OTA 1999 Posters

Reliability of Radiographic Assessment of Tibial Fracture Healing and Reduction following Intramedullary Fixation

Daniel B. Whelan, MD; Mohit Bhandari, MD; Michael D. McKee, MD; Hans Kreder, MD; David Stephen, MD; Stephen Connolly, MD; Emil H. Schemitsch, MD, St. Michael's Hospital, Toronto, Ontario, CANADA

Purpose: Intramedullary nailing of tibial fractures is a well-described and frequently employed modality of treatment. Following fixation, determination of tibial fracture healing is based on regular clinical and radiographic assessments. Despite the number of studies on tibial intramedullary nailing that have utilized fracture union as an outcome measure, no comprehensive, reliable, and universal guidelines exist to radiographically evaluate tibial fracture healing. Similarly, although there are accepted criteria to evaluate tibial fracture reduction from standard radiographs, the assessment of displacement and angulation has not previously been examined for reliability or reproducibility among several observers. Thus this study examines the intra- and interobserver agreement among orthopaedic traumatologists in assessing stages of tibial fracture healing and acceptability of reduction.

Methods: Thirty sets of anterior-posterior and lateral radiographs of tibial shaft fractures treated with intramedullary fixation were selected from a trauma database to represent fractures at various stages of healing. Four surgeons independently reviewed the radiographs on two separate occasions (8 weeks apart). The examiners were blinded to patient history, clinical examination, and age of the fracture at the time of each radiograph. For each radiograph, the surgeon scored the degree of union using a scale developed by Hammer et al. . ¹ Moreover, the observers were asked to comment on their general impression of fracture healing, to quantify the number of cortices definitively bridged by callus, the extent of the callus, and the number of cortices with a recognizable fracture line. A second portion of the assessment of each radiograph dealt with fracture reduction. The observers were to comment on general fracture reduction as well as specify the amount of shortening and/or distraction, amount of varus or valgus angulation, and amount of anterior or posterior angulation. Inter-observer and intra-observer reliabilities among surgeons were reported with quadratically weighted kappa (k) values (± standard error). A weighted kappa value of > 0.65 was considered good agreement.

Results: In general, the agreement between observers in assessing fracture healing was better than that for fracture reduction. The radiographic union scale showed moderate overall agreement (k = 0.6 ± 0.04); however, interobserver reliability in the surgeons' general impression of fracture healing was slightly higher (k = 0.67 ± 0.04). Individual kappa values for each grade in the scale suggested that agreement was best for the highest (grade 5, k = 0.55) and lowest (grade 1, k =0.55) grades. According to the scale, these grades represented fractures that were definitively not healed or definitively healed respectively. All other grades in the scale, which corresponded to relatively indeterminate states of healing, agreed less well (k = 0.31). Agreement was highest for the number of cortices bridged by callus (k= 0.75 ± 0.07) and the number of cortices with a visible fracture line (k = 0.70 ± 0.07). The extent of the callus was not as reliable in the assessment of fracture healing (k = 0.57 ± 0.07). When asked to indicate their impression of general fracture reduction, the overall percentage agreement was 85.6% that reduction was adequate. Overall agreement among the surgeons was poor regarding the amount of shortening and/or distraction (k = 0.31 ± 0.04), varus/valgus angulation(k = 0.46 ± 0.07), and anterior/posterior angulation (k = 0.36 ± 0.05). As with the fracture union assessments however, agreement on reduction was best for those fractures at the extremes of displacement or angulation. Intraobserver agreement, based on the observers second viewing of the films, approached excellent for the radiographic union scale (k = 0.76) as did that for the overall impression of healing (k = 0.89) and for the number of cortices bridged by callus (k = 0.82). As the results for interobserver reliability were generally poor for assessment of fracture reduction, intraobserver kappa values were not determined for these data.

Discussion and Conclusion: Much work has been done in recent years on interobserver reliability and the intraobserver reproducibility of previously well accepted fracture classification systems. The rationale for this scrutiny is to ensure that the information these classification systems provide is consistent and can be used in guiding treatment and suggesting prognosis with relatively predictable results. The same principle of ensuring reproducibility and reliability can be applied to other aspects of orthopaedic fracture management, including the evaluation of healing fractures. There are currently no validated scales that enable surgeons to grade fracture healing nor have the accepted criteria for assessment of reduction been analyzed for reliability. In this study, the agreement among surgeons in determining stage of fracture union was found to be better than that for assessing fracture reduction. The scale by Hammer et al. was not found to have substantial reliability; however, surgeons were able to reliably agree upon the number of cortices bridged by bone (or the number of cortices with a visible fracture line). Thus, studies, which use these two parameters as endpoints, may decrease variability in the assessment of fracture healing.

1. Hammer RR, Hammerby S, Lindholm B, Accuracy of radiologic assessment of tibial shaft fracture union in humans., Clin Orthop, 199:233-238,1985.