UPDATED CODING ALGORITHMS FOR DEFINING CHARLSON COMORBIDITIES USING LARGE RESTROSPECTIVE HOSPITAL ADMINISTRATIVE DATA

• Published in: Value in health 2017-05, Vol.20 (5), p.A48
• Main Authors: Rosenthal, N, Cao, Z, Chung, J, Magee, G, Gundrum, J, Dreyfus, J, Robinson, S, Baumer, D, James, B
• Format: Article
• Language: English
• Subjects: Acquired immune deficiency syndrome; AIDS; Algorithms; Classification; Clinical decision making; Clinical outcomes; Comorbidity; Complications; Data processing; Dementia; Dementia disorders; Deviation; Diabetes; Diabetes mellitus; Health care; Heart diseases; Heart failure; Hospitals; Inpatient care; Medicaid; Medicare; Myocardial infarction; Statistical analysis; Teams
• ISSN: 1098-3015; 1524-4733
• DOI: 10.1016/j.jval.2017.05.005

OBJECTIVES: To update Deyo’s International Classification of Disease, 9th revision, Clinical Modification (ICD-9) coding algorithm for Charlson Comorbidity Index (CCI) with corresponding ICD-10 codes and assess the concordance between algorithms using a large retrospective hospital administrative database. METHODS: Deyo’s ICD-9 algorithm for CCI was updated referencing Quan’s modified version of Deyo’s algorithm and clinical judgment made by the Premier team. The ICD-10 algorithm for CCI was developed using the mapped ICD-10 codes based on the Centers for Medicare and Medicaid Services (CMS) 2016 General Equivalence Mappings, with further adjustment according to the 2016 CMS ICD-10 codes tabular list. The ICD-10 algorithm was validated by a concordance analysis using data from hospitals that submitted both ICD-9 and ICD-10 data to the Premier Healthcare Database during Quarter 4 of 2015. Descriptive analysis was conducted to compare the distribution of CCI scores and prevalence for the individual conditions between the two algorithms. Concordance was assessed by Kappa statistics calculated for each of the CCI comorbidities. RESULTS: The analysis included 26,792 inpatient discharges from 79 hospitals that reported both ICD- 9 and ICD-10 diagnoses codes between 10/1/2015-12/31/2015. Modification to Deyo’s ICD-9 algorithm included: 1) inclusion of 294.1x (dementia in conditions classified elsewhere) and 294.2x (dementia, unspecified) to Dementia codes; 2) removal of 250.7x (Diabetes with peripheral circular disorders) from Diabetes without chronic complication group and inclusion of 250.7x, 250.8x (diabetes with other specified manifestations) and 250.9x (Diabetes with unspecified complication) to Diabetes with chronic complication group. Mean CCI was 1.42 with standard deviation of 2.10 for both algorithms. The Kappa statistics ranged from 0.82 for diabetes with chronic complications category to 0.98 for AIDS, chronic heart failure, dementia, and myocardial infarction. CONCLUSIONS: The ICD-10 CCI score had excellent concordance with the ICD-9 algorithm and both algorithms are available for use in outcome studies.