The admission radiographs for 404 patients with a displaced intracapsular hip fracture treated by reduction and internal fixation were classified using five different variables. These were the Garden grade, a modified Garden grading, a ratio of fracture displacement, direct measurement of fracture shortening and trochanteric shortening. Inter-observer reliability of the various classifications was also studied.

Only trochanteric shortening had an acceptable degree of inter-observer variation. For the Garden grading equal numbers of grade III and IV fracture healed. For the modified Garden grading 36% of Grade III fractures developed non-union against 48% of grade IV fractures (p value =0.02). The ratio method and fracture shortening were related to fracture healing complications, but trochanteric shortening was predictive of fracture healing (15.2 mms versus 11.0 mm), although the usefulness of this measure in clinical practice has to be questioned.