3099 — Inpatient Quality Indicators: Comparison of Facility-Level Rates Across the VA

Borzecki AM (COE - Bedford), Chew PW (COE- Bedford), Loveland SA (COE - Bedford), Loya PM (COE - Bedford), Hartmann CW (COE - Bedford), Rosen AK (COE - Bedford)

Objectives:
The Agency for Healthcare Research and Quality (AHRQ) Inpatient Quality Indicators (IQIs) are evidence-based measures that screen for potential inpatient quality problems using administrative data. They measure procedure volume (counts), utilization (rates), and in-hospital procedure and condition-specific mortality (rates). Despite extensive non-VA use for quality improvement, IQIs have received little VA attention. This study examined: 1)the feasibility of applying the IQIs to VA data and 2)variation in counts and rates across facilities and years.

Methods:
We used FY03-06 VA inpatient files and AHRQ IQI software (v3.1) to examine facility-level IQI counts and ratios of observed (unadjusted) to expected rates (risk-adjusted using age, gender and APR-DRGs), as appropriate for 23 IQIs, across facilities and years.

Results:
Rarer volume IQIs varied across facilities (e.g., pancreatic resection: annual range 1-9, median 2), fluctuating slightly by facility over time; more common procedures such as coronary bypass (facility range 1-325, median 130) showed an annual decrease in total VA-performed procedures (5,806-5,328 in FY03 vs FY06). Observed facility mortality rates were highest for complex procedures such as pancreatic resection (FY03 mean 0.24+/-0.44) but decreased annually (FY06 mean 0.15+/-0.42). Overall annual observed to expected ratios (O/Es) were < 1.0 (i.e., rate < expected). However, 2 of 55 facilities had O/Es > 1.0 at least 2 of 4 years. For more common procedures, observed mortality rates decreased annually (e.g., coronary bypass: FY03 mean 3.98+/-2.69 to 2.62+/-2.17 in FY06); all facilities had O/Es < 1.0 across years. VA-wide mortality rates from conditions such as stroke were fairly stable over time with O/Es < 1.0. In any given year there were a few outlier facilities (O/Es > 1.0) but no consistent trend. Utilization indicators showed the most facility variation, e.g., incidental appendectomy (FY03 facility mean 1.08+/-1.87 per 100 laparotomies). Several facilities had O/Es > 1.0 across all years.

Implications:
Facility-level mortality rates were lower than expected over time for all medical condition-related IQIs, and all but the most complex procedure-related IQIs. Rates of potentially inappropriate procedure use were more likely to exceed expected rates with certain facilities consistently higher over time.

Impacts:
Preliminary analyses suggest that, especially with respect to potentially inappropriate procedure use, on-site facility examination may provide opportunities for provider training and cost-saving.