This Article  • References  • Full text PDF  • Send to a friend  • Save in My Folder  • Save to citation manager  • Permissions  Citing Articles  • Citation map  • Citing articles on HighWire  • Contact me when this article is cited  Related Content  • Similar articles in this journal  Social Bookmarking   What's this?

Leonard Leibovici, MD; Shai Greenshtain, MD; Ohad Cohen, MD; Arjeh J. Wysenbeek, MD

Arch Intern Med. 1992;152(12):2481-2486.

Abstract
Background.—
Guidelines to show whether a patient hospitalized because of a urinary tract infection (UTI) has a severe infection, and whether he or she is at high risk for harboring a multiresistant pathogen, are scant. The aims of the present study were to find (1) clinical and laboratory variables known within 24 hours of admission that, com bined in a logistic model, will point to a high or low prob ability of bacteremia and (2) variables that can be used to define patients at high risk for the subsequent isolation of a multiresistant uropathogen.

Methods.—
In a set of patients consecutively admitted to a department of medicine because of UTI, we compared bacteremic vs nonbacteremic patients, and patients with a multiresistant uropathogen vs others, on logistic regression analysis. The logistic models derived were validated in a second set of patients with UTI.

Results.
Among 247 patients with UTI (median age, 75 years), 80 of them with bacteremia, five factors were signif icantly and independently associated with bacteremia: se rum creatinine level, leukocyte count, temperature, diabe tes mellitus, and low serum albumin level. A logistic model incorporating those factors was used to divide the patients into three groups with increasing prevalence of bacteremia (6%, 39%, and 69%) and of death (3%, 6%, and 20%). Three factors were predictive of the subsequent isolation of a resistant uropathogen: use of antibiotics before admission, advanced age, and male gender. The combination of those factors was used to divide patients into two groups, with resistance to cefuroxime of 9% vs 28%, to gentamicin of 7% vs 20%, and to sulfamethoxazole-trimethoprim of 30% vs 50%. In a second set of 144 patients with UTI, the per centages of bacteremia in the three groups were 5%, 16%, and 55%, and those of death, 2%, 6%, and 17%. When di vided by the second model, the resistance to cefuroxime in the two groups was 16% vs 30%; to gentamicin, 16% vs 28%; and to sulfamethoxazole-trimethoprim, 28% vs 59%.

Conclusion.
If prospectively validated in other set tings, the models can be used to define groups of patients with UTI at low and high risk for bacteremia, and to help in the choice of empiric antibiotic treatment.

(Arch Intern Med. 1992;152:2481-2486)

Author Affiliations
From the Department of Medicine B, Beilinson Medical Center, Pe tah Tiqva, Israel; and the Sackler Faculty of Medicine, Tel Aviv (Israel) University.

Footnotes
Accepted for publication February 28, 1992.

Reprint requests to Department of Medicine B, Beilinson Medical Center, Petah Tiqva 49100, Israel (Dr Leibovici).

CiteULike    Connotea    Del.icio.us    Digg    Reddit    Technorati    Twitter     What's this?

THIS ARTICLE HAS BEEN CITED BY OTHER ARTICLES

Acute Pyelonephritis in Adults: Prediction of Mortality and Failure of Treatment
Efstathiou et al.
Arch Intern Med 2003;163:1206-1212.
ABSTRACT | FULL TEXT  

The Clinical Host Response to Microbial Infection in Medical Patients With Fever
Bossink et al.
Chest 1999;116:380-390.
ABSTRACT | FULL TEXT