Abstract
Background
Children with gastrointestinal infections caused by Escherichia coli O157:H7 are at risk for the hemolytic–uremic syndrome. Whether antibiotics alter this risk is unknown.
Methods
We conducted a prospective cohort study of 71 children younger than 10 years of age who had diarrhea caused by E. coli O157:H7 to assess whether antibiotic treatment in these children affects the risk of the hemolytic–uremic syndrome and to assess the influence of confounding factors on this outcome. Estimates of relative risks were adjusted for possible confounding effects with the use of logistic-regression analysis.
Results
Among the 71 children, 9 (13 percent) received antibiotics and the hemolytic–uremic syndrome developed in 10 (14 percent). Five of these 10 children had received antibiotics. Factors significantly associated with the hemolytic–uremic syndrome were a higher initial white-cell count (relative risk, 1.3; 95 percent confidence interval, 1.1 to 1.5), evaluation with stool culture soon after the onset of illness (relative risk, 0.3; 95 percent confidence interval, 0.2 to 0.8), and treatment with antibiotics (relative risk, 14.3; 95 percent confidence interval, 2.9 to 70.7). The clinical and laboratory characteristics of the 9 children who received antibiotics and the 62 who did not receive antibiotics were similar. In a multivariate analysis that was adjusted for the initial white-cell count and the day of illness on which stool was obtained for culture, antibiotic administration remained a risk factor for the development of the hemolytic–uremic syndrome (relative risk, 17.3; 95 percent confidence interval, 2.2 to 137).
Conclusions
Antibiotic treatment of children with E. coli O157:H7 infection increases the risk of the hemolytic–uremic syndrome.
Escherichia coli O157:H7 causes sporadic and epidemic gastrointestinal infections worldwide. In approximately 15 percent of the children in North America who are infected with E. coli O157:H7, the hemolytic–uremic syndrome develops soon after the onset of diarrhea.1–4 This syndrome is characterized by thrombocytopenia, hemolytic anemia, and nephropathy and is believed to be caused by Shiga toxins elaborated by E. coli O157:H7 or other infecting E. coli that have been absorbed into the systemic circulation.5
Treatment with antibiotics does not ameliorate E. coli O157:H7 infections,1,6,7 and in some studies, it has been associated with worse clinical outcomes.8–10 In one prospective, randomized, controlled trial, antibiotic treatment was neither harmful nor beneficial, but treatment was administered late in the course of illness.11 In a child infected with E. coli O157:H7 who was treated with antibiotics before the onset of diarrhea, fecal shedding of the organisms ceased, but the treatment did not prevent the hemolytic–uremic syndrome.12
When analyzing the role of antibiotic administration as a risk factor for the development of the hemolytic–uremic syndrome in children infected with E. coli O157:H7, it is important to consider that the severity of illness might confound the association with antibiotic treatment. For example, antibiotics might be administered to more severely ill infected children in whom the hemolytic–uremic syndrome is destined to develop independently of antibiotic treatment. We therefore examined data from a prospective cohort study to determine whether antibiotic treatment alters the risk of the hemolytic–uremic syndrome among children infected with E. coli O157:H7.
METHODS
A network of 47 cooperating laboratories in Washington, Oregon, Idaho, and Wyoming prospectively identified 73 children younger than 10 years of age who had E. coli O157:H7 infections between April 1, 1997, and August 31, 1999. These children were identified on the basis of stool cultures done on sorbitol–Mac-Conkey agar to detect this pathogen.13 One of the investigators was contacted immediately by telephone after the identification of each infected child by the laboratory. This investigator then immediately telephoned the child’s physician seeking permission to approach the child’s family about participation in the study. Written informed consent was obtained from the parents or guardians of each child, and, if appropriate, assent was obtained from the child. The study was approved by the institutional review board at each participating hospital. Only the 71 children whose stool cultures were obtained within the first seven days after the onset of illness (the first day of diarrhea was considered to be the first day of illness) were included in the analysis.
A standardized questionnaire was administered to the caregivers of each enrolled child within two days after enrollment. Questions were included about the child’s sex, age, and race or ethnic group; the duration of symptoms and signs; the presence or absence of visible blood in the stool, vomiting, and fever; and the names of prescription and nonprescription medications taken during the illness. Prescription medications were administered at the discretion of each child’s primary care, inpatient, or emergency department physician. Nonprescription medications were administered on the recommendation of each patient’s treating physician or on the basis of the caregiver’s decision. Only medications taken on or before the seventh day of illness were considered. Medications administered after the hemolytic–uremic syndrome was diagnosed were not analyzed. Medications were classified as antibiotics; antimotility drugs (if they inhibited intestinal peristalsis), including opioid narcotics; acetaminophen; and nonsteroidal antiinflammatory drugs.
The administration of all reported prescription drugs was verified by the medical provider who prescribed them or by inspecting the child’s medical record. Isolates of E. coli O157:H7 were sent to the Children’s Hospital and Regional Medical Center in Seattle for determinations of their susceptibility to the antibiotics administered to the children. The disk-diffusion technique was used for susceptibility testing.14
To reduce concern about observer error and the ability to validate temperature measurements, the analysis included only the initial temperature determinations for children evaluated at Children’s Hospital and Regional Medical Center in Seattle. The laboratory data that were analyzed consisted of white-cell counts and serum urea nitrogen and creatinine concentrations. Only the initial laboratory test result for each child was analyzed as a potential risk factor for the development of the hemolytic–uremic syndrome.
Daily blood counts and renal-function tests were performed in all infected children until the hemolytic–uremic syndrome developed and resolved or until the diarrhea resolved and the hemolytic–uremic syndrome clearly did not develop. The hemolytic–uremic syndrome was defined as hemolytic anemia (a hematocrit below 30 percent, with evidence of the destruction of erythrocytes on a peripheral-blood smear), thrombocytopenia (a platelet count of less than 150,000 per cubic millimeter), and renal insufficiency (a serum creatinine concentration that exceeded the upper limit of the normal range for age).
On the basis of previous studies in Washington of children infected with E. coli O157:H7,1,2 the period of risk for the hemolytic–uremic syndrome was considered to be 14 days from the onset of diarrhea among children with a positive culture for E. coli O157:H7, and this represented the period of clinical observation. The incidence of the hemolytic–uremic syndrome was expressed in terms of the risk of development of the syndrome for an individual child during these 14 days. Children with the syndrome were categorized as having oligoanuria if they excreted less than 0.5 ml of urine per kilogram of body weight per hour for 48 or more hours during the course of the syndrome. Medical records were reviewed to verify the fulfillment of the classification criteria.
Differences between the children who were treated with antibiotics and those who were not were analyzed with the use of independent-sample t-tests for continuous variables and Fisher’s exact or chi-square tests for categorical variables and with use of univariate logistic regression for linear trend.15 Multivariate logistic-regression analysis was used to examine the risk of the hemolytic–uremic syndrome after adjustment for the initial white-cell count and the day of illness on which the initial stool culture was obtained for analysis.15 These factors were chosen because they were associated with the risk of the hemolytic–uremic syndrome.1,8 The day of submission of the stool sample was the most common point in the illness at which antibiotics were prescribed for the children in this study.
The association between the exposure to medication during the diarrheal phase of infection and the subsequent development of the hemolytic–uremic syndrome was estimated by odds ratios and reported as the relative risk. Demographic factors, the presence or absence of specific symptoms, the duration of the diarrhea before the laboratory assessments and the administration of antibiotics, and initial laboratory values were studied to determine whether these factors confounded the association between exposure to medication and the development of the hemolytic–uremic syndrome. We selected the variables that were significantly associated with the hemolytic–uremic syndrome, and were a priori considered to be potential confounders, to yield the most parsimonious model. After adjustment, the final models were interpreted as the adjusted relative risk of the hemolytic–uremic syndrome among children infected with E. coli O157:H7. Adjusted relative risks and their 95 percent confidence intervals were calculated from the logistic-regression coefficients and their standard errors. Results are reported as adjusted relative risks with 95 percent confidence intervals. All P values are two-tailed. Computations were performed with use of Intercooled Stata software (version 6.0 for Windows, Stata, College Station, Tex.).
RESULTS
The hemolytic–uremic syndrome developed in 10 of the 71 children with E. coli O157:H7 infection (14 percent). Four of these 10 children were classified as having oligoanuria, and each of these 4 required dialysis. Seven children required erythrocyte transfusions, platelet transfusions, or both. None of the children died during hospitalization. The incidence of the hemolytic–uremic syndrome according to the children’s characteristics is summarized in Table 1.
Table 1.
Frequency of the Hemolytic–Uremic Syndrome According to the Characteristics of 71 Children Infected with Escherichia coli O157:H7.*
| Characteristic | Frequency of the Hemolytic– Uremic Syndrome % |
No. of Cases/Total No. of Patients | P Value |
|---|---|---|---|
| Sex | 0.96 | ||
| Male | 14 | 5/36 | |
| Female | 14 | 5/35 | |
| Age | 0.95† | ||
| 0–1 yr | 5 | 1/22 | |
| 2–3 yr | 29 | 4/14 | |
| 4–6 yr | 14 | 3/22 | |
| 7–10 yr | 15 | 2/13 | |
| Race or ethnic group | |||
| White | 15 | 9/62 | 0.63 |
| Hispanic | 0 | 0/3 | — |
| Black | 0 | 0/1 | — |
| Asian or Pacific Islander | 33 | 1/3 | 0.37 |
| Native American | 0 | 0/2 | — |
| Bloody diarrhea | 0.26 | ||
| Yes | 12 | 8/64 | |
| No | 29 | 2/7 | |
| Vomiting | 0.25 | ||
| Yes | 18 | 8/45 | |
| No | 8 | 2/26 | |
| Fever‡ | 0.22 | ||
| Yes | 25 | 2/8 | |
| No | 6 | 1/16 | |
| Initial white-cell count | 0.005† | ||
| 3200–8700/mm3 | 0 | 0/18 | |
| 8800–11,800/mm3 | 6 | 1/18 | |
| 11,900–14,200/mm3 | 17 | 3/18 | |
| 14,300–24,600/mm3 | 35 | 6/17 | |
| No. of medications taken for E. coli infection | 0.002† | ||
| 0 | 4 | 2/46 | |
| 1 | 25 | 5/20 | |
| 2 | 60 | 3/5 | |
| Acetaminophen | 0.09 | ||
| Yes | 33 | 3/9 | |
| No | 11 | 7/62 | |
| Nonsteroidal antiinflammatory drugs | — | ||
| Yes | 0 | 0/1 | |
| No | 14 | 10/70 | |
| Antimotility drugs | 0.19 | ||
| Yes | 27 | 3/11 | |
| No | 12 | 7/60 | |
| Antibiotics | 0.001 | ||
| Yes | 56 | 5/9 | |
| No | 8 | 5/62 | |
| Stool culture obtained§ | 0.01† | ||
| Days 1–2 of illness | 33 | 8/24 | |
| Day 3 of illness | 9 | 2/22 | |
| Days 4–7 of illness | 0 | 0/25 | |
| Antibiotics started§¶ | 0.69† | ||
| Day 1 of illness | 50 | 2/4 | |
| Day 2 of illness | 67 | 2/3 | |
| Day 3 of illness | 50 | 1/2 | |
| Initial white-cell count obtained§ | 0.009† | ||
| Days 1–3 of illness | 28 | 7/25 | |
| Days 4–5 of illness | 12 | 3/25 | |
| Days 6–10 of illness | 0 | 0/21 | |
| Stool culture positive for E. coli O157:H7§ | 0.04† | ||
| Days 2–4 of illness | 25 | 6/24 | |
| Day 5 of illness | 16 | 3/19 | |
| Days 6–10 of illness | 4 | 1/28 |
Of the 18 children for whom the initial temperature was documented at Children’s Hospital and Regional Medical Center (2 of whom had the hemolytic–uremic syndrome), all had temperatures of less than 38.0°C except for 1, who had a temperature of 39.4°C and in whom the hemolytic–uremic syndrome did not develop. All children had serum urea nitrogen concentrations of less than 20 mg per deciliter (7.1 mmol per liter), except for two whose serum urea nitrogen concentrations were 22 mg per deciliter (7.9 mmol per liter) and 47 mg per deciliter (16.8 mmol per liter), neither of whom had the hemolytic–uremic syndrome. All children had initial serum creatinine concentrations of less than 0.9 mg per deciliter (79.6 μmol per liter) except for one, who had a serum creatinine concentration of 1.0 mg per deciliter (88.4 μmol per liter) and in whom the hemolytic–uremic syndrome did not develop.
The P value is for the linear trend for the variable.
The category of fever was based on the caregivers’ reports and was added to the standardized questionnaire on December 1, 1998; thus, only 24 children were evaluated for this variable.
The onset of diarrhea was considered to be the first day of illness.
This variable was assessed in the nine children who received antibiotics.
There were no significant differences in the frequency of the hemolytic–uremic syndrome with respect to sex and the presence or absence of bloody diarrhea, caregivers’ reports of fever, and vomiting. Nor was the frequency significantly related to the initial serum urea nitrogen or creatinine concentration or, for the subgroup of children evaluated at Children’s Hospital and Regional Medical Center, the initial temperature. However, the frequency of the syndrome was significantly related to the initial white-cell count (P=0.005), with a rate of 0 percent among children with an initial count of 3200 to 8700 per cubic millimeter, a rate of 6 percent among those with an initial count of 8800 to 11,800 per cubic millimeter, a rate of 17 percent among those with an initial count of 11,900 to 14,200 per cubic millimeter, and a rate of 35 percent among those with an initial count of 14,300 to 24,600 per cubic millimeter. In addition, the frequency of the syndrome was higher among children who had laboratory studies soon after the onset of illness (according to the day of illness on which the stool was obtained for culture, the day on which the culture was positive, and the day on which the initial white-cell count was obtained) and among those who were treated with antibiotics.
The hemolytic–uremic syndrome developed in 5 of the 9 children given antibiotics (56 percent), as compared with 5 of the 62 children who were not given antibiotics (8 percent, P<0.001). The characteristics of the children who were given antibiotics and those who were not were similar (Table 2).
Table 2.
Characteristics of Children Infected with Escherichia coli O157:H7 According to Whether They Were Treated with Antibiotics.*
| Characteristic | Antibiotics (N=9) | No Antibiotics (N=62) | P Value |
|---|---|---|---|
| Female sex — no. (%) | 6 (67) | 29 (47) | 0.30 |
| Age — yr | 5.5±3.4 | 4.1±2.6 | 0.15 |
| Race or ethnic group — no. (%) | |||
| White | 8 (89) | 54 (87) | 1.00 |
| Hispanic | 0 | 3 (5) | 1.00 |
| Black | 0 | 1 (2) | 1.00 |
| Asian or Pacific Islander | 1 (11) | 2 (3) | 0.34 |
| Native American | 0 | 2 (3) | 1.00 |
| Bloody diarrhea — no. (%) | 7 (78) | 57 (92) | 0.21 |
| Vomiting — no. (%) | 5 (56) | 40 (65) | 0.72 |
| Fever — no. (%)† | |||
| Yes | 0 | 8 (38) | 0.53 |
| No | 3 (100) | 13 (62) | |
| Initial temperature — °C‡ | 37.0±1.0 | 36.9±0.8 | 0.88 |
| Initial white-cell count — ×10−3/mm3 | 13.1±4.7 | 11.6±4.2 | 0.34 |
| Initial serum urea nitrogen — mg/dl§ | 9.2±2.7 | 10.2±6.1 | 0.63 |
| Initial serum creatinine —mg/dl¶ | 0.4±0.1 | 0.4±0.1 | 0.71 |
| Stool culture obtained — day of illness|| | 2.7±1.4 | 3.3±1.4 | 0.19 |
| Antibiotics started — day of illness|| | 2.8±1.6 | — | — |
| Initial white-cell count obtained — day of illness|| | 4.0±2.0 | 4.7±2.0 | 0.30 |
| Stool culture positive for E. coli O157:H7 — day of illness|| | 4.9±1.3 | 5.3±1.6 | 0.42 |
| Hemolytic–uremic syndrome case definition fulfilled — day of illness||** | 7.8±1.5 | 7.2±1.1 | 0.49 |
| Progression to hemolytic–uremic syndrome — no. (%) | |||
| Hemolytic–uremic syndrome | 5 (56) | 5 (8) | 0.002 |
| Oligoanuric hemolytic–uremic syndrome | 2 (22) | 2 (3) |
Plus–minus values are means ±SD.
The category of fever was based on the caregivers’ reports and was added to the standardized questionnaire on December 1, 1998; thus, only 24 children were evaluated for this variable.
The initial temperature was documented in 18 children who presented to Children’s Hospital and Regional Medical Center in Seattle (3 of whom received antibiotics).
To convert values for serum urea nitrogen to millimoles per liter, multiply by 0.357.
To convert values for serum creatinine to micromoles per liter, multiply by 88.4.
The onset of diarrhea was considered to be the first day of illness.
This variable was assessed in the 10 children in whom the hemolytic–uremic syndrome developed.
We used logistic-regression analysis to assess demographic variables, symptoms, laboratory values, and medication use as potential risk factors for the hemolytic–uremic syndrome (Table 3). The initial white-cell count and the day of the initial stool culture — two surrogates for the severity of illness — were independently associated with the development of the syndrome. In the logistic-regression analysis of both factors, the initial white-cell count remained significantly associated with the hemolytic–uremic syndrome (P=0.02), with risk being proportional to the white-cell count. Also, the interval from the onset of diarrhea to the day on which the initial stool culture was obtained remained strongly associated with the risk of the hemolytic–uremic syndrome (P=0.008), with the risk being inversely proportional to the number of days in this interval.
Table 3.
Relative Risk of the Hemolytic–Uremic Syndrome among Children Infected with Escherichia coli O157:H7.
| Covariate* | Univariate Relative Risk (95% CI)† | P Value | Adjusted Relative Risk (95% CI)‡ | P Value |
|---|---|---|---|---|
| Age (continuous) | 1.0 (0.8 to 1.3) | 0.95 | 1.1 (0.8 to 1.5) | 0.43 |
| 0 to 5 yr§ | 1.0 | 1.0 | ||
| >5 yr | 0.7 (0.2 to 2.8) | 0.57 | 1.1 (0.2 to 5.7) | 0.93 |
| Sex | ||||
| Male§ | 1.0 | 1.0 | ||
| Female | 1.0 (0.3 to 4.0) | 0.96 | 1.1 (0.2 to 5.0) | 0.95 |
| Presence of bloody diarrhea | 0.4 (0.1 to 2.2) | 0.26 | 0.4 (0.1 to 3.0) | 0.36 |
| Presence of vomiting | 2.6 (0.5 to 13.3) | 0.25 | 3.2 (0.5 to 20.6) | 0.21 |
| Presence of fever¶ | 5.0 (0.4 to 66.0) | 0.22 | 6.2 (0.2 to 169) | 0.23 |
| Temperature (continuous)|| | 1.9 (0.5 to 7.4) | 0.37 | 14.1 (0.5 to 419) | 0.13 |
| Initial serum urea nitrogen (continuous)** | 1.0 (0.9 to 1.1) | 0.73 | 1.1 (0.9 to 1.3) | 0.53 |
| 2–10 mg/dl§ | 1.0 | 1.0 | ||
| ≥11 mg/dl | 2.2 (0.6 to 8.6) | 0.25 | 2.8 (0.5 to 14.4) | 0.23 |
| Initial serum creatinine (continuous)†† | 0.7 (0.01 to 91.5) | 0.89 | 4.7 (0.01 to >500) | 0.62 |
| 0.1–0.5 mg/dl§ | 1.0 | 1.0 | ||
| ≥0.6 mg/dl | 1.2 (0.1 to 11.9) | 0.85 | 3.0 (0.1 to 69.4) | 0.50 |
| Initial white-cell count (continuous) | 1.3 (1.1 to 1.5) | 0.005 | 1.5 (1.1 to 2.1) | 0.02 |
| 3200–12,900/mm3§ | 1.0 | 1.0 | ||
| ≥13,000/mm3 | 3.9 (1.0 to 15.5) | 0.06 | 6.0 (1.2 to 29.8) | 0.03 |
| Day initial white-cell count obtained | 0.5 (0.3 to 0.8) | 0.009 | 0.7 (0.4 to 1.5) | 0.38 |
| Day stool culture obtained | 0.3 (0.2 to 0.8) | 0.01 | 0.3 (0.1 to 0.7) | 0.008 |
| Day stool culture positive | 0.5 (0.3 to 1.0) | 0.04 | 1.0 (0.5 to 2.1) | 1.00 |
| Acetaminophen given | 3.9 (0.8 to 19.3) | 0.09 | 2.0 (0.3 to 12.9) | 0.46 |
| Antimotility drugs given | 2.8 (0.6 to 13.3) | 0.19 | 3.0 (0.5 to 19.4) | 0.25 |
| Antibiotics given | ||||
| Within first 7 days after onset of illness | 14.3 (2.9 to 70.7) | 0.001 | 17.3 (2.2 to 137) | 0.007 |
| Within first 3 days after onset of illness | 15.0 (1.3 to 174) | 0.03 | 32.3 (1.4 to 737) | 0.03 |
Continuous covariates were analyzed and reported on continuous scales and are also presented as dichotomous groupings around the median. CI denotes confidence interval.
Odds ratios were used to estimate relative risk.
The relative risk was adjusted for the initial white-cell count and the interval from the onset of diarrhea to the day on which the stool culture was obtained.
This category served as the reference group.
The category of fever was based on the caregivers’ reports and was added to the standardized questionnaire on December 1, 1998; thus, data on only 24 children were included in the analysis.
The initial temperature was documented in 18 children who presented to Children’s Hospital and Regional Medical Center in Seattle (2 of whom had the hemolytic–uremic syndrome).
To convert values for serum urea nitrogen to millimoles per liter, multiply by 0.357.
To convert values for serum creatinine to micromoles per liter, multiply by 88.4.
Multivariate analysis, adjusted for the day on which the initial stool culture was obtained and the initial white-cell count, showed that children treated with antibiotics had a higher risk of the hemolytic–uremic syndrome than did children who did not receive antibiotics (adjusted relative risk, 17.3; 95 percent confidence interval, 2.2 to 137; P=0.007). Among the children who received antibiotics within the first three days after the onset of illness, the risk of the syndrome was increased in the univariate analysis (relative risk, 15.0; 95 percent confidence interval, 1.3 to 174; P=0.03) and after adjustment for the day on which the initial stool culture was obtained and the initial white-cell count in the multivariate analysis (adjusted relative risk, 32.3; 95 percent confidence interval, 1.4 to 737; P=0.03). Adjustment for other base-line covariates did not affect the significance of the association between antibiotic treatment and the development of the hemolytic–uremic syndrome (data not shown).
Of the 10 children in whom the hemolytic–uremic syndrome developed, 2 were treated with trimethoprim–sulfamethoxazole and 3 were treated with cephalosporins. Among the four antibiotic-treated children in whom the syndrome did not develop, one received trimethoprim–sulfamethoxazole, one a cephalosporin, and two amoxicillin. The relative risk of the hemolytic–uremic syndrome as a function of the class of antibiotic used, after adjustment for the initial white-cell count, was 17.7 for trimethoprim–sulfamethoxazole (95 percent confidence interval, 1.2 to 261; P=0.04) and 13.4 for β-lactam antibiotics (95 percent confidence interval, 1.9 to 96; P=0.01). In each case, all E. coli O157:H7 strains isolated from the nine children who received antibiotics were susceptible to the drug taken by the child from whom the organisms were recovered.
DISCUSSION
This prospective cohort study demonstrates that among children infected with E. coli O157:H7, the hemolytic–uremic syndrome developed more often in those who were given antibiotics than in those who were not. We were unable to identify any characteristics that differentiated children who received antibiotics from those who did not.
Our data confirm that administering sulfa-containing antibiotics to children infected with E. coli O157:H7 increases their risk of the hemolytic–uremic syndrome8,9 and indicate that β-lactam antibiotics are associated with a similar degree of risk. An extensive analysis of the E. coli O157:H7 outbreak in Sakai City, Japan, suggested that treatment with fos-fomycin was associated with a significantly decreased risk of the hemolytic–uremic syndrome,16 but only for children who received the drug on the second day, but not on other days, of their illness. Also, fos-fomycin was compared only with other antibiotics, not with the absence of treatment with antibiotics, because as in other reports describing this outbreak,16–25 almost all children received antibiotics.
We noticed an inverse relation between the risk of the hemolytic–uremic syndrome and the length of the interval between the onset of diarrhea and the day on which the initial stool culture was obtained. This inverse relation could represent the presence of a more fulminant course that prompted earlier evaluation and therefore reflected more severe extraintestinal injury in infected children who were at greater risk for the syndrome whether or not antibiotics were prescribed. However, multivariate analysis did not diminish the risk associated with antibiotic administration when this factor was considered.
We did not find an association between the reported presence of fever and the risk of the hemolytic–uremic syndrome, as has been found in some studies.8,16 However, prehospitalization temperatures were determined by a variety of caregivers, using different instruments and techniques. Moreover, of the 18 children who were evaluated at Children’s Hospital and Regional Medical Center, only 1 had a temperature exceeding 38.0°C on initial evaluation, and the hemolytic–uremic syndrome did not develop in this child. Also, despite the absence of an association between treatment with antimotility drugs or opioid narcotics and the risk of the hemolytic–uremic syndrome, we recommend against the use of these drugs in children with acute diarrhea, because of their reported association with complications of E. coli O157:H7 infection and with the prolongation of symptoms.1,24,26–28
Our data may have been confounded by selection bias, because we studied only children whose stools yielded E. coli O157:H7 as identified by our surveillance network. However, we are unable to assess the effect of antibiotic treatment in children infected with this pathogen whose illness was not diagnosed by our system of participating laboratories.
There are considerable data to support an association between the hemolytic–uremic syndrome and enteric infection with Shiga-toxin–producing bacteria.2,5 Cytotoxin, presumably Shiga toxin, has been found in the stools of patients with both E. coli O157:H7 infection and the hemolytic–uremic syndrome.5,29 In in vitro experiments, exposure to various antibiotics causes E. coli to release Shiga toxin.30–33 Antibiotics might increase the risk of the hemolytic–uremic syndrome by causing the release of Shiga toxin from injured bacteria in the intestine, making the toxin more available for absorption.
An observational study cannot completely eliminate other potential biases with respect to the association of antibiotics and the risk of the hemolytic–uremic syndrome, but our analysis has several strengths in comparison with previous attempts to assess the risks or benefits of antibiotic treatment of E. coli O157:H7 infections. Children’s caregivers were questioned prospectively during the acute illness, reducing problems with recall that may occur when interviews are performed weeks or months after infection. Such a bias might pertain especially to the use of non-prescription drugs. Also, our study addressed the use of antibiotics in children infected with a variety of strains of E. coli O157:H7 and was not limited to children infected with the same strain, as is the case in analyses of outbreaks.1,6,8,10,16–22 Thus, our data are broadly applicable to clinical practice, where physicians encounter children infected with one of many subtypes of this pathogen. Indeed, the variability among E. coli O157:H7 strains in the antibiotic-induced release of Shiga toxin30 also suggests that conclusions from analyses of outbreaks may not be easily generalizable.
The association between antibiotic treatment and the development of the hemoltyic–uremic syndrome in children with E. coli O157:H7 infections is strong and plausible. We therefore recommend against giving antibiotics to children who may be infected with E. coli O157:H7 until the results of a stool culture indicate that the pathogen responsible is one that is appropriately treated by an antibiotic. Even if the small advantage associated with empirical fluoroquinolone therapy in some adults with acute diarrhea34–36 holds true for children, we believe that the risk of administering antibiotics to children who might be infected with pathogens for which antibiotics are contraindicated (i.e., E. coli O157:H7) exceeds the potential benefit. However, it is important to remember that the hemolytic–uremic syndrome can occur even in infected children who have not been treated with antibiotics. Therefore, we believe that the hemolytic–uremic syndrome is best prevented by avoiding primary infection.
Acknowledgments
Supported by a grant (1RO1DK52081) from the National Institutes of Health.
We are indebted to Christine A. Merrikin and Kaye Green for expert secretarial assistance, to Dr. Yukiko Ikeda and Dr. Dong-Ki Lee for translating Japanese studies, and to participating patients, families, microbiologists and other laboratorians, nurses, and physicians for providing data used in the study.
Footnotes
Presented in part at the 35th United States–Japan Cholera and Related Diarrheal Diseases Conference, Baltimore, December 3–5, 1999.
References
- 1.Bell BP, Griffin PM, Lozano P, Christie DL, Kobayashi JM, Tarr PI. Predictors of hemolytic uremic syndrome in children during a large outbreak of Escherichia coli O157:H7 infections. Pediatrics. 1997;100:127. doi: 10.1542/peds.100.1.e12. abstract. (See http://www.pediatrics.org/cgi/content/full/100/1/e12.) [DOI] [PubMed] [Google Scholar]
- 2.Tarr PI, Neill MA, Clausen CR, Watkins SL, Christie DL, Hickman RO. Escherichia coli O157:H7 and the hemolytic uremic syndrome: importance of early cultures in establishing the etiology. J Infect Dis. 1990;162:553–6. doi: 10.1093/infdis/162.2.553. [DOI] [PubMed] [Google Scholar]
- 3.Ostroff SM, Kobayashi JM, Lewis JH. Infections with Escherichia coli O157:H7 in Washington State: the first year of statewide disease surveillance. JAMA. 1989;262:355–9. [PubMed] [Google Scholar]
- 4.Rowe P, Orrbine E, Lior H, et al. Risk of hemolytic uremic syndrome after sporadic Escherichia coli O157:H7 infection: results of a Canadian collaborative study. J Pediatr. 1998;132:777–82. doi: 10.1016/s0022-3476(98)70303-8. [DOI] [PubMed] [Google Scholar]
- 5.Karmali MA, Steele BT, Petric M, Lim C. Sporadic cases of haemolytic-uraemic syndrome associated with faecal cytotoxin and cytotoxin-producing Escherichia coli in stools. Lancet. 1983;1:619–20. doi: 10.1016/s0140-6736(83)91795-6. [DOI] [PubMed] [Google Scholar]
- 6.Ryan CA, Tauxe RV, Hosek GW, et al. Escherichia coli O157:H7 diarrhea in a nursing home: clinical, epidemiological, and pathological findings. J Infect Dis. 1986;154:631–8. doi: 10.1093/infdis/154.4.631. [DOI] [PubMed] [Google Scholar]
- 7.Neill MA. Treatment of disease due to Shiga toxin-producing Escherichia coli: infectious disease management. In: Kaper JB, O’Brien AD, editors. Escherichia coli O157:H7 and other Shiga toxin-producing E. coli strains. Washington, D.C: ASM Press; 1998. pp. 357–63. [Google Scholar]
- 8.Pavia AT, Nichols CR, Green DP, et al. Hemolytic-uremic syndrome during an outbreak of Escherichia coli O157:H7 infections in institutions for mentally retarded persons: clinical and epidemiologic observations. J Pediatr. 1990;116:544–51. doi: 10.1016/s0022-3476(05)81600-2. [DOI] [PubMed] [Google Scholar]
- 9.Slutsker L, Ries AA, Maloney K, Wells JG, Greene KD, Griffin PM. A nationwide case-control study of Escherichia coli O157:H7 infection in the United States. J Infect Dis. 1998;177:962–6. doi: 10.1086/515258. [DOI] [PubMed] [Google Scholar]
- 10.Carter AO, Borczyk AA, Carlson JAK, et al. A severe outbreak of Escherichia coli O157:H7–associated hemorrhagic colitis in a nursing home. N Engl J Med. 1987;317:1496–500. doi: 10.1056/NEJM198712103172403. [DOI] [PubMed] [Google Scholar]
- 11.Proulx F, Turgeon JP, Delage G, Lafleur L, Chicoine L. Randomized, controlled trial of antibiotic therapy for Escherichia coli O157:H7 enteritis. J Pediatr. 1992;121:299–303. doi: 10.1016/s0022-3476(05)81209-0. [DOI] [PubMed] [Google Scholar]
- 12.Igarashi T, Inatomi J, Wake A, Takamizawa M, Katayama H, Iwata T. Failure of pre-diarrheal antibiotics to prevent hemolytic uremic syndrome in serologically proven Escherichia coli O157:H7 gastrointestinal infection. J Pediatr. 1999;135:768–9. doi: 10.1016/s0022-3476(99)70100-9. [DOI] [PubMed] [Google Scholar]
- 13.March SB, Ratnam S. Sorbitol-MacConkey medium for detection of Escherichia coli O157:H7 associated with hemorrhagic colitis. J Clin Microbiol. 1986;23:869–72. doi: 10.1128/jcm.23.5.869-872.1986. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Performance standards for antimicrobial disk susceptibility tests: approved standard. 7. 1. Vol. 20. Wayne, Pa: National Committee for Clinical Laboratory Standards; Jan, 2000. (NCCLS no. M2-A7.) [Google Scholar]
- 15.Colton T. Statistics in medicine. Boston: Little, Brown; 1974. [Google Scholar]
- 16.Ikeda K, Ida O, Kimoto K, Takatorige T, Nakanishi N, Tatara K. Effect of early fosfomycin treatment on prevention of hemolytic uremic syndrome accompanying Escherichia coli O157:H7 infection. Clin Nephrol. 1999;52:357–62. [PubMed] [Google Scholar]
- 17.Fukushima H, Hashizume T, Morita Y, et al. Clinical experiences in Sakai City Hospital during the massive outbreak of enterohemorrhagic Escherichia coli O157 infections in Sakai City, 1996. Pediatr Int. 1999;41:213–7. doi: 10.1046/j.1442-200x.1999.4121041.x. [DOI] [PubMed] [Google Scholar]
- 18.Higami S, Nishimoto K, Kawamura T, Tsuruhara T, Isshiki G, Ookita A. Retrospective analysis of the relationship between HUS incidence and antibiotics among patients with Escherichia coli O157 enterocolitis in the Sakai outbreak. Kansenshogaku Zasshi. 1998;72:266–72. doi: 10.11150/kansenshogakuzasshi1970.72.266. (In Japanese.) [DOI] [PubMed] [Google Scholar]
- 19.Kohsaka T, Tagawa M, Suzuki T, Ito H. The old and new problems of the antibiotics treatment in hemorrhagic enterocolitis and hemolytic uremic syndrome, according to Japanese epidemiological studies. Nippon Rinsho. 1997;55:706–14. (In Japanese.) [PubMed] [Google Scholar]
- 20.Moriguchi N, Yagi K, Yamamoto T, Yoshioka K, Kubo S. The drug sensitivity of enterohemorrhagic Escherichia coli and antibiotics treatment for hemorrhagic enterocolitis — from an outbreak of enterocolitis in Sakai City. Jpn J Antibiot. 1997;50:591–6. (In Japanese.) [PubMed] [Google Scholar]
- 21.Oshima T. Predictive factors for development of hemolytic uremic syndrome (HUS) and early intensive treatments for prevention of HUS enterohemorrhagic Escherichia coli infection. Jpn J Antibiot. 1997;50:855–61. (In Japanese.) [PubMed] [Google Scholar]
- 22.Shiomi M, Togawa M, Fujita K, Murata R. Effect of early oral fluoroquinolones in hemorrhagic colitis due to Escherichia coli O157:H7. Pediatr Int. 1999;41:228–32. doi: 10.1046/j.1442-200x.1999.4121038.x. [DOI] [PubMed] [Google Scholar]
- 23.Chisaki T, Hinotani K, Shimizu T, Miyazaki K. Initial treatment at an outbreak of E. coli O-157:H7 infection: especially with respect to therapy in the emergency. Jpn J Antibiot. 1997;50:821–8. (In Japanese.) [PubMed] [Google Scholar]
- 24.Yoh M, Aoki T, Akao M, Sakaue Y, Tsubura E, Honda T. Report of questionnaire about enterohaemorrhagic Escherichia coli cases caused in the area including Sakai City in 1996. Kansenshogaku Zasshi. 1997;71:1144–54. doi: 10.11150/kansenshogakuzasshi1970.71.1144. (In Japanese.) [DOI] [PubMed] [Google Scholar]
- 25.Kawamura N, Yamazaki T, Tamai H. Risk factors for the development of Escherichia coli O157:H7 associated with hemolytic uremic syndrome. Pediatr Int. 1999;41:218–22. doi: 10.1046/j.1442-200x.1999.4121040.x. [DOI] [PubMed] [Google Scholar]
- 26.Cimolai N, Basalyga S, Mah DG, Morrison BJ, Carter JE. A continuing assessment of risk factors for the development of Escherichia coli 0157:H7-associated hemolytic uremic syndrome. Clin Nephrol. 1994;42:85–9. [PubMed] [Google Scholar]
- 27.Cimolai N, Morrison BJ, Carter JE. Risk factors for the central nervous system manifestations of gastroenteritis-associated hemolytic-uremic syndrome. Pediatrics. 1992;90:616–21. [PubMed] [Google Scholar]
- 28.Cimolai N, Carter JE, Morrison BJ, Anderson JD. Risk factors for the progression of Escherichia coli O157:H7 enteritis to hemolytic-uremic syndrome. J Pediatr. 1990;116:589–92. doi: 10.1016/s0022-3476(05)81609-9. [Erratum, J Pediatr 1990;116:1008.] [DOI] [PubMed] [Google Scholar]
- 29.Karmali MA, Petric M, Lim C, Fleming PC, Arbus GS, Lior H. The association between idiopathic hemolytic uremic syndrome and infection by verotoxin-producing Escherichia coli. J Infect Dis. 1985;151:775–82. doi: 10.1093/infdis/151.5.775. [DOI] [PubMed] [Google Scholar]
- 30.Grif K, Dierich MP, Karch H, Allerberger F. Strain-specific differences in the amount of Shiga toxin released from enterohemorrhagic Escherichia coli O157 following exposure to subinhibitory concentrations of antimicrobial agents. Eur J Clin Microbiol Infect Dis. 1998;17:761–6. doi: 10.1007/s100960050181. [DOI] [PubMed] [Google Scholar]
- 31.Kimmitt PT, Harwood CR, Barer MR. Induction of type 2 Shiga toxin synthesis in Escherichia coli O157 by 4-quinolones. Lancet. 1999;353:1588–9. doi: 10.1016/s0140-6736(99)00621-2. [DOI] [PubMed] [Google Scholar]
- 32.Walterspiel JN, Ashkenazi S, Morrow AL, Cleary TG. Effect of subinhibitory concentrations of antibiotics on extracellular Shiga-like toxin I. Infection. 1992;20:25–9. doi: 10.1007/BF01704889. [DOI] [PubMed] [Google Scholar]
- 33.Karch H, Strockbine NA, O’Brien AD. Growth of Escherichia coli in the presence of trimethoprim-sulfamethoxazole facilitates detection of Shiga-like toxin producing strains by colony blot assay. FEMS Microbiol Lett. 1986;35:141–5. [Google Scholar]
- 34.Goodman LJ, Trenholme GM, Kaplan RL, et al. Empiric antimicrobial therapy of domestically acquired acute diarrhea in urban adults. Arch Intern Med. 1990;150:541–6. [PubMed] [Google Scholar]
- 35.Wistrom J, Jertborn M, Ekwall E, et al. Empiric treatment of acute diarrheal disease with norfloxacin: a randomized placebo-controlled study. Ann Intern Med. 1992;117:202–8. doi: 10.7326/0003-4819-117-3-202. [DOI] [PubMed] [Google Scholar]
- 36.Dryden MS, Gabb RJ, Wright SK. Empirical treatment of severe acute community-acquired gastroenteritis with ciprofloxacin. Clin Infect Dis. 1996;22:1019–25. doi: 10.1093/clinids/22.6.1019. [DOI] [PubMed] [Google Scholar]