Methods: A cluster-randomized intervention trial was conducted in 12 community-based NHs located in Southeast Michigan from May 2010 to May 2013. Four hundred and eighteen residents (six control sites, n=215 and six intervention sites, n=203) with an indwelling urinary catheter and/or enteral feeding tube were enrolled. This nested prospective longitudinal study focuses on NH residents with more than one follow-up visit from the six control facilities (n=162). Patient-specific clinical and demographic data— including information on incident infections, antibiotic use, and results of diagnostic tests— were obtained at study enrollment, after 14 days, and monthly thereafter for up to one year. Predictors of diagnostic testing were evaluated using multilevel random effects logit models.
Results: More than half (61.7%) of NH residents received antibiotics over 857 visits. Incident infections occurred in 27.7% of follow-up visits. While antibiotics were used in 97.4% of visits in which an infection occurred, they were also used in 17.0% of visits without documented clinical or laboratory evidence of infection. Besides presence of infection-specific symptoms and using multilevel random effects model, a change in functional status was the most significant predictor for ordering a urinalysis to detect UTI (OR = 3.24; P = .020) and a chest X-ray to detect pneumonia (OR = 3.53; P = .013). Empiric antibiotic use before having laboratory data occurred in 21.5% of suspected UTIs and 30.2% of suspected pneumonias.
Conclusion: Change in functional status is a significant predictor of ordering laboratory testing to diagnose infections. Antibiotic use remains persistent after negative tests and should be a target for future interventions.
M. A. Rogers, None
L. Mody, VA Healthcare System Geriatric Research Education and Clinical Care Center, NIA-Pepper Center, NIA R01AG032298, and R01AG041780: Grant Investigator , Grant recipient and Research grant