Alcohol misuse is common, and implementation of evidence-based alcohol screening and brief intervention (BI) in medical settings is a high prevention priority. Experts agree that valid measures of BI are needed in order to evaluate and incentivize BI implementation. BI quality measures that evaluate rates of BI in a defined population of patients have therefore been developed based on clinical guidelines and expert opinion. The Veterans Affairs Healthcare System (VA), Joint Commission, and American Medical Association have each endorsed different BI quality measures, all based on documentation of BI by medical providers ("1st generation BI quality measures"). However, other organizations that develop healthcare quality measures have concluded that there is insufficient evidence on the validity of documented BI to endorse 1st generation BI quality measures. In addition, some BI quality measures may unintentionally reward healthcare systems for lower quality clinical alcohol screening that identifies fewer patients. This study aims to: 1) estimate the sensitivity and specificity of provider-documentation as a measure of BI and evaluate variability in sensitivity and specificity across 21 VA networks, using patient-report on surveys as the criterion standard;and 2) evaluate whether better performance on 1st generation BI quality measures is associated with lower quality clinical alcohol screening. This study uses unique secondary data from the nationwide VA Healthcare System. The study sample includes VA patients who received care at any of the 139 facilities in the 21 VA networks (1/2008-1/2011). Patients will be eligible if they screened positive for alcohol misuse on a confidential mailed survey within 12 months of having clinical alcohol screening documented in their VA medical records (n ~23,915).
Aim 1 analyses will estimate the sensitivity and specificity of BI documented in the medical record compared to patient report of BI on surveys and evaluate whether there are differences in the sensitivity and specificity across the 21 VA networks.
Aim 2 analyses will evaluate the association between facility-level performance on 1st generation BI quality measures and the quality of documented clinical alcohol screening, defined as the proportion of patients who screened positive for alcohol misuse on surveys who also screened positive when screened in VA clinical settings. All analyses will be adjusted for patient-level covariates and account for correlated data. BI quality measures will play a central role in the implementation of alcohol screening and BI. However, 1st generation BI quality measures that rely on documentation by providers may have critical limitations and may not encourage high quality alcohol screening and BI. Several organizations are therefore planning to develop "2nd generation" BI quality measures. Results of this study evaluating 1st generation BI quality measures will provide essential knowledge to inform development of evidence-based 2nd generation BI quality measures that incentivize high quality alcohol screening and BI.
Implementation of alcohol screening and brief alcohol interventions (BIs) is a high US prevention priority and experts agree that valid measures of the quality of brief interventions (BI quality measures) will be required for successful implementation. However, there are concerns about the validity and undesirable consequences of current 1st generation BI quality measures. The proposed study evaluates 1st generation BI quality measures using unique data from the Veterans Affairs (VA) Healthcare System to inform the development of evidence-based 2nd generation measures.
|Bradley, Katharine A; Chavez, Laura J; Lapham, Gwendolyn T et al. (2013) When quality indicators undermine quality: bias in a quality indicator of follow-up for alcohol misuse. Psychiatr Serv 64:1018-25|