Survey of practices for the clinical management of febrile neutropenia in children in hematology-oncology units in Latin America

We conducted a multinational, anonymous, cross-sectional, and descriptive survey approved by the St. Jude Children’s Research Hospital (St. Jude) Institutional Review Board. The survey met the criteria of minimal risk and was administered from mid-September to the end of December 2017. An e-mail questionnaire was sent to all identified potential participants.

We targeted physicians who completed their pediatric training and were working in general or pediatric hospitals throughout Latin America, including Mexico, Central and South America, and Spanish-speaking Caribbean countries. Eligible participants were physicians; most were pediatricians, several of whom had subspecialty training (oncologists or infectious diseases consultants) and provided direct healthcare to children with cancer at the time of the survey. These participants were recruited mainly from three sources: One source was the Prevencionistas e Infectólogos para Cáncer Pediátrico en América Latina (PRINCIPAL) network, which joins healthcare providers involved in infection care and prevention in children with cancer at their institutions in Latin America. Members of the PRINCIPAL network aim to collaborate to better understand infection care and prevention in children with cancer in their region and find solutions to identified gaps in care to improve outcomes of cancer treatment. The second source was contacts of St. Jude Global, which is composed of a network of oncologists and pediatricians collaborating with St. Jude to improve rates of survival of children with cancer worldwide [https://​www.​stjude.​org/​global.​html]. The third source was members of the Sociedad Latinoamericana de Infectologia Pediátrica (SLIPE). Candidates from these three sources, using a snowball-sampling technique, were encouraged to provide the names of qualified potential candidates or to pass survey information to potentially interested participants. Participant eligibility was determined if the volunteer met the following criteria: (1) the participant works in a unit or facility that provides pediatric cancer care; (2) the participant provides direct clinical care to patients with FN; and (3) the participant has read the informed consent and agrees to complete the survey. Eligible participants were asked to answer questions under the assumption that all patients were actively receiving treatment for cancer and were clinically stable. Cancer was any malignant disease that required focused oncologic care appropriate for the type of malignancy; clinical stable was defined as not requiring a critical care support; and infectious diseases were diseases attributable to infections clinically and/or microbiologically documented.

Our survey, which was adapted from a previously published survey [5], was prepared in Spanish and had 58 questions organized into five sections addressing risk-group stratification (2 questions), routine clinical management of FN (37 questions), management modification (1 question), use of guidelines (7 questions), and respondent demographic information (11 questions). Most (81%) questions were multiple-choice selections, yes/no answers, or Likert scale rating. The rest of the questions required ranking the answers, selecting all those that apply, and free texts. Before distribution, the survey tool was reviewed by six Spanish-speaking physicians, who were oncologists, infectious diseases specialists, or pediatricians. The tool was modified based on their feedback. The final version of the survey could be completed in less than 20 min. The survey was distributed electronically and hosted on the Epi Info™ Web Survey platform.

All analyses were performed using SAS software [version 9.4]. Descriptive statistics of the data were summarized accordingly.

In 2.5 months, we sent 220 surveys and received 109 responses (49.54%) (Fig. 1). One respondent did not treat children with cancer and did not fulfill our inclusion criteria. Therefore, for the analysis, we included 108 respondents. Respondents were from 19 Latin American countries (Table 1 and Fig. 2), and 45% of those were from South America. The most represented country was Mexico (n = 26), followed by Argentina (n = 11), Colombia (n = 9), Paraguay (n = 9), and Guatemala (n = 8). Most respondents were 31–40 years old (48%). Thirty-six (33.33%) respondents worked in cancer units located within general hospitals; 51 (47.22%) worked in cancer units within pediatric hospitals; and 15 (13.89%) worked in independent oncology units. Seventy (64.81%) participants worked in a national or regional reference unit. Among the respondents, 26 (24%) worked in a hospital with a bone marrow transplantation unit. Respondents’ institutions had 5–1000 new cancer diagnoses per year, with a mean of 139 cases, a median of 80, and a mode of 50. Practice-years of respondents ranged from 0 to 31, with a mean of 8, a median of 5, and a mode of 3. Other respondent characteristics are shown in Table 1.

Among the respondents who answered the survey questions about the FN guidelines, those from oncology units operating within a general hospital had access to FN guidelines more often (83%) compared to those from specialized oncology centers (60%) and from oncology units within a pediatric hospital (65%). When asked about using a risk stratification system, over two-thirds (69–79%) of the respondents of all these types of centers acknowledged using a type of risk stratification in a FN episode. The risk-stratification systems used were either published, such as the Santolaya et al. [6] (n = 21, 27%) or the SLIPE [7] (n = 20, 25%) schema, or were built locally (n = 27, 34%). Eleven (15%) respondents used other guidelines. The respondents perceived that FN guidelines were used by the treating physicians (oncologists and hematologists, 88% agreed or strongly agreed; and by the emergency department physicians, 73% agreed or strongly agreed). When asked about the percentage of patients at high risk of FN that are treated per the guidelines, 61 (58.10%) respondents answered that most or all (76%–100%) of the patients are treated following the guidelines. However, for those patients with low risk of FN, fewer respondents (n = 45, 42.45%) followed guideline recommendations to treat most or all (76–100%) of their patients (Fig. 3). When asked about specific laboratory studies done routinely as initial management of FN, most respondents did blood cultures, including in those patients with a catheter (from a peripheral vein and from the catheter; n = 98, 94%) and in those without a catheter (two blood cultures from different venipunctures; n = 71, 73%). Also, a high percentage performed a C-reactive protein test (n = 98, 91.59%), urinalysis (n = 88, 84.62%), and urine culture (n = 75, 72.12%). Chest X-rays were done for more than half of the patients (n = 57, 54.81%) (Table 2). Most respondents indicated that they did not routinely perform the following studies: β-d-glucan testing (n = 102, 100%), galactomannan testing (n = 99, 97.06%), nasal swab for respiratory viral studies (n = 88, 85,44%), fungal culture (n = 73, 70.19%), and procalcitonin level (n = 62,59.62%).

Most (94%) respondents used axillary temperature to determine the presence of fever. According to the survey, fever definition in FN varied: 39 (36%) respondents considered fever if the temperature exceeded 38.3 °C or two temperature measurements were greater than 38 °C at 1 h apart in a 24-h period; 35 (32%) respondents considered fever when the temperature was greater than 38 °C; and 34 (31%) respondents were split and considered fever if the temperature was either greater than 38.3 °C or greater than 38.5 °C. The definition of neutropenia also varied, but for patients at low risk of FN, 68 (64%) respondents considered neutropenia to be an absolute neutrophil count (ANC) ≤ 500/μL, or ANC 500–1000/μL and anticipated to decline due to recent chemotherapy; 15 (14%) of the respondents considered neutropenia if the ANC was ≤ 1500/μL and 23 (21%) respondents considered neutropenia if the ANC was ≤ 1000/μL. For patients at high risk of FN, 70 (67%) of the respondents had similar answers as they did for those at low risk; 8 (7%) of the respondents considered neutropenia if the ANC was ≤ 1500/μL and 27 (25%) of the respondents considered neutropenia if the ANC was below 1000.

According to our findings, more than one-third of our respondents used a higher than accepted ANC level to trigger a clinical decision. Respondents treated patients with FN as inpatients in 56% of low-risk cases and in 98% of high-risk cases (Table 2). The treatment for patients at low risk of FN was intravenously (IV) administered antibiotic monotherapy, which was given by 53% of respondents; ceftriaxone was the most frequently used drug (34%). In patients at high risk of FN, double-drug antibiotics administered IV were given by more than half of the respondents; the top selections included a combination of amikacin and primarily ceftazidime (n = 32, 30%), piperacillin/tazobactam (n = 27, 25%), or cefepime (n = 26, 24%).

Subtle differences were seen in discharge criteria for patients at low vs. high risk of FN. For low-risk cases, 56% of respondents preferred that the patient was afebrile for more than 48 h; 60% preferred that blood cultures were negative for more than 72 h; and 55% preferred 72-h inpatient monitoring before discharge. For high-risk cases, 57% of respondents required that patients be afebrile for more than 72 h; 67% required negative blood cultures for 72 h; and 67% required inpatient monitoring for 72 h before discharge (Table 3). An ANC > 500/μL and rising was a criterion for discharge in 74% of patients at low risk of FN and in 81% at high risk of FN (Table 3).

The use of oral antimicrobials in discharged patients at low risk of FN was similar, whether they had ANC < 500/μL (52%) or ANC < 1000/μL (56%). However, for patients at high risk of FN, IV antibiotics were preferred in 60% of cases with ANC < 500/μL, and oral antibiotics were preferred in 48% with ANC < 1000/μL (Table 3).

When an invasive fungal infection is suspected, 80 (75.47%) respondents indicated that they routinely perform computed tomography (CT) scan of the abdomen; 95 (89.62%) routinely perform CT scan of the thorax; 98 (90.74%) analyze fungal cultures; 76 (70.37%) assess serum galactomannan; and 79 (73.83%) perform a CT scan of the paranasal sinuses. Respondents indicated the use of empirical antifungal therapy in patients with FN, at high risk for infection more frequently and sooner than in those at low risk who persist febrile beyond 72 h on broad-spectrum antimicrobials (n = 38, 35.19% vs. n = 29, 27.10%) (Table 4). The preferred antifungal for empiric therapy was fluconazole in patients who were low risk for invasive fungal infection (62.96%) and amphotericin B deoxycholate (42.59%) or amphotericin B lipid formulations (26.85%) in those who were at high risk for invasive fungal infection (Table 4).