Quality improvement of International Classification of Diseases, 9th revision, diagnosis coding in radiation oncology: single-institution prospective study at University of California, San Francisco

Pract Radiat Oncol. Jan-Feb 2015;5(1):e45-51. doi: 10.1016/j.prro.2014.03.007. Epub 2014 May 22.


Purpose: Accurate International Classification of Diseases (ICD) diagnosis coding is critical for patient care, billing purposes, and research endeavors. In this single-institution study, we evaluated our baseline ICD-9 (9th revision) diagnosis coding accuracy, identified the most common errors contributing to inaccurate coding, and implemented a multimodality strategy to improve radiation oncology coding.

Methods and materials: We prospectively studied ICD-9 coding accuracy in our radiation therapy--specific electronic medical record system. Baseline ICD-9 coding accuracy was obtained from chart review targeting ICD-9 coding accuracy of all patients treated at our institution between March and June of 2010. To improve performance an educational session highlighted common coding errors, and a user-friendly software tool, RadOnc ICD Search, version 1.0, for coding radiation oncology specific diagnoses was implemented. We then prospectively analyzed ICD-9 coding accuracy for all patients treated from July 2010 to June 2011, with the goal of maintaining 80% or higher coding accuracy. Data on coding accuracy were analyzed and fed back monthly to individual providers.

Results: Baseline coding accuracy for physicians was 463 of 661 (70%) cases. Only 46% of physicians had coding accuracy above 80%. The most common errors involved metastatic cases, whereby primary or secondary site ICD-9 codes were either incorrect or missing, and special procedures such as stereotactic radiosurgery cases. After implementing our project, overall coding accuracy rose to 92% (range, 86%-96%). The median accuracy for all physicians was 93% (range, 77%-100%) with only 1 attending having accuracy below 80%. Incorrect primary and secondary ICD-9 codes in metastatic cases showed the most significant improvement (10% vs 2% after intervention).

Conclusions: Identifying common coding errors and implementing both education and systems changes led to significantly improved coding accuracy. This quality assurance project highlights the potential problem of ICD-9 coding accuracy by physicians and offers an approach to effectively address this shortcoming.

MeSH terms

  • Clinical Coding / methods*
  • Clinical Coding / standards*
  • Cohort Studies
  • Electronic Health Records
  • Female
  • Health Facilities
  • Humans
  • International Classification of Diseases / standards*
  • Male
  • Neoplasms / classification*
  • Neoplasms / diagnosis
  • Neoplasms / radiotherapy*
  • Prospective Studies
  • Quality Improvement
  • Radiation Oncology / methods*