The purpose of this study was to develop a convolutional neural network (CNN) for fracture detection, classification, and identification of greater tuberosity displacement ≥ 1 cm, neck-shaft angle (NSA) ≤ 100°, shaft translation, and articular fracture involvement, on plain radiographs. The CNN was trained and tested on radiographs sourced from 11 hospitals in Australia and externally validated on radiographs from the Netherlands. Each radiograph was paired with corresponding CT scans to serve as the reference standard based on dual independent evaluation by trained researchers and attending orthopaedic surgeons. Presence of a fracture, classification (non- to minimally displaced; two-part, multipart, and glenohumeral dislocation), and four characteristics were determined on 2D and 3D CT scans and subsequently allocated to each series of radiographs. Fracture characteristics included greater tuberosity displacement ≥ 1 cm, NSA ≤ 100°, shaft translation (0% to < 75%, 75% to 95%, > 95%), and the extent of articular involvement (0% to < 15%, 15% to 35%, or > 35%).Aims
Methods
Accurate characterisation of fractures is essential in fracture management trials. However, this is often hampered by poor inter-observer agreement. This article describes the practicalities of defining the fracture population, based on the Neer classification, within a pragmatic multicentre randomised controlled trial in which surgical treatment was compared with non-surgical treatment in adults with displaced fractures of the proximal humerus involving the surgical neck. The trial manual illustrated the Neer classification of proximal humeral fractures. However, in addition to surgical neck displacement, surgeons assessing patient eligibility reported on whether either or both of the tuberosities were involved. Anonymised electronic versions of baseline radiographs were sought for all 250 trial participants. A protocol, data collection tool and training presentation were developed and tested in a pilot study. These were then used in a formal assessment and classification of the trial fractures by two independent senior orthopaedic shoulder trauma surgeons.Objectives
Methods
Controversy surrounds the management of displaced
three- and four-part fractures of the proximal humerus. The percutaneous
Resch technique of stabilisation involves minimal soft-tissue dissection
and a reduced risk of stiffness and avascular necrosis. However,
it requires a second operation to remove Kirschner wires and the
humeral block. We describe a modification of this technique that
dispenses with the need for this second operation and relies on
a sequential pattern of screw placement. We report the outcome of
32 three- or four-part fractures of the proximal humerus treated
in this way at a mean follow-up of 3.8 years (2 to 8)). There were
14 men and 18 women with a mean age of 56 years (28 to 83). At final
follow-up the mean Oxford shoulder scores were 38 (31 to 44) and
39 (31 to 42), and the mean Constant scores were 79 (65 to 92) and
72 (70 to 80) for three- and four-part fractures, respectively. We
further analysed the results in patients aged <
60 years with
high-energy fractures and those aged ≥ 60 years with osteoporotic
fractures. There were no cases of nonunion or avascular necrosis. The results were good and comparable to those previously reported
for the Resch technique and other means of fixation for proximal
humeral fractures. We would recommend this modification of the technique
for the treatment of displaced three-part and four-part fractures
in patients both younger and older than 60 years of age.
We evaluated the cost and consequences of proximal femoral fractures requiring further surgery because of complications. The data were collected prospectively in a standard manner from all patients with a proximal femoral fracture presenting to the trauma unit at the John Radcliffe Hospital over a five-year period. The total cost of treatment for each patient was calculated by separating it into its various components. The risk factors for the complications that arose, the location of their discharge and the mortality rates for these patients were compared to those of a matched control group. There were 2360 proximal femoral fractures in 2257 patients, of which 144 (6.1%) required further surgery. The mean cost of treatment in patients with complications was £18 709 (£2606.30 to £60.827.10), compared with £8610 (£918.54 to £45 601.30) for uncomplicated cases (p <
0.01), with a mean length of stay of 62.8 (44.5 to 79.3) and 32.7 (23.8 to 35.0) days, respectively. The probability of mortality after one month in these cases was significantly higher than in the control group, with a mean survival of 209 days, compared with 496 days for the controls. Patients with complications were statistically less likely to return to their own home (p <
0.01). Greater awareness and understanding are required to minimise the complications of proximal femoral fractures and consequently their cost.
We treated 47 patients with a mean age of 57 years (22 to 88) who had a proximal humeral fracture in which there was a severe varus deformity, using a standard operative protocol of anatomical reduction, fixation with a locking plate and supplementation by structural allografts in unstable fractures. The functional and radiological outcomes were reviewed. At two years after operation the median Constant score was 86 points and the median Disabilities of the Arm, Shoulder and Hand score 17 points. Seven of the patients underwent further surgery, two for failure of fixation, three for dysfunction of the rotator cuff, and two for shoulder stiffness. The two cases of failure of fixation were attributable to violation of the operative protocol. In the 46 patients who retained their humeral head, all the fractures healed within the first year, with no sign of collapse or narrowing of the joint space. Longer follow-up will be required to confirm whether these initially satisfactory results are maintained.
The surgical treatment of three- and four-part fractures of the proximal humerus in osteoporotic bone is difficult and there is no consensus as to which technique leads to the best outcome in elderly patients. Between 1998 and 2004 we treated 76 patients aged over 70 years with three- or four-part fractures by percutaneous reduction and internal fixation using the Humerusblock. A displacement of the tuberosity of >
5 mm and an angulation of >
30° of the head fragment were the indications for surgery. Of the patients 50 (51 fractures) were available for follow-up after a mean of 33.8 months (5.8 to 81). The absolute, age-related and side-related Constant scores were recorded. Of the 51 fractures, 46 (90.2%) healed primarily. Re-displacement of fragments or migration of Kirschner wires was seen in five cases. Necrosis of the humeral head developed in four patients. In three patients a secondary arthroplasty had to be performed, in two because of re-displacement and in one for necrosis of the head. There was one case of deep infection which required a further operation and one of delayed healing. The mean Constant score of the patients with a
Secure fixation of displaced proximal fractures of the humerus is a challenging problem. A total of 32 patients with acutely displaced three- or four-part proximal fractures of the humerus were treated by open reduction and internal fixation using the proximal humeral internal locking system (PHILOS) plate. There were 23 women and nine men with a mean age of 59.9 years (18 to 87). Data were collected prospectively and the outcomes were assessed using the Constant score. The mean follow-up was for 11 months (3 to 24). In 31 patients (97%) the fracture united clinically and radiologically at a mean of 10 weeks (8 to 24). The mean Constant score at final review was 66.5 (30 to 92). There was no significant difference in outcome when comparing patients aged more than 60 years (18 patients) with those aged less than 60 years (14 patients) ( This plate provides an alternative method of fixation for fractures of the proximal humerus. It provides a stable fixation in young patients with good-quality bone sufficient to permit early mobilisation. Failure of the screws to maintain fixation in the elderly remains a problem.
We used an inverted shoulder arthroplasty in 43 consecutive patients with a mean age of 78 years (65 to 97) who had sustained a three- or four-part fracture of the upper humerus. All except two were reviewed with a mean follow-up of 22 months (6 to 58). The clinical outcome was satisfactory with a mean active anterior elevation of 97° (35° to 160°) and a mean active external rotation in abduction of 30° (0° to 80°). The mean Constant and the mean modified Constant scores were respectively 44 (16 to 69) and 66% (25% to 97%). Complications included three patients with reflex sympathetic dystrophy, five with neurological complications, most of which resolved, and one with an anterior dislocation. Radiography showed peri-prosthetic calcification in 36 patients (90%), displacement of the tuberosities in 19 (53%) and a scapular notch in ten (25%). Compared with conventional hemiarthroplasty, satisfactory mobility was obtained despite frequent migration of the tuberosities. However, long-term results are required before reverse shoulder arthroplasty can be recommended as a routine procedure in complex fractures of the upper humerus in the elderly.
Our study was undertaken to assess the inter- and intra-observer variability of the classification system of Sanders for calcaneal fractures. Five consultant orthopaedic surgeons with different subspecialty interests classified CT scans of 28 calcaneal fractures using this classification system. After six months, they reclassified the scans. Kappa statistics were used to analyse the two groups. The interobserver variability of the classification system was 0.32 (95% confidence interval (CI) 0.26 to 0.38). The subclasses were then combined and assessment of agreement between the general classes as a whole gave a kappa value of 0.33 (95% CI 0.25 to 0.41). The mean kappa value for intra-observer variability of the classification system was 0.42 (95% CI 0.22 to 0.62). When the subclasses were combined, it was 0.45 (95% CI 0.21 to 0.65). Our results show that, despite its popularity, the classification system of Sanders has only fair agreement among users.
