Session I - Upper Extremity

The Epidemiology and Outcome of Proximal Humeral Fractures

Charles M. Court-Brown, MD; Margaret M. McQueen, MD; Ashima Garg, MD, Royal Infirmary of Edinburgh, Edinburgh, Scotland

Purpose: To investigate the epidemiology and outcome of proximal humeral fractures

Materials and Methods: In a five-year period, 1,027 consecutive adult proximal humeral fractures were prospectively reviewed. Each fracture was classified according to the Neer and OTA classifications, these being applied by a single surgeon to prevent inter-observer error. Those patients that were treated nonoperatively were managed in a collar-and-cuff or sling for 2 weeks, following which a physiotherapy regime consisting of graded exercises was introduced. Operatively managed patients were treated by the same regime, which was instituted about two weeks after surgery.

The patients' pre-fracture functional status was determined by direct questioning within two weeks of injury by an independent research physiotherapist who was blinded to the fracture types. The patients were asked about their domicile and their degree of independence. Their ability to undertake a number of daily tasks such as housework, shopping, dressing, personal hygiene and driving was assessed. All patients were then followed up at 6, 13, 26 and 52 weeks after fracture. At these times the Neer score was assessed by the same research physiotherapist who also assessed the patient's abduction and flexion strength using a spring balance technique. The time taken for the patients to return to their pre-fracture activities was also noted. Radiological assessment was undertaken at each presentation using antero-posterior and modified axial radiographs.

Results: The average age was 66 years and the highest age-specific incidence was seen between 80-89 years. Using the Neer classification, 49.9% were minimally displaced fractures and 27.7% were 2-part surgical neck fractures. Applying the OTA classification showed that 67% were Type A unifocal fractures, 21.1% were Type B bifocal fractures and 5.9% were Type C anatomical neck fractures. Four individual fracture types accounted for 50% of all fractures. The commonest fracture was the B1.1 impacted valgus fracture (14.6%) followed by the A2.2 impacted varus fracture (13.1%), the A3.2 translated surgical neck fracture (12.7%) and the A1.2 displaced greater tuberosity fracture (10.3%). Age was the main determinant of fracture complexity with older patients having more severe fractures.

The average Neer score at 6 weeks was 57 with 72 being recorded at 13 weeks, 80 at 26 weeks and 85 at 52 weeks. Statistical analysis showed that the Neer score and age were independently predictive of outcome but the OTA classification was not. One hundred and sixteen (11.3%) fractures were treated operatively. Sixty-three (9.2%) of the type A fractures were treated operatively in addition to 20 (7.2%) of the type B fractures and 33 (54.1%) of the type C fractures. The commonest type A fracture to be treated surgically was the A3.2 translated two-part fracture. Of the 20 surgically treated type B fractures seven were B2.2 fractures, these being the classic Neer three-part fracture associated with rotation of the humeral head fragment. The other 13 type B fractures that were treated surgically involved a number of different fracture sub-groups. Thirty one of the 33 operations undertaken to treat type C fractures involved the insertion of a hemiarthroplasty prosthesis and all except one of these operations were undertaken in C2 or C3 fractures.

Analysis showed no difference in the age or pre-fracture level of function between patients treated nonoperatively and those treated by surgery. However analysis of the Neer score, abduction and flexion strength and the time taken to return to normal functional activities showed no evidence that surgery helped any elderly patient. Even in the OTA Type C2 and C3 fractures patients treated with a hemiarthroplasty prosthesis fared less well than those treated nonoperatively.

There were 11 nonunions with the highest incidence of nonunion occurring in the A3.2 translated surgical neck fracture (3%). No patient with a non-union had recordable flexion or abduction power at one year.

Discussion: It is likely that the age-specific incidence of proximal humeral fractures is rising and that surgeons will have to treat increasingly older patients. Most of the techniques that are in current use were designed for younger people with better bone stock, and our results suggest that surgery in this elderly group does not improve outcome. This is not surprising when one considers that normal shoulder function in patients in their ninth decade is only 65 - 70% of that seen in 30-year-old patients. Elderly patients also have lower expectations, and our results suggest that they return to their normal activities as quickly as younger patients despite a lower Neer score.

Conclusion: Proximal humeral fractures in the elderly are best treated nonoperatively regardless of the morphology of the fracture. There is no evidence that surgery improves shoulder function. The main determinant of outcome is age, and while the Neer classification predicts outcome the OTA classification does not.