Demystifying Healthcare Data Governance: An Executive Report

Phase 2 involves knocking down technical and cultural barriers that stand in the way of access to data. Typically, the most visible sign of this phase is the establishment of an EDW. While the data warehouse is being built or procured, the data governance committee can work in parallel on the establishment of the principles, policies, and procedures for gaining access to the EDW and maximizing its value to the organization. Key to deriving that value is the establishment of a data stewardship program.

In Phase 3, the data governance committee identifies data stewards who can function as domain experts in specific data content areas of the organization. Data stewards are typically from the director or manager level of the organization, and have leadership responsibility for staff who originate, collect, and enter data into the organization’s primary information systems such as registration, the EMR, laboratory information system, and revenue cycle system. In the early lifecycle of a healthcare EDW, approximately eight to 10 data stewards will play a critical role, and thus should be identified early.

Shown in the chart is an excerpt from a data steward policy written for a large academic medical center. It identifies the key source systems and data stewards required in the initial development of the organization’s data warehouse. Over time, as the EDW matures and the data content grows, additional source data information systems and data stewards will be assigned. It is not unusual for a mature data warehouse environment, one that has been in operation for more than five years, to be associated with up to 40 different source information systems and data stewards. For this particular academic medical center, its EDW became operational in 2006. As of October 2014, there are over 100 different source data systems feeding their EDW.

Source Data Information System Responsible Data Steward
Patient Identifiable Data/PHI Director of Medical Records
Patient Registration Director of Ambulatory Operations
Patient Scheduling Director of Ambulatory Operations
Electronic Medical Record – Physicians Chief Medical Information Officer
Electronic Medical Record – Nurses Chief Nursing Informatics Officer
General Ledger CFO
Accounts Payable Director of Finance
Payroll Director of Payroll
Billing and Accounts Receivable Director of Business Operations
Ancillary Departmental Systems Affiliated Department Administrator

The previous phases established cultural tone for data; defined the processes and technology to enable greater access to data; and established data stewards that could support data quality issues and serve as data domain experts in their areas. Phase 4 shifts to data quality. All of the cultural themes, people, and technological tools have been established to address the inevitable data quality concerns that will arise as a consequence a greater access and visibility to data.

In Phase 5, the data governance committee can shift its attention to the full exploitation and utilization of the data and the data management processes that surround it. The organization’s board and executive teams will now have access to data and processes that were not available previously. Developing internal management dashboards will be less labor-intensive and more accurate. External reporting to government and professional organizations will also improve and become more consistent and less error-prone. Clinical and financial process improvement initiatives will begin to thrive. Contract negotiations with payers will become more data-driven and transparent. Opportunities to identify wasteful care processes and supply chain management issues will emerge. Research grants will be easier to submit and complete, and high-quality researchers will be easier to recruit, as they are attracted to the availability of high-quality data.

Finally, in Phase 6 of the data governance committee’s evolution, members will be operating at a very strategic level with the deliberate development of roadmaps to acquire the data necessary to achieve accountable care and population health in the truest sense. In the future of healthcare, mergers, acquisitions, and partnerships will be characterized as much by the acquisition of critical data as by people and facilities. The geographic shift from delivering care in hospitals and clinics, to people’s homes and workplaces, will also move healthcare toward becoming a digital knowledge delivery industry. Data governance committees will play a critical role in the procurement of data and information systems that add content to the data warehouse and facilitate new forms of healthcare. The committee members will play a part in the acquisition of activity-based cost accounting systems, genomic information systems, patient-reported outcomes systems, 24 x 7 biometric monitoring systems for the home, and patient social interaction systems— all motivated by the value of data that they contribute to the analytics of the organization.

Healthcare Data Governance

What Is the Data Governance Committee Governing?

Shown in the graphic are 16 different categories of data that require some form of governance. There are undoubtedly more in today’s healthcare environment and more are coming. But at the present time, these are the categories of data that most organizations should be concerned about acquiring, managing, and integrating into the EDW. Organizations should address the first six of these data types immediately. These six core data sources are: billing data, lab data, imaging data, inpatient EMR data, outpatient EMR data, and claims data.

The next six, data types 7 through 12, should be on every organization’s strategic data acquisition roadmap for acquisition within the next one to two years. These include health information exchange (HIE) data, bedside monitoring data, external pharmacy data, familial data, and home monitoring data.

Over the next two to four years, the data governance committee should sponsor the acquisition of patient-reported outcomes data, long-term care facility data, genomics data, and 24 x 7 biometric monitoring data.

A significant amount of money and time have been invested in our country’s acquisition of EMRs and HIEs. Many healthcare executives exhibit a sense of achievement, and in some cases, fatigue, as a consequence of these acquisitions and projects. However, EMRs and HIEs are just the beginning of healthcare’s digital transformation. Now is not the time to celebrate false summits or plan for a break in activity; the real climb and summits are still ahead.

Healthcare Data Governance

Master Data Management

Master data management is comprised of processes, governance, policies, standards, and tools that consistently define and manage the critical data of an organization to provide a single point of reference. Tangibly, that means codes such as ICDs, CPTs, SNOMED, LOINC, RxNorm, facility and department codes, master patient identifiers, cost accounting codes, and master clinician identifiers are standardized and sequestered in a data warehouse where they can be referenced and utilized.

There are no organizations in U.S. healthcare that have a perfect adherence to local, regional, national, or international master reference standards throughout the organization. Master data management is a long-term journey in healthcare, with no end. Adherence to master reference data can seem overwhelming, and it will become so if not approached carefully, in small and manageable objectives. Establishment of master patient identifiers and master physician identifiers are the two most critical forms of master data management. Without those two identifiers, healthcare analytics is left to very broad, population-based use cases, not specific clinical outcomes transformation. The transition to ICD-10 and annual updates to CPT codes are other examples of master data management. Mapping an organization’s SNOMED, LOINC, and RxNorm data for submission to federal agencies such as CMS and NLM is also a form of master data management.

The data governance committee’s role in master data management is, once again, at the executive level. Rarely will the committee members engage in the details of master data management, unless it requires resolution of a priority or dispute. The data governance committee will endorse and support the concept of a master data management function and a master reference area to support that function, in the EDW, but they will leave the details of implementation up to a subcommittee and the data stewards who are most affected by master data. CIOs, chief data officers, and chief analytics officers can play a very hands-on role in the development and implementation of a master data management program.

While standard codes and vocabularies are critical to a master data management strategy, they also might be the easiest to manage. The more complicated aspect of master data management is related to the standardization and management of the algorithms and rules that bind data together.

Data Binding and Data Governance

Data binding goes hand-in-hand with good data governance. For example, two sets of data composed of the numbers 115 and 60, by themselves, are meaningless pieces of data. They are simply numbers. The numbers themselves do not mean anything until they are “bound” through analytic software, to a vocabulary and the clinical or business rules about that data. In the example, as seen in the diagram below, the values 115 and 60 are bound to the vocabulary “systolic” and “diastolic” and “blood pressure.” With that vocabulary binding, there is meaningful context for the numbers; now the context must be further bound to another form of context that represents knowledge and understanding of the data in a clinical sense— Are those blood pressure numbers low, normal, or high? More specifically to chronic disease management, does the patient with whom these numbers are associated exhibit signs of hypertension? Furthermore, the definition of “hypertensive patient” requires yet another rule for data binding, such as “three concurrent readings of high over a period of one month.”

Healthcare Data Governance

Knowing when to bind data, and how tightly, to vocabularies and rules is critical to analytic success and agility. Data governance committees in healthcare will play an increasingly important role in facilitating the definition of these clinical rules that bind to data. As an industry, healthcare is amazingly immature and nonstandard in these definitions. There is significant variability across organizations, and even within organizations, in the data definition of common disease states such as hypertension and diabetes. Data governance committees will encourage, stimulate, and identify these standard data binding rules by catalyzing comprehensive and persistent agreement about those rules, within their organizations and within the industry. When comprehensive and persistent agreement is achieved, the data governance committee should act quickly to establish that rule or rules as a standard and ensure that the analytic software programming of the data binding follows that standard.

Vocabulary:  Where to Start?

Data governance committees are frequently overwhelmed by the notion of master data management. The easiest place to start the master data management journey is in vocabulary management. In today’s data ecosystem, only 19 vocabulary elements constitute 80 percent of the analytic use cases in healthcare. The data governance committee and data stewards should focus their master data management efforts on these 19 vocabulary areas first.

Page 5 of 6
1 2 3 4 5 6
Loading next article...