Evaluation of treatment-related mortality among paediatric cancer deaths: a population based analysis.

BACKGROUND: Objectives were to describe the proportion of deaths due to treatment-related mortality (TRM) and to identify risk factors and probable causes of TRM among paediatric cancer deaths in a population-based cohort. METHODS: We included children with cancer ⩽18 years diagnosed and treated in Ontario who died between January 2003 and December 2012. Deaths were identified using a provincial registry, the Pediatric Oncology Group of Ontario Networked Information System. Probable causes of TRM were described. RESULTS: Among the 964 deaths identified, 821 were included. The median age at diagnosis was 6.6 years (range 0-18.8) and 51.8% had at least one relapse. Of the deaths examined, TRM occurred in 217/821 (26.4%) while 604/821 (73.6%) were due to progressive cancer. Deaths from TRM did not change over time. Using multiple regression, younger age, leukaemia diagnosis and absence of relapse were independently positively associated with TRM. The most common probable causes of TRM were respiratory, infection and haemorrhage. CONCLUSIONS: TRM was responsible for 26.4% of deaths in paediatric cancer. Underlying diagnosis, younger age and absence of relapse were associated with TRM and causes of TRM differed by diagnosis group. Future work should evaluate TRM rate and risk factors among newly diagnosed cancer patients.

Impact of registration on clinical trials on infection risk in pediatric acute myeloid leukemia.

Little is known about the impact of enrollment on therapeutic clinical trials on adverse event rates. Primary objective was to describe the impact of clinical trial registration on sterile site microbiologically documented infection for children with newly diagnosed acute myeloid leukemia (AML). We conducted a multicenter cohort study that included children aged ≤18 years with de novo AML. Primary outcome was microbiologically documented sterile site infection. Infection rates were compared between those registered and not registered on clinical trials. Five hundred seventy-four children with AML were included of which 198 (34.5%) were registered on a therapeutic clinical trial. Overall, 400 (69.7%) had at least one sterile site microbiologically documented infection. In multiple regression, registration on clinical trials was independently associated with a higher risk of microbiologically documented sterile site infection [adjusted odds ratio (OR) 1.24, 95% confidence interval (CI) 1.01-1.53; p = 0.040] and viridans group streptococcal infection (OR 1.46, 95% CI 1.08-1.98; p = 0.015). Registration on trials was not associated with Gram-negative or invasive fungal infections. Children with newly diagnosed AML enrolled on clinical trials have a higher risk of microbiologically documented sterile site infection. This information may impact on supportive care practices in pediatric AML.

Invasive Rothia infections in children with acute myeloid leukemia: A report from the Canadian infections in AML research group.

Rothia spp. (previously termed Stomatococcus) are normal flora that can cause invasive infections in immunocompromised hosts. The objective of this study was to describe infection characteristics and outcomes of Rothia spp. infections in a large cohort of children with newly diagnosed acute myeloid leukemia (AML). This retrospective chart review is a subanalysis of a larger study in which the aim was to identify factors associated with infection in pediatric patients with AML. Only sterile site infections occurring during chemotherapy were included. Among 578 children with AML, 17 (2.9%) children with at least 1 Rothia spp. infection were identified. All children were neutropenic at the time of infection. Eight (47%) had antecedent colitis or mucositis. Of the 17 infections, 16 were bacteremia and 1 was meningitis. Sepsis occurred in 4 patients, and 1 patient died due to infection. Rothia spp. infections are rare in pediatric AML but can cause significant morbidity and mortality. Future studies should describe trends in incidence and resistance patterns over time.

Lack of clarity in the definition of treatment-related mortality: pediatric acute leukemia and adult acute promyelocytic leukemia as examples.

Treatment-related mortality (TRM) is important in acute lymphoblastic leukemia and acute myeloid leukemia (AML); however, little is known about how TRM is defined across trials. Two major problems are related to what constitutes treatment versus disease-related cause of death and to TRM attribution (for example, death because of infection or hemorrhage). To address the former, we conducted a systematic review of randomized therapeutic pediatric acute leukemia and adult/pediatric acute promyelocytic leukemia trials and any study type focused on TRM in pediatric acute leukemia. We described definitions used for TRM. Sixty-six studies were included. Few therapeutic pediatric acute lymphoblastic leukemia studies (2/32, 6.3%) provided definitions for TRM, whereas more therapeutic pediatric AML studies (6/9, 66.7%) provided definitions. There was great heterogeneity in TRM classification. The authors of most studies relied on deaths during induction or in remission to delineate whether a death was TRM. However, 44.4% of therapeutic AML studies used death within a specific time frame to delineate TRM. We suggest that a consistent approach to defining and determining attribution for TRM in acute leukemia is an important future goal. Harmonization of definitions across the age spectrum would allow comparisons between pediatric and adult studies.

Infections in pediatric acute promyelocytic leukemia: from the Canadian infections in acute myeloid leukemia research group.

BACKGROUND: It is not known whether children with acute promyelocytic leukemia (APL) have an infection risk similar to non- APL acute myeloid leukemia. The objective was to describe infectious risk in children with newly diagnosed APL and to describe factors associated with these infections. METHODS: We conducted a retrospective, population-based cohort study that included children ≤ 18 years of age with de novo APL treated at 15 Canadian centers. Thirty-three children with APL were included; 78.8% were treated with APL -specific protocols. RESULTS: Bacterial sterile site infection occurred in 12 (36.4%) and fungal sterile site infection occurred in 2 (6.1%) children. Of the 127 chemotherapy courses, 101 (79.5%) were classified as intensive and among these, the proportion in which a sterile site microbiologically documented infection occurred was 14/101 (13.9%). There was one infection-related death. CONCLUSIONS: One third of children with APL experienced at least one sterile site bacterial infection throughout treatment and 14% of intensive chemotherapy courses were associated with a microbiologically documented sterile site infection. Infection rates in pediatric APL may be lower compared to non- APL acute myeloid leukemia although these children may still benefit from aggressive supportive care during intensive chemotherapy.

Incidence, risk factors, and outcomes of enteritis, typhlitis, and colitis in children with acute leukemia.

This retrospective chart review describes pediatric patients with acute lymphoblastic leukemia or acute myeloid leukemia diagnosed between January 1999 and January 2008, who were identified with enteritis, typhlitis, or colitis. Among the acute leukemia patients, 33/449 (7.3%) with acute lymphoblastic leukemia and 13/89 (14.6%) with acute myeloid leukemia experienced 51 episodes of enteritis (n=8), typhlitis (n=15), colitis (n=19), or enterocolitis (n=9). Twenty-five (49%) patients were exposed to corticosteroids within 14 days of the episode and 35 (68.6%) had fever and neutropenia concurrent with the episode. Forty-eight (94%) patients were treated with complete bowel rest and broad-spectrum antibiotics. However, 3 patients received no therapy and had uneventful courses. Complications included sepsis in 7/51 (13.7%) and intestinal obstruction in 3/51 (5.9%). One child required surgery for abscess drainage and 2 children died of causes unrelated to their colitis. Enteritis, typhlitis, or colitis occurred in 8.6% of children treated for leukemia. The optimal management approach is uncertain.

Association between corticosteroids and infection, sepsis, and infectious death in pediatric acute myeloid leukemia (AML): results from the Canadian infections in AML research group.

BACKGROUND: Infection continues to be a major problem for children with acute myeloid leukemia (AML). Objectives were to identify factors associated with infection, sepsis, and infectious deaths in children with newly diagnosed AML. METHODS: We conducted a retrospective, population-based cohort study that included children ≤ 18 years of age with de novo, non-M3 AML diagnosed between January 1995 and December 2004, treated at 15 Canadian centers. Patients were monitored for infection from initiation of AML treatment until recovery from the last cycle of chemotherapy, conditioning for hematopoietic stem cell transplantation, relapse, persistent disease, or death (whichever occurred first). Consistent trained research associates abstracted all information from each site. RESULTS: 341 patients were included. Median age was 7.1 years (interquartile range [IQR], 2.0-13.5) and 29 (8.5%) had Down syndrome. In sum, 26 (7.6%) experienced death as a first event. There were 1277 courses of chemotherapy administered in which sterile site microbiologically documented infection occurred in 313 courses (24.5%). Sepsis and infectious death occurred in 97 (7.6%) and 16 (1.3%) courses, respectively. The median days of corticosteroid administration was 2 per course (IQR, 0-6). In multiple regression analysis, duration of corticosteroid exposure was significantly associated with more microbiologically documented sterile site infection, bacteremia, fungal infection, and sepsis. The only factor significantly associated with infectious death was days of corticosteroid exposure (odds ratio, 1.05; 95% confidence interval, 1.02-1.08; P = .001). CONCLUSIONS: In pediatric AML, infection, sepsis, and infectious death were associated with duration of corticosteroid exposure. Corticosteroids should be avoided when possible for this population.

A systematic review of symptom assessment scales in children with cancer.

BACKGROUND: The objective was to describe symptom assessment scales that have been used in children with cancer. METHODS: We conducted electronic searches of OVID Medline and EMBASE in order to identify all symptom assessment scales that have been used in pediatric cancer. Two reviewers abstracted information from each identified study. Data collected included study demographics and information related to the instrument and children enrolled. We also collected information about the purpose of instrument administration and whether treatment was altered as a result of this information. RESULTS: Fourteen studies were identified which evaluated eight different symptom assessment scales. Eight studies used child self-report and all studies included children on active treatment for cancer although 4 studies also included children following completion of treatment. The most common purpose of instrument administration was to measure the prevalence of symptom burden (n = 8). None of the 14 studies used the scale to screen for symptoms and none changed patient management on the basis of identified symptoms. CONCLUSIONS: We failed to identify any symptom assessment scales that were used as a symptom screening tool. There is a need to develop such a tool for use in children with cancer.

Pneumatosis intestinalis in children with acute lymphoblastic leukemia and acute myeloid leukemia.

PURPOSE: To describe symptoms, diagnostic features, treatments, and outcomes of pneumatosis intestinalis (PI) in pediatric patients being treated for acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML). METHODS: This retrospective chart review included 514 patients ≤ 18 years of age diagnosed with ALL or AML between January 1999 and December 2007. PI episodes were identified by radiology report reviews. RESULTS: Twenty patients with ALL and four patients with AML presented 31 PI episodes. The median time between diagnoses of leukemia and PI was 1.0 month (interquartile range 0.8-6.4 months). Plain radiographs diagnosed all PI episodes. Computerized tomography (CT) and ultrasound were performed in 6 and 13 episodes, respectively. All CT and three ultrasounds demonstrated PI. Thirty episodes occurred exclusively in the colon, most commonly in the ascending (n = 26) and transverse (n = 18) segments. Treatment included complete bowel rest in 27 (87.1%) and intravenous broad-spectrum antibiotics in 29 (93.5%). One case required colectomy. Two episodes were untreated. There were no fatalities associated with PI. CONCLUSIONS: PI is uncommon in children with ALL or AML. Ultrasound is less sensitive than plain radiographs for diagnosis. PI occurred almost exclusively in the colon. With conservative management, most patients had excellent outcome.

Psychometric properties of the Oral Mucositis Daily Questionnaire for child self-report and importance of mucositis in children treated with chemotherapy.

PURPOSE: The objectives of this study were to examine the psychometric properties of the self-report Oral Mucositis Daily Questionnaire (OMDQ) and to measure the importance of mucositis in children receiving intensive chemotherapy. METHODS: Children ≥ 12 years of age receiving intensive chemotherapy for leukemia/lymphoma or undergoing stem cell transplantation were asked to complete the OMDQ daily for 21 days after chemotherapy. Other measures of mucositis obtained concurrently with OMDQ included the World Health Organization (WHO) mucositis scale, the pain visual analog scale (VAS), and the Functional Assessment of Cancer Therapy Esophageal Cancer Sub-scale (FACT-ECS). The importance of mucositis was estimated using a VAS, time trade-off technique, and willingness to pay to avoid mucositis. RESULTS: Fifteen children participated. Test-retest reliability demonstrated at least moderate correlation for all questions within the OMDQ. Assessment of construct validity of the OMDQ revealed at least moderate correlation with WHO, VAS, and FACT-ECS for questions regarding pain, swallowing, drinking, and eating. Effect on sleeping and talking had lower correlations than that expected a priori. The diarrhea question of the OMDQ did not correlate with other measures of mucositis. Severe mucositis is important to children, while mild mucositis is less important to them. Children were willing to pay moderate amounts of money to prevent mucositis. CONCLUSIONS: The OMDQ exhibits test-retest reliability, and most questions show construct validity with the exceptions of the sleep, talking, and diarrhea questions. Therefore, the OMDQ should not be used unmodified as a self-report instrument in children with cancer. Severe mucositis is of importance to these children.

Utility of peripheral blood cultures in patients with cancer and suspected blood stream infections: a systematic review.

PURPOSE: The utility of peripheral blood cultures in patients with cancer and/or hematopoietic stem cell transplantation (HSCT) recipients with central venous lines (CVL) and suspected blood stream infection (BSI) is controversial. Our main objective was to describe the proportion of bacteremia detected only by the peripheral blood (PB) culture in order to define its role in the evaluation of patients in this setting. METHODS: We performed electronic searches of OVID Medline, EMBASE, and the Cochrane Central Register of Controlled Trials for studies of adults or children with cancer and/or HSCT that evaluated concurrent PB and CVL cultures and reported sufficient data to permit calculation of the primary outcome. The proportion of bacteremia identified by site of sample was used as the effect measure. The review was registered in PROSPERO: CRD42011001610. RESULTS: From 149 reviewed articles, 7 were included in the meta-analysis. In a total number of 10,370 paired blood cultures, bacteremia was detected in 17 %. Thirteen percent of BSI were only identified by PB, while 28 % of infections were only identified by CVL. CONCLUSIONS: PB cultures identified many episodes of bacteremia not detected in the CVL culture. This finding suggests that PB culture should be considered in the evaluation of patients with cancer and/or HSCT with suspected BSI.

Perspectives toward oral mucositis prevention from parents and health care professionals in pediatric cancer.

PURPOSE: The objectives of this study were: (1) to describe parents and health care professionals (HCPs) perceived importance of oral mucositis prevention in children with cancer; (2) To describe utilities and willingness-to-pay (WTP) to prevent mucositis. METHODS: Respondents included parents of children receiving intensive chemotherapy for leukemia/lymphoma or undergoing stem cell transplantation and HCPs caring for children with cancer. Importance of mild and severe oral mucositis was estimated using a visual analogue scale (VAS). Mucositis-associated utilities were elicited using the time trade-off technique (TTO). WTP to avoid mucositis was obtained using contingent valuation. These techniques quantify how much time or money the participant is willing to relinquish in order to prevent mucositis. RESULTS: Eighty-two parents and 60 HCPs were included. Parents and HCPs believed mild mucositis to be of similar importance (median VAS 2.5 versus 3.6; P = 0.357) while parents considered severe mucositis less important than HCPs (median VAS 8.3 versus 9.0; P < 0.0001). No differences in parent versus HCP responses were seen with TTO (mild or severe mucositis) and most parents were not willing to trade any survival time to prevent severe mucositis. Parents were willing to pay significantly more than HCPs to prevent mild mucositis (average median WTP $1,371 CAN vs. $684 CAN, P = 0.031). No differences were seen in WTP to prevent severe mucositis. CONCLUSIONS: Parents and HCP believe severe mucositis to be important, although it is more important to HCPs. Parents would not be willing to reduce life expectancy to eliminate mucositis.

Defining bloodstream infections related to central venous catheters in patients with cancer: a systematic review.

The objective of this review was to determine whether consistent definitions were used in published studies of bloodstream infections due to central venous catheters in patients with cancer (ie, catheter-related or catheter-associated bloodstream infections). Review of 191 studies reporting catheter-related or catheter-associated bloodstream infections in patients with cancer revealed a lack of uniformity in these definitions. We grouped definitions by type, with 39 articles failing to cite or report a definition. Definitions included those of the Centers for Disease Control and Prevention (n = 39) and the Infectious Diseases Society of America (n = 18). The criteria included in the definitions in studies were also tabulated. Clinical manifestations were frequently included. Definitions used have been highly variable; comparability of risk factors, incidence, management, and outcomes of such infections is difficult to achieve across studies. Future research should focus on development of a common definition of catheter-related and catheter-associated bloodstream infections for both adults and children with cancer.

Complementary and alternative medicine use in pediatric cancer reported during palliative phase of disease.

PURPOSE: The objectives of this study were to assess the frequency, types, and potential determinants of complementary and alternative medicine (CAM) use, and consideration of CAM use, collected from parents with children during the palliative phase of disease. METHODS: Eligible parent respondents were identified by their primary care team. Demographic information and questionnaires were completed by the parent in the presence of a research nurse (DT). We conducted univariate logistic regression to identify predictors of parents who considered CAM use and children who actually used CAM. Descriptions of types of CAM were categorized according to the National Center for Complementary and Alternative Medicine. RESULTS: A total of 77 parents participated. Only 22 children (29%) had received some type of CAM, with 42 parents (55%) having considered its use for their child. Whole medical systems (n = 17) and biologically based therapies (n = 15) were the most frequently considered CAM, with whole medical systems (n = 6) being the most frequently used CAM. Family and disease variables were not indicative of CAM use. However, parents with higher education and those with a family member with cancer were more likely to consider CAM use, while parents were less likely to consider CAM as children were farther from time of relapse. CONCLUSIONS: The study provides initial insight into CAM use, and consideration of use, in children with cancer receiving palliative care. Further research is required to determine if the gap between CAM use and consideration is important, why this gap exists, and whether CAM has beneficial effects in this population.

Association between tumor necrosis factor-alpha promoter -308 A/G polymorphism and susceptibility to sepsis and sepsis mortality: a systematic review and meta-analysis.

OBJECTIVE: The tumor necrosis factor (TNF)-alpha promoter -308 A/G polymorphism has been reported to be associated with sepsis with inconsistent results. We conducted a systematic review and meta-analysis to determine whether the TNF-alpha -308 A/G polymorphism TNF2 (G/A or A/A) confers susceptibility to sepsis or is associated with increased risk of death from sepsis. DATA SOURCES: We performed an electronic search of OVID MEDLINE from 1950 to June 2008 and EMBASE from 1980 to June 2008. STUDY SELECTION: From 1935 reviewed study articles, 25 were included based on predefined inclusion criteria. DATA EXTRACTION: Two reviewers independently extracted data onto standardized forms. DATA SYNTHESIS: An association between development of sepsis and the TNF2 genotype was found in the overall population (odds ratio, 2.15; 95% confidence interval, 1.45-3.19; p < .01). Stratification by ethnicity indicated that the contribution to this observation may be stronger among the Asian population (odds ratio, 3.16; 95% confidence interval, 1.92 to 5.20; p < .01) compared with other ethnicities. There was no association between the TNF2 genotype and mortality from sepsis (odds ratio, 1.48; 95% confidence interval, 0.81 to 2.70; p = .20). However, when stratified for ethnicity, there may be an increased risk for fatal outcomes among Asians (odds ratio, 10.75; 95% confidence interval, 2.98 to 38.78; p < .01). CONCLUSIONS: Our systematic review and meta-analysis demonstrates that the TNF2 polymorphism is associated with sepsis. However, TNF2 is not associated with sepsis mortality.

Utility of peripheral blood cultures in bacteremic pediatric cancer patients with a central line.

PURPOSE: The utility of peripheral blood cultures in febrile neutropenic children with cancer and central venous catheters (CVC) is controversial. Our primary objective was to describe true bloodstream infections detected only by peripheral culture. Our secondary objectives were to describe true bloodstream infections detected only by CVC culture and to describe probable contaminants detected in both types of blood cultures. METHODS: We included children with cancer who had peripheral and CVC cultures obtained on the same day in which at least one culture was positive. Only cultures obtained prior to the initiation of broad-spectrum antibiotics were included. We defined true bloodstream infections due to common contaminants (such as coagulase-negative Staphylococcus) as occurring if multiple cultures were positive for the same organism or if sepsis was present. RESULTS: Between January 2002 and July 2007, 318 episodes of bloodstream infection from 224 children were included. Of these, 228/318 (71.7%) were classified as true bloodstream infections while 90/318 (28.3%) were classified as contaminants. Importantly, 28/228 (12.3%) true bloodstream infections were detected only in peripheral culture while 85/228 (37.3%) true bloodstream infections were detected only by CVC cultures. Contaminants were identified in peripheral culture in 45/318 (14.2%) of episodes and in CVC culture in 45/318 (14.2%) episodes. CONCLUSIONS: True bloodstream infections frequently are only detected in the peripheral culture. These data support continuation of the practice of routine peripheral cultures in addition to CVC cultures at the onset of fever for children with cancer who are not already receiving broad-spectrum antibiotics.

International variations in infection supportive care practices for paediatric patients with acute myeloid leukaemia.

This study described infection-related supportive care practices amongst centres participating in two large paediatric acute myeloid leukaemia (AML) cooperative groups, Children's Oncology Group (COG) and Berlin-Frankfurt-Muenster (BFM). We surveyed 216 COG and 55 BFM institutions. The overall survey response rate was 83.8%. Antibacterial prophylaxis was more common among BFM (15/46, 32.6%) compared to COG (24/180, 13.3%, P < 0.0001) institutions. Antifungal prophylaxis also was more common among BFM (42/46, 91.3%) compared to COG (137/178, 77.0%, P = 0.03). There were systematic differences in infection-related supportive care practices. This information may be used to encourage harmonization of supportive care practices and future randomized trials.

Leukapheresis and low-dose chemotherapy do not reduce early mortality in acute myeloid leukemia hyperleukocytosis: a systematic review and meta-analysis.

The role of leukapheresis and low-dose chemotherapy is unclear in decreasing early mortality in acute myeloid leukemia (AML) patients with hyperleukocytosis. This systematic review was conducted to describe early mortality (deaths during first induction) in patients with AML with an initial white blood count≥100×10(9)L(-1) stratified by the approach to leukapheresis and hydroxyurea/low-dose chemotherapy. Twenty-one studies were included. Weighted mean early deaths rate (20 studies, 1354 patients) was 20.1% (95% confidence interval 15.0-25.1). Neither leukapheresis strategy (p=0.67) nor hydroxyurea/low-dose chemotherapy (p=0.23) influenced the early death rate. Early mortality related to hyperleukocytosis in AML is not influenced by universal or selected use of leukapheresis or hydroxyurea/low-dose chemotherapy.

Predictors and outcomes of viridans group streptococcal infections in pediatric acute myeloid leukemia: from the Canadian infections in AML research group.

BACKGROUND: Viridans group streptococci (VGS) cause significant morbidity in children treated for acute myeloid leukemia (AML). Our goals were to determine the occurrence and impact of these infections in children treated for AML and to understand the factors that increase the risk of VGS infections and viridans streptococcal shock syndrome (VSSS) in this population. METHODS: We conducted a retrospective, population-based cohort study that included children ≤18 years of age with de novo AML treated at 15 Canadian centers. We evaluated factors related to VGS infection and VSSS. RESULTS: Among 341 children with AML, VGS occurred in 78 (22.9%) children over the entire course of therapy and 16 had recurrent episodes. VGS infection occurred in 97 of 1277 courses of chemotherapy (7.6%). VSSS occurred in 19.6% of these episodes and included 11 patients who required intensive care services with 2 VGS infections resulting in death. In multiple regression analysis, factors independently related to VGS included treatment on a Medical Research Council-based protocol (odds ratio (OR) 2.87, 95% confidence interval (CI) 1.53-5.39; P = 0.001), cytarabine dose per gram/m² (OR 1.04, 95% CI 1.01-1.07; P = 0.002) and prolonged neutropenia (OR 1.58, 95% CI: 0.97-2.56; P = 0.06). None of the evaluated factors were predictive of VSSS. CONCLUSIONS: VGS infections occur in 7.6% of chemotherapy courses and remain an important cause of morbidity and even mortality in children being treated for AML. Interventions to reduce VGS need to be identified.

Second Bacteremia During Antibiotic Treatment in Children With Acute Myeloid Leukemia: A Report From the Canadian Infections in Acute Myeloid Leukemia Research Group.

BACKGROUND: The risk of second bacteremia during antibiotic treatment for initial bacteremia is unknown in high-risk populations. Our objectives were to describe the prevalence of second bacteremia during treatment and identify risk factors in children with acute myeloid leukemia (AML). METHODS: We conducted a retrospective, population-based cohort study that included children and adolescents with de novo, non-M3 AML who were diagnosed and treated between January 1, 1995 and December 31, 2004 at 15 Canadian centers. Patients were monitored for bacteremia during chemotherapy until completion of treatment, hematopoietic stem cell transplantation, relapse, refractory disease, or death. RESULTS: There were 290 episodes of bacteremia occurring in 185 (54.3%) of 341 children. Eighteen (6.2%) had a second bacteremia while receiving antibiotic treatment. Two episodes of second bacteremia were complicated by sepsis; there were no infection-related deaths. Eleven episodes (61.1%) had either an initial Gram-positive and subsequent Gram-negative bacteremia or initial Gram-negative followed by Gram-positive bacteremia. Days receiving corticosteroids (odds ratio [OR], 1.09; 95% confidence interval [CI], 1.07-1.12; P < .0001), cumulative dose of corticosteroids (OR, 1.04; 95% CI, 1.00-1.08; P = .035), and days of neutropenia from start of course to initial bacteremia (OR, 1.07; 95% CI, 1.02-1.12; P = .007) were significantly associated with second bacteremia. CONCLUSIONS: In pediatric AML, 6% of patients will experience a second bacteremia during antibiotic treatment; duration of corticosteroid exposure and neutropenia are risk factors. These patients remain at high risk for second bacteremia after identification of the initial bacteremia and warrant continued broad-spectrum treatment during profound neutropenia.