Critically ill children often receive antibiotics. However, few data are available on the extent of antibiotic prescribing among children hospitalized in PICUs and how such prescribing differs from other hospitalized patients (1–7). Previous studies of hospitalized children reported that 57–79% of those admitted to a PICU received antibiotics (2–7), and 34% of antibiotics were broad-spectrum agents. Although patient factors including diagnosis, illness severity, and underlying diagnosis influence the decision to use antibiotics in the PICU, hospital-level factors such as patient case mix, regional patterns of antibiotic resistance, published studies, physician experience, and presence of antibiotic stewardship programs also contribute to antibiotic prescribing (1,2,8,9). Significant variability in the use of antibiotics for hospitalized children has been documented but not for children admitted to the PICU.
Physicians often overestimate the effect of therapies that they prescribe (10). In studies of hospitalized patients, it has been estimated that 20–50% of antibiotic use is unnecessary or inappropriate (6,11–16). Inappropriate antibiotic prescribing including overuse contributes to the development of antibiotic resistance and to more frequent adverse drug events, increased costs of medical care, and disturbances of the normal human microbiome (17–20). By optimizing antibiotic use, antimicrobial stewardship programs can contribute to improved patient outcomes, at lower costs (16,21). Yet, for antimicrobial stewardship programs to be most effective, detailed and accurate data on antibiotic usage are needed and may then permit patient-level analysis (e.g., treatment according to published guidelines) and institutional-level recommendations (e.g., formulary restriction) (11,16). The goal of this study was to characterize variability of antibiotic use in the PICU, potentially shedding light on inappropriate antibiotic use. A secondary goal was to compare antimicrobial use in the PICU with other hospitalized patients.
MATERIAL AND METHODS
Data Sources and Patient Population
The Pediatric Health Information System (PHIS) database (Children’s Hospital Association, Overland Park, KS) provided the data for this multicenter retrospective cross-sectional study. The PHIS database contains clinical and billing data from 48 tertiary care children’s hospitals in the United States. Participating hospitals are located in markets of 27 states and the District of Columbia. Data are deidentified prior to inclusion in the database, and quality is ensured through a joint effort between the Children’s Hospital Association and participating hospitals as described previously (22). This study has been exempted by the Children’s Mercy Hospital-Kansas City Institutional Review Board.
Hospitals that did not continuously submit PHIS data (n = 7) throughout the study period (January 1, 2010, and December 31, 2014) were excluded. Among participating hospitals, children 30 days to 18 years old were eligible for inclusion, and then antibiotic use stratified to PICU use or non-ICU use. Any infant admitted to the neonatal ICU (NICU) was excluded.
The primary exposure was hospital discharges with at least 1 day in the PICU including patients who died. All hospitalized children except those admitted to the NICU were included as a comparison group. Patients who received antibiotics after transfer from the PICU to the hospital ward were included in the group of non-ICU patients, so PICU patients could be included in both groups.
Patient-level variables identified for each hospital admission included age at admission (30 d up to 1, 1–4, 5–11, and 12–18 yr), sex, race (non-Hispanic white, non-Hispanic black, Hispanic, Asian, and other), transfers into the hospital (yes/no), and payer status (private, government, or other payer). International Classification of Diseases, 9th revision (ICD-9) diagnosis or procedure codes were used to identify surgical cases (yes/no). In addition, the presence of concurrent chronic illnesses was assessed by using a previously described and validated method for characterizing ICD-9–based pediatric complex chronic conditions, represented by 11 categories: neurologic and neuromuscular, cardiovascular, respiratory, renal and urologic, gastrointestinal, hematology and immunodeficiency, metabolic, malignancy, other congenital or genetic defect, neonatal, technology dependency, and transplantation (23).
Antibiotic use was defined by the presence of hospital billing data for any systemic antibacterial drug. Broad-spectrum antibiotics included cefepime, doripenem, imipenem, ertapenem, meropenem, ciprofloxacin, gatifloxacin, levofloxacin, moxifloxacin, piperacillin-tazobactam, ticarcillin-clavulanate, and ceftazidime. Antibiotics active against methicillin-resistant Staphylococcus aureus (MRSA) included vancomycin (IV-use only), clindamycin, daptomycin, linezolid, tedizolid, minocycline, quinupristin-dalfopristin, trimethoprim-sulfamethoxazole, and tigecycline. The dependent variables included the days of therapy (DOT) the patient received any antibiotic(s) or broad-spectrum antibiotic(s) during the PICU encounter. If a patient received two antibiotics on a single day, this was considered 2 DOT. However, for PICU patients, DOT included only those antibiotics received in the PICU antibiotics received on the hospital ward were included in non-PICU DOT.
Hospital-level characteristics were summarized using frequencies and percentages for categorical variables. Antibiotic receipt and the change in use over time were characterized among PICU and non-ICU patients during the study period. Adjusted trends in annual utilization and hospital-specific use rates (per 1,000 patient-days) were calculated using Poisson generalized linear mixed effects models, controlling for hospital clustering and allowing for the presence of correlated data (within hospitals), nonconstant variability (across hospitals), and responses that are not normally distributed (6). Quasi-likelihood estimation for proportions with unknown distributions was used to model the duration of antibiotic use, defined as the proportion of total days of an admission or total days in the PICU that a patient received antibiotic therapy for all patients and for only those patients within the PICU, respectively. These models were used to compare DOT/1,000 patient day values for PICU patients and all hospitalized patients (except those in the NICU) across institutions. In addition, comparisons of antibiotic use by institution was limited to the 20 All Patient Refined-Diagnosis Related Groups (APR-DRGs) with the highest antibiotic use by DOT in order to control for potential variability in disease severity by PICU. To further account for variation in patient populations across the various hospitals, we performed a sensitivity analysis including only seven APR-DRGs that were among the top 20 APR-DRGS for each of the units. Additionally to address potential differences in case mix, we reviewed data from the institutions that performed transplantation (lung, heart-lung, heart, hematopoietic stem cell, liver, and kidney) and evaluated the number of transplants, the median percentage of total DOT, and overall percentage of total DOT. We also assessed for correlation between hospital-level and PICU antibiotic using Pearson’s correlation coefficient. All analyses were performed using SAS v.9.4 (SAS Institute, Cary, NC) with a p value of less than 0.001 being considered statistically significant because of the large sample sizes used to conduct the analysis (386,914 patients in the PICU group and 3,101,201 hospitalized children).
During the study period from January 1, 2010, to December 31, 2014, from 41 institutions included in the analysis, 386,914 of 3,101,201 children (12.5%) discharged from a hospital stay received care in the PICU. Of these, 69.5% received at least 1 day of an antibiotic. Thirty-seven included institutions (90%) had antibiotic stewardship programs. Table 1 depicts the demographic characteristics and antimicrobial use of PICU patients. Demographic characteristics for all non-ICU patients are in Supplemental Table 1 (Supplemental Digital Content 1, http://links.lww.com/PCC/A637). In total, there were 2,191,997 PICU antibiotic DOT for 1,892,349 PICU hospital days, or 1,158.3 PICU DOT per 1,000 patient-days. Patients who underwent surgical procedures received more antibiotics including broad-spectrum agents both by percentages and by DOT than those who did not. The overall usage of antibiotics among PICU and non-ICU patients declined significantly each year during the study period as did the use of broad-spectrum antibiotics and agents active against MRSA (Fig. 1). Table 2 lists the 10 most commonly used antibiotics.
The median adjusted use of systemic antibiotics among PICU patients was 1,043 DOT/1,000 patient-days (interquartile range [IQR], 977–1,147 DOT/1,000 patient-days), and the median adjusted non-ICU use of systemic antibiotics was 893 DOT/1,000 patient-days (IQR, 805–968 DOT/1,000 patient-days). For broad-spectrum antibiotics, the median adjusted use among PICU patients was 176 DOT/1,000 patient-days (IQR, 152–217 DOT/1,000 patient-days) compared with 153 DOT/1,000 patient-days (IQR, 130–182 DOT/1,000 patient-days) for non-ICU patients. For anti-MRSA agents, the median adjusted use was 302 DOT/1,000 patient-days (IQR, 220–351 DOT/1,000 patient-days) compared with 244 DOT/1,000 patient-days (IQR, 203–270 DOT/1,000 patient-days) for non-ICU patients. Days of antibiotic therapy (overall and broad spectrum) among PICU patients showed significant variation (p < 0.001) when evaluated by institution with high consuming units using approximately twice that of the low prescribing ones; this difference was nearly three-fold for broad-spectrum antibiotics and anti-MRSA agents (Fig. 2, respectively). This significant variability (p < 0.001) also occurred in non-ICU use of systemic antibiotics (Fig. 2).
Table 3 details clinical characteristics among ICU and non-ICU patients for the 20 APR-DRGs with the highest use of antibiotics. In this subgroup, antibiotic use also varied significantly across institutions for PICU and non-ICU patients (Fig. 3A) (p < 0.001). Similar variability also occurred in the administration of broad-spectrum antibiotics and anti-MRSA antibiotics (Fig. 3, B and C, respectively) (p < 0.001). For all antibiotics, broad-spectrum agents, and anti-MRSA antibiotics, the highest prescribing PICUs used more than double that of the lowest using PICUs. The median adjusted use among PICU patients with the top 20 APR-DRGs was 1,236 DOT/1,000 patient-days (IQR, 1,115–1,366 DOT/1,000 patient-days) for all antibiotics compared with median adjusted use among non-ICU children of 959 DOT/1,000 patient-days (IQR, 887–1,061 DOT/1,000 patient-days). The median adjusted use of broad-spectrum antibiotics among PICU and non-ICU patients with these APR-DRGs were 185 DOT/1,000 patient-days (IQR, 152–220 DOT/1,000 patient-days) and 140 DOT/1,000 patient-days (IQR, 104–162 DOT/1,000 patient-days), respectively, whereas the median adjusted use of anti-MRSA antibiotics among PICU and non-ICU patients with these APR-DRGs were 347 DOT/1,000 patient-days (IQR, 259–407 DOT/1,000 patient-days) and 241 DOT/1,000 patient-days (IQR, 196–264 DOT/1,000 patient-days). When the analysis was limited to the top seven APR-DRGs shared by all 41 PICUs, the difference in the use of all antibiotics for PICUs and for all patients reached almost twofold for all antibiotics, and three fold for broad spectrum and anti-MRSA agents (data not shown). A subanalysis of transplant services by institution showed a fair amount of variability in the presence of transplant programs, but total percentage of antibiotic use was under 10% (Supplemental Table 2, Supplemental Digital Content 2, http://links.lww.com/PCC/A638).
Positive correlation was observed between antibiotic use for PICU and non-ICU patients for all antibiotics, for broad-spectrum antibiotics, and for anti-MRSA agents (R2 = 0.29, 0.55, and 0.62, respectively; p < 0.001 for both). Stronger correlation was noted between antibiotic use for PICU and non-ICU patients for all antibiotics (R2 = 0.56; p < 0.001), broad-spectrum antibiotic therapy and anti-MRSA agents, (R2 = 0.73 and 0.84, respectively; p < 0.001) for the top 20 PICU APR-DRGs.
From 2010 to 2014, antibiotic use for children admitted to PICUs of freestanding children’s hospitals in the United States declined significantly. Reductions occurred in overall, broad-spectrum, and anti-MRSA antibiotic use, mirrored by similar declines among all non-ICU hospitalized children. Despite these decreases, significant variability in prescribing of all antibiotics, broad-spectrum, and anti-MRSA agents was observed across the 41 PICUs studied; the use among the highest prescribing units was approximately twice that of those with the lowest prescribing, and this reached three-fold and four-fold differences for broad-spectrum and anti-MRSA antibiotics, respectively. This degree of variability persisted even after adjusting for clinical and demographic factors as well as in a subgroup analysis of patients with the 20 most common APR-DRGs or the seven most common APR-DRGs shared by all 41 institutions. The persistence of the high variability in each of these patient and antibiotic subanalyses makes case-mix variability unlikely to be the sole driver of these prescribing differences.
Despite the decline in antibiotic prescribing during the study period, antibiotic use in PICUs remains higher than for other hospitalized children. Critically ill children receive invasive medical technology, demonstrate hemodynamic and/or respiratory instability, have acute or chronically altered immune systems, and are exposed to resistant pathologic microorganisms. The high use of antibiotics in this population is almost certainly justified to some extent as studies have shown the importance of early institution of appropriate antibiotic therapy (8,9). In a single, point prevalence study, PICU patients (approximately 71%) were found to be more likely to receive antibiotics than those in NICUs (43%) although the latter group received a higher median number of antibiotics (2 vs 1) than the former group (2).
In addition to high rates of antibiotic use significant variability occurred across PICUs. Previous studies demonstrate that antibiotic use is influenced by a complex set of variables that includes disease severity, pathogens, radiographic and laboratory data, institutional factors, and limited evidence-based recommendations (4,24,25). To account for potential differences in case mix and disease severity by institution, we evaluated antibiotic use among children who were diagnosed among the 20 APR-DRGs with the greatest antibiotic usage by total DOT. Even in this restricted cohort, greater than two-fold differences in overall and broad-spectrum antibiotic prescribing were observed across centers. However, because some of these APR-DRGs were not treated in all units, we further restricted the analysis to the seven APR-DRGs shared by each of the 41 PICUs, excluding others (e.g., hematopoietic stem cell transplantation) that could bias the results. This subanalysis produced similar variability in antibiotic use. Additionally, organ transplant services were compared across the 41 PICUs. The differences in antibiotic use among the highest and lowest transplant services could only account for small fraction of the variability that we observed. We did not specifically examine data among extracorporeal membrane oxygenation (ECMO) patients, but discussions with a number of programs reveal that programs have differing protocols regarding prophylactic antibiotics. So, ECMO patients almost certainly contribute to variability in antibiotic use, but in most institutions represent a small patient population. Finally, we also evaluated antibiotic use by broad-spectrum agents and those active against MRSA, but similar degrees of variability were observed.
Although differences in disease severity and case-mix certainly account for some of the observed differences in antibiotic use even among these subsets of patients, a study of non-ICU hospitalized children obtained similar results in that significant variability in antibiotic use was observed in three of the four analyzed specific APR-DRG conditions (26). The authors adjusted for patient-level factors and concluded that unexplained variation in therapy may have identified areas for standardization of care and/or comparative-effectiveness research to inform stewardship recommendations.
A high degree of variability in antibiotic prescribing was also observed among all non-ICU patients. Overall and broad-spectrum antibiotic usages were correlated between PICU and non-ICU patients by institution. Again this may reflect institutional differences in case mix and disease severity, local hospital culture, or regional patterns of antibiotic resistance (4,19). A study among hospitalized children found that in children’s hospitals with high antibiotic use, patients were nearly 50% more likely to receive antibiotics, and antibiotic duration was 37% longer compared with institutions that had the lowest antibiotic use (6). Although regional differences exist in antibiotic resistance, this should primarily affect antibiotic choice rather than duration of therapy and so should have minor influence on DOT. Additionally, guidelines may help standardize duration of antibiotic courses. In fact, a single institution study showed that short course (4–5 d) monotherapy adequately treated bacteremia and may have contributed to lower rates of antibiotic resistance (27). Antibiotic stewardship programs have been associated with reductions in antibiotic use though almost all institutions in this study had such a program but variability persisted (28). The persistent variability by institution despite adjusting for differing patient diagnoses suggests that antibiotic use is determined, in part, by modifiable institutional level factors including local culture and practices.
This study has limitations. Despite examination of adjusted antibiotic use broadly and within a narrow group of APR-DRG conditions, variability in disease severity almost certainly exists across institutions even among the smaller group of APR-DRGs studied that are shared across all 41 PICUs in the study. PHIS does not contain specific markers of disease severity such as Pediatric Risk of Mortality or Pediatric Index of Mortality, which would have permitted more precise comparisons of patients across institutions. Furthermore, we could not assess if differences in prescribing rates were driven by treatment, prophylaxis, or both. For example, institutions that perform a high number of surgeries for which patients receive prophylactic antibiotics might result in a smaller number of antibiotic DOT. PHIS data are also limited by the nature of the database. We could not evaluate the hospital-level factors such as the impact of laboratory data, influence of antibiotic stewardship programs, physician mix or experience, and culture results on antibiotic therapy or duration. We did not examine specific diagnoses individually nor did we correlate usage with outcomes such as length of ICU stay or Clsotridium difficile rates. However, when limited to a narrow range of patients in a limited number of APR-DRGs with the highest antibiotic use, the variability was no less than observed among all PICU patients across the institutions studied.
Antibiotic use in PICUs across children’s hospitals varied significantly. Significant variation occurred even when the antibiotic use among a small subset of APR-DRGs was examined. Because PICU antibiotic usage is high compared with other inpatients, this variability represents a potentially rewarding target for antibiotic stewardship, and decreased use in the PICU may impact prescribing in other areas of the hospital.
1. Dassner AM, Nicolau DP, Girotto JE. Management of pneumonia in the pediatric critical care unit: An area for antimicrobial stewardship. Curr Pediatr Rev 2017; 13:4966
2. Grohskopf LA, Huskins WC, Sinkowitz-Cochran RL, et al.; Pediatric Prevention Network: Use of antimicrobial agents in United States neonatal and pediatric intensive care patients. Pediatr Infect Dis J 2005; 24:766773
3. Fischer JE, Ramser M, Fanconi S. Use of antibiotics in pediatric intensive care and potential savings. Intensive Care Med 2000; 26:959966
4. Fontela PA, Quach C, Karim ME, et al. Determinants of antibiotic tailoring in pediatric intensive care: A national survey. Pediatr Crit Care Med 2017; 18:e395e405
5. Blinova E, Lau E, Bitnun A, et al. Point prevalence survey of antimicrobial utilization in the cardiac and pediatric critical care unit. Pediatr Crit Care Med 2013; 14:e280e288
6. Gerber JS, Newland JG, Coffin SE, et al. Variability in antibiotic use at children’s hospitals. Pediatrics 2010; 126:10671073
7. Ciofi Degli Atti ML, Raponi M, Tozzi AE, et al. Point prevalence study of antibiotic use in a paediatric hospital in Italy. Euro Surveill 2008; 13:1215
8. Kumar A, Roberts D, Wood KE, et al. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med 2006; 34:15891596
9. Ferrer R, Martin-Loeches I, Phillips G, et al. Empiric antibiotic treatment reduces mortality in severe sepsis and septic shock from the first hour: Results from a guideline-based performance improvement program. Crit Care Med 2014; 42:17491755
10. Cassarrett D. The science of choosing wisely – overcoming the therapeutic illusion. N Engl J Med 2016; 374:12031205
11. Schuts ED, Hulscher ME, Mouton JW, et al. Current evidence on hospital antimicrobial stewardship objectives: A systematic review and meta-analysis. Lancet Infect Dis 2016; 16:847856
12. Davey P, Brown E, Charani E, et al. Interventions to improve antibiotic prescribing practices for hospital inpatients. Cochrane Database Syst Rev 2013; 4:CD003543
13. Hulscher MEJL, Grol RPTM, van der Meer JWM. Antibiotic prescribing in hospitals: A social and behavioural scientific approach. Lancet Infect Dis 2010; 10:167175
14. Laine N, Vaara M, Anttila VJ, et al. Evaluation of antimicrobial therapy of blood culture positive healthcare-associated infections in children. PLoS One 2015; 10:e0141555
15. Goycochea-Valdivia WA, Moreno-Ramos F, Paño-Pardo JR, et al. Identifying priorities to improve paediatric in-hospital antimicrobial use by cross-sectional evaluation of prevalence and appropriateness of prescription. Enferm Infecc Microbiol Clin 2017; 35:556562
16. Dellit TH, Owens RC, McGowan JE Jr, et al.; Infectious Diseases Society of America; Society for Healthcare Epidemiology of America: Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America guidelines for developing an institutional program to enhance antimicrobial stewardship. Clin Infect Dis 2007; 44:159177
17. Marchaim D, Chopra T, Bhargava A, et al. Recent exposure to antimicrobials and carbapenem-resistant Enterobacteriaceae: The role of antimicrobial stewardship. Infect Control Hosp Epidemiol 2012; 33:817830
18. Cosgrove SE. The relationship between antimicrobial resistance and patient outcomes: Mortality, length of hospital stay, and health care costs. Clin Infect Dis 2006; 42(Suppl 2):S82S89
19. Murk W, Risnes KR, Bracken MB. Prenatal or early-life exposure to antibiotics and risk of childhood asthma: A systematic review. Pediatrics 2011; 127:11251138
20. Kronman MP, Zaoutis TE, Haynes K, et al. Antibiotic exposure and IBD development among children: A population-based cohort study. Pediatrics 2012; 130:e794e803
21. Smith MJ, Gerber JS, Hersh AL. Inpatient antimicrobial stewardship in pediatrics: A systematic review. J Pediatric Infect Dis Soc 2015; 4:e127e135
22. Mongelluzzo J, Mohamad Z, Ten Have TR, et al. Corticosteroids and mortality in children with bacterial meningitis. JAMA 2008; 299:20482055
23. Feudtner C, Feinstein JA, Zhong W, et al. Pediatric complex chronic conditions classification system version 2: Updated for ICD-10 and complex medical technology dependence and transplantation
. BMC Pediatr 2014; 14:199
24. Neuman MI, Hall M, Hersh AL, et al. Influence of institutional guidelines on diagnostic testing and outcomes for children hospitalized with pneumonia. Pediatrics 2012; 130: e823830
25. Bradley JS, Byington CL, Shah SS, et al.; Pediatric Infectious Diseases Society and the Infectious Diseases Society of America: The management of community-acquired pneumonia in infants and children older than 3 months of age: Clinical practice guidelines by the Pediatric Infectious Diseases Society and the Infectious Diseases Society of America. Clin Infect Dis 2011; 53:e25e76
26. Gerber JS, Kronman MP, Ross RK, et al. Identifying targets for antimicrobial stewardship in children’s hospitals. Infect Control Hosp Epidemiol 2013; 34:12521258
27. De Santis V, Gresoiu M, Corona A, et al. Bacteraemia incidence, causative organisms and resistance patterns, antibiotic strategies and outcomes in a single university hospital ICU: Continuing improvement between 2000 and 2013. J Antimicrob Chemother 2015; 70:273278
28. Hersh AL, De Lurgio SA, Thurm C, et al. Antimicrobial stewardship programs in freestanding children’s hospitals. Pediatrics 2015; 135:3339