Introduction
Introduction
2020-2023
Supervisor
Co-Supervisor Co-Supervisor
Co-Supervisor
Associate Professor,
Department of Radiodiagnosis,
Maulana Azad Medical College &
I, Dr. Prajwal Gupta, hereby declare that the contents of this thesis entitled
FRACTURES” has not been submitted earlier in the candidature of any degree in Delhi
University. The observations embodied in this thesis and the interpretations therein have
been done by me and the thesis fulfills the rules and regulations of Delhi University. I
hereby give my consent for permission of availability of the thesis for photocopying and
G.N.E.C and G.I.P.M.E.R, New Delhi, under our guidance and supervision.
Supervisor
Co Supervisor Co Supervisor
Co Supervisor
Every mission needs a spirit of hard work and dedication to be put in the right path
to meet its destination and in my case, this credit goes to my guide, Dr Sumit Arora
It gives me utmost pleasure in expressing my words of gratitude for all the help I
received during this thesis work.I shall be forever indebted to my supervisor, Dr.
Medical College, for his constant supervision and guidance throughout the course of
care of such a mentor who has always encouraged my endeavours, kept a constant
vigilance on my mistakes and helped me correct them at the same time.I also express
Delhi, for his valuable guidance and assistance in the technological aspect of my
thesis.I owe this thesis work to my parents Dr. Vilas Bagwe and Mrs. Chhaya
Bagwe, my sister Mrs. Tejal Bagwe, who have given me love and unconditional
support in all my endeavours. No words are sufficient enough, nor can truly express,
the feeling of gratitude and respect I have for them. I am extremely thankful to my
seniors Dr. Abhay Meena, Dr Siddharth Trivedi, Dr Prateek Goel, Dr. Palash Gupta,
Dr. Gaurang Agarwal, Dr. Shekhar Tomar and Dr. Raj Kumar for sharing their
study. I would also like to thank Dr. Vaibhav Anand, from IIT Delhi for
and Dr. Piyush who helped me in statistical analysis of my thesis data. All this
would not have been possible if not for the co-operation and understanding of the
patients who agreed to be a part of this study, for whom I reserve my biggest share
of gratitude.
No.
1 INTRODUCTION 1-2
6 DISCUSSION 88-100
7 CONCLUSION 101
9 APPENDIX
139
✔ MASTER CHART
10 SUMMARY 140-142
GLOSSARY OF ABBREVIATIONS
2 D: 2 Dimensional
3 D: 3 Dimensional
A-P: Anteroposterior
AVN: Avascular
Necrosis
IL: Ilio-inguinal
Resonance Imaging
Pre-op: Pre-operative
Post-op: Post-operative
QR code: Quick Response code
curing
The increase in the life expectancy in the past decades has led to a growing elderly population
across the world, substantially increasing the incidence of fragility fractures around the hip.
Intertrochanteric fractures are the most common type of hip fractures often resulting from simple
falls in elderly osteoporotic individuals. Since these patients are frail often having coexistent
comorbid conditions, these fractures are associated with increased morbidity and mortality and
pose a unique challenge to health systems worldwide. Internal fixation with early mobilisation is
a widely accepted method of treatment of both stable and unstable intertrochanteric fractures(1–
3). With an increasing shift towards operative management of these fractures the incidence of
complications like postoperative cut out of lag screws in unstable fractures have increased
Stability of the fracture is an important predictor for the risk of implant cut out; therefore, pre-
operative evaluation using classification systems that accurately access the stability and
morphology of fracture patterns to determine the optimum choice of implant is an important step
to reduce the incidence of cut out(5). An extensive classification based preoperative assessment
is also essential to predict difficulty in achieving reduction, fracture patterns requiring open
reduction and to reduce the intra-operative time especially in elderly patients with coexistent
comorbidities who are at an increased risk for perioperative complications and subsequent
mortality.
A good classification should be simple enough to allow communication between surgeons while
being comprehensive enough to account for a majority of fracture patterns to guide the choice of
treatment, and should possess a high degree of inter- and intra-observer reliability. Radiographs
based classification systems like Boyd and Griffin(2), Evans(3), Jensen(6) and AO/ASIF(7)
despite their routine use, have poor inter-observer reliability due to difficulty in evaluation of the
trajectory of fracture line (especially in the coronal plane), the degree of comminution and the
severity of involvement of the lateral wall in radiographs. However, this information is essential
due to its impact on the mechanical stability of the fracture and the choice of implant(8–11).
AO/ASIF(7) is a standard classification system that is widely used in trauma settings. However,
studies by Crijns et al(11) and Van Emden et al(8) have shown a reduction in inter-observer
reliability at the subgroup level. In addition, it fails to take into account the instability of the
fracture due to lateral wall involvement. Lateral wall thickness is an important factor in
determining the stability of fracture and the choice of implant. A lateral wall thickness of less
than 20.5 mm is associated with a high risk of secondary fracture on placement of a sliding hip
screw(12). Recognition of lateral wall fracture is important as it is associated with high rates of
implant failure and re-operation following dynamic hip screw fixation due to uncontrolled
medialisation of the distal fragment(13,14). Coronal split extending into the greater trochanter
can lead to the loss of superolateral support and result in failure of fixation especially when a
sliding hip screw is being used as it is difficult to detect on radiographs and fractures which
might appear stable on plain radiographs with an adequate lateral wall thickness might in fact be
Some drawbacks of the conventional AO/ASIF(7) classification have been addressed in the new
31-A2 groups, and classifies isolated trochanteric fractures that were not classified in the original
classification. The differentiation based on lateral wall thickness has implications on the choice
anteroposterior radiographs and has similar problems with evaluation of fracture morphology,
detection of coronal split, lateral wall fracture and its effect on stability.
Posterior cortical integrity is essential to withstand the varus, retroversion stress transmitted by
the femoral neck, and to prevent varus displacement and retroversion of the proximal
fragment(9); however, the posterior cortex is thin and frequently comminuted and if bone on
bone impaction using sliding hip screw is applied on this, the fracture site may collapse.
Disruption of the calcar or posteromedial communition can lead to collapse with axial loading
The thicker anterior cortex is of utmost importance in these cases to provide stability; therefore,
any comminution at the anterior or posterior cortex can cause instability at the fracture site and
intact anteromedial fracture plane is essential in obtaining stable fixation for an unstable
Flexion of the proximal fragment indicates a difficult close reduction and is a predictor for the
possible need of open reduction(23). Ikuta et al(24) defined a ‘subtype-p’ in their classification
system based on lateral radiographs in which they determined that posterior displacement of the
head-neck fragment with respect to the shaft is unstable leading to difficult reduction and
excessive sliding of the head-neck fragment along with collapse of the fracture in patients with
Coronal split with en block dissociation of the greater trochanter along with the lesser trochanter
posteriorly as a single fragment was labelled as the ‘Banana’ fragment by Futamura et al(25),
and was recognised by shoda et al(26) as an inherently unstable fracture fragment frequently
Due to the complexity of intertrochanteric fractures it is easy to misinterpret the radiograph and
make an error in understanding the crucial points that help in deciding the stability and
With the advancement in technology, computed tomography (CT) is used to visualise the precise
morphology and trajectory of the fracture line. The process of 3-dimensional (3D) reconstruction
has the ability to detect fracture fragments that are not seen clearly on conventional radiographs.
Further, 3D reconstruction with segmentation and mapping of fracture lines can better detect the
involvement of the lateral wall, anterior and posterior comminution, flexion of the proximal
fragment, coronal plane fracture lines and the banana fragment. Fracture mapping can improve
our understanding of the morphology of fracture lines, which would facilitate the choice of
implant, improve pre-operative planning and help in determining the ease of reduction and risk
of fixation failure(21). These are of vital importance in the modality of fixation, intraoperative
Therefore, this study aims to assess the effect of addition of advanced imaging modalities like
2D-CT with multiplanar reconstruction and 3D reconstruction with segmentation on the inter-
Proximal femur fractures are the most common type of fractures in elderly with increasing
almost half of these fractures(8). The trochanteric region of the femur is enclosed by a line along
the intertrochanteric crest and another transverse line at the inferior border of the lesser
independence and improve their quality of life is the preferred method of treatment.
Classification systems are important for assessing the morphology, stability and the ease of
plan for their management and ascertaining the type of implant to be used for fixation. An ideal
classification system should be able to guide the choice of treatment and predict its outcome. It
should possess a high degree of inter-observer and intra-observer reliability while being simple
Historically, various classification systems have been proposed based on the anatomical location
In 1939, Moore(27) classified them based on the anatomical location of the fracture lines into
fractures of the upper end of femur and fracture dislocations. Fractures of the upper end of femur
were further subdivided into fractures of the head, neck and trochanters.
In 1946, Briggs and Keats(28) classified intertrochanteric fractures into five types in accordance
with the orientation of the fracture lines and the risk of coxa vara deformity due to impaction of
lines, they provide very little practical information regarding further management and prognosis.
Classifications were also proposed on the basis of the stability of intertrochanteric fractures by
In 1949, Boyd and Griffin(2) classified these fractures into four types considering the instability
of the fracture in the coronal and the sagittal planes along with the ease of achieving and
maintaining reduction.
In 1949, Evans(3) also proposed a classification based on the stability of fracture pattern, the
importance of calcar femorale and the propensity of the fracture to collapse leading to coxa vara.
He laid stress on the importance of the calcar to act as the medial cortical buttress and provide
fracture fragments, ability to achieve anatomical reduction in both planes along with the risk of
secondary fracture dislocation. He also used lateral radiographs for evaluation of the greater
trochanter and calcar femorale to predict the possibility of obtaining stable reduction and the risk
undisplaced two part fractures that are stable, unstable 3-fragment fractures with greater or lesser
The original AO/ASIF classification(7) (appendix 1) divided intertrochanteric fractures into three
major groups with each group being further divided into three subgroups based on the obliquity
of the fracture line and the degree of comminution. The A1 group having simple two fragment
fractures with oblique fracture line extending from the greater to the medial cortex. The A2
group includes multifragmentary fractures with posteromedial wall involvement along with
involvement of the adjacent medial cortex, the lateral cortex, however, remains intact. Fractures
in this group are generally unstable, depending on the size of the medial fragment. The A3 group
has the fracture line extending across both medial and lateral cortices, labelled as the reverse
oblique pattern. The differentiation in between the A1 and A2 groups and also in between the
In 2018, the original AO/ASIF(7) classification was modified with the differentiation between
the 31 A1 group and the 31 A2 based on the lateral wall thickness according to the method given
by Hsu et al(12). Thickness of greater than 20.5 mm has been classified as group A1 (competent
lateral wall) and less than 20.5 mm (incompetent lateral wall) as group A2. The A1 group was
also revised to include multiple fracture lines and the 31 A1.3 subgroup was modified to include
an independent lesser trochanteric fragment. The differentiation based on lateral wall thickness
has implications on the choice of implants. However, like the previous AO/ASIF(7)
classification, it utilises only anteroposterior radiographs and has similar problems with
evaluation of fracture morphology and stability. Despite these changes, the study by Chan et
al(29) and Zarie et al(30) did not demonstrate an adequate increase in reliability. The drawback
across the entire spectrum of radiographs based classifications is their inability to accurately
detect the fracture morphology, orientation of fragments, integrity of the lateral wall, flexion of
comminution, which are critical factors for determining fracture stability and vital to the aspect
The lateral wall of femur as defined by Gao et al(31) in 2017 is the area enclosed by the vastus
lateralis ridge and the intersection of the tangent along the inferior neck of femur to the lateral
wall of femur as shown in the figure 3. Both intra- and extra-medullary fixation devices are used
for fixation of intertrochanteric fractures. The choice of implant depends upon the stability of the
fracture and the integrity of the lateral wall with extra-medullary fixation (Dynamic Hip Screw)
being effective in stable fractures with an intact lateral wall and intra-medullary nail or
DHS(Dynamic Hip Screw) with modifications in unstable fractures(13,14). Fixed angle sliding
hip screws (Dynamic Hip Screw) that are commonly used in the fixation of these fractures,
function by allowing the controlled impaction of the proximal head-neck fragment with the distal
shaft fragment to achieve bone on bone stability. The use of a sliding hip implant, where the
lateral wall thickness is less than 20.5 mm, has a high risk of postoperative lateral wall fracture
Figure 3: Showing the area of lateral wall enclosed by (1) tangent along the inferior border of the
Intact lateral femoral wall provides a lateral buttress for the proximal fragment and allows for
controlled fracture impaction which prevents varus malposition and collapse providing a
conducive environment for fracture healing by allowing cyclic loading and remodelling after
Recognition of lateral wall fracture is important as it is associated with high rates of implant
failure and re-operation following dynamic hip screw fixation due to uncontrolled medialisation
of the distal fragment(13,14). These fractures are better treated by augmentation with a
trochanteric buttress plate or using intra-medullary implants which can act like a lateral
Coronal plane fracture line was first included in the classification given by Boyd and Griffin(2)
who in their original article described it as “One deceptive form of type II fracture is that which
lateral view reveals an additional fracture in the coronal plane.” In 2016, Cho et al(16) in their
article described the various morphological patterns of coronal splits originating from the ridge
on the most cephalic part of the trochanter labelled as the ‘trochanteric summit’ and exiting
through the intertrochanteric crest, through or adjacent to the lesser trochanter or through the
postero-medial cortex distal to the lesser trochanter(figure 4). Coronal splits involving the
trochanteric area are frequently missed on radiographic evaluation leading to the possibility of
failure of sliding hip screw if used in such fractures due to the loss of superolateral support.
Therefore, intertrochanteric fractures which might appear stable on plain radiographs with an
adequate lateral wall thickness might in fact be inherently unstable due to an undetected coronal
split(17). Coronal splits extending inferomedially and exiting through the posteromedial cortex
and lesser trochanter involve the lateral wall leading to an added element of instability.
Figure 4: Showing the various patterns of coronal split.
Calcar femorale is a region of dense compact bone which is vertically oriented and lies deep to
the lesser trochanter in the posteromedial part of the proximal femur (figure 5) which is involved
in the transmission of forces across the femoral neck to the shaft of the femur (34). The calcar is
oriented obliquely upwards and closely associated with both the principal tensile and
compressive trabeculae which helps in the re-distribution of the shear and rotatory stress that are
comminution will not be able to bear and redistribute the axial stress acting on it leading to the
collapse of the calcar with resultant rotational instability, coxa vara and weakening of the
trabeculae.
a primary fracture line, and the anterior wall is defined as comminuted if there is a third fracture
fragment along the main fragment line as shown in figure 7. An intact anteromedial cortical
support acts like a buttress allowing for controlled impaction of the head-neck fragment with the
shaft of femur, it becomes the primary stabilizer in cases where there is loss of the supporting
calcar(21). Anteromedial cortical apposition is necessary to arrest the excessive sliding of the
head-neck fragment, more so in cases with an incompetent lateral wall. Comminution along the
anterior cortex can lead to an increased risk of screw cut out, implant breakage and shaft
medialization(22).
A single large fragment created by a coronal split involving the trochanter and exiting
posteroinferiorly below the lesser trochanter such that the posterior part of the greater trochanter
and the lesser trochanter are fractured en block was first described by Shoda et al(26) as an
occult fragment leading to instability undetected on radiographs in 40 percent of cases but
instability was later labelled by Futamura et al(25) as the ‘banana fragment’ as shown in figure 8.
Ikuta et al(24) in 2019 classified the fractures into 3 subtypes based on the relationship of the
head-neck fragment with the shaft and their association with difficulty in achieving closed
reduction preoperativetly. Subtype-A, where the head-neck fragment is located anterior to the
shaft (in flexion with respect to the axis of the shaft of femur) indicates a difficult closed
reduction and may be irreducible if associated with a displaced lesser trochanter often requiring
open reduction. The subtype-N, where the head-neck fragment lies in the axis of the femur can
be reduced easily with traction on the fracture table, while the third subtype; subtype-P
represents the posterior displacement(extension) of the head-neck fragment with respect to the
shaft indicating a difficult close reduction. Fractures of this subtype if associated with an intact
greater trochanter are considered irreducible by closed methods, therefore, the presence of
flexion or extension of the head-neck fragment serves as a predictor for the possible need of open
Figure 9: Showing flexion and extension of the head-neck fragment with respect to the shaft of
femur.
With the advancement in technology, computed tomography(CT) is used to visualise the precise
morphology and trajectory of the fracture line. The process of 3-dimensional (3D) reconstruction
with segementation has the ability to detect fracture fragments that are not seen clearly on
conventional radiographs. Further, mapping of fracture lines can better detect the involvement of
the lateral wall, anterior and posterior comminution, flexion or extension of the proximal
fragment and coronal plane fracture lines. Fracture mapping can improve our understanding of
the morphology of fracture lines, which would facilitate the choice of implant, improve pre-
operative planning and help in determining the ease of reduction and risk of fixation failure(21).
These are of vital importance in the modality of fixation, intraoperative reduction manoeuvres
and prognostication.
In order to address the limitations of the existing classifications based on radiographs, newer CT
based classifications were proposed. Kijima et al(37) in 2014 proposed the area classification for
proximal femoral fractures based on the anatomical location of the fracture lines to help in
identifying the “dangerous” variants that are associated with increased instability along with a
higher risk of failure of fixation and pseudoarthrosis. In 2017, Shoda et al(26) put forward a
fragment based classification system based on 3D-CT reconstruction that described various
instability patterns as a function of combination of these fragments. Wada et al(38) and Futamura
et al(25) further developed classifications including 5 fragments to more accurately represent the
various morphological patterns of fracture lines. However, these newer CT based classification
systems have not been tested clinically and there is paucity of literature on their inter- and intra-
observer reliability.
For our study we added advanced imaging modalities to evaluate their effect on the inter and
concurrently understanding their influence on the detection of parameters that play a pivotal role
STUDY DESIGN: It was a descriptive observational study and the subjects were enrolled to
participate in the study after approval from the Institutional Ethics Committee, Maulana
PLACE OF STUDY: The study was conducted in the Department of Orthopaedics, Maulana
Consents were obtained from all patients after explaining the procedure.
the emergency and OPD and satisfying the below mentioned inclusion and exclusion criteria
Inclusion criteria-
3. Closed injury
Exclusion criteria-
3. Patients who have already undergone CT scans elsewhere and DICOM files of that
At 95% confidence level and 80% power, taking inter-observer kappa agreement for
and 3D CT as 0.28, 0.33 and 0.28 (Cavaignac et al(39)) and with an absolute error of 5%, the
Where,
n = sample size
k =no. of observer - 3
Total of 50 patients presenting to the emergency and OPD were enrolled in the study.
Step 1: Enrolment in the study
OPD of Lok Nayak Hospital were screened for their possible inclusion in the study using the
above-mentioned criteria.
Anteroposterior and lateral radiographs along with the CT scan of the hip with proximal femur
were done for each patient enrolled in the study. The CT scans were acquired using 128 slice
single source CT scanner (Erlangren, Germany). The images were acquired with the patient in
supine position and thin axial sections of 1 mm thickness were obtained for each patient. Thin
multiplanar images along coronal and sagittal planes were reconstructed from these axial set of
images. The acquired images were saved in the Digital Imaging and Communications in
The DICOM files were imported into the InVesalius software (Version 3.0, Centro de
fragments was done after thresholding and segmentation of the fracture fragments.
The command of thresholding and region growing was used to attenuate and separate soft tissue
structures from the pelvis and the femur. The pelvis was subtracted for better visualization of the
proximal femur and the fracture fragments (as shown in figure 10 and 11).
Figure 10: Thresholding. Figure 11: Region growing.
Segmentation:
Fracture fragments were segmented using split mask feature, differentially coloured and saved as
independent entities using the segmentation function in the software (figure 12 and 13).
3D Reconstruction:
3D reconstruction of the fracture fragments were done by exporting the segmented parts in the
The collected and processed data was segregated into patient specific folders which were given a
unique code (the first digit of the code denotes the reading and the second digit denotes the
patient number) in order to hide the patient identity from the observers (figure 15).
Two orthopaedic consultants (one senior orthopaedic surgeon and one junior orthopaedic
surgeon) and one consultant radiologist were chosen for the assessment, they were not provided
any clinical details regarding presentation or management of the patients presenting with these
fracture.
A training session was conducted to familiarise the observers with the 2018 AO/OTA
classification system, lateral wall fracture, coronal split, anterior and posterior comminution,
flexion of the proximal fragment and the presence of the banana fragment with the help of a
The examiners were also familiarised with the study design and the evaluation sheet which was
The observers were presented with the antero-posterior radiographs, lateral radiographs and were
asked to record the findings on the printed evaluation sheet provided to them. They were
provided as much time as needed to evaluate the radiographs independently, once each section of
the evaluation was complete they were not allowed to change their answers.
sections) and the findings were recorded on the second section of the evaluation sheet, taking
into consideration the radiographs along with the 2D-CT of the patient.
The previous two modalities were augmented with the 3D reconstructed, segmented digital
model of the proximal femur and the findings were recorded. Once the findings were recorded
After completion of all the 3 sections, the case was considered complete and the observer could
Following the first reading (assessment) the order of the patient folders was randomized to avoid
recall. The second evaluation was undertaken in the same manner as the first and the data
collected was organized in a tabular format and saved to a data bank for further statistical
evaluation.
A normal proximal femoral STL (stereolithography) file was imported into the software and
positioned in the four anatomical planes anterior, medial, lateral, posterior and exported as
separate files to act as templates for the creation of fracture lines (figure 16) .
Figure 16: Proximal femoral templates arranged in the four orthotropic views.
The 3D reconstructed fracture models were imported into the software, and with the help of its
move and rotate function the fragments were virtually reduced and aligned in the same
Figure 17: Virtually reduced fracture fragments arranged in the same orthotropic views as the
templates.
The image of the fracture model was imported into Adobe Photoshop (Version 20.0. 8, Adobe
Inc., United States of America) as the background layer as shown in figure 18 and the template
The opacity of the template layer was reduced to view the underlying fracture lines and using the
size adjustment tool the template was superimposed on the fracture model as shown in figure 19.
Once adequate superimposition was obtained, the fracture lines were marked on the template
layer and it was exported as the fracture map in the four orthotropic views as shown in figure 20.
.
Figure 19: Opacity of the template layer reduced and the underlying fracture line mapped onto
the template.
Figure 20: Showing the exported fracture map with four orthotropic views.
The observations were recorded as a function of instances of evaluation, since 50 cases were
INTER-OBSERVER RELIABILITY:
OBSERVER 2 3
1 0.37 0.47
(0.17-0.56) (0.31-0.62)
2 0.48
(0.30-0.65)
Mean 0.44
Table 1: Inter-observer Kappa values at the end of first reading, when only radiographs were
OBSERVER 2 3
1 0.48 0.34
(0.28-0.67) (0.31-0.62)
2 0.37
(0.21-0.52)
Mean 0.39
Table 2: Inter-observer Kappa values at the end of first reading, when radiographs and 2D-CT
1 0.51 0.41
(0.33-0.68) (0.27-0.54)
2 0.46
(0.28-0.63)
Mean 0.46
Table 3: Inter-observer Kappa values at the end of first reading, when radiographs, 2D-CT with
evaluation.
OBSERVER 2 3
1 0.38 0.31
(0.20-0.55) (0.13-0.48)
2 0.36
(0.16-0.55)
Mean 0.35
Table 4: Inter-observer Kappa values at the end of second reading, when only radiographs were
1 0.41 0.28
(0.23-0.58) (0.08-0.47)
2 0.38
(0.18-0.57)
Mean 0.35
Table 5: Inter-observer Kappa values at the end of second reading, when radiographs and 2D-CT
OBSERVER 2 3
1 0.49 0.41
(0.29-0.68) (0.23-0.58)
2 0.44
(0.22-0.65)
Mean 0.44
Table 6: Inter-observer Kappa values at the end of first reading, when radiographs, 2D-CT with
evaluation.
Subgroup Analysis:
The inter-observer reliability of the 31 A 1 subgroup decreased slightly as compared to the inter-
observer reliability for AO grouping. However, for the 31 A 2 subgroup analysis revealed a
The average kappa coefficient of inter-observer reliability in this subgroup was 0.356 (fair).
The average kappa coefficient for inter-observer reliability in this subgroup (A2) = 0.19
(slight)
The overall average kappa coefficient for inter-observer reliability of AO subgroups = 0.27 (fair)
INTRA-OBSERVER RELIABILITY:
OBSERVER 2 3
1 0.53 0.42
(0.35-0.70) (0.22-0.61)
2 0.42
(0.22-0.61)
Mean 0.45
Table 9: Intra-observer Kappa values when only radiographs were available for evaluation.
OBSERVER 2 3
1 0.44 0.47
(0.26-0.61) (0.27-0.66)
2 0.38
(0.20-0.55)
Mean 0.43
Table 10: Intra-observer Kappa values when radiographs and 2D-CT with multiplanar
OBSERVER 2 3
1 0.57 0.53
(0.39-0.74) (0.32-0.71)
2 0.42
(0.24-0.59)
Mean 0.51
Table 11: Intra-observer Kappa values when radiographs, 2D-CT with multiplanar reconstruction
Graph 1: Showing the comparison of inter- and intra-observer reliabilities on the three
radiological modalities.
Table 12: Showing the values of kappa coefficient for AO group and subgroups.
Graph 2: Showing the comparison of the mean inter- and intra-observer reliabilities between AO
OUTCOME ASSESSMENT
Kappa coefficient for intra- and inter-observer reliability was calculated and evaluated based on
the Landis and Koch scale(40). Reproducibility and agreement was considered better as the
It was deemed:
Slight (0-0.2)
Fair (0.21-0.4)
Moderate (0.41-0.6)
Substantial (0.61-0.8)
Both the inter- and intra-observer reliability was moderate for AO main grouping (0.41 and 0.46
respectively) but decreased to fair for AO sub grouping (0.27 and 0.32 respectively). There was a
substantial decrease in the inter-observer reliability to slight (0.19) from moderate (0.41) for the
AO 31-A2 subgroup due to the difficulty in calculation of lateral wall thickness along with the
instances (54%) on 2D-CT with multiplanar reconstruction and 180 instances (60%) on 3D
Graph 3: Showing the number of instances of detection (out of 300) and its percentage of
reconstruction with segmentation, as shown in the graph below. The occurrence of extension of
Extension of the proximal fragment which indicates a difficult closed reduction is relatively less
common as compared to the flexion of the head and neck fragment as shown in graph 4. While
the detection of both these entities is difficult on radiographs, as it requires an adequate lateral
view which is difficult to obtain in emergency settings, they are better detected on 2D-CT and
Anterior wall comminution was detected in 67 instances (22%) on radiographs, 201 instances
(67%) on 2D-CT with multiplanar reconstruction and 206 instances (68%) on 3D reconstruction
with segmentation, as shown in the graph below. Radiographs could detect comminution in a
majority of cases, however, the differentiation between the anterior and posterior comminution
and ascertaining its severity was challenging for observers. Higher modalities significantly
increased the detection of anterior and posterior wall comminution and helped in gauging its
severity.
Graph 6: Showing the number of instances of detection (out of 300) and its percentage of
Posterior wall comminution was detected in 125 instances (41%) on radiographs, 247 instances
(82%) on 2D-CT with multiplanar reconstruction and 255 instances (85%) on 3D reconstruction
Graph 7: Showing the number of instances of detection (out of 300) and the percentage of
Postero-medial comminution has been recognised as the most common form of instability noted
in intertrochanteric fractures which can lead to varus collapse it is more frequently observed than
wall comminution.
Coronal split was detected in 113 instances (37%) on radiographs, 189 instances (63%) on 2D-
segmentation, as shown in the graph below. Detection of coronal plane fracture lines is essential
for the assessment of stability of intertrochanteric fractures since they play an important role in
providing supero-lateral support when DHS implant is used for fixation, failure to detect them
Lateral wall fracture was detected in 45 instances (15%) on radiographs, 80 instances (26%) on
2D-CT with multiplanar reconstruction and 138 instances (46%) on 3D reconstruction with
Graph 10: Showing the number of instances of detection (out of 300) and its percentage of
Banana fragment was detected in 5 instances (1%) on radiographs, 16 instances (5%) on 2D-CT
Banana fragment occurs on the posterior aspect when the coronal split involves the greater
trochanter and exits postero-inferiorly to form a fragment having the posterior part of greater
trochanter and the lesser trochanter, this banana fragment represents a highly unstable type of
fracture pattern which is essential to detect. However, it is very difficult to detect on radiographs
Graph 11: showing the number of instances of detection (out of 300) and its percentage of
The graph given below demonstrates the occurrence of the banana fragment in comparison to
coronal split.
Graph 12: showing the relative instances of detection of the coronal splits and banana fragment.
PATIENT DEMOGRAPHICS:
Gender distribution:
The gender distribution of the patients and their percentages are represented in the table 13 and
graph 13.
Male 34 68%
Female 16 32%
TOTAL 50 100%
Table 13: Showing the gender distribution of the patients in various groups along with their
percentages.
Graph 13: Showing the gender distribution in the study.
Age distribution:
The age distribution of the patients enrolled in the study is represented in table 14 and graph 14.
<30 YEARS 1 2%
TOTAL 50 100%
Table 14: Showing the age distribution of the patients in various groups along with their
percentages.
Graph 14: Showing the age distribution of the patients in various age groups along with their
percentages.
Mode of injury:
The majority of injuries occurring in the elderly group were osteoporotic fractures secondary to
low energy trivial trauma like fall from bed or slip and fall on the floor. Fall from stairs was also
a common mode of injury in elderly patients, in younger individuals the most common mode of
Graph 15: pie chart representing the distribution of various modes of injury in the study shown
below. (RTA- road traffic accidents, bed-fall from bed, floor-slip and fall on floor, stairs-fall
from stairs)
Graph 16: Line graph showing the comparison of instances of detection of all the seven
independent entities on the three radiological modalities (on the next page).
Graph 17: Bar graph showing the comparison of percentage of instance detection of all the seven
With an increasing elderly population, the incidence of fragility fractures around the hip has also
increased with intertrochanteric fractures of femur forming a bulk of these patients. These
patients often have multiple coexistent conditions that require multidisciplinary optimization
followed by prompt fixation and early mobilization as an attempt to decrease the morbidity and
mortality associated with these fractures. Detailed pre-operative planning is essential for the
systems help surgeons in the recognition and prediction of the challenges associated with various
fracture types.
The original 1990 AO/ASIF(7) classification was based primarily upon the integrity of the
posteromedial cortical buttress and the various fracture patterns were classified in accordance of
increasing severity of comminution involving this region. Studies by Van embden et al(8) and
Crijns et al(11) demonstrated moderate agreement on inter- and intra-observer reliability at the
subgroup level which increased substantially when only groups were considered. Pervez et al(41)
in his study in 2002, also demonstrated moderate and fair agreement of inter- and intra-observer
reliability respectively. Cavaignac et al(39) in his study demonstrated a fair agreement which did
not improve substantially even with the use of 2D-CT with multiplanar reconstruction and 3D
reconstruction. While the original AO classification helps in differentiating between stable and
unstable fracture patterns based on posteromedial comminution, it fails to take into account the
other entities contributing to instability including the lateral wall thickness that has emerged as
an important factor in determining the choice of implant for allowing controlled impaction of
fracture fragments.
Some of the shortcomings of the original AO classification were addressed in the new 2018
the choice of implant and classifying fractures according to the 2018 AO/OTA(18) classification
which differentiates between the A1 group and A2 group based upon the competency of the
lateral wall. Thickness of greater than 20.5 mm has been classified as type A1 (competent lateral
wall) and less than 20.5 mm (incompetent lateral wall) as type A2. The A1 group was also
revised to include multiple fracture lines and the 31-A1.3 subgroup was modified to include an
independent lesser trochanteric fragment. However, like the previous AO/ASIF(7) classification,
it utilises only anteroposterior radiographs and has similar problems with detection of coronal
This study evaluated the 2018 AO/OTA(18) classification along these lines also assessing the
effect of application of higher imaging modalities (2D-CT with multiplanar reconstruction and
reliability and also evaluating their impact on the detection of the seven independent entities
described above.
Some variants that were actually not classifiable as per the existing described subgroupss:
descriptive and specific of the fracture patterns as compared to other classifications described
previously, certain patterns of fractures and their variations have not been accounted for in this
classification system. In this study we recognised four such types that have been summarised in
Type 1:
type of two part intertrochantreic fracture which is very difficult to reduce by either closed or
open means due to the interposition of the capsule and the iliopsoas ligament between the
fracture fragments(42). This is a two part fracture with extension of the fracture line lateral to the
pyriform fossa with a long posterior spike and the lesser trochanter being majorly a part of the
proximal fragment leading to its external rotation and valgus displacement due to the attachment
of the external rotators of the hip. Fractures lateral to the pyriform fossa had the attachment of
the external rotators to the proximal fragment and such fractures were termed as ‘extradigital’ by
are reduced by axial traction with adduction and internal rotation, however, this particular type of
intertrochanteric facture is highly unstable and not amenable to reduction by routine manoeuvre
or closed reduction, they require open reduction with retraction of the interposed iliopsoas
tendon and the capsule. Although they are suitable for fixation using the dynamic hip screw
system due to an adequate lateral wall thickness, If classified by the AO classification they
would fall under the A1.2 subgroup and be labelled as ‘simple’ two part fracture which might
mislead the surgeon in failing to recognise the instability and challenges associated with this
complex pattern.
Figure 21: Case of epsilon variant in this study which is potentially irreducible by closed means.
Type 2:
fragment with a competent lateral wall, however, in our study, we came across several fracture
patterns where the lesser trochanter was intact and independent facture fragments were present
along the fracture line over the intertrochanteric crest or a greater trochanter fragment associated
with a competent lateral wall, which created a dilemma in classification for the observer as such
a pattern could not be classified under the A 1.2 subgroup since it includes only two part
fractures and neither under the A 1.3 subgroup which includes an independent lesser trochanter
fragment with competent lateral wall; but fractures with fragments along the superior aspect of
the fracture line with competent lateral wall have not been described in the classification system.
3D reconstruction images of some cases with this pattern are shown in figure 22 .
Since the AO classification is based solely on antero-posterior radiographs and fails to consider
coronal splits, some cases in the study with coexistent posteromedial comminution in the coronal
plane could not be classified under A1 subgroup even though the lateral wall was adequate on
calculation.
Figure 22: Showing 3D reconstruction with segemntation of some cases in this study with an
intact lesser trochanteric fragment but other fragments present along the fracture line over the
Type 3:
In this study we encountered several fractures which had both intertrochanteric and reverse
obliquity elements which are not described in the classification and hence, could not be
conclusively classified into either 31-A2 or 31-A3 subgroups. Figure 23 shows the particular
pattern observed in this study on 3D reconstruction and segmentation with both intertrochanteric
and reverse oblique fracture lines creating a dilemma for the observer whether to classify these
as A2 group or A3 group.
Figure 23: Fracture patterns having both intertrochanteric and reverse oblique fracture lines.
Type 4:
During the course of our study we also came across a few cases where there were multiple
fragments along the intertrochanteric line with an incompetent lateral wall but the lesser
trochanteric fragment was intact, thus, creating a predicament on classifying these fractures as all
the patterns included under the A2 group have an associated lesser trochanteric fragment.
Figure 24: Showing example cases in this study with incompetent lateral wall and intermediate
closest pattern matching them in the classification was marked. The various non-classifiable
patterns and their relative instances observed in this study are summarised in table 15 given
below. Fractures with both reverse oblique and intertrochanteric fracture lines had the highest
Types of non Reason for being non classifiable Instances on Instances Instances on 3D
explained above
fracture line.
Graph 18: Comparing the detection of instances of non-classifiable patterns on the various
imaging modalities.
Graph 19: Showing the different proportions of detection of non classifiable patterns on the three
imaging modalities.
While the AO system provides an anatomical classification which is useful for collecting and
presenting information for research and audit purposes, it’s inherent complexity and
alphanumeric nature lacks the ease of application and communication required in routine
clinical practise, thus falling short of playing a useful role in planning and management of
patients(43). The AO compendium has an anatomical basis for deciding the long bone involved,
the location of the fracture (proximal end, diaphyseal or distal end), the type of involvement
fracture specific topographical characteristics, and the subgroup which is based on the various
fracture morphologies, it also recommends the use of universal modifiers after the classification
for better representation of the patient’s condition Although these factors improve the
incidence of error due to complexity when the fractures were coded on individual basis.
Similarly Martin and Marsh(45) in their study noted that the overall inter- and intra-observer
reliability of the classification drops to unacceptable levels at the group level in some instances
and the subgroup level in almost all fractures. In this research study, we provided the observers
with evaluation sheets depicting the various fracture patterns in order to factor out these
difficulties in classifying the fractures to better gauge the classification method objectively on its
inherent properties.
Figure 25: Showing the complex alphanumeric hierarchy of the AO/OTA classification.
In 2012 Hsu et al(12) conducted a study to determine the relationship between the thickness of
the lateral wall and the incidence of post-operative lateral wall fractures when DHS was used for
fixation in 208 patients with intertrochanteric fractures. They concluded that the risk of lateral
wall fracture increased substantially if DHS was used in patients with lateral wall thickness less
than 20.5mm. Intact lateral wall plays an integral role in preventing the excessive sliding of the
head-neck fragment by acting like a lateral buttress preventing varus collapse and allowing a
controlled impaction by cyclic loading and remodelling after fixation. This concept was utilized
by the new AO classification in differentiating between the A1 and A2 groups, and postulating a
probable choice of implant to be applied in each group. However, this requires a clear distinction
to be made in between the groups as any ambiguity in this differentiation can mislead the
surgeon in choosing the wrong implant for the fracture type leading to postoperative lateral wall
fractures.
Chan et al(29) and Klaber et al(46) in their studies attributed the decrease in reliabilities among
observers, and the disparity of reliabilities between consultants and residents, to the difficulty in
calculation of the lateral wall thickness which plays an important part in deciding groups in the
In this study too, the observers faced multiple challenges while trying to calculate the lateral
wall thickness by the method given by Hsu et al(12) which utilizes the innominate tubercle as the
proximal point for the 3cm vertical drop as shown in the figure 26. Firstly, there is ambiguity on
radiographs thus creating an impediment in choosing the proximal reference point. Furthermore,
since the innominate tubercle is situated on the intertrochanteric crest, which is frequently
Secondly, in the method described by Hsu et al(12), lateral wall thickness is taken from the
midpoint in between the anterior and the posterior cortex, during the course of our study we
encountered various instances where the posterior cortex thickness was reduced substantially due
to a coronal fracture line extending posterior-inferiorly to involve to involve the lesser trochanter
or the posteromedial cortex, often associated with a large banana fragment, however, the
majority of these cases had a thick anterior wall which could be misinterpreted as a fracture
Thirdly, for better delineation and accurate measurement of the lateral wall thickness AO
recommends traction view with leg in neutral rotation, however, this is not always feasible in
acute cases due to the pain and distress faced by the patient along with the lack of facilities in the
emergency radiology suite for obtaining traction views. Similar challenges are faced while trying
to obtain lateral views which may often end up being inadequate due to improper positioning as a
result of pain and muscular spasms, repeated attempts to obtain proper lateral views can
Figure 26: Showing the method described by Hsu et al for calculation of lateral wall thickness.
The evaluation sheet had a field where the observer entered if they faced difficulty while trying
to calculate the lateral wall thickness; these observations are presented in table 16. The observers
also faced difficulty while trying to calculate the lateral wall thickness on 2D-CT with
multiplanar reconstruction when there was associated comminution with the fracture line
extending across it. In all these cases the innominate tubercle was localised by approximation
Table 16: Showing the number of instances (out of 300) when the observers faced difficulty in
These problems were reflected in our study as decreased inter- and intra-observer reliability on
subgroup analysis of the A1.3, A2.2 and A2.3 subgroups. We suggest a three-fold approach to
tackle these problems, these methods to localise the innominate tubercle are enumerated in the
Firstly, use of vastus lateralis ridge instead of the innominate tubercle as the proximal
landmark:
Vastus lateralis ridge is a clear and prominent landmark easily recognizable on radiographs and
however, since these fractures are classified as highly unstable injuries with coexistent lateral
wall fracture, lateral wall thickness is of little significance. We calculated the distance between
the vastus lateralis ridge anad the innominate tubercle in 2D-CT of the 50 subjects enrolled in the
study and found the mean (for both males and females) to be 10.02 mm (see table 17 for the
average distance in between the vastus lateralis ridge and the innominate tubercle for patients
enrolled in the study with varying heights). In the method published by Hsu et al(12) the
reference point is taken 3 cm below the innominate tubercle (figure 27), after due corrections for
changing the proximal landmark we can take the reference point from the vastus lateralis ridge.
Calculation of the distance in between the innominate tubercle and the vastus laterlais ridge:
Step 1: On the axial sections of 2D-CT the mid axial section showing the innominate tubercle
was identified.
Step 2: A plane was created along the X and Y axis enclosing the axial mid section showing the
innominate tubercle.
Figure 28: The innominate tubercle is marked within the plane drawn along the X and Y axis.
Step 3: on the coronal sections the mid coronal section showing the vastus lateralis ridge in full
profile is selected and a line drawn across its most distal point, the distance between this line and
Figure 29: Showing the calcuation of difference in between the innominate tubercle and vastus
lateralis ridge.
Secondly, use of a line intersecting the inferior aspect of the head of femur of the contralateral
In undisplaced or mildly displaced intertrochanteric fractures, a line drawn from the inferior most
part of the head of femur of the contralateral side approximately intersects with the position of
the innominate tubercle on the affected side. We evaluated 50 bilateral hip 2D-CT with
multiplanar reconstruction by identifying the innominate tubercle on the axial sections bilaterally
and creating planes to mark them on the corresponding coronal section as explained in the
method below. After marking the corresponding position of the innominate tubercle on the
coronal section, a line drawn between the two markings intersected the inferior-most point of the
head of femur on both sides within a range of 0-2 mm depending upon the height of the patient
and the femur length as shown in the table 17. Therefore, in undisplcaed or marginally displaced
intertrochanteric fractures the innominate tubercle on the affected side may be localised with
considerable accuracy on plain radiographs as the point of intersection of the line drawn from the
from the inferior-most point of the head of femur of the unaffected side and the intertrochanteric
crest of the involved side provided there is no abduction or adduction at the hip bilaterally which
could then change the relative orientation of the innominate tubercle with respect to the
contralateral side, considering this limitation we recommend that the preference should be given
Calculation of distance in between the line intersecting the inferior most point of the head of
Steps 1 and 2 of recognizing the mid axial section of the innominate tubercle and marking it on
the coronal plane will remain the same as described above. However, this method should only be
used when both the hips are in neutral coronal alignment (no adduction or abduction at bilateral
hip joints), which should be checked before proceeding with this method.
Step 3: selecting the coronal section showing the inferior most part of the head of femur and
making a line intersecting it bilaterally, followed by calculation of the distance in between this
intersecting the inferior most part of the head of femur bilaterally. Also note the orientation of
the pelvis with the tip of both the teardrops at the same level and both the hips in neutral position
Thirdly, use of the inter-teardrop line of pelvis for localization of the innominate tubercle:
This method also helps with the localisation of the innominate tubercle by using the inter-
teardrop line as a landmark, However, similar to the second method the prerequisite for
application of this method is the neutral orientation of the pelvis (the inter-tear drop line should
be horizontal with both the tear drops approximately at similar levels) and bilateral hips also in
neutral coronal plane orientation. In this study we evaluated the 2D-CT with multiplanar
reconstruction of 50 patients and found the mean overall distance to be 8.4 mm (the average
distance for the various height groups of patients is given in table 17).
Calculation of distance in between the line intersecting the inter-teardrop line and the innominate
tubercle:
Steps 1 and 2 of recognizing the mid axial section of the innominate tubercle and marking it on
the coronal plane will remain the same as described above. However, this method should only be
used when both the hips and the pelvis are in neutral coronal alignment (no adduction or
abduction at bilateral hip joints), which should be checked before proceeding with this method.
Step 3: selecting the coronal section showing the teardrop in full profile bilaterally and making
an intertear drop line, followed by calculation of the distance in between this line and the plane
Figure 31: Showing the calcuation of distance in between the innominate tubercle and the inter-
teardrop line. Also note the orientation of the pelvis with the tip of both the teardrops at the same
level and both the hips in neutral position with respect to adduction and abduction.
As already mentioned we recommend the use of ipsilateral vastus lateralis ridge followed by the
other two methods in the order that they have been described.
HEIGHT NUMBER FEMUR Distance from Distance from Distance from
femur)
(Average)
Table 17: Showing the various methods of localisation of the innominate tubercle and the
measurnements with respect to the height and femur length of the patients.
After the publication of the new AO classification in 2018, various studies were performed to
evaluate its inter- and intra-observer reliability. In 2020, Chan et al(29) conducted a multicentre
observational study for assessment of the inter- and intra-observer reliability of the new AO
classification. Radiographs of 150 patients were classified by six orthopaedic surgeons (two
consultants and four residents) on 2 occasions 3 months apart. It demonstrated a mean inter- and
intra-observer reliability of 0.479 and 0.661 at the group level, 0.376 and 0.587 at the subgroup
level respectively. Consultants had better reliabilities than residents. The finding of inter-
observer reliability was in tune with our study which also demonstrated a moderate inter-
observer reliability (0.41) at the group level which decreased to slight (0.27) on subgroup
analysis. However, the intra-observer reliabilities were not concurrent to those observed in our
study, 0.46 for the groups which decreased to 0.32 on subgroup analysis.
Klaber et al(46) in 2020 conducted a study for comparison of inter- and intra-observer reliability
of the 2018 AO/OTA and original AO/ASIF classification. Radiographs of 67 patients were
observer agreement as inferred from this study was 0.128 (slight) for the original AO/ASIF
classification and 0.250 (fair) for the new AO/OTA system. Intra-observer agreement for the
original AO/ASIF and new AO/OTA classification system was 0.350 (fair) and 0.295 (fair)
respectively. However, this study showed a greater reliability among residents as compared to
specialists. These findings did not co-relate with the findings observed in our study where the
inter-observer and intra-observer reliability was moderate (0.41 and 0.46) respectively.
In the study conducted by Zarie et al(30) in 2020, 96 plain radiographs were evaluated by four
observers on two occasions at an interval of one month. Their results showed substantial mean
inter- and intra-observer reliability 0.61 and 0.56 respectively for the AO groups while the
agreements were fair 0.321 and 0.314 for the AO subgroups. The findings of mean inter- and
intra- observer reliability for the AO groups were in contrast to those observed in this study,
which were moderate 0.41 and 0.46 respectively. However, the findings of inter- and intra-
observer reliability regarding AO subgroups were in conjunction to the findings of our study
which also demonstrated a decrease in the inter- and intra-observer reliabilities in AO sub
Bo Yin et al(47) in 2021 conducted a study comparing the inter- and intra-observer reliability of
the Evans(3), Jensen(6), 2018 AO/OTA(18) and Tang(48) classification systems on radiographs
and 2D-CT scans of 258 patients. The evaluation was done by six orthopaedic surgeons on two
occasions one month apart. It showed moderate agreement of inter- and intra observer reliability
0.46
and 0.45 respectively on radiographs, 0.44 and 0.41 on 2D-CT with multiplanar reconstruction
respectively. This was in contrast to this study which had an inter- and intra-observer of 0.39 and
0.37 for radiographs and 2D-CT with multiplanar reconstruction respectively, however, the intra-
observer reliabilities for both the studies were similar with the intra-observer reliability in this
study being 0.45 and 0.43 for radiographs and 2D-CT with multiplanar reconstruction
respectively . According to their study only the Tang classification showed substantial agreement
in inter- and intra-observer reliability on both radiographs and 2D-CT. However, they did not
Another study by Yildrim et al(49) in 2022 conducted a retrospective analysis of the radiographs
of 60 patients by five residents and five orthopaedic surgeons, who classified them according to
classification systems. It demonstrated substantial inter- and intra-observer agreement (0.669 and
0.744 respectively) for AO grouping between both the residents and the surgeons, however, the
agreement decreased to moderate (0.444 and 0.516 respectively) when AO subgroups were
considered the classification having the most inter- and intra-observer agreement was the 2018
AO/OTA(18) when only the main groups were considered, however, the findings were in
contrast to the findings of our study which only demonstrated a moderate inter- and intra-
observer reliability (0.41 and 0.46 respectively) when considering AO main grouping which
decreased to fair for AO sub grouping (0.27 and 0.32 respectively) as shown in table 12.
The decrease in inter-observer reliability noted in the subgroup analysis reflects the difficulty
faced by observers in calculating the lateral wall thickness, which is the key determinant in
classifying fractures between the subgroup A1 and A2, and in classifying the patterns that have
not been described in the AO/OTA classification (the four types have been described above) and
were therefore, marked as the subgroup described by the AO classification which most closely
Table 18 provides a brief summary and reiterates the salient features of the inter- and intra-
As shown on the graphs 16 and 17, the instances of detection of flexion or extension of proximal
fragment, anterior or posterior wall comminution, coronal plane fracture lines, lateral wall
fractures and the banana fragment increased substantially on application of higher modalities.
Since the AO classification only utilizes plain antero-posterior radiographs it fails to recognise
lateral wall fractures and instabilities created by large posteromedial coronal splits involving the
posterior part of the lateral wall rendering it incompetent. The integrity of the posteromedial
hinge is essential to prevent varus displacements and retroversion of the proximal fragments.
Plain radiographs not only underestimate the complexity of the fracture and the degree of
essential for the distinction of superimposed intermediate fragments along the primary fracture
line for differentiation in between the A2.2 and A2.3 subgroups for better assessment of stability
and prognostication. Adequate lateral views are essential for the detection of flexion or extension
of the head-neck fragment with respect to the shaft; however, it is a herculean task in an acute
traumatic setting, often requiring multiple radiographic exposures, due to the pain and distress
caused to the patient while trying to achieve an appropriate position for these radiographs.
Higher imaging modalities offer an easier, faster and a more patient friendly alternative for
fragments:
As shown in graphs 16 and 17, significant increase was observed in detection of instances of
coronal plane fracture line, lateral wall fracture and especially the banana fragment in between
the two modalities. The presence or absence of banana fragments could also not be commented
upon conclusively on plain radiographs or 2D-CT with multiplanar reconstruction, while it was
2D-CT with multiplanar reconstruction was comparatively better in detection of location and
segmentation, since very small fragments could not be segmented leading to an underestimation
of severity of comminution on reconstruction; small fragments either in contact with the parent
fragment or in very close proximity to it, could not be differentially segmented and appeared in a
classification in some cases. Similarly, in cases where the fracture lines are closely apposed to
Axial and coronal sections of 2D-CT were most informative regarding the location and severity
of comminution, coronal splits and lateral wall fractures. Flexion-extension of the head-neck
fragment was better detected on the sagittal sections by 2D-CT and on lateral and medial
reconstruction of fracture fragments is easier, faster, with better and consistent detection of the
banana fragments that are mostly undetected on plain radiographs and 2D-CT as shown in figure.
This study demonstrated a moderate and fair overall interobserver reliability for the new
AO/OTA group and sub-group classification of intertrochanteric hip fractures, with moderate to
Fracture mapping:
Fracture mapping allows for detailed visualisation and evaluation of the morphology of fracture
lines, it provides a ‘Bird’s-eye view’ for observing the distribution of the fracture lines. On
plotting the fracture lines of the cases in this study, we were able to appreciate the diverse
morphology of the fracture lines with different variations that were far wider than the existing
AO classification could include. Plotting fracture maps with four orthotropic planes provides
more information about the characteristics of the fracture and its morphology which can be used
in future to describe other fracture patterns and understand fracture mechanisms. The method for
plotting the fracture lines is explained in detail in the methodology section. The figure shows the
radiograph based classification systems. It also takes into account the thickness of lateral wall
which is a key determinant for the choice of implant, risk of post-operative lateral wall fracture
and screw cut out. It utilizes the lateral wall thickness for differentiation in between the A1 and
A2 groups, and for postulating a probable choice of implant in these groups. However, this
requires a clear distinction to be made in between the groups as any ambiguity in this
differentiation
Various studies evaluating the classification demonstrated only moderate agreement of inter- and
intra-observer reliability which decreased substantially when subgroup analysis was performed.
Similar finding were echoed in our study as a reflection of the multiple challenges faced by the
observers for classification, namely, difficulty in recognition of the location of the innominate
wall thickness as per the method described by Hsu et al(12) and recommended by the
classification; four types of fracture patterns were encountered during the course of our study
which did not match the fracture morphologies described in the classification. This ambiguity in
the method of calculation of lateral wall thickness has serious clinical implications as it can
mislead the surgeon in choosing the wrong implant for the fracture type leading to postoperative
lateral wall fracture or screw cut out. To tackle this problem, we proposed the use of vastus
lateralis ridge which is a clearly visible landmark rarely involved in inter-trochanteric fractures,
for calculation of lateral wall thickness. We also proposed the use of a line intersecting the
inferior-most aspect of the head of femur with the intertrochanteric crest bilaterally and distance
from the inter teardrop line as accessory beacons for helping the localization of the innominate
tubercle.
Plain radiographs not only underestimate the complexity of the fracture and the degree of
categorization as anterior wall or posterior wall comminution. Adequate lateral views are
essential to comment upon the flexion or extension of the head-neck fragment, which are
difficult to obtain in acute trauma setting and repeated attempts can lead to significant patient
distress and pain. Even when adequate lateral views are obtained, it is difficult to detect coronal
This study clearly showed the advantages of application of higher imaging modalities like 2D-
CT with multiplanar reconstruction and 3D reconstruction with segmentation for better detection
coronal plane fracture line, lateral wall fracture and the banana fragment. While 2D-CT with
MPR was able to better detect the location and severity of comminution, 3D reconstruction and
segmentation was most useful for detection of Banana fragments which were missed on both
Fracture mapping allows for better evaluation of the morphology and distribution of fracture
lines enabling a better understanding of fracture mechanisms that can be used to describe other
1. Sheehan SE, Shyu JY, Weaver MJ, Sodickson AD, Khurana B. Proximal Femoral Fractures:
What the Orthopedic Surgeon Wants to Know. Radiogr Rev Publ Radiol Soc N Am Inc.
2015;35(5):1563–84.
2. Boyd HB, Griffin LL. Classification and treatment of trochanteric fractures. Arch Surg Chic
3. Evans EM. The treatment of trochanteric fractures of the femur. J Bone Joint Surg Br.
1949 ;31B(2):190–203.
4. Baer M, Neuhaus V, Pape HC, Ciritsis B. Influence of mobilization and weight bearing on
5. Bojan AJ, Beimel C, Taglang G, Collin D, Ekholm C, Jönsson A. Critical factors in cut-out
7. Marsh JL, Slongo TF, Agel J, Broderick JS, Creevey W, DeCoster TA, et al. Fracture and
133.
8. van Embden D, Rhemrev SJ, Meylaerts S a. G, Roukema GR. The comparison of two
classifications for trochanteric femur fractures: the AO/ASIF classification and the Jensen
comminution in the elderly: a series of 110 cases. Eur J Orthop Surg Traumatol Orthop
Traumatol. 2015;25(7):1165–70.
10. Cho YC, Lee PY, Lee CH, Chen CH, Lin YM. Three-dimensional CT Improves the
2018;10(3):212–7.
11. Crijns TJ, Janssen SJ, Davis JT, Ring D, Sanchez HB; Science of Variation Group.
Reliability of the classification of proximal femur fractures: Does clinical experience matter?
12. Hsu CE, Shih CM, Wang CC, Huang KC. Lateral femoral wall thickness. A reliable
2013;95-B(8):1134–8.
13. Palm H, Jacobsen S, Sonne-Holm S, Gebuhr P, Hip Fracture Study Group. Integrity of the
14. Parker MJ. Trochanteric hip fractures. Fixation failure commoner with femoral
16. Cho JW, Kent WT, Yoon YC, Kim Y, Kim H, Jha A, et al. Fracture morphology of
17. Hecht G, Saiz AM, Shelton TJ, Haffner MR, Delman C, Shieh A, et al. CT scans better
assess lateral wall morphology of “stable appearing” intertrochanteric (IT) femur fractures
and predict early failure of sliding hip screw (SHS) fixation. OTA Int Open Access J Orthop
Trauma. 2021;4(3):e140.
18. Meinberg EG, Agel J, Roberts CS, Karam MD, Kellam JF. Fracture and Dislocation
19. Bendo JA, Weiner LS, Strauss E, Yang E. Collapse of intertrochanteric hip fractures fixed
20. Poole KES, Treece GM, Ridgway GR, Mayhew PM, Borggrefe J, Gee AH. Targeted
21. Rha JD, Kim YH, Yoon SI, Park TS, Lee MH. Factors affecting sliding of the lag screw in
2012;132(6):839–46.
23. Díaz VJ, Cañizares ACP, Martín IA, Peinado MA, Doussoux PC. Predictive variables of
open reduction in intertrochanteric fracture nailing: a report of 210 cases. Injury. 2016;47
Suppl 3:S51–5.
25. Futamura K, Baba T, Homma Y, Mogami A, Kanda A, Obayashi O, et al. New classification
focusing on the relationship between the attachment of the iliofemoral ligament and the
26. Shoda E, Kitada S, Sasaki Y, Hirase H, Niikura T, Lee SY, et al. Proposal of new
and relationship to usual plain X-ray classification. J Orthop Surg Hong Kong.
2017;25(1):2309499017692700.
27. Moore JR. Fractures of the upper end of the femur including fracture dislocations at the hip
29. Chan G, Hughes K, Barakat A, Edres K, da Assuncao R, Page P, et al. Inter- and intra-
observer reliability of the new AO/OTA classification of proximal femur fractures. Injury.
2021;52(6):1434–7.
30. Zarie M, Mohamoud MF, Farhoud AR, Bagheri N, Khan FMY, Heshmatifar M, et al.
Intertrochanteric Fractures and the Device Choice (DHS, PFNA, and DCS) of Fixations.
31. Gao Z, Lv Y, Zhou F, Ji H, Tian Y, Zhang Z, et al. Risk factors for implant failure after
fixation of proximal femoral fractures with fracture of the lateral femoral wall. Injury.
2018;49(2):315–22.
32. Gotfried Y. Integrity of the lateral femoral wall in intertrochanteric hip fractures: an
reply 2553.
33. Tawari AA, Kempegowda H, Suk M, Horwitz DS. What makes an intertrochanteric fracture
unstable in 2015? Does the lateral wall play a role in the decision matrix? J Orthop Trauma.
34. Zhang Q, Chen W, Liu H jun, Li Z yong, Song Z hui, Pan J she, et al. The role of the calcar
35. Hammer A. The calcar femorale: A new perspective. J Orthop Surg Hong Kong.
2019;27(2):2309499019848778.
36. Haidukewych GJ. Intertrochanteric fractures: ten tips to improve results. Instr Course Lect.
2010;59:503–9.
37. Kijima H, Yamada S, Konishi N, Kubota H, Tazawa H, Tani T, et al. The reliability of
39. Cavaignac E, Lecoq M, Ponsot A, Moine A, Bonnevialle N, Mansat P, et al. CT scan does
40. Landis JR, Koch GG. The measurement of observer agreement for categorical data.
Biometrics. 1977;33(1):159–74.
41. Pervez H, Parker MJ, Vowler S. Prediction of fixation failure after sliding hip screw fixation.
Injury. 2004;35(10):994–8.
“epsilon sign” and its significance in reduction in highly unstable variant of intertrochanteric
43. Newey ML, Ricketts D, Roberts L. The AO classification of long bone fractures: an early
44. Johnstone DJ, Radford WJ, Parnell EJ. Interobserver variation using the AO/ASIF
AO/ASIF fracture classification for the distal tibia. J Orthop Trauma. 1997;11(7):477–83.
46. Klaber I, Besa P, Sandoval F, Lobos D, Zamora T, Schweitzer D, et al. The new AO
classification system for intertrochanteric fractures allows better agreement than the original
2021;52(1):102–5.
47. Yin B, He Y, Wang D, Zhou J. Classification of femur trochanteric fracture: Evaluating the
48. Tang ZH, Yeoh CSN, Tan GMJ. Radiographic study of the proximal femur morphology of
elderly patients with femoral neck fractures: is there a difference among ethnic groups?
49. Yıldırım C, Muratoğlu OG, Turan K, Ergün T, Mısır A, Aydın M. The intra- and
50. Tronzo RG. Symposium on fractures of the hip. Special considerations in management.
The method of measurement of lateral wall thickness as mentioned in the 2018 AO/OTA 18
classification.
The differentiation between groups is defined by the lateral wall height (d) of the greater
trochanter. Lateral wall height or thickness is defined as the distance in millimeters (mm) from a
reference point 3 cm below the innominate tubercle of the greater trochanter angled 135° upward
to the fracture line on the anteroposterior x-ray. The thickness (d) must be less than 20.5 mm for
the fracture to be considered an A2 fracture. It is recommended that the measurement for the
lateral wall be taken using the traction view with the leg in neutral rotation.
ANNEXURE-3
TRAINING SESSION.
fracture of either the greater trochanter(1) or lesser trochanter(2) As shown in figure below.
fragment with a competent lateral wall of thickness greater than 20.5 mm.
The lateral wall thickness is defined as the distance in millimetres from the midpoint of both the
fractured cortices along a line that is angled 135 degrees upwards to the fracture line from a
reference point that lies 3 cm below the inominate tubercle of the greater trochanter. B – is the
midpoint of both the cortices. A – point on the lateral wall. AB – lateral wall thickness
31 A2 : This subgroup consists of multifragmentary intertrochanteric fractures with
intermediate fragments at the trochanteric region along with an incompetent lateral wall of
Reverse oblique fractures are highly unstable fractures due to the propensity of
medialisation of the distal fragment as a result of the strong pull of medial fiber bundles of
These fractures are associated with difficulty in achieveing and maintaining reduction and
LATERAL WALL FRACTURE: Intact lateral femoral wall provides a lateral buttress for the
proximal fragment and allows for controlled fracture impaction which prevents varus
malposition and collapse and provides a conducive environment for fracture healing by allowing
cyclic loading and remodelling after fixation with a Dynamic Hip Screw.
Recognition of lateral wall fracture is important as it is associated with high rates of implant
failure and re-operation following dynamic hip screw fixation due to uncontrolled medialisation
Even when other intramedullary implants are used the blade or screw used for fixation may pass
through the fracture line leading to the separation of the fractured fragments.
lead to collapse with axial loading leading to coxa vara, rotational instability and weakened
abductors.
CORONAL SPLIT: oronal split extending into the greater trochanter can lead to the loss of the
superolateral support and result in failure of fixation especially when a sliding hip screw is being
used as the coronal split is difficult to detect on radiographs and the instability due to this
fracture line over the anterior aspect apart from a primary fracture line and the anterior wall is
defined as comminuted if there is a third fracture fragment along the main fragment line.An
intact anteromedial fracture plane is essential in obtaining stable fixation for an unstable
intertrochanteric fracture as it limits the sliding of the head neck fragment.Anterior comminution
proximal fragment indicates a difficult close reduction and is a predictor for the possible need of
open reduction
inherently unstable after 3D-CT reconstruction due to the involvement of the trochanteric region
trochanter as a single fragment are considered unstable. This pattern of fracture was labelled as
ANNEXURE-4
cm below the innominate tubercle of the greater trochanter angled 135° upward to the fracture line on
the anteroposterior x-ray. The thickness (d) must be less than 20.5 mm for the fracture to be considered
an A2 fracture. It is recommended that the measurement for the lateral wall be taken using the traction