In the pages of Health Management Technology, experts frequently speak of the heaps data being gathered by apps and providers alike – data that will supposedly change the face of healthcare forever by allowing for the identification of vital population trends, boosting revenue by streamlining care. But, as so-called “big data” rises, navigating through the sea of information is easier said than done.
HMT asked a panel of experts from the Berkeley Research Group (BRG) for their take on patient data analytics and the strategies providers can adopt to leverage big data as a means to better know their members, reduce costs, and improve services across the board.
Editor’s Note: The following has been lightly edited for the purpose of concision and clarity.
Q. Gathering and analyzing consumer data is typically associated with retail and service companies. As the healthcare industry moves away from fee-for-service and into the realm of value-based models, isn’t treating patients like traditional customers counter intuitive?
Healthcare has never fit the typical retail consumer model. Patients generally seek care when they do not feel well and need access to care as a necessity – although modern medicine has many exceptions.
Additionally, physicians tell the patient what care is needed – so the patient needs a professional to both assess what is needed and provide necessary care. Finally, a large majority of patients do not bear the full cost because insurance takes care of most of the cost of treatment, with the patient paying deductibles and copayments that may cover only a fraction of the care. With that understanding, however, providers have been able to market directly to consumers; direct-to-consumer advertising for drugs is a prime example.
Beyond the retail marketing use, however, data is crucial to improving patient care and reducing cost under value-based models. Having data across the healthcare spectrum to understand where patients currently receive their care and where well-managed patients should receive their care is key to understanding how care can be better managed, particularly for patients with chronic conditions. Data is essential to reducing preventable emergency room visits, hospital admissions, and readmissions. Further, value-based payments will require providers to see opportunities for streamlined care delivery, as in the bundling initiatives under CMS. Data is necessary to reduce the cost of care delivery while maintaining quality of care.
Q. If you’re a smaller, independent practice, does buying outside consumer data help you to better understand your own members, or does that simply muddy the waters?
While more information is generally a good thing, a small practice has to weigh benefits against the cost of information. Outside data is expensive. The data set is then likely to require analysis with statistical programs and interpretation based on unique characteristics of the data set. If the data can be used to develop performance benchmarks or standards against which the practice can measure itself, a small practice may see large differences between the benchmarks – higher or lower – because of its size. For small practices, outside consumer data can muddy the waters without a clear plan for the analysis, a specific purpose to be achieved, and clarity around the limitations of what the analysis may yield.
Q. With the passage of the Affordable Care Act, Medicare and Medicaid patients often find themselves on managed care plans. Their choice in terms of what healthcare practice to patronize is limited. Is analyzing patient data as necessary in a situation where patients choose one organization over another out of pure necessity?
From a public policy perspective, analyzing the data and understanding the results is more important than ever to ensure that the managed care approach delivers improved care coordination, appropriate access to care, and the cost savings that should occur from those activities. If patients retain the choice of a fee-for-service model, sorting and selection of patients may result in different risks in managed care organization patients than in fee-for-service patients. Specifically, if high-risk patients stay in fee-for-service while low-risk patients choose managed care, the overall cost to the public may increase.
Even if patients are limited to a managed care option, data to understand care access, utilization, and outcomes are necessary to determine the impact of managed care and assess the quality of care provided to patients.
Q: Retail companies have all sorts of indirect methods to gather customer data – loyalty cards, surveys with rewards, etc. Beyond what’s put into an EMR, what can healthcare organizations do to gather information directly from their members more effectively?
Healthcare organizations can capture patient-generated health data for use through emerging approaches that leverage both existing investments and new technologies.
Many health systems engage with patients and collect additional information with existing investments in patient portal technology. Patient portals now integrate with medical devices, such as glucometers and consumer fitness trackers, to provide access to near real-time health data. They can also monitor post-acute care patients as they recover at home. Hospitals can reduce readmissions and improve post-acute care surveillance and coordination with questionnaires designed to identify unexpected complications in the recovery process, securely distributed to patients through a portal’s notifications system.
Many well-funded startups in the rapidly growing digital health space offer new opportunities to capture patient-generated health data. They are piloting platforms designed to improve remote patient-monitoring capabilities, such as mobile apps that engage with chronically ill patients between visits through app notifications and text-message alerts that provide medication and appointment reminders, as well as daily disease-monitoring questionnaires.
Q: For those organizations moving away from paper, is there a reasonable strategy for compiling all of those filing cabinets into consumable digital information?
When organizations transition from paper to an electronic record with the goal of optimizing consumable data, they should first examine and analyze the processes and existing paper forms used to capture documentation. An optimized strategy will then recreate as many of these forms as possible as electronic versions that use standardized data fields. This will provide the facility with consumable data to be used in conjunction with a data repository or warehouse solution.
For existing paper records, organizations can best accomplish the migration of historical patient data via a scanning and archiving solution that allows batch scanning of the paper forms into the electronic record. An organization with this capability can use optical character recognition technology, which provides the ability to search these scanned forms for discreet data.
Q: With every action leaving a digital trail, gathering data seems like the easy part. How can an organization ensure that the information they’re receiving is reliable?
Obtaining reliable clinical data is an age-old problem for healthcare organizations. Inconsistent processes between those that write, type, and enter this data can erode its reliability. Some may perceive the adoption of electronic health records as worsening this existing problem because of the increased visibility it brings. In reality, the EHR can ensure the integrity and accuracy of clinical information.
The EHR allows for structured documentation templates for healthcare providers that can help capture data in a standardized format with embedded alerts. For example, allergies, medications, and problems can be captured in a standard format linked to standard nomenclature for easy data exchange.
Leveraging the EHR and associated technology can reduce data-entry errors by acquiring patient data directly from bedside hemodynamic monitors. Further, patient-care errors may be reduced with techniques such as bedside electronic medication administration and verification that uses barcoding technology.
Q. Are analytics software solutions limiting in terms of what type of actionable data they can produce?
Analytic software solutions are only as good as the data entered into the product; as such, these solutions are not the limiting factor. It is up to providers, analysts, and administration to ensure consistency of processes and methods for collecting the data, and ultimately to take that data and make decisions about what actions should be taken. The data can help prioritize projects, but there must be an interpretation as to the actions to take. The people and processes take precedence in actionable data over analytics software solutions.
Q. After the data is analyzed and a conclusion is made, who turns it into actionable change? Is it the providers? The administrators? In a hospital, there often isn’t a traditional hierarchy when it comes to clinicians that work by contract.
All of the above and more! The adage “It takes a village” applies. Actionable, sustainable change requires the full support of administration. Providers also must lead these changes, especially if they are clinically related. Others, such as nursing, pharmacy, information technology, finance, and other clinical ancillary services, will be part of the team implementing the changes. It is up to each organization to design processes that account for this team effort and involve all departments in delivering systemwide changes to the benefit of the patient, organization, and individual teams involved.
Q. What about population separation?
There are several ways to identify patient populations and, ultimately, trends within them. A basic segmentation can be by payer, such as Medicare, Medicaid, or “dual eligibles.” Patients can be viewed as segments, such as pediatric or geriatric patients, or even by disease state. Ultimately, with consistent data, hospitals may segment patient populations with high utilization rates, readmissions, emergency department visits, and observation visits. The idea is to have complete, accurate data across segments, not allowing an artificially designed segment to dictate how data is gathered or used.
Making data personal
Healthcare leaders sound off on how big data can be used to help
providers see patients as people.
Lee Rivas, CEO, Public Sector and Healthcare, LexisNexis Risk Solutions
Clinical analytics provide us with a lens through which we can understand health risks, motivation, and specific care-planning needs at an individual level. The ability to drill down to the individual level is critical to understanding what resources to deploy, when to deploy them, and how to deploy them. Analytics that take into account unique population characteristics are superior to those that simply group populations together based on broad categories like age, condition, etc. This is why tapping into non-traditional sources of data is necessary to move the needle in healthcare.
Imagine one of many single parents who has just gone through a divorce and had to move to a new neighborhood that has high crime rates. Up until then, they were healthy – and from a clinical perspective, there is nothing to indicate otherwise – but living in a less safe place than they did before, needing to find a job after caring for a child at home, and dealing with emotional effects of divorce all make them susceptible to serious stress that can set off a series of adverse health challenges. It is this socio-determinant data that affords a unique view into an individual’s clinical risks and can allow either the plan or provider to be proactive in getting the patient and her child the care they need.
This type of insight is equally critical for diagnosing and treating low or nonusers of the healthcare system. In this case, a plan or provider has scarce clinical information and cannot adequately predict the individual’s health risks and would need to rely on a picture presented by the socioeconomic data.
Q: Presumably, data warehouses have a lot more information about patients than just their medical history. At what point in time does data collection on patients become a violation of privacy – or a HIPAA violation?
Regulation exists – such as HIPAA, CMS guidelines, parts of the Accountable Care Act – so that data collection doesn’t violate privacy. HIPAA mandates aggregation and de-identification of data and prohibits re-linking data that would lead to identification of an individual, especially in reference to analytics. Even clinical analytics models themselves can and should be HIPAA certified.
Whether through a member portal or EHR, once a payer or provider has access to analyzed information they are well positioned to help their patients and society improve clinical outcomes, such as addressing why 20 percent of the population do not come back for follow-up visits nor fill their prescriptions; enhancing participation in wellness management programs; reducing avoidable costs by identifying actions or care plans that can lower the chance of complications or the continuation of adverse outcomes; and minimizing fraud and abuse – for example, by identifying an orthopedic surgeon overprescribing Xanax or another controlled substance.
Saeed Aminzadeh, CEO,
Decision Point Healthcare Solutions
Segmentation is part science and part art. The science part has to do with ensuring that there are enough patients in the segment for the segment to be meaningful, and to segment patients based on logical categories, such as age and type of insurance. The more nuanced segmentation is in trying to segment members based on their “barriers to engagement” in order for the healthcare organization to be able to be more focused and personalized in communicating with the patient. For example, segmenting by health literacy enables the healthcare organization to divert healthcare educational resources to patients who need it the most.
Q: How do you ensure the conclusion an organization comes to after data is analyzed is able to be fine-tuned on demand and continuously updated as new needs arise and new information is gathered?
Effective data analytics is a continuous learning experience, enabling healthcare organizations to get smarter and more targeted over time. At a micro level, it is important to understand whether the organization was effective in achieving its clinical and business goals, and in what areas and across which populations it was most effective. This type of understanding is a key part of any analytics platform. This way the organization can continue to deploy the most effective strategies, while modifying the less effective strategies in order to promote continuous improvement. At a macro level, it is also important to understand the relative changes in the population in order to ensure that the services offered are in line with the population’s needs.
