SLR - June 2021 - Douglas J. Cooper
Inter- and Intraobserver Reliability of Non-Weight-Bearing Foot Radiographs Compared with CT in Lisfranc Injuries
Reference: Ponkilainen VT, Partio N, Salonen EE, Riuttanen A, Luoma EL, Kask G, Laine HJ, Mäenpää H, Päiväniemi O, Mattila VM, Haapasalo HH. Inter- and Intraobserver Reliability of Non-Weight-Bearing Foot Radiographs Compared with CT in Lisfranc Injurie. Arch Orthop Trauma Surg. 2020 Oct;140(10):1423-1429.
Level of Evidence: Level III
Scientific Literature Review
Reviewed By: Douglas J. Cooper, DPM
Residency Program: Southern Arizona Veterans Affairs Health Care System – Tucson, AZ
Podiatric Relevance: The ability to accurately and consistently diagnose a Lisfranc injury is extremely important to properly treat the patient. The focus of this study was to better understand the interobserver reliability and intraobserver reliability of a non-weight-bearing foot radiograph, compared with CT scans.
Methods: A level III study was performed analyzing 456 patients who had received foot radiographs and CT scans. Data was received from two separate hospitals within a four-year window. CT scans were taken numbering 174 patients with non-displaced Lisfranc injury, and 59 patients with a displaced Lisfranc injury. Each patient also had radiographs taken, in anteroposterior, lateral, and oblique views. The non-weight-bearing foot radiographs were then evaluated twice at three-month intervals. The observers included six orthopaedic surgeons (three senior attendings and three residents). They were each asked a series of questions for each film set: “Is there an injury at the Lisfranc joint?”; (Yes/No), “If you answered yes, describe the findings”, and “Are there any other injuries”; (Yes/No). The radiographs were then randomly mixed for second observation. Again, the six observers independently assessed the radiographs and CT scans of the same patients in two different moments at an interval of three months.
Results: The Fleiss Kappa coefficient for interobserver reliability demonstrated moderate correlation (0.50, 95 percent CI 0.45-0.55). Substantial correlation using Cohen Kappa was found for intraobserver reliability (0.71, 95 percent CI 0.64 – 0.85). The mean sensitivity for all 6 observers was 76.1% and specificity measured 85.3 percent. The number of missed cases was higher among non-displaced injuries than in the displaced injuries. Three cases with a non-displaced Lisfranc injury were missed by all evaluators. The false negative rate was 23.9 percent and false positive rate was 14.7 percent. Additionally, there was no statistically significant differences in findings between senior attendings and residents in sensitivity, specificity, positive predictive value, or negative predictive value.
Conclusions: Lisfranc injuries were able to be diagnosed on conventional radiographs with moderate agreement between observers, and substantial agreement in the same observer at different time points, however, substantial number of injuries (24 percent) are missed if only non-weight-bearing radiographs are used for diagnosis. The authors recommended that CT imaging be implemented, especially if there are “subtle signs” of injury. Additionally, given the large data sample involved in this study, the value of the results achieved should not be overlooked. A possible limitation within the study was identified being that the imaging was evaluated by orthopaedic surgeons and residents, who are familiar with Lisfranc injuries. If an additional study were to be performed with primary healthcare physicians, the results and data obtained might be of more value since they are usually the first to diagnose and refer to subspecialties. Overall, this article helps to confirm the need and efficacy of additional imaging to normal foot films, in specific, CT images.