(Editor’s note: Because many healthcare organizations are still not incorporating data governance into their data initiatives, we’ve updated and republished this post. Read the updated post.)
Like going to the gym or skipping ice cream, most of us involved directly or indirectly with data in healthcare are at least aware that we “should” do something about data governance. Some of us have even told ourselves, or our colleagues, that we’re going to make sure that our healthcare data governance initiative gets our attention next quarter. Or maybe in the fall, right after we finish this big project.
But the warning signs that you need to make data governance a priority may not be as clear as the result of too much ice cream. Unless you know where to look. So in homage to those irresistible summer beach-read magazine lists, I present “5 Signs Your Healthcare Organization Needs Data Governance.”
1. The data is wrong; this is <insert another department>’s fault.
Interdepartmental finger point over data issues is one of the most telltale signs your healthcare organization is struggling with data governance problems – especially when that finger pointing involves IT and another department. The frustrated clinician knows the data they are looking at doesn’t reflect what really happened to the patient, and the overworked IT analyst knows that they didn’t change the data. She just moved it from system A to system B and made sure it was backed up and available for this week’s report.
Data governance can help get everyone on the same page about the data question before the finger pointing starts. Start with the fundamentals such as being clear about or answering questions such as:
- What is the list of AMI patients from last month?
- Is that the right definition to apply to this data?
- Is that the right data to answer that question?
- When was that data updated?
2. A never ending queue of requests for new reports
Despite having more than 2500 production reports across your EMR vendor’s multiple reporting tools and your custom analytics platform, your overworked analytics team and business intelligence analysts have all but given up hope of keeping up with the onslaught of new requests pouring in each week from unsatisfied users who can’t find the data they need. Most frustrating is that 90% of these requests are for reports that either already exist or differ only by a column or two from an existing report. These requests are something your users should easily be able to adapt in order to meet their needs since they have completed your BI self-service training program (but more on that later).
Perhaps the issue isn’t your self-service analytics program or even user laziness, but is instead a lack of an easily searchable, documented, and well governed report catalog that spans the myriad of BI and reporting tools that you’ve acquired over the years. A single, easily-searchable catalog containing reports from across your enterprise enables users to find reports related to relevant business concepts or critical data elements that are directly or indirectly used in the report they need.
3. Focusing on the data instead of the analysis and conclusions
When managers and executives at all levels start talking about the data instead of taking about what the data means to patients or the operations of the institution, you’ve uncovered a sure sign that your organization needs to spend more time on governance. Typically, someone presents data or a report to support a controversial analysis or conclusion. But instead of the data becoming the factual reference point that the presenter intended, the discussion around the table quickly shifts focus to where the numbers came from or if the numbers presented are the “right” numbers.
Too often, what’s really underlying these objections about the data is a lack of understanding about the concepts being presented. You can trace this problem back to a lack of consensus on definitions and meaning. Even if everyone has a common understanding, the presenter may have provided little in the way of transparency on the data itself such as documenting where the data came from or what was included or excluded in the data used for their analysis. Providing this additional data lineage can deter those who’d rather argue over the “data” than face the possibility of there’s a clear case for change.
4. Business intelligence self-service programs not taking flight
Business intelligence (BI) tools have come a long way in the last 10 years. Box plots, tree maps, even predictive visuals are just a few clicks away in a modern BI tool, assuming the user understands the tool and the data. Self-service BI initiatives have focused on teaching users how to use these new BI tools, but the returns on these initiatives are often underwhelming. Sure, there will always be those who would rather have someone do it for them than take the time to find the answer, but most people like to be able to answer their own questions.
Too often the root cause slowing the adoption of these self-service BI programs is lack of context for the data itself. If a user can’t determine if a list of providers includes both in-house hospitalists and referring physicians, they are unlikely to be able to use that data set within their cool BI dashboard effectively, or worse, accurately. Governance decisions that are available alongside the self-service BI environment can provide the critical context often missing for users to be truly self-sufficient with a BI tool and some training.
5. “Mary” is the most important individual in the organization
Every organization has a “Mary,” and thankfully, most healthcare organizations have a couple. Mary is the person who “knows the data.” She is the person without whom that board report would never get done, and without whom, we would never know where to get that data that Dr. Wallid needs tomorrow. Mary has been around for years and is not only the “data expert,” but also the data librarian, custodian, and part-time data archeologist. Mary is the person whose resignation or retirement would send chills through the CIO’s spine on a hot summer day.
Much of what Mary has learned over her years of service are the unwritten governance policies and practices – or at least what serves as governance – for an organization that’s relied far too long on the expertise of a few experienced analysts. While we all love Mary dearly, we also know that our dependence on her isn’t healthy. Data has become far too critical to allow tribal knowledge to be the only way policies, processes, and methods for successful data use are passed from generation to generation within the organization. Don’t get rid of Mary, but know that someday she is going to want to go on that cruise-of-a-lifetime, and she might not be there when you need her. Get her engaged now. Ask her to share what she knows. It will most likely take longer than you think.
Don’t wait to see all five of these signs to get started with healthcare data governance. Any one of these signs indicates that there’s real work to do. And if you don’t know where to get started, reach out to Collibra. We would be happy to share how other healthcare organizations crossed data governance off their to-do list.