Background: In 2016, an academic detailing (AD) intervention took place in 115 nursing homes in British Columbia. AD meetings, attended by physicians, nurses, and nursing home staff, were organized to reduce unnecessary antibiotic treatment of urinary tract infections (UTI), and in particular, asymptomatic bacteriuria. Meta-analysis of AD indicates general effectiveness for creating small prescribing changes; however, there are no large-scale evaluations of AD for nursing home antimicrobial stewardship (AMS).
Methods: UTI-linked prescriptions for nursing home residents were extracted from Pharmanet, an administrative database of prescriptions dispensed in community pharmacies. Changes in the days of supplied (DOS) prescriptions were assessed with an ecologic interrupted time series analysis. 82 local health areas (LHAs) were included with 50 intervention LHAs (61%). The study period was June 2015 to March 2017 and the intervention began on July 2016. Multilevel segmented regression was used for statistical analysis.
Results: During the study period, 9,822 residents received 23,141 UTI-linked prescriptions. Intervention and control had an overall average of 101 and 15 DOS, respectively. Both intervention and control had a decreasing pre-intervention trend (average of -1.4 and -0.2 DOS per month, respectively). While the expected post-intervention rate for the intervention group was -1.1 [-1.8, -0.3] DOS per month, the observed trend was -2.8 [-2.8, -0.7] DOS per month; 169.9% lower than expected [-59.7%, 663.7%]. The controls average post-intervention trend was unchanged, -0.1 [-0.6, 0.2] DOS per month. For the intervention group, there were 4,714 [1,921, 6,113] fewer days of UTI prescriptions in the intervention period.
Conclusion: In this pragmatic ecologic evaluation, AD was associated with reductions in UTI-coded antibiotic prescribing. The lack of large-scale AMS studies in nursing homes has hindered AMS implementation in this setting. Thus, these preliminary results address a key gap in the AMS literature. Further evaluation of this intervention with a multiple baseline design is warranted.
J. Sutherland, None
M. Maclure, None
D. Patrick, None