Healthcare Analytics & Data Governance
By Bonnie Chung, Vanna Trieu and Matthew Trisic
Data is the new oil of the 21st century. It is such a valuable commodity for companies that we have entire teams solely dedicated to working with it, whether in building the infrastructure or performing analysis. Like how oil fields are protected from sabotage and gasoline should be stored in proper containers, data needs similar protection, refinement protocols, and regulations governing its use. In healthcare, careful treatment of patient data isn’t new — it is the law. If you Google searched “data governance in healthcare” and stumbled upon this blog post, then you’ve already taken your first step towards high quality and robust data. In this post, we will explore what data governance is, why it is important in a healthcare setting, and dive deeper into the people aspect of data governance.
What is data governance?
The term “data governance” covers a lot of different sins within an organization, from data quality to data security. According to Wikipedia, “the key focus areas of data governance include availability, usability, consistency, data integrity, and data security. It also includes establishing processes to ensure effective data management throughout the enterprise such as accountability for the adverse effects of poor data quality and ensuring that the data which an enterprise has can be used by the entire organization.” Within the realm of healthcare analytics, data governance plays an especially important role due to HIPAA regulations and the need for trustworthy data. Data governance allows for consistent definitions and transparency into the source and the methodology used to create metrics, increasing trust in the data at the point of consumption. Analytics practitioners — on the technical and business side — all have an interest in adhering to their organization’s data governance framework.
Why data governance?
Poor data governance leads to slower decision making. Once decisions are made, they are of lower quality. These lower quality decisions are a result of the principle of Garbage In, Garbage Out (GIGO). Even worse, it engenders diminished trust in the data and proliferates the creation of data siloes — the opposite outcome of an analytics team’s goals. Healthcare organizations cannot let the latter happen. Why? Because the stakes are higher in healthcare. The aim isn’t to create marketing funnels, increase conversion rates, or drive demand for the coolest juicer ever.
Instead, healthcare organizations are focused on understanding their patient populations and providing better outcomes. Here are some key insights that can be derived from their data, while keeping in mind that data governance underpins the answers generated:
· What’s the average reimbursement rate for procedures?
· What activities, comorbidities, etc. factor into readmission risk for a given condition?
· How are doctors performing against CMS’ value-based care measures?
· What’s the stratification of the patient population by disease cohort?
· Does an apple a day really keep the doctor away?
Data governance provides a great opportunity for risk mitigation. A patient data leak or a bad decision steered by bad data can cause an organization to take a large hit to their reputation. Through data governance, an organization can minimize the chances of data issues and allow their employees to focus on creating better outcomes for their patients or on making decisions to grow the business.
In one example, a healthcare system had many different groups each generating their own metrics. Each group had slightly different definitions, making it impossible to consolidate the data to get accurate organizational level reporting as the organization scales. Proper data governance allows organization to scale and grow.
Getting Started with the Right People
The Health Information Technology for Economic and Clinical Health Act (HITECH Act) of 2009 served as the impetus for enacting and refining our modern definition of data governance. This act ushered in the implementation of electronic health records (EHR) which later enabled the application of data analytics to healthcare. You might be thinking at this point, “Great, we’re already on Epic across all facilities so let’s push that patient data into Tableau dashboards and prove we’re at Stage 1 Meaningful Use!”
Before getting to the fun part — analysis — think about what your organization needs to drive data governance forward.
The work involved in data governance can be divided into three main segments: defining rules and process, ownership, and execution. Depending on size and maturity of a given organization, these roles can be flexible. Typically, the business leaders will define rules and processes at a high level, and the technology leaders will own the implementation of these rules on a macro level. A key factor in the success of data governance is whether there are sponsors on both the business and technology sides. Without support from both groups, there can be rules that are implemented inconsistently or inaccurately, or metrics that don’t truly capture the needs of the business.
A Data Steward is a critical role in following through on the actual execution of the data governance. Very often, the Data Steward is played in a part-time role or collaboratively as a committee. When it’s played as a committee, you can run into contrasting opinions and delayed decisions. When it’s a part-time role, the job never gets the focus and attention that it needs. For maximum efficiency, we recommend that the data steward is a full-time role filled by a person, who is responsible for:
· Executing to ensure data quality
· Capturing and documenting metadata and other information for critical data entities
· Documenting the provenance and source of truth of data entities
· Providing a structure around access control policies
· Working closely with analysts, data architects, and other organization members to maintain data governance standards
Healthcare Analytics
Anyone working in analytics will tell you that 90% of the analytics lifecycle is data preparation. Extending that model, we can treat data governance as the most important part of data preparation. Before a team can even start on analysis, there are myriad steps and hurdles the data must go through. There should be teams that hold ultimate accountability for each data source, ensuring its use conforms to data governance policies.
When it comes to healthcare, the teams who have accountability for each data source are even more critical, because they are ultimately responsible for the regulations and requirements that are applied to whatever the data by their respective governing body.
When data governance is in place, you can feel comfortable starting to run the most interesting and impactful types of analytics. Many healthcare organizations are paralyzed by their own fear of the government regulations that surround the sensitive data that they collect. However, if you have a strong data governance team, their ability to consolidate and analyze data from any source turns from weeks to days to hours, providing an incredible impact to the people they are trying to help the most: their patients.
