The value of skull radiography in identifying intracranial injury has not yet been satisfactorily defined. A multidisciplinary panel of medical experts was assembled to review the issue of skull radiography for head trauma. The panel identified two main groups of patients--those at high risk of intracranial injury and those at low risk of such injury--and developed a management strategy for imaging in the two groups. The high-risk group consists primarily of patients with severe open or closed-head injuries who have a constellation of findings that are usually clinically obvious. These patients are candidates for emergency CT scanning, neurosurgical consultation, or both. The low-risk group includes patients who are asymptomatic or who have one or more of the following: headache, dizziness, scalp hematoma, laceration, contusion, or abrasion. Radiographic imaging is not recommended for the low-risk group and should be omitted. An intermediate moderate-risk group is less well defined, and skull radiography in this group may sometimes be appropriate. A prospective study of 7035 patients with head trauma at 31 hospital emergency rooms was conducted to validate the management strategy. No intracranial injuries were discovered in any of the low-risk patients. Therefore, no intracranial injury would have been missed by excluding skull radiography for low-risk patients, according to the protocol. We conclude that use of the management strategy is safe and that it would result in a large decrease in the use of skull radiography, with concomitant reductions in unnecessary exposure to radiation and savings of millions of dollars annually.

No abstract

The chest roentgenograms of 128 consecutive ambulatory children with radiologic pneumonia were read independently and without clinical information by a faculty general pediatrician (Ped), a pediatric radiologist (R-P) and a general radiologist (R-G). The films were classified as normal, indicative of a viral or bacterial process, or indeterminate. Readings were compared with results of viral titers and bacterial cultures. Agreement between any two observers in classifying films, measured by unweighted Kappa, while statistically significant (p less than 0.001) for any pair, was low. There was no significant difference between the agreement scores of Ped/R-P, Ped/R-G, and R-P/R-G. Twenty-one patients had fourfold viral titer increases (N = 16) or positive bacterial cultures of blood or pulmonary aspirate (N = 5). The sensitivity of viral readings for titers increases varied from 19% to 68% depending on observer type; the sensitivity of bacterial readings for positive bacterial cultures varied from 60% to 80%. The three observers agreed on a correct reading in only three children with viral and three with bacterial pneumonia. Because of poor observer agreement and appreciable false-negative errors when viral and bacterial readings were compared to titer increases and positive bacterial cultures, respectively, we conclude that radiographic findings are poor indicators of etiology diagnosis in ambulatory childhood pneumonias and, of themselves, are an insufficient data base for making therapeutic decisions.