Guest post written by Davide Zaccagnini, MD., Director of Medical Informatics at Nuance on coding complexity, meaningful use and the value of innovation
The recently proposed Meaningful Use Stage 2 provisions just might be the disruptive force that the industry has been waiting for since ARRA, in 2009. Following the first round of Meaningful Use legislation, which aimed mostly at putting systems in place, this second set of criteria is in essence demanding that electronic health record (EHR) systems actually deliver on their grand promise: an integrated, accessible and synchronized patient information ecosystem where actionable, live data becomes available through every branch of the patient care environment. It’s bold. In the end, if the final legislation resembles the majority of the current proposal, health-IT and the industry at large will be taken to the next level.
The proposed criterion specifies the use of SNOMED-CT as a means to code patient problem lists. This is an interesting proposition, mainly because there is a glaring gap that exists today between what is documented and our ability to apply SNOMED codes to it. The issue is that the skills and tools necessary to associate and assign SNOMED codes with information captured as part of patients’ charts are rare (if not imaginary) animals. Furthermore, when considering SNOMED coding, it’s difficult not to let your mind wander toward the industry’s much feared ICD-10 coding transition – another major leap provider organizations and providers are faced with and forced to make. While revenue cycle, when it comes to SNOMED is not an issue, there is still the overwhelming challenge of linking each patient condition to a unique SNOMED code. Much like with the shift to ICD-10, the industry will struggle. The combination of a too-small skilled workforce to manually abstract codes, the absence of software tools that are capable of applying the complex logic needed to automate coding, and a general tendency in healthcare to wait until it’s simply too late to take effective action will plague the industry on this issue. To make things even more interesting (complex), there is yet another requirement that calls for making all coded data available in the EHR within 36 hours of a patient’s visit. Even if manual abstractors could be found to staff HIM departments, this fast turnaround would be near impossible for the majority of organizations to meet.
It is compounding pressures like these that make the demand for automation and intelligence, driven by healthcare specific technologies, that much more important. Clinical language understanding (CLU) technologies (a clinical-specific form of natural language processing) are available and already enabling healthcare professionals to process, interpret and then leverage what becomes meaningful data (knowledge) vs. a deep sea of unnavigable data. CLU-powered technologies use highly-complex, text analysis algorithms to extract diagnoses and other patient information from clinicians’ dictated notes. Once data is identified and extracted, CLU also automates the coding of data – in line with SNOMED and other terminologies – precisely what Meaningful Use requires. Given these capabilities and the high relevance of CLU to current regulatory mandates, it is no surprise that many system vendors are increasingly introducing CLU as a key piece of their documentation systems. In addition to dictation as a means to data capture (supported in most instances by speech recognition), CLU can create a streamlined information pipeline that transforms voice (patients’ health stories) into actionable data (knowledge), in real time. Preparing for Meaningful Use will still be no walk in the park, but CLU can make the journey much easier.
Article first appeared in Nuance’s For the Health of IT blog