26.
High Reliability of the Hvidovre Algorithm for Hip Fracture Surgery
Henrik Palm, Eva Posner, Hans-Ulrik Ahler-Toftehøj, Silas Gylvin, Tobias
Aasvang, Kenneth Brian Holtz
Department of Orthopaedic Surgery Hvidovre University Hospital;
Department of Orthopaedic Surgery, Hvidovre University Hospital;
Department of Orthopaedic Surgery, Herlev University Hospital; Department
of Orthopaedic Surgery, Hvidovre University Hospital; Department of
Orthopaedic Surgery, Herlev University Hospital; Department of Orthopaedic
Surgery, Herlev University Hospital
Background:
Hip fracture treatment is controversial with high complication
rates. The Hvidovre Algorithm for choice of surgery has proven to be easy to
implement and has reduced the reoperation rate. The algorithm is however
based on the commonly used fracture classifications known to be unreliable,
which could result in different choice of implant between surgeons/hospitals.
Purpose / Aim of Study:
To investigate the reliability of the Hvidovre
Algorithm.
Materials and Methods:
4
+4 (consultant, fellow, resident, intern) observers
from two different hospitals used the algorithm to classify into 15 hip fracture
types (Garden 1-4 collum femoris incl. posterior tilt, Vertical collum femoris,
Basocervical and AO31 A1.1 to A3.3 trochanteric fractures) and choose
between 5 surgical procedures (Parallel implants, Prosthesis, 2 or 4 hole SHS
and IM nail). After individual assessment, the 4 observers made a hospital
decision for each patient. Observations were performed twice 10 weeks apart,
on pelvic, AP and axial x-rays from 100 consecutive hip fracture patients
(
F73/M27, mean age 80,) admitted Sept - Nov 2011 to Hvidovre Hospital.
Findings / Results:
Fracture type intra-reader agreement was 0.56-0.81 and
the inter-reader was 0.63-0.72, with 0.70 between the two hospitals. Posterior
tilt intraclass coefficient was 0.92-0.94 and the interclass coefficient was 0.90-
0.97.
Choice of implant intra-reader was 0.83- 0.94, including 0.90 and 0.92
for the two hospital decisions; inter-reader was 0.88-0.94, with 0.91 between
the two hospitals. Younger surgeons had the lowest agreement. Basocervical
and vertical collum femoris fractures were most difficult to classify.
Conclusions:
Al though fracture classification confirmed to be unreliable,
posterior tilt measurement and the Hvidovre Algorithm appears to be very
reliable with treatment agreement above 90% between hospitals.
