Strategic Development of Enterprise Analytics for Healthcare Delivery Systems

By By Joe Kimura, Deputy Chief Medical Officer, Atrius Health

By Joe Kimura, Deputy Chief Medical Officer, Atrius Health

Healthcare analytics is increasingly recognized as a critical core competence for accountable-care delivery systems. Analytics drive efficient and effective delivery system innovation and process improvement. Predictive and prescriptive analytics serve as the engine of proactive, prioritized population management activities. But perhaps more fundamentally, timely and reliable data are the lifeblood of active organizational learning and essential for meaningful evaluation of Microsystems and organizational innovations. Delivery systems limited by rudimentary analytic capabilities will be at a competitive disadvantage in the race to produce, deliver, and demonstrate superior value to the market.

As executives explore options to strategically augment internal capabilities, it may be helpful to always examine the complete organizational data value chain and to assess barriers and opportunities to create the maximum amount of organizational value from a specific investment in analytics. From an investment in analytic data warehousing infrastructure to new advanced statistical methods to new data visualization software, a narrow examination (i.e. the immediate proximal and distal step in the value chain) may underestimate the barriers and overestimate the organizational value.

The primary mission for most accountable care delivery systems is to provide outstanding and efficient patient care. Hence, the core value from data analytics stems not only from the ability to capture and transform raw data into actionable information, but from the actual use of that information to make better and faster organizational and clinical decisions. While value is clearly created throughout the chain, the most direct value is created and captured through actions taken on the analytic output.

A quick inquiry of the data value chain can provide an estimate of the likelihood of creating value from an investment in analytics. A simple summary of the data value chain links five major activities: Capture, Integration, Analysis, Reporting, and Use. These five steps are the technical steps that support the effective translation of raw data into action and organizational learning. While each activity should add value, lack of strategic development across the chain can lead to upstream or downstream gaps that can stymie overall value creation. As such, the entire value chain should be collectively assessed.

Working from the end, here is a quick description of activities in the value chain.

Data Use: The adoption of analytics can be considered a decision making innovation. As with any innovation, it helps if the decision-making problem to be addressed with analytics is clearly defined. A concise statement of the target business problem will crystalize the expected created value. It also helps identify the target user and the ability of this user to reliably, effectively, and efficiently use analytics to address and learn from the decision-making problem. Careful exploration of data use is often overlooked as people focus on the technical elements, but it can be highly informative. Busy frontline clinical users may not have the time, skills, or incentives to leverage robust self-service business intelligence tools with sophisticated embedded advanced analytics.

Data Reporting: Next is to identify the best way to get the right information to the right user at the right time and in the right way. A firm understanding of expected data use is a prerequisite. The most technically elegant approach to reporting information may not be the best way to get end users to use the information. Even if users recognize the inherent value in the information, if barriers to access and use are too high, then analytic reports will likely be seldom used. Report format (visual dashboards vs tables), timing (patient data at point of care vs patient data one week later), and access (report requires flipping into a new software program) are elements to consider.

Data Analysis: Data analysis usually garners the most attention in the data value chain. Is there the ability to understand the problem and select the right analytic methods and tools to use? Then, are the required technical and programming skills to properly execute the analysis present? As more advanced methods are leveraged in healthcare, valid and responsible application of those methods are critical for creating value. Given the complexity of methods, inappropriate analysis is frequently difficult for end users to detect and can lead to inappropriate care to patients.

Data Integration: As disparate data sources are brought together for analytics, the organization, cleaning, and staging of data can impact the ability of downstream analytics to leverage those data sources. Inconsistent data standardization and inconsistent business definitions can result in the inadvertent use of heterogeneous business concepts in analytics. Such imprecision can constrain the ability for analytics to detect signal from the noise. Effective master data management and governance systems can thus help prevent analytics from becoming an exercise of garbage-in, garbage-out.

Data Capture/Creation – For any analytic investment to yield meaningful results, the right data must be captured and/or created. Healthcare analytics is awash with secondary use of data with the ongoing use of claims data to assess clinical quality as one glaring example. However, on a practical level, most delivery systems must try to leverage existing and available data sources as much as possible. The critical analysis here is to understand if the right data are captured to support the maximum amount of value from the end use of the data.

In summary, once the core business problem is defined, the following set of questions can be asked to explore the data value chain:

1) Data Capture/Creation - Do we have the right data?

2) Data Integration - Do we have the data ready for analysis?

3) Data Analysis - Do we have the ability to analyze the data?

4) Data Reporting - Do we have the ability to get the analytic output to the target users?

5) Data Use - Do the end users have the ability to use the analytics to make better decisions?

Through this sort of broad inquiry, executives can begin to assess the likelihood that a proposed improvement in healthcare analytics will produce the expected value to the organization. It widens the lens through which to assess analytics and provides a way to map relevant organizational assets and capabilities around analytics. Such understanding may help to 1) develop a strategic plan for developing organizational analytic capabilities, 2) assess the value of third party tools & technologies, 3) diagnose issues contributing to suboptimal value from existing investments in healthcare analytics, and 4) look for opportunities to optimize value capture, including potential commercialization opportunities.