Variability in Ceftriaxone Dosing Across 32 US Acute Care Children’s Hospitals

Abstract

Ceftriaxone is one of the most common antibiotics prescribed for hospitalized children in the United States. However, ceftriaxone is not dosed consistently. Sepsis/serious bacterial infection had high dosing variability. Dosing for central nervous system infection was frequently suboptimal. Future efforts should focus on optimizing and standardizing ceftriaxone dosing.

This study examines variability in antibiotic dosing at US children’s hospitals. There is substantial variability in ceftriaxone dosing for sepsis or serious bacterial infection and a high percentage of suboptimal dosing for the central nervous system.

Most of the children admitted to US children’s hospitals receive an antibiotic during hospitalization [1]. While antibiotics are necessary in many instances, excessive and inappropriate use can lead to needless adverse drug events and contribute to the emergence of drug-resistant bacteria. Examples of inappropriate use include duplicating coverage, prescribing for an excessive duration, or dosing suboptimally.

Variability in care helps identify suboptimal practices that may be targets for quality improvement. Variability in inpatient pediatric antibiotic use has been studied broadly [1–4]. However, these studies relied on administrative datasets that can characterize general trends but leave some areas of antibiotic prescribing, such as antibiotic dosing, unexplored. There have been no published studies on dosing variability for antibiotics prescribed at US children’s hospitals.

In 2016, the Sharing Antimicrobial Reports for Pediatric Stewardship (SHARPS) Collaborative began conducting point prevalence surveys to capture more granular antibiotic utilization data, including dosing data [5]. Utilizing this dataset, we sought to understand if there was significant variability in dosing. Specifically, we examined ceftriaxone because it is one of the most frequently used antibiotics in hospitalized children, and anecdotal evidence suggests variability in dosing practices.

METHODS

Study Design

We analyzed data from a serial, cross-sectional survey of antibiotic prescribing at 32 US acute care children’s hospitals from July 2016 through December 2017 [5]. For this analysis, we limited the broader dataset to those receiving parenteral ceftriaxone. This study was approved by the institutional review board (IRB) at Children’s Mercy-Kansas City (the coordinating center). Washington University School of Medicine in St. Louis IRB approved this sub-analysis of ceftriaxone.

Data Source

The SHARPS Collaborative point prevalence survey (PPS) collected data on antimicrobial use in 32 US acute care children’s hospitals in 23 states [6]. The hospitals that submitted data were either freestanding or tertiary-care children’s hospitals within larger institutions. Anonymized patient data were collected and managed using the Research Electronic Data Capture (REDCap) tool hosted at Children’s Mercy-Kansas City [7]. On a single day from 6 quarterly time periods in 2016–2017, antimicrobial stewardship program representatives (physicians and/or pharmacists) from each hospital recorded and evaluated antibiotic use in hospitalized children.

Study Population

Inclusion criteria were patients age <18 years old with any active systemic antimicrobial order at 8 am local time. Chart review was performed to collect patient-specific clinical data and antimicrobial data, including dose, route, dosing interval, and indication.

For this sub-analysis, listwise deletion was used to exclude participants missing critical data (Supplementary Figure 1). Patients were also excluded if there was a concern for clerical error, such as if they likely received 2000 mg based on weight but the entry listed 200 mg or if their weight-for-age z-score seemed biologically implausible according to Centers for Disease Control and Prevention growth curves. Our cutoff was a z-score greater than ±10 after adjusting for prematurity [8]. Patients were excluded if their frequency of administration was not every 12 hours (Q12), every 24 hours (Q24), or one-time-only as these may also have been clerical errors.

Outcomes

We compared each total daily dosage (TDD) in milligrams per kilogram body weight per day (mg/kg/day) to Lexicomp’s indication-specific recommended daily doses (RDD) to classify if each TDD was optimal [9]. Prescriptions were classified as suboptimal if the TDD was less than the minimum range of the RDD –20% or greater than the maximum range of the RDD +20%. The 20% margin was chosen in consultation with clinical infectious diseases pharmacists to reflect typical practices for dose rounding.

We assumed that most of the prescriptions in this pediatric dataset were weight-based. If a patient was prescribed a maximum dose because their weight-based dose exceeded the maximum dose, our method for calculating TDD would lead to an overestimate of the frequency of underdosing. Depending on the indication, ceftriaxone has a maximum dose of 1000, 2000, or 4000 mg. Therefore, patients were excluded if they received TDD <40 mg/kg/day along with weight >20 kg and exactly 1000 mg per day, weight >40 kg and exactly 2000 mg per day, or weight >40 kg and exactly 4000 mg per day.

Statistical Analysis

Statistical analysis was performed using StataIC (version 16.0.806).

RESULTS

Ceftriaxone Use and Indications

Ceftriaxone was the second most prescribed antimicrobial agent in the database. In total, 1517 ceftriaxone prescriptions were included in our descriptive statistics analysis. Patients’ median age was 3.3 years (interquartile range: 0.8-8.4 years) (Table 1). The 4 most common indications were bacterial lower respiratory tract infection (LRTI) (448/1517, 29.5%), sepsis/serious bacterial infection (SBI) (186/1517, 12.3%), urinary tract infection (UTI) (156/1517, 10.3%), and central nervous system (CNS) infection (153/1517, 10.1%). While most of the prescriptions were dosed Q24, 60.8% of CNS infections (93/153) and 25.8% of sepsis/SBI (48/186) were dosed Q12.

Table 1.

Characteristics of Ceftriaxone Prescriptions in 32 US Acute Care Children’s Hospitals

	Did Not Receive Max TDD (n = 1312)	Received Max TDD (n = 205)	Total Received Parenteral Ceftriaxone (n = 1517)
Characteristic	No. (%)	No. (%)	No. (%)
Patient age
<= 30 d	20 (1.5)	0	—
1-5 mo	224 (17.1)	0	—
6-11 mo	126 (9.6)	0	—
1-4 y	399 (30.4)	0	—
5-11 y	398 (30.3)	34 (16.6)	—
≥12 y	145 (11.1)	171 (83.4)	—
Frequency
Every 12 h (Q12)	238 (18.1)	6 (2.9)	—
Every 24 h (Q24)	1061 (80.9)	197 (96.1)	—
One-time only	13 (1.0)	2 (1.0)	—
Daily dose (mg/d)
1000	71 (5.4)	24 (11.7)	—
2000	191 (14.6)	178 (86.8)	—
4000	60 (4.6)	3 (1.5)	—
Other	990 (75.5)	0	—
Type of infection
Community-acquired infection	—	—	1285 (84.7)
Healthcare-associated infection	—	—	159 (10.5)
Prophylaxis	—	—	47 (3.1)
Not used for prophylactic or treatment	—	—	3 (0.2)
Unknown	—	—	6 (0.4)
Not recorded	—	—	17 (1.1)
Medical service type
Medicine	—	—	819 (54.0)
Intensive care	—	—	449 (29.6)
Surgery	—	—	240 (15.8)
Not recorded	—	—	9 (0.6)
Indication
Bacterial LRTI	—	—	448 (29.5)
Sepsis/SBI	—	—	186 (12.3)
UTI	—	—	156 (10.3)
CNS infection	—	—	153 (10.1)
GI infection	—	—	123 (8.1)
Surgical treatment	—	—	94 (6.2)
CRBSI	—	—	50 (3.3)
Surgical prophylaxis	—	—	45 (3.0)
Febrile neutropenia	—	—	43 (2.8)
SSTI	—	—	42 (2.8)
URI	—	—	38 (2.5)
Other	—	—	34 (2.2)
AOM	—	—	26 (1.7)
Bone/joint infection	—	—	20 (1.3)
Viral LRTI	—	—	19 (1.3)
FUO	—	—	16 (1.1)
Cardiac infection	—	—	10 (0.7)
Unknown	—	—	6 (0.4)
Medical prophylaxis	—	—	5 (0.3)
Lymphadenitis	—	—	3 (0.2)
Empiric vs targeted
Targeted: pathogen and antimicrobial resistance confirmed first	—	—	185 (12.2)
Empiric and pathogen identified later	—	—	125 (8.2)
Empiric without positive cultures yet	—	—	379 (25.0)
Empiric but culture and susceptibility testing pending	—	—	513 (33.8)
Empiric without culture(s) performed	—	—	291 (19.2)
Not recorded	—	—	24 (1.6)
Number of additional antimicrobials
0	—	—	720 (47.5)
1	—	—	688 (45.4)
2	—	—	103 (6.8)
3	—	—	3 (0.2)
4 or more	—	—	3 (0.2)

The first column includes prescriptions that were not maximum doses and thus have calculated total daily dosages (TDD; mg/kg/d) used for variability analysis. The second column includes prescriptions that were maximum doses. The third column highlights all ceftriaxone prescriptions in the database.

Abbreviations: LRTI, lower respiratory tract infection; SBI, serious bacterial infection; UTI, urinary tract infection; CNS, central nervous system; GI, gastrointestinal; CRBSI, catheter-related bloodstream infection; SSTI, skin/soft tissue infection; URI, upper respiratory infection; AOM, acute otitis media; FUO, fever of unknown origin.

Dosage Variability by Indication

After excluding prescriptions with maximum daily doses (n = 205), we examined variability in TDD. Sepsis/SBI had the most variability (median 74.6, interquartile range width 49.8).

Suboptimal Ceftriaxone Dosing

TDD for each patient was compared with Lexicomp’s RDD to identify suboptimal dosing by indication (Figure 1). Nine of 20 indications had explicit dosing recommendations in Lexicomp. For the remaining 11 indications, Lexicomp’s general dosing guideline (50-75 mg/kg/day) was used. Sepsis/SBI did not have explicit dosing recommendations in Lexicomp. Many UTI dosages were above the recommended range (51/127, 40.2%). Dosing below the recommended range occurred most frequently in the treatment of CNS infections (35/151, 23.2%).

Figure 1.

Dot plot of ceftriaxone total daily dosage (mg/kg/day) by treatment indication. Each dot represents one prescription. Red dots represent dosages out of Lexicomp’s recommended total daily dosage range [9]. Indications are ordered from most to fewest prescriptions. Abbreviations: LRTI, lower respiratory tract infection; SBI, serious bacterial infection; CNS, central nervous system; UTI, urinary tract infection; GI, gastrointestinal; CRBSI, catheter-related bloodstream infection; URI, upper respiratory infection; SSTI, skin/soft tissue infection; AOM, acute otitis media; FUO, fever of unknown origin. The indications “Other” and “Unknown” are excluded from this figure.

DISCUSSION

Ceftriaxone was the second most prescribed antibiotic across 32 US children’s hospitals, but ceftriaxone was inconsistently dosed. The largest variability occurred for prescriptions for sepsis/SBI and a large number of prescriptions were outside of Lexicomp’s guidelines for CNS infection. Minimizing suboptimal prescriptions could reduce antibiotic exposure at both the individual patient level (leading to fewer adverse events including treatment failures) and the population level (reducing the development and spread of antimicrobial resistance). Some cases of suboptimal dosing may be prevented through clinician reminders of indication-specific dosing and electronic health record decision support.

The dosing generally recommended for CNS infection is 100 mg/kg/day [9, 10]. In our study, 23% of prescriptions were below 100 mg/kg/day, even after allowing for a 20% margin of TDD and after excluding patients who received maximum daily doses that would make their TDD falsely appear low. Subtherapeutic dosing is theoretically dangerous for infections at this site due to the need for high levels of free (not protein-bound) ceftriaxone to accumulate in the CNS [11]. While our dataset does not allow for assessment of clinical outcomes, these patients may have experienced undue sequelae secondary to inappropriately low dosing.

Sepsis/SBI, the indication with the most dosing variability, does not have a recommended dose in Lexicomp. This wide variability and a lack of recommendation for optimal dosing suggest an opportunity to better understand dosing in these situations. Minimizing the instances of supratherapeutic ceftriaxone dosing may limit the number of adverse drug reactions such as liver injury and hematologic abnormalities [12]. Further research is needed on optimal ceftriaxone dosing for sepsis/SBI.

This study had several limitations. First, this database only allowed for one documented indication per prescription, but clinicians may have been prescribing ceftriaxone to cover for more than one indication. For example, children receiving ceftriaxone for sepsis/serious bacterial infection may have received a high dose classified as “supratherapeutic” because the clinician could have been covering for meningitis. However, since data were reviewed and entered by infectious diseases clinical experts, we think it is reasonable to presume in most instances that the indication with the highest dosing was selected by the reviewer. Second, these data were manually entered so data entry errors were possible; we made efforts to exclude such errors. Additionally, though Lexicomp is widely used, clinicians may employ other clinical reference materials (eg, other tertiary reference compendiums, clinical practice guidelines, and primary literature) for guidance on dosing decisions. Next, pathogen-specific CNS infections (ie, gonococcal meningitis or Lyme meningitis) have a recommended dose of 50–75 mg/kg/day but our dataset did not allow us to identify these subtypes of CNS infections. Finally, patients who received the maximum TDD may have received a therapeutic dose, but they were excluded from the variability analysis.

CONCLUSION/SUMMARY

Ceftriaxone dosing is variable across US children’s hospitals. Large dosing variability was observed for sepsis/serious bacterial infection. Dosages for CNS infection were often suboptimal compared with Lexicomp’s recommendations. Future efforts to establish and standardize optimal ceftriaxone dosing may decrease suboptimal and variable ceftriaxone dosing.

Notes

Acknowledgments. Thank you to clinical pharmacists Miranda Nelson, PharmD, and Sarah Jones, PharmD, and SHARPS Collaborative members.

Financial support. This work was supported by Merck, which provided funding for the development of this point prevalence survey study. J. C. H. was supported by an National Institutes of Health training grant from the National Library of Medicine (grant number T15LM007092; Principal Investigators: Alexa T. McCray, PhD & Nils Gehlenborg PhD).

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.