An article from the Kansas City Star outlines how the Kansas City Veterans Affairs (VA) Medical Center published quality data online, found systemic issues in their care, and set about resolving those systemic issues. (Also related “Data Spur Changes in VA Care” from the Health Reform Report)
To summarize, the VA provides care for United States veterans. Hospital administrators decided to exceed Medicare guidelines and publish extensive quality information on their care levels. Veterans can compare quality statistics from different hospitals. Admirable. However, public scrutiny of this data revealed that the number of deaths among surgery patients at the Kansas City hospital was 79% higher than what was expected. 79%. Wowie.
Immediately, the VA started following a core information governance practice: root cause analysis. (You can find more details on the root cause analysis process in the blog post from Carol Newcomb, “A Data Governance Primer.”) Normally, the process goes something like this:
- Houston, we have a problem. Either a business process stopped functioning sufficiently, or a data analyst noticed discrepancies in reports or dashboards. (See this excellent post from TDWI on the Six Myths about Data Analysts). In this case, let’s assume that the issue has high enough priority to proceed with the investigation.
- The data issue is passed to a Data Analyst or Data Steward, who immediately starts asking questions. These questions don’t follow a particular order, but customers I talk to many times follow this path:
- Where did this data come from? Are multiple fields contributing to this single data element?
- When was the last time the data was refreshed?
- Where was the data transformed? And how? By whom? When? Why?
- Is the data complete enough to be useful for decision-making?
- Now that the Data Steward is armed with facts about the data in question, they need to evaluate the data system. Essentially, they’re seeing to answer where the problem got introduced in the business process, what kind of issue it is (training, controls, etc.), and if they can do anything to stop the problem from perpetuating.
- Review and approve the findings. Depending on your organization, sometimes this is very quick. But if you are decentralized and do not yet have a formal information governance organization, this could take awhile.
- Quick-fix the data issues in question. And use those same business rules to establish a preventative program so the same issue doesn’t crop up next month.
- Monitor that your fixes are indeed working through dashboards that incorporate trends over time. These dashboards will help you pinpoint system events that impact your quality.
The VA followed a process much like this one. In fact, the article says that “By compiling data on patients, making it public and holding administrators’ feet to the fire when numbers don’t turn in the right direction, the VA is trying to prod improvement at its 150-plus hospitals.” Now that’s a lofty information governance project (which should turn into a long-term program)!
The root cause analysis outlined above can be very time consuming, especially if some of your operational data is in personal spreadsheets or Access databases. I’m sure none of you have that problem. Right? RIGHT? J
Tools can help you more holistically manage the movement of the data, and provide visibility on how that data is transformed and used. In this case, the VA was only able to provide the additional quality data analytics because of governmental program. The article outlines the motivation:
“The VA did away with much of its paperwork years ago and replaced it with electronic medical records. The system can track in detail the condition patients were in when they arrived at the hospital, treatments they received and how they fared.
Several years ago, the VA started putting its databases together so that hospitals could more easily see how they were doing.”
SAP tools can help you with this process. We have tools that help in the on-boarding stage of your information. In this stage, focus on understanding and quantifying the information you are getting AND on establishing policies for how long you need to retain this information. In particular, Information Steward can automatically show the data lineage (where did this data come from, and how was it transformed/combined). Information Steward also helps you quickly determine if the information is complete enough o support decision-making.
Root cause analysis usually inspires data integration, data cleansing, or even master data management projects. Tools to highlight here are Information Steward, Data Services, Master Data Management, Extended Enterprise Content Management (ECM), and Business Process Management (BPM). BPM is a significant aid to root cause analysis, as you need to determine points of failure or optimization opportunities in your business process. To do so, you need to have an orchestrated, managed business process with audit and traceability features.
Keep in mind that these projects should also inspire preventative information governance. In fact, you can use the same information policies applied in your project to enforcing standards and policies at point-of-entry. This flavor of data quality firewall can jumpstart an on-going information governance program.
Of course, in the case of the Kansas City VA, all of this data was surfaced through analytics. For analytics, SAP provides Business Intelligence tools.
So what did the root cause analysis reveal at the Kansas City VA? To quote the article, “Sanders said the hospital has increased its surgical staffing and is looking at ways to improve patient counseling about palliative care.” Both training and business process changes were required.