical data. For most providers, regular data – not big data – is enough. A lot of high-investment tech solutions aren’t profitable. It’s a sad fact that many organizations who try to tackle predictive analytics run into cash problems.
Anil Jain, M.D., FACP, Senior Vice President & Chief Medical Information Officer, Explorys Inc.
Most healthcare systems vary in the amount of data they collect. With storage and computing costs being relatively low, the cost of collecting data is mainly labor either through abstraction or by providers at the point of care in electronic health records. In addition, health systems are dealing with a host of new data sources including imaging data, genomics, telemetry, billing and claims data, as well as smart devices. What we see with our clients at Explorys is that although most health systems collect quite a bit of data, the data is often not standardized, not collected consistently, not aligned to strategic imperatives or not analyzed due to a lack of data science skills.
Data without a strategy is just data. At best, it is wasted effort. At worst, it will lead to poor decisions. In a well-designed, sustainable analytics strategy, collecting data is never the end goal – it is simply the raw material used to build the final product: actionable information to drive the best decisions.
Dan Riskin, M.D., CEO, Health Fidelity
Healthcare organizations are collecting the wrong data. That’s why the data we have currently is so hard to use to make meaningful decisions. One more template that supports the doubling of the documentation volume for a patient encounter does not actually improve care. Healthcare organizations should focus on getting the highest quality content in the record, whether through typing, template or dictation. Next, find ways to use the high-quality content to improve care. High-quality data combined with well-designed technology can provide answers and opportunities. Doctors know what’s important – that high-quality clinical data is most valuable in a high-powered, well-designed analytics system.
Eric Mueller, Director, Product Management, Lumeris
There’s really no such thing as too much data. I think people may get hung up on that concept. If the right tools, information and incentives are not provided, there is no motivation or way to use data, no matter the amount.
Incentives are not enough. Tools are not enough. A massive amount of information is not enough. It’s the combination of the three that allow data to do what everyone wants it to do – help make better-informed decisions that will improve quality and lower cost.
In terms of technology, the key is making data usable to those that it impacts most. For example, reports need to reach decision makers immediately so they can then be shared with physicians. Once this happens, healthcare organizations can achieve the Triple Aim Plus One: better health outcomes, lower costs and improved patients plus physician satisfaction.
Patricia Birch, Vice President and Healthcare Consulting Practice Head, Cognizant
Healthcare organizations across the value chain do collect hundreds of thousands of diverse data elements, both in structured and unstructured format, from across the healthcare ecosystem. The structured data elements are collected in the form of demographic and historical medical data for risk assessment (claims and encounters), clinical data as part of medical care delivery through EHRs (lab results, radiology, EHR data), and lifestyle and behavioral data (remote patient monitoring devices, sensors, personal health records, etc.). Unstructured data is collected from consumer databases, social networking sites, doctors’ notes, customer care logs, etc.
Whether the large amount of data collected per consumer is a waste or not is determined by how this data is used and what information is derived from the available data. Data collection practices need to be revised on a continuous basis to assess the usefulness of certain data types and sources. Most stakeholders are challenged by data gaps, insufficient data sources and data quality gaps, in part due to challenges around security and privacy, multiple data standards and inconsistency in data collection practices. This situation, though challenging, can be turned around to develop consumer and patient insights by embracing analytics, ushering in a culture of collaboration among healthcare players to contextualize the insights generated with the societal behavior and lifestyle pattern of the consumer.
Bonnie Cassidy, Senior Director of HIM Innovation, Nuance
It all comes down to collecting the right data for the right reason, at the right time. Regardless of the type of decision making, the rule of thumb is to only collect the data that you are going to use that meets your needs. So in healthcare, that means carefully culling only information that is relevant to your organization and patient population. This requires strategic planning and goal identification.
In order to improve healthcare quality, we need to be able to measure it. The U.S. is a leader in developing reliable measures of healthcare quality, so we need to be collecting the right data. To this point, Dr. John Halamka, reflecting on the nation’s financial issues, has noted:
“Given the recent challenges of Lehman, Merrill, AIG, Washington Mutual, and others, you wonder just how effective the IT systems of these companies have been.
“Of course they had great transactional systems, disaster recovery, infrastructure and data warehouses. However, did they have the business intelligence tools and dashboards that could have alerted decision makers about the looming collapse of the industry? Did the financial services industry have controls, risk analysis, or a memory of previous crisis – the Depression, the Japanese banking crisis, Enron/Worldcom? Was it greed, irrational expectations or too much data and not enough information that brought down these great institutions?”
To learn from Dr. Halamka’s wisdom, we must challenge the healthcare industry to collect the right amount of information in an effort to meet the needs of everything from quality measures to research and more. [Editor’s Note: John Halamka, M.D., is Chief Information Officer of Beth Israel Deaconess Medical Center, Chairman of the New England Healthcare Exchange Network (NEHEN), Co-Chair of the HIT Standards Committee, a full professor at Harvard Medical School, a practicing emergency physician and a noteworthy author and blogger.
Tony Jones, M.D., CMO, Philips Healthcare’s Patient Care and Clinical Informatics’ Business Unit
Almost no information can be hurtful to have on hand, but the storage of data and information can certainly be challenging. Aside from clinical information about a patient’s symptoms, vital signs and the treatments and procedures they undergo, think about what would happen if we introduced gene sequencing into the picture. Today, gene sequencing is used primarily to determine the course of treatment for cancer patients. As we reach an inflection point in the cost of gene sequencing, this data will be routinely added to a patient’s health record. Imagine the kind of impact this data will have on treating infectious diseases – where hours and even minutes matter. The next time there’s a disease outbreak, we could potentially know the genome of the infectious organism, the susceptibility of the organism to various antibiotic therapies, and determine the correct course of action without wasting precious resources in trial and error.
An important factor in managing the growth of data will be eliminating redundancy to avoid storing the same data more than once. If someone receives an MRI of their knee and that data is stored and available to be shared among appropriate providers (at the patient’s direction, of course), it eliminates the need and the cost of repeating the test and generating yet another image to be stored.
Thankfully, technologies to store data and information are constantly evolving, and healthcare technology vendors are coming up with new, creative and easy-to-use solutions that enable healthcare organizations to access and analyze that data.
Karen Handmaker, MPP, Vice President, Population Health Strategies, Phytel
The question is not so much whether they are collecting too much information but whether they are making that wealth of information easily digestible and actionable by their providers. If they are not, their analytical reports are not being optimized and they are likely missing key functionality that will make the difference between meeting quality and savings targets or not.
