Association of Intrapartum Antibiotic Exposure and Late-Onset Serious Bacterial Infections in Infants
http://www.100md.com
《小儿科》
Divisions of General PediatricsInternal Medicine Infectious Diseases Pediatric Infectious Diseases, University of Utah, Salt Lake City, Utah
ABSTRACT
Objective. Recommendations to prevent vertical transmission of group B Streptococcus (GBS) infections have resulted in many women's receiving antibiotics during labor with an associated reduction in early-onset GBS infections in their newborn infants. However, a potential relationship of intrapartum antibiotics (IPA) to the occurrence of late-onset (7–90 days) serious bacterial infections (SBIs) in term infants has not been reported. The objectives of this study were to determine whether infants with late-onset SBI were more likely than healthy control infants to have been exposed to IPA and whether there was a greater likelihood of antibiotic resistance in bacteria that were isolated from infants who had an SBI and had been exposed to IPA compared with those who had not.
Methods. We used a case-control design to study the first objective. Cases were previously healthy full-term infants who were hospitalized for late-onset SBI between the ages of 7 and 90 days. Control subjects were healthy full-term infants who were known not to have an SBI in their first 90 days. Cases and control subjects were matched for hospital of delivery. In the second part of the study, rates of antibiotic resistance of bacteria that were isolated from infected infants were compared for those who had and had not been exposed to IPA.
Results. Ninety case infants and 92 control subjects were studied. Considering all types of IPA, more case (41%) than control infants (27%) had been exposed to IPA (adjusted odds ratio [OR]: 1.96; 95% confidence interval [CI]: 1.05–3.66), after controlling for hospital of delivery. The association was stronger when IPA was with broad-spectrum antibiotics (adjusted OR: 4.95; 95% CI: 2.04–11.98), after controlling for hospital of delivery, penicillin IPA, maternal chorioamnionitis, and breastfeeding. Bacteria that were isolated from infected infants who had been exposed to IPA were more likely to exhibit ampicillin resistance (adjusted OR: 5.7; 95% CI: 2.3–14.3), after controlling for hospital of delivery, but not to other antibiotics that are commonly used to treat SBI in infants.
Conclusions. After adjusting for potential confounders, infants with late-onset SBI were more likely to have been exposed to IPA than noninfected control infants. Pathogens that cause late-onset SBI were more likely to be resistant to ampicillin when the infant had been exposed to intrapartum antibiotics.
Key Words: intrapartum antibiotics serious bacterial infection term infant antibiotic resistance group B Streptococcus
Abbreviations: GBS, group B Streptococcus SBI, serious bacterial infection IPA, intrapartum antibiotic PCMC, Primary Children's Medical Center OR, odds ratio UTI, urinary tract infection CI, confidence interval
National recommendations designed to prevent the vertical transmission of group B Streptococcus (GBS) have resulted in many women's receiving antibiotics during labor.1,2 Data obtained from the largest insurer in the state of Utah (Intermountain Health Care) indicate that, between 1998 and 2002, 35% of mothers who delivered term infants had received intrapartum antibiotics (B.M. James, MD, MStat, Intermountain Health Care Institute for Health Care Delivery Research, personal communication, 2003). Intrapartum antibiotic treatment has been very successful in preventing early-onset neonatal GBS infection.3 However, some investigators have reported an increase in early-onset neonatal infection in preterm infants caused by antibiotic-resistant Gram-negative organisms.4–7 It has been suggested that this increase may be attributable, in part, to maternal antibiotic exposure before delivery of the infant,8,9 and several studies have reported an association between maternal antibiotic therapy and early-onset infection in preterm infants with pathogens resistant to antibiotics.10–12 These studies all focused on early-onset infection, leaving the question of late effects of maternal antibiotic treatment unaddressed. Antibiotics are known to alter gastrointestinal and vaginal bacterial flora, selecting for resistant pathogens.13 Early-onset serious bacterial infections (SBIs) occur within the first week and are attributed to vertical transmission of pathogens from mother to infant during labor and delivery. Late-onset SBIs occur after the seventh day and have been considered to be community-acquired infections. The incidence of late-onset SBIs in former term infants is 1.6 per 1000 births.14
In the process of caring for infants with late-onset SBI, the investigators observed what seemed to be an increase in the proportion of these infections that were caused by bacteria resistant to antibiotics and wondered whether this could be attributable, in part, to the large-scale delivery of broad-spectrum antibiotics during labor. Although the Centers for Disease Control and Prevention recommends penicillin for GBS prophylaxis, the majority of women in Utah receive the broader spectrum ampicillin.15 To explore this potential relationship, we developed 2 research questions: (1) Could exposure to intrapartum antibiotics increase the likelihood of an infant's acquiring a late-onset SBI (2) Are the organisms that cause SBIs more likely to demonstrate antibiotic resistance when the infant has been exposed to intrapartum antibiotics Although a prospective study that follows 2 cohorts of infants (those who had and who had not been exposed to intrapartum antibiotics [IPAs]) and notes their respective rates infection might be an appropriate design to answer these questions, the relative rarity of late-onset SBI (1.6 per 1000 births14) makes this approach impractical. We therefore used a case-control design to study these questions.16 We compared the proportion of a cohort of infants with SBI (cases) who had been exposed to IPAs with the proportion of a cohort of healthy infants (control subjects) who had been born during the same time period and at the same hospitals. We hypothesized that the proportion of infants who had late-onset SBI and had a history of IPA exposure would be higher than control subjects. In the second part of the study, we focused on the infected infants and examined the bacterial resistance patterns of the organisms that were isolated from them. We hypothesized that the proportion of bacteria that demonstrated antibiotic resistance would be higher in the infants who were exposed to IPAs.
METHODS
Approval to conduct this research was obtained by the Institutional Review Boards of the University of Utah, Intermountain Health Care, and St Marks Hospital (Salt Lake City, UT).
Study Patients
Cases were previously healthy former term (37 weeks' gestation) infants who were between the ages of 7 and 90 days and were hospitalized at Primary Children's Medical Center (PCMC) with am SBI between July 1999 and May 2003. PCMC is a 252-bed children's hospital that serves as the only pediatric hospital for children from the Salt Lake City metropolitan area. None of the other hospitals in the area provides nonnursery, inpatient care for infants. Infants with SBIs (cases) were identified using 2 methods. Between July 1, 2002, and May 31, 2003, cases were identified by daily examination of the PCMC microbiology culture logs (see below). Additional cases were identified by reviewing the database of a previous 6-year study of late-onset infections in infants between 7 and 90 days of age conducted by 1 of the authors (C.L.B.).17 To facilitate contact with the parents of this group of infants, we limited our study to those who had been enrolled in that study between 1999 and 2002. Control infants were recruited by cooperating primary care physicians in the Salt Lake City area. Parents of former term infants, who were being seen for their 4-month well-child visit and who had not had an SBI, were invited to participate. Parents of case and control infants who agreed provided informed consent for the investigators to review maternal and infant medical records.
Review of Microbiologic Records
The methods and the criteria used for identifying infants with a late-onset SBI have been described previously17 and were identical to the methods used to identify cases seen after July 1, 2001. The computerized microbiologic records maintained by PCMC were reviewed daily to identify infants with a possible SBI. SBI was defined as the presence of pathogenic bacteria cultured from blood, urine, or cerebrospinal fluid. Urine cultures were considered positive when >50000 colony-forming units/mL of a single organism were isolated from a catheterized specimen.18 For bacteria that could be considered contaminants, such as Viridans group streptococci, SBI was diagnosed only when the infant had 2 separate positive cultures. We considered omphalitis to be an SBI when there was evidence of a clinical abdominal wall infection and when a single pathogenic organism was isolated from an umbilical swab. The antibiotic susceptibility for each pathogen was determined following recommendations of the National Committee of Clinical Laboratory Standards.19,20
Maternal Interview
Once cases of SBI were obtained from the PCMC records and after consent was obtained, the infant's mother was interviewed using a structured format. Data regarding the mother's age, the infant's race, hospital of delivery, the mother's recollection of receiving IPA treatment, antibiotic treatment of the infant, and breastfeeding were obtained. Interviews were conducted in person for the 2002–2003 cases and by telephone for the cases from 1999 to 2002. The same information was obtained from mothers of control infants by a questionnaire that was given to them at the well-child visit. In-person and telephone interviews, questionnaires, and consent forms were administered in Spanish when appropriate.
Delivery Hospital Medical Record Review
Two reviewers (T.G. and C.K.) reviewed the peripartum medical records of all mothers and infants using a structured abstraction format. Maternal records were reviewed for documentation of IPA administration, the type of antibiotic administered, the number of doses, reason for antibiotic treatment (eg, chorioamnionitis), GBS culture results, maternal fever during labor, prolonged rupture of membranes, method of delivery, receipt of prenatal care, mother's race/ethnicity, and type of health insurance. We considered IPAs to have been given when the record indicated that an antibiotic agent had been administered to the mother during labor at any time before the delivery of her infant. The category broad-spectrum IPAs was applied whenever an antibiotic other than penicillin had been given. Infant records were reviewed for gender, presence of breastfeeding, and a history of antibiotic treatment after delivery.
Sample Size Estimates and Data Analysis
On the basis of the literature and local data, we estimated that 30% of women would have a history of IPA exposure. On the basis of this estimate, a sample size of 93 in each cohort would be required to have 80% power to detect an absolute 20% difference in either direction between IPA rates for cases and control subjects. Our achieved sample size of 90 cases and 92 control subjects provided 80% power to detect a 20.1% difference. Data were analyzed (G.S. and S.F.) using STATA statistical software (Stata Corp, College Station, TX). To test the first hypothesis, we compared the proportion of infants who were exposed to an IPA for cases and control subjects using conditional logistic regression stratified by hospital of delivery. One delivery hospital was overrepresented in the cohort of control subjects. The slight imbalance in numbers of cases and control subjects, 90 and 92, did not create a problem for analysis, because the conditional logistic regression model compared the cases as a group with the control subjects as a group within the same hospital stratum, rather than pairwise case-control comparisons. Odds ratios (ORs) for IPA exposure were calculated for the 2 cohorts. The other study variables including race, type of insurance, prenatal care, type of delivery, GBS colonization, maternal fever, chorioamnionitis, prolonged rupture of membranes, infant's gender, and breastfeeding then were entered individually into the model. All reported ORs are adjusted for hospital of delivery, as all ORs come from a conditional logistic regression that stratifies on this variable. Any variables that altered the hospital of delivery adjusted OR by >10% were placed into the full conditional logistic regression model.
Next, we performed a separate analysis of the cohort of infected infants. Associations between receipt of IPAs, type of antibiotic, number of doses (1, 2, or >2), and antibiotic-resistant pathogens were identified by 2 or Fisher exact test as appropriate. Other study variables, including race, type of insurance, prenatal care, type of delivery, GBS colonization, maternal fever, chorioamnionitis, hospital of delivery, gender, and breastfeeding, were analyzed individually to determine whether they were associated with the presence of antibiotic resistance. Variables that altered the OR by 10% or more were placed into a multiple logistic regression model. ORs for infection with a resistant pathogen for infants who were exposed to IPAs compared with those who were not exposed were generated. Finally, we performed a similar analysis comparing each individual type of IPA with no IPAs.
RESULTS
Patient Enrollment
A total of 90 infants with SBI and their mothers were enrolled. Forty were identified from the database of the previous study17 that contained 101 infants who were between the ages of 7 and 90 days, had an SBI, and had been admitted to PCMC between July 1999 and June 2002. Sixty of the mothers of these infants could not be contacted, and 1 who was contacted refused participation. There were no significant differences between these 61 infants and the 40 enrolled infants with regard to gender, type of SBI, type of pathogen, or antibiotic resistance. Fifty infants with SBI were identified from the PCMC microbiology logs between July 2002 and May 2003; the parents of all of these infants consented for the study. Community pediatricians identified a total of 187 potential control infants, and the parents of all 187 verbally agreed to participate. When control subjects were matched to cases at the hospital level, which involved randomly selecting from the available control subjects for 1 hospital that was overrepresented, we ended up with 90 cases and 92 control subjects who completed the formal consenting process and became enrolled in the study.
Infant deliveries were distributed among 7 Salt Lake City area hospitals. The demographics of the mothers and infants are listed in Table 1. There were no differences between cases and control subjects with respect to the demographic variables. Notably, all but 1 of the mothers had received prenatal care. The types of infection are listed in Table 2. The majority (72%) of infections were uncomplicated urinary tract infections (UTIs). However, 28% of infants had potentially life-threatening infections, including bacteremia and meningitis. The pathogens isolated are shown in Table 3.
IPA Exposure in Infants With SBI Compared With Control Subjects
As hypothesized, more of the cases (41%) than control subjects (27%) had been exposed to IPAs (OR: 1.96; 95% confidence interval [CI]: 1.05–3.66). There were no significant differences between cases and control subjects with respect to maternal fever (OR: 0.50; 95% CI: 0.13–1.9), prolonged rupture of membranes (OR: 2.16; 95% CI: 0.41–11.27), GBS colonization (OR: 0.90; 95% CI: 0.36–2.25), cesarean delivery (OR: 0.76; 95% CI: 0.34–1.67), gender of infant (OR: 0.99; 95% CI: 0.55–1.78), and private insurance compared with Medicaid (OR: 1.12; 95% CI: 0.43–2.97). Table 4 shows the final regression model comparing case and control infants and includes the variables (type of IPA [broad-spectrum or penicillin], breastfeeding, and maternal chorioamnionitis) that altered the hospital of delivery adjusted IPA OR by >10%. In the full model, only the use of a broad-spectrum antibiotic remained significant with an OR of 4.95 (95% CI: 2.04–11.98).
Antibiotic Resistance and IPAs
In the cohort of 90 infants with SBI, 41% of the responsible isolates were resistant to ampicillin. None was resistant to gentamicin or to third-generation cephalosporins. Resistance to ampicillin was strongly associated with IPA exposure. As shown in Table 5, 24 (65%) of pathogens that were isolated from the 37 infants who were exposed to IPAs were resistant to ampicillin compared with 13 (25%) of the pathogens that were isolated from the 43 infants who were not exposed to IPAs (OR: 5.7; 95% CI: 2.3–14.3). This relationship seems to be accounted for by the use of broad-spectrum antibiotics, because, as shown, rates of bacterial resistance for infants whose mothers had received penicillin were the same as that for infants whose mothers had not received IPAs. When either ampicillin or another broad-spectrum antibiotic had been used, the associations with ampicillin resistance were significant (OR: 6.1; 95% CI: 1.9–19.7) for ampicillin, and for other broad-spectrum antibiotics, the OR was 12.3 (95% CI: 2.3–65.5). Proximity to the IPA exposure did not seem to account for resistance because it was as likely to be present in infants who developed an SBI before or after their first month. There were no differences in resistance rates related to the number of doses that the infant's mother had received (data not shown).
Infants with UTIs and other types of SBI were analyzed separately. UTIs in infants who were exposed to IPAs were more likely to be caused by an ampicillin-resistant pathogen compared with those who were not exposed (OR: 4.3; 95% CI: 1.6–11.7). Even in the relatively small number of infants with other SBIs (meningitis, omphalitis, and bacteremia without UTI), the relationship between infection with an ampicillin-resistant organism and IPAs achieved statistical significance, although with wide CIs (OR: 25; 95% CI: 1.8–346; Fisher exact P = .013). Gender, type of insurance, race, prenatal care, cesarean delivery, maternal fever, prolonged rupture of membranes, chorioamnionitis, GBS colonization, treatment of the infant with antibiotics, hospital of delivery, and breastfeeding had no significant effect on rates of infection with an antibiotic-resistant pathogen in this group of infants.
Eleven infants were infected with constitutively ampicillin-resistant organisms: Klebsiella species, Enterobacter species, and Staphylococcus aureus. Nine had been exposed to IPAs with broad-spectrum agents. Our review of the mother's medical record indicated that, in all, the reason for the IPAs had been GBS prophylaxis; none had a diagnosis of chorioamnionitis, pyelonephritis, or cellulitis.
We compared the mothers' recollection during the interview of whether they had received IPAs with data in the medical record. Of the 277 women in the total sample, the recollection of 235 (85%) was accurate. The 15% of women whose recall differed from the data in the medical record were as likely to over- as underreport IPA treatment.
DISCUSSION
Having noted that 35% of mothers were receiving antibiotics during labor, we wondered whether this practice could be resulting in a change in either the frequency of late-onset SBIs or the antimicrobial resistance patterns of the bacteria that cause these infections in former term infants. The findings of this case-control study support our first hypothesis that more infants with late-onset SBI would have a history of IPA exposure than control infants. The second hypothesis, that, in infected infants rates of antibiotic resistance would be higher in those who were exposed to IPAs, was only partially supported. We found that the responsible organisms in these infants were more likely to be resistant to ampicillin but not to other antibiotics, such as gentamicin or third-generation cephalosporins. Both of these findings seem to be related to the use of broad-spectrum antibiotics during labor rather than penicillin, which is the recommended agent. For example, when we compared the rates and types of IPA exposure in infants with and without SBI, we found that the rates of exposure to penicillin were the same for the infants with SBI and control subjects but that the rates of exposure to broad-spectrum IPAs were much higher for cases compared with control subjects (data not shown). This increase seems to be attributable largely to the impact of ampicillin-resistant organisms. In the infants with an SBI, the rates of infection with an ampicillin-resistant pathogen were higher in the group that was exposed to IPAs than in nonexposed infants, but the rates of infections with ampicillin-susceptible pathogens were the same. We believe that these findings have important implications for both the selection of an appropriate choice of antibiotics for GBS prophylaxis and the management of late-onset SBI in infants. First, penicillin should be the preferred agent for GBS prophylaxis; second, a history of exposure to IPAs should be obtained whenever a clinician is evaluating an infant between 7 and 90 days for the possibility of an SBI. Maternal recall of either receiving or not receiving IPAs is likely to be accurate: 85% in this heterogeneous population.
This is the first study to evaluate the association between IPAs and late-onset SBI in term infants, but our findings are consistent with previous studies that have demonstrated an increase in antibiotic-resistant pathogens and an association of IPAs with infection with antibiotic-resistant pathogens in preterm infants with early-onset sepsis (first 7 days of life).6,8,10
Normal neonates become colonized with bacterial flora that mirror maternal organisms.21,22 Antibiotics have been shown to decrease normal gastrointestinal flora such as Lactobacillus species and Bifidobacteria species and select for antibiotic-resistant pathogens such as Klebsiella species, Enterobacter species, and S aureus.13,23–27 In our study population, almost all of the infants who were infected with these pathogens were exposed to broad-spectrum IPAs. We believe that these organisms are transferred to the infant both vertically at delivery and, more important, horizontally during the first weeks after birth.21,28 Thus, broad-spectrum IPAs may alter both maternal and infant bacterial flora and select for antibiotic-resistant pathogens. We suspect that these antibiotic-resistant pathogens are more invasive and that, as a result, infants who are colonized with these organisms are more likely to develop an SBI.
In our study, IPAs were associated with an increased risk for infections with ampicillin-resistant bacteria but not with infections caused by ampicillin-susceptible organisms. The acquisition of antibiotic-resistant infections in community settings is complex. Factors such as population density and antibiotics in the environment (eg, animal feed, cleaning products, child care) contribute to community resistance patterns.29–33 Recent exposure to antibiotics (30–120 days before infection) has consistently been identified as a risk factor for acquisition of resistant pathogens.34–36 In neonates, antibiotic exposure can disturb normal initial intestinal colonization. This effect may be persistent. In our study population, infants who were between 60 and 90 days of age and exposed to IPAs were just as likely to have an ampicillin-resistant pathogen as the infants who were younger than 30 days. Although many factors may be involved in the acquisition of antibiotic-resistant pathogens in neonates, the possible association of IPAs cannot be ignored.
The majority of the mothers who received IPAs in this study were treated with broad-spectrum antibiotics such as ampicillin rather than penicillin. Because of its narrower spectrum, penicillin was recommended for GBS prophylaxis in both the 1997 and 2002 Centers for Disease Control and Prevention guidelines.1,2 Unlike broad-spectrum antibiotics, penicillin does not perturb normal gastrointestinal flora.26,27 Infected infants whose mothers had received intrapartum penicillin were no more likely to be infected with a resistant organism than infected infants whose mothers did not receive an IPA. In a separate analysis from the case-control portion of the study, we compared the rates of intrapartum penicillin treatment between cases and control subjects and found that they were the same. This finding should be interpreted cautiously, however, because the numbers of mothers who received penicillin was small, and there may be insufficient power to detect a small but possibly significant difference. However, we believe that our findings do provide support for choosing penicillin rather than ampicillin for GBS prophylaxis.
Clinicians who are faced with an infant in whom an SBI may be present are required to choose an antibiotic regimen before culture and sensitivity results become available. Our findings suggest that a history of IPA exposure should influence this choice. In a previous study from our institution, initial antibiotic therapy was changed in more than half of the cases of late-onset SBI after identification of the pathogenic organism.17 In addition, 64% of patients with bacteremia or meningitis caused by ampicillin-resistant organisms had documented treatment failure of their initial antibiotic regimen.17 In the current study, 62% of the 20 infants with bacteremia were infected with an ampicillin-resistant pathogen. Because infants with a history of IPA exposure and suspicion of meningitis are at increased risk for infection with an ampicillin-resistant pathogen, clinicians should consider adding a third-generation cephalosporin to the standard ampicillin and gentamicin regimen as the initial choice.
We were somewhat surprised that 85% of the mothers in our study (54% of whom did not speak English) accurately recalled whether they had received IPAs. When an infant is being evaluated for suspected SBI, maternal recollection of IPAs can be a valuable historical tool and may help to guide initial antibiotic treatment, reducing the risk for treatment delay and failure.
This study has several limitations. A case-control design can indicate only associations rather than cause and effect. However, the strength of the associations, particularly for broad-spectrum antibiotic exposure and the biological plausibility for the development of ampicillin-resistant organisms' passing from the mother to her infant in the weeks after birth, suggest that the relationship may, in fact, be causal. Another limitation is that we were unable to contact and enroll many of the potential cases identified from 1999–2002 study. However, the types of infection, pathogens, and resistance patterns of the cases enrolled were not different from those that were not enrolled. There is no reason to suspect that they differed in a systematic way in IPA exposure from those whom we were able to contact and enroll. The individuals who reviewed the medical records were not blinded with respect to whether the record belonged to a case or control mother. This could introduce a bias with respect to recording of the data; however, the data obtained from the records were straightforward variables and were not subject to interpretation. One issue of more potential significance is that we were unable to determine whether mothers had received antibiotic treatment before the onset of labor or after discharge from the delivery hospitalization. Some infants in the non-IPA group may have been exposed to antibiotics at times other than intrapartum, which could bias the study toward the null. If mothers who had received IPAs were more or less likely to have received antibiotics at other times, then there could be an effect on the infant's flora. We could find no suggestion in the literature that this would be the case, but this or other unknown confounders that we did not take into account could affect the outcomes that we observed. We did stratify by hospital of delivery for the conditional logistic regression model. We did this to account for possible variation among the hospitals in the use of IPAs or other, unknown practices that might have an impact on the observed outcomes.
One potential problem with a case-control design, incomplete ascertainment of cases, was avoided by the unique environment of the Salt Lake City health care environment. As the only hospital in the area in which infants are hospitalized, all infants with an SBI would have been cared for at PCMC. It is possible that a few infants' families moved from the area and that the infant developed an SBI and was hospitalized out of the region, but, given the low incidence of SBI and the stability of the population, this is unlikely to have a significant impact on the results.
Taking into account the potential limitations of our study, we believe that our findings have uncovered a potentially important, unintended consequence of the widespread use of broad-spectrum IPAs to prevent early-onset GBS infection. Currently, the benefits of this practice clearly outweigh the potential negative consequences. However, other approaches, such as immunization, may be more beneficial and have fewer consequences. In the interim, physicians who are responsible for choosing an IPA agent should be encouraged to select penicillin whenever possible, and those who are responsible for evaluating infants for a possible SBI should determine whether the infant was exposed to IPAs.
CONCLUSION
Infants with late-onset SBI were more likely than healthy control subjects to have been exposed to IPAs. Those with late-onset SBI were more likely to be infected with ampicillin-resistant pathogens when they had been exposed to IPAs. Maternal recollection of IPA treatment was accurate in this patient population. The choice of initial antibiotic therapy for infants with suspected SBI should include consideration of a history of IPAs.
ACKNOWLEDGMENTS
Dr Glasgow and Dr Young are supported by a BRIC grant through the Agency for Healthcare Quality and Research (AHRQ) grant P20HS11826. The project was funded by an innovative research grant through the Primary Children's Medical Center Foundation. Dr Carrie Byington was supported by the Robert Wood Johnson Generalist Physician Faculty Scholar Program and by Public Health Service Research Grant M01-RR0064 from the National Center of Research Resources.
FOOTNOTES
Accepted Dec 30, 2004.
No conflict of interest declared.
Parts of this work were presented at the Pediatric Academic Societies Meetings, May 5, 2003, Seattle, Washington; and May 2, 2004, San Francisco, California.
REFERENCES
Schrag S, Gorwitz R, Fultz-Butts K, Schuchat A. Prevention of perinatal group B streptococcal disease. Revised guidelines from CDC. MMWR Recomm Rep.51 :1 –22, 2002
Revised guidelines for prevention of early-onset group B streptococcal (GBS) infection. American Academy of Pediatrics Committee on Infectious Diseases and Committee on Fetus and Newborn. Pediatrics. 1997;99 :489 –496
Schrag SJ, Zywicki S, Farley MM, et al. Group B streptococcal disease in the era of intrapartum antibiotic prophylaxis. N Engl J Med. 2000;342 :15 –20
Friedman S, Shah V, Ohlsson A, Matlow AG. Neonatal Escherichia coli infections: concerns regarding resistance to current therapy. Acta Paediatr. 2000;89 :686 –689
Hyde TB, Hilger TM, Reingold A, Farley MM, O'Brien KL, Schuchat A. Trends in incidence and antimicrobial resistance of early-onset sepsis: population-based surveillance in San Francisco and Atlanta. Pediatrics. 2002;110 :690 –695
Stoll BJ, Hansen N, Fanaroff AA, et al. Changes in pathogens causing early-onset sepsis in very-low-birth-weight infants. N Engl J Med. 2002;347 :240 –247
Shah SS, Ehrenkranz RA, Gallagher PG. Increasing incidence of gram-negative rod bacteremia in a newborn intensive care unit. Pediatr Infect Dis J. 1999;18 :591 –595
Towers CV, Carr MH, Padilla G, Asrat T. Potential consequences of widespread antepartal use of ampicillin. Am J Obstet Gynecol. 1998;179 :879 –883
Towers CV, Briggs GG. Antepartum use of antibiotics and early-onset neonatal sepsis: the next 4 years. Am J Obstet Gynecol. 2002;187 :495 –500
Mercer BM, Carr TL, Beazley DD, Crouse DT, Sibai BM. Antibiotic use in pregnancy and drug-resistant infant sepsis. Am J Obstet Gynecol. 1999;181 :816 –821
Terrone DA, Rinehart BK, Einstein MH, Britt LB, Martin JN Jr, Perry KG. Neonatal sepsis and death caused by resistant Escherichia coli: possible consequences of extended maternal ampicillin administration. Am J Obstet Gynecol. 1999;180 :1345 –1348
Moore MR, Schrag SJ, Schuchat A. Effects of intrapartum antimicrobial prophylaxis for prevention of group-B-streptococcal disease on the incidence and ecology of early-onset neonatal sepsis. Lancet Infect Dis. 2003;3 :201 –213
Edwards RK, Clark P, Sistrom CL, Duff P. Intrapartum antibiotic prophylaxis 1: relative effects of recommended antibiotics on gram-negative pathogens. Obstet Gynecol. 2002;100 :534 –539
Baraff LJ, Oslund SA, Schriger DL, Stephen ML. Probability of bacterial infections in febrile infants less than three months of age: a meta-analysis. Pediatr Infect Dis J. 1992;11 :257 –264
Harbarth S, Gundlapalli AV, Stockdale W, Samore MH. Shortage of penicillin G: impact on antibiotic prescribing at a US tertiary care centre. Int J Antimicrob Agents. 2003;21 :484 –487
Newman T. Designing an observational study: cross-sectional and case-control studies. In: Hulley SB, Cummings S, Browner WS, Grady D, Newman TB, eds. Designing Clinical Research. 2nd ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2001:107 –121
Byington CL, Rittichier KK, Bassett KE, et al. Serious bacterial infections in febrile infants younger than 90 days of age: the importance of ampicillin-resistant pathogens. Pediatrics. 2003;111 :964 –968
Hoberman A, Wald ER, Reynolds EA, Penchansky L, Charron M. Pyuria and bacteriuria in urine specimens obtained by catheter from young children with fever. J Pediatr. 1994;124 :513 –519
(NCCLS) NCoCLS. M7A45-Methods for Dilution and Antimicrobial Susceptibility Testing for Bacteria that Grow Aerobically. In; 2000
(NCCLS) NCoCLS. M2-A7-Performance for Antimicrobial Disk-Susceptibility Testing: Approved Standard. In; 2000
Mackie RI, Sghir A, Gaskins HR. Developmental microbial ecology of the neonatal gastrointestinal tract. Am J Clin Nutr. 1999;69 :1035S –1045S
Brook I, Barrett CT, Brinkman CR 3rd, Martin WJ, Finegold SM. Aerobic and anaerobic bacterial flora of the maternal cervix and newborn gastric fluid and conjunctiva: a prospective study. Pediatrics. 1979;63 :451 –455
Finegold SM. Changes in Human Intestinal Flora Related to the Administration of Antimicrobial Agents. 1st ed. Lubbock, TX: Academic Press; 1983
Spaetgens R, DeBella K, Ma D, Robertson S, Mucenski M, Davies HD. Perinatal antibiotic usage and changes in colonization and resistance rates of group B streptococcus and other pathogens. Obstet Gynecol. 2002;100 :525 –533
Gaya H, Adnitt PI, Turner P. Changes in gut flora after cephalexin treatment. Br Med J. 1970;3 :624 –625
Heimdahl A, Kager L, Malmborg AS, Nord CE. Impact of different betalactam antibiotics on the normal human flora, and colonization of the oral cavity, throat and colon. Infection. 1982;10 :120 –124
Heimdahl A, Nord CE. Effect of phenoxymethylpenicillin and clindamycin on the oral, throat and faecal microflora of man. Scand J Infect Dis. 1979;11 :233 –242
Brunel A, Gouet P. Influence of the destabilisation of the maternal digestive microflora on that of the newborn rat. Biol Neonate. 1993;63 :236 –245
Bruinsma N, Hutchinson JM, van den Bogaard AE, Giamarellou H, Degener J, Stobberingh EE. Influence of population density on antibiotic resistance. J Antimicrob Chemother. 2003;51 :385 –390
Shea KM. Antibiotic resistance: what is the impact of agricultural uses of antibiotics on children's health Pediatrics. 2003;112 :253 –258
Aiello AE, Larson E. Antibacterial cleaning and hygiene products as an emerging risk factor for antibiotic resistance in the community. Lancet Infect Dis. 2003;3 :501 –506
Holmes SJ, Morrow AL, Pickering LK. Child-care practices: effects of social change on the epidemiology of infectious diseases and antibiotic resistance. Epidemiol Rev. 1996;18 :10 –28
Spach DH, Black D. Antibiotic resistance in community-acquired respiratory tract infections: current issues. Ann Allergy Asthma Immunol. 1998;81 :293 –302; quiz 302–303
Samore MH, Magill MK, Alder SC, et al. High rates of multiple antibiotic resistance in Streptococcus pneumoniae from healthy children living in isolated rural communities: association with cephalosporin use and intrafamilial transmission. Pediatrics. 2001;108 :856 –865
Lodise TP Jr, McKinnon PS, Rybak M. Prediction model to identify patients with Staphylococcus aureus bacteremia at risk for methicillin resistance. Infect Control Hosp Epidemiol. 2003;24 :655 –661
Finkelstein JA, Huang SS, Daniel J, et al. Antibiotic-resistant Streptococcus pneumoniae in the heptavalent pneumococcal conjugate vaccine era: predictors of carriage in a multicommunity sample. Pediatrics. 2003;112 :862 –869(Tiffany S. Glasgow, MD, P)
ABSTRACT
Objective. Recommendations to prevent vertical transmission of group B Streptococcus (GBS) infections have resulted in many women's receiving antibiotics during labor with an associated reduction in early-onset GBS infections in their newborn infants. However, a potential relationship of intrapartum antibiotics (IPA) to the occurrence of late-onset (7–90 days) serious bacterial infections (SBIs) in term infants has not been reported. The objectives of this study were to determine whether infants with late-onset SBI were more likely than healthy control infants to have been exposed to IPA and whether there was a greater likelihood of antibiotic resistance in bacteria that were isolated from infants who had an SBI and had been exposed to IPA compared with those who had not.
Methods. We used a case-control design to study the first objective. Cases were previously healthy full-term infants who were hospitalized for late-onset SBI between the ages of 7 and 90 days. Control subjects were healthy full-term infants who were known not to have an SBI in their first 90 days. Cases and control subjects were matched for hospital of delivery. In the second part of the study, rates of antibiotic resistance of bacteria that were isolated from infected infants were compared for those who had and had not been exposed to IPA.
Results. Ninety case infants and 92 control subjects were studied. Considering all types of IPA, more case (41%) than control infants (27%) had been exposed to IPA (adjusted odds ratio [OR]: 1.96; 95% confidence interval [CI]: 1.05–3.66), after controlling for hospital of delivery. The association was stronger when IPA was with broad-spectrum antibiotics (adjusted OR: 4.95; 95% CI: 2.04–11.98), after controlling for hospital of delivery, penicillin IPA, maternal chorioamnionitis, and breastfeeding. Bacteria that were isolated from infected infants who had been exposed to IPA were more likely to exhibit ampicillin resistance (adjusted OR: 5.7; 95% CI: 2.3–14.3), after controlling for hospital of delivery, but not to other antibiotics that are commonly used to treat SBI in infants.
Conclusions. After adjusting for potential confounders, infants with late-onset SBI were more likely to have been exposed to IPA than noninfected control infants. Pathogens that cause late-onset SBI were more likely to be resistant to ampicillin when the infant had been exposed to intrapartum antibiotics.
Key Words: intrapartum antibiotics serious bacterial infection term infant antibiotic resistance group B Streptococcus
Abbreviations: GBS, group B Streptococcus SBI, serious bacterial infection IPA, intrapartum antibiotic PCMC, Primary Children's Medical Center OR, odds ratio UTI, urinary tract infection CI, confidence interval
National recommendations designed to prevent the vertical transmission of group B Streptococcus (GBS) have resulted in many women's receiving antibiotics during labor.1,2 Data obtained from the largest insurer in the state of Utah (Intermountain Health Care) indicate that, between 1998 and 2002, 35% of mothers who delivered term infants had received intrapartum antibiotics (B.M. James, MD, MStat, Intermountain Health Care Institute for Health Care Delivery Research, personal communication, 2003). Intrapartum antibiotic treatment has been very successful in preventing early-onset neonatal GBS infection.3 However, some investigators have reported an increase in early-onset neonatal infection in preterm infants caused by antibiotic-resistant Gram-negative organisms.4–7 It has been suggested that this increase may be attributable, in part, to maternal antibiotic exposure before delivery of the infant,8,9 and several studies have reported an association between maternal antibiotic therapy and early-onset infection in preterm infants with pathogens resistant to antibiotics.10–12 These studies all focused on early-onset infection, leaving the question of late effects of maternal antibiotic treatment unaddressed. Antibiotics are known to alter gastrointestinal and vaginal bacterial flora, selecting for resistant pathogens.13 Early-onset serious bacterial infections (SBIs) occur within the first week and are attributed to vertical transmission of pathogens from mother to infant during labor and delivery. Late-onset SBIs occur after the seventh day and have been considered to be community-acquired infections. The incidence of late-onset SBIs in former term infants is 1.6 per 1000 births.14
In the process of caring for infants with late-onset SBI, the investigators observed what seemed to be an increase in the proportion of these infections that were caused by bacteria resistant to antibiotics and wondered whether this could be attributable, in part, to the large-scale delivery of broad-spectrum antibiotics during labor. Although the Centers for Disease Control and Prevention recommends penicillin for GBS prophylaxis, the majority of women in Utah receive the broader spectrum ampicillin.15 To explore this potential relationship, we developed 2 research questions: (1) Could exposure to intrapartum antibiotics increase the likelihood of an infant's acquiring a late-onset SBI (2) Are the organisms that cause SBIs more likely to demonstrate antibiotic resistance when the infant has been exposed to intrapartum antibiotics Although a prospective study that follows 2 cohorts of infants (those who had and who had not been exposed to intrapartum antibiotics [IPAs]) and notes their respective rates infection might be an appropriate design to answer these questions, the relative rarity of late-onset SBI (1.6 per 1000 births14) makes this approach impractical. We therefore used a case-control design to study these questions.16 We compared the proportion of a cohort of infants with SBI (cases) who had been exposed to IPAs with the proportion of a cohort of healthy infants (control subjects) who had been born during the same time period and at the same hospitals. We hypothesized that the proportion of infants who had late-onset SBI and had a history of IPA exposure would be higher than control subjects. In the second part of the study, we focused on the infected infants and examined the bacterial resistance patterns of the organisms that were isolated from them. We hypothesized that the proportion of bacteria that demonstrated antibiotic resistance would be higher in the infants who were exposed to IPAs.
METHODS
Approval to conduct this research was obtained by the Institutional Review Boards of the University of Utah, Intermountain Health Care, and St Marks Hospital (Salt Lake City, UT).
Study Patients
Cases were previously healthy former term (37 weeks' gestation) infants who were between the ages of 7 and 90 days and were hospitalized at Primary Children's Medical Center (PCMC) with am SBI between July 1999 and May 2003. PCMC is a 252-bed children's hospital that serves as the only pediatric hospital for children from the Salt Lake City metropolitan area. None of the other hospitals in the area provides nonnursery, inpatient care for infants. Infants with SBIs (cases) were identified using 2 methods. Between July 1, 2002, and May 31, 2003, cases were identified by daily examination of the PCMC microbiology culture logs (see below). Additional cases were identified by reviewing the database of a previous 6-year study of late-onset infections in infants between 7 and 90 days of age conducted by 1 of the authors (C.L.B.).17 To facilitate contact with the parents of this group of infants, we limited our study to those who had been enrolled in that study between 1999 and 2002. Control infants were recruited by cooperating primary care physicians in the Salt Lake City area. Parents of former term infants, who were being seen for their 4-month well-child visit and who had not had an SBI, were invited to participate. Parents of case and control infants who agreed provided informed consent for the investigators to review maternal and infant medical records.
Review of Microbiologic Records
The methods and the criteria used for identifying infants with a late-onset SBI have been described previously17 and were identical to the methods used to identify cases seen after July 1, 2001. The computerized microbiologic records maintained by PCMC were reviewed daily to identify infants with a possible SBI. SBI was defined as the presence of pathogenic bacteria cultured from blood, urine, or cerebrospinal fluid. Urine cultures were considered positive when >50000 colony-forming units/mL of a single organism were isolated from a catheterized specimen.18 For bacteria that could be considered contaminants, such as Viridans group streptococci, SBI was diagnosed only when the infant had 2 separate positive cultures. We considered omphalitis to be an SBI when there was evidence of a clinical abdominal wall infection and when a single pathogenic organism was isolated from an umbilical swab. The antibiotic susceptibility for each pathogen was determined following recommendations of the National Committee of Clinical Laboratory Standards.19,20
Maternal Interview
Once cases of SBI were obtained from the PCMC records and after consent was obtained, the infant's mother was interviewed using a structured format. Data regarding the mother's age, the infant's race, hospital of delivery, the mother's recollection of receiving IPA treatment, antibiotic treatment of the infant, and breastfeeding were obtained. Interviews were conducted in person for the 2002–2003 cases and by telephone for the cases from 1999 to 2002. The same information was obtained from mothers of control infants by a questionnaire that was given to them at the well-child visit. In-person and telephone interviews, questionnaires, and consent forms were administered in Spanish when appropriate.
Delivery Hospital Medical Record Review
Two reviewers (T.G. and C.K.) reviewed the peripartum medical records of all mothers and infants using a structured abstraction format. Maternal records were reviewed for documentation of IPA administration, the type of antibiotic administered, the number of doses, reason for antibiotic treatment (eg, chorioamnionitis), GBS culture results, maternal fever during labor, prolonged rupture of membranes, method of delivery, receipt of prenatal care, mother's race/ethnicity, and type of health insurance. We considered IPAs to have been given when the record indicated that an antibiotic agent had been administered to the mother during labor at any time before the delivery of her infant. The category broad-spectrum IPAs was applied whenever an antibiotic other than penicillin had been given. Infant records were reviewed for gender, presence of breastfeeding, and a history of antibiotic treatment after delivery.
Sample Size Estimates and Data Analysis
On the basis of the literature and local data, we estimated that 30% of women would have a history of IPA exposure. On the basis of this estimate, a sample size of 93 in each cohort would be required to have 80% power to detect an absolute 20% difference in either direction between IPA rates for cases and control subjects. Our achieved sample size of 90 cases and 92 control subjects provided 80% power to detect a 20.1% difference. Data were analyzed (G.S. and S.F.) using STATA statistical software (Stata Corp, College Station, TX). To test the first hypothesis, we compared the proportion of infants who were exposed to an IPA for cases and control subjects using conditional logistic regression stratified by hospital of delivery. One delivery hospital was overrepresented in the cohort of control subjects. The slight imbalance in numbers of cases and control subjects, 90 and 92, did not create a problem for analysis, because the conditional logistic regression model compared the cases as a group with the control subjects as a group within the same hospital stratum, rather than pairwise case-control comparisons. Odds ratios (ORs) for IPA exposure were calculated for the 2 cohorts. The other study variables including race, type of insurance, prenatal care, type of delivery, GBS colonization, maternal fever, chorioamnionitis, prolonged rupture of membranes, infant's gender, and breastfeeding then were entered individually into the model. All reported ORs are adjusted for hospital of delivery, as all ORs come from a conditional logistic regression that stratifies on this variable. Any variables that altered the hospital of delivery adjusted OR by >10% were placed into the full conditional logistic regression model.
Next, we performed a separate analysis of the cohort of infected infants. Associations between receipt of IPAs, type of antibiotic, number of doses (1, 2, or >2), and antibiotic-resistant pathogens were identified by 2 or Fisher exact test as appropriate. Other study variables, including race, type of insurance, prenatal care, type of delivery, GBS colonization, maternal fever, chorioamnionitis, hospital of delivery, gender, and breastfeeding, were analyzed individually to determine whether they were associated with the presence of antibiotic resistance. Variables that altered the OR by 10% or more were placed into a multiple logistic regression model. ORs for infection with a resistant pathogen for infants who were exposed to IPAs compared with those who were not exposed were generated. Finally, we performed a similar analysis comparing each individual type of IPA with no IPAs.
RESULTS
Patient Enrollment
A total of 90 infants with SBI and their mothers were enrolled. Forty were identified from the database of the previous study17 that contained 101 infants who were between the ages of 7 and 90 days, had an SBI, and had been admitted to PCMC between July 1999 and June 2002. Sixty of the mothers of these infants could not be contacted, and 1 who was contacted refused participation. There were no significant differences between these 61 infants and the 40 enrolled infants with regard to gender, type of SBI, type of pathogen, or antibiotic resistance. Fifty infants with SBI were identified from the PCMC microbiology logs between July 2002 and May 2003; the parents of all of these infants consented for the study. Community pediatricians identified a total of 187 potential control infants, and the parents of all 187 verbally agreed to participate. When control subjects were matched to cases at the hospital level, which involved randomly selecting from the available control subjects for 1 hospital that was overrepresented, we ended up with 90 cases and 92 control subjects who completed the formal consenting process and became enrolled in the study.
Infant deliveries were distributed among 7 Salt Lake City area hospitals. The demographics of the mothers and infants are listed in Table 1. There were no differences between cases and control subjects with respect to the demographic variables. Notably, all but 1 of the mothers had received prenatal care. The types of infection are listed in Table 2. The majority (72%) of infections were uncomplicated urinary tract infections (UTIs). However, 28% of infants had potentially life-threatening infections, including bacteremia and meningitis. The pathogens isolated are shown in Table 3.
IPA Exposure in Infants With SBI Compared With Control Subjects
As hypothesized, more of the cases (41%) than control subjects (27%) had been exposed to IPAs (OR: 1.96; 95% confidence interval [CI]: 1.05–3.66). There were no significant differences between cases and control subjects with respect to maternal fever (OR: 0.50; 95% CI: 0.13–1.9), prolonged rupture of membranes (OR: 2.16; 95% CI: 0.41–11.27), GBS colonization (OR: 0.90; 95% CI: 0.36–2.25), cesarean delivery (OR: 0.76; 95% CI: 0.34–1.67), gender of infant (OR: 0.99; 95% CI: 0.55–1.78), and private insurance compared with Medicaid (OR: 1.12; 95% CI: 0.43–2.97). Table 4 shows the final regression model comparing case and control infants and includes the variables (type of IPA [broad-spectrum or penicillin], breastfeeding, and maternal chorioamnionitis) that altered the hospital of delivery adjusted IPA OR by >10%. In the full model, only the use of a broad-spectrum antibiotic remained significant with an OR of 4.95 (95% CI: 2.04–11.98).
Antibiotic Resistance and IPAs
In the cohort of 90 infants with SBI, 41% of the responsible isolates were resistant to ampicillin. None was resistant to gentamicin or to third-generation cephalosporins. Resistance to ampicillin was strongly associated with IPA exposure. As shown in Table 5, 24 (65%) of pathogens that were isolated from the 37 infants who were exposed to IPAs were resistant to ampicillin compared with 13 (25%) of the pathogens that were isolated from the 43 infants who were not exposed to IPAs (OR: 5.7; 95% CI: 2.3–14.3). This relationship seems to be accounted for by the use of broad-spectrum antibiotics, because, as shown, rates of bacterial resistance for infants whose mothers had received penicillin were the same as that for infants whose mothers had not received IPAs. When either ampicillin or another broad-spectrum antibiotic had been used, the associations with ampicillin resistance were significant (OR: 6.1; 95% CI: 1.9–19.7) for ampicillin, and for other broad-spectrum antibiotics, the OR was 12.3 (95% CI: 2.3–65.5). Proximity to the IPA exposure did not seem to account for resistance because it was as likely to be present in infants who developed an SBI before or after their first month. There were no differences in resistance rates related to the number of doses that the infant's mother had received (data not shown).
Infants with UTIs and other types of SBI were analyzed separately. UTIs in infants who were exposed to IPAs were more likely to be caused by an ampicillin-resistant pathogen compared with those who were not exposed (OR: 4.3; 95% CI: 1.6–11.7). Even in the relatively small number of infants with other SBIs (meningitis, omphalitis, and bacteremia without UTI), the relationship between infection with an ampicillin-resistant organism and IPAs achieved statistical significance, although with wide CIs (OR: 25; 95% CI: 1.8–346; Fisher exact P = .013). Gender, type of insurance, race, prenatal care, cesarean delivery, maternal fever, prolonged rupture of membranes, chorioamnionitis, GBS colonization, treatment of the infant with antibiotics, hospital of delivery, and breastfeeding had no significant effect on rates of infection with an antibiotic-resistant pathogen in this group of infants.
Eleven infants were infected with constitutively ampicillin-resistant organisms: Klebsiella species, Enterobacter species, and Staphylococcus aureus. Nine had been exposed to IPAs with broad-spectrum agents. Our review of the mother's medical record indicated that, in all, the reason for the IPAs had been GBS prophylaxis; none had a diagnosis of chorioamnionitis, pyelonephritis, or cellulitis.
We compared the mothers' recollection during the interview of whether they had received IPAs with data in the medical record. Of the 277 women in the total sample, the recollection of 235 (85%) was accurate. The 15% of women whose recall differed from the data in the medical record were as likely to over- as underreport IPA treatment.
DISCUSSION
Having noted that 35% of mothers were receiving antibiotics during labor, we wondered whether this practice could be resulting in a change in either the frequency of late-onset SBIs or the antimicrobial resistance patterns of the bacteria that cause these infections in former term infants. The findings of this case-control study support our first hypothesis that more infants with late-onset SBI would have a history of IPA exposure than control infants. The second hypothesis, that, in infected infants rates of antibiotic resistance would be higher in those who were exposed to IPAs, was only partially supported. We found that the responsible organisms in these infants were more likely to be resistant to ampicillin but not to other antibiotics, such as gentamicin or third-generation cephalosporins. Both of these findings seem to be related to the use of broad-spectrum antibiotics during labor rather than penicillin, which is the recommended agent. For example, when we compared the rates and types of IPA exposure in infants with and without SBI, we found that the rates of exposure to penicillin were the same for the infants with SBI and control subjects but that the rates of exposure to broad-spectrum IPAs were much higher for cases compared with control subjects (data not shown). This increase seems to be attributable largely to the impact of ampicillin-resistant organisms. In the infants with an SBI, the rates of infection with an ampicillin-resistant pathogen were higher in the group that was exposed to IPAs than in nonexposed infants, but the rates of infections with ampicillin-susceptible pathogens were the same. We believe that these findings have important implications for both the selection of an appropriate choice of antibiotics for GBS prophylaxis and the management of late-onset SBI in infants. First, penicillin should be the preferred agent for GBS prophylaxis; second, a history of exposure to IPAs should be obtained whenever a clinician is evaluating an infant between 7 and 90 days for the possibility of an SBI. Maternal recall of either receiving or not receiving IPAs is likely to be accurate: 85% in this heterogeneous population.
This is the first study to evaluate the association between IPAs and late-onset SBI in term infants, but our findings are consistent with previous studies that have demonstrated an increase in antibiotic-resistant pathogens and an association of IPAs with infection with antibiotic-resistant pathogens in preterm infants with early-onset sepsis (first 7 days of life).6,8,10
Normal neonates become colonized with bacterial flora that mirror maternal organisms.21,22 Antibiotics have been shown to decrease normal gastrointestinal flora such as Lactobacillus species and Bifidobacteria species and select for antibiotic-resistant pathogens such as Klebsiella species, Enterobacter species, and S aureus.13,23–27 In our study population, almost all of the infants who were infected with these pathogens were exposed to broad-spectrum IPAs. We believe that these organisms are transferred to the infant both vertically at delivery and, more important, horizontally during the first weeks after birth.21,28 Thus, broad-spectrum IPAs may alter both maternal and infant bacterial flora and select for antibiotic-resistant pathogens. We suspect that these antibiotic-resistant pathogens are more invasive and that, as a result, infants who are colonized with these organisms are more likely to develop an SBI.
In our study, IPAs were associated with an increased risk for infections with ampicillin-resistant bacteria but not with infections caused by ampicillin-susceptible organisms. The acquisition of antibiotic-resistant infections in community settings is complex. Factors such as population density and antibiotics in the environment (eg, animal feed, cleaning products, child care) contribute to community resistance patterns.29–33 Recent exposure to antibiotics (30–120 days before infection) has consistently been identified as a risk factor for acquisition of resistant pathogens.34–36 In neonates, antibiotic exposure can disturb normal initial intestinal colonization. This effect may be persistent. In our study population, infants who were between 60 and 90 days of age and exposed to IPAs were just as likely to have an ampicillin-resistant pathogen as the infants who were younger than 30 days. Although many factors may be involved in the acquisition of antibiotic-resistant pathogens in neonates, the possible association of IPAs cannot be ignored.
The majority of the mothers who received IPAs in this study were treated with broad-spectrum antibiotics such as ampicillin rather than penicillin. Because of its narrower spectrum, penicillin was recommended for GBS prophylaxis in both the 1997 and 2002 Centers for Disease Control and Prevention guidelines.1,2 Unlike broad-spectrum antibiotics, penicillin does not perturb normal gastrointestinal flora.26,27 Infected infants whose mothers had received intrapartum penicillin were no more likely to be infected with a resistant organism than infected infants whose mothers did not receive an IPA. In a separate analysis from the case-control portion of the study, we compared the rates of intrapartum penicillin treatment between cases and control subjects and found that they were the same. This finding should be interpreted cautiously, however, because the numbers of mothers who received penicillin was small, and there may be insufficient power to detect a small but possibly significant difference. However, we believe that our findings do provide support for choosing penicillin rather than ampicillin for GBS prophylaxis.
Clinicians who are faced with an infant in whom an SBI may be present are required to choose an antibiotic regimen before culture and sensitivity results become available. Our findings suggest that a history of IPA exposure should influence this choice. In a previous study from our institution, initial antibiotic therapy was changed in more than half of the cases of late-onset SBI after identification of the pathogenic organism.17 In addition, 64% of patients with bacteremia or meningitis caused by ampicillin-resistant organisms had documented treatment failure of their initial antibiotic regimen.17 In the current study, 62% of the 20 infants with bacteremia were infected with an ampicillin-resistant pathogen. Because infants with a history of IPA exposure and suspicion of meningitis are at increased risk for infection with an ampicillin-resistant pathogen, clinicians should consider adding a third-generation cephalosporin to the standard ampicillin and gentamicin regimen as the initial choice.
We were somewhat surprised that 85% of the mothers in our study (54% of whom did not speak English) accurately recalled whether they had received IPAs. When an infant is being evaluated for suspected SBI, maternal recollection of IPAs can be a valuable historical tool and may help to guide initial antibiotic treatment, reducing the risk for treatment delay and failure.
This study has several limitations. A case-control design can indicate only associations rather than cause and effect. However, the strength of the associations, particularly for broad-spectrum antibiotic exposure and the biological plausibility for the development of ampicillin-resistant organisms' passing from the mother to her infant in the weeks after birth, suggest that the relationship may, in fact, be causal. Another limitation is that we were unable to contact and enroll many of the potential cases identified from 1999–2002 study. However, the types of infection, pathogens, and resistance patterns of the cases enrolled were not different from those that were not enrolled. There is no reason to suspect that they differed in a systematic way in IPA exposure from those whom we were able to contact and enroll. The individuals who reviewed the medical records were not blinded with respect to whether the record belonged to a case or control mother. This could introduce a bias with respect to recording of the data; however, the data obtained from the records were straightforward variables and were not subject to interpretation. One issue of more potential significance is that we were unable to determine whether mothers had received antibiotic treatment before the onset of labor or after discharge from the delivery hospitalization. Some infants in the non-IPA group may have been exposed to antibiotics at times other than intrapartum, which could bias the study toward the null. If mothers who had received IPAs were more or less likely to have received antibiotics at other times, then there could be an effect on the infant's flora. We could find no suggestion in the literature that this would be the case, but this or other unknown confounders that we did not take into account could affect the outcomes that we observed. We did stratify by hospital of delivery for the conditional logistic regression model. We did this to account for possible variation among the hospitals in the use of IPAs or other, unknown practices that might have an impact on the observed outcomes.
One potential problem with a case-control design, incomplete ascertainment of cases, was avoided by the unique environment of the Salt Lake City health care environment. As the only hospital in the area in which infants are hospitalized, all infants with an SBI would have been cared for at PCMC. It is possible that a few infants' families moved from the area and that the infant developed an SBI and was hospitalized out of the region, but, given the low incidence of SBI and the stability of the population, this is unlikely to have a significant impact on the results.
Taking into account the potential limitations of our study, we believe that our findings have uncovered a potentially important, unintended consequence of the widespread use of broad-spectrum IPAs to prevent early-onset GBS infection. Currently, the benefits of this practice clearly outweigh the potential negative consequences. However, other approaches, such as immunization, may be more beneficial and have fewer consequences. In the interim, physicians who are responsible for choosing an IPA agent should be encouraged to select penicillin whenever possible, and those who are responsible for evaluating infants for a possible SBI should determine whether the infant was exposed to IPAs.
CONCLUSION
Infants with late-onset SBI were more likely than healthy control subjects to have been exposed to IPAs. Those with late-onset SBI were more likely to be infected with ampicillin-resistant pathogens when they had been exposed to IPAs. Maternal recollection of IPA treatment was accurate in this patient population. The choice of initial antibiotic therapy for infants with suspected SBI should include consideration of a history of IPAs.
ACKNOWLEDGMENTS
Dr Glasgow and Dr Young are supported by a BRIC grant through the Agency for Healthcare Quality and Research (AHRQ) grant P20HS11826. The project was funded by an innovative research grant through the Primary Children's Medical Center Foundation. Dr Carrie Byington was supported by the Robert Wood Johnson Generalist Physician Faculty Scholar Program and by Public Health Service Research Grant M01-RR0064 from the National Center of Research Resources.
FOOTNOTES
Accepted Dec 30, 2004.
No conflict of interest declared.
Parts of this work were presented at the Pediatric Academic Societies Meetings, May 5, 2003, Seattle, Washington; and May 2, 2004, San Francisco, California.
REFERENCES
Schrag S, Gorwitz R, Fultz-Butts K, Schuchat A. Prevention of perinatal group B streptococcal disease. Revised guidelines from CDC. MMWR Recomm Rep.51 :1 –22, 2002
Revised guidelines for prevention of early-onset group B streptococcal (GBS) infection. American Academy of Pediatrics Committee on Infectious Diseases and Committee on Fetus and Newborn. Pediatrics. 1997;99 :489 –496
Schrag SJ, Zywicki S, Farley MM, et al. Group B streptococcal disease in the era of intrapartum antibiotic prophylaxis. N Engl J Med. 2000;342 :15 –20
Friedman S, Shah V, Ohlsson A, Matlow AG. Neonatal Escherichia coli infections: concerns regarding resistance to current therapy. Acta Paediatr. 2000;89 :686 –689
Hyde TB, Hilger TM, Reingold A, Farley MM, O'Brien KL, Schuchat A. Trends in incidence and antimicrobial resistance of early-onset sepsis: population-based surveillance in San Francisco and Atlanta. Pediatrics. 2002;110 :690 –695
Stoll BJ, Hansen N, Fanaroff AA, et al. Changes in pathogens causing early-onset sepsis in very-low-birth-weight infants. N Engl J Med. 2002;347 :240 –247
Shah SS, Ehrenkranz RA, Gallagher PG. Increasing incidence of gram-negative rod bacteremia in a newborn intensive care unit. Pediatr Infect Dis J. 1999;18 :591 –595
Towers CV, Carr MH, Padilla G, Asrat T. Potential consequences of widespread antepartal use of ampicillin. Am J Obstet Gynecol. 1998;179 :879 –883
Towers CV, Briggs GG. Antepartum use of antibiotics and early-onset neonatal sepsis: the next 4 years. Am J Obstet Gynecol. 2002;187 :495 –500
Mercer BM, Carr TL, Beazley DD, Crouse DT, Sibai BM. Antibiotic use in pregnancy and drug-resistant infant sepsis. Am J Obstet Gynecol. 1999;181 :816 –821
Terrone DA, Rinehart BK, Einstein MH, Britt LB, Martin JN Jr, Perry KG. Neonatal sepsis and death caused by resistant Escherichia coli: possible consequences of extended maternal ampicillin administration. Am J Obstet Gynecol. 1999;180 :1345 –1348
Moore MR, Schrag SJ, Schuchat A. Effects of intrapartum antimicrobial prophylaxis for prevention of group-B-streptococcal disease on the incidence and ecology of early-onset neonatal sepsis. Lancet Infect Dis. 2003;3 :201 –213
Edwards RK, Clark P, Sistrom CL, Duff P. Intrapartum antibiotic prophylaxis 1: relative effects of recommended antibiotics on gram-negative pathogens. Obstet Gynecol. 2002;100 :534 –539
Baraff LJ, Oslund SA, Schriger DL, Stephen ML. Probability of bacterial infections in febrile infants less than three months of age: a meta-analysis. Pediatr Infect Dis J. 1992;11 :257 –264
Harbarth S, Gundlapalli AV, Stockdale W, Samore MH. Shortage of penicillin G: impact on antibiotic prescribing at a US tertiary care centre. Int J Antimicrob Agents. 2003;21 :484 –487
Newman T. Designing an observational study: cross-sectional and case-control studies. In: Hulley SB, Cummings S, Browner WS, Grady D, Newman TB, eds. Designing Clinical Research. 2nd ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2001:107 –121
Byington CL, Rittichier KK, Bassett KE, et al. Serious bacterial infections in febrile infants younger than 90 days of age: the importance of ampicillin-resistant pathogens. Pediatrics. 2003;111 :964 –968
Hoberman A, Wald ER, Reynolds EA, Penchansky L, Charron M. Pyuria and bacteriuria in urine specimens obtained by catheter from young children with fever. J Pediatr. 1994;124 :513 –519
(NCCLS) NCoCLS. M7A45-Methods for Dilution and Antimicrobial Susceptibility Testing for Bacteria that Grow Aerobically. In; 2000
(NCCLS) NCoCLS. M2-A7-Performance for Antimicrobial Disk-Susceptibility Testing: Approved Standard. In; 2000
Mackie RI, Sghir A, Gaskins HR. Developmental microbial ecology of the neonatal gastrointestinal tract. Am J Clin Nutr. 1999;69 :1035S –1045S
Brook I, Barrett CT, Brinkman CR 3rd, Martin WJ, Finegold SM. Aerobic and anaerobic bacterial flora of the maternal cervix and newborn gastric fluid and conjunctiva: a prospective study. Pediatrics. 1979;63 :451 –455
Finegold SM. Changes in Human Intestinal Flora Related to the Administration of Antimicrobial Agents. 1st ed. Lubbock, TX: Academic Press; 1983
Spaetgens R, DeBella K, Ma D, Robertson S, Mucenski M, Davies HD. Perinatal antibiotic usage and changes in colonization and resistance rates of group B streptococcus and other pathogens. Obstet Gynecol. 2002;100 :525 –533
Gaya H, Adnitt PI, Turner P. Changes in gut flora after cephalexin treatment. Br Med J. 1970;3 :624 –625
Heimdahl A, Kager L, Malmborg AS, Nord CE. Impact of different betalactam antibiotics on the normal human flora, and colonization of the oral cavity, throat and colon. Infection. 1982;10 :120 –124
Heimdahl A, Nord CE. Effect of phenoxymethylpenicillin and clindamycin on the oral, throat and faecal microflora of man. Scand J Infect Dis. 1979;11 :233 –242
Brunel A, Gouet P. Influence of the destabilisation of the maternal digestive microflora on that of the newborn rat. Biol Neonate. 1993;63 :236 –245
Bruinsma N, Hutchinson JM, van den Bogaard AE, Giamarellou H, Degener J, Stobberingh EE. Influence of population density on antibiotic resistance. J Antimicrob Chemother. 2003;51 :385 –390
Shea KM. Antibiotic resistance: what is the impact of agricultural uses of antibiotics on children's health Pediatrics. 2003;112 :253 –258
Aiello AE, Larson E. Antibacterial cleaning and hygiene products as an emerging risk factor for antibiotic resistance in the community. Lancet Infect Dis. 2003;3 :501 –506
Holmes SJ, Morrow AL, Pickering LK. Child-care practices: effects of social change on the epidemiology of infectious diseases and antibiotic resistance. Epidemiol Rev. 1996;18 :10 –28
Spach DH, Black D. Antibiotic resistance in community-acquired respiratory tract infections: current issues. Ann Allergy Asthma Immunol. 1998;81 :293 –302; quiz 302–303
Samore MH, Magill MK, Alder SC, et al. High rates of multiple antibiotic resistance in Streptococcus pneumoniae from healthy children living in isolated rural communities: association with cephalosporin use and intrafamilial transmission. Pediatrics. 2001;108 :856 –865
Lodise TP Jr, McKinnon PS, Rybak M. Prediction model to identify patients with Staphylococcus aureus bacteremia at risk for methicillin resistance. Infect Control Hosp Epidemiol. 2003;24 :655 –661
Finkelstein JA, Huang SS, Daniel J, et al. Antibiotic-resistant Streptococcus pneumoniae in the heptavalent pneumococcal conjugate vaccine era: predictors of carriage in a multicommunity sample. Pediatrics. 2003;112 :862 –869(Tiffany S. Glasgow, MD, P)