Today is the first day healthcare organizations can report meaningful use data to the government for electronic health records (EHR) systems reimbursements.  So you might think that with all this newly digitized patient data, it shouldn’t be long until doctors and clinicians start applying data mining and predictive analytics to improve patient care.

Well, sort of. Most EHR systems have native reporting tools built-in, and some boast more advanced analytics capabilities. PracticeFusion, a web-based EHR system, allows clinicians to create reports based on patient data to identify high-risk H1N1 patients, for example. Almost all EHR’s provide some type of revenue reporting capabilities.

But dig a little deeper and limitations become clear. It’s a two-part problem, as I see it. First, most EHR analytic capabilities are rudimentary compared to analytics in other fields like finance, retail and manufacturing. Second, virtually all EHRs are proprietary platforms that make it difficult if not impossible to share patient data with other EHRs and third-party sources.

Let’s take the second part first. Each EHR system stores and process patient data in its own unique way making data integration a nightmare. Data models, hierarchies and definitions vary from EHR to EHR. So a primary care doctor using a GE Healthcare EHR can’t share patient data with a specialist across town using a PracticeFusion EHR, for example. As a result, neither has a complete set of patient data on which to run analytics to, say, improve diagnosis or identify treatments most likely to work. (The reason is that most EHRs were developed independently of one another, with hospitals forcing their own ways of working with patient data into the development process, but that’s another discussion altogether.)

The lack of data standards also hampers the ability of third-party data analytics vendors to create really robust healthcare-specific tools to sit on top of EHRs. Wouldn’t it be great if SAS Institute, for example, developed a predictive analytics tool that was interoperable with most major EHRs on the market and could help doctors forecast likely patient outcomes based on different treatment options? But as it stands there’s no way for a software company to produce a data analytics tool compatible with most EHRs because each has its own data models and standards.

So what’s the solution? Obviously, somebody needs to develop industry-wide, accepted data standards for storing and exchanging electronic patient data. The difficult part of this job isn’t the technology. It’s deciding which standards to adopt. Each EHR vendor will want to push its own standards, each of which have their own set of advantages and disadvantages. Doctors and clinicians that have already purchased an expensive EHR system will want to see their vendors’ standards adopted, lest their investment go to waste. Patients, I suspect, probably don’t care which standards are adopted, as long as they include robust privacy controls.

From my perspective, there are three options here. Either the government can develop and mandate electronic patient data standards from on high, as a recent White House report recommends; The market can decide which standards gain adoption if one or another proprietary EHR gains a dominant share of the market; Or EHR, middleware and data management vendors can voluntarily come together to develop patient data standards and improve data integration capabilities.

The bottom-line is that EHR data standards need to be developed for healthcare providers to get the benefits of powerful data analytics. Which road we take to get there is far from clear, but CIO’s and IT directors at healthcare providers should ask their EHR vendors how they plan to tackle EHR interoperability and data analytics. Push them for as direct an answer as you can get and impress upon them the importance/potential of sophisticated healthcare data analytics.