Identifying prevalence of cardiovascular conditions in women vetrans with varying lengths of observation period

Background: Women veterans enrolled for VHA care has been increasing since 1992. However, women veterans still make up a small minority of the patients seeking care. Because of their numerical minority status and unique nature of their health problems, it is important to assess identification rates of cardiovascular conditions (CVC) from administrative data with varying observation periods.

Objective: To estimate the rates of identification of CVC and analyze differences in identification over various follow up periods in a cohort of women veteran clinic users in fiscal year (FY) 2000 and FY2001.

Methods: Analysis of 189,307 women VHA users with diabetes, hypertension, and heart disease derived from the FY2000 and FY2001 VHA administrative data and Medicare claims files. One inpatient or two outpatient ICD-9-CM codes indicative of CVC were used to identify women with CVCs. Chi-square tests, logistic and multinomial logistic regressions were used to analyze subgroup differences.

Results: Of the 189,307 women veterans, 57.9% were identified with CVC with 12 months, 65.2% with 18 months and 70.3% with 24 months observation period. Of the women with CVC, 82% women were identified within 12 months. Likelihood of identification increased with length of observation for some subgroups; whites, younger, not married individuals and those without psychiatric disorders were less likely to be identified using shorter lengths of observation period.

Conclusions: Subgroup variations by length of observation period suggest that studies using administrative data to identify CVC among women veterans need to be cautious in selecting the lengths of observation period.