Community-acquired febrile urinary tract infection in diabetics could deserve a different management: a case control study

Transcription

1 Journal of Internal Medicine 2003; 254: Community-acquired febrile urinary tract infection in diabetics could deserve a different management: a case control study J. P. HORCAJADA 1, I. MORENO 1, M. VELASCO 1, J. A. MARTÍNEZ 1, A. MORENO- MARTÍNEZ 1, M. BARRANCO 2, J. VILA 2 & J. MENSA 1 Services of 1 Infectious Diseases and 2 Microbiology, Institut Clínic Infeccions i Immunologia (ICII), Hospital Clínic Universitari-IDIBAPS, Barcelona, Spain Abstract. Horcajada JP, Moreno I, Velasco M, Martínez JA, Moreno-Martínez A, Barranco M, Vila J, Mensa J (Hospital Clínic Universitari- IDIBAPS, Barcelona, Spain) Community-acquired febrile urinary tract infection in diabetics could deserve a different management: a case control study. J Intern Med 2003; 254: Objective. To investigate if there are relevant differences in clinical, microbiological and outcome characteristics of community-acquired febrile urinary tract infection (UTI) between diabetic and nondiabetic patients. Design. A prospectively matched case control study. Setting. An 800-bed tertiary care universityaffiliated hospital. Subjects. A total of 108 patients (54 diabetic and 54 nondiabetic patients matched by age and gender) admitted between January 1996 and September 1999 with febrile UTI. Methods. Clinical, analytical, microbiological and outcome variables were analysed by means of McNemar test (categorical) or Wilcoxon matched pairs signed rank test (continuous). Results. Mean age (SD) in both groups was 67.9 (14.4) years. In comparison with controls, diabetic patients were more likely to have fever without localizing symptoms (27% vs. 9%, P ), diminished consciousness level at admission (25% vs. 10%, P ¼ 0.03), aetiological microorganism different from Escherichia coli (17% vs. 0, P ¼ ), and quinolone-resistant bacteria (17% vs. 3.7%, P ¼ 0.07). Duration of fever after the onset of treatment was 1.75 (1) days in diabetics and 1.5 (1.1) days in nondiabetics (P ¼ 0.17). However, diabetic patients had a longer hospitalization [5.2 (3.3) days] than nondiabetics [3.9 (2.6) days, P ¼ 0.006]. Conclusions. In diabetic patients, febrile UTIs have clinical and microbiological peculiarities that may have diagnostic and therapeutic implications. Keywords: case control study, diabetes, febrile urinary tract infection. Introduction Urinary tract infections (UTI) is of concern in diabetic patients because of their higher incidence and tendency to cause severe complications, such as emphysematous pyelonephritis and abscess formation [1, 2]. Although it is believed that the clinical presentation of acute pyelonephritis is similar in diabetic and nondiabetic patients [2], some indirect data, such as the higher rates of hospitalization because of pyelonephritis in diabetics, suggest that, besides the eventual need of hospitalization for metabolic control, some differences in clinical presentation may actually incline the physician s attitude towards patient admission [3]. Up to now, little information describing the impact of diabetes on the characteristics and evolution of UTI has been generated, and studies analysing possible differences with nondiabetics are scarce [4]. In patients with UTI, diabetes has been found to be a risk factor for multi-drug resistant uropathogens [5]. Currently, quinolones are considered by some experts to be the first-line empiric antimicrobial therapy for diabetics with UTI [2]. However, during 280 Ó 2003 Blackwell Publishing Ltd

2 FEBRILE UTI IN DIABETICS 281 the last years, there has been a significant increase in quinolone-resistant uropathogenic Escherichia coli strains isolated from patients with UTI in several countries [6, 7] and therapeutic failures have been reported [8, 9]. Hence, a better understanding of the clinical presentation and microbiology of febrile UTI in diabetics could improve the current management of these frequent and potentially serious infections in this patient population. In order to compare the clinical, microbiological and outcome characteristics of community-acquired febrile UTI in patients with and without diabetes, a case control study was conducted. Patients and methods Patients Clinical, microbiological and outcome characteristics of febrile UTI in diabetic versus nondiabetic patients were studied and compared by means of a prospectively matched case control study. Cases were defined as diabetic patients with febrile UTI consecutively admitted to the infectious diseases ward of a tertiary care university affiliated hospital between January 1996 and September Controls were selected in a 1 : 1 ratio among 590 nondiabetic patients with febrile UTI admitted during the same period, using gender and exact age as matching criteria. When more than one control could be matched to a given case, random selection was made. Patients with previously documented urinary tract abnormalities, recent (within last month) urinary tract instrumentation, and use of antibiotics or immunosuppressant therapy during the previous month were excluded. The following clinical data were recorded: type of diabetes, current medication, antecedents of cystitis, pyelonephritis, nephritic colic and previous urinary incontinence, and clinical data of the current episode at admission, including duration of symptoms before admission, presence of lower urinary tract symptoms, chills, vomiting, lumbar tenderness, prostatic tenderness at rectal examination, consciousness state and presence of septic shock. The following laboratory tests were obtained on admission: serum creatinine, white blood cell count and differential, C-reactive protein, erythrocyte sedimentation rate, urinalysis, urine culture and two blood cultures. Urine samples were collected using the clean-catch midstream method. Urine specimens were platted onto MacConkey agar and cystine lactose electrolyte deficient (CLED) agar and incubated aerobically at 37 C. The isolated microorganisms were identified to the species level using standard methods. Blood samples were inoculated into aerobic and anaerobic BACTEC blood culture bottles and placed in the BACTEC 9240 blood culture instrument (BACTEC 9240; Beckton Dickinson Diagnostic Instrument Systems, Sparks, MD, USA). Identification of positive cultures was performed using conventional methods. An abdominal ultrasound was performed whenever there was evidence of shock or acute renal failure, suspicion of renal obstruction or other urological abnormality or persistence of fever for more than 3 days after the onset of adequate antimicrobial therapy. All patients received empiric treatment with a parenteral third generation cephalosporin until susceptibility testing was available. After that, and upon improvement of patient s clinical condition, an appropriate oral antibiotic was prescribed to complete a 2-week therapy cycle. The following outcome data were recorded: duration of fever, lumbar tenderness and lower urinary tract symptoms after the onset of antimicrobial therapy, length of hospitalization, and bacteriological and clinical recurrences 1 month after the acute episode. Definitions. Febrile UTI was defined by an armpit temperature of >38 C (fever) with pyuria (>8 leucocytes hpf )1 ) and a urine culture growing >10 5 cfu ml )1 of a single microorganism, in the absence of other infectious source. Acute pyelonephritis was defined as a febrile UTI with lumbar or flank tenderness. Acute prostatitis was defined by the presence of lower urinary tract symptoms and either the finding of a tender prostate at rectal examination or a prostatic specific antigen blood level >10 ng ml )1. Localizing symptoms are referred to symptoms that localize the source of fever, i.e. the kidney (lumbar tenderness) or the prostate (prostatic tenderness in men). Diminished consciousness was considered when the patient had a diminished or absent response to verbal and/or painful stimuli. Patients were considered to be afebrile when armpit temperature remained below 37 C during more than 24 h. Bacteriological recurrences were defined as the isolation of an uropathogen in the follow-up urine culture performed 1 month after the acute episode. Diabetes mellitus was defined according to standard criteria

3 282 J. P. HORCAJADA et al. [10]. Type 1 diabetes was defined as the deficiency of insulin secretion [11]. Type 2 diabetes was defined as the combination of resistance to insulin action and an inadequate compensatory insulin secretory response [11]. Patients were considered as having type 1 or 2 diabetes based on their medical history. The study was performed following the standards of the local ethics committee. Statistical analysis The comparison of categorical, dichotomous variables between cases and controls was carried out by the McNemar test, using the odds ratio with their corresponding 95% confidence interval as the measure of association. Continuous variables were compared by the Wilcoxon matched pairs signed rank test. P < 0.05 was considered to be significant. Results A total of 64 diabetic patients with febrile UTI were admitted during the study period. Of these, 10 did not meet the inclusion criteria and were excluded (three because of recent urinary tract instrumentation, two because of current use of indwelling urinary catheter, and five as a result of recent antibiotic exposure). Fifty-four patients (16 with type 1 and 38 with type 2 diabetes) met the inclusion criteria and could be appropriately matched to 54 control patients. Mean age was 67.9 (14.4) years and there was a predominance of women. A history of urinary incontinence was significantly more frequent in diabetic (45%) than in control (24%) patients (P ¼ 0.02). There were no differences in the other recorded antecedents (Table 1). At presentation, case patients were more likely to have fever without focal symptoms (27% vs. 9% in controls, P < ) and a diminished consciousness level (25% vs. 10% in controls, P ¼ 0.03). Only one diabetic and two nondiabetic patients had shock at admission. There were no differences also in the other recorded clinical data (Table 1). Among the 15 studied diabetic men, three (20%) had acute prostatitis, four (27%) had acute pyelonephritis and eight (53%) had febrile UTI. Among the 15 studied nondiabetic men, seven (47%) had acute prostatitis, three (20%) had acute pyelonephritis and five (33%) had febrile UTI (P ¼ 0.12 for acute prostatitis and 0.26 for febrile UTI). At admission, cases and controls did not differ by serum creatinine, total leucocyte count, number of band neutrophils, erythrocyte sedimentation rate or serum C-reactive protein (Table 2). Escherichia coli grew in all urine cultures from nondiabetic patients, and in 45 (83%) of the 54 diabetics. In the other nine diabetic patients (17%), a microorganism different from E. coli was isolated (P ¼ ). These pathogens included Klebsiella pneumoniae (three patients), Citrobacter spp. (two patients), Proteus mirabilis (two patients), Diabetes (n ¼ 54) Nondiabetes (n ¼ 54) P values* Age (years), mean (SD) 67.9 (14.4) 67.9 (14.5) 1 Sex: women/men 39/15 39/15 1 Previous renal colic, n (%) 9 (17) 10 (18) 1 Previous cystitis, n (%) 27 (54) 19 (38) 0.15 Previous pyelonephritis, n (%) 8 (15) 15 (29) 0.14 Urinary incontinence, n (%) 24 (45) 13 (24) Days of any symptom before admission, mean (SD) 3.6 (3.4) 4.2 (4) 0.45 Days of fever before admission, mean (SD) 1.5 (1.5) 1 (0.8) 0.23 Lower urinary tract symptoms, n (%) a 25 (67) 30 (81) 0.22 Lumbar tenderness, n (%) 30 (60) 35 (70) 0.35 Absence of localizing symptoms, n (%) 20 (37) 3 (5) < Vomits, n (%) a 15 (43) 10 (28) 0.30 Chills, n (%) 47 (88) 48 (87) 0.14 Shock, n (%) 1 (2) 2 (4) 1 Diminished consciousness level, n (%) b 11 (24) 4 (8.7) 0.03 Table 1 Antecedents and clinical presentation of febrile urinary tract infection (UTI) in the studied population *Exact McNemar significance probability for qualitative data. Wilcoxon matched pairs signed ranks test for continuos variables. a Calculated over 37 pairs. b Calculated over 46 pairs.

4 FEBRILE UTI IN DIABETICS 283 Table 2 Analytical, microbiological and outcome data in the studied population Diabetes (n ¼ 54) Nondiabetes (n ¼ 54) P values* Serum creatinine (mg dl )1 ), mean (SD) 1 (0.32) 1.1 (0.43) 0.83 Leucocyte count ( 10 9 L )1, mean (SD) (5385) (7139) 0.75 Blood immature forms (%), mean (SD) 8.8 (4.9) 8.4 (7.5) 0.79 C-reactive protein (mg dl )1 ), mean (SD) 14 (10.5) 15 (8.7) 0.55 Erythrocyte sedimentation rate (mm h )1 ), mean (SD) 78 (33.4) 66 (27.9) 0.05 Non-E. coli in the urine culture*, n (%) 9 (17) 0 (0) Quinolone resistant uropathogens, n (%) 9 (17) 2 (3.7) 0.07 Bacteraemia, n (%) 32 (59) 23 (43) 0.09 Duration of fever after antibiotic was started 1.75 (1) 1.5 (1.1) 0.17 (days), mean (SD) Duration of low urinary tract symptoms after antibiotic 1.2 (1.6) 1.4 (1.1) 0.54 was started (days), mean (SD) Duration of lumbar tenderness, mean (SD) 1.1 (1.3) 1.2 (1.1) 0.94 Hospital stay (days), mean (SD) 5.2 (3.3) 3.9 (2.6) Bacteriological recurrence 1 month later, n (%) a 15 (41) 9 (21) 0.22 Clinical recurrence 1 month later, n (%) a 5 (14) 4 (9) 0.21 *Klebsiella pneumoniae (3), Citrobacter spp. (2), Proteus mirabilis (2), Staphylococcus aureus (1), and S. agalactiae (1). a Calculated over 33 pairs. Staphylococcus aureus (one patient) and Streptococcus agalactiae (one patient). Thirty-two of the 54 diabetics (59%) had bacteraemia versus 23 (43%) of the nondiabetics (P ¼ 0.09). In control patients, bacteraemia was always the result of E. coli, while in three patient cases a microorganism other than E. coli was involved (Citrobacter sp., S. aureus and S. agalactiae each in one patient). In the subgroup of 45 case control pairs with E. coli UTI, bacteraemia occurred in 28 (62%) diabetic and 23 (51%) nondiabetic patients (P ¼ 0.28). In nine diabetic (17%) and in two (3.7%) nondiabetic patients, a quinolone-resistant microorganism was isolated from urine (P ¼ 0.07) (Table 2). Abdominal ultrasound was performed in 43 patients, 17 diabetics and 26 nondiabetics. Among diabetics, the reasons for performing an abdominal ultrasound were: suspicion of renal obstruction or other urological abnormalities in 12 (70%) patients, persistence of fever for more than 3 days after the onset of antibiotic therapy in two (12%) patients, septic shock in one (6%) patient and undefined in two (12%) patients. Abdominal ultrasound findings in diabetics were as follows: without abnormal findings in eight (47%) patients, nonobstructive renal litiasis in two (12%) patients, renal obstruction in three (18%) patients, and pyelocaliceal dilatation compatible with acute pyelonephritis in four (23%) patients. Among nondiabetic patients the indications for requesting an abdominal ultrasound were: suspicion of renal obstruction or other urological abnormalities in 18 (69%) patients, persistence of fever for more than 3 days after the onset of antibiotic therapy in one (4%) patient, septic shock in two (8%) patients, acute renal insufficiency in one (4%) patient, and undefined in four (15%) patients. Abdominal ultrasound findings in nondiabetics were as follows: without abnormal findings in 15 (58%) patients, nonobstructive renal litiasis in three (11%) patients, renal obstruction in three (11%) patients, pyelocalicial dilatation compatible with acute pyelonephritis in two (8%) patients, congenital anatomical abnormality in one (4%) patient, and renal cysts in two (8%) patients. Complications, such as emphysematous pyelonephritis, necrotizing papilitis or renal abscess, were not detected in any of the studied patients. After the onset of antimicrobial therapy, there were no differences in the mean duration of fever, lumbar tenderness or lower urinary tract symptoms. Diabetic patients had a significantly longer hospitalization [5.2 (3.3) days] than nondiabetics [3.9 (2.6) days, P ¼ 0.006]. None of the patients studied died. Bacteriological and clinical recurrence rates 1 month after the acute episode were similar in both groups (Table 2). Discussion In this study, the clinical characteristics of community-acquired febrile UTI in diabetic and nondiabetic

5 284 J. P. HORCAJADA et al. patients were compared. It was found that diabetics were more likely to have a previous history of urinary incontinence, clinical presentation of fever without localizing symptoms and diminished consciousness level, longer hospitalization, higher incidence of non-e. coli uropathogens and a trend towards a higher quinolone-resistance rate among the uropathogens isolated from this patient population. Previous reports have indicated less differences between diabetics and nondiabetics with UTI, but those reports focused on all forms of UTI, including patients with asymptomatic bacteriuria and nosocomial UTI [4]. In an attempt to assess the differences truly caused by diabetes, we focused on communityacquired febrile UTI in patients without obvious risk factors that could increase the likelihood of UTI and the development of complications or resistance to antimicrobials. The clinical differences observed in our study have not been previously reported. The absence of localizing symptomatology could be related to diabetic neuropathy and could be the cause of delayed diagnosis and more severe clinical state. The diminished consciousness level could reflect a more severe septic state, metabolic disturbances associated with diabetes or cerebral hypoperfusion which seems to be more frequent in diabetics [12]. However, we cannot attribute this putative marker of organ dysfunction to a delayed diagnosis as shown by the similar duration of any symptom of febrile UTI before admission in diabetic and nondiabetic patients. As noted in previous studies [13, 14, 15], we also observed a trend towards a higher prevalence of bacteraemia in diabetic patients. Bacteraemia may indicate an intrinsic inability from diabetics to contain the microorganisms at the infectious site which, in turn, may increase the severity of febrile UTI. Therefore, diabetes by itself seems to make a difference in terms of clinical presentation of febrile UTI. In our series, it was observed that, although the duration of fever and urinary symptoms after starting therapy was similar in both groups, the duration of hospitalization was significantly longer in diabetics than in nondiabetics. Although there is not accurate explanation for this finding, we suspect that diabetic patients required a slightly longer hospitalization (1.3 days on average) than nondiabetics either to reach an adequate metabolic control of their underlying disease or to recover from a more severe infection. Some defects in leucocyte function and a lower pro-inflammatory cytokine production in diabetics could contribute to these clinical differences [16, 17]. The degree of glycosuria and metabolic control have been proposed as risk factors for an increased likelihood of UTI in diabetics, however, some studies did not find them to be major contributors [18, 19]. To the contrary, confirmed factors related to an increased frequency of UTI in diabetics are a high prevalence of urinary tract anatomical defects and neurogenic bladder (27 85%) [18, 20, 21]. This is consistent with our finding that diabetics had a significantly higher prevalence of urinary incontinence than nondiabetics, probably reflecting a higher prevalence of neurogenic bladder. It should be noted that in our series non-e. coli uropathogens were isolated more frequently in diabetics than nondiabetics. Previous series have also observed this in women with diabetes versus nondiabetic women [22, 23]. Antimicrobials and urinary tract instrumentation, which tend to be more frequent in diabetics, and neurogenic bladder and other anatomical defects, could be related to this finding. However, in our study we excluded patients with instrumentation within the previous month and those treated with antimicrobials in the previous month in order to avoid their influence in the aetiology of UTI. This is probably because neurogenic bladder is an independent element influencing the aetiology of UTI in diabetics. Moreover, quinolone-resistant uropathogens tended to be isolated more frequently in diabetics. This could be related to previous antibiotic cycles, although we cannot demonstrate that, because we did not record whether patients had taken antimicrobials before the previous month. However, as it has been reported recently, quinoloneresistant E. coli selected by previous therapy with quinolones in the faecal flora are displaced by quinolone-susceptible strains in about 2 months after finishing therapy. Beyond this point, the possibility of having an UTI produced by quinolone-resistant E. coli strains seems to be less likely [24]. Urinary incontinence is more frequent in diabetics. This could be one cause of the presence of more quinolone-resistant uropathogens in diabetics with UTI as quinolone-resistant E. coli seems to have a decreased invasive capacity and is isolated more frequently in patients with anatomical or

6 FEBRILE UTI IN DIABETICS 285 functional abnormalities or manipulations of the urinary tract [25]. The high prevalence of quinolone-resistant uropathogens in diabetics could have therapeutic implications. It has been recommended that individual episodes of UTI in diabetics should be treated in the same way as in nondiabetics [21, 26]. Currently, quinolones are the first choice empiric antimicrobial therapy in diabetics with UTI [2]. However, our findings suggest that, in Spain, more than 15% of community-acquired febrile UTI in diabetics can be caused by uropathogens resistant to quinolones. Therefore, there is a question of whether these patients should be empirically treated with quinolones when they have febrile UTI in areas where quinolone-resistant E. coli strains are prevalent in the community. The strength of our study relies on the strict selection of controls in order to avoid the influence of other risk factors in the comparison performed. On the downside, a limitation of this study is that we did not record baseline characteristics of diabetes (years of evolution, glycosylated haemoglobin levels, presence of macro-microangiopathy). However, it is improbable that the objective of the study is influenced by this lack of information. Another limiting factor is that we did not follow patients beyond 1 month after the acute episode. Recurrence rates could have been higher in diabetics later, as it has been previously reported [21]. Our results have two possible implications. First, they support a high level of suspicion of febrile UTI in diabetics with fever without localizing symptoms. Secondly, they suggest that empirical antibiotic therapy in diabetics with febrile UTI could be different than in nondiabetics, because of the more frequent involvement of non-e. coli uropathogens in diabetics and the trend towards a higher quinoloneresistance rate in the uropathogens isolated in this patient population, especially in areas where quinolone-resistant E. coli strains are prevalent in the community. Conflict of interest statement No conflict of interest was declared. References 1 Patterson JE, Andreoli VT. Bacterial urinary tract infections in diabetes. Infect Dis Clin North Am 1997; 11: Joshi N, Caputo GM, Weitekamp MR, Karchmer AW. Infections in patiens with diabetes mellitus. N Engl J Med 1999; 341: Nicolle LE, Friesen D, Harding GKM, Roos LL. Hospitalization for acute pyelonephritis in Manitoba, Canada, during the period from 1989 to 1992: impact of diabetes, pregnancy, and aboriginal origin. Clin Infect Dis 1996; 22: Bonadio M, Meini M, Gigli C, Longo B. Urinary tract infection in diabetic patients. Urol Int 1999; 63: Wright SW, Wrenn KD, Haynes M, Haas DW. Prevalence and risk factors for multidrug resistant uropathogens in ED patients. Am J Emerg Med 2000; 18: Garau J, Xercavins M, Rodriguez-Carballeira M et al. Emergence and dissemination of quinolone-resistant Escherichia coli in the community. Antimicrob Agents Chemother 1999; 43: Goettsch W, van Pelt W, Nagelkerke N et al. Increasing resistance to fluoroquinolones in Escherichia coli from urinary tract infections in the Netherlands. J Antimicrob Chemother 2000; 46: Deguchi T, Kawamura T, Yasuda M et al. In vivo selection of Klebsiella pneumoniae strains with enhanced quinolone resistance during fluoroquinolone treatment of urinary tract infections. Antimicrob Agents Chemother 1997; 41: Nakano M, Yasuda M, Yokoi S, Takahashi Y, Ishihara S, Deguchi T. In vivo selection of Pseudomonas aeruginosa with decreased susceptibilities to fluoroquinolones during fluoroquinolone treatment of urinary tract infection. Urology 2001; 58: Wahl PW, Savage PJ, Psaty BM, Orchard TJ, Robbins JA, Tracy RP. Diabetes in older adults: comparison of 1997 American Diabetes Association classification of diabetes mellitus with 1985 WHO classification. Lancet 1998; 352: Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Diabetes Care 2002; 25 (Suppl. 1): S Meyer JS, Rauch G, Rauch RA, Haque A. Risk factors for cerebral hypoperfusion, mild cognitive impairment, and dementia. Neurobiol Aging 2000; 21: MacFarlane IA, Brown RM, Smyth RW, Burdon DW, FitzGerald MG. Bacteraemia in diabetics. J Infect 1986; 12: Carton JA, Maradona JA, Nuno FJ, Fernandez-Alvarez R, Perez-Gonzalez F, Asensi V. Diabetes mellitus and bacteraemia: a comparative study between diabetic and non-diabetic patients. Eur J Med 1992; 1: Bryan CS, Reynolds KL, Metzger WT. Bacteremia in diabetes patients: comparison of incidence and mortality with nondiabetic patients. Diabetes Care 1985; 8: Delamaire M. Impaired leucocyte functions in diabetic patients. Diabet Med 1997; 14: Geerlings SE, Brouwer EC, van Kessel KC, Gaastra W, Stolk RP, Hoepelman AI. Cytokine secretion is impaired in women with diabetes mellitus. Eur J Clin Invest 2000; 30: Zhanel GC, Nicolle LE, Harding GKM and the Manitoba Diabetic Urinary Infection Study Group. Prevalence of asymptomatic bacteriuria and associated host factors in women with diabetes mellitus. Clin Infect Dis 1995; 21: Ingberg CM, Palmer M, Schvarcz E, Aman J. Prevalence of urinary tract symptoms in long-standing type 1 diabetes mellitus. Diabetes Metab 1998; 24:

7 286 J. P. HORCAJADA et al. 20 Kaplan SA. Vesical disfunction. In: Lebovitz HE, ed. Therapy for Diabetes Mellitus and Related Disorders. Alexandria: American Diabetes Association, 1994; Forland M. The treatment of urinary tract infections in women with diabetes mellitus. Diabetes Care 1985; 8: Lye WC, Cahn RK, Lee EJC, Kumarasinghe G. Urinary tract infections in patients with diabetes mellitus. J Infect 1992; 24: Geerlings SE, Stolk RP, Camps MJL et al. Asymptomatic bacteriuria may be considered a complication in women with diabetes. Diabetes Care 2000; 23: Horcajada JP, Vila J, Moreno-Martínez A et al. Molecular epidemiology and evolution of resistance to quinolones in Escherichia coli after prolonged administration of ciprofloxacin in patients with prostatitis. J Antimicrob Chemother 2002; 49: Velasco M, Horcajada JP, Mensa J et al. Decreased invasive capacity of quinolone-resistant Escherichia coli in patients with urinary tract infections. Clin Infect Dis 2001; 33: Warren JW, Abrutyn E, Hebel JR, Johnson JR, Schaeffer AJ, Stamm WE. Guidelines for antimicrobial treatment of uncomplicated acute bacterial cystitis and acute pyelonephritis in women. Clin Infect Dis 1999; 29: Received 18 October 2002; revision received 3 April 2003; accepted 28 April Correspondence: Juan P. Horcajada, Service of Infectious Diseases, Institut Clínic Infeccions i Immunologia (ICII), Hospital Clínic Universitari-IDIBAPS, Villarroel 170, Barcelona, Spain (fax: ; jhorcaja@clinic.ub.es).