点击显示 收起
Center for Health Studies, Group Health Cooperative, and Departments of Epidemiology and Medicine, University of Washington, and the VA Puget Sound Health Care System, Seattle, Washington
Background.
Although Escherichia coli is a well-recognized cause of urinary tract infection in seniors, little is known about the burden of invasive E. coli infection in this population.
Methods.
We conducted a population-based cohort study of 46,238 noninstitutionalized Group Health Cooperative members 65 years of age to ascertain incidences of community-onset E. coli bacteremia and, for comparison, pneumococcal bacteremia, and we then performed a case-control study to identify risk factors for community-onset E. coli bacteremia.
Results.
The overall rate of community-onset E. coli bacteremia in the study cohort was 150 cases/100,000 person-years, which was 3 times higher than the rate of pneumococcal bacteremia. In the case-control study, urinary catheterization and urinary incontinence were the only factors associated with an increased risk of E. coli bacteremia in men (62 cases), whereas cancer, renal failure, congestive heart failure, coronary artery disease, and urinary incontinence were associated with an increased risk of E. coli bacteremia in women (119 cases).
Conclusions.
E. coli appears to be the leading cause of community-onset bacteremia in seniors, and, on the basis of these rates, we estimate that 53,476 cases occur in noninstitutionalized seniors each year in the United States. Community-onset E. coli bacteremia in seniors is, therefore, an infection of public health importance.
Escherichia coli is a well-recognized cause of urinary tract infection and community-onset bacteremia in seniors [17], and hospital-based case studies indicate that E. coli accounts for 30% of hospitalizations for community-onset bacteremia in seniors [811]. However, because population-based assessments of community-onset E. coli bacteremia have not been reported, the epidemiological profile of this infection is not well defined.
To estimate the burden of community-onset E. coli bacteremia in seniors, we conducted 2 population-based studies of noninstitutionalized Group Health Cooperative members 65 years of age. To determine the rate of community-onset E. coli bacteremia, we conducted a retrospective cohort study of 46,238 Group Health Cooperative seniors who were followed over a 3-year period. We compared the rate of community-onset E. coli bacteremia with the rate of pneumococcal bacteremia, as determined by a prior study of pneumococcal vaccine effectiveness conducted in the same study population [12]. To identify potential risk factors for community-onset E. coli bacteremia, we conducted a case-control study, ascertaining characteristics of the study population primarily through a review of medical records.
SUBJECTS, MATERIALS, AND METHODS
Population and Setting
Group Health Cooperative is a staff-model health maintenance organization with 400,000 members in western Washington State. Group Health Cooperative maintains administrative databases that record results of laboratory tests, prescriptions for medications, and International Classification of Diseases, Ninth Revision, Clinical Modification (ICD9)coded diagnoses associated with outpatient visits and hospitalizations. Each member also has a paper medical record that includes copies of hospital admission and discharge summaries and notes from outpatient and emergency department visits.
Retrospective Cohort Study
Study population.
The study cohort was defined as Group Health Cooperative members 65 years of age as of 1 March 1998 who had been enrolled in Group Health Cooperative for at least 1 year prior to that date and who were not residents of a nursing home. Study cohort members were followed from 1 March 1998 until the study end date of 28 February 2001, death, termination of Group Health Cooperative membership, or admission to a nursing home, whichever occurred first.
Identification of episodes of E. coli bacteremia.
Cases of community-onset E. coli bacteremia were identified primarily from blood culture test results reported by the Group Health Cooperative central laboratorywhich processes all blood culture specimens collected at Group Health Cooperative outpatient clinics, emergency departments, and hospitalsand results reported by the laboratory at an affiliated hospital. To identify episodes of E. coli bacteremia that occurred during hospitalizations at nonaffiliated facilities, the medical records of study cohort members who had a hospital discharge diagnosis of E. coli bacteremia (ICD9 codes 038.42 and 041.4) were also reviewed. An index date based on the date of the first positive blood culture was assigned for each episode. Nosocomial infections, defined as episodes in which the first E. coli isolate was obtained from a blood sample collected during a hospitalization and >2 days after the date of admission or from a sample collected within 7 days after discharge from the hospital, were excluded.
Identification of episodes of pneumococcal bacteremia.
Episodes of Streptococcus pneumoniae bacteremia in the study cohort had been identified, as part of a previous assessment, by laboratory test reports and by review of medical records pertaining to hospitalizations associated with a diagnosis of pneumonia (ICD9 codes 480.0 through 487.0) or pneumococcal or streptococcal bacteremia (ICD9 codes 038.0, 038.2, 041.0, 041.2, and 320.1) [12]. An index date based on the date of the first positive blood culture was assigned for each episode.
Analysis.
Crude event rates were calculated by dividing the number of incident cases of either community-onset E. coli bacteremia or pneumococcal bacteremia by the cumulative person-years that elapsed during the study period. To allow the calculation of rates of community-onset E. coli bacteremia in persons with diabetes mellitus, diabetes status was defined by use of administrative data sources, as described elsewhere [12]. Incidence rate ratios were computed to compare rates between sex and diabetes groups.
Case-Control Study
Selection of case patients and control subjects.
Case patients were defined as Group Health Cooperative members 65 years of age who had an episode of community-onset E. coli bacteremia between 1 January 1998 and 31 December 2001. This group included the cohort study case patients, additional persons who had an episode of community-onset E. coli bacteremia that occurred either before or after the cohort study period of 1 March 199828 February 2001, and persons who had an episode of community-onset E. coli bacteremia and were not included in the cohort study because they were <65 years of age on 1 March 1998.
Three frequency-matched control subjects were randomly selected for each case patient. Specifically, case patients were categorized jointly according to their age group (6569, 7074, 7579, 8084, 8589, or 90 years of age), sex, and index year of the E. coli bacteremia episode (1998, 1999, 2000, or 2001). For each group of case patients, we identified the population of Group Health Cooperative members of the same sex and in the same age group who were enrolled as of the last day of the case patients' index year of the E. coli bacteremia episode and had at least 1 outpatient visit or hospitalization recorded during that index year. For each case patient, 3 control subjects were then randomly sampled from the corresponding population of eligible control subjects. Each control subject was assigned an index date, which was a randomly selected date during the case patient's index year.
Data collection.
Information on the characteristics of the bacteremic infections for case patients and on underlying conditions for case patients and control subjects was abstracted from the paper medical record. The presence of coronary artery disease, congestive heart failure, diabetes mellitus, emphysema or chronic obstructive pulmonary disease, asthma, leukemia/lymphoma, lung cancer, nonlung metastatic cancers, multiple myeloma, immunocompromising therapy (including chemotherapy), alcohol abuse, chronic renal failure, presence of an indwelling venous or arterial access device, urinary obstruction, urinary catheterization within 2 months preceding the index date, and urinary incontinence were defined from information recorded in the paper medical record within 1 year preceding the index date. Use of oral estrogen (including progesterone combinations) and intravaginal estrogen therapy within the 6 months preceding the index date was identified from the Group Health Cooperative computerized pharmacy database, which contains records of all prescriptions filled through Group Health Cooperative pharmacies.
Analysis.
We hypothesized that predisposing factors for bacteremic infections from a urinary source may differ from those for bacteremic infections from other (primarily gastrointestinal) sources and that many of the bacteremic infections classified as being due to an unknown source during a review of medical records would have actually resulted from a urinary source. For these reasons, the primary risk factor analyses were restricted to cases of E. coli bacteremia classified as being due to a urinary source or to an unknown source. Alternate analyses that included only case patients with infections due to a known urinary source were not substantively different from the primary analyses. Preliminary analyses revealed differences between men and women in risk factors, so sex-stratified analyses are presented.
Multivariate analyses were conducted using unconditional logistic regression with backward stepwise selection. A variable was removed from the model if the probability of its likelihood-ratio statistic was >0.10, and it was reentered if the probability was <0.05, which was the entry value. Variables entered into the model-building process included diabetes mellitus, emphysema or chronic obstructive pulmonary disease, asthma, alcohol abuse, urinary tract obstruction, arterial or venous indwelling access device, and topical estrogen therapy (female model only), which were not selected for either final model, and the variables shown in table 3, which were selected for at least 1 final model. Age as a categorical variable of 5-year age groups (6569 to 90 years of age) was forced into both models.
RESULTS
Retrospective cohort study.
The 46,238 members of the study cohort contributed 122,365 person-years of observation during the 3-year study period, during which time 184 incident cases of community-onset E. coli bacteremia were identified. Of those, 165 cases were identified from examination of laboratory reports, and the remaining 19 cases were identified by review of medical records of 202 hospitalizations assigned a discharge diagnosis code of E. coli bacteremia. Of the 184 persons with community-onset E. coli bacteremia, 121 (66%) were women, and 43 (23%) were treated on an outpatient basis. The fatality rate within 14 days of the culture date was 7%.
Overall, women were at a 38% increased risk of infection, compared with men (table 1). In both women and men, rates of bacteremia tended to be higher in persons with diabetes mellitus than in those without it, although a statistically significant association between diabetes mellitus and higher risk of bacteremia was found only in women 6569 years of age, women overall, and men 8084 years of age.
Case-control study.
A review of the paper medical record was completed for 211 eligible case patients and 769 control subjects. Of the case patients, 150 (71%) had an identified urinary source for E. coli bacteremiadefined as isolation of E. coli from urine (134/150) or a clinical diagnosis of urosepsis, pyelonephritis, or prostatitisand 31 (15%) had E. coli bacteremia due to an unknown source. Of the remaining 30 case patients, 15 had biliary sepsis, 10 had pneumonia, and 1 each had appendicitis, ischemic colitis, necrotizing fasciitis, septic arthritis, or line sepsis. The case-control analysis was restricted to the 181 case patients who had E. coli bacteremia from a known urinary source or from an unknown source. Characteristics of those case patients, and of control subjects, are shown in table 2.
In men, urinary catheterization and urinary incontinence were the only factors independently associated with the risk of E. coli bacteremia in the multivariate analysis (table 3). In male case patients, 39% were reported to have urinary incontinence or to have had urinary catheterization in the 2 months preceding the index date (table 2).
In contrast, urinary catheterization was uncommon in female case patients. In women, cancer, chronic renal failure, congestive heart failure, coronary artery disease without congestive heart failure, and urinary incontinence were independently associated with an increased risk of E. coli bacteremia, and oral estrogen therapy was associated with a decreased risk of E. coli bacteremia (table 3). Although diabetes mellitus was associated with an increased risk of E. coli bacteremia from a urinary source in women in an age-adjusted model (odds ratio, 2.42 [95% confidence interval, 1.404.17]), this factor was not selected for the final multivariate model.
DISCUSSION
In our study cohort of 46,238 Group Health Cooperative seniors, the incidence of community-onset E. coli bacteremia was 150 cases/100,000 person-years, which is 3 times higher than the rate of pneumococcal bacteremia in this population. The rate of community-onset E. coli bacteremia in seniors at Group Health Cooperative is also substantially higher than the rates of community-onset bacteremia due to Staphylococcus aureus, Group A streptococci, and Group B streptococci in persons 65 years of age, as estimated from surveillance studies of other populations [1315]. These comparisons suggest that E. coli is the most common cause of community-onset bacteremia in seniors. Extrapolating our rate to the noninstitutionalized population of persons 65 years of age, we estimate that there are 53,476 cases of community-onset E. coli bacteremia annually in seniors in the United States.
The majority of the cases of community-onset E. coli bacteremia were due to an identified urinary source, which is consistent with findings of previous reports [11, 16]. As would be expected, we found that several factors previously associated with the risk of urinary tract infection in adults were also associated with the risk of community-onset E. coli bacteremia in seniors. In men, urinary catheterization and urinary incontinence were strongly associated with an increased risk of E. coli bacteremia, and urinary catheterization is a major risk factor for nosocomial urinary tract infection in men. Assessments of the use of urinary catheters in hospitals indicate that catheters are frequently used for inappropriate indications or are used longer than is necessary [17]. Less is known regarding the use of urinary catheters in an ambulatory setting, but it is possible that the pattern of use may be a potentially modifiable risk factor for E. coli bacteremia in men 65 years of age in that setting.
In women, several conditions were independently associated with an increased risk of community-onset E. coli bacteremia, including incontinence, congestive heart failure, and coronary artery disease without congestive heart failure. Incontinence has been identified as a risk factor for urinary tract infection in healthy postmenopausal women [18]. Community-based surveys indicate that at least 30% of women 65 years of age report the presence of incontinence but that only a minority of women with incontinence seek medical care for this condition [1921]. Thus, our detected rate of incontinence of 9% in female control subjects is likely an underestimate of the true prevalence of this condition, which could be a more important contributing factor to the risk of community-onset E. coli bacteremia than was estimated in our analyses.
In contrast, to our knowledge, congestive heart failure and coronary artery disease have not been reported as risk factors for urinary tract infection in women. One previous studythe Heart and Estrogen/Progestin Replacement Study, which was a randomized trial of the effects of hormone therapy on the secondary prevention of coronary eventsincluded an evaluation of the relationship between congestive heart failure at baseline and the risk of urinary tract infection, but no association was identified [22]. In our study, heart disease may be a surrogate marker of poor health status. Alternatively, it is possible that an increased risk of infection could be due to specific biological factors, such as the disruption of urinary tract integritypossibly as a result of diuretic useor impaired hepatic clearance of circulating bacteria.
In our case-control study, oral estrogen therapy was associated with a significantly decreased risk of E. coli bacteremia in women. Although 2 randomized trials have reported a reduction in the risk of recurrent urinary tract infection in postmenopausal women who use intravaginal estrogen preparations [23, 24], assessments of the use of oral estrogen therapy and the risk of urinary tract infection have yielded mixed results [18, 22, 25, 26]. It is possible that oral estrogen therapy modifies the risk of E. coli bacteremia, but in our retrospective study we cannot exclude the possibility that preferential selection of estrogen therapy by relatively healthy women accounts for some or all of the observed association.
To our knowledge, these are the first population-based assessments of community-onset E. coli bacteremia in seniors, and they provide initial estimates of disease rates and risk factors for infection. Limitations include the restriction of the study cohort to the Group Health Cooperative population, which may not be representative of all populations of seniors. However, because the rate of pneumococcal bacteremia of 44 cases/100,000 person-years in our study population is essentially identical to the reported rate of 45 cases/100,000 person-years in community-living seniors (according to a multistate population-based Centers for Disease Control and Prevention surveillance system [27]) findings from our population may be generalizable to other senior populations. Another limitation of our study was that information on underlying medical conditions and other exposures for case patients and control subjects was collected primarily by a review of the paper medical record. Such records are likely an incomplete source of information for potentially important risk factors, such as urinary incontinence and catheterization. Prospective studies that include more detailed assessments of the risk factors suggested by the present study would provide additional information.
E. coli strains causing bacteremia and urinary tract infection appear to be genetically distinct from other gastrointestinal commensal E. coli strains [28]. Russo and Johnson have proposed the designation "ExPEC," for "extraintestinal pathogenic E. coli," for these strains [29]. The invasion of the urinary tract or the bloodstream by ExPEC strains from the gastrointestinal tract is facilitated by a variety of genetically encoded virulence factors, including adhesions, toxins, and polysaccharide coatings. As has been noted, the specificity of ExPEC strains for extraintestinal E. coli infection is important, because this implies that interventions specifically targeting ExPEC strains could be effective preventive measures [28].
Prevention of E. coli bacteremia has not been a focus of vaccine development, but vaccines designed to prevent E. coli urinary tract infection are being evaluated. One of these is a subunit vaccine that stimulates the production of antibodies against FimH, the adhesin on pili that mediates attachment of E. coli type 1 to bladder cells [30]. If this vaccine is effective, it could theoretically prevent some proportion of invasive infections arising from a urinary source. It is also possible that alternate vaccine formulations could prevent bacteremic disease via other mechanisms of action [3133]. Our estimates of the magnitude of the burden of disease due to community-onset E. coli bacteremia in seniors suggest that an effective preventive intervention targeted toward high-risk groups could have a potentially meaningful public health and economic impact.
Acknowledgments
We thank Lisa Mahoney, for her assistance with collection of the laboratory data, and Jennifer Nelson, for her helpful comments on the manuscript.
References
1. Rayner BL, Willcox PA. Community-acquired bacteraemia: a prospective survey of 239 cases. Q J Med 1988; 69:90719. First citation in article
2. Gosbell IB, Newton PJ, Sullivan EA. Survey of blood cultures from five community hospitals in south-western Sydney, Australia, 19931994. Aust N Z J Med 1999; 29:68492. First citation in article
3. Diekema DJ, Pfaller MA, Jones RN, et al. Survey of bloodstream infections due to gram-negative bacilli: frequency of occurrence and antimicrobial susceptibility of isolates collected in the United States, Canada, and Latin America for the SENTRY Antimicrobial Surveillance Program, 1997. Clin Infect Dis 1999; 29:595607. First citation in article
4. Gransden WR, Eykyn SJ, Phillips I, Rowe B. Bacteremia due to Escherichia coli: a study of 861 episodes. Rev Infect Dis 1990; 12:100818. First citation in article
5. Weinstein MP, Towns ML, Quartey SM, et al. The clinical significance of positive blood cultures in the 1990s: a prospective comprehensive evaluation of the microbiology, epidemiology, and outcome of bacteremia and fungemia in adults. Clin Infect Dis 1997; 24:584602. First citation in article
6. Ispahani P, Pearson NJ, Greenwood D. An analysis of community and hospital-acquired bacteraemia in a large teaching hospital in the United Kingdom. Q J Med 1987; 63:42740. First citation in article
7. Scheckler WE. Septicemia in a community hospital 1970 through 1973. JAMA 1977; 237:193841. First citation in article
8. Meyers BR, Sherman E, Mendelson MH, et al. Bloodstream infections in the elderly. Am J Med 1989; 86:37984. First citation in article
9. Diekema DJ, Pfaller MA, Jones RN. Age-related trends in pathogen frequency and antimicrobial susceptibility of bloodstream isolates in North America: SENTRY Antimicrobial Surveillance Program, 19972000. Int J Antimicrob Agents 2002; 20:4128. First citation in article
10. Whitelaw DA, Rayner BL, Willcox PA. Community-acquired bacteremia in the elderly: a prospective study of 121 cases. J Am Geriatr Soc 1992; 40:9961000. First citation in article
11. Esposito AL, Gleckman RA, Cram S, Crowley M, McCabe F, Drapkin MS. Community-acquired bacteremia in the elderly: analysis of one hundred consecutive episodes. J Am Geriatr Soc 1980; 28:3159. First citation in article
12. Jackson LA, Neuzil KM, Yu O, et al. Effectiveness of pneumococcal polysaccharide vaccine in older adults. N Engl J Med 2003; 348:174755. First citation in article
13. Morin CA, Hadler JL. Population-based incidence and characteristics of community-onset Staphylococcus aureus infections with bacteremia in 4 metropolitan Connecticut areas, 1998. J Infect Dis 2001; 184:102934. First citation in article
14. Davies HD, McGeer A, Schwartz B, et al. Invasive group A streptococcal infections in Ontario, Canada. Ontario Group A Streptococcal Study Group. N Engl J Med 1996; 335:54754. First citation in article
15. Farley MM, Harvey RC, Stull T, et al. A population-based assessment of invasive disease due to group B Streptococcus in nonpregnant adults. N Engl J Med 1993; 328:180711. First citation in article
16. McCue JD. Gram-negative bacillary bacteremia in the elderly: incidence, ecology, etiology, and mortality. J Am Geriatr Soc 1987; 35:2138. First citation in article
17. Saint S, Lipsky BA. Preventing catheter-related bacteriuria: should we Can we How Arch Intern Med 1999; 159:8008. First citation in article
18. Hu KK, Boyko EJ, Scholes D, et al. Risk factors for urinary tract infections in postmenopausal women. Arch Intern Med 2004; 164:98993. First citation in article
19. Molander U, Sundh V, Steen B. Urinary incontinence and related symptoms in older men and women studied longitudinally between 70 and 97 years of age: a population study. Arch Gerontol Geriatr 2002; 35:23744. First citation in article
20. Stoddart H, Donovan J, Whitley E, Sharp D, Harvey I. Urinary incontinence in older people in the community: a neglected problem Br J Gen Pract 2001; 51:54852. First citation in article
21. Roberts RO, Jacobsen SJ, Rhodes T, et al. Urinary incontinence in a community-based cohort: prevalence and healthcare-seeking. J Am Geriatr Soc 1998; 46:46772. First citation in article
22. Brown JS, Vittinghoff E, Kanaya AM, Agarwal SK, Hulley S, Foxman B. Urinary tract infections in postmenopausal women: effect of hormone therapy and risk factors. Obstet Gynecol 2001; 98:104552. First citation in article
23. Raz R, Stamm WE. A controlled trial of intravaginal estriol in postmenopausal women with recurrent urinary tract infections. N Engl J Med 1993; 329:7536. First citation in article
24. Eriksen B. A randomized, open, parallel-group study on the preventive effect of an estradiol-releasing vaginal ring (Estring) on recurrent urinary tract infections in postmenopausal women. Am J Obstet Gynecol 1999; 180:10729. First citation in article
25. Kirkengen AL, Andersen P, Gjersoe E, Johannessen GR, Johnsen N, Bodd E. Oestriol in the prophylactic treatment of recurrent urinary tract infections in postmenopausal women. Scand J Prim Health Care 1992; 10:13942. First citation in article
26. Foxman B, Somsel P, Tallman P, et al. Urinary tract infection among women aged 40 to 65: behavioral and sexual risk factors. J Clin Epidemiol 2001; 54:7108. First citation in article
27. Kupronis BA, Richards CL, Whitney CG. Invasive pneumococcal disease in older adults residing in long-term care facilities and in the community. J Am Geriatr Soc 2003; 51:15205. First citation in article
28. Russo TA, Johnson JR. Medical and economic impact of extraintestinal infections due to Escherichia coli: focus on an increasingly important endemic problem. Microbes Infect 2003; 5:44956. First citation in article
29. Russo TA, Johnson JR. Proposal for a new inclusive designation for extraintestinal pathogenic isolates of Escherichia coli: ExPEC. J Infect Dis 2000; 181:17534. First citation in article
30. Langermann S, Ballou WR Jr. Vaccination utilizing the FimCH complex as a strategy to prevent Escherichia coli urinary tract infections. J Infect Dis 2001; 183(Suppl 1):S846. First citation in article
31. O'Hanley P, Marcus R, Baek KH, Denich K, Ji GE. Genetic conservation of HlyA determinants and serological conservation of HlyA: basis for developing a broadly cross-reactive subunit Escherichia coli -hemolysin vaccine. Infect Immun 1993; 61:10917. First citation in article
32. Russo TA, McFadden CD, Carlino-MacDonald UB, Beanan JM, Olson R, Wilding GE. The siderophore receptor IroN of extraintestinal pathogenic Escherichia coli is a potential vaccine candidate. Infect Immun 2003; 71:71649. First citation in article
33. O'Hanley P, Lalonde G, Ji G. -Hemolysin contributes to the pathogenicity of piliated digalactoside-binding Escherichia coli in the kidney: efficacy of an -hemolysin vaccine in preventing renal injury in the BALB/c mouse model of pyelonephritis. Infect Immun 1991; 59:115361. First citation in article