My top 5 papers of 2017 about clinical informatics and digital health implications for clinical coding

My top 5 papers of 2017 about clinical informatics and digital health implications for clinical coding

• 1.
  CLINICAL INFORMATICS ANDDIGITAL HEALTH IMPLICATIONS FOR CLINICAL CODING: MY TOP 5 PAPERS OF 2017 SARAH DORWARD HEALTH INFORMATION MANAGER
• 2.
  SEARCH STRATEGY • Databases:PubMed, Proquest Central, Embase, Google Scholar, University of Melbourne Discovery Search Tool • Journal Search: JAMIA, JAHIMA, HIMJ • Key words: Clinical informatics, health informatics, digital health, electronic medical records, clinical coding, computer-assisted clinical coding, automated clinical coding, ICD, clinical classification • Articles retrieved: 799 • Number of articles shortlisted for review: 19
• 3.
  RATIONALE • How ishealth informatics, in particular EHRs, contributing to the development of computer-assisted clinical coding applications? • Does the study use a version of ICD, ICHI or SNOMED CT? • Could the results of the study be applied to Australian EHRs to improve clinical coding system integration? • How could the results of the study improve clinical coding accuracy and efficiency in Australia through health informatics innovations?
• 4.
  Scheurwegs, E., Cule,B., Luyckx, K., Luyten, L., & Daelemans, W. (2017). Selecting relevant features from the electronic health record for clinical code prediction. Journal of Biomedical Informatics, 74, 92-103. doi:10.1016/j.jbi.2017.09.004 • AIM: To compare feature selection methods to identify the most reliable information within an EHR to predict diagnosis and procedure codes. • METHOD: Different feature selection methods were applied to structured and unstructured data across 6 medical specialities across Antwerp University Hospital. • FINDINGS: Confidence coverage feature selection applied to unstructured data was the most successful method of predicting clinical codes. • Feature selection system was designed to be the foundation of a CAC application to support clinical coders. • LIMITATIONS: Dutch-language data used, sample from one hospital setting, ICD-10 variations, implications for clinical coding were hypothetical.
• 5.
  Jia, Z., Qin,W., Duan, H., Lv, X., & Li, H. (2017). A hybrid method for ICD-10 Auto-Coding of Chinese diagnoses. In A.V. Gundlapalli, M-C. Jaulent, & D. Zhao (Eds.), MEDINFO 2017: Precision Healthcare through Informatics (pp.427-431). Amsterdam: IMIA and IOS Press. • AIM: To develop a hybrid auto-coding system to improve coding efficiency in light of China’s shortage of skilled clinical coders. • METHOD: Three integrated approaches to automated-code assignment. The results were matched against 1537 principle diagnosis codes assigned by human coders. • FINDINGS: It was found that the combination of the three approaches lead to 96.9% precision of auto-coding. • The character-matching approach was not efficient, had poor accuracy and was to be abandoned from future trials. • LIMITATIONS: The system was found to have poor interoperability and transferability due to variations in clinical terminology across different clinical departments and other hospitals.
• 6.
  Azadmanjir, Z., Safdari,R., Ghazisaeedi, M., Mokhtaran, M., & Kameli, M.E. (2017) A three- phase decision model of computer-aided coding for the Iranian Classifications of Health Interventions (IRCHI). Acta Infomatica Medica, 25(2), 88-93. doi: 10.5455/aim.201725.88-93 • AIM: To develop a CAC system suitable for use in Iranian health services. • METHOD: Literature review of current CAC systems currently in use. • FINDINGS: 41 CAC systems are in use, 11 are fully-automated. • Fully-automated CAC systems are dependent of NLP, integration with EHR, unique language and standard terminology in medical documentation. • LIMITATIONS: Accuracy of code assignment was not considered. • Semi-automated CAC was not trialled and further research is needed.
• 7.
  Lu, M., Chacra,W., Rabin, D., Rupp, L.B., Trudeau, S., Li, J., & Gordon, S.C. (2017). Validity of an automated algorithm using diagnosis and procedures codes to identify decompensated cirrhosis using electronic health records. Clinical Epidemiology,9, 369- 376. doi:10.2147/CLEP.S136134 • AIM: Extraction of diagnosis and procedure codes to identify patients with decompensated cirrhosis. • To see if coding systems for diagnosis and procedures used within an EHR could be applied to clinical coding. • METHOD: Algorithm was applied to a random sample of 296 patient EHRs. Cluster coding of manifestations were used to identify decompensated cirrhosis. • FINDINGS: Algorithm was found to have high reliability of predicting decompensated cirrhosis. • Algorithm could be applied to EHRs to trigger alerts of possible conditions. • APPLICATIONS: Cluster coding code be used to assist with code allocation of conditions poorly represented by ICD-10. • While codes generated by an EMR could support clinical coding, it could not replace it due to the strict coding criteria that is in place in Australia.
• 8.
  Lawley, M., Truran,D., Hansen, D., Good, N., Staib, A., & Sullivan, C. (2017). SnoMAP: Pioneering the path for clinical coding to improve patient care. In A. Ryan, L.K. Schaper, & S. Whetton (Eds.), Integrating and Connecting Care: Selected papers from the 25th Australian national health Informatcis Conference (HIC 2017) (pp.55- 62). Unknown: IOS Press • AIM: Tool to map SNOMED CT codes to ICD-10-AM codes for administrative purposes. • METHOD: Mapping occurred through a web-based tool. Patient data encoded in SNOMED CT where uploaded and converted to ICD-10-AM codes. • INDINGS: The tool allowed clinicians to document freely within the EHR without compromising administrative reporting requirements. • LIMITATIONS: Mapping is not a long term solution.
• 9.
  CONCLUSIONS • Even withEHR, fully integrated and automated CAC applications are still only in their early development and testing stages. • Studies were limited to single sites, specialities or diagnosis which has implications for portability and transferability. • Standardised clinical terminologies and structures within EHR are essential to support automated CAC applications. • Strong results using data from EHR for code mapping and identification of specific diagnosis. • Fully automated CAC has been successful where the clinical coding is based around principle diagnosis assignment only.
• 10.
  FUTURE DIRECTIONS • Integratingclinical coding programs to EHRs. • Australian-based research on automated coding of routine procedures. • Better understanding of clinical terminology and data structure across clinical disciplines. • Development or modification of automated clinical coding tools that can be used in Australia.