In Part 2 of the “Cutting through the Hype” series, we discussed the importance, especially in healthcare, of getting the data matched, merged, cleansed, mastered and enriched/tagged with the proper HLI code sets and clinical ontologies. This is the work represented in the first 4 columns (Connect, Move, Fix, and Relate) of the “Data Value Chain” diagram below:
Following “Relate” is “Govern”. For healthcare organizations, data governance is regularly discussed, rarely understood, and hardly attempted. In my experience working for and collaborating with some of the top analytics groups in the country, data governance failures (or non-starts) are the result of two primary causes:
- The columns to the left of “Govern” are not being done with automated, scalable technology
- Lack of clarity – what is data governance, what is its purpose, why is it needed?
We’ve reviewed item one with sufficient detail in previous blog posts. Item number two compels us to define data governance in the context of healthcare to understand the “whats” and the “why”:
Data governance is the ensemble of people, processes and technology that creates a harmonized administration of an organization's data quality (availability, usability, consistency) and ensures data are transformed into meaningful information that is used to effectively achieve the quadruple aim (better outcomes, lower cost, satisfied patients, happy clinicians).
Governance should be the supporting structure dedicated to “solving Data Rich and Information Poor”. If everything in the left two thirds of the data value chain is automated, leveraging self-documenting technology that creates a complete dictionary and data lineage for all transformations, then governance becomes much easier. Governance systems are needed that can monitor, visualize what if scenarios for quality rules, alert, remediate, report, and manage history. If these things are in place or in progress, we can focus on the heart of data governance – people and processes.
Where should you start? Ideally, data governance begins before purchasing and developing technology. However, in my experience, governance usually catches up after purchasing the related technology. Starting at the top, governance will need an executive sponsor. Some organizations know that they need to take steps to become data driven, so the executive sponsor can focus on strategy and organization. Other organizations may not be as centralized in their culture, approach to systemness, or geographic location, and will need an executive sponsor who is more of an influencer who can effectively communicate the value proposition of treating data as an organizational asset and using effective data governance to achieve that end.
The executive sponsor will chair the data governance executive committee which should include leadership over component areas responsible for each aspect of the quadruple aim, as well as information systems leadership. The executive committee is responsible for:
- Developing the strategy and vision of causing data to be transformed into information that can help achieve the quadruple aim
- Organizing the data steward committees
- Making decisions when steward committees escalate to them for executive guidance
How do you organize? Generally speaking, data steward committees are the business and/or clinical owners of source systems and applications plus the technology support person(s) for that source system. For example, one committee over patient access and scheduling, another around systems for revenue cycle management, and yet another over the EHR and closely connected clinical applications such as radiology and lab. The number of committees and scope will correspond to the size and complexity of your organization. I’ve been at institutions that had over a dozen chartered steward committees with six active at any given time.
It is absolutely KEY that the business or clinical owner lead the steward committee (e.g. – Director of Patient Access for the committee over scheduling and registration data, or the VP of this area if many hospitals or clinics are involved). They need to be subject matter experts for those data and have the ability to impact front end data quality, change source system applications (if needed) and make decisions about data field priorities and match/merge/cleanse rules in collaboration with their committee. They should have the expertise to provide definitions for data at the element level as well as give input to definitions for department and enterprise metrics and KPIs.
Sustaining, and day to day guidance – there should be one or two stewards in each committee, and part of their ongoing daily activity is interacting with the data quality systems to monitor, work alerts, remediate issues, manage definitions/history, and analyze “what if” scenarios for data quality.
They should have the latitude to make decisions within their areas of responsibility, presenting an informal summary report in the committee meetings.
They should also have guidelines for bringing decisions to that area’s committee when needed.
Data governance, when formally chartered, treats data as a valued asset, much as HR exists to ensure good treatment of staff resources. Indeed, it may be desirable one day to have a “Department of Data Governance” with full time professionals that serve in administrative and facilitation roles. However, it is critical that daily stewardship never becomes a full time job. Previous blog posts clearly illustrate how healthcare data are more complex than nearly any other industry. It is therefore vital that data stewards forever remain within the business or clinical area as their primary job so that their subject matter expertise does not wane over time!
In my next blog post, I review ROI and leveraging governance to prioritize clinical and business objectives. In the meantime, read our white paper about data governance in healthcare.