Diagnosis and empirical treatment of fever of unknown origin (FUO) in adult neutropenic patients: guidelines of the Infectious Diseases Working Party (AGIHO) of the German Society of Hematology and Medical Oncology (DGHO)

1Neutropenic cancer patients have a high risk of infectious complications, depending on the extent and duration of neutropenia, as well as on additional cellular and/or humoral immunosuppression and disruption of skin and mucosal barriers. Duration and nadir of neutropenia are correlated with the incidence of fever and infections, which not only cause significant morbidity and mortality but may also compromise further chemotherapy. At onset of fever of unknown origin (FUO) in neutropenic patients, prompt and effective evidence-based intervention is required.

2In the past decade, an increasing rate of resistance among bacterial pathogens to widely used antibacterial agents, particularly beta-lactams and fluoroquinolones, has been noted. At the same time, only a small number of newer antimicrobial agents have become available, e.g., tigecycline, linezolid, moxifloxacin, cefozopran, telavancin, oritavancin, ceftazidime-avibactam, ceftolozan-tazobactam, or micafungin; however, data on their empirical use in neutropenic patients are limited or not existent. Efficacy and safety of anti-infective strategies have to be reassessed in this context. Newly introduced diagnostic tests and data on the usefulness of biomarkers for therapeutic decisions are to be critically reconsidered.

3This guideline, an update of a version from 2003, focuses on risk-adapted diagnostic procedures and empirical antimicrobial treatment in neutropenic cancer patients with FUO according to their likelihood of a complicated course of infection.

4Separate AGIHO guidelines for hematological and oncological patients have been published on sepsis, primary prophylaxis of bacterial and fungal infections, prevention of infections after allogeneic hematopoietic stem cell transplantation (HSCT), diagnosis and treatment of invasive fungal infections, management of pulmonary infiltrates, abdominal complications, venous catheter related infections, central nervous system infections, infections after autologous HSCT, and community respiratory viral infections.

5A group of hematologists, oncologists, and infectious disease specialists was built within the AGIHO, which after thorough literature search (including only full publications but excluding allogeneic HSCT), created a set of core slides with statements and recommendations, discussed in face-to-face meetings, telephone conferences, and by electronic correspondence. The final version was approved in an AGIHO plenary meeting on 10 February 2017. This manuscript was reviewed by all co-authors. A detailed methodological report is provided in the Electronic supplementary material.

6Consistent with recently updated AGIHO guidelines, the grading system currently used by the European Society for Clinical Microbiology and Infectious Diseases (Table 1) was adapted.

7There is no defined cut-off value for the neutrophil count clearly separating patients with or without increased risk of infections and mortality. In line with most recommendations and risk stratifications in clinical trials, a neutrophil count (segments and bands) < 500/μl or < 1000/μl with a predicted decline to < 500/μl within the next 2 days defines neutropenia.

8Different definitions of fever in neutropenia have been used in guidelines and clinical trials, and several methods and sites to determine the body temperature are available. In general, either a temperature measured orally of ≥ 38.3 °C once or ≥ 38.0 °C lasting for at least 1 h or being measured twice within 12 h or a method shown to be equivalent to these results may be used to define fever. In the absence of a definite non-infectious cause, such as a febrile reaction to cytokines, cytotoxic drugs (e.g., cytarabine or bleomycin), or a transfusion of blood products, this clinical symptom has to be regarded as a sign of an infectious complication. It should be kept in mind that fever may be obscured by antipyretic drugs used for analgesia or cancer treatment, such as prednisone, non-steroidal anti-inflammatory agents, or metamizole (dipyrone).

9While it is widely accepted that the incidence of infections in cancer patients is directly related to nadir and duration of neutropenia, it is difficult to exactly predict this in an individual patient. Clinical trials on treatment of fever and infections in patients with short periods of neutropenia, e.g., below 5 or 7 days, are limited, and some patients with neutropenia lasting for more than 5 days have also been enrolled in studies including oral therapy and outpatient care. As a result of a literature review, we have agreed upon stratification into two risk groups, i.e.,Standard risk: expected duration of neutropenia of up to 7 days andHigh risk: expected duration of neutropenia of at least 8 days.

10Standard risk: expected duration of neutropenia of up to 7 days and

11High risk: expected duration of neutropenia of at least 8 days.

12However, while all patients with neutropenia lasting eight or more days are regarded as high-risk patients with respect to a complicated course of a febrile episode, those assigned to the standard-risk group may exhibit individual characteristics justifying their allocation to the high-risk population as well. These individual factors can be identified by the use of the Multinational Association of Supportive Care in Cancer (MASCC) criteria, which have been repeatedly validated and are shown in Table 2.

13Standard-risk patients with a MASCC score of ≥ 21 constitute a group of patients with a high likelihood of a non-complicated clinical course of infection. Provided they meet all individual criteria listed in Table 3, primary outpatient management of neutropenic fever is possible (BIIr).

14At onset of fever, antibiotic therapy needs to be started immediately, and because of the time needed for microbiological tests, it will have to be empirical in the beginning in patients who also do not show a suspected clinical focus of infection. In about half of patients with febrile neutropenia, the antibiotic therapy will remain empirical, since no relevant pathogen or focus of infection can be identified during the following days. The grounds for selection of empirical antimicrobial agents are (a) reported results of prospective, randomized clinical studies and (b) microorganisms identified in patients with microbiologically documented infections by analogy. Here, Staphylococcus aureus, Streptococcus spp., enterococci, coagulase-negative staphylococci, gram-negative enterobacteria, and Pseudomonas aeruginosa are the most frequent and relevant pathogens. While numerically, coagulase-negative staphylococci are the most frequent microbial isolates in many institutions, a single blood culture positive for those commensal skin pathogens, in lack of a corresponding clinical focus of infection, should be considered contamination. The same is true for other potential contaminants like Corynebacterium, Bacillus cereus, Propionibacterium, or Micrococcus spp. Among fungal pathogens, Candida spp. and Aspergillus spp. are predominant, the latter typically being associated with a prolonged duration of neutropenia in high-risk patients.

15A recent history of antibiotic prophylaxis or therapy increases the risk of infections due to bacterial pathogens resistant to the antibiotic used. After ciprofloxacin prophylaxis, a relative predominance of infections caused by gram-positive cocci compared with gram-negative bacteria has been observed. Quinolones have been reported as being associated with an increased rate of colonization by vancomycin-resistant enterococci (VRE) or methicillin-resistant S. aureus (MRSA) and with a higher prevalence of multidrug resistance among enterobacteria via extended-spectrum beta-lactamases (ESBL). Colonization by ESBL, VRE, or MRSA has been associated with an increased rate of bacteremia with these pathogens. As a consequence, the use of quinolones for interventional treatment in febrile neutropenic patients should be limited to microbiologically documented infections caused by in vitro susceptible microorganisms.

16The local epidemiology must be taken into account for the appropriate choice of empirical antimicrobial therapy. Microbiological findings from patients treated in a defined hematology-oncology institution should be discussed on a regular basis, i.e., at least once a year, with infection-control and antimicrobial stewardship experts (BIII). Baseline screening of newly or re-admitted patients for multidrug-resistant pathogens, i.e., MRSA (BIII), VRE (BIII), and ESBL (BIIt), should be considered.

17Before starting myelosuppressive therapy, patients must be thoroughly explored for relevant previous or prevalent infections, which may become relevant during treatment-induced neutropenia (AIII). Clinical examination should be performed with special attention paid to skin, mucosa, puncture, and vascular catheter exit sites, paranasal sinuses, lungs, and the perianal region (AIII). In patients with a self-reported penicillin allergy, skin testing is recommended (BIIt), as a negative result (which is to be expected in the vast majority of cases) helps to avoid unnecessary first-line use of carbapenems, aztreonam, or vancomycin.

18Baseline laboratory tests include a blood count, liver enzymes (ASAT/SGPT, ALAT/SGOT, gGT), total bilirubin, alkaline phosphatase, LDH, creatinine, blood urea nitrogen, coagulation tests (INR, aPTT), C-reactive protein, and urinalysis (BIII). Except for urinalysis, it is recommended to repeat these tests regularly, e.g., twice a week, during long-lasting neutropenia (BIII). Procalcitonin or cytokine levels (such as interleukin-6) are not recommended for routine baseline diagnostics (DIII).

19If neither computed nor magnetic resonance tomography of chest and abdomen have been performed for staging of the underlying disease, chest radiographs (two views) and abdominal ultrasound may be considered a baseline examination before first chemotherapy to check for pre-existing abnormalities and facilitating comparison with subsequent studies (CIII). Particularly in high-risk patients, a thoracic CT scan prior to chemotherapy appears desirable for documentation of baseline status. However, in the absence of prospective studies, no recommendation can be made. In patients with a history of an invasive infection, appropriate imaging is recommended even in the absence of clinical symptoms of recurrence (BIII).

20For high-risk patients with an expected duration of profound neutropenia > 7 days, serial (at least twice weekly) monitoring for Aspergillus galactomannan in serum has been recommended. Monitoring patients with 1,3-beta-d-glucan in blood samples is being discussed as an alternative but rarely used due to higher costs. A sensitive, validated Aspergillus PCR may also be helpful (CIII) for screening of blood samples in specific high-risk populations. However, the sensitivity of these tests is strongly reduced in patients given systemic mold-active antifungals, and false-positive results may be caused by beta-lactam antibiotics, parenteral nutrition, severe intestinal mucositis, or transfusion of blood products. Therefore, screening of afebrile and asymptomatic patients should be restricted to those not receiving systemic mold-active prophylaxis (BIIu). These non-culture-based procedures do not replace clinical, imaging, endoscopic, or other microbiological diagnostics (BIII). Details on early diagnosis of fungal infections are discussed in a separate guideline.

21Diagnostic measures at first fever in neutropenic patients aim atRuling out non-infectious causes of feverIdentifying a clinical focus and/or causative pathogens, andAssessing the severity of inflammatory response in order to early identify patients in need for intensive care

22Ruling out non-infectious causes of fever

23Identifying a clinical focus and/or causative pathogens, and

24Assessing the severity of inflammatory response in order to early identify patients in need for intensive care

25They must not delay the start of appropriate antibiotic therapy (AIIt). In a clinically unstable patient, eventually presenting in the emergency room, prompt start of antimicrobial therapy is required (AI) and immediate referral to an intensive care unit must be considered.

26Thorough clinical examination must be updated (AIII) and repeated at least daily as long as a hospitalized patient is febrile (AIII). It may reveal a presumable focus of infection and enable a pre-emptive antimicrobial treatment targeting typically involved pathogens rather than purely empirical treatment (Table 4).

27A minimum of two separate pairs of blood cultures must be taken prior to initiation of antibiotic therapy (AIII). There is no need to wait between sampling of cultures; separate sets can be achieved by venipuncture of both arms. If the patient has an indwelling central venous catheter (CVC), one pair should be drawn from a peripheral vein and at least one from the CVC. The diagnostic yield of this approach can be increased by taking a blood sample from each lumen of a CVC and by taking three pairs of blood cultures (60 ml blood) (BIIt). A “differential time to positivity” of ≥ 2 h between CVC and peripheral blood cultures may indicate a CVC-related infection and give reason for pre-emptive treatment described in a separate guideline (BIIu). Multiplex PCR-based methods do not replace the standard microbiology (CIIu) but may improve turnaround time, sensitivity, and specificity of pathogen detection. A reduction in morbidity or mortality in febrile neutropenic patients through the use of PCR-based methods supplementing blood cultures has not yet been shown.

28In addition to repeat baseline laboratory tests described above, determination of lactate, blood gas analysis, and coagulation tests, in order to early identify severe sepsis, should be considered (BIII). Biomarkers such as procalcitonin or interleukin-6 are widely used for assessing the severity of inflammation, but data on their prognostic or predictive value in adult patients with febrile neutropenia are conflicting. In high-risk patients who did receive systemic mold-active antifungal prophylaxis and were not screened for Aspergillus galactomannan, beta-d-glucan, or fungal PCR, such a test (preferably galactomannan) should be ordered at this time to enable early detection of a breakthrough invasive fungal disease (BIII).

29At onset of fever, a CT scan of the lungs is recommended in the case of respiratory tract symptoms (BIII). Conventional chest radiographs are discouraged (DIIt), as they show abnormalities in less than 2% of febrile neutropenic patients who have no clinical signs of lower respiratory tract infection. Nasal congestion or signs and symptoms of sinusitis should give reason for a CT scan of paranasal sinuses (BIII). First data on PET-CT indicate a potential use for early identification of the source of fever/infection, particularly abdominal foci. Despite these positive reports, an explicit recommendation for its routine use cannot be given due to the lack of systematic studies.

30Gastrointestinal complaints or laboratory abnormalities should prompt abdominal ultrasonography (BIIu). An abdominal CT scan is an alternative if neutropenic enterocolitis is suspected (BIIu).

31The following recommendations are based on evidence from controlled studies and clinical experience. The status of approval by regulatory agencies and reimbursement policies have not been taken into account.

32A clinical treatment algorithm for high-risk patients is depicted in Fig. 1.

33In high-risk patients, the spectrum of first-line antibacterial agents should include gram-negative enterobacteria, P. aeruginosa, S. aureus, and streptococci (AI), while local epidemiology must be taken into account. Treatment must be started within 2 h after onset of fever, without awaiting the results of microbiology (AIIt). If oral fluoroquinolone prophylaxis has been given, it should be discontinued at the start of interventional antimicrobial therapy (AIII). Emergency treatment algorithms for this clinical situation as well as supervision or audits have been shown to improve timely and adequate therapy. Piperacillin/tazobactam, imipenem, meropenem, cefepime,1 and ceftazidime are suitable for first-line empirical antibacterial monotherapy in severely neutropenic high-risk patients with FUO (AI). Most patients with a history of penicillin allergy will tolerate imipenem, meropenem, or an antipseudomonal cephalosporin. For the small cohort of patients with severe immediate-type hypersensitivity reaction, aztreonam may be used as a—less well-studied—alternative (CIIu). In this setting, the addition of a vancomycin or teicoplanin to aztreonam may be considered due to the lack of activity of aztreonam against gram-positive bacteria (CIII).For newer broad-spectrum antibacterial agents, such as ertapenem, which has insufficient activity against Pseudomonas spp., doripenem, ceftazidime-avibactam, ceftolozane-tazobactam, or cefozopran, there is very limited data on their safety and efficacy for empirical treatment in adult febrile neutropenic cancer patients available so far. Tigecycline in combination with an antipseudomonal beta-lactam has shown benefit in terms of a lesser need for treatment modification in institutions with excess rates of multidrug-resistant pathogens and for 2nd- or 3rd-line treatments.

34In high-risk patients, there is no evidence for superior efficacy of a combination of antibacterial agents compared with monotherapy (AIIr). Combining antibacterial agents in this indication has also not been shown to prevent the development of resistance. A combination might be useful in institutions with a high prevalence of multidrug-resistant bacteria (AIIr). An antipseudomonal beta-lactam should always be included, with an aminoglycoside or a fluoroquinolone such as levofloxacin and ciprofloxacin as the combination partner (AIIt). For standard-risk patients without critically impaired renal function, the combination of an aminoglycoside with a third- or fourth-generation cephalosporin can be considered (AI). When aminoglycoside antibiotics are given, therapeutic drug monitoring is mandatory (AIIu) and once-daily dosing is appropriate (AIIr).

35A combination therapy including vancomycin or teicoplanin (DIIr) or linezolid (DIII) is generally discouraged for empirical first-line therapy but might be considered in the case of (CIII) severe mucositis, skin or soft tissue infection, foreign body infection, or documented colonization of a patient with MRSA. The use of vancomycin is associated with an increased risk of nephrotoxicity, which should be diminished by therapeutic drug monitoring (BIIt). Although a higher rate of VRE infections has been found in VRE-colonized patients, the addition of linezolid to empirical first-line treatment has not shown a significant benefit. Beyond this, the risk of thrombocytopenia as one of the major potential side effects of linezolid has to be considered.

36The recommendations in this paragraph affect patients with an expected duration of neutropenia ≤ 7 days who typically are not receiving systemic fluoroquinolone prophylaxis.

37For patients with ≤ 7 days of neutropenia and no high-risk criteria identified by the MASCC score (Table 2) and without practical obstacles to outpatient management (Table 3), oral and outpatient therapy is recommended (AI). Hospitalization and in-patient start of antibacterial therapy, with a change to an outpatient treatment after defervescence or stabilization within the first 2 days, is an alternative approach validated in clinical studies (AI).

38Safety and feasibility of oral outpatient therapy for standard-risk patients identified by the MASCC score has recently been confirmed in a randomized prospective multicenter study, with less than 10% complications in the total study population and only 5% of patients requiring readmission under moxifloxacin monotherapy compared with a twice-daily combination of amoxicillin/clavulanate and ciprofloxacin. For oral antibiotic therapy in standard-risk patients, a combination of amoxicillin/clavulanate with ciprofloxacin or monotherapy with moxifloxacin appears feasible (AI). Moxifloxacin is not active against P. aeruginosa. In the case of confirmed penicillin allergy, amoxicillin/clavulanate may be replaced by clindamycin or cefuroxime axetil (BIIu). Primary intravenous therapy for patients of the standard-risk group may consist either of a monotherapy with ceftazidime, cefepime, or piperacillin/tazobactam, or, in the case of an increased rate of multiresistant gram-negative bacteria, of a combination of a third- or fourth-generation cephalosporin with an aminoglycoside (AI).

39Increasing CRP on day 5, compared with day 1, might indicate an unfavorable course of the episode. Elevated procalcitonin may point at severe infection or sepsis in high-risk patients. Also, monitoring of procalcitonin on day 2 after onset of fever may help to detect a minority of patients with potentially severe infections, and in the case of persistent fever, it may contribute to early diagnosis of invasive mycoses. Rising interleukin-6 typically indicates an unresolved infection/inflammation, while low levels have a high negative predictive value making severe septic infection unlikely. None of these parameters should be used alone for clinical decision-making (BIII).

40After ≥ 96 h of persistent or recurrent fever despite adequate therapy, a multislice pulmonary CT scan (AIIu) should be performed (preferably within 24 h after indication), independent of respiratory symptoms (AIIu). Other imaging procedures are indicated according to clinical signs or symptoms of a localized infection (BIIu). A thorough physical examination must be reiterated, with inspection of the oropharynx, skin lesions with particular attention to venous access and puncture sites and the perianal region, as well as painfulness of paranasal sinuses or other signs of upper airway infection (Table 4). Blood cultures from peripheral vein and indwelling central venous catheters should be repeated, while other microbiological cultures are only useful if clinical signs or symptoms indicate a possible site of infection (BIIu).

41If diagnostic procedures reveal a clinically documented infection or if a causative pathogen has been isolated, the empirical antibacterial approach should be changed to targeted or pre-emptive therapy (AIIt). Pre-emptive antimicrobial treatment is chosen according to the spectrum of microorganisms typically involved in the respective clinically documented infection (Table 4).

42A change of the empirical antimicrobial treatment regimen can be considered in patients with fever recurrent or persisting for more than 96 h; however, a general change of antibacterial agents is not recommended (DIIr). In clinical studies on antibiotic therapy of neutropenic fever, median time to defervescence was 4 to 5 days. The empirical addition of vancomycin after non-response to piperacillin/tazobactam or teicoplanin after non-response to imipenem has not been more effective than placebo, and defervescence after another 72 h of the unmodified beta-lactam regimen (placebo arms) was 45%. A modification or escalation of antimicrobial therapy only because of persistent elevation of inflammatory laboratory parameters has not been successful as well. A change of antimicrobial therapy is recommended in patients with recurrent or persisting fever and clinical deterioration, instability, or other signs of progressive infectious disease (AIIu). In the case of severe sepsis and/or signs of critical organ failure, modification of antimicrobial therapy along with intensive further medical support is required (AIIu).

43As prospective studies for second-line antimicrobial therapy in neutropenic patients with persistent FUO under clearly specified 1st-line treatment regimens are sparse, recommendation of treatment modification are partially based on clinical expertise.

44A change of empirical antimicrobial therapy aims to cover a broader range of bacteria and/or to overcome resistance among pathogens principally included in the spectrum of the first-line regimen. Again, the local prevalence of vancomycin-resistant enterococci, methicillin-resistant S. aureus, and extended-spectrum beta-lactamase-producing gram-negative bacilli, as well as the rate of primary resistance to piperacillin/tazobactam among Escherichia coli must be reconsidered (AIIt). Antimicrobial agents recommended for empirical second-line treatment are included in Table 5. In standard-risk patients initially treated with a cephalosporin plus/minus an aminoglycoside, a change to piperacillin/tazobactam, meropenem, or imipenem is recommended (AIIt).

45Empirical antifungal therapy is not recommended in patients of the standard-risk group (DIII). In high-risk patients, a prospective randomized trial showed a higher defervescence rate after addition of empirical mold-active antifungal therapy as compared with modification of antibacterial agents only. These data were confirmed in a meta-analysis, yet a significant survival benefit could not be demonstrated. No benefit, however, could be shown for high-risk neutropenic patients, if antifungal therapy was applied already at onset of first fever instead of fever persisting for more than 72 h.

46For high-risk patients without prior systemic antifungal prophylaxis, mold-active empirical antifungal therapy is recommended, if fever persists for ≥ 96 h or if fever relapses despite adequate antibacterial therapy (AI). This also includes patients given either oral itraconazole prophylaxis but not achieving sufficient serum or plasma trough concentrations (> 500 ng/ml) or a mold-inactive prophylaxis, i.e., fluconazole (BIIt). For patients receiving oral voriconazole or posaconazole prophylaxis, no prospective trial on the efficacy of a switch to another mold-active agent for empirical antifungal therapy is available. Thus, such a switch may be judicious in the setting of persistent FUO (CIII), but if a patient shows no clinical sign of invasive fungal disease despite adequate diagnostic work-up, blood samples were negative for Aspergillus galactomannan, and levels of posaconazole or voriconazole are within the target range, unmodified continuation of oral antifungal prophylaxis is reasonable. In the case of clinical deterioration, a change to an intravenously applied antifungal agent is recommended (AIII). For empirical mold-active antifungal therapy in febrile neutropenic patients, caspofungin and liposomal amphotericin B (AmB) are approved. Liposomal amphotericin B is preferred in patients at increased risk of fungal infection with non-Aspergillus molds (AI).

47Several studies have aimed at a reduction of antifungal therapy in high-risk patients by not empirically treating all patients with persisting fever in prolonged neutropenia but only those with additional findings indicating the presence of a fungal disease. Utilizing pulmonary CT scan and testing for galactomannan, Aspergillus-specific PCR or both have been used for this so-called diagnostic-driven or pre-emptive approach. An increased number of invasive fungal infections and a substantially reduced consumption of antifungals were found in the pre-emptive as compared with the empirical treatment groups, without a significant increase in mortality rates. This approach cannot be recommended as a routine standard but might provide an alternative to empirical antifungal therapy (BIIr).

48Numerous studies have compared efficacy and safety of empirical antifungal treatment. The most robust data are available for caspofungin or liposomal AmB (AI) (Table 6). Conventional AmB deoxycholate is not recommended because of its renal toxicity and other adverse events (DI). The use of the two lipid AmB formulations ABCD and ABLC is not supported due to the lack of appropriate clinical studies in this setting. For voriconazole (BI), a prospective trial could not demonstrate non-inferiority to liposomal AmB in a composite endpoint, although it was associated with a lower rate of breakthrough fungal infections. For itraconazole, relevant data are available for the intravenous formulation. As the oral application is associated with impaired bioavailability, only the intravenous application can be recommended (BI). For micafungin (CI), a comparison with intravenous itraconazole showed superior response rates, but no study with a standard of care (caspofungin or liposomal AmB) as the comparator is available. No data are available for anidulafungin or posaconazole in the empirical indication. Fluconazole, with no effect on mold infections, is not recommended for empirical antifungal therapy in persistently febrile neutropenic high-risk patients.

49Empirical antiviral therapy in febrile neutropenic patients without signs or symptoms typical for a viral infection is discouraged (DIII).

50The adjunctive use if granulocyte colony-stimulating factor (G-CSF) is not recommended for routine clinical practice in febrile neutropenic patients (DIIr). If G-CSF has not been started before the onset of neutropenia, its interventional use can be considered in patients with fever and neutropenia who are at high risk for infection-associated complications or who have prognostic factors that are predictive of poor clinical outcomes, including expected prolonged (> 10 days) and profound (< 100/μl) neutropenia, age > 65 years, uncontrolled primary disease, or hospitalization at the time of fever development (BIIr).

51Supportive therapy with polyclonal immunoglobulin is recommended only in select neutropenic patients with proven immunoglobulin deficiency (BIIt).

52In a high-risk setting such as neutropenic fever, potential sources of infection should be identified and removed if possible. Central venous catheters not indispensable for patient care should be taken off. In patients with persistent FUO in whom no focus of infection has been found, empirical removal or change of the catheter may be justified (CIII).

53The appropriate duration of antimicrobial therapy in neutropenic patients after onset of stable defervescence, i.e., body temperature below 38 °C without the use of antipyretic drugs, has not been prospectively studied. A continuation until neutrophil recovery has been questioned by results from several studies on discontinuation of parenteral antibiotics in responding, but persistently neutropenic patients, showing no substantial rates of recurrence of fever or documented infections. Challenges of antimicrobial stewardship in an era of globally increasing multidrug-resistance and missing development of new broad-spectrum anti-infectives gave reason for the recommendation to drastically shorten the administration of antibiotics in neutropenic patients after treatment response.

54In persistently neutropenic patients, empirical therapy may be discontinued, but not earlier than 7 days after the onset of stable defervescence (without the use of antipyretic agents) and only in the absence of clinical signs or symptoms of infection (BIII). In this setting, a re-institution of systemic antibacterial prophylaxis, if given before the onset of fever, may be considered (CIII). In the case of hematopoietic recovery to a neutrophil count of >500/μl, empirical antimicrobial therapy can be safely discontinued after 2 days of stable defervescence (BIII).