Presentation (CTI)

OBJECTIVES: HEOR investigations rely on secondary data sources with validated operational definitions of target populations to assess the economic burden of disease, real-world outcomes, and value of healthcare interventions. However, research can be complicated by limitations in clinical coding systems. Examples of conditions without referenceable diagnostic codes include generalized myasthenia gravis (indistinguishable from ocular), idiopathic underactive bladder disorder (identifiable by non-specific urinary symptoms), and desmoplastic small round cell tumor (rare condition bundled under more general diagnostic codes). Procedures including surgical interventions can require combinations of numerous ICD-10-PCS and CPT codes, systems which vary in specificity and bill type (periprosthetic joint infection interventions). In these situations, study populations are subject to significant misclassification limiting interpretability and validity of results. The authors propose a 7-step process to generate credible results: 1) convene clinical experts to describe the clinical entity regarding presenting symptoms, relevant diagnostics, and a priori expectations of healthcare utilization and disease progression; 2) link administrative healthcare data with clinical documentation to investigate the association between patterns of diagnostic/procedure/treatment coding and clinician’s documentation of the clinical entity; 3) assess heterogeneity in real-world coding practices; 4) develop working computable definitions of the clinical entity; 5) assess and characterize the definition using administrative data, 6) review empirical results with clinical experts and assess the alignment/divergence of anticipated vs. observed results; and 7) revise definitions based upon feedback to reduce misclassification. Including these methods and the corresponding insights in the research design mitigates the limitation and supports clinical face validity of results. Using the examples of UAB and PJI surgical intervention, the authors will present examples of the information crosswalk between clinical progress notes, diagnosis and procedure codes, and grouping cases in claims data to illustrate the value of this process to increase the utility and validity of administrative healthcare data sources.
METHODS: X
RESULTS: X
CONCLUSIONS: X