Q&A: BI Helps Healthcare Meet Huge Challenges
"BI and analytics are new concepts to many healthcare organizations," says Ted Corbett, whose consulting company, Vizual Outcomes, focuses on how BI and data visualization can help healthcare providers meet huge future challenges.
- By Linda L. Briggs
- March 12, 2013
Healthcare is on the cusp of massive change, much of it tied to how data is collected, stored, and analyzed. In this interview, Visual Outcomes founder Ted Corbett discusses the challenges faced by healthcare organizations and how better tools for data visualization can help. "The need for robust analytical tools to address the various users in a healthcare organization is now critical," Corbett says.
BI This Week: What transformations are we going to see in healthcare in the next few years, and how will it affect the use of business intelligence by healthcare providers?
Ted Corbett: Healthcare is on the cusp of massive changes in how hospitals and providers are compensated by commercial payers and Medicare. Traditionally, hospitals and providers have worked on a fee-for-service approach, in which they are paid when a patient goes to the doctor or hospital. There is little incentive for providers to help keep people well and out of the doctor's office or hospital.
With healthcare reform, ideas such as accountable care organizations (ACOs) and "capitation" are gaining traction. Under these models, providers and hospitals agree to care for a specific population and are paid an agreed-upon amount for their care. If the population stays out of the hospital, the providers still get paid.
This represents a shift in risk from the payers to the hospitals and providers. Traditionally, the data to assess risk was housed either with CMS (Centers for Medicare and Medicaid Services) or commercial payers. Understanding these risks will require more data and analysis than has been used to date. In fact, the need for robust analytical tools for the various users in a healthcare organization is now critical.
How are BI, analytics, and big data challenges different in healthcare than in other fields?
One of the biggest challenges in delivering BI to healthcare organizations is that business intelligence is a relatively new idea for most organizations. Through electronic medical records and "meaningful use" (a term referring to effective use of electronic health records within a healthcare organization), organizations are now stockpiling vast quantities of data that they haven't had access to historically. Despite all of this data, organizations aren't sure how to get access to it or make sense of it. The traditional approach of using Excel spreadsheets and manual processes are time-intensive and leave executives and administrators waiting in long lines to get access.
Electronic medical record (EMR) vendors have delivered many bolt-on data warehouse and reporting capabilities, but those solutions still require extensive training and expertise to get access to data. Healthcare companies are buried under a sea of reports, and users don't have the skills or time to integrate and transform data into actionable information.
In addition, healthcare leaders, doctors, and nurses have extensive experience in patient care, but not in using computers and technology to make decisions. In many cases the training required is not about something as sophisticated as how to use big data -- it's about basics such as how to leverage browsers and spreadsheets.
Leaders struggle with how to prioritize analytics when faced with strained budgets due to EMR adoption, meaningful use, and looming ICD-10 conversions. (ICD-10 refers to a revision of the system of disease codes used by physicians and hospitals.)
BI and analytics are a new concept to many healthcare organizations. Often, organizations don't have the staff, budget, or expertise to plan and adopt solutions that have been in place for 20 years in other industries such as financial services, retail, and technology companies.
How big a challenge is HIPAA and related security issues when we talk about BI in healthcare?
Securing patient data is a core tenet of all healthcare BI and analytics efforts. Specifically, there are 18 elements of personal health information (PHI) that need to be protected for patients. One challenge organizations face is how to remove those core elements while still providing relevant information for analytics.
All healthcare organizations are mandated to deliver the minimum necessary amount of information for operations, researchers, and others to analyze patient data. Traditionally, many healthcare organizations use Microsoft Excel spreadsheets and Access databases to house critical data for analytics. This approach puts many organizations at risk, however, because sensitive personal health information (PHI) is then housed on computers across the organization and on laptops and USB drives.
The use of a centralized business intelligence architecture and modern BI tools can help minimize risk by keeping all PHI behind the scenes, removing key PHI information while still enabling analysts and researchers to analyze trends and track operational performance.
With a centralized architecture, all key elements can be preserved in a warehouse, and security and privacy controls can be put in place. In my experience both at Seattle Children's Hospital and Swedish Health System, we have delivered robust reporting and analytical capabilities without exposing any PHI to end users.
We can devise ways to show providers how many patients they've seen, as well as [necessary treatment and diagnosis] codes, timestamp information about patient flow, and financial performance. All of that can be analyzed without exposing any PHI. If a user needs to know specific patient-data related elements, existing controls can be leveraged.
When we use the term "users" in healthcare BI, what different groups are we talking about, and what are the different challenges in reaching each?
There are several different user groups, all of whom require access to more data to help make fact-based decisions, each with their own considerations. These include:
- Executives: They want information delivered in a highly summarized format, either e-mailed or printed for review. Executive users usually don't have the time or aptitude to use robust online reporting tools.
- Clinical administrators: Mid-level managers struggle to summarize information for executives and to deliver detailed information to providers who are generally skeptical of any information delivered. These leaders don't have the technical skills or time to build conduct analysis on their own but are asked to provide vastly different types of answers.
- Providers: Doctors and nurses who care for patients every day want to know more about their own performances and how they relate to similar providers. Providers are demanding and detail-oriented but often cannot get information on their own due to time or technical capability.
- Analysts: Most healthcare organizations have analysts everywhere. They collect data in spreadsheets and deliver answers to providers, clinical administrators, and executives, generally through the use of traditional Microsoft Office tools such as Excel, Access, and PowerPoint. Many are trained in how to pull data from core system reporting tools such as Business Objects or Crystal reports but cannot create reports on their own.
- IT analysts or developers: These users are inundated with requests from users above them. They often have the technical capabilities to write queries but lack the functional knowledge to act independently. They struggle to keep up with the demands of all of the users listed here and are stymied by the lack of time, tools, and resources to meet growing demand.
How does unstructured data fit into the picture with healthcare and BI?
Unstructured data is an important component because it houses some of the richest data in the hospital system. Providers enter vast quantities of data into clinical notes, either through telephone transcription or directly online. These notes serve as the basis for patient care, medical coding, and billing. Unfortunately, there is little consistency across providers in note format, which makes it difficult to access this rich store of information.
Providers don't have time to enter all of the data required into structured databases, so the clinical note provides a large and critical piece of the puzzle. Many technologies, such as optical character recognition (OCR) and natural language processing, are in the research and development stages to help harness this data. Until such technologies are more advanced, most data used in BI in healthcare comes from billing and coding systems.
What are some examples of innovative things being done with analytics in healthcare?
What excites me most in healthcare is what can be done by getting data into the hands of users who can directly affect change. As director of knowledge management at Seattle Children's Hospital, I brought in Tableau Software to help improve self-service access to data. My team enabled power users by providing initial training, then turned them loose while helping empower them with better, cleaner, integrated data.
My favorite story is about the business manager of the operating room, who met with the chief of surgery once a month prepared with several spreadsheets and presentations. Inevitably, the chief would say, "Why are you telling me this? All I want to know is x and y." She'd return the next month with the answers to the previous month's questions, and the cycle would continue.
When the business manager started using Tableau, along with a data mart of clinical and operational data, she began attending the meeting with nothing prepared. She'd open the computer and say, "What do you want to know?" After a lively discussion, that useless 15 minutes a month turned into hours of productive conversation, enabling the chief of surgery to make actionable decisions in a more timely way.
Can you briefly describe your consulting firm, Vizual Outcomes, and its focus? Why healthcare and BI?
Vizual Outcomes, a Tableau Software partner, is a consulting company focused on delivering expertise in BI and Tableau-based solutions to healthcare and research organizations. I started Vizual Outcomes because I love using Tableau and helping to solve challenging problems in a dynamic and important industry. Prior to joining Seattle Children's Hospital, I spent about 10 years in consulting, delivering BI solutions to a variety of companies and industries, most outside of healthcare. I enjoy learning about new organizations and how I can help harness the power of tools like Tableau to transform data into actionable decisions.