Attaining a 360 Degree View of the Patient

By Jack Plotkin, CTO, Virtual Health
Twitter: @VirtualHealth_

Over the last 10 years, the healthcare industry has experienced a major shift in both the prevalence and importance of information technology. Electronic health records have progressed from bleeding edge, to basic requirement, to competitive differentiator. Claims and authorizations have made the leap from the fax machine to the cloud. Patient engagement has transitioned from printed questionnaires to connected devices and wearables. Care team communication has evolved from paper memoranda to secure mobile communication.

The rapid growth in digital solutions has made specialized tasks and records easier and more transparent, but it has also resulted in a highly-fragmented information landscape. Complex healthcare entities such as managed care organizations (MCOs) are operating ponderous IT infrastructures with separate systems for mission-critical activities. These may include data warehousing, medical and pharmacy claims, clinical data, enrollment and benefits information, provider networks, authorizations, care management, quality data, educational content, and member and provider portals, among numerous others. In turn, they interface with health systems and other provider organizations that typically run disparate electronic medical record (EMR), practice management, billing and data management systems.

This siloed digital environment results in partial and overlapping fragments of patient data spread across dozens of landlocked IT systems. As a result, the industry’s pursuit of a singular and complete record of the patient’s journey through the health continuum faces multiple roadblocks, leaving efforts to acquire and embed the most up-to-date patient information in care pathways and analytics short of their potential.

The quest for a fully integrated and seamless 360-degree patient view has never been more important. Transparency regarding a patient’s medical history and recommended treatments enhances decision-making, supports care management, limits avoidable errors, and improves outcomes. Access to timely, detailed information also makes it possible to identify non-compliance and address care gaps. In turn, this supports proactive rather than reactive care, reducing costs and readmissions alongside higher quality of care and better quality of life.

When our team at VirtualHealth set out to tackle this challenge for a multi-state Blues plan, we defined it as, foundationally, a data problem. Akin to a detective looking for evidence, we knew the data was out there, but it had to be found, extracted, and filed. In this vein, we deconstructed the challenge into three basic components: (1) data integration; (2) data normalization; and (3) data binding.

Data Integration. Step one was finding a way to aggregate patient data from myriad disparate sources maintained by the plan and its delegated entities. The term “interoperability” is one of the most often used and most misleading in healthcare IT. Many vendors will create a single mode of data extraction for their systems and then make the marketing claim that they are “interoperable.” Sure, if your system supports a particular vendor’s mode of data extraction, then it may be relatively more straightforward to integrate the system as a data source. But that is a far cry from true plug-and-play interoperability. Because the healthcare IT landscape contains so many legacy systems, and because data standards such as HL7 are neither universally adopted nor identically implemented, the only way to effectively integrate with the multitude of existing healthcare systems and formats in the wild was to create a dedicated health information exchange (HIE).

Key Insight. Our team took a close look at HIEs and tried to understand why so many had failed. There were issues of poor planning and operational mismanagement, but from a technical standpoint, the key problem was a combination of inability and vendor unwillingness to comply with each HIE’s requirements for data integration. Based on this analysis, we realized that the only practical path forward was to ensure our system could both consume and generate data in the preferred format of each of the Blues plan’s other vendors.

On the back of this realization, we built or customized our existing interfaces for every standard type of counterparty extract, including enrollment/eligibility, EMR, claims, authorizations, practice management, pharmacy, lab, vendor management, and provider network management, among others. We ensured support for both standards—such as HL7 and X12—and proprietary formats. Our interfaces were grounded in a modular architecture that followed best practice object-oriented patterns such as SOLID and DRY, which meant that we could rapidly adjust interfaces for format variations and for new systems. This approach made unparalleled integration timelines possible for the Blues plan around connecting state, claims, authorization, and provider management systems and ensured that the challenging regulatory timetable was met.

Data Normalization. Step two was ensuring that all the disparate data being consumed by our system was consistently mapped, indexed, and stored. To start, the data schema had to be sufficiently comprehensive to encompass all major categories of healthcare data and sufficiently flexible to support variability across vendor systems and the plan’s needs. Then, we had to develop an approach to defining data dictionaries, validation criteria, and business rules that could apply across data domains.

Key Insight. Although the structure, consistency, and integrity of a relational database works well for most healthcare data, a lot of critical information is also captured in an unstructured format. To ensure that the system could optimally support all data categories, we designed our data repository to contain the elements of both a data warehouse and a data lake, something we termed as an “adaptive data model”.

Data Binding. While aggregating and indexing cross-domain data about a patient is valuable, it is the binding of that data to workflows and analytics that empowers care teams to fundamentally work smarter. Working with the Blues plan’s clinical operations teams, we built data-driven workflows that iteratively identified the next best action for each patient based on that patient’s unique profile and latest information streaming into the system.

Key Insight. Most population health systems focus on calculating quality metrics and care gaps across the entire patient population. While this is useful in identifying broad trends and areas of improvement, the approach does not provide actionable feedback to the providers, care managers, and service coordinators on the front lines. By calculating these measures in real time at the individual patient level and tying these calculations to tasks and dashboards directly leveraged by care teams, it is possible to ensure more optimal and timely interventions at the point of care.

Our approach resulted in an operationally successful implementation of the VirtualHealth platform in the government programs space for the plan where interoperability, scalability, and flexibility are requisite for successful operation between the rock of government regulations and the hard place of very challenging and complex populations. The key lesson learned was that the 360-degree patient view is a key driver for success in the value-based world.