137 - Evaluating a Novel Antibiogram Format for use in Wisconsin Nursing Homes

Background: Nursing homes (NHs) increasingly use antibiograms to track antibiotic-related outcomes and guide antibiotic choice. Creation of a facility-specific antibiogram is hampered by low number of cultures collected in NHs. A weighted-incidence syndromic combination antibiogram (WISCA) is an alternative approach that may provide more stable estimates of antibiotic activity. In this study, we compare traditional antibiograms and WISCAs in a sample of Wisconsin NHs.

Methods: We created urine-specific antibiograms using traditional and WISCA approaches at facility and regional levels using culture data collected in study NHs from 01/01/2018 – 12/31/2018. Susceptibility results were standardized across laboratories using CLSI breakpoints. Traditional antibiograms were deemed reliable when ≥ 20 isolates were recovered for at least one species and species exceeding this threshold comprised 75% of all isolates. WISCAs were deemed reliable if ≥ 20 urinary isolates were recovered. Bootstrapped regional mean susceptibilities and confidence intervals for traditional antibiograms and WISCAs were calculated. Susceptibilities calculated at the facility-level were compared to regional estimates. Facility-level susceptibility estimates were deemed concordant if within 1 SD, moderately discordant if between 1 and 2 SDs, and severely discordant if greater than 2 SDs of the regional estimate.

Results: 462 urine isolates were obtained from 23 NHs in 2 regions. None of the facility-specific traditional antibiograms met reliability criteria. 10 of 23 facility-specific WISCAs were reliable and increased to 19 of 23 when 2-years of microbiology data were utilized (table). Severe discordance between facility-specific and regional estimates was identified with 62/107 NH species-antibiotic means and 98/119 NH urine isolate-antibiotic means falling outside of 2 SD of corresponding bootstrap regional susceptibility means (figure).Table. Reliability analysis of facility-specific urinary WISCAs and traditional antibiograms. 2-year projection was created using the assumption of similar culture results over 2-years.
Figure. Proportion of concordant, moderately discordant, and severely discordant NH mean susceptibilities in comparison to bootstrap regional mean susceptibilities for traditional antibiograms and WISCAs. NH mean susceptibilities from 5 isolates of more were included.


Conclusion: WISCAs are more reliable than traditional antibiograms for estimating antibiotic susceptibilities using facility-specific data. The high degree of discordance observed between facility-specific and regional antibiograms raises concerns about pooling culture data from multiple facilities.