P-045 Validation of American Gastroenterological Association (AGA) Crohn's Diseases Care Pathway Risk Assessment Metrics Against Crohn's Related Costs. (March 2016)

• Record Type: Journal Article
• Authors: Kosinski, Lawrence; Brill, Joel; Sorensen, Michael; Baum, Charles; Lissoos, Trevor; Metcalfe, Leanne; Underwood, Howard; Blumberg, Pamela; Eaddy, Michael; Brining, Marie
• Is Part Of: Inflammatory bowel diseases. Volume 22(2016:Mar.)Supplement 1
• Publication Date: 2016-03
• Subjects: Inflammatory bowel diseases -- Periodicals; Colitis, Ulcerative -- Periodicals; Crohn Disease -- Periodicals; Inflammatory Bowel Diseases -- Periodicals; 616.344
• DOI: 10.1097/01.MIB.0000480091.25715.73
• Languages: English
• ISSNs: 1078-0998

Abstract

Abstract : Background: The AGA's Crohn's Disease Care Pathway (CDCP) (1) proposes 26 individual clinical markers organized into 3 groups—disease burden, inflammation, and co-morbidities—to predict a patient's risk for adverse outcomes. Project Sonar (PS) (2) uses a technology driven patient engagement platform linked to an intensive medical home within the Illinois Gastroenterology Group to identify patients at high risk for clinical failure using the CDCP thereby allowing early treatment intervention. The predictive value of the metrics was assessed against Crohn's related medical costs for participating PS patients to validate the CDCP. (1) Gastroenterology: Volume 147, Issue 3, Pages 702–705; (2) Healthleaders June 2014. Methods: For a subset of 282 Blue Cross Blue Shield Illinois Crohn's patients enrolled in PS, Crohn's related medical costs were identified from their 2014 claims using ICD-9-CM 555.X plus select ICD-9-CM symptom codes. Total medical costs for each patient were linked to their 26 CDCP measures. After reviewing the total medical cost distribution, costs were divided into 3 groups: low (min to 5th percentile), middle (5th to 90th percentile) and high (90th percentile to max). The patient sample was randomly split into model build (1A) and expected cost (1B) cohorts. A linear regression model was run on 1A with the 26 CDCP risk factors, age, gender and member months as covariates. The resulting linear regression model was then applied to cohort 1B. The differences between the 1B observed and expected costs provided the explanatory precision of the model. As member months was a significant predictor of medical costs and most predictive for those continuously enrolled for ≥12 months (n = 130), the linear regression was rerun with these patients and a multinomial ordered logit model was run to identify those CDCP risks explanatory of 2014 Crohn's related costs (i.e., statistically significant at P < 0.05) Results: Considering only those patients continuously enrolled for ≥12 months resulted in an R 2 of .22; thus 22% of the model variation was explained by the 26 CDCP risk factors. The logit model demonstrated good model fit (c-statistic = 0.758), odds ratios (OR) and 95% confidence limits for those specific risks significantly associated with increasing Crohn's related medical costs, which included: Serum Albumin (OR 19.4), Comorbidity Risk—Inflammation (OR 11.5), Inflammation Risk—Joint Pain (OR 5.7), and Inflammation Risk—Stricturing (OR 5.4). Conclusions: This study is the first of many planned to validate the AGAs CDCP and evaluate the performance of the PS program. Four of the 26 CDCP risk factors were found to be significant drivers of Crohn's related medical costs. Until further validation is performed, the factors may be used to identify patients at risk for both clinical failure and increased Crohn's related medical costs.