2006 HSR&D National Meeting Abstract
3013 — Impact of Self-Reported Social/Health Data on Facility Level Ranking of Improved Glycemic Control
Maney M (VANJHCS)
Tseng C (VANJHCS)
Safford M (Birmingham VAMC)
Miller D (CHQOER,Bedford VAMC)
Pogach L (VANJHCS)
Barriers to optimal glycemic control include lower educational, social, health, and economic status. Our objective was to evaluate the impact of self-reported sociodemographic and health status domains upon facility level performance in achieving good (<8% A1c) or excellent (<7%A1c) glycemic control.
Administrative records of VHA clinical users identified as having diabetes in FY99 were linked to the Large Veterans Health Survey. We used linear regression models to adjust individual FY00 A1c levels for variables reflecting socioeconomic circumstances (education level; employment; economic hardship, i.e., concerns over food sufficiency); social support (marital status; whether living alone); health behaviors (smoking; alcohol use; exercise level); health status (physical and mental component summary scores from the Short Form-36 veterans version; body mass index; duration of diabetes). We determined facility outliers in the proportion of veterans achieving <8% or <7% A1c with and without risk adjustment.
There were 56,900 subjects; 105 facilities had at least 100 subjects [111 to 2298] and used certified A1c methodology. There were substantial facility level variations in significant variables; for example, only grade school education (mean 15.4%, [range 2.3%-32.7%]); retired ( 38.3%, [19.9% -59.7%]); food sufficiency (13.9%, 7.0%- 25.6%); no reported exercise ( 43.2%, [31.1%-53.6%]); married (65.2,% [43.7%-77.8%]; and mental health function (43.8%, [38.7%-48.9%]). The final model had an R-squared of 10.1%. When facilities were ranked on achieving A1c <7%, 5 of the 11 best performing facilities changed at least 3 deciles with risk adjustment, and 1 of 11 of the worst facilities changed more than two deciles. Using the <8% threshold resulted in 6 of 11 best facilities changing at least 3 deciles, and 1 of 11 of the worst facilities changing more than two deciles.
There are marked facility level variations in socioeconomic circumstances, social support, health behaviors, and health status. Adjustment of A1c using self-reported information on these domains had a marked impact upon assessment of which facilities were best achieving improved glycemic control for their patients.
Lower A1c thresholds should not be used for comparative purposes without adjustment for socio-economic factors, health status, and health behaviors. Interventions to improve glycemic control should consider social circumstances.