Antibiotic treatment duration for bloodstream infections in critically ill children-A survey of pediatric infectious diseases and critical care clinicians for clinical equipoise.

OBJECTIVE: To describe antibiotic treatment durations that pediatric infectious diseases (ID) and critical care clinicians usually recommend for bloodstream infections in critically ill children.
DESIGN: Anonymous, online practice survey using five common pediatric-based case scenarios of bloodstream infections.
SETTING: Pediatric intensive care units in Canada, Australia and New Zealand. PARTICIPANTS: Pediatric intensivists, nurse practitioners, ID physicians and pharmacists. MAIN OUTCOME MEASURES: Recommended treatment durations for common infectious syndromes associated with bloodstream infections and willingness to enrol patients into a trial to study treatment duration.
RESULTS: Among 136 survey respondents, most recommended at least 10 days antibiotics for bloodstream infections associated with: pneumonia (65%), skin/soft tissue (74%), urinary tract (64%) and intra-abdominal infections (drained: 90%; undrained: 99%). For central vascular catheter-associated infections without catheter removal, over 90% clinicians recommended at least 10 days antibiotics, except for infections caused by coagulase negative staphylococci (79%). Recommendations for at least 10 days antibiotics were less common with catheter removal. In multivariable linear regression analyses, lack of source control was significantly associated with longer treatment durations (+5.2 days [95% CI: 4.4-6.1 days] for intra-abdominal infections and +4.1 days [95% CI: 3.8-4.4 days] for central vascular catheter-associated infections). Most clinicians (73-95%, depending on the source of bloodstream infection) would be willing to enrol patients into a trial of shorter versus longer antibiotic treatment duration.
CONCLUSIONS: The majority of clinicians currently recommend at least 10 days of antibiotics for most scenarios of bloodstream infections in critically ill children. There is practice heterogeneity in self-reported treatment duration recommendations among clinicians. Treatment durations were similar across different infectious syndromes. Under appropriate clinical conditions, most clinicians would be willing to enrol patients into a trial of shorter versus longer treatment for common syndromes associated with bloodstream infections.

Introduction

Bloodstream infections (BSIs) cause significant morbidity and mortality in critically ill pediatric patients. In a three-year surveillance of BSIs in a pediatric intensive care unit, there were 39 cases of BSIs per 1000 admissions, and these patients experienced a crude mortality three-fold higher than uninfected patients [1]. Optimal treatment of infections involves both timely initiation of adequate treatment [2-4] as well as continuation of treatment for an appropriate duration. However, the optimal treatment duration with antimicrobials for BSIs has not been determined [5-8]. Shortened treatment durations might offer potential benefits in patient-centred outcomes including earlier hospital discharge and reduced antimicrobial adverse events, C. difficile infections, antimicrobial resistance and healthcare costs. Therefore, it is important to know whether shorter treatment durations would lead to similar clinical cure and survival. A trial of shorter versus longer treatment duration for bacteremia is underway for critically ill adults [9]. Given the paucity of evidence guiding current practices on treatment duration, a trial among pediatric patients may be warranted, in order to explore generalizability and examine pediatric-relevant outcomes.

Previous surveys of critical care and infectious diseases (ID) physicians in Canada, Australia and New Zealand have reported wide practice variation in recommended antibiotic treatment durations for five common bacteremic syndromes among adults: pneumonia, skin/soft tissue, urinary tract, intra-abdominal and central vascular catheter-associated infections. They found that the most common treatment durations recommended were 7, 10 and 14 days and most respondents would not modify their recommendations about treatment duration based on host characteristics or measures of clinical response [10,11]. Pediatric clinicians generally view infants and children as a distinct group from adults and self-reported treatment recommendations for adults may not reflect practices for infants and children. Pediatric patients can experience different disease and developmental issues, and potentially respond differently to medical therapies [2,12-14].

To examine pediatric-specific BSI treatment duration recommendations and explore potential factors that might influence clinical decisions on treatment, we conducted a survey among pediatric ID physicians, critical care clinicians and pharmacists using pediatric-based case scenarios. The objectives of this survey study were to describe the self-reported durations of antibiotic therapy that pediatric ID and critical care clinicians usually recommend for treatment of BSIs in critically ill children, and to determine if they would be willing to enrol such patients into a trial comparing shorter versus longer antibiotic treatment. We hypothesized that there would be practice heterogeneity among clinicians but that the majority of respondents are currently recommending longer treatment courses of at least 10 days for each of the common bacteremic syndromes.

Methods

Study setting and population

We conducted an anonymous, online practice survey via SurveyMonkey among pediatric ID and critical care clinicians in Canada, Australia and New Zealand between December 2020 and February 2021. Critical care clinicians and pharmacists in Canadian pediatric intensive care units were contacted by email with invitations to participate in the survey. The ID clinicians surveyed in Canada belonged to the Pediatric Investigators Collaborative Network on Infections in Canada (PICNIC), and those in Australia and New Zealand belonged to the Australia and New Zealand Paediatric Infectious Diseases Group (ANZPID) of the Australasian Society of Infectious Diseases (ASID). Respondents were provided with a link to the survey webpage for further study information and were requested to provide informed consent on the first page before they proceeded to the next page to start answering the survey questions. The survey was conducted anonymously online, and participants could withdraw participation at any time by closing the survey prior to final submission. Continuation beyond the first page of the survey, followed by completion and submission of the survey constituted informed consent for participation. Two reminder email invitations were sent after the initial invitation to participate in the survey. Research Ethics Board approval for the survey and consent process was granted at Sunnybrook Health Sciences Centre, University of Toronto (Toronto, Canada).

Survey design

The survey was based on a survey tool previously used in a Canadian study conducted with clinicians caring for critically ill adults [10] but modified to include cases with a pediatric focus, suitable for pediatric clinicians. The survey consisted of 5 pediatric scenarios describing common infectious syndromes associated with BSIs in the pediatric intensive care unit: pneumonia, skin/soft tissue, urinary tract, intra-abdominal and central vascular catheter-associated infections. For each scenario, clinicians were asked what total duration of antibiotic therapy they would usually recommend for the patient. In the case of bacteremia involving a central vascular catheter, clinicians were asked to specify their recommendations if the causative pathogens were: Enterococcus faecalis, Staphylococcus aureus, Klebsiella pneumoniae, coagulase negative staphylococci, Escherichia coli, Enterobacter cloacae and Pseudomonas aeruginosa. For every case scenario, survey respondents were also asked whether they would be willing to enrol a patient like the one described in the case into a study comparing 7 versus 14 days of antibiotic therapy. After the 5 pediatric scenarios, respondents were provided a list of patient outcome measures and asked to indicate which ones they considered important trial outcomes that would influence their antibiotic treatment practices for critically ill children with BSIs. They were also asked whether they would be willing to shorten their treatment duration recommendations based on evidence extrapolated from an adult trial demonstrating that the mortality rate in adults receiving shorter duration therapy was no worse than adults receiving longer duration therapy.

The survey was pilot-tested by 2 ID physicians, 3 critical care physicians, 1 critical care nurse practitioner and 1 critical care pharmacist. They assessed the flow, acceptability and ease of administration of the survey tool, and the clarity and interpretation of the questions [15]. After pilot-testing, two additional questions regarding the existence and participation in antimicrobial stewardship programs (ASP) were added. The final survey included 5 scenarios and 33 items (S1 File).

Sample size

The target sample size for the survey was 97 respondents to allow a 95% two-sided confidence interval to extend ±10% around an expected proportion of 50% of respondents that would recommend a longer course (e.g. ≥10 days) versus a shorter course of antimicrobial therapy for any specific infectious syndrome (α = 0.05) [10,16].

Statistical analyses

We described recommended antibiotic treatment durations using frequencies and percentages, medians and interquartile ranges and graphical displays of the responses. We defined longer treatment duration as ≥10 days of antimicrobial therapy [17]. Univariate analyses (Wilcoxon Rank Sum Test for 2 groups and Kruskal-Wallis Test for >2 groups) were performed to determine if clinician characteristics (practice specialty, country of practice, years of experience and ASP activity) were associated with recommended treatment durations. The willingness of clinicians to enrol patients into a trial of 7 versus 14 days of antimicrobials, and to shorten their treatment duration recommendations based on evidence extrapolated from an adult trial were described using frequencies and percentages. Clinical trial outcomes that would influence clinicians’ practices on treatment duration were counted and graphically illustrated. We conducted multivariable linear regression analyses with treatment duration for each bacteremia syndrome as the dependent variable and clinician specialty, country of practice, years since graduation and ASP at practice site as independent variables. For the intra-abdominal infection scenario, source control was included as an additional predictor variable. For the central vascular catheter-associated infection scenario, catheter removal and pathogen type were additional predictor variables. Statistical analyses were conducted using SAS statistical software version 9.4M6 (SAS Institute, Cary, NC) and R version 4.0.2.

Results

Characteristics of survey respondents

There were 136 survey respondents, including 77 (57%) critical care clinicians (intensivists, fellows and nurse practitioners), 32 (24%) ID physicians, 20 (15%) critical care pharmacists and 7 (5%) clinicians with combined training in critical care, ID, pediatrics, pharmacy, surgery, cardiology or clinical microbiology. The overall response rate was 26%—with higher response rates among pharmacists (51%) and critical care clinicians (50%) and lower response rates among ID physicians (16%). All respondents practiced in academic institutions and 97% treated only pediatric patients. The respondents had a broad range of clinical experience: ≤5 years (7%), to 6–10 years (24%), 11–15 years (23%), 16–20 years (22%) and ≥21 years (24%) of clinical practice. Half (50%) of the respondents reported managing more than 20 cases of BSIs per year. Most respondents were clinicians who were currently practicing in Canada (85%), the remainder were practicing in Australia and New Zealand (15%). There was an ASP program established in 75% of respondents’ institutions. Among those with ASP programs, 42% of respondents reported being an active member of the ASP team.

Treatment duration recommendations

The majority of clinicians recommended at least 10 days of antibiotics for bacteremia associated with each of the infectious syndromes: pneumonia (84/129, 65%), skin/soft tissue (80/108, 74%), urinary tract (68/107, 64%), intra-abdominal (drained) (97/108, 90%) and intra-abdominal (undrained) (107/108, 99%). The most common treatment durations recommended for most infections were either 7, 10 or 14 days (full distributions displayed in Figs 1 and 2). Treatment durations of 21 days or longer were sometimes recommended for undrained intra-abdominal infections and central vascular catheter-associated infections without catheter removal.

Fig 1

Violin plot distribution of recommended treatment durations for bacteremia from different sources.

The shaded area represents the density of responses for treatment duration. Inside each violin, the thick line marks the median, the boxes indicate the interquartile range (IQR) and the lines mark 1.5xIQR for the treatment duration recommendation.

Fig 2

Violin plot distribution of recommended treatment durations for central vascular catheter-related bacteremia caused by different pathogens.

For central vascular catheter-associated BSIs where the catheter was not removed, over 90% of clinicians recommended at least 10 days of antibiotics for all pathogens, except for coagulase negative staphylococci (86/109, 79%). When the infected catheter was removed, 64 to 78% (depending on the pathogen) of clinicians still recommended at least 10 days of antibiotics irrespective of the causative pathogen, except for E. faecalis (53/110, 48%) and coagulase negative staphylococci (35/110, 32%).

Table 1 summarizes the overall recommended treatment durations by infectious syndrome. There were significant differences in median recommended treatment durations depending on whether intravenous catheters were removed or not (10 [IQR 7–14] days vs. 14 [IQR 14–14] days) and whether intra-abdominal infections were drained or only partially/not drained (14 [IQR 10–14] days vs. 21 [IQR 14–21] days).

Table 1

Recommended treatment duration (days) by infectious syndrome.

Infectious syndrome	n	Median	IQR	Range
Pneumonia
	129	10	7–10	7–14
Skin/soft tissue
	108	10	7.5–14	7–21
Urinary tract
	107	10	7–14	7–14
Intra-abdominal a
Drained	108	14	10–14	5–21
Partial/not drained	108	21	14–21	7–30
Central vascular catheter-associated b
Catheter removed (overall)	110	10	7–14	0–28
Enterococcus faecalis		7	7–10	2–14
Staphylococcus aureus		10	7–14	2–28
Klebsiella pneumoniae		10	7–14	2–14
Coagulase negative staphylococci		7	5–10	0–14
Escherichia coli		10	7–14	2–14
Enterobacter cloacae		10	7–14	2–21
Pseudomonas aeruginosa		14	10–14	2–21
Catheter not removed (overall)	109	14	14–14	0–42
Enterococcus faecalis		14	10–14	3–21
Staphylococcus aureus		14	14–14	7–42
Klebsiella pneumoniae		14	14–14	7–21
Coagulase negative staphylococci		14	10–14	0–42
Escherichia coli		14	14–14	7–21
Enterobacter cloacae		14	14–14	7–28
Pseudomonas aeruginosa		14	14–14	7–42

Difference between intra-abdominal infection drained vs. partial/no drain: p<0.05 (Wilcoxon Rank Sum Test).

Difference between catheter removed (overall) vs. catheter not removed (overall): p<0.05 (Wilcoxon Rank Sum Test).

Overall, there were no significant differences in treatment durations recommended by critical care clinicians, ID physicians and pharmacists for bacteremia associated with pneumonia, skin/soft tissue, urinary tract and intra-abdominal infections (S1 Table). For central vascular catheter-associated bacteremia with source control (infected catheter removed), ID physicians recommended significantly shorter treatment durations than critical care clinicians and pharmacists for E. faecalis and coagulase negative staphylococci infections. When there was no removal of infected catheters, ID physicians recommended significantly shorter durations than critical care clinicians and pharmacists for all pathogens, except for S. aureus, K. pneumoniae and P. aeruginosa infections (Table 2).

Table 2

Median (IQR) treatment duration (days) of central vascular catheter-associated bacteremia by clinician specialty.

	Critical care (n = 68)	Infectious diseases (n = 29)	Pharmacy (n = 13)	p-valuea,b
Catheter removed (n = 110)
Enterococcus faecalis	10 (7–10)	7 (7–10)	7 (7–10)	0.004
Staphylococcus aureus	10 (7–14)	14 (7–14)	10 (7–14)	0.31
Klebsiella pneumoniae	10 (7–14)	10 (7–10)	7 (7–10)	0.31
Coagulase negative staphylococci	7 (7–10)	5 (3–7)	7 (7–7)	<0.001
Escherichia coli	10 (8.5–14)	10 (7–10)	10 (7–10)	0.2
Enterobacter cloacae	10 (10–14)	10 (7–10)	10 (7–10)	0.07
Pseudomonas aeruginosa	14 (10–14)	10 (7–14)	10 (7–14)	0.21
Catheter not removed (n = 109)
Enterococcus faecalis	14 (14–14)c	10 (10–14)	14 (10–14)	<0.001
Staphylococcus aureus	14 (14–21)c	14 (14–14)	14 (14–21)	0.55
Klebsiella pneumoniae	14 (14–14)c	14 (10–14)	14 (14–14)	0.03
Coagulase negative staphylococci	14 (10–14)c	10 (7–10)	14 (10–14)	<0.001
Escherichia coli	14 (14–14)c	14 (10–14)	14 (14–14)	0.009
Enterobacter cloacae	14 (14–14)c	14 (10–14)	14 (14–14)	0.005
Pseudomonas aeruginosa	14 (14–21)c	14 (14–14)	14 (14–14)	0.04

Kruskal-Wallis Test.

Bonferroni adjusted p-value threshold = 0.02.

Missing = 1.

Recommendations for treatment durations for the various bacteremia syndromes in the survey scenarios were generally similar between clinicians based on years of clinical practice, the number of BSIs managed per year and whether an ASP program was established at practice sites or not (S2–S4 Tables).

Multivariable predictors of recommended treatment durations

No association was found between clinician specialty and recommended treatment duration for the bacteremic syndromes presented in the survey scenarios, except for skin/soft tissue-related infections where critical care clinicians recommended 2.3 days (95% CI: -3.8 to -0.8 days) shorter treatment compared to ID clinicians (Table 3). For bacteremia associated with pneumonia, clinicians in Canada recommended 1.7 days (95% CI: 0.3 to 3.2 days) longer treatment than clinicians in Australia and New Zealand. Recommended treatment duration of skin/soft tissue-related bacteremia was also 3.4 days (95% CI: 1.6 to 5.2 days) longer among Canadian clinicians (Table 3).

Table 3

Multivariable regression models for predictors of recommended treatment duration for different infectious syndromes.

Predictor	Adjusted β-coefficienta	95% confidence interval	p-valueb
PNEUMONIA (Omnibus F-test: 1.28 (8, 120), p = 0.26, R-square = 0.08)
Specialty
Pharmacist	-0.8	-2.3 to 0.6	0.26
Critical care	-1	-2.2 to 0.2	0.09
Infectious diseases	Reference	--	--
Country
Canada	1.7	0.3 to 3.2	0.02
Australia/New Zealand	Reference	--	--
Years since graduation
0–5	-0.6	-2.4 to 1.1	0.46
6–10	0.6	-0.6 to 1.9	0.32
11–15	-0.1	-1.4 to 1.1	0.85
16–20	0.3	-0.9 to 1.6	0.58
≥21	Reference	--	--
Antimicrobial stewardship program
Yes	-0.3	-1.3 to 0.7	0.55
No	Reference	--	--
SKIN/SOFT TISSUE (Omnibus F-test: 2.42 (8, 99), p <0.05, R-square = 0.16)
Specialty
Pharmacist	0.1	-1.8 to 2	0.91
Critical care	-2.3	-3.8 to -0.8	0.004
Infectious diseases	Reference	--	--
Country
Canada	3.4	1.6 to 5.2	<0.001
Australia/New Zealand	Reference	--	--
Years since graduation
0–5	0.4	-2.1 to 3	0.73
6–10	-0.4	-1.9 to 1.2	0.65
11–15	-0.5	-2 to 1.1	0.54
16–20	0.1	-1.5 to 1.6	0.92
≥21	Reference	--	--
Antimicrobial stewardship program
Yes	0.2	-1.1 to 1.5	0.77
No	Reference	--	--
URINARY TRACT (Omnibus F-test: 0.65 (8, 98), p = 0.74, R-square = 0.05)
Specialty
Pharmacist	-1	-2.9 to 0.9	0.31
Critical care	-1.1	-2.6 to 0.4	0.15
Infectious diseases	Reference	--	--
Country
Canada	0.8	-1 to 2.6	0.39
Australia/New Zealand	Reference	--	--
Years since graduation
0–5	-0.7	-3.3 to 1.8	0.56
6–10	0.7	-0.9 to 2.2	0.4
11–15	-0.4	-1.9 to 1.2	0.64
16–20	0.1	-1.4 to 1.7	0.88
≥21	Reference	--	--
Antimicrobial stewardship program
Yes	-0.6	-1.9 to 0.8	0.39
No	Reference	--	--
INTRA-ABDOMINAL (AIC = 1187.6)
Specialty
Pharmacist	-2.2	-4.6 to 0.3	0.08
Critical care	-0.8	-2.6 to 1.1	0.43
Infectious diseases	Reference	--	--
Country
Canada	1.3	-1 to 3.6	0.26
Australia/New Zealand	Reference	--	--
Years since graduation
0–5	-0.04	-3.2 to 3.1	0.98
6–10	-0.8	-2.7 to 1.2	0.45
11–15	-1.4	-3.3 to 0.5	0.15
16–20	-0.3	-2.3 to 1.6	0.73
≥21	Reference	--	--
Antimicrobial stewardship program
Yes	-0.5	-2.2 to 1.2	0.57
No	Reference	--	--
Source control
Partial/no drainage	5.2	4.4 to 6.1	<0.001
Drained	Reference	--	--
CENTRAL VASCULAR CATHETER-ASSOCIATED (AIC = 7991.9)
Specialty
Pharmacist	0.6	-1 to 2.3	0.45
Critical care	1.1	-0.2 to 2.3	0.09
Infectious diseases	Reference	--	--
Country
Canada	1.4	-0.2 to 2.9	0.08
Australia/New Zealand	Reference	--	--
Years since graduation
0–5	2.6	0.4 to 4.7	0.02
6–10	1.2	-0.1 to 2.5	0.08
11–15	1.1	-0.2 to 2.4	0.1
16–20	1.4	0.1 to 2.7	0.03
≥21	Reference	--	--
Antimicrobial stewardship program
Yes	-0.5	-1.6 to 0.6	0.39
No	Reference	--	--
Source control
Catheter not removed	4.1	3.8 to 4.4	<0.001
Catheter removed	Reference	--	--
Pathogen
E. faecalis	-2.4	-3 to -1.9	<0.001
S. aureus	-0.2	-0.7 to 0.4	0.59
K. pneumoniae	-1.6	-2.2 to -1.1	<0.001
Coagulase negative staphylococci	-3.4	-4 to -2.8	<0.001
E. coli	-1.3	-1.9 to -0.7	<0.001
E. cloacae	-1	-1.6 to -0.4	0.001
P. aeruginosa	Reference	--	--

Adjusted β-coefficient for each predictor variable represents the change in number of days of antimicrobial therapy for the variable relative to the reference predictor variable.

Bonferroni adjusted p-value thresholds = 0.02 for ‘Specialty’, 0.005 for ‘Years since graduation’, 0.002 for ‘Pathogens’.

Treatment durations were 5.2 days (95% CI: 4.4 to 6.1 days) longer for undrained intra-abdominal infections and 4.1 days (95% CI: 3.8 to 4.4 days) longer for central vascular catheter-associated infections without catheter removal. The type of pathogen in central vascular catheter-associated BSIs was also associated with treatment duration recommendations (Table 3).

Willingness to enrol patients into a trial of shorter versus longer treatment duration

Most clinicians would be willing to enrol patients with bacteremia into a trial of 7 versus 14 days of antibiotics, when the bacteremia is associated with pneumonia (109/129, 84%), skin/soft tissue (85/108, 79%), urinary tract (97/107, 91%), intra-abdominal infections (drained) (79/108, 73%) or central vascular catheter-associated infections (105/110, 95%). For this later scenario, 61% (64/105) would be willing regardless of catheter removal, 35% (37/105) would be willing only if the catheter was removed and 4% (4/105) would be willing only if the catheter was not removed.

The majority of clinicians who were willing to enrol patients into a trial of 7 versus 14 days of antibiotics made recommendations for long treatment durations of at least 10 days for bacteremia in the case scenarios: pneumonia (79/109, 72%), skin/soft tissue (64/85, 75%), urinary tract (64/97, 66%) and drained intra-abdominal infections (70/79, 89%). For central vascular catheter-associated infections with catheter removal, depending on the pathogen, 63 to 77% of clinicians who would enrol patients in a trial recommended at least 10 days of antibiotics for all the pathogens listed (S5 Table), except for E. faecalis (48/105, 46%) and coagulase negative staphylococci (30/105, 29%). Without catheter removal, over 90% of clinicians who would enrol patients in a trial had recommended at least 10 days of antibiotics for various pathogens, except for coagulase negative staphylococci (81/104, 78%) (S5 Table).

Clinical trial outcomes in pediatrics and extrapolation of adult data

Fig 3 shows that more than 50% of respondents indicated that measures of organ dysfunction, time to clinical stability, length of stay in the intensive care unit, duration of mechanical ventilation and length of hospital stay were important outcome measures in a pediatric trial. In contrast, neurodevelopmental outcomes, functional status outcomes, measures of frailty and quality of life were selected by fewer than 25% of respondents. Most clinicians indicated that they would be willing to extrapolate results from an adult trial comparing 7 versus 14 days of antibiotic treatment for bacteremia to their own pediatric practices (84/107, 79%).

Fig 3

Outcomes in pediatric trials that could influence antibiotic treatment practices.

Discussion

In our survey of pediatric ID clinicians, critical care clinicians and pharmacists in Canada, Australia and New Zealand, we found that at least 60% of clinicians would recommend 10 days or longer of antibiotics when treating bacteremia associated with pneumonia, skin/soft tissue, urinary tract, intra-abdominal and central vascular catheter-associated infections. The recommended treatment durations and variability were similar across different infectious syndromes and between clinician specialties, supporting implicit collective equipoise for a study of shorter versus longer duration treatment for BSIs in critically ill children. Pending availability of evidence-based criteria to guide treatment duration of BSIs in children, most clinicians indicated that they would be willing to apply adult trial data to their pediatric practices. Under appropriate clinical conditions, most clinicians would be willing to enrol patients into a trial of 7 versus 14 days of antimicrobial treatment for different bacteremic syndromes. Interestingly, the majority of clinicians who would enrol patients into such a trial also recommended longer antibiotic treatment durations for at least 10 days in the survey scenarios. This suggests that they recognize that shorter duration therapy could be appropriate for some patients, despite their current self-reported practices. We hypothesize that this dissonance exists because previous guidelines relating to antibiotic management of bacteremic syndromes endorsed these longer durations [18,19]; however, recent evidence has demonstrated that shorter-duration therapy is preferable for a number of different infectious syndromes [8,20-25]. In addition, earlier switch from intravenous to suitable oral antibiotics with patient defervescence and clinical improvement is another potential strategy to optimize treatment duration and reduce antimicrobial overuse [8]. Clinicians may realize that their current practices are overly prolonged but need evidence to support modifications to management guidelines. Our results demonstrate explicit support for equipoise to enrol pediatric patients into a trial that could potentially help to decrease unnecessary antimicrobial use and prevent harms associated with antibiotic overuse. While outcomes like organ dysfunction, clinical stability, length of stay and duration of mechanical ventilation are objective trial outcome measures that could influence clinician practices, additional outcomes like neurodevelopmental milestones, functional status, quality of life and frailty measures that are important to patients and their families should also be considered.

The variability in antibiotic duration recommendations found in our survey is similar to patterns found among adult ID and critical care clinicians [10,11]. However, in contrast to the adult results, our survey of pediatric clinicians did not detect a difference between ID and critical care clinicians in their recommendations for treatment durations for bacteremia associated with pneumonia, skin/soft tissue, urinary tract and intra-abdominal infections. Recommendations by pediatric pharmacists were also similar to ID and critical care clinicians.

In our regression models of treatment duration recommendations for bacteremia associated with pneumonia, skin/soft tissue and urinary tract infections, we found that clinicians in Australia/New Zealand recommended significantly shorter treatment durations for BSIs associated with pneumonia and skin/soft tissue infections compared to Canadian clinicians. However, the overall number of respondents in our subgroups were small and this difference among countries may be confounded by survey respondents in Australia and New Zealand being primarily ID clinicians. Lack of source control could increase the potential for continued seeding of infection—this was a significant predictor of longer treatment duration for intra-abdominal and central vascular catheter-associated bacteremia, as was the type of causative pathogen in catheter-related infections. The presence of an institutional antimicrobial stewardship program was not associated with a significant difference in treatment duration recommendations made by survey respondents.

Limitations of this survey study include the overall low response rate and inherent biases of self-reporting in survey studies (as compared to actual practice). Survey responses on self-reported practices may not be an accurate reflection of actual clinical practice. We also acknowledge that only ID clinicians were surveyed in Australia and New Zealand via ANZPID and the ID physicians surveyed in Canada all belonged to PICNIC so respondents may not be representative of all ID clinicians in pediatric practices. Furthermore, three quarters of survey respondents had established ASP programs at their institutions and almost half of them were active members of ASP teams. Therefore, participants in our survey might have been more inclined to support ASP-related initiatives like reducing unnecessary antibiotic use and the actual median treatment durations that would be recommended by ID and critical care clinicians could be longer than what we detected.

Conclusions

The results of this survey study confirm that there is variability in recommended antibiotic treatment durations for critically ill pediatric patients with BSIs, but most clinicians recommend treating for 10 days or longer. While pediatric ID and critical care clinicians would consider extrapolating results from an adult trial, we also demonstrated both implicit and explicit clinical equipoise among clinicians to enrol their own pediatric patients into a trial of 7 versus 14 days of antibiotic treatment for bacteremia. Ultimately, a generalizable trial of shorter versus longer antibiotic treatment is likely needed to inform treatment practices, so that the benefits of antibiotic therapy can be optimized while the harms are minimized for critically ill children.

Survey document.

(PDF)

Click here for additional data file.

Median (IQR) treatment duration (days) by clinician specialty.

(DOCX)

Click here for additional data file.

Median (IQR) treatment duration (days) by years since graduation.

(DOCX)

Click here for additional data file.

Median (IQR) treatment duration (days) by number of bloodstream infections managed per year.

(DOCX)

Click here for additional data file.

Median (IQR) treatment duration (days) by presence of institutional antimicrobial stewardship program.

(DOCX)

Click here for additional data file.

Respondents who were willing to enrol patients into a trial of 7 versus 14 days of antimicrobial therapy and recommended at least 10 days of antimicrobial therapy for central vascular catheter-associated infections in case scenarios.

(DOCX)

Click here for additional data file.

Comparison of respondents recommending longer (≥10 days) treatment durations between infectious syndromes.

(DOCX)

Click here for additional data file.

(XLSX)

Click here for additional data file.

19 May 2022

3 Jul 2022

Dear Dr. Vadivelu,

Thank you for your consideration of our manuscript, entitled “Antibiotic treatment duration for bloodstream infections in critically ill children—a survey of pediatric infectious diseases and critical care clinicians for clinical equipoise” for publication in PLOS ONE. We are pleased to hear that your Editorial team believes our manuscript is of interest to your readership. We greatly appreciated your feedback and reviewer comments and have strived to incorporate the suggestions into the manuscript. We believe the paper has been strengthened in the process. The detailed responses to the Editor and Reviewers’ comments are provided below.

We look forward to hearing from you.

Sincerely,

Sandy Pong

Corresponding Author:

Sandra Pong, PharmD

Clinical Pharmacist

The Hospital for Sick Children

Department of Pharmacy

555 University Avenue

Toronto, Ontario M5G 1X8

Canada

sandra.pong@sickkids.ca

Journal Requirements:

In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found.

We have added our dataset to “Supplementary materials” (S7 Dataset).

We note that you have included the phrase “data not shown” in your manuscript. Unfortunately, this does not meet our data sharing requirements. PLOS does not permit references to inaccessible data. We require that authors provide all relevant data within the paper, Supporting Information files, or in an acceptable, public repository.

We have added these data to “Supplementary materials” (S2-S6 Tables).

Please review your reference list to ensure that it is complete and correct. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

We have reordered and added additional references according to changes in the main body of the manuscript. No references were deleted or changed.

Reviewer #1: Review of article entitled “Antibiotic treatment duration for bacteremia in critically ill children--a survey of pediatric infectious diseases and critical care clinicians for clinical equipoise” by S. Pong et al.

Major comments

Interesting study. The project is well designed and the results are presented clearly. The use of violin plots is judicious. I have the following comments to help improve the manuscript.

1. Abstract: could we indicate how many scenarios were presented? Perhaps mentioning that the survey focused on 5 common pediatric scenarios.

We have revised this sentence to say “…five common pediatric-based case scenarios of bacteremia.” (page 2, line 47-48)

2. Respondents: could an individual answer the survey twice?

To maintain anonymity of survey participants, we did not track individual respondents and could not identify if an individual answered the survey more than once, but we do not expect that a survey respondent would complete this survey multiple times.

3. Invitation to participate: were reminders to participate sent to potential participants to improve response rate?

Yes, two reminder email invitations were sent to participants after the initial invite. We have added this clarification to the Methods section. (page 6, line 124-125)

4. Could people forward the survey to other potential respondents? This can sometimes bias the response rate.

We did not request participants to forward the survey to other potential respondents. There were instances when invitees inquired if they could forward the survey to colleagues. In those cases, we asked if they could let us know the number of additional colleagues who they were going to send the survey to, or if they could send us their email addresses so that we could contact them directly with an invitation to complete a survey. However, there was no mechanism in place that would prevent someone from forwarding the survey without our knowledge.

5. Analyses: From the survey, I understand that the answers regarding the number of days are continuous numerical values (i.e. respondents indicated the number of days they would recommend), not a multiple-choice question (e.g. 7, 10, 14 days). Consequently, I’m not sure to understand why the main outcome is presented as “the proportion of respondents who recommend >=10 days of antibiotics” (page 8, line 187). Presenting the results in this manner would be expected if the question had been a multiple choice question, but considering the type of data available, I would expect these results to be presented as a median number of days with IQR. Saying that “65% recommend at least 10 days of treatment for pneumonia” is not the same (and not as informative) as saying that the median duration is 10 days (IQR, 7-10). Actually, the former sentence gives the impression that the treatment duration recommendation is longer than it really is (looking at the violin chart, 50% of respondents recommend 10 days or less of treatment). This comment would also apply to the abstract.

We used ≥10 days as a cut-off to arbitrarily define longer duration treatment, which has also been done in previous work by our group in adult populations. In our analyses of whether respondents would be willing to enrol patients into a trial of shorter vs. longer antibiotic treatment, we also used this dichotomy to determine the proportion who had also recommended ‘longer’ treatment durations in the case scenarios.

We feel this dichotomization also facilitates the cross-correlation analyses that were suggested by Reviewer #1 to assess if recommendations for longer treatment duration for a given infectious syndrome are associated with longer treatment duration for another infectious syndrome.

We have added that we defined ‘longer duration treatment” as ≥10 days and referenced the previous use of this in Daneman et al. Crit Care Med 2016;44:256-264. (page 8, line 161-162)

6. Analyses: the authors state that they conducted a multivariable linear regression analysis using specialty, country, years of practice and ASP practice as independent variables. However, when I look at table 3 and read the discussion section (page 18, lines 329-332), I have the impression that only bivariate (univariate) analyses were performed. Indeed, conducting multiple univariate analyses is not the same thing as conducting a multivariate analysis. Please clarify whether multiple variables were entered simultaneously into a single model to identify independent predictors after adjusting for covariates.

Yes, the multiple variables (specialty, country, years since graduation and antimicrobial stewardship program) were entered simultaneously into single models for each infectious syndrome (pneumonia, skin/soft tissue, urinary tract, intra-abdominal and central vascular catheter). In the case of intra-abdominal infections, we included source control (partial/not drained vs. drained) as an additional variable in the model. In the case of central vascular catheter-associated infections, we included source control (catheter removed vs. not removed) and pathogen type as additional variables in the model. These models are summarized in Table 3.

We have clarified this by specifying these are multivariable regression models in the heading of Table 3. “Table 3. Multivariable regression models for predictors of recommended treatment duration for different infectious syndromes.” (page 14, line 261-262)

7. As an additional analysis, I would be curious to know whether there is cross-correlation in terms of treatment duration recommendation for a given respondent. For example: is a respondent who recommend a longer treatment for a given infection also more likely to recommend a longer treatment for another type of infection? Identifying cross-correlation would allow to identify subgroups of “long-prescribers” and “short-prescribers”.

We have added “Supplement Table 6 (S8 Table)” showing pairwise comparisons of respondents’ recommendations for longer treatment duration for bacteremia due to pneumonia, skin/soft tissue infection, urinary tract infection, intra-abdominal infection (drained, partial/not drained) and central venous catheter-associated infection, due to E. coli, as an example organism (catheter removed, catheter not removed).

8. Table 3. A footnote to explain how to interpret the B-coefficient would be helpful.

We have added a footnote to Table 3: “Adjusted �  -coefficient for each predictor variable represents the change in number of days of antimicrobial therapy for the variable relative to the reference predictor variable.” (page 16, line 264-265)

9. Considering the small sample size and the multiple comparisons that were conducted between groups, I wonder whether the p-value should be adjusted for multiple comparisons and be lower than 0.05. This could prevent the identification of spurious findings.

We have made Bonferroni correction to the p-values in Tables 2 and 3.

Table 2—in the comparison of treatment duration between specialities, the adjusted p-value threshold is 0.02. (page 12, line 244)

Table 3—in the multivariable regression models: for ‘Specialty’, the adjusted p-value threshold is 0.02, for ‘Years since graduation’ the adjusted p-value threshold is 0.005 and for ‘Pathogens’, the adjusted p-value threshold is 0.002. (page 16, line 266)

Reviewer #2: In this study, Dr. Pong and colleagues aimed to understand typical practice patterns surrounding duration of therapy for bacteremia in pediatric patients. They designed an online survey that was administered to pediatric intensivists, nurse practitioners, ID physicians, and pharmacists. They report interesting data that should motivate RCTs in this area. Some minor comments below.

Abstract

Lines 59-60: This is a little unclear. I think that it would be helpful avoid the nested parentheses and perhaps use brackets instead. Also, make it more clear that the first set of numbers refers to lack of source control with intra-abdominal infections and second set of numbers refers to patients with central line infections

We have revised the sentence to clarify: “In multivariable linear regression analyses, lack of source control was significantly associated with longer treatment durations (+5.2 days [95% CI: 4.4-6.1 days] for intra-abdominal infections and +4.1 days [95% CI: 3.8-4.4 days] for central vascular catheter-associated infections).” (page 2, line 60-63)

Lines 60: It would be helpful to clarify why there is a range here. Perhaps “73-95%, depending on source of bacteremia”

We have revised the sentence: “Most clinicians (73-95%, depending on the source of bacteremia) would be willing to enrol patients into a trial of shorter versus longer antibiotic treatment duration.” (page 2, line 63-65)

Lines 63-64: I don’t think that this concept of implicit versus explicit equipoise is going to be familiar to most readers and so it either warrants a brief explanation or just remove it. It would be important to point out how the results presented in the abstract support this statement.

We have removed references to implicit and explicit equipoise in the abstract and replaced with: “There is practice heterogeneity in self-reported treatment duration recommendations among clinicians. Treatment durations were similar across different infectious syndromes. Under appropriate clinical conditions, most clinicians would be willing to enrol patients into a trial of shorter versus longer treatment for common syndromes associated with bloodstream infections.” (page 3, line 67-71)

Introduction:

Why transition in terminology between bacteremia (Abstract) and then BSIs (the rest)?

We have replaced ‘bacteremia’ with ‘bloodstream infections’ in the abstract and the manuscript title.

Results:

Lines 188-190: Why the different denominators for each infectious syndrome, and why none equal to 136? Participants had the option of not answering questions?

The different denominators were the result of a few participants not answering every question.

Table 3: It would be helpful to note that the beta-coefficient and CI represent days of therapy

We have added clarification as a footnote to Table 3: “Adjusted �  -coefficient for each predictor variable represents the change in number of days of antimicrobial therapy for the variable relative to the reference predictor variable.” (page 16, line 264-265)

Discussion:

Lines 327-335: Could consider omitting this paragraph. Identifying differences in treatment duration between the various subgroups is challenging given the overall low numbers in the subgroups and the different compositions of the subgroups (e.g., more ID providers in the New Zealand/Australia group).

Although we had small overall numbers, we feel that it is worthwhile to mention the subgroups with respect to the potential problems of confounding given that respondents in Australia and New Zealand were primarily ID clinicians.

We have added a statement to acknowledge that the overall numbers in the subgroups are small: “However, the overall number of respondents in our subgroups were small and this difference among countries may be confounded by survey respondents in Australia and New Zealand being primarily ID clinicians.” (page 19, line 345-347)

Reviewer #3: This manuscript nicely describes the variability in AB prescription patterns for SBI in children. As expected, there is a wide range of treatment duration as beautifully showed in the violin plots, but the general duration is long (10-14 days) and to me always intriguing, durations are usually 5, 7, 10, 14 or 21 days, never a number in between.

The manuscript is very clear and well written. One question that came into my mind is how are the prescription patterns related to the national protocols? Did authors look at national protocols for paediatric SBI and compared those protocols with the answers of the respondents? This probably would be interesting to add although there is also variability within a country.

We did not compare survey responses with national protocols. We expected significant practice variability to exist across institutions and across countries, so comparisons of national protocols may not be meaningful. There may also be greater reliance on provincial/state or local regional protocols rather than national level protocols.

The manuscript is a good base for starting of a trial into duration of SBI treatment. I am wondering if there is any place for IV-oral switch in those SBIs (see review DOI: 10.1016/S1473-3099(16)30024-X) , this is nowhere mentioned by the authors and might be addressed in the discussion section.

We have added to our discussion: “In addition, earlier switch from intravenous to suitable oral antibiotics with patient defervescence and clinical improvement is another potential strategy to optimize treatment duration and reduce antimicrobial overuse,” and referenced this systematic review and guideline. (page 18, line 323-326)

Specific comments:

The introduction is very well written and covers the most important aspects of questions concerning antibiotic use. The review of McMullan could be referred to (DOI: 10.1016/S1473-3099(16)30024-X)

We have included a reference to this systematic review and guideline. (page 4, line 79)

Line 228/229: is this sentence correct? Is should probably read longer durations after no removal of catheter?

This sentence is correct, but we agree that it is unclear as written. The meaning of “shorter” is in comparison to durations recommended by critical care clinicians and pharmacists. We have revised the sentences in this section to make the meaning clearer.

“For central vascular catheter-associated bacteremia with source control (infected catheter removed), ID physicians recommended significantly shorter treatment durations than critical care clinicians and pharmacists for E. faecalis and coagulase negative staphylococci infections. When there was no removal of infected catheters, ID physicians recommended significantly shorter durations than critical care clinicians and pharmacists for all pathogens, except for S. aureus, K. pneumoniae and P. aeruginosa infections (Table 2).” (page 12, line 233-238)

Table 3: very interesting an ASP present does not lead to differences in treatment duration - authors could highlight this in the discussion.

We have added to our discussion: “The presence of an institutional antimicrobial stewardship program was not associated with a significant difference in treatment duration recommendations made by survey respondents.” (page 19-20, line 350-352)

Submitted filename: Response_to_Reviewers.docx

Click here for additional data file.

12 Jul 2022

Antibiotic treatment duration for bloodstream infections in critically ill children—a survey of pediatric infectious diseases and critical care clinicians for clinical equipoise

PONE-D-22-04577R1

Dear Dr Pong,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Jamunarani Vadivelu

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

18 Jul 2022

PONE-D-22-04577R1

Antibiotic treatment duration for bloodstream infections in critically ill children—a survey of pediatric infectious diseases and critical care clinicians for clinical equipoise

Dear Dr. Pong:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Jamunarani Vadivelu

Academic Editor

PLOS ONE