Nov 12 2012
By Sarah Guy, medwireNews Reporter
Serum procalcitonin measurements could improve the accuracy of pneumonia diagnoses when chest radiographs give indeterminate results, indicate US study findings.
Patients in the study with pneumonia had higher procalcitonin levels than their counterparts with exacerbations of chronic obstructive pulmonary disease, acute bronchitis, or asthma.
High levels of the hormone also had moderate accuracy for distinguishing the presence of pneumonia infiltrates in receiver operating characteristic (ROC) analysis compared with a pulmonologist's evaluation, report the researchers.
"Thus, a low ProCT [procalcitonin] may help clinicians feel more confident to delay or stop antibiotics in a patient in whom clinical suspicion for bacterial pneumonia is low, yet the radiographic report states 'pneumonia cannot be ruled out'," say Edward Walsh (Rochester General Hospital, New York) and co-researchers.
"Such strategies may assist in reducing unnecessary antibiotic use in the United States," they write in the Journal of Hospital Medicine.
Levels of procalcitonin - a calcitonin precursor secreted in the body in response to bacterial infection - were determined in 530 patients admitted to hospital over the winters of 2008 to 2009 and 2009 to 2010, with 528 diagnoses compatible with respiratory tract infection.
The primary admission diagnoses were predominantly pneumonia and exacerbations of chronic obstructive pulmonary disease, at 31% and 27%, respectively.
Pneumonia patients had significantly higher median levels of procalcitonin, at 0.27 ng/mL, than their counterparts with exacerbations of chronic obstructive pulmonary disease, at approximately 0.90 ng/mL; asthma, at approximately 0.80 ng/mL; bronchitis, at approximately 0.10 ng/mL; and viral influenza, at 0.11 ng/mL.
Radiology reports for the cohort indicated that 40% (n=213) of patients had no acute disease, 14% (n=76) had "other" definitive findings (ie, pleural effusion), 14% (n=75) had infiltrates consistent with pneumonia, and 31% (n=164) had indeterminate findings.
However, while the pulmonologist agreed with the majority of these diagnoses, he only agreed that four were indeterminate, classifying half of the remaining 160 as infiltrates.
The authors calculated the ROC curve using cases in which the pulmonologist agreed with the radiologists' classification of infiltrates, and, giving an area under the curve (AUC) score of 0.80 (where 1.00 denotes perfect discrimination). The likelihood of having an infiltrate increased with increasing procalcitonin thresholds.
When this analysis was repeated using data for patients with indeterminate radiographs, the AUC was lower, but still with moderate predictive value, at 0.72, says the research team.
Licensed from medwireNews with permission from Springer Healthcare Ltd. ©Springer Healthcare Ltd. All rights reserved. Neither of these parties endorse or recommend any commercial products, services, or equipment.