HIT Happenings: Patient-Generated Health Data — Managing the Inflow
By Karly Rowe
For The Record
Vol. 33 No. 5 P. 30
Health care providers face an ever-growing mountain of data, much of which is being driven by remote monitoring and other tools and apps that allow patients to collect and record their own data, known as patient-generated health data (PGHD). These circumstances are challenging systems to better manage patient identity processes that were already under stress.
Fortunately, there are steps providers can take to lessen the impact of this surge and improve their performance (both health outcomes and bottom line) while positioning themselves for more innovation down the road.
A Huge Challenge
Calling the last few years of growth in health data “explosive” risks being an understatement. One estimate published at the start of the pandemic suggests there has been an 878% increase in data since 2016.
Given the diversity in the types of data being produced, qualifying this volume increase is a more recent challenge. For instance, nearly one-half of hospitals surveyed in this research expect that remote patient monitoring rates, which spiked over the past year, will match or exceed inpatient data collection within five years. Use of health wearables, which also surged during the pandemic, are expected to grow by 16% annually between now and 2026.
In addition to this generation of new patient data, there’s an emerging consensus that social determinants of health can yield meaningful insights into patient health outcomes. While the Centers for Disease Control and Prevention’s Healthy People 2020 project identifies 42 topics that correlate with health journeys and outcomes provides some visibility into anonymized data, there remain vast reams of data that could be associated with patients that exist outside traditional HIT systems. And they grow with every passing day.
Another component to this challenge is the emergence of patient expectations. Today’s health care consumers want greater visibility and agency over all of their data, including PGHD. Further complicating matters are regulations governing data sharing and privacy protections. Not only has Health and Human Services issued new rules but there are also evolving needs and requirements (such as those included in the Cures Act) for deciding which data—and in what form (raw vs processed, for instance)—can be accessed by patients, providers, and payers.
Not All Data Are Created Equal
Because their content is inexorably tied to their format, not all data are equal, technically speaking. Data have a structure—literally, the pattern for how the digits 1 and 0 are presented—that make them readable and thereby combinable. In the IT world, the term for data that aren’t structured to meet the expectations of a system are called “unstructured.” Trying to use them is like trying to make a piece from one puzzle fit into another puzzle; they’re incompatible, irrespective of what they’re made from.
Because traditional health care systems weren’t designed to accommodate patients’ active participation in the management of their health, much of the PGHD inflow is unstructured. Lifestyle insights, biometrics, reports of symptoms, and other data collected outside a hospital’s or clinic’s four walls don’t conform to the established processes let alone the technical structures in which clinical data are captured and stored. Data on social determinants are also formatted outside of clinical conventions and norms, as well as collected at touchpoints outside the reach of traditional HIT systems.
An additional layer of complexity arises from the challenge of matching data to the right person. While there’s no dearth of data being created and stored, there’s also no reliable causal connection between more data and better outcomes. In fact, the growing volume of health data can create risks of their own which are characterized primarily as “filter failure” or the inability to parse data for specific purposes. This makes the linking process crucial, since it underlies the subsequent efficacy of any efforts to prioritize data used for description, diagnosis, predictions, and/or prescriptions.
You can’t prioritize unless you’re sure you “know” your patient.
The Three Deadly Sins of PGHD
To avoid falling victim to this avalanche of data, health care organizations should keep in mind the following points.
Prioritize what you want to use: The volume and diversity of PGHD is truly an embarrassment of riches and can quickly add up to significant time and resource challenges for overwhelmed health care providers. It’s important to identify which data can be most efficiently appended to patient files, qualified by a direct relationship to improving health outcomes.
“More” is less important than “necessary,” which means that prioritizing your approach to these data is as important as using them. There’s a concept in new product development called “minimum viable product,” which is the simplest working version of a new idea. The concept offers intriguing ways to think about prioritizing the use of PGHD.
Integration is far more important than addition: The availability of PGHD and social determinant data does not directly translate into usability. Many nascent data management tools (such as consumer-level tablet computers to collect and explore patient data in clinics) often incorporate data points in “notes” fields or formats such as image attachments that can’t be searched. The same challenges exist for video records, which also are unsearchable.
The key to harnessing the value of this or any other nontraditional health data is to incorporate it into your processes and systems. This starts by deciding what PGHD you want to use for what activity—description, diagnosis, predictions, and/or prescriptions—and then keying the data to those purposes. Ensuring that the processes and systems are tuned to accept and store data in a way that makes it searchable and usable is a critical foundational step.
While data sources and structure differ, the knowledge of the integration process makes each future integration easier.
Don’t rely on single sources of information: There are variables inherent in the ways PGHD are gathered, including patient state and the quality or reliability of their technology, that can make it unreliable, especially when compared with the data collected by controlled clinical testing. Therefore, it’s important to put these data into context with clinical data as well as other sources of information, such as social determinants of health provided by third parties, and weigh data appropriately.
In the broader data world, data are considered more trustworthy if they’re unfiltered or interpreted—closer to the “original source compiler”—which means they’ve passed through fewer hands and therefore have a lower probability of error or misuse. Or, if they are removed from the source, the adaptations have been made by reliable providers.
The volume of PGHD is only going to increase over time, as will the types, depth, and nuance it presents. With this continued growth will come the capacity for making earlier and more reliable diagnoses, as well as better management of treatment (both by providers and patients themselves). Much of these data will remain unstructured, and for good reason: They will continue to come from new and often unrelated sources, which is why correlating them can be so powerfully useful. Patient condition will need to be understood in an ever-evolving detail of context.
The challenge for the health care industry is to embrace this transformation, often labeled as “digitization” (translating paper or physical world activities into digital formats) and “digitalization” (creating new processes to accomplish tasks). The good news is that health care isn’t the first industry to face this new world—it’s been underway for years in the financial services industry, with nearly all firms surveyed in a 2019 research report citing varying levels of activity on this front.
This means that there are both tested models and proven resources available to help address the PGHD explosion. Data management strategies and insights from outside of the four walls of a health care entity can come precertified for specific outcomes and prechecked for glitches. It also means that the health care response to PGHD can not only utilize the data for better patient outcomes but also reduce health and financial risks. Risk reduction is a meaningful outcome of the digitization seen in industries such as financial services.
The phenomenon of more data, and the greater expectations not just to use them but also share them and give patients more control over their courses of treatment, is part of a far larger transformation underway. The corresponding challenges weren’t invented in health care. The solutions won’t solely originate there, either.
— Karly Rowe is vice president of patient access, identity, and care at Experian Health.