Rational antibiotic use


A large retrospective multi-centre US cohort study examined early antibiotic use in inborn neonates: 104 803 were treated with ampicillin plus gentamicin, while 24 111 were treated with ampicillin and cefotaxime.7 Only 2% of babies treated had proven early-onset sepsis. After adjusting for possible confounders, neonates treated with ampicillin plus cefotaxime were more likely to die (4.7%) than neonates treated with ampicillin plus gentamicin (2.3%), (adjusted OR 1.5, 95% CI 1.4–1.7) and less likely to be discharged from hospital. The increased mortality occurred in both pre-term and full-term babies. Different units used different regimens, so these results do not prove that cefotaxime increases mortality, but the size of the study and the degree of the difference in mortality led the authors to conclude that cefotaxime use in the first 3 days after birth, compared with gentamicin, is a surrogate for an unrecognized factor or is itself associated with an increased risk of death.7


As discussed in Chapter 16, broad spectrum antibiotics, particularly third generation cephalosporins, are associated with an approximate doubling of the risk of neonatal candidiasis, which is a particular problem for low birth-weight infants and babies with predisposing gastro-intestinal abnormalities.8


Listeria and enterococci (faecal streptococci), which can both cause early-onset infection, are inherently resistant to third generation cephalosporins, so even if it is decided to use a third generation cephalosporin as empiric therapy for early-onset sepsis, it should be used in conjunction with an agent like ampicillin to which these organisms are sensitive.


The disadvantages of using aminoglycosides are predominantly concerns about toxicity and the need to monitor drug levels. Drug levels are monitored to make sure therapeutic levels are achieved and/or to avoid accumulation. If the baby stops antibiotics after 48–72 hours, drug levels are unnecessary because toxicity is cumulative and associated with prolonged aminoglycoside use.


Hearing loss in low birth-weight babies could be due to aminoglycoside toxicity, but other possible causes are genetic factors unrelated to aminoglycosides and environmental factors such as perinatal asphyxia, post-natal hypoxia, acidosis, bilirubin toxicity and noise. It is difficult to assess the relative contribution of aminoglycoside toxicity to neonatal deafness, particularly in pre-term infants. We could find no RCTs comparing hearing in babies who did and did not receive aminoglycosides. A Cochrane systematic review of 11 studies of 574 neonates treated with once daily or multiple daily doses of aminoglycosides reported minimal ototoxicity and nephrotoxicity with either regimen.9 A 5-year New Zealand study compared hearing tested by otoacoustic emissions (OAE) in infants managed in a neonatal intensive care unit who received gentamicin and/or vancomycin (a glycopeptide antibiotic, not an aminoglycoside). Somewhat surprisingly, they found that babies who received gentamicin but no vancomycin had a significantly lower risk of OAE failure than babies who received neither antibiotic, while vancomycin use increased the risk of OAE failure.10


Aminoglycoside ototoxicity has been described in association with mutations of the mitochondrial 12s rRNA genes, which occur with a frequency of approximately one in 500 Europeans.11 However, in a prospective case-cohort US neonatal study, four of 378 neonates exposed to gentamicin had mitochondrial mutations, but only one failed the initial hearing test.12. A German study addressed possible gentamicin ototoxicity with two approaches.13 A cohort study of 8333 children examined for hearing disorders, found 134 (1.6%) had received previous treatment with gentamicin, and only eight of 134 (6%) had any degree of sensorineural hearing impairment. All eight had an alternative cause for hearing loss, such as perinatal asphyxia, acidosis, severe neonatal jaundice or meningitis. In addition, the author compared vestibular function in 30 children with normal hearing who received gentamicin in the neonatal period and 30 matched controls and found no difference.13


There are insufficient data to say whether any of the frequently used aminoglycosides, amikacin, gentamicin, netilmicin and tobramycin, is more toxic for neonates than the others, but there is most safety data for gentamicin.8–13 Aminoglycosides accumulate in renal tissue and in the perilymph of the inner ear causing progressive toxicity, so it is recommended not to use them for longer than 7 days, if possible.


Antibiotics select for organisms that carry ancient resistance genes.14 Antibiotic resistance is an example of Darwinian selection. The term ‘unnatural selection’ emphasizes the iatrogenic nature of rapid selection of resistant organisms from misusing broad spectrum antibiotics.15 To minimize resistance, narrow spectrum antibiotics should always be preferred if the choice does not jeopardize babies.15


On the basis of the above data suggesting possible increased mortality with third generation cephalosporins and relatively reassuring data about aminoglycoside toxicity, the following recommendations are suggested:






Recommendations for choice of empiric antibiotics for possible early-onset sepsis


  • The empiric antibiotics for babies with possible early-onset neonatal infection should be based on the organisms causing infections locally and their antibiotic susceptibility.
  • The antibiotic regimen should be as narrow spectrum as possible.
  • A penicillin (ampicillin or penicillin) and an aminoglycoside (e.g. gentamicin) is the regimen of choice if this covers the local organisms adequately.
  • Broad-spectrum antibiotics such as third generation cephalosporins and meropenem should not be included in empiric regimens unless essential because of local epidemiology.




5.1.2 Choice of antibiotics for suspected late-onset sepsis


As for early-onset sepsis, the rational choice of antibiotics for possible late-onset sepsis should be based on the likely organisms causing sepsis and their antibiotic susceptibility. Like early sepsis, all countries surveyed report late-onset infections with both Gram-positive cocci and Gram-negative bacilli, so empiric antibiotic regimens need to cover these groups of organisms.


Possible regimens, in order from the narrowest to the broadest spectrum, include the following.



1. A semi-synthetic penicillin with anti-staphylococcal activity, for example, cloxacillin, flucloxacillin, methicillin, nafcillin, oxacillin, to cover Gram-positive cocci plus an aminoglycoside (e.g. gentamicin, netilmicin, tobramycin) to cover Gram- negative bacilli;

2. Penicillin or ampicillin to cover Gram-positive cocci plus an aminoglycoside to cover Gram-negative bacilli if staphylococcal infections are very rare;

3. A broad spectrum β-lactam antibiotic such as a third generation cephalosporin (e.g. cefotaxime, ceftriaxone, ceftazidime) or an extended-spectrum penicillin with Pseudomonas activity (e.g. piperacillin, ticarcillin) used alone or with an aminoglycoside;

4. A broad-spectrum carbapenem, for example, meropenem, as monotherapy to cover ESBL-producing Gram-negative bacilli; or

5. A combination of an extended spectrum penicillin with a β-lactamase inhibitor (e.g. piperacillin–tazobactam, ticarcillin–clavulanate) as monotherapy. Specific anti-anaerobic cover is not used empirically in general.

A Cochrane systematic review failed to find any studies that effectively compared different antibiotic regimens,16 with the exception of the study already quoted which compared ticarcillin–clavulanic acid with piperacillin and gentamicin for both early- and late-onset sepsis and found no difference in mortality.3


Continuing surveillance of late-onset infections is needed to monitor for changes in prevailing organisms and/or their susceptibility.17


In Western industrialized countries, the major need in empiric treatment of possible late-onset sepsis is to cover against fulminant infections, which are usually caused by Gram-negative bacilli or S. aureus (methicillin-sensitive S. aureus (MSSA) or methicillin-resistant S. aureus (MRSA)). Although CoNS are the most common single organism associated with late-onset sepsis, it is extremely rare for CoNS septicaemia to cause fulminant infection.18,19


The mortality from Pseudomonas septicaemia is 52–74%, compared with 10–25% for other Gram-negative bacilli,19-21 so empiric regimens should cover Pseudomonas, particularly if babies on the neonatal unit are known to be colonized.22


There is strong evidence that broad-spectrum antibiotics, as in regimens (3), (4) and (5) above, are major drivers of the emergence of antibiotic- resistant Enterobacteriaceae, particularly ESBL-producing Gram-negative bacilli.23-26 Their use should be avoided if possible. In some developing countries, however, ESBL-producing Gram-negative bacilli circulate as community organisms as well as in hospital, and are an important cause of both early- and late-onset neonatal infection.27,28 In Kolkata, for example, more than half of all early- and late-onset neonatal infections were caused by multi-resistant Gram-negative bacilli.28






Question: For Gram-positive cover, should empiric regimens for suspected late sepsis in Western countries include vancomycin?

In a 10-year multicentre study from Australia and New Zealand, four of 1281 babies (0.3%) with CoNS infections died, and CoNS may have contributed to the deaths of 20 other babies (1.6%).18 A 10-year study in a single US neonatal unit reported a total of 49 episodes of fulminant sepsis (lethal within 48 hours), of which 34 (69%) were caused by Gram-negative bacilli and only four by CoNS.19 There were 277 episodes of CoNS infection, so the proportion of fatal CoNS infections was 1.4%. During the study, a decision was made after 6.7 years to change the empiric regimen for suspected late-onset sepsis from vancomycin plus cefotaxime to oxacillin plus gentamicin. When vancomycin was used for Gram-positive cover there were 141 episodes of CoNS sepsis, of which two were fulminant. For the remaining period when oxacillin was used, there were 136 episodes and two were fulminant.19

The main reason for changing from vancomycin was to avoid selecting for vancomycin-resistant enterococci (VRE).19,29 There are a number of reasons that oxacillin and gentamicin was apparently no less safe than vancomycin and cefotaxime. First, even for supposedly fulminant CoNS infections, the baby may be dying from an unrelated cause and the CoNS is a contaminant. While this is difficult to prove or disprove, neonatal clinicians will understand the argument. Second, some CoNS will respond to oxacillin or flucloxacillin. Third, central line-associated CoNS infections often respond to removal of the central line, so if the line is removed, the choice of antibiotic may be less critical. Finally, although aminoglycosides are not recommended to treat staphylococcal infections, they do have some anti-staphylococcal activity, so it is possible that gentamicin provides some additional cover.

The glycopeptide antibiotic vancomycin provides broad cover against Gram-positive organisms, but there are concerns about the selection of VRE29 about possible ototoxicity,10 and also concerns that vancomycin use actually predisposes to Gram-negative sepsis.30 In a Dutch case-control study of 105 children (39 of them neonates) with Gram-negative bacteraemia, the strongest association was with prior use of vancomycin (OR 8.1; 95% CI 3.1–20.9). In a multiple logistic regression model, the use of vancomycin remained positively and strongly associated with Gram-negative bacteraemia (OR 3.9; 95% CI 1.3–11.2).30 S. aureus, MSSA or MRSA, are intermediate in virulence between CoNS and Gram-negative bacilli, although there is strain-to-strain variability in MRSA virulence.31 If MRSA is prevalent on a neonatal unit, the empiric late-onset regimen probably needs to include vancomycin, but if not the use of a semi-synthetic penicillin is much preferable.

Teicoplanin is a complex of five glycopeptides with a similar spectrum to vancomycin. It was mainly used in the 1990s in parts of Europe and has been suggested to have a better safety profile than vancomycin.32,33 However, it is less active than vancomycin against some CoNS, especially Staphylococcus haemolyticus.34 A search found only one small comparative study in neonates of low-dose prophylactic teicoplanin and vancomycin, which showed no difference in toxicity or efficacy.35 Although teicoplanin might be as effective as vancomycin, the limited pharmacologic data in neonates and concerns about antibiotic spectrum mean that it has been used rarely and cannot be advised as a suitable alternative to vancomycin for empiric regimens.

Recommendation: Vancomycin only needs to be included in the empiric antibiotic regimen for late-onset sepsis in neonatal units with a significant risk of MRSA infection.








Question: For Gram-negative cover in Western countries, what is the preferred regimen?

As discussed for early-onset sepsis above, there are theoretical concerns about the safety of aminoglycosides but the limited evidence available suggests they do not cause significant ototoxicity or nephrotoxicity in neonates. Aminoglycoside levels only need to be measured if they are continued for longer than 2–3 days. Aminoglycosides accumulate in the ears and kidneys causing progressive toxicity, so it is recommended not to use them for longer than 7 days, if possible. However, their use is preferred to broad-spectrum antibiotics, which are associated with the selection of multi-resistant organisms.15,36,37

If a strain of Gram-negative bacillus that is resistant to the aminoglycoside recommended in the empiric regimen becomes prevalent colonizing and/or infecting babies on a neonatal unit, one possibility is to change to another aminoglycoside to which the organism is sensitive. For example, when a gentamicin-resistant strain of Klebsiella colonized a number of babies and caused episodes of sepsis in a neonatal unit, a temporary change to netilmicin as empiric aminoglycoside of choice was successful, before returning to gentamicin.38








Recommendations for choice of empiric antibiotics for possible late-onset sepsis


  • The empiric antibiotics for babies with possible late-onset neonatal infection should be based on the organisms causing infections locally and their antibiotic susceptibility.
  • The antibiotic regimen should be as narrow spectrum as possible.
  • A semi-synthetic penicillin with anti-staphylococcal activity (e.g. cloxacillin, dicloxacillin, flucloxacillin, methicillin, nafcillin, oxacillin) and an aminoglycoside (e.g. gentamicin) is the regimen of choice if this covers the local organisms adequately.
  • Vancomycin and an aminoglycoside (e.g. gentamicin) should be the chosen empiric regimen if there is a high prevalence of MRSA.
  • Broad-spectrum antibiotics such as third generation cephalosporins and meropenem should not be included in empiric regimens unless essential because of local epidemiology.




5.1.3 Antibiotics for suspected community-acquired sepsis in developing countries


The WHO currently recommends ampicillin or penicillin G plus gentamicin to treat suspected community-acquired late-onset sepsis in developing countries. However, a non-Cochrane systematic review and meta-analysis of 19 studies from 13 countries found resistance or reduced susceptibility to ampicillin/penicillin and gentamicin and equally commonly to third generation cephalosporins in up to 40% of cases.39 An editorial on combating antimicrobial resistance subtitled ‘The war against error’ counselled against ‘one size fits all’ antibiotic recommendations and recommended tailoring antibiotic regimens to local epidemiology. The editorial also emphasized that bacteria have evolved resistance genes over millions of years, so escalating use of ever broader spectrum antibiotics is doomed to failure and alternative strategies to prevent infection are needed.40


5.1.4 Antibiotic rotation


Rotation of antibiotic regimens in neonatal units might reduce the selection of resistant strains by reducing antibiotic pressure. An adult surgical intensive care unit reported an improvement in antibiotic susceptibility of some Gram-negative bacilli following the introduction of an antibiotic rotation protocol, while the susceptibilities did not improve in the adjoining medical intensive care unit which did not introduce antibiotic rotation.41 An adult intensive care unit which introduced antibiotic rotation reported a reduced incidence of ventilator-associated pneumonia but no significant change in antibiotic susceptibilities.42 In contrast, two adult intensive care units reported that antibiotic rotation had no measurable effect on antibiotic susceptibility patterns.43,44


A systematic review published in 2005 found 11 studies of antibiotic rotation. Only four had attempted to evaluate the effects and there were considerable methodological problems, both in the reviews and implementation. The authors advised against the use of antibiotic rotation to attempt to reduce antibiotic resistance rates.45


Only one NICU study is reported.46 The unit was divided into two similar populations of babies. The empiric antibiotic regimen for suspected Gram-negative sepsis for one population (rotation team) was rotated monthly through gentamicin, piperacillin–tazobactam and ceftazidime. The other control team used an unrestricted choice of antibiotic according to the choice of the attending physician. After 1 year, 10.7% of infants in the rotation team and 7.7% of controls were colonized with resistant Gram-negative bacilli (p > 0.05).46


5.2 Prophylactic antibiotics


Prophylactic antibiotic use is when antibiotics are given to an uninfected baby with risk factors for infection to prevent the baby developing infection. Prophylaxis means prevention. This is different from empiric antibiotic use which means giving antibiotics to babies who may already be infected.






Question: Do prophylactic antibiotics compared with no prophylactic antibiotics improve outcome for any group of neonates?

All infants: An Italian RCT compared one with 3 days of ampicillin plus netilmicin for 130 infants admitted consecutively to the NICU. There was no difference in incidence of late-onset sepsis.47

Central venous catheters: We found a Cochrane systematic review48 and a separate non-Cochrane systematic review,49 both of which identified three small RCTs. Two studies used vancomycin and one used amoxicillin. Prophylactic vancomycin was associated with a statistically significant reduction in the incidence of bacteraemia but there was no change in mortality and no data on neurodevelopmental outcome. There was documented change in antibiotic susceptibility, although the methods for determining these were not robust. Both sets of authors cautioned that the routine use of prophylactic antibiotics for neonates with central venous catheters cannot be recommended because of concerns about the long-term safety.

Umbilical artery catheters: A Cochrane systematic review found only two small quasi-randomized trials and concluded there was insufficient evidence to recommend or advise against the use of prophylactic antibiotics.50

Umbilical venous catheters: A Cochrane systematic review found only one small trial which showed no effect of a short 3-day course of prophylactic antibiotics.51

Mechanical ventilation: A Cochrane systematic review found only one trial of fair quality which showed no effect of prophylactic antibiotics in any outcome, although bacteraemia rates were not reported.52

Neutropenia: As discussed in Chapter 4, neutropenia in the first few days of life may be due to overwhelming infection, but particularly in low birth-weight infants is much more commonly secondary to growth retardation or maternal pre-eclampsia.53 Neutropenia at birth (defined as <;2.2 neutrophils<;1000/μL) was associated with a risk of 14% of proven or presumed early-onset sepsis (<;48 hours) compared with 2% in non-neutropenic babies.52 Sepsis was proven in 6% of neutropenic babies.54

Only gold members can continue reading. Log In or Register to continue

Related posts:

  1. Urinary tract infections
  2. Infection control
  3. Bacterial infections
  4. Viral infections

Stay updated, free articles. Join our Telegram channel
Tags:
Jun 18, 2016 | Posted by in PEDIATRICS | Comments Off on Rational antibiotic use

Full access? Get Clinical Tree
Get Clinical Tree app for offline access