Kalpana Gupta

International Clinical Practice Guidelines for the Treatment of Acute Uncomplicated Cystitis and Pyelonephritis in Women: A 2010 Update by the Infectious Diseases Society of America and the European Society for Microbiology and Infectious Diseases

A Panel of International Experts was convened by the Infectious Diseases Society of America (IDSA) in collaboration with the European Society for Microbiology and Infectious Diseases (ESCMID) to update the 1999 Uncomplicated Urinary Tract Infection Guidelines by the IDSA. Co-sponsoring organizations include the American Congress of Obstetricians and Gynecologists, American Urological Association, Association of Medical Microbiology and Infectious Diseases-Canada, and the Society for Academic Emergency Medicine. The focus of this work is treatment of women with acute uncomplicated cystitis and pyelonephritis, diagnoses limited in these guidelines to premenopausal, non-pregnant women with no known urological abnormalities or co-morbidities. The issues of in vitro resistance prevalence and the ecological adverse effects of antimicrobial therapy (collateral damage) were considered as important factors in making optimal treatment choices and thus are reflected in the rankings of recommendations.

Increasing antimicrobial resistance and the management of uncomplicated community-acquired urinary tract infections.

Uncomplicated community-acquired urinary tract infections (UTIs) are among the most common infections in women, accounting for more than 8 million office visits per year in the United States as well as significant morbidity and health care costs (1). Current management of these infections is usually empirical, without the use of a urine culture or susceptibility testing to guide therapy. The rationale for this approach is based on the narrow and predictable spectrum of etiologic agents that cause acute cystitis and their susceptibility patterns (2). However, as with many community-acquired infections, antimicrobial resistance among the pathogens that cause community-acquired UTIs is increasing (3-6). As the problem of antimicrobial resistance becomes more widespread, the use of narrow-spectrum, inexpensive antimicrobial agents becomes less feasible, affecting both the cost of and access to health care for patients. In addition, such infections as uncomplicated community-acquired UTI, which have traditionally been readily treatable, are now becoming therapeutic challenges. Perhaps the most significant change in resistance among uropathogens has been the increase in the prevalence of resistance to trimethoprimsulfamethoxazole (TMPSMX), the current drug of choice in the United States for empirical therapy for uncomplicated UTI in women. In addition, TMPSMX resistance has been associated with concurrent resistance to other antibiotics, resulting in multidrug-resistant uropathogens (7, 8). However, although several reports have focused on changing patterns of in vitro resistance of uropathogens to TMPSMX, the clinical significance of this resistance has not been well studied. Therefore, implications for health care providers are not yet clear. This review highlights the problem of antimicrobial resistance in acute uncomplicated community-acquired UTI, focusing on TMPSMX resistance, and summarizes the few available data regarding clinical outcomes associated with in vitro resistance. Finally, we outline recommendations for empirical treatment of uncomplicated UTI in a period of evolving antimicrobial resistance. Etiologic Agents in Uncomplicated Community-Acquired UTIs Although susceptibility patterns have changed, the spectrum of agents causing community-acquired UTI has remained relatively constant. Escherichia coli accounts for 75% to 90% of cases; Staphylococcus saprophyticus accounts for 5% to 15% (particularly in younger women); and enterococci and nonE. coli aerobic gram-negative rods, such as Klebsiella species and Proteus mirabilis, account for the remaining 5% to 10% (2, 9). Although less well studied, the spectrum of agents causing uncomplicated pyelonephritis is similar to that causing acute cystitis (10, 11). Pharmacologic Issues The antimicrobial agents used to treat uncomplicated community-acquired UTIs include the -lactams, TMPSMX, nitrofurantoin, fosfomycin, and the fluoroquinolones. All of these agents achieve high urinary concentrations, usually greatly exceeding the expected serum levels (Table 1). Of note, susceptibility breakpoints from the National Committee for Clinical Laboratory Standards are based on serum rather than urine concentrations of these antimicrobial agents, except for nitrofurantoin and fosfomycin, which are exclusively used for treatment of cystitis (12). Table 1. Pharmacokinetic Characteristics of Selected Antimicrobial Agents Used To Treat Community-Acquired Urinary Tract Infections The aminopenicillins, ampicillin and amoxicillin, and most cephalosporins are rapidly excreted into the urine and attain high urinary concentrations (Table 1) (12-15). The peak serum and urinary concentrations of amoxicillin are higher than those achieved with a similar dose of ampicillin (12). In the 1970s, ampicillin was commonly used for treatment of acute cystitis. However, because of increasing in vitro resistance, as well as lower efficacy and more adverse effects than are seen with other available UTI antimicrobial agents, the -lactams in general are no longer recommended for empirical UTI therapy (1). In certain settings, such as during pregnancy or when enterococci are suspected, ampicillin or amoxicillin may still be an appropriate choice for acute UTI (1). The combination of TMP and SMX has been shown to be synergistic against a variety of organisms, including such aerobic gram-negative rods as E. coli (16). This combination has been used for treatment of UTI for more than two decades, although the commonly used 3-day regimen has not been formally approved by the U.S. Food and Drug Administration. After a single oral dose of one double-strength tablet (TMP, 160 mg; SMX, 800 mg), the peak urine concentrations are approximately 35 and 3 to 4 times higher, respectively, than serum levels (Table 1). Trimethoprim also concentrates in prostatic tissue at levels 2 to 3 times higher than those found in serum (12). Nitrofurantoin is one of the oldest urinary antiinfective agents in use. The macrocrystalline formulation requires frequent dosing (every 6 hours); however, a modified monohydratemacrocrystal form delays gastric uptake and allows twice-daily dosing. Nitrofurantoin is used primarily for treatment of cystitis because it does not attain appreciable serum levels (Table 1). It is 90% renally excreted, and therefore the urine concentration is very high, making it an effective urinary anti-infective agent for most gram-positive and gram-negative uropathogens (17). Fosfomycin tromethamine is a phosphonic acid derivative that is newly licensed for treating uncomplicated cystitis caused by E. coli or Enterococcus faecalis (18). It is not approved for use for cystitis caused by S. saprophyticus or for treatment of pyelonephritis. It achieves very high concentrations in the urine and persists in the urine for more than 24 hours (Table 1). The fluoroquinolones each have individual pharmacodynamic properties (12, 19, 20). The first fluoroquinolones widely used for treatment of UTI, namely norfloxacin, ciprofloxacin, ofloxacin, and levofloxacin, all have excellent bioavailability and achieve high urinary concentrations (Table 1). Their penetration into prostatic and renal tissue is also excellent. Some of the newer fluoroquinolones, such as sparfloxacin and trovafloxacin, are not excreted in high urinary concentrations and thus should not be used for treatment of uncomplicated pyelonephritis or complicated UTI. The fluoroquinolones also have a significant postantibiotic effect against gram-negative organisms (12). In Vitro Susceptibility Data Many recent studies have reported the resistance profiles of uropathogens to antimicrobial agents commonly used to treat UTI (3-8). Much of this in vitro data comes from laboratory-based surveys that often do not define the sex, age, clinical syndrome, or location (inpatient vs. outpatient) of the patients from whom the urine specimens are collected. Therefore, the reported rates of resistance may vary depending on whether the study sample consists primarily of outpatients with uncomplicated UTIs or patients with complicated nosocomial UTIs. The data presented here focus on studies that clearly define the study sample as women with uncomplicated community-acquired UTI. Although susceptibility profiles vary by each specific organism and antimicrobial combination, some general trends have clearly emerged. Resistance of E. coli and other uropathogens to -lactams, such as ampicillin, and the first-generation cephalosporins has continued to increase in the past decade and now approaches 40% in most studies (3, 6). Most gram-negative uropathogens are still susceptible to the combination of amoxicillinclavulanate, but the expense and gastrointestinal side effects of this drug make it a less desirable choice for empirical treatment of uncomplicated UTI (2). Moreover, it has been suggested that the failure rate with this drug is high when the uropathogen is resistant to ampicillin but susceptible to amoxicillinclavulanate (22). Nitrofurantoin and the fluoroquinolones have retained in vitro activity against most E. coli isolates that cause uncomplicated community-acquired UTI (>99% in most studies) (Table 2). However, nitrofurantoin is less active against nonE. coli gram-negative rods (3, 6) and inactive against Proteus and Pseudomonas species. The fluoroquinolones have had consistently high activity against essentially all gram-negative uropathogens seen in women with uncomplicated community-acquired UTI but are active against only 60% to 70% of enterococci, depending on the study (3, 6). Both nitrofurantoin and the fluoroquinolones retain good in vitro activity against S. saprophyticus (6), although increased failure rates have been reported with the use of single doses of fluoroquinolones to treat S. saprophyticusrelated UTIs (1). Table 2. in Vitro Susceptibility of Escherichia coli from U.S. Studies of Urinary Tract Infection As mentioned, the most substantial change in resistance prevalence has been to TMPSMX. Resistance to TMPSMX among uropathogens in the community was relatively infrequent in the United States in the early 1990s (Table 2). At that time, McCarty and colleagues (9) conducted a multicenter trial of low-dose ciprofloxacin compared with standard-dose ofloxacin and TMPSMX for treatment of acute uncomplicated cystitis in women. They reported only a 7% prevalence of TMPSMX resistance among the E. coli isolates. The rates of resistance did not increase to levels that might compromise clinical effectiveness until the mid-1990s (Table 2). We conducted a cross-sectional survey of urine isolates from a well-defined sample of outpatient women in western Washington State who had acute uncomplicated cystitis. We found that the prevalence of TMPSMX resistance among E. coli was 9% in 1992 but had increased to 18% by 1996, the last year of the study (3). Of interest, in a California student health sample, Dyer and coworkers found that the prevalence of UTI isolate

Risk Factors for Recurrent Urinary Tract Infection in Young Women

To define host factors associated with an increased risk of recurrent urinary tract infection (RUTI), a case-control study was conducted in 2 populations: university women and health maintenance organization enrollees. Case patients were 229 women 18-30 years old with RUTIs; control subjects were 253 randomly selected women with no RUTI history. In a multivariate model, independent risk factors for RUTI included recent 1-month intercourse frequency (odds ratio [OR], 5.8; 95% confidence interval [CI], 3.1-10.6 for 4-8 episodes), 12-month spermicide use (OR, 1.8; 95% CI, 1.1-2.9), and new sex partner during the past year (OR, 1.9; 95% CI, 1.2-3.2). Two newly identified risk factors were age at first urinary tract infection (UTI) </=15 years (OR, 3.9; 95% CI, 1.9-8.0) and UTI history in the mother (OR, 2.3; 95% CI, 1.5-3.7). Blood group and secretor phenotype were not associated with RUTI. In young women, risk factors for sporadic UTI are also risk factors for recurrence. Two predictors suggest that genetic/long-term environmental exposures also predispose to RUTI.

Antimicrobial Resistance Among Uropathogens that Cause Community-Acquired Urinary Tract Infections in Women: A Nationwide Analysis

Current recommendations for empirical therapy for community-acquired urinary tract infection (UTI) in women hinge on knowledge of antimicrobial susceptibility patterns in the geographic region of the practitioner. We conducted a survey of antimicrobial susceptibilities of 103,223 isolates recovered from urine samples that were obtained in 1998 from female outpatients nationally and within 9 geographic regions in the United States. Resistance of Escherichia coli isolates to trimethoprim-sulfamethoxazole varied significantly according to geographic region, ranging from a high of 22% in the western United States to a low of 10% in the Northeast (P<.001). There were no clinically significant age-related differences in the susceptibility of E. coli to any of the study drugs, but the susceptibility to fluoroquinolones of non-E. coli isolates that were recovered from women who were aged >50 years was significantly lower than that of isolates recovered from younger women (P<.001). The in vitro susceptibility of uropathogens in female outpatients varies according to age and geographic region.

A Prospective Study of Asymptomatic Bacteriuria in Sexually Active Young Women

BACKGROUND Asymptomatic bacteriuria is common in young women, but little is known about its pathogenesis, natural history, risk factors, and temporal association with symptomatic urinary tract infection. METHODS We prospectively evaluated 796 sexually active, nonpregnant women from 18 through 40 years of age over a period of six months for the occurrence of asymptomatic bacteriuria (defined as at least 10(5) colony-forming units of urinary tract pathogens per milliliter). The women were patients at either a university student health center or a health maintenance organization. Periodic urine cultures were taken, daily diaries were kept, and regularly scheduled interviews were performed. Escherichia coli strains were tested for hemolysin, the papG genotype, and the ribosomal RNA type. RESULTS The prevalence of asymptomatic bacteriuria (the proportion of urine cultures with bacteriuria in asymptomatic women) was 5 percent (95 percent confidence interval, 4 percent to 6 percent) among women in the university group and 6 percent (95 percent confidence interval, 5 percent to 8 percent) among women in the health-maintenance-organization group. Persistent asymptomatic bacteriuria with the same E. coli strain was rare. Symptomatic urinary tract infection developed within one week after 8 percent of occasions on which a culture showed asymptomatic bacteriuria, as compared with 1 percent of occasions when asymptomatic bacteriuria was not found (P<0.001). Asymptomatic bacteriuria was associated with the same risk factors as for symptomatic urinary tract infection, particularly the use of a diaphragm plus spermicide and sexual intercourse. CONCLUSIONS Asymptomatic bacteriuria in young women is common but rarely persists. It is a strong predictor of subsequent symptomatic urinary tract infection.

Proton pump inhibitors and risk for recurrent Clostridium difficile infection.

BACKGROUND Proton pump inhibitors (PPIs) are widely used gastric acid suppressants, but they are often prescribed without clear indications and may increase risk of Clostridium difficile infection (CDI). We sought to determine the association between PPI use and the risk of recurrent CDI. METHODS Retrospective, cohort study using administrative databases of the New England Veterans Healthcare System from October 1, 2003, through September 30, 2008. We identified 1166 inpatients and outpatients with metronidazole- or vancomycin hydrochloride-treated incident CDI, of whom 527 (45.2%) received oral PPIs within 14 days of diagnosis and 639 (54.8%) did not. We determined the hazard ratio (HR) for recurrent CDI, defined by a positive toxin finding in the 15 to 90 days after incident CDI. RESULTS Recurrent CDI was more common in those exposed to PPIs than in those not exposed (25.2% vs 18.5%). Using Cox proportional survival methods, we determined that the adjusted HR of recurrent CDI was greater in those exposed to PPIs during treatment (1.42; 95% confidence interval [CI], 1.11-1.82). Risks among exposed patients were highest among those older than 80 years (HR, 1.86; 95% CI, 1.15-3.01) and those receiving antibiotics not targeted to C difficile during follow-up (HR, 1.71; 95% CI, 1.11-2.64). [corrected] CONCLUSIONS Proton pump inhibitor use during incident CDI treatment was associated with a 42% increased risk of recurrence. Our findings warrant further studies to examine this association and careful consideration of the indications for prescribing PPIs during treatment of CDI.

Patient-Initiated Treatment of Uncomplicated Recurrent Urinary Tract Infections in Young Women

Urinary tract infection (UTI) is an exceedingly common outpatient problem, accounting for more than 8 million office visits and

Inverse Association of H2O2-Producing Lactobacilli and Vaginal Escherichia coli Colonization in Women with Recurrent Urinary Tract Infections

1 billion in health care costs per year. Approximately 50% to 70% of women will have a UTI sometime during their lifetimes, and 20% to 30% of women will have recurrent episodes (1-3). Thus, safe and effective management strategies that have the potential to improve patient convenience and decrease costs are of considerable interest to patients, providers, and health care organizations. The most well-studied and commonly accepted approach to managing recurrent UTIs uses low-dose antimicrobial prophylaxis given postcoitally, three times per week, or daily (4-6). This strategy has been shown to be safe and highly effective, even after 5 years of use. However, studies of the natural history of recurrent UTI demonstrate substantial variability in the number of recurrences experienced per woman (range, 0.3 to 7.6 episodes per year) (7). In addition, recurrences often cluster in time. Thus, continuous prophylaxis may result in unnecessary antimicrobial use in women who have infrequent recurrences or clustered recurrences. An alternative strategy, namely patient self-diagnosis and self-treatment of recurrent UTIs, may decrease antimicrobial use and improve patient convenience. However, this strategy has been evaluated in only two previous studies, both in relatively small groups of older women referred to specialty clinics for management of recurrent UTI (8, 9). To establish the safety and feasibility of this approach in a larger and more generalized sample, we assessed the accuracy of self-diagnosis and the cure rates seen with self-treatment of UTIs in 172 women who had a history of recurrent infection and were attending a university-based primary care clinic. Methods Study Sample We studied women at least 18 years of age who were attending a university-based primary care clinic and had had at least two UTIs in the previous 12 months. Women were recruited through advertisements in the campus newspaper and through referrals by primary health care providers. Potentially eligible women were screened by the study coordinator and were invited to participate if eligible. At enrollment, all participants were asymptomatic, were not pregnant, were regularly using contraception if sexually active, and were under the care of a primary provider. Women with a known allergy to fluoroquinolones; a full-term pregnancy in the past 12 months; or a history of diabetes, hypertension, or renal disease were excluded. Study Procedures Eligible women were briefly interviewed, instructed on how to obtain a clean-catch urine specimen, and given a urine collection kit containing a sterile urine cup and wipes for perineal cleaning to take home. Participants also received six 200-mg ofloxacin tablets to take twice daily for 3 days if UTI symptoms developed. Those who enrolled in the study after levofloxacin, the L-isomer of ofloxacin, became available were instead given three 250-mg levofloxacin tablets to take once daily for 3 days. The women were instructed to self-initiate antimicrobial therapy after collecting a clean-catch urine sample if they developed new symptoms suggestive of a UTI. They were asked to use symptoms experienced with previous UTIs as a guide for self-diagnosis during the study period. Women were prospectively followed for up to 12 months. The only interactions with study staff occurred during telephone checks performed monthly to review study procedures and during follow-up visits after each UTI self-diagnosis and self-treatment event. Women who developed presumptive symptoms of a UTI collected and refrigerated a clean-catch urine sample and then self-initiated antimicrobial therapy. Participants were asked to deliver the refrigerated urine sample to the study nurse within 24 hours of collection. On days 10 and 30 after initiation of therapy, participants returned to the clinic for follow-up urinalysis and culture and a brief interview. They were also given a new urine collection kit and a refill of medication. Women who self-diagnosed a UTI but whose urine cultures were negative were evaluated for alternative diagnoses with urine testing for chlamydia by ligase chain reaction, a pelvic examination, vaginal wet mount, and repeated urinalysis and culture. If the work-up was nondiagnostic and UTI symptoms resolved after therapy, no further treatment was provided. If the symptoms persisted in the setting of a nondiagnostic work-up, the participant was referred to a primary care provider for further evaluation. Participants whose microbiologically confirmed UTIs did not resolve after 3 days of self-therapy were re-treated on the basis of results of a repeated urinalysis, results of previous microbiological tests, and the discretion of the study nurse, who consulted with study investigators as needed. Data on adverse events were collected by using an open-ended questionnaire after each self-treatment episode. Satisfaction with the study protocol and subjective reductions in symptom days and restricted-activity days were assessed by questionnaire after each self-diagnosis event. Written informed consent was obtained from all participants, and study procedures were approved by the University of Washington Human Subjects Review Committee. Laboratory Methods Pyuria was determined by using a hemocytometer counting chamber. Cultures were performed on clean-catch midstream urine specimens. Bacteria were isolated by using standard techniques, as described elsewhere (10). Significant bacteriuria was considered present when uropathogens (Escherichia coli, other aerobic gram-negative rods, enterococci, or Staphylococcus saprophyticus) were present in quantities of at least 102 colony-forming units (CFU)/mL. Group B streptococcus was considered a uropathogen if it was the predominant organism and was present in quantities of at least 105 CFU/mL. In vitro susceptibility testing was performed by using the KirbyBauer disc method (11). The urine ligase chain reaction for chlamydia was done as described elsewhere (12). Statistical Analysis The main outcomes of interest were the accuracy of self-diagnosis and the microbiological and clinical cure rates seen with self-treatment of each UTI event. Symptomatic episodes associated with bacteriuria ( 102 CFU/mL of a uropathogen on a pretreatment urine culture) were defined as definite UTIs, while symptomatic episodes without significant bacteriuria but with pyuria ( 10 leukocytes per high-power field) and no other established diagnosis were defined as probable UTIs. Definite (culture-positive) episodes of UTI were evaluated for clinical and microbiological cure rates after self-treatment. Clinical failure was defined as persistent symptoms requiring additional therapy. Microbiological failure was defined as isolation, at or before the 10-day follow-up visit, of at least 102 CFU of the original uropathogen per mL in symptomatic participants or at least 105 CFU of the original uropathogen per mL in asymptomatic participants. Women with both microbiological and clinical cure at 10-day follow-up who had less than 105 CFU of the original uropathogen per mL and were asymptomatic at day 30 were defined as having continued microbiological cure. Confidence intervals adjusted for multiple episodes per woman were calculated by using the method of Donner and Donald (13). Comparisons of cure rates between subgroups and of factors associated with accurate diagnosis were performed by using logistic regression analyses with generalized estimating equations to adjust for multiple observations per woman (14). Role of the Funding Source The study protocol was initiated and written by the authors. Ortho-McNeil Pharmaceutical funded the study in part and provided the 3-day packets of ofloxacin and levofloxacin but had no role in the design, conduct, analyses, or reporting of the study. Results One hundred seventy-two women were enrolled and prospectively followed for 2 to 12 months each, with a mean follow-up period of 242 days (approximately 8 months) per woman and total follow-up of 114 person-years. The demographic characteristics and genitourinary history of the study sample are shown in Table 1. The mean age was 23 years, and most women were never married, were white, and were sexually active. Eighteen percent of women had their first UTI before 16 years of age, and 11% had had more than 10 UTIs in their lifetimes. Approximately 20% of women reported a history of one or more sexually transmitted diseases, including trichomoniasis, herpes simplex, genital warts, chlamydial cervicitis, or pelvic inflammatory disease. None reported a history of syphilis or gonorrhea. Table 1. Baseline Characteristics of Study Women (n = 172) Accuracy of Patient Self-Diagnosis During the study period, 88 of the 172 women (51%) self-diagnosed at least one episode of UTI. These 88 women experienced a total of 172 symptomatic events, with a mean of two episodes per woman. Most symptomatic events occurred in the first 6 months after study enrollment (Figure 1). Pretherapy urinalysis and culture performed to confirm these presumed UTI episodes showed a uropathogen in 144 cases (84% [95% CI, 77% to 90%]), sterile pyuria in 19 cases (11% [CI, 6% to 16%]), and no pyuria or bacteriuria in 9 cases (5% [CI, 1% to 10%]) (Figure 2). Twenty-five of the 28 culture-negative symptomatic episodes were further evaluated for alternative diagnoses; 3 episodes of sterile pyuria were not evaluated because of protocol error. A positive chlamydia ligase chain reaction was identified in one woman with acute dysuric symptoms and sterile pyuria; she was treated with a 7-day course of doxycycline. None of the remaining 24 evaluable culture-negative episodes had alternative diagnoses confirmed; all responded clinically to self-treatment, and none required further therapy (Figure 2). The microbiologically confirmed episodes were considere

Risk factors associated with acute pyelonephritis in healthy women.

Women with recurrent urinary tract infection (UTI) often demonstrate persistent vaginal colonization with Escherichia coli. Since strains of lactobacilli that produce hydrogen peroxide inhibit the growth of E. coli, the absence of these strains may predispose to E. coli colonization and to UTI. To test this hypothesis, vaginal introital cultures were obtained from 140 women, 65 with recurrent UTI (case-patients) and 75 without (controls). Vaginal E. coli colonization was significantly more frequent in case-patients than controls (35% vs. 11%; P < .001) and in women without H2O2-positive lactobacilli than in women with (odds ratio [OR], 4.0; P = .01). Spermicide use was associated with greater risk of vaginal E. coli colonization (OR, 12.5; P < .001) and with absence of H2O2-positive lactobacilli (OR, 2.9; P = .04). The inverse association between H2O2-positive lactobacilli and vaginal E. coli colonization remained in case-patients after controlling for spermicide use (OR, 6.5; P = .02). Thus, absence of H2O2-positive lactobacilli may be important in the pathogenesis of recurrent UTI by facilitating E. coli introital colonization.

Emerging antibiotic resistance in urinary tract pathogens

Context Little information is available about risk factors for pyelonephritis among healthy, community-dwelling women. Contribution In a population-based casecontrol study of women with pyelonephritis 18 to 49 years of age, intercourse history variables, including frequency, new sexual partners, and spermicide use, were strongly associated with pyelonephritis. Personal and family histories of urinary tract infection, presence of diabetes, and stress incontinence were also associated with pylonephritis on multivariable analysis. Escherichia coli was the predominant infecting organism. Implications Risk factors for pyelonephritis were similar to those for acute and recurrent cystitis and asymptomatic bacteriuria, supporting the concept that pyelonephritis is usually caused by the ascent of organisms from the bladder. The Editors Acute pyelonephritis, a potentially severe infection of the upper urinary tract, is estimated to account for more than 250000 office consultations with physicians and nearly 200000 hospital admissions annually in the United States (1, 2). The vast majority of these infections occur in women, and most of these women are treated in ambulatory care settings (3). While numerous studies have evaluated factors predisposing to acute cystitis, most studies of pyelonephritis have been treatment studies or descriptive studies that focused on hospitalized patients (3-6). To date, pyelonephritis in community-dwelling, healthy adults has not been extensively investigated, and, to our knowledge, risk factors have not been evaluated. We thus undertook a population-based, casecontrol study to increase our understanding of the epidemiology of acute pyelonephritis in adult women 18 to 49 years of age. Specifically, we evaluated whether the risk factors that predispose women to acute cystitis also predispose them to pyelonephritis and whether additional exposures are associated with upper urinary tract infection. The estimated incidence of infection and, among case-patients, the infecting organisms and their antimicrobial susceptibility profiles were also of interest. Methods Study Setting and Participants We performed our study at Group Health Cooperative, Seattle, Washington, a mixed-model health maintenance organization. During study recruitment (April 2000 to October 2001), approximately 87000 of 475000 enrollees in Group Health Cooperative were women between 18 and 49 years of age (our target population). Group Health Cooperatives Human Subjects Committee reviewed and approved all study procedures. We selected potential cases of pyelonephritis by using the health plans computerized enrollment, ambulatory care, inpatient, and laboratory databases. Each month we selected all women 18 to 49 years of age who had received a diagnosis of acute pyelonephritis (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM], codes 590.1 [acute pyelonephritis, without lesion of renal medullary necrosis], 590.11 [acute pyelonephritis, with lesion of renal medullary necrosis], or 590.8 [pyelonephritis, unspecified]) and who had a positive urine culture (103 colonies of a urinary pathogen, as previously defined [7]). Because laboratory data were not available for some inpatients who were treated in affiliate hospitals, we also selected any additional inpatients with acute pyelonephritis as their primary reason for hospitalization. We reviewed 58 randomly chosen medical records of cases selected in this manner and found that 57 (98%) patients had a diagnosis of pyelonephritis, defined as a diagnosis of pyelonephritis recorded in the chart, at least 1 symptom or sign of pyelonephritis (fever, chills, severe back or flank pain, nausea or vomiting, or costovertebral angle tenderness), and antibiotic treatment consistent with pyelonephritis. Most reviewed patients also had other symptoms of urinary tract infection (UTI). We selected potential controls concurrently each month by sampling randomly from age-eligible women in Group Health Cooperative enrollment files; controls were frequency-matched to the case-patients by 5-year age groups. We sent potential participants a letter that described the study and invited them to participate in a telephone interview. Interviewers began calling 1 week thereafter. During telephone screening, we excluded women who were pregnant within 12 months of the reference date (the diagnosis date for case-patients and the midpoint of the sampling month for controls), who had ever received kidney dialysis, who had had problems with their bladder or kidneys that required surgery (such as urogenic bladder or reflux) or kidney stones that did not pass or were not removed before the reference date, or who were nonambulatory at the reference date. We excluded potential control participants who reported a diagnosis of pyelonephritis within 5 years of the reference date. We identified 386 potential case-patients during recruitment, 54 (14%) of whom were not eligible. Of the remaining 332 women, 242 (73%) agreed to participate and were interviewed, 46 (14%) declined to participate, and 44 (13%) could not be contacted. Of the 242 case-patients, 18 (7.4%) were hospitalized. We randomly selected 960 similar-age women from the enrollment database as potential controls, 109 (11%) of whom were ineligible. Of the 851 women remaining, 546 (64%) were interviewed, 163 (19%) declined, and 142 (17%) could not be contacted. Data Collection After providing oral consent, participants received a 15- to 25-minute structured interview that was conducted by using computer-assisted telephone interviewing software (Raosoft, Inc., Seattle, Washington), which allowed all responses to be entered directly into a computer database. We programmed item wording, skip patterns, and range checks into the instrument to minimize errors and standardize administration. We pretested the instrument for length, flow, and comprehension. After training in the study instrument (interviewers were already experienced in epidemiologic interviewing), the project manager monitored interviewers for accuracy and completeness during initial fielding by an audit telephone line and continued periodic monitoring throughout the study. The interview included items on demographic characteristics, sexual behavior, contraceptive practices, genitourinary infection history, history of other medical conditions, and other behaviors (contraceptive practices, personal and family genitourinary infection history, history of other health conditions, and other health habits). We also asked case-patients about signs and symptoms close to their index pyelonephritis episode and collected data on infecting organism and antimicrobial susceptibility from Group Health Cooperative automated laboratory files. Statistical Analysis We characterized the study group on selected variables of interest by casecontrol status, further characterizing the case group by symptoms and infecting organisms. We assessed data on potential risk factors of interest for their univariate association with the study outcome, pyelonephritis, by calculating odds ratios and 95% CIs. We developed multivariable logistic regression models to identify independent risk factors for pyelonephritis from univariately associated exposures and to evaluate their relative contributions. We considered variables for the models by incorporating exposures that were important in earlier studies of cystitis in this or similar populations and, in addition, considered other variables that seemed to be associated with pyelonephritis in our data set. When several variables were highly correlated, principally sexual activity and UTI history variables, we selected 1 variable for inclusion in the model. Substitution of other related variables into the final model did not substantially alter the results. We also examined how the risk factors in the final model for the entire study group performed in selected subgroups of interest: 2 age subgroups (age 30 years and age > 30 years), women who reported no previous UTI history, and inpatients. Because we selected this study group from a defined population, an estimated incidence of pyelonephritis may be derived with some assumptions. We determined the numerator (the estimated number of case-patients) by applying the proportion of women eligible among the contacted and screened case-patients (242 of 296 participants [81.8%]) to the remaining potential case-patients (participants who declined or could not be contacted [n= 90]), which yielded an estimated total number of 316 case-patients over the 19-month sampling interval. To estimate the denominator (the number of at-risk enrollees), we identified the total number of age-eligible women enrolled at the midpoint of recruitment (n= 86738) and applied the proportion of women eligible among participating controls (546 of 655 participants [83.4%]) to this estimate, which yielded an estimated 72306 women who were eligible and at risk for pyelonephritis. We expressed the estimated rate (number of cases/at-risk women) as an annual incidence rate. Role of the Funding Source The National Institute of Diabetes and Digestive and Kidney Diseases provided financial support for this study. The agency had no role in the design, conduct, or reporting of the study or in the decision to submit the manuscript for publication. Results Study Sample Characteristics Case-patients and controls were similar with respect to age and ethnicity (Table 1). Case-patients were likely to be less educated, to report an annual household income less than