Abstract: The authors developed a novel clinical pathway in the pediatric emergency department (PED) of Doernbecher Children’s Hospital, Oregon Health & Science University, that endorses administering intravenous antibiotics to febrile pediatric patients on chemotherapy or recently post–stem cell transplant before the absolute neutrophil count (ANC) is resulted. Three cohorts of consecutively selected febrile patients on chemotherapy who presented to the PED (pre-, post-, and late cohorts) were retrospectively evaluated. Time to antibiotic administration (TTA), number of episodes that met goal TTA (ie, <60 minutes after initial presentation to the PED), and PED length of stay (LOS) were calculated. Median TTA was significantly shorter in the post-cohort than in the pre-cohort (60 minutes vs 115 minutes; P<.0001). More patients met goal TTA in the post-cohort than in the pre-cohort (72% vs 15%; P<.0001). However, median PED LOS was similar between the pre- and post-cohorts (200 minutes vs 196 minutes; P=.8). The three outcome measures in the post-implementation cohort were similar to those in the late cohort. In pediatric patients at risk for fever and neutropenia, the implementation of a clinical pathway to endorse administration of antibiotics before the availability of ANC results was effective in reducing TTA for febrile pediatric patients on chemotherapy in the PED, but it did not reduce PED LOS.
Key Words: fever, neutropenia, pediatric, oncology, clinical pathway, antibiotic administration, emergency department length of stay.
Citation: Journal of Clinical Pathways. 2015;1(2):33–42.
Received September 18, 2015; accepted November 3, 2015.
The Infectious Diseases Society of America and the National Comprehensive Cancer Network currently recommend that patients with fever and neutropenia (FN) receive antibiotics within 60–120 minutes of emergency department (ED) presentation.1,2 This recommendation is based on research findings indicating that the longer an individual has FN before receiving empiric treatment, the greater the risk of serious infection.3,4 Although only a few studies have demonstrated a reduction in adverse outcomes when time to antibiotic administration (TTA) is minimized in patients with FN,5–7 numerous studies in individuals without neutropenia presenting with infection show a correlation between shorter TTA and improved clinical outcomes.8,9 Despite the importance of prompt antibiotic administration, however, multiple studies demonstrate that individual hospitals are inadequate at consistently meeting this goal in patients who are immunocompromised, such as those receiving chemotherapy.6,10–13
Our institution treats a large population of children receiving chemotherapy who develop FN. These patients are triaged through our pediatric hematology/oncology (PHO) clinic during clinic hours and are evaluated and treated in their local EDs, including our institution’s pediatric ED (PED), after clinic hours and on weekends. Presentation to a PED instead of a PHO clinic presents several unique challenges: for example, PED healthcare providers are not specifically trained in oncology; point-of-care complete blood counts (CBCs) are not readily available for patients presenting to the PED; and oncology patients have to compete with other patients for PED staffing and resources.
When the PHO clinic is closed, our standard of care for pediatric patients with fever (defined as one episode of temperature >38.3°C or two episodes of temperature >38.0°C measured 1 hour apart) who are on chemotherapy or post–stem cell transplant (SCT) and at risk for neutropenia is to direct these patients to the PED. Parents of such patients are advised to call the on-call PHO provider, who, in turn, calls the PED to communicate about the patient to PED staff.
Prior to late 2009, the PED staff would draw blood from the patient’s central line(s) promptly after arrival to carry out complete blood count (CBC) and blood culture(s) and then wait for absolute neutrophil count (ANC) results before proceeding with antibiotic preparation and administration in the PED. Neutropenia is determined when ANC <500/µL; ANC is calculated as the total number of white blood cells per 1 µL, multiplied by the combined percentage of “segs,” or mature neutrophils, and “bands,” or immature neutrophils. Based on the ANC results, patients with confirmed FN were admitted to the hospital’s PHO unit where broad spectrum antipseudomonal antibiotics were continued; patients without neutropenia were typically administered a dose of ceftriaxone in the PED and discharged to home with close PHO clinic follow-up.
For patients with confirmed FN, the goal in place at our institution has been a TTA within <60 minutes of the patient’s arrival to the PED. It was generally perceived, however, that this goal was not consistently met prior to 2009, although it was never studied. Previous research has shown that a delay in obtaining laboratory results was the chief cause of TTA delays.14
In anticipation of the H1N1 influenza virus epidemic in late 2009 and the related expected increased PED utilization and infectious exposure, we aimed to streamline the flow of patients with fever and cancer through the PED and improve the quality of care by decreasing TTA. Notably, a subsequent study from another institution showed a statistically significant increase in TTA during the 2009 epidemic of H1N1 influenza virus.14 A clinical care pathway was developed and implemented to endorse the administration of antibiotics without waiting for the patient’s ANC results. The goal of this pathway was to reduce TTA and PED length of stay (LOS).
This retrospective study was approved by the Oregon Health & Science University Research Integrity Office as expedited research prior to data collection in October 2012. Data for the present study were acquired with assistance from the National Institutes of Health–funded Clinical and Translational Science Award: Oregon Clinical & Translational Research Institute.
The authors formulated the following primary hypotheses: (1) TTA would be significantly reduced in patients seen after the implementation of the pathway (post-implementation cohort) compared with those seen prior to the implementation of the pathway (pre-implementation cohort); (2) the percentage of patients achieving goal TTA of <60 minutes after PED arrival would be significantly greater for the post-implementation cohort than the pre-implementation cohort; and (3) PED LOS would be significantly shorter for the post-implementation cohort compared with the pre-implementation cohort. The secondary objectives of the study were to evaluate whether these hypothesized improvements were sustained over time, whether neutropenic status impacted results, and whether patient outcomes differed before and after pathway implementation.
Clinical Care Pathway Implementation
A clinical care pathway was created, developed, and approved collaboratively by the hospital’s PED and division of PHO (Figure 1). The decision-making algorithm included instructions to administer antibiotics to the patient within 60 minutes of arrival at the PED and without waiting for ANC results; the standard procedures were otherwise unchanged. This pathway did not discriminate between patients with low- and high-risk FN.
In preparation for pathway implementation, the algorithm was reviewed by nurses and physicians at routine PED meetings, physically posted in the PED, and e-mailed to staff. A new electronic medical record (EMR) order set for order entry by nurses or physicians was created to reflect the new algorithm. Orientation of new PED staff subsequently included training on this algorithm.
During the year following pathway implementation, a formulary change from ceftazidime to cefepime as the cephalosporin with antipseudomonal coverage was instituted hospital-wide.
To determine the optimal sample size and statistical power, a random sample population of 60 pediatric patients with fever on chemotherapy or post-SCT (30 patients seen before pathway implementation and 30 patients seen after pathway implementation) was initially obtained to serve as an a priori. This assessment determined that 100 episodes for each patient cohort would be sufficient for meaningful statistical analysis.
A query was created to systematically obtain computerized data from the Research Data Warehouse, a repository of data from our institution’s electronic medical records (EMR), and Clarity (a data store). Search terms for the query were the same as the inclusion criteria. The desired population was captured by querying all patients with an oncologic diagnosis (by International Classification of Diseases, Ninth Revision [ICD-9] code) and at least one previous encounter within our division of PHO. Once this population was identified, it was narrowed down to include only those individuals who were examined in the PED for a medical reason pertinent to our study. Pertinent patient encounters included those in which fever, neutropenia, or FN were coded as the chief report, were specified in the problem list, and/or were the primary diagnosis at presentation (by ICD-9 code). Finally, patients had to have a CBC obtained at their PED visit in order to be included in the study.
All patient encounters that met the inclusion criteria from the 12-month period from September 1, 2008, to August 31, 2009 were included in the pre-implementation cohort (n=121), and those from the 12-month period from January 1, 2010, to December 31, 2010 were included in the post-implementation cohort (n=158), in order to control for seasonal variability in the PED. A 4-month washout period surrounded the implementation date of October 31, 2009. Patient encounters that met the inclusion criteria from the 12-month period from November 1, 2011, to October 31, 2012, were assigned to a late cohort (n=197), in order to evaluate the sustainability of pathway implementation. Within each cohort, we delineated patient encounters by whether or not individuals had neutropenia.
To determine the TTA, the EMR was searched for pertinent antibacterial antibiotics (ie, ceftriaxone, cefepime, ceftazidime, vancomycin, and meropenem), and the PED arrival time was subtracted from the time when the antibiotics were administered. PED LOS was calculated as the time interval between the PED arrival time and the discharge time. All of these times were collected from the patients’ EMRs as timestamps, which had been entered once the respective action had been performed.
A second query was performed to capture individual timestamps to examine workflow through the PED. These timestamps included the time the patient was checked into the PED, the time the patient was roomed, the time patient’s laboratories were collected, the time the antibiotics were administered, the time the laboratory results were available, and the time the patient was discharged from the PED. For this study, pathway deviation was defined as encounters in which antibiotics were not administered at all or were administered at a time later than the time when the final CBC results were available.
For descriptive analysis, median/interquartile range for quantitative variables and frequency (%) for qualitative/categorical variables were computed. A nonparametric Wilcoxon rank-sum test was used to calculate the median comparisons between groups (eg, pre-implementation cohort vs post-implementation cohort, post-implementation cohort vs late cohort) when examining TTA and PED LOS variables. A chi-square/Fisher’s exact test was used to determine frequency distribution comparisons between groups (eg, pre-implementation cohort vs post-implementation cohort, post-implementation cohort vs late cohort) when examining achievement of goal TTA. Further multivariable regression analysis was applied to examine the potential confounding factors of age, diagnosis, gender, and admission status, all of which may affect group comparisons of TTA, PED LOS, and/or achievement of goal TTA.
Cohort demographics and characteristics are listed in Table 1. Mean age differed significantly between all three cohorts (pre-implementation cohort, 7.9 years; post-implementation cohort, 6.2 years; late cohort, 7.7 years; P<.01 for both comparisons). The proportions of patients who were post-SCT also differed significantly between the pre-implementation cohort (2%) and the post-implementation cohort (13%; P<.005). All other demographics were similar between the patient cohorts.
Patients received antibiotics in similar percentages of patient encounters in each cohort (pre-implementation cohort, 90%; post-implementation cohort, 92%; late cohort, 90%), although the specific antibiotics that were used differed (Table 2). Protocol adherence was similar in the post-implementation cohort and the late cohort; antibiotics were not administered or were administered after ANC results were available in 26% and 28% of patient encounters in the post-implementation cohort and the late cohort, respectively.
Median TTA significantly declined between the pre-implementation cohort (115 minutes) and the post-implementation cohort (60 minutes; P<.0001; Table 3). This decrease was sustained in the late cohort (59 minutes; vs post-implementation cohort, P=.69). Median TTA was similar between encounters with neutropenic patients and with non-neutropenic patients within the pre-implementation cohort (111 minutes vs 117 minutes; P =.18) and within the post-implementation cohort (60 minutes vs 58.5 minutes; P=.68). Within the late cohort, however, median TTA among encounters with neutropenic patients was significantly shorter than among encounters with non-neutropenic patients (53.5 minutes vs 64.5 minutes; P<.0001).
The percentage of patient encounters meeting the goal TTA of <60 minutes after the patient’s arrival to the PED increased significantly following pathway implementation (pre-implementation cohort, 12%; post-implementation cohort, 46%; P<.0001; neutropenic subset: preimplementation cohort, 15%; post-implementation cohort, 44%; P=.0002) and was sustained in the late cohort (46%; vs post-implementation cohort, P=.9). The percentages of patient encounters that met goal TTA were similar for patients with and without neutropenia within the pre-implementation cohort (15% vs 11%; P=.13) and within the post-implementation cohort (44% vs 46%; P=.84). Within the late cohort, however, the percentage of patient encounters that met goal TTA was higher for patients with neutropenia than for patients without neutropenia (69% vs 34%; P<.0001), and a higher percentage of encounters in the late cohort met goal TTA than in the post-implementation cohort (69% vs 44%; P=.048).
Despite a reduction in median TTA, no differences were observed in median PED LOS between the pre- and post-cohorts and between the post- and late cohorts (pre-implementation cohort, 200 minutes; post-implementation cohort, 196.5 minutes; late cohort, 187 minutes; Table 4). The median PED LOS was similar for patients with and without neutropenia within the pre-implementation cohort (203 minutes vs 197.5 minutes; P=0.8) and within the post-implementation cohort (192 minutes vs 203.5 minutes; P=0.9). Within the late cohort, however, PED LOS was significantly shorter for neutropenic patients than for non-neutropenic patients (177 minutes vs 198 minutes; P=.01).
Using the calculated median times (in minutes) for intake, lab collection, antibiotic preparation, antibiotic
administration, review of CBC results, and medical decision-making, we constructed timelines of patient encounters in the PED for each of the three cohorts (Figure 2). The timelines indicate that elimination of wait time for CBC results, along with reductions in intake and laboratory collection times, contributed to decreased TTA in the two cohorts seen after pathway implementation. The increased time between antibiotic administration and PED discharge for the two cohorts seen after pathway implementation contributed to the lack of a corresponding decrease in PED LOS.
The percentages of patients hospitalized in each cohort did not differ significantly between the cohorts. Similarly, the percentages of patients with positive blood cultures were equivalent in all cohorts. Within and between each of the three cohorts, a similar percentage of blood cultures were positive in encounters with and without neutropenia.
Table 5 lists all isolated pathogens and the PED antibiotic received. Among patients with positive cultures, 2 patients from the post-implementation cohort and 1 patient from the late cohort required initial stabilization in the pediatric intensive care unit (PICU). No patients died during their infectious episode.
Due to the relatively early versions of the EMR during the time periods when patients in the pre- and post-implementation cohorts were seen, we were unable to collect data on allergic reactions and Clostridium difficile infections. These data, however, were captured among patients in the late cohort, and we identified three episodes of C. difficile colitis. Each of these infections occurred after multiple days of hospitalization on antibiotics.
When comparing the pre-implementation and post-implementation cohorts, age, diagnosis, gender, and admission status were not found to be confounding variables by multivariable regression for TTA, PED LOS, and proportion of patient encounters meeting goal TTA. When comparing the post-implementation cohort with the late cohort, however, age and diagnosis were confounding variables for TTA and PED LOS; age, diagnosis, gender, and admission status were confounding variables for the proportion of patient encounters meeting goal TTA. Neutropenic status was not a confounding variable for TTA, PED LOS, or the proportion of patient encounters meeting goal TTA among the different cohorts based on an adjusted estimate.
We hypothesized that both median TTA and PED LOS would significantly decrease after implementation of a pathway that endorses administration of broad-spectrum antipseudomonal antibiotics to febrile and potentially neutropenic patients without waiting for the patient’s ANC results. However, only TTA was improved after implementation of the pathway. The significant decrease in TTA led to a direct increase in the percentage of patients meeting our institution’s goal TTA of <60 minutes after PED presentation. Both the reduction in TTA and the increased percentage of patients meeting our institution’s goal TTA were sustained over the 2–3 years after pathway implementation.
Despite the significant increase in percentage of patient encounters reaching goal TTA after pathway implementation, TTA remained >60 minutes after PED presentation in half the cases. We found that 25% of patient encounters deviated from the pathway, in that antibiotics were either not administered or were administered only after CBC results became available. We were unable to determine from the data we collected whether these late antibiotic administrations were due to intentional health care provider deviations from the pathway or system delays. Notably, the preparation of antibiotics during the time frame of our study required either an additional PED nurse to prepare the antibiotic dose for administration or the central pharmacy to prepare and deliver the antibiotic to the PED.
Interestingly, there were several unforeseen beneficial outcomes associated with implementation of the pathway. First, the decrease in TTA and the increase in percentage of patients meeting goal TTA were greater among patients in the late cohort with neutropenia. The cause of this observation is unclear, as the pathway did not distinguish between patients with neutropenia and patients without neutropenia. It is possible that patients with neutropenia might have presented with a more toxic appearance, leading to more expedited care. Further, it is possible that recent documentation of neutropenia was included in a greater proportion of the EMRs of patients with neutropenia than of patients without neutropenia and that the former group of patients were triaged more quickly. The second unforeseen outcome was that, in the late cohort, patients with neutropenia experienced shorter PED LOS when compared with patients without neutropenia. The cause of this disparity is again unclear, but it may reflect a period of PED observation before making the final decision to discharge the febrile patient without neutropenia to home.
Our finding that PED LOS was not affected by the implementation of the pathway was contrary to our hypothesis. To explore this result further, we performed a query to define the timelines of PED encounters in each of the three cohorts. We found that, despite a shorter TTA after pathway implementation, the time interval between the availability of a patient’s CBC results and his or her discharge increased. Thus, a sole reduction in TTA was not sufficient to reduce PED LOS. Notably, a reduction in intake and laboratory collection times after pathway implementation was observed in our timelines, suggesting that adherence to a standardized pathway, regardless of the specific practice change, may in and of itself have clinical benefit.
When designing the FN pathway, we recognized the potential risk of greater antibiotic exposure than existed previously in the subgroup of patients without neutropenia. However, we found that the same percentage of patients without neutropenia received a dose of antibiotics in our PED before and after pathway implementation. Based on our outcomes data for the subset of patients without neutropenia, the change from ceftriaxone pre-implementation to one dose of ceftazidime or cefepime after pathway implementation appeared to have no clinical impact on these patients. It is well documented that the implementation of a consistent FN pathway when none previously existed can reduce TTA.5,15,16 However, the majority of relevant published studies focus on adjusting existing pathways to improve quality metrics rather than creating new ones.6,17 Studies from 201218 and 201519,20 relied on pathway revision through multidisciplinary approaches. Although multidisciplinary approaches have been successful in reducing TTA in individual centers, their implementation may not be generalizable given the number of parties and variables involved. Our FN pathway changed a single practice variable, which was sufficient to significantly effect a desired outcome metric. This pathway appears to be novel; to our knowledge, an FN approach to administer antibiotics before the CBC results are reviewed has been described in the literature only once, and in concert with multiple modifications of a clinical practice in adult cancer patients.20 In addition, our study is the first to explore and document sustainability of outcomes (>2 years) after implementing a practice change involving an FN pathway.
The retrospective nature of our data collection is a study limitation. However, we used a robust data-mining program to capture >100 separate encounters in each of the three 1-year cohorts. Because of the large number of patient encounters with potential FN in the PED at our institution available for query, we were able to ensure sufficient statistical power based on our a priori hypotheses. Although we controlled for PED seasonal variability by including a full year of consecutive patients in each cohort, we did not control for the specific time of day and we did not have access to information on PED volume during each patient’s PED encounter. Confounding variables did not influence the pre- and post-implementation cohort comparisons of our three primary outcome measures. Although we identified age and diagnosis as confounding variables between the post-implementation and late cohorts, the large cohort size and robust statistical power minimized the impact of these variables. We chose not to control for patients with repeat PED encounters, as each individual episode initiated a care pathway and the three cohorts had similar percentages of repeat encounters.
The study findings allowed us to recognize that pathway deviation was 28% in the late cohort. Since then, our PHO clinic and PED explored the potential barriers to reaching goal TTA, one of which was the time required for the pharmacy to prepare and deliver antibiotics. Our recently revised clinical pathway allows for a PED nurse or healthcare provider to order antibiotics in the patient’s EMR before his or her arrival to the PED, based on his or her most recent documented weight. Since the implementation of this change, 100% of pediatric cancer patients with fever and potential neutropenia who have presented to our PED (n=16) have met the goal TTA of <60 minutes after arrival.
The implementation of this FN pathway significantly shortened TTA in our PED, as hypothesized. There were no deaths and an insignificant number of PICU admissions in our three cohorts; thus, we could not demonstrate a benefit of this FN pathway on morbidity and mortality. This pathway, wherein antibiotics are administered in the PED to oncology patients with a fever prior to resulted ANC, caused no documented increases in adverse effects. Because it significantly reduced the mean TTA and increased the percentage of encounters achieving goal TTA, this clinical pathway remains our institution’s standard approach in the PED.
1. Freifeld AG, Bow EJ, Sepkowitz KA, et al. Clinical practice guideline for the use of antimicrobial agents in neutropenic patients with cancer: 2010 update by the Infectious Diseases Society of America. Clin Infect Dis. 2011;52(4):e56-e93.
2. Segal BH, Freifeld AG, Baden LR, et al. Prevention and treatment of cancer-related infections. J Natl Compr Canc Netw. 2008;6(2):122-174.
3. Meckler G, Lindemulder S. Fever and neutropenia in pediatric patients with cancer. Emerg Med Clin North Am. 2009;27(3):525-544.
4. Bodey GP, Buckley M, Sathe YS, Freireich EJ. Quantitative relationships between circulating leukocytes and infection in patients with acute leukemia. Ann Intern Med. 1966;64(2):328-340.
5. Pakakasama S, Surayuthpreecha K, Pandee U, et al. Clinical practice guideline for children with cancer presenting with fever to the emergency room. Pediatr Int. 2011;53(6):902-905.
6. Fletcher M, Hodgkiss H, Zhang S, et al. Prompt administration of antibiotics is associated with improved outcomes in febrile neutropenia in children with cancer. Pediatr Blood Cancer. 2013;60(8):1299-1306.
7. Salstrom JL, Coughlin RL, Pool K, et al. Pediatric patients who receive antibiotics for fever and neutropenia in less than 60 min have decreased intensive care needs. Pediatr
8. Battleman DS, Callahan M, Thaler HT. Rapid antibiotic delivery and appropriate antibiotic selection reduce length of hospital stay of patients with community-acquired pneumonia: link between quality of care and resource utilization. Arch Intern Med. 2002;162(6):682-688.
9. Rivers E, Nguyen B, Havstad S, et al. Early goal-directed therapy in the treatment of severe sepsis and septic shock. N Engl J Med. 2001;345(19):1368-1377.
10. Szwajcer D, Czaykowski P, Turner D. Assessment and management of febrile neutropenia in emergency departments within a regional health authority—a benchmark analysis. Curr Oncol. 2011;18(6):280-284.
11. Nirenberg A, Mulhearn L, Lin S, Larson E. Emergency department waiting times for patients with cancer with febrile neutropenia: a pilot study. Oncol Nurs Forum. 2004;31(4):711-715.
12. Perrone J, Hollander JE, Datner EM. Emergency Department evaluation of patients with fever and chemotherapy-induced neutropenia. J Emerg Med. 2004;27(2):115-119.
13. Baltic T, Schlosser E, Bedell MK. Neutropenic fever: one institution’s quality improvement project to decrease time from patient arrival to initiation of antibiotic therapy. Clin J Oncol Nurs. 2002;6(6):337-340.
14. Burry E, Punnett A, Mehta A, Thull-Freedman J, Robinson L, Gupta S. Identification of educational and infrastructural barriers to prompt antibiotic delivery in febrile neutropenia: a quality improvement initiative. Pediatr Blood Cancer. 2012;59(3):431-435.
15. Zuckermann J, Moreira LB, Stoll P, Moreira LM, Kuchenbecker RS, Polanczyk CA. Compliance with a critical pathway for the management of febrile neutropenia and impact on clinical outcomes. Ann Hematol. 2008;87(2):139-145.
16. Cohen C, King A, Lin CP, Friedman GK, Monroe K, Kutny M. Protocol for reducing time to antibiotics in pediatric patients presenting to an emergency department with fever and neutropenia: efficacy and barriers [published online ahead of print March 27, 2015]. Pediatr
Emerg Care. doi: 10.1097/PEC.0000000000000362.
17. Hawley EL, Loney M, Wiece M. Development of tools and processes to improve treatment times in patients with febrile neutropenia. Clin J Oncol Nurs. 2011;15(5):E53-E57.
18. Volpe D, Harrison S, Damian F, et al. Improving timeliness of antibiotic delivery for patients with fever and suspected neutropenia in a pediatric emergency department. Pediatrics. 2012;130(1):e201-e210.
19. Jobson M, Sandrof M, Valeriote T, Liberty AL, Walsh-Kelly C, Jackson C. Decreasing time to antibiotics in febrile patients with central lines in the emergency department. Pediatrics. 2015;135(1):e187-e195.
20. Keng M, Thallner R, Elson P, et al. Reducing time to antibiotic administration for febrile neutropenia in the emergency department. J Onc Pract. 2015;11(6):450-455.