1002 — Measuring the Quality of Care for Alcohol Misuse: Limitations of Screening-Based Measures
Bradley KA (HSR&D Northwest Center for Outcomes Research in Older Adults (NW COE), VAPSHCS; University of Washington), Johnson ML
(HSR&D NW COE, VAPSHCS), Lapham GT
(HSR&D NW COE, VAPSHCS; University of Washington), Williams EC
(HSR&D NW COE, VAPSHCS; University of Washington), Achtmeyer C
(HSR&D NW COE, VAPSHC), Rubinsky AD
(HSR&D NW COE, VAPSHCS; University of Washington), Hawkins EJ
(HSR&D NW COE, VAPSHCS; University of Washington), Saitz R
(Boston University), Kivlahan DR
(HSR&D NW COE, VAPSHCS; University of Washington)
Screening-based performance measures evaluate the quality of care among patients who screen positive on validated screening questionnaires. The VA healthcare system uses screening-based measures for follow-up of alcohol misuse, depression, and PTSD. This method of performance measurement assumes that screening identifies comparable samples of patients across health systems being compared. This study evaluated a screening-based performance measure of brief intervention and referral to treatment (BI/RT) for alcohol misuse.
This cross-sectional study included outpatients whose medical records were reviewed for the VA Office of Quality and Performance (7/1/2006-3/31/2009) who had AUDIT-C alcohol screening documented (positive screen > = 5). VA networks were compared regarding: prevalence of positive screens; proportion of screen-positive patients with alcohol-related diagnoses in the previous year (“recognized” misuse) as a measure of severity; proportion of screen-positive patients with BI/RT documented (“screening-based measure of BI/RT”); and the number of patients with positive screens and BI/RT documented per 100,000 screened (“standardized number identified and offered BI/RT”).
The prevalence of positive alcohol screens [4.9% (95% CI 4.3-5.5%) to 11.2% (10.3-12.1)] and proportion of screen-positive patients with recognized misuse [36.1% (32.5-39.7) to 54.4% (48.2-60.5)] varied across networks. Screen-positive patients with recognized misuse were more likely to have BI/RT documented: mean 62.8 (46.1-77.8%) versus 35.1 (18.1-59.8%). As a result of such differences, screening-based measures of BI/RT did not reflect the standardized number of patients identified and offered BI/RT. Networks with similar rates of BI/RT on the screening-based measure (40-48%) had 2-fold differences in the standardized number of patients identified and offered BI/RT (2055-4948/100,000 screened). Networks with similar standardized numbers of patients identified and offered BI/RT (3563 and 3584/100,000 screened) had markedly different screening-based rates of BI/RT: 67.9% (63.8-72.0) and 36.0% (32.8-39.3), respectively. Results were unchanged after adjustment for demographics and regional drinking patterns.
The prevalence and severity of alcohol misuse identified by clinical alcohol screening differed across networks despite use of a common screening measure. As a result, a screening-based performance measure did not allow valid comparisons of quality of care across networks.
VA should consider alternative performance measures for alcohol misuse follow-up.