1030 — A Longitudinal Analysis of the Effects of Using Data from One System for Dual VA and Medicare Users with Dementia

Zhu CW (James J. Peters Bronx VAMC), Penrod J (James J. Peters Bronx VAMC), Dellenbaugh C (James J. Peters Bronx VAMC), Ross J (James J. Peters Bronx VAMC), Sano M (James J. Peters Bronx VAMC)

Objectives:
To compare changes over time in diagnoses and comorbidities in VA and Medicare data for dual users with dementia.

Methods:
We used the National Longitudinal Caregiver Survey of male veterans age 65 and older with a formal diagnosis of Alzheimer’s disease or vascular dementia in the VA in 1997 to create a longitudinal sample (1998-2001) of VA and Medicare dual users alive at year 4 (669 patients with 2,676 observations). We obtained all ICD-9-CM diagnosis codes (truncated to 3 digits) from the VA National Patient Care Database and Medicare Standard Analytic Files. We constructed an Elixhauser comorbidity index for each patient using VA data only, Medicare data only, and VA and Medicare data combined.

Results:
From combined VA and Medicare data, patients averaged 21.6 unique diagnosis codes per year: 19.5 were identified in only one system, 2.1 were identified in both. In 1998, on average, 10.6 unique diagnosis codes per patient were identified in VA data, 8.8 in Medicare data. By 2001, this distribution had reversed: 10.7 unique diagnosis codes were identified in Medicare data, 8.7 in VA data. Between 1998 and 2001, percentage of diagnoses captured using VA data alone decreased significantly from 62.7% to 55.7% (p < 0.001); in contrast, percentage of diagnoses captured using Medicare data alone increased significantly from 60.7% to 67.2% (p < 0.001). Between 1998 and 2001, Elixhauser comorbidity index increased significantly using both combined data (3.6 vs. 3.9; p=0.017) and Medicare data alone (2.2 vs. 2.6; p < 0.001), but remained stable using VA data alone (~2.2 each year).

Implications:
Using VA data alone for VA and Medicare dual users with dementia underestimated the total number of diagnoses and comorbidities, the magnitude of which worsened over time.

Impacts:
Not taking into account information generated by veterans obtaining healthcare outside the VA underestimates the VA’s assessment of the burden of disease and healthcare needs of this population. Strategies are needed to routinely monitor veterans obtaining care outside the VA to ensure that the VA has adequate resources to provide necessary care to all veterans.