Predictors of radiographic pneumonia in febrile children with cancer presenting to the emergency department

Abstract

Background

Fever is a common presenting complaint to the pediatric emergency department (PED), especially among oncology patients. While bacteremia has been extensively studied in this population, pneumonia has not. Some studies suggest that chest X-ray (CXR) does not have a role in the investigation of neutropenic fever in the absence of respiratory symptoms, yet non-neutropenic pediatric oncology patients were excluded from these studies.

Objective

We aimed to determine the incidence of CXRs ordered for febrile pediatric oncology patients, irrespective of their absolute neutrophil count (ANC), and to evaluate the rates of radiographic pneumonia as well as predictors of the latter in this group.

Method

This study was conducted in the PED at the American University of Beirut Medical Center (AUBMC), an Eastern Mediterranean tertiary-care hospital. We conducted a retrospective cohort study of acutely febrile pediatric cancer patients, younger than 18 years, presenting to a tertiary center from 2014 to 2018. We included one randomly selected febrile visit per patient. Fever was defined as a single oral temperature ≥38 °C within 24 h of presentation. We collected data on patient characteristics and outcomes. Our primary outcome was radiographic pneumonia; our secondary outcome was whether a CXR was done or not. We defined radiographic pneumonia as a consolidation, pleural effusion, infiltrate, pneumonia, “infiltrate vs. atelectasis,” or possible pneumonia mentioned by the radiologist. SPSS was used for the statistical analysis.

Results

We reviewed a total of 664 medical charts and included data from 342 febrile pediatric patients in our analysis. Of these, 64 (18.7%) had a CXR performed. Overall, 16 (25%) had radiographic pneumonia while 48 (75%) did not. Patients were significantly more likely to have a CXR performed if they presented with upper respiratory tract symptoms, cough (p < 0.001 for both), or abnormal lung auscultation at the bedside (p = 0.004). Patients were also less likely to have a CXR done if they were asymptomatic upon admission to the PED (p < 0.001). However, neither cough nor shortness of breath nor abnormal lung examinations were significant predictors of a positive CXR (p = 0.17, 0.43, and 0.669, respectively). Patients with radiographic pneumonia were found to be significantly younger (4.29 vs. 6 years, p = 0.03), with a longer time since their last chemotherapy (15 vs. 7 days, p = 0.005), and were given intravenous (IV) bolus in the PED (87.5% vs. 56.3%, p = 0.02). Interestingly, patients with higher white blood cell (WBC) counts were more likely to have radiographic pneumonia (4850 vs. 1750, p = 0.01). Having a cough and an abnormal lung examination on presentation increased the odds of having a CXR (adjusted odds ratio [aOR]: 6.6; 95% confidence interval [CI]: 3.4–12.8 and aOR: 4.5; 95% CI: 1.1–18.3, respectively). Returning to the PED for the same complaint within 2 weeks was associated with lower odds of a CXR at the index visit (aOR: 0.3; 95% CI: 0.1–0.6). For every year the child is older, the odds of having radiographic pneumonia decreased by 0.8 (95% CI: 0.6–0.98). However, for every day since the last chemotherapy session, the odds increased by 1.1 (95% CI: 1.01–1.12).

Conclusion

In our sample, CXR was not commonly performed in the initial assessment of febrile cancer patients in the PED, unless respiratory symptoms or an abnormal lung examination was noted. However, these were not significant predictors of radiographic pneumonia. Further studies are needed to identify better predictors of pneumonia in this high-risk population.