Interobserver variability among radiologists for diagnosis of scaphoid fractures by computed tomography

Diagnosis of Suspected Scaphoid Fractures

»

Suspected scaphoid fractures are a diagnostic and therapeutic challenge despite the advances in knowledge regarding these injuries and imaging techniques. The risks and restrictions of routine immobilization as well as the restriction of activities in a young and active population must be weighed against the risks of nonunion that are associated with a missed fracture.

»

The prevalence of true fractures among suspected fractures is low. This greatly reduces the statistical probability that a positive diagnostic test will correspond with a true fracture, reducing the positive predictive value of an investigation.

»

There is no consensus reference standard for a true fracture; therefore, alternative statistical methods for calculating sensitivity, specificity, and positive and negative predictive values are required.

»

Clinical prediction rules that incorporate a set of demographic and clinical factors may allow stratification of secondary imaging, which, in turn, could increase the pretest probability of a scaphoid fracture and improve the diagnostic performance of the sophisticated radiographic investigations that are available.

»

Machine-learning-derived probability calculators may augment risk stratification and can improve through retraining, although these theoretical benefits need further prospective evaluation.

»

Convolutional neural networks (CNNs) are a form of artificial intelligence that have demonstrated great promise in the recognition of scaphoid fractures on radiographs. However, in the more challenging diagnostic scenario of a suspected or so-called "clinical" scaphoid fracture, CNNs have not yet proven superior to a diagnosis that has been made by an experienced surgeon.

The interobserver reliability of the diagnosis and classification of scaphoid fractures using high-resolution peripheral quantitative CT

Aims

Besides conventional radiographs, the use of MRI, CT, and bone scintigraphy is frequent in the diagnosis of a fracture of the scaphoid. However, which techniques give the best results remain unknown. The investigation of a new imaging technique initially requires an analysis of its precision. The primary aim of this study was to investigate the interobserver agreement of high-resolution peripheral quantitative CT (HR-pQCT) in the diagnosis of a scaphoid fracture. A secondary aim was to investigate the interobserver agreement for the presence of other fractures and for the classification of scaphoid fracture.

Methods

Two radiologists and two orthopaedic trauma surgeons evaluated HR-pQCT scans of 31 patients with a clinically-suspected scaphoid fracture. The observers were asked to determine the presence of a scaphoid or other fracture and to classify the scaphoid fracture based on the Herbert classification system. Fleiss kappa statistics were used to calculate the interobserver agreement for the diagnosis of a fracture. Intraclass correlation coefficients (ICCs) were used to assess the agreement for the classification of scaphoid fracture.

Results

A total of nine (29%) scaphoid fractures and 12 (39%) other fractures were diagnosed in 20 patients (65%) using HR-pQCT across the four observers. The interobserver agreement was 91% for the identification of a scaphoid fracture (95% confidence interval (CI) 0.76 to 1.00) and 80% for other fractures (95% CI 0.72 to 0.87). The mean ICC for the classification of a scaphoid fracture in the seven patients diagnosed with scaphoid fracture by all four observers was 73% (95% CI 0.42 to 0.94).

Conclusion

We conclude that the diagnosis of scaphoid and other fractures is reliable when using HR-pQCT in patients with a clinically-suspected fracture. Cite this article: Bone Joint J 2020;102-B(4):478–484.

Systematic Review of Diagnosis of Clinically Suspected Scaphoid Fractures

Background  Scaphoid fracture accounts for approximately 15% of acute wrist fractures. Clinical examination and plain X-rays are commonly used to diagnose the fracture, but this approach may miss up to 16% of fractures in the absence of clear-cut lucent lines on plain radiographs. As such, additional imaging may be required. It is not clear which imaging modality is the best. The goal of this study is to summarize the current literature on scaphoid fractures to evaluate the sensitivity, specificity, and accuracy of four different imaging modalities. Case Description  A systematic-review and meta-analysis was performed. The search term "scaphoid fracture" was used and all prospective articles investigating magnetic resonance imaging (MRI), computed tomography (CT), bone scintigraphy, and ultrasound were included. In total, 2,808 abstracts were reviewed. Of these, 42 articles investigating 51 different diagnostic tools in 2,507 patients were included. Literature Review  The mean age was 34.1 ± 5.7 years, and the overall incidence of scaphoid fractures missed on X-ray and diagnosed on advanced imaging was 21.8%. MRI had the highest sensitivity and specificity for diagnosing scaphoid fractures, which were 94.2 and 97.7%, respectively, followed by CT scan with a sensitivity and specificity at 81.5 and 96.0%, respectively. The sensitivity and specificity of ultrasound were 81.5 and 77.4%, respectively. Significant differences between MRI, bone scintigraphy, CT, and ultrasound were identified. Clinical Relevance  MRI has higher sensitivity and specificity than CT scan, bone scintigraphy, or ultrasound. Level of Evidence  This is a Level II systematic review.

Moderate Interrater Reliability in the Diagnosis of Scaphoid Waist Fractures

Background  Conventional radiographs have been shown to yield unreliable results in classifying scaphoid fractures. Computed tomography (CT) has been claimed to be the tool of choice in determining the treatment as well as fracture displacement. Purpose  The purpose of the study was to examine the interrater reliability and intrarater reproducibility in the decision-making of the treatment of scaphoid waist fractures. Patients and Methods  Fifty-one CT scans of scaphoid waist fractures were utilized. Seven orthopaedic surgeons with a particular interest in hand surgery independently scrutinized the scans classifying each in undisplaced, < 2 mm displaced, or > 2 mm displaced, and suggested a treatment of immobilization in cast or screw fixation. The Fleiss' and Cohen's kappa values using SPSS (Statistical Package for Social Science) version 24 were calculated and interpreted according to Landis and Koch. Results  The kappa value representing interrater reliability when choosing between operative or nonoperative treatment was 0.58. Interrater reliability of the distinction between < 2 mm displaced or > 2 mm displaced fractures was 0.61. On average 79.5% of the fractures were suggested treated nonoperatively and 20.5% operatively. Overall, intrarater reproducibility was 0.75 when classifying between < 2 mm displaced or > 2 mm displaced fractures. When choosing between operative or nonoperative treatment, intrarater reproducibility was 0.69. Conclusion  Moderate interrater reliability was found when choosing between nonoperative and operative treatment. The use of CT showed substantial reliability in the distinction between < 2 mm displaced and > 2 mm displaced fractures. Intrarater reproducibility was substantial when classifying between < 2 mm displaced and > 2 mm displaced fractures as well as when choosing between operative or nonoperative treatment. Level of Evidence  This is a Level III study.

The diagnostic accuracy of cross-sectional imaging for detecting acute scaphoid fractures in children: a systematic review

OBJECTIVE

To determine the diagnostic accuracy of cross-sectional imaging for the diagnosis of acute scaphoid fractures in children.

METHODS

A systematic review of Medline, Embase and Cochrane databases between 1980 and July 2017 was independently performed by two observers. Criteria for study inclusion in a meta-analysis and assessment of the quality of such studies using the QADAS tool, were predetermined.

RESULTS

No studies were eligible for inclusion in a meta-analysis. Three studies (of low quality when assessed against the STARD guidelines for reporting of studies of diagnostic accuracy) assessed MRI (performed between Days 2 and 10 after acute injury) for the diagnosis of scaphoid fractures in a total of 119 children (age range 6 to 16 years). Study 1 (45 children) reported inter-observer reliability of radiographs and MRI of 0.53 and 0.95 respectively. Study 3 (18 children) reported a negative predictive value of MRI (even as early as Day 2), of 100%. No measure of diagnostic accuracy or observer reliability was reported in Study 2 (56 children). In all 3 studies, MRI identified more scaphoid fractures (and other carpal injuries) than radiographs. Study 3 showed that follow-up MRI between Days 38 and 45 added no new information compared to initial MRI.

CONCLUSION

Based on a systematic review of the literature, there is currently no evidence on which to suggest an imaging protocol for suspected scaphoid fracture in children. Until such evidence is available, existing guidelines (which are based on findings from adult studies) should be followed. Advances in knowledge: (1) There is low quality evidence regarding the diagnostic accuracy of cross-sectional imaging for suspected scaphoid fractures in children and no evidence on which to propose an optimal imaging strategy. (2) Until such evidence is available, current guidelines (based predominantly on findings in adults and expert opinion) should be followed.