Health Language Blog

Terminology & Analytic Challenges for Accountable Care Organizations

Posted on 05/15/13


As the United States strives to get more value out of its health care expenditures, Accountable Care Organizations (ACOs) make sense. As both a clinician and a consumer of health services, I’m enthusiastic about providers, hospitals, and health plans jointly taking responsibility for providing holistic, coordinated, effective and efficient care. But given the at-risk nature of payments to ACOs, only organizations with the right scale of providers and patients – and the right information technology resources – are positioned to be successful ACOs.

ACOs need to analyze diverse streams of health information. Many organizations already make excellent use of administrative and claims data. As the quality of care is further emphasized in the ACO model, organizations increasingly need to analyze clinical data as well. These data are likely to come from a variety of EHRs using different data representations. For example:

  • Lab data can be crucial indicators of quality of care (for instance, hemoglobin A1c in diabetes, microbiology data for surveillance of postoperative infections). In today’s world, these data are largely represented as local laboratory codes. Although Meaningful Use 2 (MU-2) should increase the availability of LOINC-encoded laboratory results, organizations can’t wait for MU-2 to mine the wealth of laboratory data in existence.
  • For patients with chronic diseases, providing personalized support on medication use will be key to improving quality and outcomes. ACOs don’t just need to know whether doctors prescribed the right drugs – they need to know whether patients with chronic diseases are filling those drugs, and if not, why not? With electronic prescribing, organizations can access burgeoning data on drugs prescribed and drugs actually dispensed. But these data often come in incompatible proprietary formats (Medi-Span, FDB, Multum) that need to be reconciled.

Many ACOs will find it strategic to incorporate data like these into data warehouses to feed their business intelligence systems. But before they can be converted into actionable knowledge, these data must be normalized. ACOs must also carefully manage the development, maintenance, and distribution of higher-level analytic entities, such as the clinical phenotypes defining “patient with type 2 diabetes” and “patient at high risk for readmission.” Executives, clinicians, and staff will reject reports that aren’t credible – and credibility is impossible without consistent enterprise-wide definitions.

In the large, multi-system environments that will characterize ACOs, terminology services will be an essential tool for supporting analytics. Terminology services have long been used to maintain the terminologies used in clinical documentation and claims processing. Scalable products like HL’s Language Engine can also normalize high-volume flows of data, supporting auto-mapping in real-time and batch modes and providing workflow tools such as LEAP for data that must be manually reviewed. These products will increasingly be used to provide organizations with modular value sets used in quality metrics, which organizations can adapt and repurpose for a variety of analytic purposes.

ACOs are the right prescription for a healthcare system with a history of rising costs and uncertain quality and value. Management requires actionable information – information that can’t wait for a future state of interoperability in MU-2 and beyond. We look forward to continually enhancing the terminology services that allow ACOS to use their wealth of clinical and administrative data to improve care and the bottom line.

Do you have questions about terminology services? Contact Health Language - we have robust, innovative solutions for payers, providers, government payers, and healthcare IT and EMR vendors. The HLI solution can assist with your ICD-10 conversion, analytics, and clinical research needs, as well as with pharmaceutical and international applications.

data normalization


Topics: Analytics, Coding Challenges, terminology services, health information, medical data

About the Author

Dr. Steve Ross, MD is a physician informaticist with Health Language, part of Wolters Kluwer Health. Dr. Ross joined Health Language after 16 years as faculty in the University of Colorado Division of General Internal Medicine, researching personal health records and health information exchanges.