Evaluating the Impact of Urine Reflex Testing on Antibiotic Prescribing for Urinary Tract Infections
Tigran Khachatryan, Judy San, Kimi Pontoppidan, Hasmig EkmekdjianAbstract
Purpose
Urinary tract infections (UTIs) are among the most common indications for outpatient antibiotic prescribing. Despite clinical guidelines promoting appropriate antibiotic use, unnecessary urine cultures and treatment of asymptomatic bacteriuria continue to contribute to adverse drug reactions and antimicrobial resistance. To improve diagnostic stewardship, Kaiser Permanente Southern California (KPSC) region implemented urinalysis (UA) with reflex to microscopy and urine culture (UC), in which microscopy and culture are automatically performed when predefined UA parameters are positive. This strategy aims to improve diagnostic accuracy and reduce inappropriate antibiotic use.
Methods
This retrospective, pre- and post-implementation observational study evaluated the impact of urine reflex testing on outpatient antibiotic prescribing for suspected UTI across KPSC. Patients were included if they were KPSC members aged 2 years with either a UA or UC collected between July 1, 2021 and June 30, 2025. Patients who were pregnant, immunocompromised, had an indwelling catheter, or were hospitalized were excluded.
Results
Over 3.7 million outpatient encounters were analyzed. Overall antibiotic prescribing within 7 days remained similar (23.2% vs. 23.4%; p<0.0001). However, prescribing was more aligned with culture results, with decreased antibiotic use following negative cultures (25.0% to 23.3%; p<0.0001) and increased use following positive cultures (75.9% to 77.6%; p<0.0001). Antibiotic switched within 7 days of index encounter increased (9.1% to 10.4%; p<0.0001). Diagnostic testing patterns shifted with UA, stand-alone UC, and positive UC decreased, while urine reflex testing increased.
Conclusion
Implementation of UA with reflex testing standardized diagnostic workflows and improved alignment of antibiotic prescribing with microbiologic findings.