Demonstration of a ProcessOutcome Link for Smoking Cessation Melissa M. Farmer, PhD1,2 Elizabeth M. Yano, PhD1,2 Brian S. Mittman, PhD1,2 Scott E. Sherman, MD, MPH1,3 1VA HSR&D Center of Excellence for the Study of Healthcare Provider Behavior; UCLA Schools of Public Health2 and Medicine3 AcademyHealth Annual Meeting • Boston • June 2005 Background: Smoking Cessation (SC) Nearly 25% of all Americans are smokers SC has one of the lowest treatment delivery rates of all preventive services. Many SC interventions work – NRT and/or buproprion – SC programs – Brief provider tobacco counseling Background: Why the VA? Smoking is more prevalent among veterans. VA is the largest health care system in U.S. – Currently ~6 million VA users: ~2 million smokers Increased SC guideline adherence was among VA’s first national performance goals – Substantial improvements in detecting and counseling smokers QUITS Intervention Evidence-based quality improvement (QI) guideline implementation study at 18 Southwestern VA’s in 1998-2002 Intervention did not result in higher quits or quit attempts More research needed on contribution of setting and the process of care Funded by VA Health Services Research & Development Service Objectives Examine the relationship between discrete site-level process-of-care measures and clinical outcomes for smoking cessation (30-day abstinence) in a representative sample of VA primary care patients Sample and Methods Site-level data: Primary/Ambulatory care manager survey (n=18) – Assessment, counseling, referral to SC programs, treatment & resources available Patient-level data: Telephone survey data from QUITS sample of veterans smokers – Baseline (n=1941), 12 month follow-up (n=1038) and 18 month follow-up (n=885). – Sociodemographics, health status, smoking status Analytic Strategy Logistic regression with clusteradjustment at the site-level to examine process-quality measures as predictors of successful smoking cessation outcomes – controlling for patient characteristics, baseline health and smoking status. Facility Characteristics (n=18) Variable Assessment of smoking status at visit % or Mean/SD 41% Counseled at visit* 2.17 (0.54) Referral to specialized SC clinic/program* 1.90 (0.84) Medications prescribed w/out restrictions 31% * 3 point scale: 1=less than half, 2=most, 3=all or almost all patients Facility Characteristics (n=18) Variable % Site measures assessment performance 19% Site measures counseling performance 25% Routine follow-up for those counseled 44% Staff reminders to assess smoking status 81% Patient Population (n=1038) Variable % Age (22-91) 57.18 Race: White 64% Self rated health (1=poor-5=excellent) 2.49 Had made quit attempt at baseline 45% Results Has not smoked cigarette in past 30 days Abstinence at 12 month follow-up % 9% Abstinence at 18 month follow-up 10% Logistic Model for 30-day abstinence 12-month N=1038 OR (95% CI) 18-month N=885 OR (95% CI) Assessment 2.10(1.15-3.83) 1.87(1.02-3.46) Counsel 1.61(1.01-2.57) 1.58(1.06-2.35) Referral 1.28(1.03-1.59) 1.08(0.86-1.36) Measures assessment 2.38(1.43-3.95) 2.03(1.35-3.03) Summary of Results: At 12- months increased odds of abstinence at sites that report more frequent assessment, counseling and referrals. At 18-months, odds were further increased by sites that report more frequent assessment and counseling. Those sites that monitored their own performance for tobacco use assessment had higher odds of abstinence at 12 and 18-months. Limitations Results limited to QUITS sample of facilities and patients Future research needed on the measurement of process-of-care Conclusions and Implications Site-level process-of-care measures predict clinical outcomes for smoking cessation. The demonstration of a process-outcome link between smoking cessation performance measures and veterans’ actual cessation supports VA’s process measurement as a valid marker for ultimately helping veterans quit smoking.