Reply to Johnson

SIR—We thank Dr. Johnson [1] for his interest in and comments on our study. To determine outcomes of relapse or reinfection we used definitions commonly applied in clinical trials of urinary tract infection (UTI). We agree that isolation of the same species after therapy may possibly result from reinfection with the same strain from an external source. This has been well documented for young women with acute uncomplicated UTI [2], although similar observations for complicated UTI have not been reported, to our knowledge. Despite this, reinfection rates identified with the study definitions were similar for both treatment arms, whereas relapse occurred only in the 3-day treatment arm. Surveillance cultures of sites of colonization, the likely source of same-strain reinfection, did not reveal persistent colonization in either arm after treatment. These observations support the conclusion that outcomes classified as relapse did represent relapse rather than reinfection.

PFGE typing was only performed for organisms associated with late relapse. The expectation was that relapse would usually be identified early after therapy, so late relapse, of which there were only 2 occurrences, would more likely represent reinfection. The majority of study subjects were inpatients on the spinal cord injury unit and exposed to nosocomial pathogens, and Acinetobacter species are well recognized causal organisms in this situation. Thus, reinfection with Acinetobacter anitratus is certainly “statistically probable” [3]. Three Acinetobacter isolates from the patient from our study [4] with late relapse—1 isolated 2 days before initiation of antimicrobial therapy, 1 immediately before therapy, and another at week 6 of therapy—were analyzed by PFGE. The 2 pretherapy isolates were similar, but the third isolate was clearly distinct, using standard criteria for interpretation. Although we cannot comment on the suggestion of genetic reassortment, PFGE typing of Acinetobacter species has been widely used in epidemiologic studies and is considered a valid method [3].

The prostate gland is certainly a potential source for relapsing UTI in men. There would be no specific way to identify this localization. The study by Ulleryd et al. [5] describes febrile UTI, so it is not applicable to our study population. As noted in the article, a previous study of UTI localization in a population of men with spinal cord injuries reported that almost 50% of subjects with bacteriuria had occult upper urinary tract involvement [6]. In fact, a 2-week course of therapy is likely not sufficient to eradicate prostatic bacteriuria, so the absence of relapses in the 14-day group suggests that the prostate was not a common source of infection.

Acknowledgments

Potential conflicts of interest.All authors: no conflicts.