Diabetes Mellitus as a Risk and Prognostic Factor for Community-Acquired Bacteremia Due to Enterobacteria: A 10-Year, Population-Based Study among Adults

Patients with diabetes mellitus may be at increased risk for infections [1, 2]; in particular, infections originating from the urinary tract [3]. Previous studies have reported a 2- to 4-fold increased prevalence of bacteriuria associated with diabetes [4] and recently also an increased risk of acute symptomatic urinary tract infection [5] and hospitalization for pyelonephritis [6]. Studies of the association with the most life-threatening consequence of urinary tract infection, bacteremia due to gram-negative bacilli, such as Escherichia coli and other enterobacteria, are limited [7].

Mechanisms predisposing diabetic patients to enterobacterial bacteremia may include cystopathy and micturition abnormalities; glucosuria, which may promote bacterial growth; and impaired function in neutrophils and macrophages due to hyperglycemia. Decreased immune response, end-organ disease that compromises tissue oxygenation, and hyperglycemic derangement may further be associated with a poorer outcome. In this study, we examined whether diabetic persons have an increased risk for community-acquired bacteremia due to enterobacteria, compared with persons without diabetes, and whether presence of diabetes was associated with a poorer prognosis.

Methods. The study was conducted from 1992 through 2001 in North Jutland County, Denmark, which has a population of ∼500,000 inhabitants, all provided with free, tax-supported health care. Through the use of a unique civil registry number, a complete hospitalization and drug prescription history can be obtained for each individual.

Patients aged >15 years with a first hospitalization for monomicrobial enterobacterial bacteremia were identified in the North Jutland County Bacteremia Registry, as described in detail elsewhere [8]. A case of enterobacterial bacteremia was defined as a clinical episode of infection in which an organism in the family Enterobacteriaceae was detected by blood culture. Enterobacteria were identified with conventional methods or with a commercial identification system [9]. Salmonella infections were excluded. The infection had to be present or incubating at the time of admission to the hospital. Concurrent bacteriological and clinical data were acquired by physicians as part of standard working practice using standardized forms, and the data were later computerized. Ongoing empirical antibiotic therapy was regarded as appropriate if given intravenously (except for fluoroquinolones) and if the blood isolate was susceptible in vitro to ⩾1 of the antibiotics given. The probable focus of infection was assessed on the basis of microbiological and clinical findings.

Using civil registry numbers, we selected 10 control subjects from the population for each case patient by electronic linkage with the Danish Central Population Registry. Case and control subjects were matched by sex, age (same year of birth), and residence (North Jutland County).

Data on diabetes among case and control subjects were obtained, using a validated method [8], from the County Prescription Database and Hospital Discharge Registry. Diabetes was considered present if (1) at least 1 prescription for insulin or an oral antidiabetic drug was recorded, and/or (2) there was a hospital discharge diagnosis of type 1 or type 2 diabetes recorded before the date of hospitalization with bacteremia.

To adjust for comorbidity, we calculated the Charlson index score [10] for each patient and control subject on the basis of the complete hospital-discharge history before the date of admission with bacteremia. Diabetes was not included in the index in our study. Three comorbidity index levels were defined, according to the Charlson index score: score of 0, low; score of 1–2, medium; score of >2, high.

For the case-control analysis of the risk for enterobacterial bacteremia, we used conditional logistic regression to estimate ORs with 95% CIs for bacteremia according to the presence of diabetes, adjusted for comorbidity. We stratified according to sex, age (15–39 years, 40–64 years, 65–79 years, and ⩾80 years), and comorbidities. ORs were estimated separately for patients with a focus of infection in the urinary tract, for patients with bacteremia due to E. coli, and for patients with bacteremia due to Klebsiella species. To examine the impact of diabetes on the overall risk for community-acquired bacteremia due enterobacteria, we also calculated the population-attributable risk for a diagnosis of diabetes [11].

For the cohort analysis of 30-day and 90-day mortality associated with enterobacterial bacteremia, we obtained data on patients with enterobacterial bacteremia from the Central Population Registry, which keeps record of all changes in vital status and migration for the entire Danish population, including the date of death. The duration of follow-up was calculated as the time from the date when the first culture-positive blood sample was drawn until death or migration, or until 90 days after that blood sample was drawn. Cox regression analyses were used to compute mortality rate ratios with 95% CIs for diabetic and nondiabetic patients with bacteremia, adjusted for sex, age, and comorbidities, with and without inclusion of data on the focus of infection. All statistical analyses were performed with STATA software, version 8.0 (STATA).

Results. The study included 1317 case patients with a first hospitalization for bacteremia due to enterobacteria. The incidence of bacteremia increased from 21 cases per 100,000 person-years, in 1992, to 34 cases per 100,000 person-years, in 2001.

With respect to the risk for developing enterobacterial bacteremia, 225 (17.1%) of the 1317 patients with bacteremia had diabetes, compared with 779 (5.9%) of 13,170 control subjects. The adjusted OR for enterobacterial bacteremia in persons with diabetes was 2.9 (95% CI, 2.4–3.4) (table 1). The relative risk of bacteremia associated with diabetes was greatest for adults aged <65 years (OR, 5.9; 95% CI, 3.9–9.0) and for persons without other comorbidities (OR, 4.9; 95% CI, 3.7–6.6). The relative risk appeared to be higher for diabetic female subjects than for diabetic male subjects, and higher for patients who had bacteremia with a urinary tract origin (OR, 3.4; 95% CI, 2.7–4.2) than for patients with other foci of infection (OR, 2.2; 95% CI, 1.6–2.9). For E. coli bacteremia, the OR was 3.0 (95% CI, 2.5–3.6), and for bacteremia due to Klebsiella species, the OR was 2.2 (95% CI, 1.3–3.7). Under the assumption that the prevalence of diabetes in the control subjects was 5.9%, the population-attributable risk was 10.1%.

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1

Mortality curves for community-acquired bacteremia due to enterobacteria in patients with diabetes (n = 225) and without diabetes (n = 1092).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1

Unstratified and stratified crude and adjusted ORs for community-acquired enterobacterial bacteremia according to presence of diabetes mellitus

With respect to the prognosis for patients with enterobacterial bacteremia, diabetic patients were more likely than nondiabetic patients to have a urinary tract focus of infection (68% vs. 59%) but less likely to have an abdominal focus (12% vs. 19%). There were no substantial differences between diabetic and nondiabetic patients with respect to the distribution of infection due to the major groups of bacteria (E. coli, 83% vs. 80%; Klebsiella species, 9% vs. 12%) or the appropriateness of the empirical antibiotic therapy given (66% vs. 63%). The mortality curves for diabetic and nondiabetic patients with bacteremia are shown in figure 1. At 30 days, the mortality was 17.3% for diabetic patients and 13.4% for nondiabetic patients; after 90 days, it was 23.6% and 19.5%, respectively. After 30 days, the adjusted mortality rate ratio for diabetic patients was 1.3 (95% CI, 0.9–1.8), and after 90 days it was 1.2 (95% CI, 0.9–1.6). Stratification for focus of infection did not materially affect the mortality estimates. When focus of infection was included in the analyses, the mortality rate ratio for diabetic patients was virtually unchanged: 1.4 (95% CI, 1.0–2.0) after 30 days and 1.3 (95% CI, 0.9–1.7) after 90 days.

Discussion. We found that presence of diabetes mellitus was associated with a 3-fold increased risk and a slightly poorer outcome of community-acquired bacteremia due to E. coli and other enterobacteria. Our findings support the hypothesis that urinary tract infections in patients with diabetes increase the risk of acquiring enterobacterial bacteremia. Canadians aged >65 years who had diabetes were 3–6 times more likely than nondiabetic persons to be hospitalized with acute pyelonephritis, and diabetic patients aged <45 years had a 15-fold increased risk [6]; these findings are in accordance with our findings of the highest relative risk in young adults. The decrease in relative risk with increasing age must be viewed in comparison with the low absolute risk for bacteremia among young persons, and may reflect different risk patterns for patients with type 1 diabetes, compared with persons with type 2 diabetes.

The poorer outcome was related to an increased risk of dying during the second week of infection (figure 1). We have reported that outcomes for diabetic patients with community-acquired pneumococcal bacteremia are similar to those for nondiabetic patients, with a tendency towards better survival during the first week of infection [8]. Among patients with pneumococcal bacteremia, a high proportion of fatalities due to uncontrolled sepsis occur during the first few days, whereas, among patients with bacteremia originating from the urinary tract, mortality seems to increase more slowly. A less-active inflammatory cascade [12] may protect diabetic patients with pneumococcal bacteremia against early, severe manifestations of sepsis and subsequent death, whereas in patients with bacteremia caused by enterobacteria, protracted multiorgan failure promoted by diabetic organ disease may play a role in the higher mortality among diabetic patients.

We used a population-based design and prospective data from registries with complete follow-up. The study itself did not introduce bias associated with differences in diagnoses. Diabetic patients may have been hospitalized and their blood obtained for culture at a lower threshold of clinical suspicion than was the case for nondiabetic individuals, leading to a potential overestimation of risk and underestimation of mortality. Nevertheless, the higher mortality among diabetic patients argues against more-meticulous case ascertainment. We may, in addition, have missed some diabetic individuals treated only with diet control. The Charlson index has been validated for prediction of mortality, including among patients with bacteremia [13], and it includes most suggested risk factors for gram-negative bacteremia, such as malignancies and liver cirrhosis. Nevertheless, there is a possibility of residual confounding by diseases not included in the index [2]. Finally, differences in the treatment received after hospitalization that were associated with the presence of diabetes may have played a role in the prognosis, although the appropriateness of antibiotic therapy appeared to be unrelated to the presence of diabetes.

We conclude that diabetes has a considerable public health impact on the risk for and prognosis of enterobacterial bacteremia acquired in the community. In our region, >10% of cases of enterobacterial bacteremia may be attributable to diabetes. Preventive measures for diabetic patients might include increased surveillance and avoidance of well-known risk factors for urinary tract infections.

• Received August 13, 2004.
• Accepted October 14, 2004.

• © 2005 by the Infectious Diseases Society of America