Collin et al,1 2016, Sweden | 44 consecutive patients with low-energy trauma and clinical suspicion of hip fracture who had negative or inconclusive radiographs followed by an inconclusive CT scans and then had MRI scans. The primary reports were classified into negative, suspicious or definite fracture. All images were independently reviewed as part of this study by two different experienced radiologists and classified into negative, suspicious or definite fracture. | Single-centre, retrospective study. | CT compared with MRI. | MRI changed the primary CT diagnosis in 27 cases (61%). 20 fractures were identified and 7 suspected fractures were found to be normal. | Small single-centre, retrospective study. MRI scanning may have been avoided if the initial CT scans had been reported or reviewed by the more experienced radiologists in the first place. |
Radiologist variability in reporting the CT/MRI. | Senior reviewers of CT disagreed with primary CT report in 24 of 44 cases. |
MRI. | Initial reports and reviews identified 20 fractures and 24 cases with no fracture. Observer agreement for MRI kappa=1.00. |
CT. | 26 CTs initially reported as no fracture—20 remained so after review. 18 suspicious CTs—10 and 9 declared negative by each of the two reviewers. Seven and eight cervical fractures identified by the two reviewers and one trochanteric fracture identified by both reviewers. Observer agreement kappa=0.87 (95% CI 0.72 to 1.00). |
Haubro et al,2 2015, Denmark | Prospective recruitment of patients with hip pain after low-energy fall with no fracture on plain X-ray between 2008 and 2012. A total of 1588 proximal hip fractures were identified. 67 patients underwent both CT and MRI, as primary X-ray was inconclusive. Junior and senior radiologists then reported the scans. | Prospective cohort study. | Occult fractures. | 15 occult fractures identified by MRI. In six cases there was a discrepancy between CT and MRI findings. Three CT-missed fractures needed surgery. | Presumed gold standard of MRI. |
Sensitivity of CT. | 0.87 (95% CI 0.6 to 0.98) vs 0.67 (95% CI 0.38 to 0.88) senior vs resident radiologist. |
Rehman et al,3 2016, Scotland | Single-centre between 2008 and 2012. Patients included had post-traumatic hip pain with no fracture on X-ray. Patients had either CT or MRI depending on physician choice, patient compliance or availability. | Retrospective review. | Occult fractures. | Occult fractures were suspected in 179 patients over a 4-year period, 71 (39.6%) of these had confirmed occult hip fracture. 37 of these were detected using CT and 34 using MRI. 51 of those had operative management. No readmissions to hospital with missed fractures in the following 12 months. | Retrospective. Single-centre. No direct comparison of images for each patient. Not known if any fractures were missed or presented to other EDs. |
Time to further imaging. | Time to MRI scan was longer than CT (3.78±3.014 days vs 1.78±1.68 days, p<0.05). |
Hakkarinen et al,4 2012, USA | Patients over 60 years old who had a discharge diagnosis of hip fracture. Notes review over 3 years. | Retrospective notes review, single-centre. | Occult fractures. | Occult hip fracture rate about 1% (24 out of a total of 235 hip fractures). | Single-centre, retrospective. No direct comparison between CT and MRI. Potential to have missed patients with negative hip plain films that were sent home without further imaging. |
Incidence of occult hip fractures missed by CT. | CT detected 18 of 24 occult fractures. Three patients had positive fracture on MRI which was missed on CT. |
Gill et al,5 2013, UK | Patients presenting with a clinical suspicion of hip fracture with inconclusive plain films. All patients were subsequently sent for either MRI or CT depending on availability and clinician choice. | Retrospective notes review, single-centre. | Occult fractures. | 34 patients had a confirmed occult hip fracture out of the 1353 hip fractures identified. | Single-centre, retrospective. No direct comparison between imaging modalities as no patient had both CT and MRI. |
CT detection vs MRI detection. | CT was performed in 61 cases, 23 of whom had a fracture (38%). 31 patients received MRI, 11 of whom had a fracture (36%). No patient had both imaging modalities. |
Lubovsky et al,6 2005, Israel | 13 patients who fell and had no evidence of a fracture on plain X-rays. Six patients had a CT scan and MRI, seven only had MRI. | Cohort. | CT vs MRI. | Four of the six CTs produced incorrect diagnoses based on MRI. | Very small cohort study. MRI assumed as the gold standard. Single radiologist viewing all images, potential for bias. |
MRI. | All MRI scans deemed accurate. |