Discussion

Treatment of gram-negative infections in children continues to be challenging due to increasing MICs and prevalence of resistance. A recent meta-analysis showed an increase in MDR P. aeruginosa from 15.4% to 25% between 1999 and 2012. ^{Reference Logan, Gandra and Mandal25} One potential method for suppressing resistance development is pharmacodynamic dosing. ^{Reference Craig4–Reference Mouton, Ambrose and Canton7} Although these methods have shown enhanced clinical response rates and improvement in surrogate markers of outcomes in adults, only a small amount of published information has shown similar outcomes in pediatric patients. ^{Reference Kuti, Florea and Nightengale8–Reference Lorente, Lorenzo and Martin10,Reference Shabaan, Nour and Elsayed22,Reference Cies, Moore and Enache23,Reference De Waele, Lipman and Akova26– Reference Nichols, Chung and Knoderer34} Unfortunately, because pediatric patients have different pharmacokinetic parameters than adults, the specifics of dosing optimization cannot be simply extrapolated from adult models. Monte Carlo simulations have been recommended as one method to help optimize pediatric antibiotic dosing. ^{Reference Bradley, Garonzik and Forrest12} Many of these simulations have found that traditional dosing of β-lactams was only predicted to provide bactericidal activity for many organisms with low MICs. Courter et al ^{Reference Courter, Kuti and Girotto20} showed the potential for continuous and prolonged infusions to optimize pharmacodynamics of time-dependent antibiotics in children. Although traditional doses with standard intermittent infusions were unlikely to provide bactericidal activity against P. aeruginosa strains, use of higher dosing or PI or CI achieved 90% likelihood of bactericidal exposures for common β-lactam antibiotics. ^{Reference Courter, Kuti and Girotto20}

In this study, most children’s hospitals use PI of cefepime, piperacillin-tazobactam, and meropenem only in patients with known high MICs or unique pharmacokinetic parameters (e.g., cystic fibrosis). In comparison, most rarely or never used CI for β-lactams. Reasons for the not using PI or CI were lack of guidelines, lack of experience, and insufficient patient venous access.

Although the pediatric Surviving Sepsis guidelines recommend the use of PI and CI in treating pediatric patients, they do not provide any guidance regarding dosing or monitoring. ^{Reference Weiss, Peters and Alhazzani35} To address these barriers, studies assessing benefits and outcomes in pediatric patients are needed.

Monte Carlo simulations have also been performed to evaluate and optimize fluoroquinolone dosing in children. ^{Reference Courter, Nichols and Kazazian21,Reference Meesters, Michelet and Mauel24} In simulations, dosages of ciprofloxacin at 30–40 mg/kg/day every 12 hours were needed to achieve high probability of meeting pharmacodynamic targets. ^{Reference Meesters, Michelet and Mauel24} For levofloxacin, the specific dosage needed for optimal bactericidal activity against gram-negative organisms is uncertain. A Courter et al ^{Reference Courter, Nichols and Kazazian21} model suggested that in children 5–14 years, even doses of levofloxacin 8 mg/kg every 12 hours only achieved pharmacodynamic targets in 90% of patients for gram-positive bacteria with an MIC ≤ 1 mg/L. ^{Reference Courter, Nichols and Kazazian21} Given that gram-negative bacteria require higher pharmacodynamic targets, optimal activity would only be seen at even lower MICs.

Based on survey data, pharmacodynamically optimized dosing of ciprofloxacin was more commonly utilized than optimized dosing of levofloxacin (by 65% and 17% of respondents, respectively). Interestingly, almost 10% of respondents said that they would not use a fluoroquinolone to treat gram-negative sepsis. These responses could possibly reflect different interpretations of the survey question, the fact that fluoroquinolones are not a first-line–recommended agent in pediatric sepsis, or limited literature or previous experience with suboptimal outcomes.

The TDM of β-lactams has been used sporadically in pediatrics, and use will likely increase thanks to new recommendations in the Pediatric Surviving Sepsis Campaign International Guidelines. ^{Reference Weiss, Peters and Alhazzani35} In this survey, TDM was reportedly used at 20% of institutions, with lack of published guidelines, testing and laboratory support identified as common barriers. Responses also reflected an overall lack of experience and confidence with employing TDM; few respondents provided an opinion on appropriate therapeutic goals for β-lactam monitoring. For this technology to see wider utilization, more literature, guidance, and timely processing are needed.

Debate continues regarding the use of certain cephalosporin antibiotics to treat ESBL-positive organisms. ^{Reference Dudley, Ambrose and Bhavnani36} Data from adults studies suggest that outcomes are closely related to the MIC of the infecting organism and that clinical failure significantly increases with increasing MIC. ^{Reference Andes and Craig37–Reference Kang, Kim and Park39} No specific pediatric data are available on this topic, making it uncertain whether lowering the MIC break points is sufficient to ensure successful outcomes with susceptible ESBL-positive organisms. This cautious outlook seems to be shared by survey respondents, who reported using cephalosporin antibiotics only on a case-by-case basis (63%) or never (36%). Example scenarios in which respondents would use a cephalosporin included low-inoculum infections and instances in which an infection is exposed to high levels of antibiotic for an extended period (e.g., cystitis without bacteremia treated with renally eliminated antibiotics).

The implementation of the SDD category by the CLSI has allowed for retained utilization of important antimicrobials in adult patients by providing specific dosing regimens to use with higher MICs. ⁴⁰ Caution should be taken when extrapolating CLSI recommended doses to pediatric patients due to differing pharmacokinetic parameters in children. and such dosing should be evaluated specifically at the SDD MIC. ^{Reference Courter, Kuti and Girotto20,Reference Nielsen, Forrester, Girotto, Dassner and Humphries41} Concern about extrapolating these standards to pediatric patients was shared by most survey respondents; only 29% of programs reported dose modification with SDD MIC organisms.

This study included a large number of participating hospitals, in part due to the utilization of the SHARPS Collaborative. As expected, most responding institutions were freestanding children’s hospitals with >100 beds. Most survey responders were ASP pharmacists. Surveyed hospitals were supported by a median of 1 FTE ASP pharmacist with some level of postgraduate training.

Programs employing a pharmacist with pediatric training, either in pediatrics or pediatric infectious diseases, were more frequent users of PI and CI β-lactams, and they performed more dose modification based on MIC and infection severity. Although this finding is likely multifactorial, it is probably related to the children’s hospitals hiring well-trained pharmacists to manage critically ill children and pharmacists with pediatric training better understanding children’s unique pharmacokinetics.

In this study, most reporting hospitals were ≥100 beds and met the recommended ASP pharmacist FTE percentage with a median of 1 FTE. ^{Reference Doernberg, Abbo and Burdette42} These factors may be why pharmacist FTE was not associated with any observed differences in responses. Physician FTE was lower, at a median of 0.3 (IQR, 0.2–0.6). Importantly, sites that reported lower ASP physician FTE were less likely to participate in dose optimization strategies, perhaps because the physician is not given sufficient time to actively assist and champion the ASP. ^{Reference Doernberg, Abbo and Burdette42}

This study has several limitations. Data collection was from respondents entering information based upon their practice. Bias may have been introduced into the data because only 1 staff member responded to the survey questions, and different staff may have had different responses based on their experiences. This study may also have been affected by sampling bias; only hospitals that were part of the SHARPS Collaborative listserv received the survey. Most responding institutions were also freestanding children’s hospitals, potentially limiting the generalizability of these findings to other institution types. Because response was voluntary, institutions that elected to participate in this survey may inherently place higher priority on antimicrobial dosing initiatives, which may have introduced further bias to the analysis.

Overall, variation exists in the optimization of dosing antimicrobials for pediatric patients in an era of increasing resistance. Further guidance by national and international groups is needed to continue to guide and increase pharmacodynamic dosing of children.