Download Full Text (603 KB)

Presentation date

Summer 8-12-2021

College, Institute, or Department

Internal Medicine

Faculty Mentor

Dr. Bryant England

Research Mentor

Dr. Bryant England


Background/Objective: Comorbidities can contribute to increased risk for mortality and disability in individuals with rheumatoid arthritis (RA)1,2. The Rheumatic Disease Comorbidity Index (RDCI) assesses 11 comorbidities and produces a weighted score (0-9) that accurately predicts several health outcomes3. The RDCI was developed with self-report data and later validated with ICD-9-CM codes collected from administrative data3,4. On October 1, 2015, the U.S. transitioned to ICD-10-CM, resulting in a nearly five-fold increase in the number of codes available to classify conditions5. Our objective was to update the RDCI by translating it into ICD-10-CM.

Methods: We defined an ICD-9-CM cohort and an ICD-10-CM cohort using patient data from the Veterans Affairs Rheumatoid Arthritis Registry (VARA). ICD-10-CM codes were generated by converting ICD-9-CM codes using tools that provide suggested crosswalks, and the codes were reviewed by a physician to assess clinical relevance. Comorbidities were collected from national VA administrative data over a two-year period in both cohorts (ICD-9-CM: October 1, 2013 to September 30, 2015; ICD-10-CM: January 1, 2016 to December 31, 2017). Comorbidity frequencies were compared using Cohen’s Kappa, and RDCI scores were compared using Intraclass Correlation Coefficients (ICC).

Results: Both the ICD-9-CM cohort (n=1,082) and ICD-10-CM cohort (n=1,446) were predominantly male (ICD-9-CM: 89%; ICD-10-CM: 87%), Caucasian (ICD-9-CM: 76%; ICD-10-CM: 73%), and middle to old-aged (ICD-9-CM: 67.3 ± 10.2 years; ICD-10-CM: 68.2 ± 10.0 years). Prevalence of comorbidities were similar between coding systems, with absolute differences less than 4% (range: 0.28 to 3.91). Myocardial infarction, hypertension, diabetes mellitus, depression, stroke, other cardiovascular, lung disease, and cancer had moderate agreement or higher (range κ: 0.47 to 0.84), while fracture and ulcer/stomach problem had slight and fair agreement, respectively (κ = 0.13; κ = 0.27)6,7. The RDCI scores were 2.95 ± 1.73 (mean ± SD) for the ICD-9-CM cohort and 2.93 ± 1.75 for the ICD-10-CM cohort. RDCI scores had moderate agreement (ICC: 0.71; 95% CI: 0.68-0.74)8 among individuals who were observed during both the ICD-9-CM and ICD-10-CM eras.

Conclusion: We have mapped the RDCI from ICD-9-CM to ICD-10-CM codes, generating comparable RDCI scores in a large RA registry. Individual comorbidity agreement varied, with more chronic conditions such as diabetes and hypertension having higher agreement and more acute conditions such as fractures and ulcer/stomach problems having lower agreement. The updated RDCI can be used in clinical outcomes research with ICD-10-CM era patient data.


Rheumatic Disease Comorbidity Index, ICD-9-CM, ICD-10-CM

Updating and Validating the Rheumatic Disease Comorbidity Index to ICD-10-CM