The Clinical Utility of Chest Radiography for Identifying Pneumonia: Accounting for Diagnostic Uncertainty in Radiology Reports release_mo4axdhc4vasjgunclaewynozu

by Alexander Makhnevich, Liron Sinvani, Stuart L. Cohen, Kenneth H. Feldhamer, Meng Zhang, Martin L. Lesser, Thomas G. McGinn

Abstract

OBJECTIVE. Currently, chest radiography is the first-line imaging test for identifying pneumonia; chest CT is considered the reference standard. The purpose of this study was to calculate the statistical measures of performance of chest radiography for identifying pneumonia when taking into account uncertain results of both chest radiography and CT examinations. MATERIALS AND METHODS. Statistical measures of performance of chest radiography, using CT as the reference standard, were calculated with 95% CIs by varying uncertain radiology report impressions of both chest radiography and CT to all negative or all positive. The resulting scenarios were as follows: scenario 1, uncertain chest radiography and CT impressions are considered positive for pneumonia; scenario 2, uncertain chest radiography impressions are positive but uncertain CT impressions are negative; scenario 3, uncertain chest radiography impressions are negative and uncertain CT impressions are positive; scenario 4, uncertain chest radiography and CT impressions are negative; and scenario 5, uncertain chest radiography and CT impressions are excluded. RESULTS. A retrospective analysis of 2411 patient visits revealed the prevalence of uncertain radiology report impressions to be 31.8% for chest radiography and 21.7% for CT. Scenario 1 yielded the following performance values: sensitivity, 51.9%; specificity, 71.3%; PPV, 59.4%; and NPV, 64.5%. Scenario 2 produced the following performance values: sensitivity, 59.6%; specificity, 67.1%; PPV, 59.6%; and NPV, 67.1%. Scenario 3 showed the following performance values: sensitivity, 13.4%; specificity, 97.7%; PPV, 82.6%; and NPV, 58.1%. Scenario 4 yielded the following performance values: sensitivity, 19.6%; specificity, 96.4%; PPV, 81.6%; and NPV, 59.5%. Scenario 5 produced the following performance values: sensitivity, 32.7%; specificity, 96.8%; PPV, 89.2%; and NPV, 63.8%. CONCLUSION. Uncertain chest radiography results for the evaluation of pneumonia are prevalent. A chest radiography impression using the strongest language in support of a pneumonia diagnosis is useful to rule in pneumonia radiographically, but a negative result performs poorly at ruling out disease.
In text/plain format

Published in American Journal of Roentgenology by American Roentgen Ray Society
ISSN-L 0361-803X
Volume 213
Issue 6
Page(s) 1-6
Release Date 2019-09-11
Container Type journal
Publisher American Roentgen Ray Society
Primary Language en (lookup)

Known Files and URLs

There are no accessible files associated with this release. You could check other releases for this work for an accessible version.

No Full Text Available "Save Paper Now"
Know of a fulltext copy of on the public web? Submit a URL and we will archive it
Type  article-journal
Stage   published
Date   2019-09-11
Container Metadata
Not Open Access
Not in DOAJ
Not in ISSN ROAD
ISSN-L:  0361-803X
Fatcat Entry
Work Entity
grouping other versions (eg, pre-print) and variants of this release
Cite This Release
Fatcat Bits

State is "active". Revision:
274655f7-1308-4df9-9fbd-3b2e3428c25c
As JSON object via API