Cohort Study of the Impact of Time to Antibiotic Administration on Mortality in Patients with Febrile Neutropenia

Abstract

The time to antibiotic administration (TTA) has been proposed as a quality-of-care measure in febrile neutropenia (FN); however, few data regarding the impact of the TTA on the mortality of adult cancer patients with FN are available. The objective of this study was to determine whether the TTA is a predictor of mortality in adult cancer patients with FN. A prospective cohort study of all consecutive cases of FN, evaluated from October 2009 to August 2011, at a single tertiary referral hospital in southern Brazil was performed. The TTA was assessed as a predictive factor for mortality within 28 days of FN onset using the Cox proportional hazards model. Kaplan-Meier curves were used for an assessment of the mortality rates according to different TTAs; the log-rank test was used for between-group comparisons. In total, 307 cases of FN (169 subjects) were evaluated. During the study period, there were 29 deaths. In a Cox regression analysis, the TTA was independently associated with mortality within 28 days (hazard ratio [HR], 1.18; 95% confidence interval [CI], 1.10 to 1.26); each increase of 1 h in the TTA raised the risk of mortality within 28 days by 18%. Patients with FN episodes with a TTA of ≤30 min had lower 28-day mortality rates than those with a TTA of between 31 min and 60 min (3.0% versus 18.1%; log-rank P = 0.0002). Early antibiotic administration was associated with higher survival rates in the context of FN. Efforts should be made to ensure that FN patients receive effective antibiotic therapy as soon as possible. A target of 30 min to the TTA should be adopted for cancer patients with FN.

INTRODUCTION

Febrile neutropenia (FN) may represent the only sign of severe infection in cancer patients, because the symptoms and signs of inflammation are typically attenuated due to a reduced absolute neutrophil count (ANC) (1, 2). The frequent need for indwelling central venous catheters in association with damage to the gastrointestinal mucosa caused by anticancer agents provides a portal of entry for pathogenic bacteria, which in turn predispose patients to bacteremia (3). This fact, in association with an impaired host response to infection due to neutropenia and decreased cellular immunity secondary to intensive chemotherapy, leads to increased risk for severe infections in cancer patients. Despite recent improvements in managing FN, infections in the context of neutropenia continue to be associated with substantial mortality, which may reach values of approximately 10% in specialized centers (4).

The time to antibiotic administration (TTA) is a well-recognized determinant of mortality in patients with severe sepsis or septic shock (5). Current guidelines for treatment of sepsis and FN (6, 7), based mainly on studies involving immunocompetent subjects, recommend administration of broad-spectrum β-lactam antibiotic monotherapy with antipseudomonal activity within 1 h after the onset of fever in the neutropenic patients; furthermore, many oncology centers use a benchmark of <60 min to antibiotic administration (8, 9). However, few studies investigating the impact of the TTA on the mortality of patients with FN are available. Moreover, the ideal TTA in this setting is not well established, which raises questions about the potential benefit of TTAs shorter than those recommended by current guidelines. Accordingly, we performed the present study to determine whether the TTA is a predictor of mortality in adult cancer patients with FN and to define the optimal TTA for patients with FN secondary to cytotoxic chemotherapy.

MATERIALS AND METHODS

Study design, patients, and setting.

A prospective cohort study was conducted at a single tertiary center. The present study followed all cancer patients >18 years of age who were consecutively admitted to the hematology ward of the Hospital de Clínicas de Porto Alegre (Porto Alegre, Brazil) with neutropenia (i.e., an ANC of <500 cells/mm³ or of <1,000 cells/mm³ with the expectation of a decrease to <500 cells/mm³ during the ensuing 48 h) and fever (i.e., a single axillary temperature measurement of ≥38.5°C or sustained temperature of ≥38.0°C over a 1-h period). Subjects who were receiving palliative treatment only, had an indication for outpatient treatment, or had neutropenia due to a specific etiology other than an adverse reaction to chemotherapy were excluded. Subjects were allowed to reenter the study after an initial episode of FN if they remained free from the signs or symptoms of infection for at least 7 days after completing the treatment for the first episode and if all causative organisms (if any) were eradicated.

Treatment protocol.

In our institution, the immediate antibiotic administration in patients with FN is recommended. To ensure the shortest possible TTA, a protocol for the management of FN was created with the involvement of nursing, medicine, and pharmacy teams. According to this protocol, all neutropenic patients were routinely screened for fever by the nursing staff (every 4 h or every 20 min if they had a temperature of ≥37.5°C) with a standardized axillary thermometer calibrated for the range of 32.0 to 42.0°C, in steps of 0.1°C. All neutropenic patients were evaluated straightaway after the onset of fever by a medical fast response team, which defined the initial antimicrobial therapy; a pharmacy unit within the hematology ward immediately dispensed the antibiotic regimen prescribed. FN patients were treated according to the 2002 guidelines of the Infectious Diseases Society of America (10). The initial antimicrobial treatment regimen was performed using β-lactam monotherapy with antipseudomonal activity (cefepime, piperacillin-tazobactam, or a carbapenem); vancomycin was recommended as part of the initial empirical regimen only in cases with hemodynamic instability, suspected catheter-related infection, or infection of the skin and soft tissue. In patients with clinically or microbiologically documented infections, the duration of antimicrobial therapy was dictated by the particular organism and site of infection; appropriate antibiotics were continued until the ANC was ≥500 cells/mm³. Empirical antifungal therapy with amphotericin B deoxycholate was administered to patients with persistent fever after 4 days of treatment with broad-spectrum antibiotics. Antibacterial prophylaxis was not administered to any patient. Antifungal prophylaxis with fluconazole was routinely performed for patients in whom the anticipated duration of neutropenia was >7 days. Acyclovir antiviral prophylaxis was administered for herpes simplex virus-seropositive patients undergoing allogeneic hematopoietic stem cell transplantation or leukemia induction chemotherapy.

Definitions.

The primary independent variable in the study was the TTA, which was defined as the time between the onset of fever in the neutropenic patients and antibiotic administration. The Multinational Association for Supportive Care in Cancer (MASCC) risk index score (11) was applied at the onset of fever to determine the risk for serious complications during FN; episodes were classified as high risk if the score was <21 points and as low risk if the score was ≥21 points. Clinical comorbidity was defined as the presence of heart failure, diabetes mellitus, chronic pulmonary disease, chronic liver disease, or chronic renal failure. The patients were divided into two groups based on their chemotherapy regimen: the high-dose chemotherapy group included patients who underwent hematopoietic stem cell transplantation or induction chemotherapy and the standard-dose chemotherapy group included patients who underwent consolidation or maintenance chemotherapy. Microbiological studies were performed at the onset of fever according to the standards of practice and included two separate blood samples drawn from two different sites for aerobic culture. In our institution, anaerobic blood culture is not routinely performed due to the low incidence of anaerobic bacteremia even in the context of FN (data not shown). In the absence of an indwelling central venous catheter, the two blood samples were obtained from two distinct peripheral veins. When an indwelling central venous catheter was present, one sample for blood culture was obtained through the indwelling central venous catheter, and the other was collected from a peripheral vein. Bacteremia caused by coagulase-negative Staphylococcus spp. was defined as two positive results from two independent cultures. Bacteremia in one positive culture was considered to be diagnostic for other microorganisms. Antibiotic susceptibilities of the isolated pathogens were evaluated according to the recommendations of the Clinical and Laboratory Standards Institute (12).

Outcome and follow-up.

The primary outcome of the study was all-cause mortality 28 days after the onset of FN. The patients were followed up through interviews and medical record reviews, using a standardized case report form, by researchers who were not associated with the assistant physician's team. The follow-up was maintained for 28 days after the onset of fever in the neutropenic patients. For subjects who were discharged before 28 days, follow-up telephone calls were made on the 28th day after the onset of FN to determine whether they remained alive. If a patient was deceased at the time of the telephone call, the survival time was calculated based on the date of death reported by the family.

Statistical analysis.

A multivariate Cox proportional hazards model was used to determine whether the TTA was a predictor of mortality within 28 days of the onset of FN. All variables with P < 0.15 in the univariate analysis were included. In the multivariate model, independent variables were eliminated from the highest to the lowest P value but remained in the model if P < 0.05. The hazard ratios (HRs) were estimated along with the 95% confidence intervals (CIs). Kaplan-Meier curves were used to evaluate the time-dependent occurrence of death according to the TTA; the log-rank test was applied for between-group comparisons. The Bonferroni correction was applied as a protection against multiple comparisons (i.e., type I error). Given that two comparisons of the mortality rate according to the TTA were planned (a TTA of 31 min to 60 min versus a TTA of ≤ 30 min in all cases of FN and a TTA of 31 min to 60 min versus a TTA of ≤ 30 min only in cases with the first episode of documented bacteremia), the adjusted α level for the log-rank test was set at 0.025. STATA version 12 (Stata Corp LP, USA) was used for statistical analysis.

Ethics issues.

The institutional review board of the Hospital de Clínicas de Porto Alegre approved the study, and written informed consent was obtained from all study participants.

RESULTS

During the study period, 307 cases of FN (169 patients) were analyzed. Seventy-one subjects (42% of the study population) had two or more episodes of FN. The characteristics of all episodes of FN are shown in Table 1. Hematological malignancies accounted for most cases of cancer (78.8%). The predominant neoplastic diseases were acute myeloid leukemia (48.5%), lymphoma (16.6%), acute lymphoblastic leukemia (14.6%), and multiple myeloma (9.7%). High-dose chemotherapy regimens were performed in 53.4% of the study population. FN occurred after 48 h of hospitalization in 81.4% of cases. Bloodstream infections were responsible for 37.4% of all episodes of FN: in descending order, the most common blood isolates were Escherichia coli, coagulase-negative staphylococci, Klebsiella pneumoniae, Pseudomonas aeruginosa, viridans streptococci, and Enterococcus spp. All patients with episodes of FN were treated as inpatients for the entire course. No case with an incorrect antibiotic dose, according to the institutional FN treatment protocol, was documented in the present cohort.

TABLE 1.

Study population characteristics and microorganisms isolated in 307 cases of febrile neutropenia

Characteristica	Patient data
Age (mean ± SD) (yr)	40.7 ± 14.2
Female sex (no. [%])	148 (48.2)
Type of cancer (no. [%])
Acute myeloid leukemia	149 (48.5)
Acute lymphoblastic leukemia	45 (14.6)
Chronic myeloid leukemia	18 (5.8)
Multiple myeloma	30 (9.7)
Lymphoma	51 (16.6)
Other solid tumors	14 (4.5)
Relapsing underlying disease (no. [%])	155 (50.4)
Clinical comorbidity (no. [%])	76 (24.7)
Phase of chemotherapy (no. [%])
Induction	76 (24.7)
Consolidation	86 (28.0)
Maintenance	57 (18.6)
HSCT	88 (28.7)
ANC at the time of diagnosis of FN (median [IQR]) (cells/mm3)	130 (260)
Duration of neutropenia (median [IQR]) (days)	9 (12)
Nosocomial acquired episodes of FN (no. [%])	250 (81.4)
Bloodstream infection (no. [%])b	115 (37.4)
Escherichia coli	48 (41.7)
Coagulase-negative staphylococci	36 (31.3)
Klebsiella pneumonia	13 (11.3)
Pseudomonas aeruginosa	11 (9.5)
Viridans streptococci	8 (6.9)
Enterococcus spp.	4 (3.4)
Serratia spp.	2 (1.7)
Enterobacter spp.	2 (1.7)
Candida spp.	2 (1.7)
Salmonella spp.	1 (0.8)
Staphylococcus aureus	1 (0.8)
Kocuria varians	1 (0.8)

^aHSCT, hematopoietic stem cell transplantation; ANC, absolute neutrophil count; FN, febrile neutropenia; IQR, interquartile range (75th to 25th percentile).

^bThere were 12 cases of polymicrobial bloodstream infections.

During the study period, there were 29 deaths (9.4% of all cases of FN or 17.1% of all study patients), although no deaths occurred during the coincident follow-up period of ≥2 FN episodes in the same patient. The assessment of whether mortality was attributable to infection was concordant in all 29 patients who died. In the univariate analysis of the risk factors for mortality within 28 days (Table 2), relapsing underlying disease status (P = 0.001), standard-dose chemotherapy regimens (P = 0.02), bloodstream infection (P = 0.001), and presentation with a high-risk MASCC score (P < 0.001) were more frequent in nonsurvivors. The TTA was also directly associated with the hazards for death (P < 0.001). After a multivariate analysis was conducted, the variables that constituted independent risk factors for mortality included relapsing underlying disease status (HR, 6.66; 95% CI, 2.36 to 18.79), a bloodstream infection (HR, 3.64; 95% CI, 1.49 to 8.88), a presentation with a high-risk MASCC score (HR, 4.21; 95% CI, 1.82 to 9.72), and the TTA (HR, 1.18; 95% CI, 1.10 to 1.26). Each hour of delay in the TTA raised the 28-day mortality risk by 18%. The distribution of TTAs for each episode of FN is shown in Fig. 1; the median TTAs were 0.33 h (interquartile range [IQR], 1.0 h) and 1.66 (IQR, 5.17 h) for survivors and nonsurvivors, respectively.

TABLE 2.

Predictors of 28-day mortality in febrile neutropenia patients according to Cox regression analysis

Variablea	Data by group:		Univariate analysis results		Multivariate analysis results
	Mortality group (n = 29)	Survival group (n = 278)	HR (95% CI)b	P	HR (95% CI)	P
Age (mean [SD]) (yr)	41.6 (15.0)	40.6 (14.1)	1.00 (0.98–1.03)	0.62
Female sex (no. [%])	13 (44.8)	135 (48.5)	0.85 (0.41–1.78)	0.68
Clinical comorbidity (no. [%])	6 (20.6)	70 (25.1)	0.82 (0.33–2.01)	0.66
Type of neoplastic disease (no. [%])
Hematologic	24 (82.8)	218 (78.4)	1.20 (0.45–3.15)	0.70
Solid tumor	5 (17.2)	60 (21.6)
Relapsing underlying disease status (no. [%])	23 (79.3)	132 (47.4)	4.30 (1.75–10.58)	0.001	6.66 (2.36–18.79)	<0.001
High-dose chemotherapy regimens (no. [%])	9 (31.0)	155 (55.7)	0.39 (0.17–0.86)	0.02
ANC at the time of FN diagnosis (median [IQR]) (cells/mm3)	130 (260)	130 (260)	1.00 (0.99–1.00)	0.37
ANC < 100 cells/mm3 at the time of FN diagnosis (no. [%])	16 (55.1)	114 (41.0)	1.68 (0.80–3.49)	0.16
Duration of neutropenia (median [IQR]) (days)	8 (16)	9.5 (11)	0.97 (0.93–1.01)	0.23
Bloodstream infection (no. [%])	20 (68.9)	95 (34.1)	3.91 (1.78–8.60)	0.001	3.64 (1.49–8.88)	0.004
High-risk MASCC score (no. [%])	18 (62.0)	65 (23.2)	5.03 (2.37–10.65)	<0.001	4.21 (1.82–9.72)	0.001
In vitro sensitivity of blood isolates to initial antibiotic treatment (no. [%])	22 (75.8)	246 (88.4)	0.47 (0.20–1.10)	0.08
Time to antibiotic (median [IQR]) (h)	1.66 (5.17)	0.33 (1.0)	1.14 (1.07–1.22)	<0.001	1.18 (1.10–1.26)	<0.001

^aANC, absolute neutrophil count; FN, febrile neutropenia; IQR, interquartile range (75th to 25th percentile); MASCC, Multinational Association for Supportive Care in Cancer.

^bHR, hazard ratio; CI, confidence interval.

FIG 1.

Distribution of time to antibiotic therapy in 307 cases of febrile neutropenia.

Figure 2 shows the subgroup analysis of mortality rate according to TTA. Patients with episodes of FN with a TTA of ≤ 30 min had lower mortality rates than those with a TTA of between 31 min and 60 min (3.0% versus 18.1%; log-rank P = 0.0002). A second analysis comparing a TTA between 31 min and 60 min versus a TTA of ≤ 30 min was performed only in cases with the first episode of documented bacteremia (Fig. 3). This procedure was conducted in order to confirm the benefit of early TTA in patients with documented bloodstream infections and to avoid possible confounding by evaluation of multiple episodes of FN; also in this analysis, a TTA of <30 min was associated with a decreased likelihood of death compared with a TTA of between 31 and 60 min (6.4% versus 27.2%; log-rank P = 0.020). The rates of in vitro sensitivity of blood isolates to the initial antibiotic treatment for cases with a TTA of ≤ 30 min and those with a TTA of between 31 min and 60 min were 87% and 90%, respectively (Fisher's exact P = 0.77).

FIG 2.

Comparison of survival curves of FN cases (all episodes of FN) with a time to antibiotic administration (TTA) of between 31 min and 60 min versus those with a TTA of ≤30 min. The Bonferroni-corrected α level is 0.025.

FIG 3.

Comparison of survival curves of FN cases (first episode of documented bacteremia) with a time to antibiotic administration (TTA) of between 31 min and 60 min versus those with a TTA of ≤30 min. The Bonferroni-corrected α level is 0.025.

DISCUSSION

In the present study, the TTA was independently associated with all-cause mortality within 28 days in the context of FN after cytotoxic chemotherapy: each hour of delay in the TTA raised the risk of death by 18%. Moreover, our study findings demonstrated that patients with a TTA of ≤ 30 min had lower mortality rates than those with a TTA of between 31 min and 60 min.

Previous research involving distinct populations has confirmed the impact of early antibiotic administration on clinical outcomes. For example, the study by Gaieski et al. (13) showed a relative risk reduction of mortality by 70% for nonneutropenic intensive care patients with severe sepsis and septic shock treated with an appropriate antibiotic within 1 h of triage compared with that for those with a TTA of >1 h (P = 0.02). The retrospective cohort in the study by Fletcher et al. (14) found that a TTA of ≤60 min, compared with a TTA of 61 min to 120 min, was associated with a lower incidence of a composite endpoint that included in-hospital mortality, intensive care unit admission, and fluid resuscitation of >40 ml/kg within 24 h of presentation in pediatric FN patients. In a study by Hamandi et al. (15), there was a significant association between increasing TTA (24-h increments) and increased hospital mortality rates in solid-organ transplant patients. Similarly, our study showed a reduction in the mortality risk from the early TTA in a specific population of high-risk neutropenic patients in whom the correct implementation of antimicrobial strategy is of paramount importance. Beyond that, we went further and evaluated the benefit of an earlier TTA compared with a TTA of ≤1 h recommended by current practice.

The considerable reduction in risk of mortality associated with a target TTA of ≤30 min (compared with the typical target of ≤1 h recommended by current guidelines) observed in the present study has scientific plausibility because a population with a high risk for morbidity and mortality due to severe and prolonged immunosuppression is expected to benefit from prompt administration of effective antimicrobial agents. This finding underscores the importance of validated institutional strategies focused on reducing any delay in starting antimicrobial treatment for neutropenic cancer patients (16–18).

The strengths of the present study include its prospective design, the implementation of follow-up by a research team that was not responsible for providing care for the patients, the proper measurement of variables and outcomes with previously defined objective criteria, and the appropriate use of subgroup analysis through a conservative approach to avoid type I error. The main limitations of the present study were related to its single-center design and possible systematic errors related to observational studies, given that we cannot be certain that we identified all potential confounding factors. Additionally, important pharmacokinetic aspects of the initial antimicrobial treatment administered (e.g., antibiotic serum levels) were not controlled.

Identifying the optimal TTA for patients with FN is of paramount importance to clinical practice because immediate antibiotic therapy is one of the few available treatments that effectively reduce mortality in the context of immunosuppression and infection. In addition, improvement in strategies focused on early, effective empirical antimicrobial administration usually requires nothing more than institutional organization and commitment. Taking into account the magnitude of these results, further research investigating strategies to ensure prompt, effective antibiotic administration within 30 min after the onset of FN in cancer patients undergoing cytotoxic chemotherapy is warranted.