Q&A: Healthcare BI Faces Challenges, Opportunities
The healthcare industry faces huge challenges, including a lag in technology implementation and a lack of standardization, says Dr. Kevin Dawson in this second of a two-part interview.
- By Linda L. Briggs
- June 25, 2013
Healthcare organizations are facing massive change as they attempt to standardize processes, turn patient data into readily useable information, and meet a number of new federal regulations. In this interview, the second of two parts, Dr. Kevin Dawson explains some of the challenges his industry faces -- including a shortage of staff with the right mix of technical, medical, and business skills. (Read Part 1 here.)
BI This Week: Why do you say that the healthcare industry is behind other industries in terms of technology?
Dr. Kevin Dawson: I think there are many reasons behind that fact. Primarily, the healthcare industry hasn't standardized its processes yet; there is still much customization going on. If two individuals go to two different hospitals in the U.S. with the same disease, they may receive two different treatments. How is that determined? Pathways are published in clinical literature but are still debated, because healthcare in general is so complex. There are accepted standards for treating certain diseases, of course, but when you are dealing with individual patients, you always need to customize that. It's extremely complex.
Also, healthcare is changing very fast. By the time you view a very complex system -- an electronic health record (HER), for instance, which is part of a system with many clinical processes -- by the time you have one of those very complex systems ready and on the market, it may already be obsolete.
Also, our diagnostic coding system is changing. The International Classification of Diseases ICD-9 is used in the U.S., while most of the rest of the world is using ICD-10.
What will changing to ICD-10 mean for data management?
What it will mean in my area, business intelligence, is that all the different analyses that are based on ICD codes will need to be changed. At the same time, a more traumatic aspect of the change is that all the source system connections will have to be redone. That is a project currently occurring at many hospitals -- to modify any systems that use ICD-9.
The problem is, it's not a one-to-one translation. There are many more ICD-10 codes than there are ICD-9 codes. They are much more refined, for both procedures and diagnoses. In terms of analytics, you can imagine that if you built a data model or a metric that uses ICD code -- which is very common -- then it will be very challenging to integrate the new codes.
Those are just some of the factors contributing to challenges in healthcare today.
What about software products? Have vendors kept up with the needs of the healthcare industry?
That's a good question. One major limitation right now in healthcare and BI is that there is no such thing as a well-defined, accepted, general data model for healthcare. There are quite a few companies that have helped establish a data model as a skeleton -- a framework -- to build a data model on. Eventually, one of the major values that an organization can build out over time is its own data model -- a formalized way of describing its business processes. That's fundamental. However, building it internally takes a lot of time.
In terms of tools, we are using general business intelligence tools. That's common in the healthcare industry. It's no different than other industries, but the major value is contributed by people who are experienced in both healthcare data and business intelligence. They can build these custom data models for the organization.
Are you finding the people you need with those skills?
Actually, it's very challenging. I mentioned that there are many new initiatives currently in healthcare, and lots of change taking place. People who have the skill sets with healthcare data in general are really a highly valued commodity. Those people are currently all working; they are not looking for jobs, so yes, finding the right talent is challenging.
What do you look for in the people you hire for those positions?
It depends on the particular position. For some of the more technical jobs, like ETL and data work, somebody who has very good technical skills and a very good understanding of data in general can be trained to understand healthcare data at that level.
When we talk about positions such as business analyst, or even data architect, then experience with healthcare data is almost a necessity. Someone can start to understand the industry and the data types, but to describe the difference between healthcare and some other industries is challenging and complex.
Why is that? What are some key differences?
In this industry, we have many different types of data -- a very heterogeneous data set, definitions, and metrics. Sometimes, we have just a few elements of each. In other industries, by comparison, the complexity is less, but there may be much more data per data element.
Going back to your skills question, a business analyst who has healthcare data experience -- say with a healthcare provider or a health insurance company -- that's a major benefit, compared to someone who has good business analysis experience, but no healthcare experience.
Basically, you're looking for people who are the most technically skilled possible, but also have a healthcare background.
Any tips for readers on where to find people with those sorts of skills?
Yes. TDWI is actually one good place to find very experienced healthcare BI people. The Healthcare Data Warehousing Association, HCDWA, is another good source -- it's a little bit more academic, but they have a job board. The Healthcare Information Management Society, HIMS, is also very helpful. CHIME is also good -- the College of Healthcare Information Management Executives.
As I said earlier, in healthcare, the easy reporting has already been done. What remains is often very complex sorts of dashboards and reports that require internal talent and a solid understanding of the business.