Antibiotic use for respiratory syncytial virus in the Middle East: A surveillance study in hospitalized Jordanian children

Abstract

Introduction: In developing countries where point-of-care testing is limited, providers rely on clinical judgement to discriminate between viral and bacterial respiratory infections. We performed a cross-sectional cohort study of hospitalized Jordanian children to evaluate antibiotic use for respiratory syncytial virus (RSV) infections.

Materials and methods: Admitting diagnoses from a prior viral surveillance cohort of hospitalized Jordanian children were dichotomized into suspected viral-like, non-pulmonary bacterial-like, and pulmonary bacterial-like infection. Stratifying by sex, we performed a polytomous logistic regression adjusting for age, underlying medical condition, maternal education, and region of residence to estimate prevalence odds ratios (PORs) for antibiotic use during hospitalization. Sensitivity and specificity of admission diagnoses and research laboratory results were compared.

Results: Children with a suspected viral-like admission diagnosis, compared to those with suspected non-pulmonary bacterial-like, were 88% and 86% less likely to be administered an empiric/first-line antibiotic (male, aPOR: 0.12; female, aPOR: 0.14; p-value = <0.001). There were slight differences by sex with males having a lower prevalence than females in being administered an expanded coverage antibiotic; but they had a higher prevalence of macrolide administration than males with non-pulmonary bacterial-like infection. Overall, children with RSV had a 34% probability (sensitivity) of being assigned to a suspected viral-like diagnosis; whereas RSV-negative children had a 76% probability (specificity) of being assigned to a suspected pulmonary bacterial-like diagnosis.

Conclusions: Hospitalized children with a suspected viral-like admission diagnosis were less likely to receive an empiric/first-line and expanded coverage antibiotic compared to suspected non-pulmonary and pulmonary infections; however, when evaluating the accuracy of admission diagnosis to RSV-laboratory results there were considerable misclassifications. These results highlight the need for developing antibiotic interventions for Jordan and the rest of the Middle East.

Fig 1. Consort diagram of hospitalized Jordanian children included in a cross-sectional cohort analysis assessing the association of admission diagnosis and antibiotic administration.

Fig 2. Percentage of antibiotics administered based on presenting clinical symptoms, by antibiotic type to children less than two hospitalized in Amman, Jordan.

Footnote: ^a Viral-Like Infection—RSV, bronchiolitis, acute respiratory infection, upper respiratory infection, influenza, apnea, asthma, croup, and wheezing. ^b Non-Pulmonary Bacterial-Like Infection—pertussis, pneumonia, bronchopneumonia, and acute otitis media. ^c Pulmonary Bacterial-Like Infection—sepsis, febrile, febrile seizure, and tonsillitis. ^d Empiric/First-Line—ampicillin, gentamicin, cefotaxime, ceftriaxone, and cefuroxime. ^e Macrolides—erythromycin, clarithromycin, and azithromycin. ^f Expanded Coverage—vancomycin, ceftazidime, and amikacin.