Critically ill neonatal and pediatric patients face a higher risk of nosocomial infection due to a combination of immature humoral and cellular immunity and the use of invasive techniques such as central venous catheterization and endotracheal intubation. Neonatal bacterial sepsis causes death in 10–20% of cases and morbidity in 25–30% of survivors. Therefore, effective and timely antimicrobial treatment regimens are often life-saving (1).

Selection of antimicrobial therapy for a critically ill patient may vary based on the nature of treatment (empirical or directed), source of infection and/or likely pathogen(s) and susceptibility. Vancomycin, a glycopeptide antibiotic, is often prescribed to cover resistant Gram-positive organisms such as methicillin-resistant Staphylococcus aureus (MRSA) and coagulase-negative Staphylococci (CoNS), particularly methicillin-resistant coagulase-negative Staphylococci (MRCoNS). MRSA is among the most threatening pathogens and the principal indicator of nosocomial risk. The incidence of neonatal sepsis has also increased over time in developed countries among very low-birthweight (VLBW) infants, and CoNS has emerged as a major pathogen in this context (2, 3). In a NICU/PICU setting, empirical vancomycin therapy is commonly initiated in a patient with suspected staphylococcal infections until the susceptibility pattern of the pathogen is confirmed.

The optimization of the initial vancomycin dosing required to achieve a therapeutic plasma level, and reducing the time taken to reach this level, have been proven to improve clinical outcomes in adults (4). However, deviations from the optimal vancomycin dosing may occur in the neonatal and pediatric population due to inconsistencies in the recommended dosing algorithms (5–8). This lack of consensus is illustrated by the wide variation in dosing regimens recommended in reference guidelines and the targeted therapeutic levels (Supplementary Materials 1, 2) (9–14). One study conducted at a pediatric tertiary care medical center in the United States observed that 35% of initial vancomycin courses were inappropriate (15). Another study observed that 74% of vancomycin dosing regimens administered according to the British National Formulary for Children (BNFc) recommendations did not achieve the target trough level (16). This issue is concerning because the inappropriate or suboptimal use of vancomycin can lead to treatment failure and enable the emergence of antibiotic resistance. Many of the above recommendations are supported by a relatively weak level of evidence, and the optimal dosing and monitoring of vancomycin remain controversial. Recent guidelines for vancomycin dosing suggest an area under the curve (AUC)/minimum inhibitory concentration (MIC) ratio >400 as the primary targeted therapeutic parameter to treat MRSA infections (17). The considerable variability in dosing recommendations of vancomycin is an ongoing challenge.

Addressing antibiotic resistance has been a global health agenda over the recent years, and is currently one of the top research priorities in Hong Kong. In 2019, a multidisciplinary team consisting neonatologist, pediatricians, and clinical pharmacists initiated a mixed-methods approach to address appropriate antimicrobial use in the pediatric population. The overarching aim is to implement personalized, pharmacokinetic-guided dosing of antimicrobials in critically ill patients in the NICU/PICU through a mixed-methods research approach (Supplementary Material 3). At the preliminary stage, this study was developed to collect the expert opinions of a representative sample of physicians and pharmacists who practice in the NICU/PICUs of 12 major medical institutions in Hong Kong. This survey aimed to describe the initial vancomycin dose selection practices used in these NICU/PICUs while focusing on the key factors influencing dose selection, preferred references, and target trough level. We also aimed to identify the challenges associated with initial vancomycin dosing and the corresponding therapeutic drug monitoring (TDM) practices.

This multicenter cross-sectional survey was conducted between January and December 2018. Approval was obtained from The Chinese University of Hong Kong (CUHK) Survey and Behavioral Research Ethics Committee prior to study initiation.

This study aimed to describe the prescribing preferences of physicians and pharmacists regarding the initial IV intermittent vancomycin dosing and TDM in multiple NICU/PICUs of Hong Kong. Our results demonstrate that these clinicians used varying drug references and vancomycin dosing guidelines in their practices. Generally, the trends in vancomycin target trough levels were similar between the physicians and pharmacists. The highest target trough level was reported for suspected severe/complicated Gram-positive and MRSA infections, followed by MRCoNS and suspected moderate/uncomplicated Gram-positive infections.

Although the AUC/MIC ratio is the primary predictive therapeutic parameter during a vancomycin regimen, the monitoring of trough concentrations as a surrogate parameter is a common practice in clinical settings. Current guidelines regarding vancomycin TDM in adult patients with MRSA infection recommend aggressive dosing when the MIC of the organism is ≤1 mg/L, with a target AUC/MIC ratio >400. This goal is to increase the vancomycin dose exposure and ensure adequate serum and tissue concentrations, particularly in critically ill patients such as those with sepsis, infective endocarditis, meningitis, and pneumonia. Extrapolating adult data to the pediatric population, an expert consensus indicated that an AUC/MIC ratio >400 corresponds to a targeted trough concentration of 15–20 mg/L at a MIC >1 mg/L (22–24). However, our study revealed that the majority of physicians tended to select a target trough level of 10–15 mg/L for MRSA infection, which is lower than the level of 15–20 mg/L suggested by clinical guidelines. This finding has a few implications. First, a clinical consensus on the targeted trough among physicians and pharmacists is needed. Second, with a set of pre-determined trough levels for different clinical scenarios, our future work is to conduct in-house drug utilization studies to evaluate whether the targets are achieved with current dosing strategies. This will shed light on the appropriateness and adequacy of existing dosing approaches. Third, to address concerns about an increase in vancomycin treatment failures, we recommend that clinicians should actively monitor MIC trends within each institution and review the targeted trough ranges. This information should be used to advise the appropriate empirical vancomycin dosing in this special population.

Our results also demonstrated that physicians and pharmacists used different drug references. This practice may lead to a variety of initial dosing regimens. This lack of consensus regarding dosing is illustrated by the wide variations in dosing regimens and targeted therapeutic levels across the reference guidelines (Supplementary Materials 1, 2). Consequently, two studies reported that ~75% of vancomycin dosing regimens designed with respect to the British National Formulary for Children (BNFc) and Neofax recommendations do not achieve the target trough levels (5, 6). Although few studies have involved the pediatric population, the optimization of initial vancomycin dosing to reduce the time required to reach a therapeutic serum level has been proven to improve clinical outcomes. Accordingly, this process should be a goal of therapy (4). However, it is difficult to achieve dose optimization in the neonatal and pediatric population in the absence of an evidence-based empirical dosing recommendation. Our results obtained from a small population of pediatric physicians and pharmacists in Hong Kong suggest a lack of standardization regarding the administration of vancomycin to NICU/PICU patients. As the pediatric critical care service in Hong Kong moves toward centralization at a new pediatric hospital, this issue may become problematic at the early stage when physicians of different institutions adopt varying prescribing preferences. However, we foresee that the gradual development of a standardized dosing guideline for empirical and targeted vancomycin use and a centralized TDM service may potentially improve the overall care for NICU/PICU patients.

The surveyed physicians and pharmacists considered several factors to be important determinants of the initial vancomycin dosing. They gave unanimous responses on crucial determinants such as post-menstrual age, body weight, severity of infection, and renal function. These parameters have also been reported as factors of concern in the literature and references regarding the selection of an initial vancomycin dosing regimen. Among critically ill neonatal and pediatric patients, changes in renal function, fluid resuscitation, and tissue perfusion can influence the clearance of vancomycin. Importantly, the existing references and guidelines differ with respect to covariates (25). Standardized dosing regimens for pediatric patients, which includes neonates, infants, and children, remain controversial because of the high pharmacokinetic variability within this diverse category. Compared with adults, neonates have a higher extracellular fluid volume and a limited but rapidly increasing capacity for renal elimination. Consequently, vancomycin pharmacokinetics evolve significantly as neonates mature (25). The current landscape of multiple dose recommendations pose difficulties on the clinicians to determine dose that can achieve target AUC/MIC goals. This has led institutions to adopt their own measures of developing internal dosing recommendations based on clinical variables collected from their own specific patient population and through stimulation of real-world data (26, 27). Recent studies have also demonstrated the value of using population pharmacokinetic data to personalize empirical vancomycin dosing. Researchers have begun to assess the validity, predictive performance, and generalizability of a published vancomycin pharmacokinetic model retrospectively through a model simulation based on a cohort of neonatal and pediatric patients (28, 29). The existing literature contains promising evidence supporting the use of pharmacokinetic parameters and computational strategies to estimate the extent of drug exposure in this special population. Our proposed future work is to examine the contribution of the identified clinical variables on serum levels using real-world data through drug utilization studies, and using these clinical variables as predictors in developing dosing calculators. We believe that this approach can be extended to establishing pharmacokinetic-guided dosing calculators of antimicrobials in children with other diseases or ethnic origins.

The TDM of medications with narrow therapeutic windows, such as vancomycin, has long been a standard of practice worldwide. Studies investigating the effects of pharmacist-managed TDM services have repeatedly demonstrated improvements in clinical outcomes. The benefits include an improvement in clinical efficacy, as indicated by reductions in the duration of antibiotic use, length of hospital stay, and mortality (4, 17). In a locally conducted pilot study, a pharmacist-managed TDM service exhibited a nearly 50% reduction in the time required to reach target therapeutic levels (30). Despite the purported benefits of TDM, the availability and scope of this clinical service varies among local public hospitals. The respondents in our study identified various obstacles to the optimization of TDM services in their institutions, including logistical barriers such as a slow laboratory turnover time and a lack of autonomy of pharmacists to modify dosing regimens. Based on our observations of local practices, we note that vancomycin dosage adjustments are usually based on incrementally derived dose adjustments, rather than individualized calculations based on pharmacokinetics. This practice suggests a misconception of the principles of TDM and a significant gap in service. The establishment of a well-defined pharmacist-led TDM service and the use of population-based pharmacokinetic modeling principles may resolve challenges such as the determination of a dose for a target serum level.

These study findings should be considered in the context of several limitations. We used a convenience sampling method and obtained a modest response rate. However, efforts were made to disseminate the survey through representatives (e.g., pharmacy managers) at the participating institutions. To highlight, these institutions include hospitals with the largest NICUs and PICUs in Hong Kong and highest usage rates of vancomycin. This approach may have helped to establish the sampling frame and likely strengthened the validity and applicability of the study findings in Hong Kong. It is inevitable that the aggregate responses might be dominated by the major acute hospitals. However, we also purposively distributed the survey to the smaller institutions to ensure that their responses are not omitted due to suboptimal sampling. We also found that the distribution of respondents across the main districts did appear reasonable. The clinicians' practices were self-reported and may not accurately reflect their actual practices in a clinical setting. Therefore, we are currently conducting drug utilization studies to evaluate antimicrobial usage patterns and the impacts on tangible clinical outcomes in the pediatric population. As this study is descriptive in nature, we emphasize that the findings should not be considered the primary and sole evidence on any aspect of this subject. Our study should be interpreted as an effort to understand the perceptions and practices of the pediatric critical care medicine community in Hong Kong. More importantly, findings of this survey will be used to direct future studies within this theme-based research. For example, we found that physicians and pharmacists collectively indicated a target trough of 15–20 mg/L for suspected severe/complicated Gram-positive infection; we will apply this target trough as the threshold when we evaluate clinical outcomes in the subsequent drug utilization studies, as well as whether the current dosing approach can effectively achieve this target trough range.

In conclusion, we believe that our findings have important implications. From a practice perspective, a standardized dosing guideline is needed to decrease variability and improve local practice. The clinicians' prescribing practices and the effectiveness of using current dosing guidelines to achieve targeted therapeutic trough levels should be evaluated in future drug utilization studies. The multidisciplinary TDM service in Hong Kong can be improved further by addressing logistic barriers and implementing a well-structured, evidence-based service protocol. From a research perspective, the next phase of this work includes conducting a multi-centered drug utilization studies to review real-world data on vancomycin use in the PICU/NICU. The clinical and outcomes data collected will be used to construct pharmacokinetics model for the PICU and NICU cohorts. Future work will also include the performance of comparative studies to demonstrate the effectiveness of a pharmacokinetics-driven dosing regimen vs. standard practice, particularly within the Chinese pediatric population. Globally, drug-related problems (particularly inappropriate dose selection) are common in NICU/PICU, hence predominating problems of suboptimal treatment. We believe that our systematic approach in identifying and addressing therapeutic gaps may potentially be extended to standardize dosing of antimicrobials and other drugs of narrow therapeutic index in the NICU/PICU. The eventual goal is to achieve personalized, pharmacokinetic-guided dosing of drugs in critically ill neonates and pediatric patients.

The datasets generated for this study are available on request to the corresponding author.

The studies involving human participants were reviewed and approved by The Chinese University of Hong Kong (CUHK) Survey and Behavioral Research Ethics Committee. Written informed consent for participation was not required for this study in accordance with the institutional requirements.

TC, JL, TL, CE, and YC: analysis. TC, JL, and YC: drafting the work. All authors: conception or design of the work, acquisition of data, interpretation of data, revising it critically for important intellectual content, and provide approval for publication of the content.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.