Not long ago, parts of our health system remained mired in technological backwaters. With patient records locked in paper charts, sharing information with other hospitals or physicians usually involved a foot chase across town.
Today, following a period of innovation and unprecedented investment, most physicians are online and managing their patient records electronically. This increase in the use of Health IT is resulting in greater connectivity between all those who work in and depend on the health system. It’s also ensuring clinicians accurately and more comprehensively capture details from patient encounters. This, however, combined with the tendency of many users to copy and paste information from prior visits into daily progress notes – ultimately duplicating information found elsewhere in the medical record – is driving an explosion in the volume of patient data and is creating new administrative challenges.
The challenges are most evident in clinical documentation improvement (CDI) efforts. Hospitals established these programs in their efforts to work with care providers to better reflect the quality and intensity of patient care while increasing accuracy in coding and external reporting.
According to a survey by the Association for Clinical Documentation Improvement Specialists, CDI specialists typically review between 8 and 12 new cases per day, with between 33 and 48 minutes spent on each initial review. Fewer than half of these reviews trigger a query to clinicians to clarify a diagnosis or procedure note or to seek more information. And up to 66 percent of these reviews may be unnecessary.
While, over the long run, CDI reviews help improve a hospital’s case mix index and ensure accurate reimbursement rates, their costs may absorb – or eclipse – those benefits. Ending CDI reviews is not an option, given their added necessity to compliance efforts and participation in quality measurement initiatives.
The good news is CDI can be automated by the same computer-assisted coding (CAC) technologies many hospitals are adopting as they gear up for a different administrative juggernaut: ICD-10.
CAC technology uses natural language processing (NLP) to automate the examination of complete clinical records, assess the medical context of potential diagnosis and procedure notes, and then render the appropriate codes.
A special initiative Optum has underway with the UPMC in Pittsburgh, Pennsylvania, applies NLP in CDI reviews. There, we are developing an application that uses our NLP engine, LifeCode, to pre-screen patient records in real-time using a set of proprietary case-finding rules to filter those cases most likely to require additional CDI work from the multitude that don’t. The software also summarizes each case, and flags potential errors to help CDI specialists quickly formulate precise queries to clinicians to make corrections and offer tips that improve documentation of future patient encounters.
By applying NLP technology to CDI workflow, we can make initial case reviews dramatically more efficient and help CDI specialists spend time on the cases that are most likely to yield a return. There is significant potential to improve productivity and decrease administrative costs to overall CDI efforts.
Next month, Healthcare Exchange will feature a guest post from our partners at UPMC describing their experiences with the application to date, and discussing how it supports their efforts to enhance CDI efforts.