Limited Utility of Routine Chest X-Ray (CXR) in Initial Evaluation of Febrile Neutropenia in Patients with Haematological Diseases Undergoing Chemotherapy

Background

Initial investigations for patients presenting to health services with febrile neutropenia (FN) aim to identify the infective source. Routine CXR is recommended by several international guidelines as a component of this assessment. However, in the era of rapid delivery of empirical broad-spectrum antibiotics, the evidence supporting use of CXR in detecting chest infection in asymptomatic patients and its role in directing management are lacking. While CXRs are considered relatively inexpensive, accessible and low-risk, they still contribute to the overall healthcare bill, expose patients to ionising radiation and may involve moving immunocompromised patients from a filtered 'clean air' environment. The aim of this study was to evaluate the use of CXR in haematology patients undergoing immunosuppressive therapy who presented for management of FN.

Methods:

A retrospective single-centre analysis of patients with haematological diseases undergoing chemotherapy admitted to a large cancer centre with FN over a 5 year period. FN was defined as fever and a neutrophil count below the lower limit of normal (<1.5 x 10⁹/L). Eligible patients were aged ≥16years with a confirmed diagnosis of a haematological disease who had received chemotherapy within the preceding 2 months. Protocolised investigations of newly diagnosed FN included full blood count, biochemistry, blood cultures, urine culture together with the prompt administration of piperacillin-tazobactam (4.5g) IV. CXRs were recommended but not mandated. Further antibiotic management and investigation were guided by results and clinical assessment. Patients were identified from electronic hospital records and baseline demographics, initial assessment, pre-existing lung conditions, investigation and treatment details and outcomes were collated. CXR were reviewed by an independent radiologist and results were recorded together with overall management decisions. Univariate analysis was used to determine any variables with a significant association with an abnormal CXR.

Results:

Between January 2011 & December 2015, 427 FN episodes were identified in321patients with a median age 58 years (range 17-90); 56% were male; 98% had an underlying diagnosis of leukemia, lymphoma or myeloma. CXR was performed in 322 episodes (75%); 137 standard PA, 182 AP, 3 unknown (of these, 163 were portable). Twenty-seven CXRs (8%) demonstrated evidence of infection, 285 (89%) were normal and 10 (3%) were indeterminate. Only 4/182 (2%) of CXRs in patients with no reported respiratory symptoms or signs had evidence of infection; in those with reported respiratory symptoms/signs, 23/130 (18%) CXRs were consistent with chest infection. The presence of ≥1 respiratory symptom/sign was associated with an increased chance of an abnormal CXR (odds ratio 17.19; p=0.007, 95%CI:2.16-136.41). Only 5/322 (1.6%) of CXR results caused a change in antibiotic management, all 5 in patients with respiratory symptoms or signs.

Conclusion:

In FN patients undergoing chemotherapy for haematological conditions, CXR rarely detected chest infection or changed management in asymptomatic patients. CXR in our institution is no longer part of routine assessment of FN in the absence of clinical respiratory features in patients with haematological conditions.