Twenty-two studies containing 41 observations were identified. There were 21 comparisons of diagnostic accuracy (n=3,655), 17 comparisons of process efficiency (n=11,567), 1 comparison of patient care and 2 observations of cost-savings (n not known).
Diagnostic accuracy.
The use of low resolution displays (approximately 0.5K x 0.5K pixels) to make a primary diagnosis from chest and bone images was not supported (1 study). Ten out of 20 observations suggested that conventional film was either equivalent to, or more accurate than digital images viewed on a workstation or printed to film; however, this was based on, at best, weak evidence and may vary according to the diagnosis. The available data suggested that conventional imaging is more accurate than digital imaging, particularly for primary diagnosis of subtle manifestations of diseases. The data also suggested that printing digital images to film does not reduce their diagnostic accuracy. As yet, there has not been a definitive demonstration that digital imaging, either viewed on a workstation or printed to film, is equivalent to analogue film for making a primary diagnosis of all the clinical conditions that present in a varied patient population.
Work process efficiency.
The evidence of work process efficiency was limited. However, there was a suggestion that PACS are more efficient at generating and delivering images than film-based systems. Two studies reported that the time interval between the completion of an examination and the image being available for viewing was longer for film images than for workstation images. Two studies noted that the time taken to retrieve archived images was significantly shorter for PACS than for films. Four of the 6 studies reported no significant difference between the time taken to interpret film versus workstation images, while one study was in favour of film and one in favour of PACS. One study found that lost images were twice as common for film as for images archived electronically in a PACS, and another found that PACS led to no statistically significant decrease in the time interval between when the imaging exam was completed to the time at which the ordering physician first encountered imaging information.
Clinical care and patient outcomes.
In one study, for the subset of patients in which an image-based clinical action was reported and timed, it was reported that the clinical action was taken significantly faster when workstation imaging was used (2.5 hours) than when films were reviewed (4.4 hours).