Predicting Antibiotic Susceptibility Among Patients with Recurrent Urinary Tract Infection Using a Prior Culture.

Abstract

Recurrent cystitis guidelines recommend relying on a local antibiogram or prior urine culture to guide empirical prescribing, yet little data exist to quantify the predictive value of a prior culture. We constructed a urinary antibiogram and evaluated test metrics (sensitivity, specificity, and Bayes' positive and negative predictive values) of a prior gram-negative organism on predicting subsequent resistance or susceptibility among patients with uncomplicated, recurrent cystitis. We performed a retrospective database study of adults with recurrent, uncomplicated cystitis (cystitis occurring 2 times in 6 months or 3 times in 12 months) from urology or primary care clinics between 11/1/2016 and 12/31/2018. We excluded pregnant females, patients with complicated cystitis, or pyelonephritis. Test metrics were calculated between sequential, paired cultures using standard formulas. We included 597 visits from 232 unique patients wherein 310 (51.2%) visits had a urine culture and 165 had gram-negative uropathogens isolated. Patients with gram-negative uropathogens were mostly females (97%), with a median age of 58.5. Our antibiogram found 38.0%, 27.9%, and 5.5%, of Escherichia coli isolates had resistance to trimethoprim-sulfamethoxazole, ciprofloxacin, and nitrofurantoin, respectively. Prior cultures (within 2 years) had good predictive value for detecting future susceptibility to first-line agents nitrofurantoin (0.85) and trimethoprim-sulfamethoxazole (0.78) and excellent predictive values (≥0.90) for cefepime, ceftriaxone, cefuroxime, ciprofloxacin, levofloxacin, gentamicin, tobramycin, piperacillin-tazobactam, and imipenem. Considerable antibiotic resistance was detected among E. coli isolates in patients with recurrent, uncomplicated cystitis. Using a prior culture as a guide can enhance the probability of selecting an effective empirical agent.