The Thai Journal of Orthopaedic Surgery Vol.27 No.1 March 2002
Observer Variation in Garden’s Classification of
Femoral Neck Fractures
Prakasit Sanguanjitra, M.D., Banchong Mahaisavariya, M.D.
Department of Orthopedic Surgery and Rehabilitation
Faculty of Medicine Siriraj Hospital
Mahidol University, Bangkok 10700
ABSTRACT
Intra- and inter-observer variation in the use of Garden’ classification of femoral
neck fractures was evaluated. Pre-operative radiographs of 90 patients were assessed
independently by 10 observers. Each observer evaluated and classified each radiaograph
three times with at least 2 weeks interval between each time of assessment. The results
showed poor agreement among observers. Median Kappa coefficient of intra-observer
variation was 0.32 and inter-observer variation was 0.35, 0.26 and 0.38 in orderly. When
the classification was modified into nondisplaced and displaced, the level of agreement
was acceptable.
INTRODUCTION
Femoral neck fractures are most commonly classified according to Garden’s
classification. This classification is based on assessment of the antero-posterior
radiograph only. 1 Because this classification is intended to provide therapeutic guideline
and enable comparison between different treatment methods. It is therefore important that
the classification is reliable. Although there are some previous reports concerning the
reliability of this classification, 2,3 there is no any study has been conducted in our real
practice in Thailand. We therefore conducted this reliabilty study to assess the level of
agreement or its variations among both inter- and intra-observers.
MATERIAL AND METHOD
A total of 90 cases of femoral neck fractures treated in our department during
1998 to 2000 were included in this study. The antero-posterior (AP) radiographs of both
hip joints of the patients were used for the assessment and classified according to
Garden’s classification (Fig. 1). There were 67 females and 23 males whose ages ranged
from 18-95 years with an average of 70.1 years. There were 10 observers ( 5 from 3rd
year residents and 5 from 4th year residents) who participated in this study. The method
and details of Garden’s classification of femoral neck fractures were explained to all
observers prior to start the study. The series of radiographs was studied by each observer
working alone. Each observer were asked to assess the Garden’s stage according to the
description made in 19611 (Fig. 2):
The Thai Journal of Orthopaedic Surgery Vol.27 No.1 March 2002
Stage I: Incomplete fracture (impact valgus fracture)
Stage II: Complete fracture without displacement
Stage III: Complete fracture with partial displacement
Stage IV: Complete fracture with full displacement
Each observer were asked to assess the series of radiograph for 3 times with at
least 2 weeks duration of interval between each time of asssessment.
Statistics
To assess the reliability of this study we a
coefficient of agreement corrected for chance a es
between +1 and a negative worst possible value> is
+1; Kappa = 0 when the observed agreement Po is Pe
. Kappa values greater than 0.75 are defined as e
0.50 as poor agreement.3
RESULTS
Median Kappa coefficient of intraobserver
coefficient among observers of first, second and th
and 0.38 respectively.
When the conditions were categorized as
and displaced fractures (Garden’s stage III & IV)
with median Kappa coefficient was 0.61. The med
for each time of assessment were 0.59, 0.46 and 0.6
Fig.1 The antero-posterior radiograph
of both hip joints was used for the study
of observer variation of Garden’s
classification.
Fig.2 Four stages of femoral neck
fractures according to Garden’s
classification.
used the Kappa statistics. Kappa is
greement. Kappa can assume valu
/= -1. For perfect agreement Kappa
equal to expected chance agreement
xcellent agreement and values below
was 0.32 (0.24-0.44). Median Kappa
ird time of assessment were 0.35, 0.26
nondisplaced (Garden’s stage I & II)
the intra-observer agreement is higher
ian Kappa coefficient of intra-observer
4 respectively (Table I).
The Thai Journal of Orthopaedic Surgery Vol.27 No.1 March 2002
Table I Overall inter- and intra-observer variations from 10 observers in assessment of
90 cases of femoral neck fractures using Garden’s classification and modified Garden’s
classification.
Kappa coefficient Garden Modified Garden
Intra-observer 0.32 0.61
Inter-observer
1st reading 0.35 0.59
2nd reading 0.26 0.46
3rd reading 0.38 0.64
DISCUSSION
Garden’s classification of femoral neck fractures seemingly consists of four well
defined stages.1 The classification serves as a guideline for treatment and it is therefore
particulary important to determine the reliability of a classification system.
From previous studies concerning the observer variation in Garden’s
classification, Frandsen et al2 found that among 8 observers who assessed radiographs of
100 cases of femoral neck fractures only 22 cases that all observers were identified
identically.
Thomsen et al reported the observer variation in using Garden’s classification
among 6 observers in evaluation of 96 cases of femoral neck fractures.3 It was found that
the level of agreement was poor for overall classification with Kappa coefficient of 0.39.
The author also found that when reducing Garden’s system into non-displaced (Stage I &
II) and displaced fractures (Stage III & IV) the level of agreement became acceptable
with Kappa coefficient of 0.68.
From this study, we found a similarly poor agreement of the Garden’s
classification of femoral neck fracture. The median Kappa coefficient of intra-observer
among 10 observers was 0.32. We also experienced the same result that when reducing
the Garden system to non-displaced and displaced femoral neck fractures we found better
agreement with median Kappa coefficient of 0.61. This improvement was very similar to
the result from the study of Thomsen et al.3
On the basis of our study, we conclude that Garden’s classification is poorly
reproducible between observers. By reducing Garden’s system to the categories, non-
displaced fractures (Stage I & II) and displaced fractures (Stage III & IV) the level of
agreement become acceptable. This will make comparison between different studies
more reliable.
REFERENCES
1. Garden RS. Low angle fixation in fractures of the femoral neck. J Bone Joint Surg [Br] 1961; 43-B:
647-63.
2. Frandsen PA, Anderson E, Madson F, Skjodt T. garden’s classification of femoral neck fracture: an
assessment of inter-observer variation. J Bone Joint Surg [Br] 1988; 70-B: 588-90.
3. Thomsen NOB, Jensen CM, Skovgaard N, Pedersen MS, Pallesen p, Soe-Nielsen NH, Rosenklint A.
Observer variation in the radiographic classification of fractures of the neck of the femur using
Garden’s system. Int Orthop 1996; 20: 326-9.
Observer Variation in Garden’s Classification of
Femoral Neck Fractures
Faculty of Medicine Siriraj Hospital
Mahidol University, Bangkok 10700
ABSTRACT
INTRODUCTION
MATERIAL AND METHOD
Statistics
RESULTS
DISCUSSION
REFERENCES