Data governance is a longstanding obligation of the healthcare industry.  And in light of the many Business Intelligence (BI), Electronic Health Record (EHR)/Electronic Medical Records (EMR) projects, data governance has become critical to the data integrity of health information. Most hospitals have multiple information systems that feed into their BI, Decision Support Systems (DSS) and Analytic / Meaningful Use Dashboard and Scorecards solutions. But, without data governance built into the solutions, the ability of these systems to authenticate and properly move data — knowing where it came from and where it needs to go — information usage is less likely to be “successful”.  Data governance helps a health system identify the source of data and store it, if necessary, in the appropriate place. Data governance needs to be done from the beginning, so systems (people, process and information systems) can identify the source of the data, the original author and the rules for the data.
Most assume that our source systems have reasonable and consistent data quality based on some uniform or adopted data dictionaries.  However, we are often taken aback during the build-out of the hospital’s Enterprise Data Warehouse (EDW).  Those that have recently built an EDW will tell you a solid Data Governance structures requires:

  • Consistent, reliable and repeatable information. A central role of data governance is to ensure that metrics are defined consistently across the organization. So when managers or analysts talk about “encounters” or “inpatient admissions,” everyone else knows precisely what they’re talking about. Without clearly documented standards around metrics, decisions may be made around false assumptions.
  • Analytics and reporting issues are most often data governance problems, not information system problems. Many hospitals are quick to blame their tools or technology when there is confusion about the meaning of analytics data or lack of clarity in reports. Typically, the tools and reports have not been configured to clarify what various metrics mean, how they align to specific goals, or where they may vary from data provided by different systems. Whole health systems end up ripping and replacing perfectly good systems before doing the necessary governance work to ensure they work properly.
  • Data governance guides all BI activities. Just as contractors would never build a house without clear blueprints, BI teams need data governance to guide and structure their most important activities. In this sense, data governance informs everything from BI implementation to page tagging to report design. Specifically, it ensures that your data capture mechanisms are set up to capture what you need to capture. It may also outline who is responsible for which BI tasks or data. Perhaps most importantly, data governance can help ensure that there is clear alignment between BI tactics and big-picture strategic goals.
  • It is about integrity. Far too many BI veterans know the nightmares that come with trying to explain or reconcile conflicting data sets to skeptical administrators, and the headaches and second-guessing that occurs when there is less than 100% confidence in data. Truly effective data governance helps eliminate those headaches by clarifying what metrics mean, which are the most important, how internal numbers relate to outside constituencies and why there may be gaps. More clarity means more confidence in decisions. There is also a compliance angle here. With more legal and financial reporting ramifications for the businesses, effective governance models can ensure all of your operations follow relevant privacy policies, data security guidelines and patient information regulations.
  • You have already paid for it. Having a firm grip on how you define “Scorecard” or a specific Key Performance Indicator (KPI) can help you when you’re negotiating or advertising.  Healthcare Financial Management Association (HFMA), for example requires data governance to:
    • Establish a leadership team and define the BI/Data program’s scope
    • Calculate the return using the confidence in data-dependent assumptions metric
    • Identify specific areas of deficiency and create a budget to address these areas

Data governance for healthcare is an ongoing process that ensures that the health system’s EDW and BI initiatives align with business objectives of today and into the future. A good data governance structure enables hospitals to prioritize, validate and manage their ongoing series of projects, enabling them to:

  • Understand and manage
    • strategic and tactical data
    • project ownership from a data perspective
    • priority setting for data projects
  • Define day-to-day activities of creating, using and retiring data
  • Describe how, when and by whom data was
    • received,
    • created,
    • accessed,
    • modified, and/or
    • formatted
  • Determine whether data is fit for its intended use
    • including completeness and business-rule compliance
  • Implement processes to
    • cleanse,
    • transform,
    • integrate, and
    • enrich fresh data across subject areas
  • Address security and privacy compliance across integrated departments
  • Manage master data by examining data assets and relationships that define enterprise operations
  • Provide a logical structure for
    • classifying,
    • organizing, and
    • communicating complex activities involved in making decisions about and taking action on enterprise data.

Implementation of a data governance framework like this requires a careful structure for data usage and ownership.

What Works and What Doesn’t

While most existing guidelines, methods and approaches to data governance can provide a valid starting point for developing a structure, active and on-going participation are a must to address ever “evolving” data. What has been implemented in the name of Data Governance, and the results received, vary dramatically from one organization to another.  But, some common findings are:

What Works

  • Obtain executive sponsorship and involvement early (and often); define roles and responsibilities
  • Empower department managers to make data decisions
  • Aligning “system” goals around local business objectives
  • Understanding there is no “one size fits all” approach to data governance
  • Driving “use case” and or “quick wins” to benefit executives and managers throughout the organization
  • Focusing on improved productivity, improved quality, improved alignment with business strategies and improve patient satisfaction
  • Identifying gaps in realization of strategic objectives
  • Escalating current risks and identifying potential risks early

What Doesn’t Work

  • Poor communications to relevant stakeholders
  • Poor monitoring and control of data (not auditing)
  • Not mediating issue resolution (don’t be shy to escalate as needed)
  • Missing the “fit” between the business strategy, culture, department, business stakeholders & resources and what is implemented to help manage data determines the level and kind of benefits realized.
  • Not integrating data governance for all data projects including
    • Training. Hospitals must understand that it is not just about “analyst” training or “data miners”; physicians and nurses need analytics training to understand how big data and tools add value to healthcare.  Recall: when bad data goes in then bad data is the result.
    • Use tools such as dashboards for clinicians to visualize incoming data. As big data moves toward real-time processing often at the point of care, organizations should strive to update processes and develop capabilities to enable tool use, and focus on real- or near-real time clinical decision support.
    • Don’t scale up, scale out. While some organizations may be tempted to replace older technology with bigger and more powerful solution, today’s trend is to “scale out” (hosted).
    • Close the quality loop. Data analytics teams must work in lockstep with quality improvement teams so that analytics tools and techniques can be integrated into the various quality-improvement methodologies which, together, can provide a framework that drives the front-line and administrative changes necessary for achieving desired improvements to health care outcomes and efficiency.

Data Governance – Challenges

  • Resource Management
    • Complex resource management needs have outgrown current data management process
    • Resource conflict (prioritization of people’s time)
    • Difficult to balance project and core service delivery responsibilities
  • Data Portfolio Management
    • Processes are immature and inconsistent, leading to frequent re-prioritization
    • Indeterminate correlation to business strategies
  • Data Quality Assurance
    • Low Project Management Methodology compliance
    • Inconsistent project delivery
  • Governance and Organization
    • Fragmented or ineffective project governance teams
    • PMO has only dotted line authority over many project manager’s leading programs and projects
  • Project Management Methodology
    • Project processes are one-size-fits all; limited scalability and do not address data governance
    • Project risks not proactively managed
    • Accuracy of project delivery reporting and project health is problematic
  • Missed estimating variance
    • Causes Resources to appear scheduled and at capacity
    • Communicates misleading milestone and delivery dates
    • Prevents additional project work from being added to the current portfolio
  • Inadequate Resource Management System, Portfolio Management, and Data Quality controls
    • No real time mechanism to observe and harvest estimates
    • No mechanism to calculate and drive early project deliveries 

An Healthcare Data Governance Model

To be sure, there are many Data Governance models, but below are the core or essential components. 

A data governance structure should feature selected leaders from the Senior Leadership Team (titles may defer), Oversight Committee, Operations Committee (including Research), and sub-Entity Groups.  The goal is to move data governance closer to appropriate levels of management & oversight and provide scalable governance based upon size and complexity.
Now, consider the various components shown in Figure 1. And notice the “top down” nature of the framework shown. Though information governance encompasses both strategy and execution, it is decidedly business driven. The top box in the framework represents the various business functions involved in information governance, from executive sponsorship to the information governance council to data stewardship. This represents the “classic” definition of information governance, namely, the business-driven policy making and oversight of corporate data.
If that’s the definition of information governance, then the lower part of the diagram, data management, is defined as the tactical execution of information governance decisions. As managers in both the business and IT organizations come to grips with the breadth of information governance, they also realize what’s at stake in terms of launching the effort. And as the early adopters of information governance have learned, kicking off an information governance program is fraught with risks.

Figure 1 Core responsibilities


The most successful data governance Committees[i] are either pieces of a larger, more strategic initiative – in which case, they are already funded, but need to be scoped – or are efforts to fix key data in support of a larger BI objective.  Based on the lessons learned here, focus your Data Governance goal to

  • Solve at least one acknowledged business problem (use case)
    • Have an easily articulated end state
    • Help define often nonexistent delivery processes
    • Be data enabled or data intensive
    • Secure executive support or funding
    • Be delivered in 90 days or less

The point of the small, controlled data project is to establish some major new information governance processes and execute them in a smaller, more nimble way. But the processes themselves should be solid and repeatable. The point isn’t simply to prove the viability of new data governance steps but to position this set of activities – including a process for establishing data ownership and procuring tools – so they can scale to the next project or set of projects.

[i] Developing a Data Governance Model in Health Care, Mary G Reeves, Rita Bowen, February 01, 2013