Purpose – This research describes the development and independent validation of prospective mortality risk adjustment indices for ambulatory Medicaid populations based on ICD-9-CM codes and assesses the incremental value of automated pharmacy data.
Methods - A retrospective review of Georgia and North Carolina Medicaid claims from 1990 to 1997 was used to identify ambulatory recipients (ages 15 to 50). Cox proportional hazards regression was used to model seven-year survival based on ICD-9-CM codes only and ICD-9-CM codes supplemented with drug exposure information. Risk factors, identified on statistical empirical evidence in the GA sample in the year prior to the index date, were subsequently submitted to a clinical panel for validation. The clinically validated GA models were then re-estimated, ‘frozen’, and prospectively validated on the external NC Medicaid cohort.
Results - We identified cohorts of 273,970 GA and 120,000 NC Medicaid recipients. In the GA cohort, c-statistic for ICD-9-CM only and combined survival models were 0.86 and 0.88, respectively. Demographic, eligibility, ICD-9-CM and drug-based comorbidity risk/protective factors were identified. Both ‘frozen’ models achieved a c-statistic of 0.89 when prospectively tested on the external NC sample.
Conclusions - With large cohorts, a model based on ICD-9-CM codes performs as well as a model supplemented with drug exposure information. Both sources of data, however, provide valuable information regarding long-term mortality risk/protective factors. These models provide a tool to Medicaid programs and health service researchers to stratify or control for varying levels of comorbidities based on ICD-9-CM codes and/or drug exposure for ambulatory Medicaid populations.
Learning Objectives: At the conclusion of the session, the participant will be able to recognize the relative value of drug exposure and ICD-9-CM code-based information in predicting long-term survival of ambulatory Medicaid populations.
Keywords: Risk Factors, Medicaid
Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: None
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.