ICD-10 pilot finds room for improvement

An ICD-10 coding pilot project found plenty of room for improvement.

Participating coders achieved an average accuracy rate of just 63 percent, according to a report from the Healthcare Information and Management Systems Society (HIMSS) and the Workgroup for Electronic Data Interchange (WEDI).

The National Pilot Program was designed to be a collaborative, industrywide effort to help those transitioning to the new coding system by sharing best practices and lessons learned. Though initially meant to be an end-to-end test, time constraints meant initiating a separate test of billing.

The pilot involved real, de-identified medical records coded against an answer key developed by certified trainers from various organizations. Providers, clearinghouses and health plans could then test their coding against the answer key.

Testing by the North Carolina Healthcare Information and Communications Alliance (NCHICA) served as a basis for the national pilot. NCHICA Executive Director Holt Anderson told the Medical Group Management Association at a conference earlier this month that results from its efforts were "scary."

The national testing was done in 12 waves, with low scores often associated with "unspecified" diagnoses, such as "chest pain, unspecified" or "congestive heart failure, unspecified." Some variation in coding was attributed to organizational variations in reporting minor procedures and codes for history and risk factors. Some hospitals, for instance, might code every indication, such as history of hemorrhoids, for example, while others include only those relevant to the case at hand, the report says.

Among the findings:

• Occasionally coders coded the diagnosis but forgot to code the procedures.
• Some coders became too dependent on the software instead of determining when they needed to override it.
• Testing organizations with fully implemented EHRs had difficulty coding medical records that were hand-written.

Among the recommendations:

• Move away from silo-based testing software and approaches that inhibit collaboration across the healthcare continuum.
• Assign all codes from a native ICD-10 perspective rather than converting from ICD-9.
• Anticipate coding discrepancies; participate in a group that allows for a national discussion regarding discrepancies.

The paper concludes by saying that all affected entities should be testing now, and should allow as much time as possible for testing to protect their bottom lines.

Read the complete report.