Affiliations: Department of Pediatrics Gentofte Hospital,
Denmark | Department of Biostatistics, University of Copenhagen,
Denmark | Department of Radiology, Gentofte Hospital,
Denmark
Note: [] Correspondence: Hans Erik Nielsen, MD, Department of Pediatrics,
Gentofte Hospital, Niels Andersens Vej 65, 2900 Hellerup, Denmark. Tel.: +45
39777354; Fax: +45 39777639; E-mail: Hanie@gentoftehosp.kbhamt.dk
Abstract: The aim of this study is to establish criteria for distinguishing
between antibiotic-sensitive and insensitive pneumonia, from data available at
the moment of deciding which treatment to give. Retrospective chart review of
169 consecutive children, aged 1–16 years, admitted to hospital with
X-ray-verified pneumonia and treated with intravenous antibiotics, 89% with
penicillin (because resistance of pneumococci and of Haemophilus influenzae to
penicillin is practically non-existent in Denmark). The temperature response to
intravenous antibiotic treatment, as we define it rigorously, was the response
variable. Its dependence on the clinical, laboratory and radiological variables
available at the moment of diagnosis was analyzed by multiple regression
analysis. From the regression model, we derived a rule for predicting the
effect of antibiotic treatment on the temperature response. We identified a
number of variables that independently predicted rapid resolution of the fever:
young age; several days of disease before admission; oral administration of
antibiotics before admission; high initial temperature; high white blood cell
count; absence of wheezing and of atelectasis; the presence of a sharply
delineated and/or spherical lung infiltrate; and a general appearance of being
ill, but only in the absence of chest wall retractions. The resulting
prediction rule explained less than 50% of the response variability. From the
data available at the moment of deciding about treatment for pneumonia, we have
identified the parameters that predicted temperature response to antibiotic
treatment, but the prediction rule was too imprecise to be clinically useful,
i.e., it could not distinguish between those children who would benefit from
antibiotic treatment, and those who would not.
