This cohort study used a comprehensive database shared by 6 hospitals to determine the association between the administration of combination therapy for serious infections and the development of acute kidney injury in hospitalized children.
Key Points
Question
Is the combination of intravenous vancomycin hydrochloride and piperacillin sodium/tazobactam sodium associated with the development of acute kidney injury in children?
Findings
In this multicenter cohort study that included 1915 hospitalized children prescribed combination therapy, the concomitant administration of intravenous vancomycin plus piperacillin/tazobactam, compared with vancomycin plus other antipseudomonal β-lactam antibiotics, was significantly associated with an increased risk of acute kidney injury.
Meaning
Pediatricians must be cognizant of the potential added risk of this combination therapy when making empirical antibiotic choices.
Abstract
Importance
β-Lactam antibiotics are often coadministered with intravenous (IV) vancomycin hydrochloride for children with suspected serious infections. For adults, the combination of IV vancomycin plus piperacillin sodium/tazobactam sodium is associated with a higher risk of acute kidney injury (AKI) compared with vancomycin plus 1 other β-lactam antibiotic. However, few studies have evaluated the safety of this combination for children.
Objective
To assess the risk of AKI in children during concomitant therapy with vancomycin and 1 antipseudomonal β-lactam antibiotic throughout the first week of hospitalization.
Design, Setting, and Participants
This retrospective cohort study focused on children hospitalized for 3 or more days who received IV vancomycin plus 1 other antipseudomonal β-lactam combination therapy at 1 of 6 large children’s hospitals from January 1, 2007, through December 31, 2012. The study used the Pediatric Health Information System Plus database, which contains administrative and laboratory data from 6 pediatric hospitals in the United States. Patients with underlying kidney disease or abnormal serum creatinine levels on hospital days 0 to 2 were among those excluded. Patients 6 months to 18 years of age who were admitted through the emergency department of the hospital were included. Data were collected from July 2015 to March 2016. Data analysis took place from April 2016 through July 2017. (Exact dates are not available because the data collection and analysis processes were iterative.)
Main Outcomes and Measures
The primary outcome was AKI on hospital days 3 to 7 and within 2 days of receiving combination therapy. Acute kidney injury was defined using KDIGO criteria and was based on changes in serum creatinine level from hospital days 0 to 2 through hospital days 3 to 7. Multiple logistic regression was performed using a discrete-time failure model to test the association between AKI and receipt of IV vancomycin plus piperacillin/tazobactam or vancomycin plus 1 other antipseudomonal β-lactam antibiotic.
Results
A total of 1915 hospitalized children who received combination therapy were identified. Of the 1915 patients, a total of 866 (45.2%) were female and 1049 (54.8%) were male, 1049 (54.8%) were identified as white in race/ethnicity, and the median (interquartile range) age was 5.6 (2.1-12.7) years. Among the cohort who received IV vancomycin plus 1 other antipseudomonal β-lactam antibiotic, 157 patients (8.2%) had antibiotic-associated AKI. This number included 117 of 1009 patients (11.7%) who received IV vancomycin plus piperacillin/tazobactam combination therapy. After adjustment for age, intensive care unit level of care, receipt of nephrotoxins, and hospital, IV vancomycin plus piperacillin/tazobactam combination therapy was associated with higher odds of AKI each hospital day compared with vancomycin plus 1 other antipseudomonal β-lactam antibiotic combination (adjusted odds ratio, 3.40; 95% CI, 2.26-5.14).
Conclusions and Relevance
Coadministration of IV vancomycin and piperacillin/tazobactam may increase the risk of AKI in hospitalized children. Pediatricians must be cognizant of the potential added risk of this combination therapy when making empirical antibiotic choices.
Introduction
Hospitalized children commonly receive multiple antibiotics.1,2 Although they are lifesaving in the setting of severe infection and sepsis, some antibiotics can result in acute kidney injury (AKI),3,4 which is associated with an increased risk of chronic kidney disease and death in children.5,6,7,8 Identifying antibiotic combinations with a decreased likelihood to cause AKI is important to reduce the negative sequelae among children treated for suspected serious infection or sepsis.
Vancomycin hydrochloride and a broad-spectrum β-lactam antibiotic are often administered in combination to empirically treat children admitted with suspected serious bacterial infections. Vancomycin is the drug of choice for suspected serious gram-positive infection, and the β-lactam antibiotic provides the necessary gram-negative coverage. Despite its effectiveness, vancomycin has been associated with AKI in up to 22% of children,9,10,11,12,13 a risk that may increase in the setting of concurrent administration of other nephrotoxic agents.9,10,11,12,13
Several adult studies have suggested that adding piperacillin sodium/tazobactam sodium (TZP) to vancomycin increases the risk of AKI, compared with the use of vancomycin alone or in combination with certain β-lactam antibiotics.14,15,16,17,18,19,20 Limited data exist, however, on whether the combination of TZP and vancomycin is associated with increased nephrotoxicity in children.21,22 Using a large administrative data set, we assessed the risk of AKI in children who concomitantly received intravenous (IV) vancomycin plus an antipseudomonal β-lactam antibiotic during the first week of hospitalization. We hypothesized that the combination of vancomycin and TZP is associated with an increased risk of AKI in children. The results are immediately informative to clinicians administering empirical antibiotics to children admitted with suspected severe infection or sepsis.
Methods
Study Design and Data Source
We conducted a retrospective cohort study of hospitalized children receiving IV vancomycin plus 1 antipseudomonal β-lactam antibiotic at 6 large pediatric hospitals between January 1, 2007, and December 31, 2012. These hospitals contribute administrative and laboratory data to the Pediatric Health Information System Plus (PHIS+) database. The PHIS+ database incorporates the clinical and financial data contained in the Pediatric Health Information System (PHIS),23 augmented with laboratory and radiology data of children seen in the ambulatory and inpatient departments of the 6 hospitals. The Children’s Hospital of Philadelphia Institutional Review Board has judged that the PHIS+ database contains data that are not readily identifiable, and thus its use exempts our study from the requirements for prospective review and approval. According to the US Department of Health and Human Services Common Rule 45 CFR 46 and the policies of The Children’s Hospital of Philadelphia institutional review board, our research did not meet the definition of human subjects research, and thus no patient informed consent was obtained. Data were collected from July 2015 to March 2016. Data analysis took place from April 2016 through July 2017. (Exact dates are not available because the data collection and analysis processes were iterative.)
Study Population
We identified children 6 months to 18 years of age who were admitted through the emergency department and were given combination therapy, which consisted of IV vancomycin plus 1 antipseudomonal β-lactam agent (ceftazidime sodium, cefepime hydrochloride, TZP, meropenem, or imipenem/cilastin sodium), on at least days 1 and 2 of hospitalization; ticarcillin disodium/clavulanate potassium was not included because it was given infrequently (n = 9) and not at each hospital. Because the first calendar day associated with a hospital admission is inherently less than 24 hours and antibiotics given in the emergency department may differ from those administered when the child becomes an inpatient, the second calendar day of an admission was considered hospital day 1. For patients admitted more than once during the study period, only the first admission was included.
We excluded patients who were hospitalized for fewer than 3 days; had an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) discharge diagnosis code of kidney disease other than acute renal failure, including chronic kidney disease (580-583, 587, 590-2, 593.1, 593.3-5, 593.7-8; eTable 1 in the Supplement) or an ICD-9-CM procedure code that may be associated with chronic kidney disease (39.27, 39.42-3, 39.50, 39.53, 39.93-4, 55.52-4, 55.6; eTable 1 in the Supplement); had an ICD-9-CM procedure code for dialysis (39.95 or 54.98) during the first 3 days of hospitalization; or had an ICD-9-CM procedure code or charge for extracorporeal membrane oxygenation (36.95) during the first 7 days of hospitalization. To reduce the potential for ascertainment bias, the level of serum creatinine (SCr) had to be measured (1) at least once during hospital days 0 to 2, (2) at least once on hospital days 3 to 7, and (3) no fewer than every 4 days during the hospitalization, as validated in the literature.24,25
Exposures
For our primary analysis, we grouped patients on the basis of the combination therapy they received: vancomycin plus TZP or vancomycin plus an antipseudomonal β-lactam antibiotic (ceftazidime, cefepime, meropenem, or imipenem/cilastin). Patients who received multiple β-lactam agents plus vancomycin on hospital day 1 or 2 were excluded. Because the choice of antipseudomonal antibiotics may vary by hospital, the groups were chosen owing to the similarity in their spectrum of activity and clinical indications for the drugs they used. In addition, children receiving non-antipseudomonal, as opposed to antipseudomonal, β-lactam agents likely differed at baseline in several ways, such as severity of illness, number of previous health care exposures, and underlying conditions. Such differences could introduce bias in the comparison of cases of AKI between patients receiving vancomycin plus TZP and patients receiving vancomycin plus any β-lactam antibiotic.
Outcomes
For each patient, the presence or absence of AKI was determined each day using SCr data from hospital day 0 through hospital day 7 (or discharge), whichever was sooner. The KDIGO (Kidney Disease: Improving Global Outcomes) criteria were used to define AKI26 as an increase in SCr level by 50% or higher from baseline or by 0.3 mg/dL or higher (to convert to micromoles per liter, multiply by 88.4) within 2 hospital days or fewer. The SCr level had to be 0.5 mg/dL to qualify as AKI if the 50% change criterion was met. A SCr-based AKI definition was used because the PHIS+ database does not include patient heights to allow for back calculation of the glomerular filtration rate; urine output was also not captured by the PHIS+ database. Because laboratory data preceding admission were unavailable, the baseline SCr level was calculated using the lowest SCr level on hospital days 0 to 2; the lowest level was chosen to allow for minor perturbations in the initial SCr level measurement secondary to dehydration, which can occur on presentation to care.
The primary outcome was antibiotic-associated AKI (AA-AKI), defined as AKI occurring on any of hospital days 3 to 7 and within 2 days of the last administration of vancomycin plus an antipseudomonal β-lactam antibiotic combination therapy. Given that the primary exposure required combination therapy on at least hospital days 1 and 2, AKI identified on hospital days 0 to 2 would, by definition, precede the exposure window. Therefore, patients with AKI on hospital days 0 to 2 or a baseline SCr level above the upper limit of published normal values27 for age and sex were excluded.
Covariates
Demographic data, such as age, sex, and race, were collected from the PHIS+ database. To account for variation in chronic medical conditions, and identify patients with medical problems that precede hospitalization, complex chronic condition codes were derived using ICD-9-CM codes and dichotomized as 0 to 1 or 2 or more. These codes have previously been used to detect the presence of a chronic condition in a hospitalized patient.28
Because patients may present to the hospital with varying degrees of acuity of illness, data from the PHIS+ database were leveraged to identify patients requiring intensive care unit (ICU) level of care each day based on resource utilization and any procedure associated with cardiovascular or respiratory support (vasopressors or mechanical ventilation).29 Physical admission locations were not used. Patients were categorized based on the ICU level of care received at any time on hospital days 0 to 2 or on any given hospital day.
Receipt of concomitant nephrotoxic medications (eTable 2 in the Supplement) the preceding day was categorized as fewer than 2 vs 2 or more each day. Because the injurious outcomes of nephrotoxins may be summative, continuous nephrotoxin exposure, defined as receipt of multiple nephrotoxic medications for more than 2 consecutive days, was also identified. Intravenous contrast administration on the preceding day was separately evaluated. Data on the first vancomycin trough concentration for each patient on hospital day 1 or 2 were also collected.
Statistical Analysis
Baseline (hospital days 0 to 2) characteristics of patients with or without subsequent AA-AKI were compared using the Wilcoxon rank sum test for continuous variables and the χ2 tests for categorical variables. Baseline characteristics were also compared across the combination therapy groups. Temporal trends in AA-AKI incidence were assessed using negative binomial regression to evaluate AA-AKI by each year of the study. Median vancomycin trough concentrations were compared for the subset with available data, as were the proportions of patients in each combination therapy group with a first vancomycin trough more than 15 µg/mL and more than 20 µg/mL (to convert to micromoles per liter, multiply by 0.690).
We performed multiple logistic regression using a discrete-time failure model to test the association between AA-AKI on any given hospital day and receipt of either vancomycin plus TZP or vancomycin plus another 1 antipseudomonal β-lactam antibiotic. Time was modeled using a day indicator. Patients contributed days to the model until the detection of AA-AKI, hospital day 7, or 2 days after the end of vancomycin plus an antipseudomonal β-lactam agent combination therapy, whichever occurred first. The association between covariates and AA-AKI was first evaluated using a model adjusting for the combination therapy group and for hospital day. Then, the final multivariable model was constructed, including ICU level of care, exposure to concomitant nephrotoxic medications, and hospital that admitted the patient a priori, along with other covariates if P < .20 or if inclusion resulted in a change in the point estimate of the combination therapy group by 10% or more. We did not evaluate vancomycin trough concentrations because we could not verify through the PHIS+ database the timing of measurement of the preceding doses. To evaluate the potential influence of confounding by an unmeasured covariate on the observed measure of association between the combination therapy group and AA-AKI, we conducted a quantitative sensitivity analysis (eAppendix in the Supplement).30,31 In addition, we examined the influence of our definition of baseline SCr level by repeating the multivariable analysis in a subset of patients whose SCr level measurements on hospital days 0 to 2 were less than the median for age and sex.27
Hospital outcomes (length of stay and in-hospital mortality) of patients with or without AA-AKI were compared using the Wilcoxon rank sum test and the χ2 tests, respectively. Multiple logistic regression determined the association of AA-AKI with in-hospital mortality, adjusting for the combination therapy group and other baseline (hospital days 0-2) covariates if P < .20 or if inclusion resulted in a change in the point estimate of AA-AKI by 10% or more. P < .05 was considered significant, and all P values reported were 2-sided. All analyses were performed using Stata, version 13.1 (StataCorp LLC).
Results
We identified 1915 admissions of patients who received the IV vancomycin plus a β-lactam agent combination therapy on at least hospital days 1 and 2 during the study period (eFigure in the Supplement). Of the 1915 patients, a total of 866 (45.2%) were female and 1049 (54.8%) were male, 1049 (54.8%) were identified as white in race/ethnicity, and the median age was 5.60 years (interquartile range [IQR], 2.12-12.65 years). In this combination therapy cohort, 157 patients (8.2%) developed AA-AKI within the first hospital week, and 74 (47%) of these patients had KDIGO26 stage 2 AKI or higher. Table 1 displays the baseline characteristics of patients in the cohort with AA-AKI and patients in the cohort without AA-AKI. The incidences of AA-AKI were similar across the years of the study, ranging from 19 (6.6%) in 2008 to 33 (9.2%) in 2010 (P = .89); no trend by study year was observed (negative binomial regression P = .85). Recipients of IV vancomycin plus a β-lactam agent who sustained AA-AKI, compared with those who did not, were older (median age, 12.64 years [IQR, 7.92-15.82 years] vs 5.03 years [IQR, 1.97-11.81 years]; P < .001), more often required ICU level of care on hospital days 0 to 2 (67/157 [42.7%] vs 497/1758 [28.3%]; P < .001), and were more often administered 2 or more nephrotoxic medications (40/157 [25.5%] vs 319/1758 [18.1%]; P = .02) or IV contrast on hospital days 0 to 2 (25/157 [15.9%] vs 168/1758 [9.6%]; P = .01). The AA-AKI incidences across the 6 hospitals ranged from 6 patients (4.2%) to 15 patients (11.5%) (P < .001).
Table 1. Baseline (Hospital Days 0 to 2) Characteristics of Patients With or Without AA-AKI Who Received Vancomycin Hydrochloride Plus 1 Antipseudomonal β-Lactam Agent.
| Variable | Patients, No. (%)a | P Value | |
|---|---|---|---|
| Without AA-AKI (n = 1758) |
With AA-AKI (n = 157) |
||
| Sex | |||
| Female | 792 (91.5) | 74 (8.6) | .62 |
| Male | 966 (92.1) | 83 (7.9) | |
| Race | |||
| White | 960 (91.5) | 89 (8.5) | .88 |
| Black | 144 (92.3) | 12 (7.7) | |
| Other, mixed race, or unknown | 654 (92.1) | 56 (7.9) | |
| Ethnicity | |||
| Hispanic or Latino | 143 (94.1) | 9 (5.9) | .29 |
| Not Hispanic or Latino/unknown | 1615 (91.6) | 148 (8.4) | |
| Age, median (IQR), y | 5.03 (1.97-11.81) |
12.64 (7.92-15.82) |
<.001 |
| Required ICU level of care on hospital days 0-2 | 497 (88.1) | 67 (11.9) | <.001 |
| Required no ICU level of care | 1261 (93.3) | 90 (6.7) | |
| Had ≥2 complex or chronic conditions | 932 (90.8) | 95 (9.3) | .07 |
| Had 0-1 complex or chronic conditions | 826 (93.0) | 62 (7.0) | |
| Received ≥2 concomitant nephrotoxins on hospital days 0-2 | 319 (88.9) | 40 (11.1) | .02 |
| Received 0-1 nephrotoxin on hospital days 0-2 | 1439 (92.5) | 117 (7.5) | |
| Received IV contrast on hospital days 0-2 | 168 (87.1) | 25 (13.0) | .01 |
| Received no IV contrast | 1590 (92.3) | 132 (7.7) | |
| Year of study | |||
| 2007 | 252 (91.6) | 23 (8.4) | .89 |
| 2008 | 270 (93.4) | 19 (6.6) | |
| 2009 | 267 (91.1) | 26 (8.9) | |
| 2010 | 325 (90.8) | 33 (9.2) | |
| 2011 | 312 (92.0) | 27 (8.0) | |
| 2012 | 332 (92.0) | 29 (8.0) | |
| PHIS hospital | |||
| 1 | 230 (94.7) | 13 (5.4) | .04 |
| 2 | 541 (90.3) | 58 (9.7) | |
| 3 | 412 (90.6) | 43 (9.5) | |
| 4 | 116 (88.6) | 15 (11.5) | |
| 5 | 322 (93.6) | 22 (6.4) | |
| 6 | 137 (95.5) | 6 (4.2) | |
Abbreviations: AA-AKI, antibiotic-associated acute kidney injury; ICU, intensive care unit; IQR, interquartile range; IV, intravenous; PHIS, Pediatric Health Information System.
Percentages are based on the total for each row.
Among recipients of any combination therapy on at least hospital days 1 and 2, AA-AKI was detected most often in recipients of vancomycin plus TZP (117 [11.7%]; Table 2). The Figure displays a Kaplan-Meier graph of AA-AKI–free survival for each combination therapy group. Baseline characteristics of recipients of combination therapy are shown in Table 3. The duration of combination therapy and the number of SCr level measurements were similar among the combination therapy groups. Among the subset of patients with vancomycin trough concentrations available on day 1 or 2 (1177 [61.5%]), the median initial troughs were similar between those who received IV vancomycin plus TZP and those who received vancomycin plus 1 other antipseudomonal β-lactam agent (7.5 µg/mL [IQR, 5.0-10.7 µg/mL] vs 7.1 µg/mL [IQR, 5.2-10.2 µg/mL]; P = .62; eTable 3 in the Supplement).
Table 2. Unadjusted Frequency of AA-AKI Among Patients Who Received Combination Therapy on at Least Hospital Days 1 and 2.
| β-Lactam Antibiotic | Patients, No. (%) | % Increase in SCr Level Among Those With AA-AKI, Median (IQR) | |
|---|---|---|---|
| Courses of Treatment (n = 1915) |
Cumulative Incidence of AA-AKI (n = 157) |
||
| Piperacillin sodium/tazobactam sodium | 1009 | 117 (11.7) | 67 (67-120) |
| Ceftazidime sodium | 295 | 17 (5.8) | 75 (55-200) |
| Cefepime hydrochloride | 422 | 17 (4.0) | 67 (60-120) |
| Meropenem/imipenem | 189 | 6 (3.2) | 69 (43-133) |
Abbreviations: AA-AKI, antibiotic-associated acute kidney injury; IQR, interquartile range; SCr, serum creatinine.
Figure. Kaplan-Meier Curves for Antibiotic-Associated Acute Kidney Injury (AA-AKI)–Free Survival in Hospitalized Children.
Combination therapy was administered to children during the first week of hospitalization. TZP indicates piperacillin sodium/tazobactam sodium.
Table 3. Characteristics of Patients Who Received Combination Therapy.
| Variable | Patients, No. (%)a | P Value | |
|---|---|---|---|
| Vancomycin Hydrochloride Plus Piperacillin Sodium/ Tazobactam Sodium (n = 1009) |
Vancomycin Plus 1 Antipseudomonal β-Lactam Agent (n = 906) |
||
| Sex | |||
| Female | 447 (44.3) | 419 (46.3) | .39 |
| Male | 563 (55.7) | 487 (53.8) | |
| Race | |||
| White | 568 (56.3) | 481 (53.1) | .36 |
| Black | 81 (8.0) | 75 (8.3) | |
| Other, mixed race, or unknown | 360 (35.7) | 350 (38.6) | |
| Hispanic or Latino ethnicity | 64 (6.3) | 88 (9.7) | .006 |
| Age, median (IQR), y | 4.42 (1.56-11.47) |
7.07 (2.77-13.72) |
<.001 |
| Required ICU level of care on hospital days 0-2 | 367 (36.4) | 197 (21.7) | <.001 |
| Had ≥2 complex or chronic conditions | 557 (54.2) | 470 (51.9) | .15 |
| Received ≥2 concomitant nephrotoxins on hospital days 0-2 | 225 (22.3) | 134 (14.8) | <.001 |
| Received IV contrast on hospital days 0-2 | 118 (11.7) | 75 (8.3) | .01 |
| PHIS hospital | <.001 | ||
| 1 | 15 (1.5) | 228 (25.2) | |
| 2 | 384 (38.1) | 215 (23.7) | |
| 3 | 371 (36.8) | 84 (9.3) | |
| 4 | 41 (4.1) | 90 (9.9) | |
| 5 | 133 (13.2) | 211 (23.3) | |
| 6 | 65 (6.4) | 78 (8.6) | |
| Patients with AA-AKI | 117 (11.6) | 40 (4.4) | <.001 |
| AA-AKI onset, median (IQR), d | 4 (3-5) | 4 (3-6) | .25 |
| Duration of combination therapy, median (IQR), d | 4 (3-7) | 4 (3-7) | .69 |
| No. of SCr level measurements on hospital days 3-7, median (IQR) | 2 (1-4) | 2 (1-3) | .66 |
Abbreviations: AA-AKI, antibiotic-associated acute kidney injury; ICU, intensive care unit; IQR, interquartile range; IV, intravenous; PHIS, Pediatric Health Information System; SCr, serum creatinine.
Percentages are based on the total within each combination therapy group.
On multivariable analysis using a discrete-time failure model, receipt of IV vancomycin plus TZP combination therapy was associated with increased odds of AA-AKI each hospital day (adjusted odds ratio [aOR], 3.40; 95% CI, 2.26-5.14; Table 4). These findings were robust for the addition of an unmeasured confounder to the model across a range of plausible clinical scenarios (eTable 4 in the Supplement). In a sensitivity analysis of patients with a baseline SCr level less than the median SCr level for age and sex, a similar association between IV vancomycin plus TZP combination therapy and AA-AKI was observed (aOR, 4.14; 95% CI, 2.34-7.33).
Table 4. Discrete-Time Failure Models of AA-AKI in Patients Who Received IV Vancomycin Hydrochloride Plus 1 Antipseudomonal β-Lactam Antibiotic.
| Covariate | OR (95% CI)a | aOR (95% CI)b |
|---|---|---|
| Vancomycin plus | ||
| 1 other antipseudomonal β-lactam agent | 1 [Reference] | 1 [Reference] |
| Piperacillin sodium/tazobactam sodium | 2.64 (1.83-3.79) | 3.40 (2.26-5.14) |
| Female sex | 1.06 (0.77-1.46) | |
| Race | ||
| Other, mixed race, or unknown | 1 [Reference] | |
| White | 1.06 (0.76-1.49) | |
| Black | 1.13 (0.60-2.13) | |
| Hispanic or Latino ethnicity | 0.77 (0.39-1.53) | |
| Age in years | 1.14 (1.12-1.17) | 1.15 (1.12-1.18) |
| Presence of ≥2 complex or chronic conditions | 1.16 (0.83-1.60) | |
| Receipt of ≥2 concomitant nephrotoxins on hospital days 0-2 | 1.17 (0.81-1.70) | |
| Receipt of ≥2 concomitant nephrotoxins on the preceding hospital day | 1.12 (0.73-1.73) | |
| Receipt of ≥2 nephrotoxins for >2 consecutive days | 1.36 (0.80-2.31) | 1.50 (0.87-2.16) |
| Receipt of IV contrast on hospital days 0-2 | 1.32 (0.85-2.06) | |
| Receipt of IV contrast on the preceding hospital day | 0.49 (0.12-2.00) | |
| Requirement of ICU level of care on hospital days 0-2 | 1.31 (0.95-1.80) | |
| Requirement of ICU level of care on a given day | 1.42 (1.01-2.00) | 1.46 (0.99-2.16) |
| Receipt of mechanical ventilation on a given day | 1.22 (0.85-1.76) | |
| Receipt of vasopressors on a given day | 1.68 (0.98-2.88) | |
| Year of study | ||
| 2007 | 1 [Reference] | |
| 2008 | 0.71 (0.38-1.32) | |
| 2009 | 1.12 (0.63-1.98) | |
| 2010 | 1.15 (0.67-1.99) | |
| 2011 | 0.93 (0.52-1.63) | |
| 2012 | 1.01 (0.58-1.63) | |
| PHIS hospital | ||
| 1 | 1 [Reference] | 1 [Reference] |
| 2 | 0.76 (0.39-1.45) | 0.90 (0.47-1.74) |
| 3 | 0.59 (0.30-1.14) | 0.69 (0.34-1.39) |
| 4 | 1.79 (0.83-3.87) | 1.88 (0.84-4.21) |
| 5 | 0.66 (0.83-3.88) | 0.66 (0.31-1.34) |
| 6 | 0.40 (0.15-1.08) | 0.40 (0.15-1.10) |
| Hospital day | ||
| 3 | 1 [Reference] | 1 [Reference] |
| 4 | 0.61 (0.41-0.93) | 0.63 (0.41-0.98) |
| 5 | 0.71 (0.44-1.13) | 0.70 (0.43-1.14) |
| 6 | 0.89 (0.53-1.50) | 0.89 (0.51-1.56) |
| 7 | 0.79 (0.41-1.52) | 0.76 (0.38-1.49) |
Abbreviations: AA-AKI, antibiotic-associated acute kidney injury; aOR, adjusted odds ratio; ICU, intensive care unit; IV, intravenous; OR, odds ratio; PHIS, Pediatric Health Information System.
Models adjusted for combination therapy group and hospital day, along with specific covariate being tested.
Receipt of nephrotoxins, ICU level of care, and PHIS hospital included in the final model a priori, along with covariates with P < .20 or that caused a 10% change in the point estimate of combination therapy group. Consecutive days of nephrotoxin receipt were most strongly associated with AA-AKI among nephrotoxin covariates and were included in the final model.
Patients who sustained AA-AKI had increased length of stay (median, 13 days [IQR, 6-18 days] vs 10 days [IQR, 9-26 days]; P < .001) and increased in-hospital mortality (7 of 157 [4.5%] vs 22 of 1758 [1.3%]; P = .002), compared with patients who did not sustain AA-AKI. On multiple logistic regression analysis, AA-AKI, ICU level of care on hospital days 0 to 2, and 2 or more nephrotoxins on hospital days 0 to 2 were associated with increased in-hospital mortality, whereas receiving combination therapy was not (eTable 5 in the Supplement).
Discussion
Vancomycin is an important antibiotic for treatment of gram-positive infections, but clinicians need to be cognizant of its potential for nephrotoxicity. When serious infections are suspected, vancomycin and a β-lactam agent are often coadministered to provide appropriate empirical broad-spectrum coverage until a pathogen is identified. Understanding how the nephrotoxic potential of vancomycin is altered by the choice of β-lactam agent can inform empirical antibiotic decisions that will reduce the risk for subsequent AKI. This study demonstrates that AKI in children receiving vancomycin appeared to be increased by coadministration of TZP compared with similar-spectrum β-lactam antibiotics. Our findings are in keeping with several recent studies in adults.14,15,16,17,18,19
Limited pediatric data are available, but previous investigations have suggested an association between TZP and AKI during courses of vancomycin. Knoderer and colleagues32 reported that among children administered vancomycin for 8 days or more, a higher percentage of patients experiencing AKI received concomitant TZP (62% vs 38%), although this difference was not significant on multivariable analyses. McQueen and Clark21 compared the incidence of AKI between children who received vancomycin alone (3.8%) and children who received vancomycin plus TZP (23.6%). This difference was significant in bivariate analysis, but a multivariate analysis was not conducted. To our knowledge, no pediatric studies have described an independent association of AKI with concurrent TZP and vancomycin administration. There is a risk of confounding by indication, but sicker patients may have been more likely to receive vancomycin plus TZP and thus were more likely to sustain AKI. We attempted to address this by adjusting for differences in patient characteristics and by comparing patients receiving similarly broad antibiotic coverage. Our results were also robust to the potential for unmeasured confounding that explains the association between vancomycin plus TZP combination therapy and AA-AKI.
The negative association of AKI with outcomes of hospitalized children receiving vancomycin has been well documented.11,12,13 As in our study, AKI has been associated with increased hospital length of stay and in-hospital mortality.11,12,13 Few therapeutic approaches exist to mitigate AKI secondary to nephrotoxic medication exposure, aside from avoidance of these potentially injurious agents.
With limited alternatives to vancomycin for treatment of serious gram-positive infections, knowledge that concomitant TZP use carries an added risk of nephrotoxicity for children may help guide empirical β-lactam antibiotic selection. When adequate substitutes to TZP exist, other broad-spectrum β-lactam antibiotics should be considered. In settings where TZP and vancomycin are both needed, clinicians should make efforts to limit the duration of combination therapy. Other β-lactam agents may less often be associated with AKI, but additional studies should evaluate how these antibiotics are associated with the development of other adverse events, such as Clostridium difficile infection, and antimicrobial resistance.
The mechanism by which TZP and vancomycin cause greater nephrotoxicity is unknown. Piperacillin inhibits tubular secretion and clearance of other drugs, and an interaction between piperacillin and vancomycin may contribute to toxic consequences for proximal tubule cells.33,34 Rutter et al18 reported that the incidence of AKI in adults receiving vancomycin plus either TZP or cefepime increased with higher dosages of TZP but not of vancomycin or cefepime. This finding suggests that TZP plays an important additive or synergistic role in vancomycin nephrotoxicity. However, because the PHIS+ data set does not include medication dosing information, we could not investigate the association between antibiotic dosage and AKI. Future studies need to determine the association between vancomycin and TZP dosing and subsequent AKI in children.
Higher vancomycin trough serum concentrations are associated with vancomycin nephrotoxicity in children.9,10,13 For the subset of patients with initial vancomycin trough levels available in our study, trough concentrations were similar across the combination therapy groups. However, we did not incorporate trough concentrations into multivariable analyses because of the lack of information about the timing of doses in the PHIS+ database. We do not suspect that vancomycin doses or trough goals vary by β-lactam agent coadministration, given that vancomycin serum trough concentrations have been similar across combination therapy groups reported in the adult literature.17,18,21
Limitations
This study has several limitations. To focus on the empirical antibiotic decisions made on patient presentation, we restricted our cohort to children admitted through the emergency department and given combination therapy on at least hospital days 1 and 2. Our results are not necessarily generalizable to other clinical scenarios. For instance, the risk factors for AKI in patients starting antimicrobial therapy later during hospitalization may be different than in patients who are treated on admission32; such patients may be sicker or have more comorbidities. In addition, we limited our analysis to AKI onset in the first week of hospitalization. The factors influencing AKI may differ in children with prolonged vancomycin administration, as suggested by Knoderer et al.32
Investigation of AKI using administrative data sets is challenging. Certain variables, such as vital signs and fluid status, are not included in the PHIS+ database. Reliance on billing codes has low sensitivity and underestimates the incidence of nephrotoxin-associated AKI in children.24,35 To overcome this potential limitation, we leveraged a large administrative database supplemented with laboratory data and applied a systematic approach to defining AKI. This method resulted in an estimated AKI incidence in our study that is similar to that previously reported in the pediatric literature describing vancomycin nephrotoxicity.10,36 The method of SCr level measurement likely differs at each of the 6 hospitals in this study, but the use of an AKI definition based on percentage change in SCr level allows for more consistent identification of AKI cases across all hospitals. Through implementation of rigorous screening criteria, we believe that we identified a cohort of hospitalized children with normal baseline kidney function and accurately captured AA-AKI on days 3 to 7 after admission.
Conclusions
This large, multicenter study using both administrative and laboratory data found that the combination of vancomycin and TZP was associated with AKI in children during the first week of hospitalization compared with vancomycin combined with 1 other similar-spectrum β-lactam agent. Because of the deleterious consequences of AKI, including increased length of stay and in-hospital mortality, clinicians must be cognizant of the potential added risk of this combination therapy when making empirical antibiotic choices. Pediatricians must limit the duration of vancomycin plus TZP combination therapy, as is feasible, and closely monitor children for whom both of these drugs are necessary.
eAppendix. Methods
eFigure. Flow Diagram of Study Cohort Derivation
eTable 1. Diagnosis Codes Associated With Diseases of the Kidney Other Than Acute Kidney Injury and Procedures Associated With Chronic Kidney Disease
eTable 2. Nephrotoxic Medications
eTable 3. Vancomycin Trough Concentrations Among Recipients of Vancomycin Plus Piperacillin/Tazobactam or Vancomycin Plus Other Antipseudomonal Beta-Lactam Agents
eTable 4. Sensitivity Analysis of the Effect of an Unmeasured Confounder on the Association Between Combination Therapy Group and AA-AKI
eTable 5. Multiple Logistic Regression of Antibiotic-Associated AKI on In-Hospital Mortality Among Recipients of Vancomycin and an Antipseudomonal Beta-Lactam
eReferences.
References
- 1.Lasky T, Ernst FR, Greenspan J, Wang S, Gonzalez L. Estimating pediatric inpatient medication use in the United States. Pharmacoepidemiol Drug Saf. 2011;20(1):-. [DOI] [PubMed] [Google Scholar]
- 2.Feudtner C, Dai D, Faerber J, Metjian TA, Luan X. Pragmatic estimates of the proportion of pediatric inpatients exposed to specific medications in the USA. Pharmacoepidemiol Drug Saf. 2013;22(8):890-898. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Pannu N, Nadim MK. An overview of drug-induced acute kidney injury. Crit Care Med. 2008;36(4 suppl):S216-S223. [DOI] [PubMed] [Google Scholar]
- 4.Patzer L. Nephrotoxicity as a cause of acute kidney injury in children. Pediatr Nephrol. 2008;23(12):2159-2173. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Menon S, Kirkendall ES, Nguyen H, Goldstein SL. Acute kidney injury associated with high nephrotoxic medication exposure leads to chronic kidney disease after 6 months. J Pediatr. 2014;165(3):522-527. [DOI] [PubMed] [Google Scholar]
- 6.Alkandari O, Eddington KA, Hyder A, et al. Acute kidney injury is an independent risk factor for pediatric intensive care unit mortality, longer length of stay and prolonged mechanical ventilation in critically ill children: a two-center retrospective cohort study. Crit Care. 2011;15(3):R146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Sutherland SM, Byrnes JJ, Kothari M, et al. AKI in hospitalized children: comparing the pRIFLE, AKIN, and KDIGO definitions. Clin J Am Soc Nephrol. 2015;10(4):554-561. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Akcan-Arikan A, Zappitelli M, Loftis LL, Washburn KK, Jefferson LS, Goldstein SL. Modified RIFLE criteria in critically ill children with acute kidney injury. Kidney Int. 2007;71(10):1028-1035. [DOI] [PubMed] [Google Scholar]
- 9.McKamy S, Hernandez E, Jahng M, Moriwaki T, Deveikis A, Le J. Incidence and risk factors influencing the development of vancomycin nephrotoxicity in children. J Pediatr. 2011;158(3):422-426. [DOI] [PubMed] [Google Scholar]
- 10.Le J, Ny P, Capparelli E, et al. Pharmacodynamic characteristics of nephrotoxicity associated with vancomycin use in children. J Pediatric Infect Dis Soc. 2015;4(4):e109-e116. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Totapally BR, Machado J, Lee H, Paredes A, Raszynski A. Acute kidney injury during vancomycin therapy in critically ill children. Pharmacotherapy. 2013;33(6):598-602. [DOI] [PubMed] [Google Scholar]
- 12.Seixas GT, Araujo OR, Silva DC, Arduini RG, Petrilli AS. Vancomycin therapeutic targets and nephrotoxicity in critically ill children with cancer. J Pediatr Hematol Oncol. 2016;38(2):e56-e62. [DOI] [PubMed] [Google Scholar]
- 13.Knoderer CA, Nichols KR, Lyon KC, Veverka MM, Wilson AC. Are elevated vancomycin serum trough concentrations achieved within the first 7 days of therapy associated with acute kidney injury in children? J Pediatric Infect Dis Soc. 2014;3(2):127-131. [DOI] [PubMed] [Google Scholar]
- 14.Burgess LD, Drew RH. Comparison of the incidence of vancomycin-induced nephrotoxicity in hospitalized patients with and without concomitant piperacillin-tazobactam. Pharmacotherapy. 2014;34(7):670-676. [DOI] [PubMed] [Google Scholar]
- 15.Kim T, Kandiah S, Patel M, et al. Risk factors for kidney injury during vancomycin and piperacillin/tazobactam administration, including increased odds of injury with combination therapy. BMC Res Notes. 2015;8:579-589. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Fodero KE, Horey AL, Krajewski MP, Ruh CA, Sellick JA Jr, Mergenhagen KA. Impact of an antimicrobial stewardship program on patient safety in veterans prescribed vancomycin. Clin Ther. 2016;38(3):494-502. [DOI] [PubMed] [Google Scholar]
- 17.Navalkele B, Pogue JM, Karino S, et al. Risk of acute kidney injury in patients on concomitant vancomycin and piperacillin-tazobactam compared to those on vancomycin and cefepime. Clin Infect Dis. 2017;64(2):116-123. [DOI] [PubMed] [Google Scholar]
- 18.Rutter WC, Cox JN, Martin CA, Burgess DR, Burgess DS. Nephrotoxicity during vancomycin therapy in combination with piperacillin-tazobactam or cefepime. Antimicrob Agents Chemother. 2017;61(2):e02089-16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Gomes DM, Smotherman C, Birch A, et al. Comparison of acute kidney injury during treatment with vancomycin in combination with piperacillin-tazobactam or cefepime. Pharmacotherapy. 2014;34(7):662-669. [DOI] [PubMed] [Google Scholar]
- 20.Hammond DA, Smith MN, Li C, Hayes SM, Lusardi K, Bookstaver PB. Systematic review and meta-analysis of acute kidney injury associated with concomitant vancomycin and piperacillin/tazobactam. Clin Infect Dis. 2017;64(5):666-674. [DOI] [PubMed] [Google Scholar]
- 21.McQueen KE, Clark DW. Does combination therapy with vancomycin and piperacillin-tazobactam increase the risk of nephrotoxicity versus vancomycin alone in pediatric patients? J Pediatr Pharmacol Ther. 2016;21(4):332-338. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.LeCleir LK, Pettit RS. Piperacillin-tazobactam versus cefepime incidence of acute kidney injury in combination with vancomycin and tobramycin in pediatric cystic fibrosis patients. Pediatr Pulmonol. 2017;52(8):1000-1005. [DOI] [PubMed] [Google Scholar]
- 23.Cameron DB, Graham DA, Milliren CE, et al. Quantifying the burden of interhospital cost variation in pediatric surgery: implications for the prioritization of comparative effectiveness research. JAMA Pediatr. 2017;171(2):e163926. [DOI] [PubMed] [Google Scholar]
- 24.Schaffzin JK, Dodd CN, Nguyen H, Schondelmeyer A, Campanella S, Goldstein SL. Administrative data misclassifies and fails to identify nephrotoxin-associated acute kidney injury in hospitalized children. Hosp Pediatr. 2014;4(3):159-166. [DOI] [PubMed] [Google Scholar]
- 25.Moffett BS, Goldstein SL. Acute kidney injury and increasing nephrotoxic-medication exposure in noncritically-ill children. Clin J Am Soc Nephrol. 2011;6(4):856-863. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Kidney International Supplements. Kidney Disease: Improving Global Outcomes (KDIGO). KDIGO clinical practice guideline for acute kidney injury. http://www.kdigo.org/clinical_practice_guidelines/pdf/KDIGO%20AKI%20Guideline.pdf. Published March 2012. Accessed May 11, 2012.
- 27.Pottel H, Vrydags N, Mahieu B, Vandewynckele E, Croes K, Martens F. Establishing age/sex related serum creatinine reference intervals from hospital laboratory data based on different statistical methods. Clin Chim Acta. 2008;396(1-2):49-55. [DOI] [PubMed] [Google Scholar]
- 28.Feudtner C, Hays RM, Haynes G, Geyer JR, Neff JM, Koepsell TD. Deaths attributed to pediatric complex chronic conditions: national trends and implications for supportive care services. Pediatrics. 2001;107(6):E99. [DOI] [PubMed] [Google Scholar]
- 29.Maude SL, Fitzgerald JC, Fisher BT, et al. Outcome of pediatric acute myeloid leukemia patients receiving intensive care in the United States. Pediatr Crit Care Med. 2014;15(2):112-120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Lash TL, Fox MP, Fink AK. Applying Quantitative Bias Analysis to Epidemiologic Data. New York, NY: Springer; 2009. [Google Scholar]
- 31.Lash TL, Fox MP, MacLehose RF, Maldonado G, McCandless LC, Greenland S. Good practices for quantitative bias analysis. Int J Epidemiol. 2014;43(6):1969-1985. [DOI] [PubMed] [Google Scholar]
- 32.Knoderer CA, Gritzman AL, Nichols KR, Wilson AC. Late-occurring vancomycin-associated acute kidney injury in children receiving prolonged therapy. Ann Pharmacother. 2015;49(10):1113-1119. [DOI] [PubMed] [Google Scholar]
- 33.Najjar TA, Abou-Auda HS, Ghilzai NM. Influence of piperacillin on the pharmacokinetics of methotrexate and 7-hydroxymethotrexate. Cancer Chemother Pharmacol. 1998;42(5):423-428. [DOI] [PubMed] [Google Scholar]
- 34.Landersdorfer CB, Kirkpatrick CM, Kinzig M, Bulitta JB, Holzgrabe U, Sörgel F. Inhibition of flucloxacillin tubular renal secretion by piperacillin. Br J Clin Pharmacol. 2008;66(5):648-659. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Grams ME, Waikar SS, MacMahon B, Whelton S, Ballew SH, Coresh J. Performance and limitations of administrative data in the identification of AKI. Clin J Am Soc Nephrol. 2014;9(4):682-689. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Cies JJ, Shankar V. Nephrotoxicity in patients with vancomycin trough concentrations of 15-20 μg/ml in a pediatric intensive care unit. Pharmacotherapy. 2013;33(4):392-400. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eAppendix. Methods
eFigure. Flow Diagram of Study Cohort Derivation
eTable 1. Diagnosis Codes Associated With Diseases of the Kidney Other Than Acute Kidney Injury and Procedures Associated With Chronic Kidney Disease
eTable 2. Nephrotoxic Medications
eTable 3. Vancomycin Trough Concentrations Among Recipients of Vancomycin Plus Piperacillin/Tazobactam or Vancomycin Plus Other Antipseudomonal Beta-Lactam Agents
eTable 4. Sensitivity Analysis of the Effect of an Unmeasured Confounder on the Association Between Combination Therapy Group and AA-AKI
eTable 5. Multiple Logistic Regression of Antibiotic-Associated AKI on In-Hospital Mortality Among Recipients of Vancomycin and an Antipseudomonal Beta-Lactam
eReferences.