Bacteremia Treatment & Management

Most infants and young children who are evaluated for occult bacteremia present with a fever. The use of antipyretics to treat fever is somewhat controversial. However, while the child is evaluated to determine a source of the fever, fever reduction with medication is reasonable and is widely accepted. Studies have shown that ibuprofen (10 mg/kg/dose every 8 h) or acetaminophen (10-15 mg/kg/dose every 4-6 h) are both effective and well tolerated. ^[73]

Low-risk criteria: Who should be treated?

As recently as 1984, guidelines for treating febrile young infants recommended evaluation, treatment, and hospitalization because of the increased risk of bacterial infection and the inability to clinically distinguish infants at an increased risk for serious bacterial infection. ^[74] Since then, numerous studies have evaluated combinations of age, temperature, history, examination findings, and laboratory results to determine which young infants are at a low risk for bacterial infection. ^{[10, 67, 75, 76, 77]}

The following are the low-risk criteria established by groups from Philadelphia, Boston, and Rochester and the 1993 American Academy of Pediatrics (AAP) guideline.

Table 11. Low-Risk Criteria for Infants Younger than 3 Months ^{[10, 75, 76, 77]} (Open Table in a new window)

How well do low-risk criteria work?

The above guidelines are presented to define a group of febrile young infants who can be treated without antibiotics. Statistically, this translates into a high NPV (ie, a very high proportion of true negative cultures is observed in patients deemed to be at low risk). The NPV of various low-risk criteria for serious bacterial infection and occult bacteremia are as follows ^{[10, 14, 16, 17, 75, 76, 77]} :

• Philadelphia NPV - 95-100%

• Boston NPV - 95-98%

• Rochester NPV - 98.3-99%

• AAP 1993 - 99-99.8%

In basic terms, even by the most stringent criteria, somewhere between 1 in 100 and 1 in 500 low-risk, but bacteremic, febrile infants are missed. Many centers still choose to admit and treat young febrile infants.

A helpful clinical finding is that of a diagnostic viral syndrome, in particular respiratory syncytial virus bronchiolitis. In this setting, the likelihood of a concomitant bacterial is lower in nearly all instances, with the exception of a concurrent UTI. ^[49]

See the image below for a treatment approach in febrile infants younger than 3 months.

Empiric antibiotics: How well do they work?

The first step in the treatment of children with FWS is to use a combination of age, temperature, and screening laboratory test results to determine the risk for serious bacterial infection or occult bacteremia. Low-risk children are generally monitored as outpatients. Children who do not fit low-risk criteria are treated with empiric antibiotics either as inpatients or as outpatients.

Numerous studies have compared the effectiveness of oral antibiotics and parenteral antibiotics in reducing complications of occult bacteremia. Many of these studies were conducted before widespread use of the conjugate Hib vaccine. ^[1] Parenteral antibiotics were generally found to be significantly more effective than oral treatment or no treatment in reducing the sequelae of occult bacteremia, most importantly meningitis. ^{[10, 39]}

Table 12. Occult Bacteremia - Relationship Between Outpatient Antibiotic Use and Complications ^{[10, 11, 12, 22, 78]} (Open Table in a new window)

Recent studies and analyses have focused on specific causes of occult bacteremia other than Hib, information more applicable to current evaluation, and treatment of febrile children.

Several studies and analyses have concluded that oral antibiotics and parenteral antibiotics are equally effective in reducing complications of pneumococcal bacteremia. ^{[1, 10]} However, a metaanalysis found no statistical change in occurrence of meningitis between patients with and without treatment with oral antibiotics. ^[79]

Table 13. Pneumococcal Bacteremia - Relationship Between Outpatient Antibiotic Use and Complications ^{[1, 2, 10, 11, 34, 35, 42, 53, 79]} (Open Table in a new window)

Meningococcal bacteremia is rare but important because of its high rates of morbidity and mortality. Studies have found that parenteral antibiotics are significantly more effective than no treatment or oral antibiotics in reducing complications. The risk of developing meningitis with no antibiotic therapy is 50%, the risk is 29% with oral antibiotic therapy, and it is 0% with intramuscular and/or intravenous antibiotic therapy. ^[2]

In young infants and debilitated or immunocompromised patients, Salmonella bacteremia can have serious complications. The risk of serious complications in previously healthy children aged 3-36 months with Salmonella bacteremia is small. ^{[1, 6]} Empiric oral antibiotics have not been proven to prevent focal complications or persistence of bacteremia in children with occult nontyphoidal Salmonella bacteremia. ^[1] However, some form of antibiotic treatment, oral or intravenous, is recommended for all children with Salmonella bacteremia and for young infants and immunocompromised children with Salmonella gastroenteritis. ^[80]

The choice of empiric antibiotic treatment is primarily based on the likely causes of bacteremia for a given patient and the likelihood of resistance.

In very young infants, bacterial causes are most commonly acquired from the mother during childbirth. For neonates younger than 1 month, Streptococcus species and E coli are the most common pathogens. Other gram-positive and gram-negative infections are also observed; including infections with Listeria species (see Causes). Treatment with ampicillin and gentamicin is widely accepted for patients in this age group; ampicillin and cefotaxime may also be used. ^{[14, 34]} This combination has good gram-positive and gram-negative coverage for the most likely pathogens, and ampicillin is effective against Listeria.

In the recent Kaiser Permanente study, fully 36% of their pathogens were ampicillin resistant, and, if so, were usually gentamicin resistant as well. They recommended ampicillin/cefotaxime as the combination of choice, and with a very low rate of Listeria and Enterococcus (the only 2 organisms for which ampicillin would be the preferred drug in that setting) called into question whether cefotaxime alone would be sufficient. They did not go so far as to recommend this, but called for further research in other centers. ^[31]

Third-generation cephalosporins are useful in older infants and children, but they are not active against Listeria and are not recommended as a single-agent therapy in the empiric treatment of neonates younger than 1 month who are at risk for occult bacteremia. ^[16]

A gradual shift toward community-acquired causes occurs as age increases; the causes of bacteremia in infants aged 1-3 months are a combination of organisms (see Etiology). Empiric antibiotics used in practice vary in this age group. Some practitioners use ampicillin and gentamicin, some use ampicillin and cefotaxime, and others use ceftriaxone. ^{[10, 14, 34]} The risk for infection with Listeria is significantly decreased in children older than 4-6 weeks; whether coverage for Listeria is required in infants aged 1-3 months at risk for occult bacteremia is controversial. All these possible antibiotic regimens have excellent coverage against the other childbirth-acquired or community-acquired bacterial pathogens in this age group.

The empiric treatment of infants and children aged 3-36 months at risk for occult bacteremia usually involves ceftriaxone. This third-generation cephalosporin has broad-spectrum gram-positive and gram-negative coverage, is active against all likely community-acquired pathogens in this age group, and is resistant to beta-lactamases produced by some pathogenic organisms. ^{[12, 16]} Ceftriaxone has the longest half-life of the third-generation cephalosporins, and high serum concentrations can be sustained for 24 hours with a single dose. Most body tissues and fluids are penetrated, including the CSF. ^[12]

Early studies of empiric coverage with oral antibiotics examined various agents, including amoxicillin and penicillin. Because of concern for infection with Hib positive for beta-lactamase, later studies focused on amoxicillin and clavulanic acid.

Other than antibiotic spectrum coverage, adverse effects and compliance are also considered when choosing an antibiotic treatment. Studies evaluating adverse effects of ceftriaxone and amoxicillin and clavulanic acid have shown that, whereas amoxicillin and clavulanic acid more commonly cause diarrhea, the overall rate of adverse effects (eg, diarrhea, vomiting, maculopapular exanthems) is similar at approximately 5%. ^{[12, 39]} Regarding compliance, the administration of antibiotic treatment is essentially witnessed when the antibiotic is intramuscularly administered. However, in a study of compliance with 2 days of amoxicillin taken 3 times per day as outpatient treatment, approximately 10% of families reported missing at least one dose. ^[12]

Antibiotic resistance, most importantly in S pneumoniae infection, also affects the choice of empiric treatment for occult bacteremia. Studies in Sweden, Greece, Israel, Portugal, Russia, and Nebraska have shown that 40-50% of cases of S pneumoniae in children attending daycare centers are resistant to penicillin. ^[81] Unlike the beta-lactamase of staphylococcal penicillin resistance, streptococcal resistance is mediated by altered penicillin-binding protein affinity for the drug. This resistance can be overcome by sufficiently high doses of antibiotic. Tissue concentrations sufficient to treat penicillin-resistant infections, other than meningitis, are achieved with oral therapy. ^[78]

To understand the role of penicillin-resistant pneumococcus in serious bacterial infection and occult bacteremia, realize that all pneumococci are not equal, antibiotic resistance patterns are not static, and resistance does not necessarily equal virulence. Penicillin resistance varies from mildly resistant (minimal inhibitory concentration [MIC] < 0.1), to intermediately resistant (MIC 0.1-1), to highly resistant (MIC >1). The prevalence of penicillin resistance is increasing over time, and no change in mortality seems to be associated with invasive pneumococcal disease due to the increase in antibiotic-resistant pneumococcus. ^{[24, 78, 82]}

Longitudinal studies of invasive pneumococcal disease show that the prevalence of intermediately penicillin-resistant pneumococcus (MIC 0.1-1) has increased from 5-10% in 1993 to 22% in 1999, and highly penicillin-resistant pneumococcus (MIC >1) has increased from 4% in 1993 to 15% in 1999. ^{[10, 39, 78]} A survey of pneumococcal meningitis in the mid 1990s found 13% intermediately penicillin-resistant pneumococcus (MIC 0.1-1) and 7% highly penicillin-resistant pneumococcus (MIC >1). ^[82]

Antibiotic pressure likely has a large role in selecting for antibiotic-resistant pneumococci, and a longitudinal study of invasive pneumococcal disease found an increased risk of penicillin resistance in patients who have used antibiotics in the last 30 days. ^[24]

Since the end of the 1980s, researchers have been concerned that penicillin-resistant pneumococcus may also be resistant to third-generation cephalosporins. ^[24] At that time, less than 1% of pneumonococci were resistant to ceftriaxone. ^[39] Since then, ceftriaxone resistance has increased but remains significantly less common than penicillin resistance. ^{[24, 39, 82]}

Longitudinal studies of invasive pneumococcal disease show that the prevalence of intermediately ceftriaxone-resistant pneumococcus (MIC 0.1-1) has increased from 3% in 1993 to 9% in 1999. ^{[10, 24, 39]} Highly ceftriaxone-resistant pneumococcus (MIC >1) has increased from 0.5% in 1993 to 2% in 1999. ^[24] A survey of pneumococcal meningitis in the mid 1990s found 4.4% intermediately ceftriaxone-resistant pneumococcus (MIC 0.1-1) and 2.8% highly ceftriaxone-resistant pneumococcus (MIC >1). ^[82]

Because of the frequency with which children with fever present to emergency departments and clinics for evaluation, the cost of evaluating and treating children with FWS can be considerable. Several authors have examined how well screening works in identifying infants and young children with occult bacteremia and how efficient empiric treatment is in preventing sequelae of bacteremia, namely meningitis. Costs of treatment and cost savings in preventing hospitalization, morbidity, and mortality have also been addressed to assess whether screening and empiric treatment are cost-effective strategies.

Screening febrile infants younger than 3 months by means of history, physical examination, and laboratory tests and treating low-risk infants as outpatients has been shown to be cost-effective. ^[10] Furthermore, an analysis of the Philadelphia criteria in 1993 found that outpatient treatment based on these low-risk criteria costs $3,100 less per patient than with inpatient treatment. ^[75]

Screening febrile infants and children aged 3-36 months based on age, degree of fever, and laboratory results has also been found to be a cost-effective and reasonable approach. ^{[1, 10, 20, 21]} See Lab Studies for statistics associated with different laboratory values used as screening tools for occult bacteremia; most studies determined that ROC curves were most favorable for WBC counts fewer than 15 per HPF or ANCs fewer than 10, criteria that were used to define low-risk children. Although these values have an NPV of approximately 99% for occult bacteremia, numerous reviews have noted that these cutoff values may still miss 25% of children with occult bacteremia because of the large numbers of febrile children presenting for evaluation. ^{[1, 20, 21]}

Determining the number needed to treat (NNT) to prevent a given event is another method used to assess the effectiveness of screening criteria. Two studies have analyzed the NNT to prevent meningitis for different laboratory screening criteria in febrile children aged 3-36 months with a temperature of more than 39°C. One used a WBC count greater than 15,000/μL and found an NNT of 500 to prevent one case of meningitis, and the other used an ANC greater than 10,000/μL and found an NNT of 240. ^{[1, 21]}

A recent formal estimate of cost-effectiveness compares the cost of screening and treatment of febrile children using numerous criteria. ^[35] This analysis also estimates the cost of complications associated with treatment and hospitalization and estimates the costs incurred while treating patients with sequelae from untreated infections. For the rate of occult bacteremia in febrile young children, this analysis uses an estimate of 1.5%, which is consistent with other current estimates. ^{[23, 21]} At this rate of bacteremia, empiric testing and treatment were found to be the most cost-effective approaches for treatment of febrile children; the cost is $72,000 per life-year saved. This strategy also favorably compares with other medical treatments that are considered cost-effective.

Many authors, including the authors of this article, anticipated that the rate of occult bacteremia would markedly decrease following widespread use of the 7-valent conjugate pneumococcal vaccine. ^{[2, 35, 38]} Using an estimate of 0.5% for the predicted rate of occult bacteremia, the authors have also calculated the cost-effectiveness of several approaches to treat febrile children. At this rate of bacteremia, the cost of empiric testing and treatment of febrile children increases markedly from $72,000 to more than $300,000 per life-year saved. With the changes in pneumococcal disease (an initial decline followed by a resurgence of nonvaccine strains), the final outcome is uncertain. What is clear is that risk-based estimates of likelihood of bacteremia were almost without exception obtained in a largely prevaccination era.

The sensitivity and specificity of clinical judgment in predicting occult bacteremia and serious bacterial infections have widely varied in previous studies, with a consensus that clinical judgment is not a reliable indicator of occult bacteremia. ^{[1, 10, 14, 20, 51]} Clinical judgment has been estimated to be 28% sensitive and 82% specific in predicting occult bacteremia, not inconsistent with previous studies performed on children aged 3-36 months. At a decreased predicted rate of occult bacteremia of 0.5%, treatment of febrile children based on clinical judgment was found to be considerably more cost-effective than other approaches; the cost is $38,000 per life-year saved.

The cost-effectiveness analysis indicating cost per life-year saved per intervention is as follows ^[35] :

• Tissue plasminogen activator (TPA) for acute myocardial infarction - $32,678

• Medical treatment for hypertension - $20,000

• Coronary artery bypass grafting (CABG) for myocardial infarction - $7,000

• Empiric testing and treatment in febrile children when rate of occult bacteremia is 1.5% - $72,300

• Empiric testing and treatment in febrile children when rate of occult bacteremia is 0.5% - Over $300,000

• Treatment based on clinical judgment, sensitivity 28% and specificity 82%, when rate of OB is 0.5% - $38,000

If the rate of bacteremia declines to 0.5%, this analysis concluded that clinicians should reevaluate their approach to highly febrile children and eliminate strategies that use empiric testing and treatment. ^[35]

In a retrospective cohort study of pediatric patients with gram-negative bacteremia, beta-lactam monotherapy resulted in a lower incidence of subsequent nephrotoxicity than combination therapy without compromising survival. ^[83] Of the 879 patients studied, 537 (61.1%) received combination treatment. Use of combination therapy had no significant effect on 30-day mortality. Of the 170 patients with evidence of acute kidney injury, 135 (25.1%) were treated with combination therapy and 35 (10.2%) with monotherapy.

A retrospective time series analysis concluded that changes in the etiology of pediatric bacteremia have implications for prompt, appropriate empirical treatment. The authors reported that pediatric bacteremia in the ED is health care associated, which increases length of inpatient stay. New tools to improve recognition are necessary for prompt, effective antimicrobial administration. ^{[84, 85]}

Treatment algorithms

The 1993 Practice Guidelines, published jointly in Pediatrics and Annals of Emergency Medicine, has dominated the US approach to febrile patients aged 3-36 months. ^[10] Considerable debate in the medical literature has followed the publication of these guidelines, and surveys indicate that considerable variation from the guidelines occurs in practice among pediatricians, family practitioners, and emergency department physicians. ^{[5, 34, 42, 53, 86]}

For febrile infants and young children aged 3-36 months, the 1993 Practice Guidelines recommended no tests or antibiotics for children with a temperature of less than 39°C and a nontoxic appearance. For children aged 3-36 months with a temperature of at least 39°C and a nontoxic appearance, a blood culture and empiric antibiotics were recommended, either for all children (option 1) or for children with a WBC count higher than 15,000/μL (option 2).

All children who appeared toxic were admitted to the hospital for sepsis workup and parenteral antibiotics pending culture results. Urine cultures were recommended for males younger than 6 months and females younger than 2 years, stool cultures were recommended for children with blood or mucus in the stool or more than 5 WBCs per HPF on stool smear, and chest radiography was recommended for children with dyspnea, tachypnea, rales, or decreased breath sounds. Follow-up in 24-48 hours was recommended for children who had cultures drawn.

In response to the 1993 Practice Guidelines, Kramer and Shapiro published an alternate approach that involved less laboratory screening and no empiric antibiotic treatment. ^[5] Febrile children aged 3-36 months were carefully assessed for bacterial foci; children with a toxic appearance were admitted to the hospital for sepsis workup, and focal infections were appropriately treated. Children who appeared well and had no focus of infection received a urinalysis if appropriate for age, whereas all children received no other laboratory tests and no antibiotics and were followed up in 24 hours to assess for worsening or persistence of signs and symptoms of infection.

A 1999 review by Kuppermann proposed an approach to the febrile child aged 3-36 months that was based on the risk of occult bacteremia during a time after Hib had been eliminated but before the introduction of pneumococcal vaccine. ^[1] His algorithm divided children into the following 2 groups based on risk: those aged 3 months to 2 years and those aged 2-3 years. He also recommended laboratory screening with ANC rather than a WBC count.

Kuppermann recommended no laboratory tests and no antibiotics for children aged 2-3 years with a nontoxic appearance and with a temperature of less than 39.5°C and for children aged 3 months to 2 years with a nontoxic appearance and with a temperature of less than 39°C. ^[1] For children aged 3 months to 2 years with a temperature of at least 39°C and a nontoxic appearance and for those aged 2-3 years with a temperature of at least 39.5°C and a nontoxic appearance, a blood culture and empiric antibiotics were recommended if the ANC was greater than 10,000/μL.

In 2000, Baraff published a review that included immunization status in the decision analysis of FWS. ^[2] Because of the low overall risk of occult bacteremia in children aged 3-36 months with FWS who have received the 7-valent conjugate pneumococcal vaccine, Baraff recommended that no blood work be performed in these patients irrespective of the degree of fever. He also recommended that no blood work be performed for FWS in children with a temperature of less than 39.5°C.

A blood culture and empiric antibiotics is recommended for children with an ANC of greater than 10,000/μL or a WBC count of greater than 15,000/μL if the child's temperature is at least 39.5°C and he or she has not received the pneumococcal vaccine. Baraff stated that for children who have received the pneumococcal vaccine, the overall prevalence of occult pneumococcal bacteremia should decrease by 90%, making screening of the WBC count or ANC impractical.

In 2004, Nigrovic and Malley published a management guideline, currently in use at the Children's Hospital Boston's emergency department. ^[3] This guideline is also based on the low risk of occult bacteremia in infants immunized against H influenza type b and S pneumoniae. It recommends that routine screening laboratory tests should not be performed for well-appearing febrile infants who have received 3 doses of 7-valent pneumococcal vaccine and 3 doses of Hib vaccine. Although acknowledging the ongoing concerns over the appropriate approach to infants and children with FWS, the authors conclude that this new approach is reasonable based on the best available information.

With the use of Prevnar 13, new studies of current risks and pathogens involved in occult bacteremia are needed, and updated practice guidelines are probably overdue.

For application of the algorithm approach to febrile infants and young children aged 3-36 months, see the image below.