1. Nursing Homes as Data Collection Sites Philip D. Sloane, MD, MPH University of North Carolina at Chapel Hill ABT team: Rosanna Bertrand, PhD; Lauren Olsho; Louise Hadden; Alrick Edwards. UNC team: Sheryl Zimmerman, PhD; Anna Beeber, PhD, GNP, RN; Christine Kistler, MD, MASc; C. Madeline Mitchell, MURP Funding provided by AHRQ contract # HHSA 290200600001I. Nursing homes are complex systems with a unique culture

2. Prior Research Experience UNC Team (CS-LTC) UNC Team (CS-LTC) Nursing home research: Over 25 funded NH studies over 20+ years involving primary data collection 7 NH intervention trials Infection research: 2 NH infection studies ABT Team Nursing home research: Over 20 years of NH QI research Over 30 federally-funded intervention/evaluation studies Infection research: Antibiogram use in NHs Multiple US and international HIV studies

3. Goals and Study Design GOALS: Implement & evaluate a multicomponent QI program on optimizing antibiotic prescribing Assess the validity of the Loeb Minimum Criteria DESIGN: LTC provider group / 2 geographically separate regions One region assigned to intervention (6 NHs) Other assigned to comparison (6 NHs) 9 months chart abstraction (3m baseline, 6m follow-up) QI program (months 4–9) Provider and NH staff training (Pocket Card/Referral Form) Brochure / meetings for residents/families Monthly data reports / meetings with facility QI teams

4. Percent Change in Abx Prescribing from Baseline to Follow-Up Intervention group: 26.8% reduction Comparison group: 5.3% reduction P = 0.036 Main Results

5. Using NH Records as Data Sources Many NHs pre-electronic; some have EHR Physician notes: –If electronic, access limited –If paper, quality often poor Nursing notes: –Often quite limited; length, quality, and consistency inferior to hospital setting Assessment data consistent, illness data not Our attempt to introduce a standardized reporting tool for infections  limited acceptance

6. Collecting Primary Data in NHs No major administrative barriers encountered: NH staff cooperative Data collection staff must be clinical (we used RNs), well trained, and supervised Records can be massive; audit time per record was longer than we’d anticipated Lack of systems for acute illness made standardization of data collection difficult