„ REVIEW ARTICLE
The measurement of patellar height
C. L. Phillips,
D. A. T. Silver,
P. J. Schranz,
V. Mandalia
From Department of
Clinical Radiology
and Department of
Trauma and
Othopaedics, Royal
Devon and Exeter
Hospital, United
Kingdom
„ C. L. Phillips, MRCS, FRCR,
Specialist Registrar in
Radiology
„ D. A. T. Silver, FRCP, FRCR,
Consultant Radiologist
Department of Clinical
Radiology
„ P. J. Schranz, FRCS(Trauma &
Orth), Consultant Orthopaedic
Surgeon
„ V. Mandalia, MS(Orth),
DNB(Orth),
FRCS(Trauma&Orth),
Consultant Orthopaedic
Surgeon
Exeter Knee Reconstruction
Unit, Department of Trauma
and Orthopaedics
Royal Devon & Exeter Hospital,
Barrack Road, Exeter EX2 5DW,
UK.
Correspondence should be sent
to Mr V. Mandalia; e-mail:
vipul.mandalia@rdeft.nhs.uk
©2010 British Editorial Society
of Bone and Joint Surgery
doi:10.1302/0301-620X.92B8.
23794 $2.00
J Bone Joint Surg [Br]
2010;92-B:1045-53.
VOL. 92-B, No. 8, AUGUST 2010
A REVIEW OF THE METHODS OF IMAGING
Many radiographic techniques have been described for measuring patellar height. They can
be divided into two groups: those that relate the position of the patella to the femur (direct)
and those that relate it to the tibia (indirect). This article looks at the methods that have
been described, the logic behind their conception and the critical analyses that have been
performed to test them.
In 1938, Blumensaat1 described the first practi-
cal radiographic technique for measuring patel-
lar height. Since then, many further attempts
have been made to establish a simple, reliable
and reproducible way of assessing the patella on
standard imaging.
Knowledge of the anatomy and biomechanics
of the patella is fundamental to understanding
the different pathologies of the anterior knee.2-8
Abnormal patellar height is seen in many condi-
tions that affect the patellofemoral joint. Patella
alta, an abnormally high patella, is associated
with anterior knee pain, patellar instability and
Osgood-Schlatter’s disease, whereas patella baja
or infera, an abnormally low patella, may be
observed with anterior knee pain and limitation
of knee flexion, mainly as a complication of sur-
gery or trauma.
Since it was first described, the techniques
devised to measure patellar height and the stud-
ies that test their validity have increased in num-
ber and complexity. Despite progression from a
simple radiographic measurement to the most
modern biomechanical and imaging technolo-
gies, the definition of patellar height and its
causal relationship to the conditions commonly
associated with anterior knee pain remains con-
troversial. We have yet to devise a truly fool-
proof method that has all the necessary criteria
for a universally accepted system providing
accurate structural information for use in a clin-
ical environment.
Methods of measurement of patellar height.
Many methods have been described with only
two specifically relating to children9,10 and only
one applied to a specific racial group.11 These
are summarised in Tables I and II and can be
divided into two groups, those that relate the
position of the patella to the femur (direct) and
those that relate it to the tibia (indirect). The
terms ‘direct’ and ‘indirect’ have been applied
more latterly, following the realisation that the
quantitative relationship between the patella
and the femur is most important, as the patello-
femoral biomechanics hold the key to explain-
ing anterior knee pathology.
Indirect methods. The most widely accepted
radiographic techniques used to measure patel-
lar height are indirect, perhaps influenced by the
pioneering method of Insall and Salvati,12 pub-
lished in 1971 with the development of a simple
ratio between the length of the patellar tendon
and that of the patella. It proved easy to mea-
sure on lateral radiographs, not requiring a
fixed-flexion angle, and the mean normal ratio
of 1.0 (0.8 to 1.2) is easy to remember. It
remains the most popular method, although the
bony landmarks are not always easy to identify,
the size of the patella can vary and pathological
bony overgrowth can distort the native anat-
omy.13-15 The method also assumes that the tib-
ial tubercle is at a constant distance below the
tibial plateau. These drawbacks can lead to
both intra- and interobserver error.16 However,
modern digital radiographs have such good
soft-tissue definition that the patellar tendon
itself is usually visible, making accurate identifi-
cation of the tibial tubercle obsolete. Further-
more, a cadaveric study by Schlenzka and
Schwesinger,17 demonstrated that absolute ana-
tomical measurements and the resultant Insall-
Salvati ratios correlated well with ratios mea-
sured on lateral radiographs.
Using a more modern approach, Miller,
Staron and Feldman18 showed that Insall-
Salvati ratios can also beapplied to MR images
but require a slight change in the normal range
of values in order to be valid. In order to
1045
1046 C. L. PHILLIPS, D. A. T. SILVER, P. J. SCHRANZ, V. MANDALIA

Table I. Summary of methods of indirect patellar height measurement in chronological order

Indirect method of
Authors Study population* Imaging method† measurement‡ Interpretation§
Insall and 114: Lateral XR Length of patellar tendon (LT); LT:LP ratios:
Salvati12 ‘Normal’ subjects: 20° to 70° flexion Length of patella (LP) PA > 1.2;
Meniscectomies Ratio = LT:LP N: 0.8 to 1.2;
No demographics given LP PI: < 0.8

LT

Blackburne and 269: Lateral XR Horizontal line projected A:B ratios:

Peel21 171 ‘Normal’ subjects: ≥ 30° flexion to anteriorly from the tibial plateau. PA: > 1.0;
Meniscal pathology or tension the patellar Height from line to inferior edge N: 0.8 to 1.0;
contralateral knees tendon of PAS (A); B PI: < 0.8
(121M; 50F); Length of PAS (B)
58 Subluxers (25M; 33F);
40 CMP (10M; 20F) A

Caton et al22 128: Lateral XR Distance between the inferior AT: AP ratios:
Clinical patella infera 10° to 80° edge of PAS to the anterosuperior PA: > 1.3;
(70M; 58F; mean age 36 flexion angle of the tibial plateau (AT); AP N: 0.6 to 1.3;
years); Length of PAS (AP) PI: < 0.6
141: Ratio = AT:AP
Normal subjects -
asymptomatic (80M;
AT
61F; no age given)

deCarvalho et al25 150: Lateral XR Shortest distance between the T:P ratio:
Normal subjects - Approximately 30° inferior edge of PAS and the P PA: > 1.11
asymptomatic flexion anterior tibial plateau (T)
(20 to 60 yrs; no gender Length of PAS (P)
given) Ratio = T:P

Micheli et al9 Paediatrics AP XR Distance from superior pole of A-B distance:

19: 0° flexion patella to tibial plateau (A); PA: > 0 mm
Physeal plate fracture; Serial XRs at 6 month Length of patella on AP film (B)
contralateral knee used to Distance = A-B
(12M; 7F; no ages given) 1 yr intervals over 2 to
10 yrs B A

THE JOURNAL OF BONE AND JOINT SURGERY

 THE MEASUREMENT OF PATELLAR HEIGHT 1047

Table I. Summary of methods of indirect patellar height measurement in chronological order

Indirect method of
Authors Study population* Imaging method† measurement‡ Interpretation§
26
Egund et al 99: Lateral XR Distance from midpoint PAS (M) No range of values
47 Normal (23M; 24F; 30° to 40° flexion to tibial plateau, perpendicular to given for PA,
mean age 34 yrs); Full weight-bearing mechanical axis of tibia; N or PI but:
left knee; with the tibia 15° to Subject height (H); M M > F for all absolute
36 Extraskeletal the vertical Tibial length (TL) measurements:
malignancy (27M; Ratio = M:TL and M:H M = F for all ratios:
9F; mean age 70 yrs); L = R;
both knees: Young = Old
16 Knee pathology
(no demographics
given); both knees
TL

Koshino and Paediatric Lateral XRs at Lines drawn through distal 0° to 30°:
Sugimoto10 36 (59 knees): 0°, 30°, 60°, 90°, 120° femoral and proximal tibial Ratio decreased from
Normal subjects flexion physes; 1.31 to 0.9; 30° to 90°.
(15M; 21F; Midpoint of longest diagonal line Ratio static between
13 to 18 yrs) of patella to midpoint of tibia 0.99 to 1.20
physis 3.4% of subjects
(P-T) and line connecting P-T FT had PA with own
midpoints method;
of physes (FT) 45.7% of subjects had
Ratio = P-T:FT at all angles PA with Blumensaat
Compared with IS and method;
Blumensaat at 60° 66.7% had PA with IS
method

Grelsamer and 300: Lateral XR The distance between the inferior A:B ratio:
Meadows19 100 ‘Normal’ subjects: 20° to 70° flexion PAS and the patellar tendon PA: > 2.0
(modified Insall- no patellar pathology; insertion (A);
Salvati method) 200 subjects with Length of the PAS (B).
patellar pathology Ratio = A:B B
No demographics given
A

Leung et al11 Ethnic (S. Chinese) Lateral XR Patellar tendon length (A1); PAI ratios:
290: 30° to 70° flexion Patellar length (A2); Length of PA: > 3.4;
173 ‘normal’ - included PAS (B) A2 N: 2.7 to 3.3;
simple meniscal tears; Patellar alta index = Patellar alta PI < 2.7 in Chinese
54 anterior knee pain; Index = B population
52 dislocators; (A1 + A2)/B
11 OS (15 to 50 yrs; Compared with IS, MIS, BP and
no gender given) DeC A1

* CMP, chondromalacia patellae; OS, Osgood-Schlatter

† XR, radiograph; AP, anteroposterior
‡ PAS, patella articular surface; IS, Insall-Salvati; MIS, modified Insall-Salvati; BP, Blackburne-Peel; DeC, deCarvalho
§ PI, patella infera

account for the potential errors arising from morphological
variations in the patella, Grelsamer and Meadows19 devel-
oped the modified Insall-Salvati ratio, for which the length of
the articular surface of the patella rather than that of the
patella itself is used. However, the inferior margin of the
articular surface is not easily identifiable and the inter-
observer error of this method is higher than for the original
ratio.15,16,20
Blackburne and Peel21 devised a technique that again used
the length of the patellar articular surface for their ratio
denominator, but exchanged the tibial tubercle for the tibial
plateau as a landmark in order to avoid misidentification.

VOL. 92-B, No. 8, AUGUST 2010

1048 C. L. PHILLIPS, D. A. T. SILVER, P. J. SCHRANZ, V. MANDALIA
They did not describe how they arrived at the ‘tibial plateau
line’ used. Is it perpendicular to the long axis of the tibia or,
if simply estimated to run parallel to the plateau, which tibial
condyle is used as a reference, bearing in mind that the two
are significantly different in their morphology? It can only be
assumed that the medial plateau is used, as the illustration
implies this, together with the fact that it is concave and thus
has better defined anterior and posterior margins. Seil et al15
and Berg et al16 both found this technique to be the most
reproducible and accurate.
Caton et al22 also modified the numerator by measuring
the distance between the inferior margin of the articular sur-
face of the patella and the anterosuperior angle of the tibial
plateau. However, the problems associated with identifying
the articular margin of the patella would apply similarly to
the tibia.23 Despite this, Berg et al16 found the Caton-
Deschamp ratio to be very reproducible, closely behind that
of Blackburne and Peel, when looking at the inter-observer
error. This may in part be explained by the fact that the nor-
mal range of values is wide (0.6 to 1.3). They did, however,
recognise that it was difficult to use in osteoarthritic knees.16
These four methods are the most popular, owing partly to
their relative simplicity. The lateral radiograph does not need
to be taken with the knee at a fixed angle of flexion as long
as the patellar tendon is under tension at 30° or more.24
These four are the ones most scrutinised in intra- and inter-
observer variability studies that followed and utilised most
often when comparing anterior knee pathologies. Further,
less widely accepted and therefore relatively unknown, novel
techniques were published in fairly quick succession during
the 1980s and 1990s.
De Carvalho et al25 developed a method in 1985 which is
almost identical to that of Caton and Deschamps, but with a
minor alteration in the tibial landmark. They measured the
shortest distance between the inferior edge of the articular
surface of the patellar and the anterior tibial plateau rather
than its anterosuperior edge. Interestingly, the authors not
only misquoted the methodology of Caton and Deschamps
but also stated that, ‘this assessment never gained wide
acceptance’, despite the close similarity to their own method.
Egund, Lundin and Wallengren26 published a method that
is very similar to the Blackburne-Peel in that the patellar
height is measured perpendicular to the tibial plateau. How-
ever, they defined the ‘condylar plane’ as being perpendicular
to the true mechanical axis of the tibial shaft in order to elim-
inate variations in the inclination of the plateau. The lateral
radiograph must be taken with the subject fully weight-
bearing with the knee flexed in order to tension the patellar
tendon and the tibia at 15° to the vertical. Although the
authors recognised the problems associated with establishing
a reliable way to measure patellar height, the solutions they
introduced into their method have resulted in a complicated
technique, making it impractical to use routinely; despite
their claims. No range of values is given for the ratios calcu-
lated in the paper, which perhaps reflects the complexity of
the method.
Direct methods. Relatively few publications describe and
analyse methods of direct measurement of patellar height.
Even the four most popular methods described above are
indirect, once again raising the point about recognising the
importance of the patellofemoral relationship in the aetiol-
ogy of anterior knee pathology. Although Blumensaat is con-
sidered the pioneer of radiographic measurement of patellar
height, an earlier method was described in 1930 by Boon-
Itt.23,27 This assessed both patellofemoral and patellotibial
relationships, but it required complicated geometrical calcu-
lations for each angle of flexion of the knee and proved too
complex for routine use.
Blumensaat1,28 perhaps had early insight into the impor-
tance of the patellofemoral relationship, instead using the
roof of the intercondylar notch as a reference line (the ‘Blu-
mensaat line’) on the lateral radiograph of a knee flexed to
30°. The perpendicular height of the inferior pole of the
patella above this line was measured, with normal being
defined as a distance of zero. Any value greater than this was
classified as patella alta. In 1955 Thestrup-Anderson29
applied this to a group of 286 subjects and found that 207
would be classified as having patella alta. He consequently
moved the normal limit to 5 mm above the Blumensaat line.
In 1970, Brattstrom28 pointed out the disadvantages of the
Blumensaat method, indicating that this distance varies with
knee flexion and also that the Blumensaat line does not form
a fixed angle with the femoral axis, with variations up to 30°.
In an attempt to rectify the latter problem, Seyahi et al30
developed the intercondylar shelf angle (ISA)-corrected Blu-
mensaat method, using a fixed ISA value of 147°, which was
the mean angle between the femoral shaft axis and the Blu-
mensaat line on 105 knees. The patellar height was deter-
mined using this fixed angle and a comparison was made
with the original Blumensaat technique, but again moving
the normal limit up to 10 mm above the line, and also with
the Insall-Salvati, modified Insall-Salvati and Blackburne-
Peel ratios. They showed that Blumensaat and the ISA-cor-
rected methods showed similar but poor correlation with the
three other methods, summarising ‘that the intercondylar
notch cannot be a beneficial landmark alone’. This statement
is likely to have prompted the idea of indirect methods that
followed.
A further variation was developed by Hepp23 in 1984,
whereby the perpendicular height of the superior pole of the
patella above the Blumensaat line is measured, rather than
the inferior pole. They also suggested using the ‘patellar
height angle’, which is the angle subtended by the Blumen-
saat line and a line extending from the superior pole of the
patella to the intersection with the posterior margin of the
femoral condyles. Considering the variation in the position
of the Blumensaat line, these methods do not appear to over-
come this problem.
The next attempt at a direct method came from Labelle et
al31 and Laurin.32 However, this appears rather too simplis-
tic and unusually requires a lateral radiograph of the knee
flexed to 90°. If the superior pole of the patella lies above the
THE JOURNAL OF BONE AND JOINT SURGERY
 THE MEASUREMENT OF PATELLAR HEIGHT 1049

Table II. Summary of methods of direct patellar height measurement in chronological order

Direct method of
Author/s Study population* Imaging method† measurement‡ Interpretation§
Blumensaat1 No details available Lateral XR Perpendicular distance Distance:
30° flexion between the inferior pole of PA > 0 mm
patella and the ‘Blumensaat’
line projected anteriorly
through the roof of the
femoral trochlea

Bernageau 44: Lateral XR Distance between inferior Distances R to T:

et al37 Normal subjects: Full extension edge of PAS (point R) and R PA: > +6 mm (i.e. R higher
No demographics superior edge of femoral than T);
given trochlea (point T) PI: < -6 mm (i.e. R lower
Distance = R to T T than T)

Labella et al31 No details Lateral XR Height of the superior pole of Distance:

available 90° flexion patella above the tangent PA: > 0 mm
of the anterior cortical line of
the femur

Norman et al33 91: Lateral XR Perpendicular distance from VIP ratio:

Meniscal pathology 0° flexion the distal edge of the PAS to VP 0.21 ± 0.02
(57 M; 34 F; 10° to 15° external the femoral condylar
mean age 33 yrs) rotation plane (line tangential to
Quads contracted femoral condyles,
perpendicular to anterior
femoral cortex) (VP);
Body height (H);
Ratio = Vertical index of the
patella = VIP = VP:H

Hepp23 560 knees Lateral XRs at 5° 2 methods: Linear decrease in both

No clinical or incremental flexion 1. Perpendicular distance
demographic from between superior edge of
details given 25° to 60° flexion PAS and the Blumensaat line;
2. Angle subtended by the
Blumensaat line and the
superior edge of the PAS
to posterior femoral condyle
margin
distance (58 to 37 mm) and
angle (53° to 32°) with
2 increasing knee flexion
No range of values given
1 for PA, N and PI

VOL. 92-B, No. 8, AUGUST 2010

1050 C. L. PHILLIPS, D. A. T. SILVER, P. J. SCHRANZ, V. MANDALIA

Table II. Summary of methods of direct patellar height measurement in chronological order

Direct method of
Author/s Study population* Imaging method† measurement‡ Interpretation§
14
Burgess 100: Lateral XR Distance from midpoint of PAS A:B ratio:
‘Normal’ subjects: No specific flexion to tibial plateau (A); N: 0.56 to 0.76 (mean 0.66)
Medial angle - enough to AP width of femoral condyle
meniscectomies; tension the patellar (B) (perpendicular to femoral B
No demographics tendon axis)
given Ratio = A:B
Compared with IS and BP A

Miller et al18 44 (46 knees): Lateral XR Length of PAS (AS); AS:PH ratios:
Meniscal/ACL 30° to 60° flexion: Length of PAS at the level of 2.1 ± 0.8 (M)
injury MRI anterior aspect of the distal AS 2.9 ± 2.4 (F)
(18M, mean age sagittal femoral physeal scar (PH) IS (XT) = 1.0
40 yrs; 0° flexion Ratio = AS:PH AS:PH to IS ratio
26F, mean age 56 Quads relaxed Compared with IS ratio on XR PH
years) and MRI

Seyahi et al30 77 (105 knees): Lateral XR Perpendicular distance Distance:

‘Normal’ 30° flexion between PA: > 10 mm
subjects - no the IPP and the ‘intercondylar Poor correlation between
degenerative shelf angle’ (ISA) corrected ISA-corrected method and
change on ‘Blumensaat’ line projected ISA IS, MIS and BP methods
plain radiograph anteriorly through the roof of
(23M, 54F; the femoral trochlea.
mean age 32 yrs, ISA = mean angle between
range 21 to 62 yrs) Blumensaat line and
femoral shaft
axis in 105 knees = 147°
Compared with IS, MIS
and BP

Biedert and 66: MRI Baseline patella (BLp) = vertical Patellotrochlear index:
Albrecht34 ‘Normal’ subjects: Sagittal length of PAS; Baseline PA: ≤ 12.5%
Most had meniscal 0° flexion trochlea (BLT)= vertical height
BLP PI: ≥ 50%
or ACL 15° external rotation of trochlear articular surface t-test showed 2nd observer
pathologies. No Quads relaxed from most superior femoral intra-observer error;
patellofemoral aspect to most inferior aspect BLT Inter-observer correlation
complaints; of PAS; high and significant
(44M; 22F; 12 to 36 Patellotrochlear index (%) =
yrs) BLT/BLP x 100
3 observers at 2 different times

* ACL, anterior cruciate ligament

† XR, radiographic
‡ PAS, patella articular surface; IS, Insall-Salvati; BP, Blackburne-Peel
§ PA, patella alta; PI, patella infera; N, normal; ISA, intercondylar shelf angle; MIS, modified Insall-Salvati

tangent of the anterior cortical line of the femur, patella alta
is diagnosed. It is synonymous with the clinical evaluation,
where alta is diagnosed if, with the knee in 90° of flexion, the
patella points to the ceiling, rather than toward the torso.15
The fact the femur is curved makes the anterior cortical line
difficult to position. Patella infera cannot be diagnosed using
this method.
Norman et al33 developed a truly direct method, originally
described in Swedish in 1976 and published in English in
1983, which assessed the height of the patella in its most
proximal position, with the knee in full extension and the
quadriceps contracted. The perpendicular distance between
the inferior margin of the patellar articular surface and the
distal aspect of the femoral condyles, expressed relative to

THE JOURNAL OF BONE AND JOINT SURGERY

 THE MEASUREMENT OF PATELLAR HEIGHT
body height rather than absolute measurements, is used
because it was found that this ratio is equal in both males
and females, unlike the absolute distances, thus requiring
only one set of normal values. This gender difference was
also observed by Egund et al.26 The biggest drawback is the
fact that the height is measured parallel to the anterior cortex
of the femur, which is curvilinear, allowing both intra- and
inter-observer errors. No subsequent studies to test this pos-
sibility have been performed. Also, the use of a lateral radio-
graph of the knee in full extension, with the quadriceps fully
contracted, is not part of the standard imaging technique,
making it less practical.
It was not until 1989 that Burgess14 devised a method
using lateral radiographs of the knee at any angle of flexion
which produced tension in the patellar tendon. The direct
ratio between the distance from the midpoint of the articular
surface of the patella to the tibial joint line and the antero-
posterior (AP) dimension of the femoral condyles perpendic-
ular to the mechanical axis of the femoral shaft, was
compared with the Insall-Salvati and Blackburne-Peel ratios.
The direct method was deemed better at determining a nor-
mal range of values. However, although Burgess applied the
problem of variation in patellar size and the position of the
tibial tubercle to recognise the inaccuracies of indirect
methods, he did not acknowledge that the femoral condyles
could also vary in size. In addition, he did not identify which
condylar width is used in the calculation: the lateral condyle
usually has a greater AP dimension.
The most recent technique was described by Biedert and
Albrecht in 200634 as the ‘patellotrochlear index’. This
involves MRI rather than radiographs and uses chondral
landmarks rather than bony ones. This is logical, because
patellofemoral biomechanics relates to articular surface
interaction. Both Staeubli et al35 and van Huyssteen et al36
have shown a morphological difference between the articular
cartilage and the underlying subchondral osseous anatomy
of the patella and the femoral trochlea. Biedert and
Albrecht34 adapted a radiographic technique, first described
in 1969 by Bernageau et al,37 where the relationship between
the superior line of the trochlea and the inferior edge of the
patella is applied to articular cartilaginous landmarks rather
than the corresponding bony ones. The original technique
of Bernageau resulted in a large range of normal values from
-6 to +6 mm. It was also criticised by Caton38 as not being
useful when planning surgery in patients with patella alta or
infera, and almost impossible to use in patients with severe
trochlear dysplasia. The need to change the imaging modal-
ity from plain film radiography to MRI does not make the
technique less practical, as most patients with anterior knee
pain will undergo an MR scan as part of their management.
A similar MRI method was described by Miller et al,18 who
calculated the ratio between the patellar articular surface and
the anterior femoral physeal height on sagittal MR scans,
and tested the accuracy of the standard Insall-Salvati ratio
when using MRI. Unfortunately, the correlation between the
two methods was only fair. Biedert34 deemed the use of the
VOL. 92-B, No. 8, AUGUST 2010
1051
physeal scar as the upper articular landmark to be inaccurate
when developing the patellotrochlear index, subsequently
finding that with the new adjustments, the intra-observer
error was low and the inter-observer correlation was signifi-
cantly high. It was also acknowledged that the effects of a
tensed quadriceps, a change in the flexion angle and weight-
bearing on this technique were not addressed. Biedert and
Albrecht’s34 method has recently been tested by Barnett et
al,39 who compared it to the Insall-Salvati, Blackburne-Peel
and Caton-Deschamps ratios in subjects with patellofemoral
dysplasia, and found that the patellotrochlear index does not
correlate well with the other ratios for patella alta, although
all had good intra- and inter-observer reliability. This sup-
ported the theory that chondral rather than osseous relation-
ships are more important and clinically relevant.
Paediatric studies. Micheli et al9 described a method of mea-
suring patellar height in children. This study is the only one
identified, where the authors have acknowledged that adult
ratios cannot be applied accurately to the paediatric popula-
tion (anterior knee pain is common in adolecents) due to lack
of ossification and subsequently developed a new method.
They investigated how the growth spurt can affect the devel-
opment of patella alta by performing serial radiographs at
six- to 12-month intervals over a period of up to ten years.
Because only a single AP view was used to assess growth in
the lower limb, they adapted the principles of Brattström28 to
fit the radiograph, thus evolving a new technique, but failed
to recognise the significance of skeletal maturity. The length
of the patellar tendon was calculated as the difference
between the inferior pole of the patella and the tibial plateau.
This differed from Brattström, who used the approximate
joint line as his reference point. If this was greater than zero,
then patella alta was diagnosed. However, not surprisingly,
all of the 19 children studied had a positive value, although
they found that patella alta is more likely to develop in girls
than in boys during the adolescent growth spurt. The omis-
sion of the ages of the subjects is an important error, as full
ossification may not have occurred and would in part
explain why all subjects had positive values. Walker, Harris
and Leicester40 also showed that young children all have a
high Insall-Salvati ratio, which decreased with age and
became the same as the adult values at approximately ten
years in girls and 12 in boys. Full ossification occurred at
15 and 17 years of age, respectively.
Koshino and Sugimoto10 described a novel technique that
uses the midpoints of the femoral and tibial physes to calcu-
late a ratio between the patellotibial distance and the
tibiofemoral distance, thereby eliminating the need to adjust
for the extent of epiphyseal ossification. Although only a
small series of 59 knees in 36 children, the method produced
stable values for all angles of flexion between 30° and 90°,
making it practical to perform in children, 3.4% of whom
had patella alta using their calculations. The authors also
determined both Blumensaat and Insall-Salvati ratios using
the same radiographs and found that 46% and 67% of the
subjects, respectively, would have been classed as having
1052 C. L. PHILLIPS, D. A. T. SILVER, P. J. SCHRANZ, V. MANDALIA
patella alta, thus emphasising how adult ratios cannot be
applied to children. They recognised that one drawback with
this method was that cautious line-drawing through the phy-
ses is required, owing to their irregular morphology. Apari-
cio et al41 compared this with the Blackburne-Peel and
Caton-Deschamps methods and found that, although all
three had low inter-observer variability, Koshino’s was not as
reproducible as the Caton-Deschamps ratio, with the latter
also not being affected by skeletal maturity.
Method limitations. A review of all the methods that have
been devised to assess patellar height has demonstrated that,
although the patellofemoral relationship is the most impor-
tant aspect of understanding the causes of anterior knee pain,
most techniques are indirect. The last to be developed, by
Biedert and Albrecht in 2006,34 is probably the closest to
perfect, to date, as it uses MRI, relates the patella directly to
the femur, and uses cartilaginous landmarks, thereby elimi-
nating osseous variation, which is possibly the greatest cause
of error in measuring patellar height accurately.
Of the studies that resulted in the description of the four
most established methods, Insall-Salvati,12 Blackburne-
Peel,21 Caton-Deschamp,22 and the modified Insall-Salvati19,
only the Caton-Deschamp included knees that were asymp-
tomatic, in an endeavour to establish a normal range of val-
ues. The subjects included in the remaining three studies had
either meniscal or cruciate ligament pathology, which was
deemed acceptable by the authors on the assumption that
these soft-tissue problems have no effect on the
patellofemoral relationship. This has been shown to be
incorrect,42,43 and any proposed method should ideally be
tested using only subjects with clinically and radiologically
normal knees, before applying it to patients with anterior
knee pathology. Of the less well-known methods, only
Egund et al,26 Koshino and Sugimoto,10 Bernageau et al37
and de Carvalho et al25 managed to use normal knees, but
even then, normality was only defined as being ‘subjects with
no previous knee symptoms’. They did not undergo a clinical
examination and none were excluded on the basis of the sub-
sequent radiographic appearance.
Furthermore, little reference is made to the inherent ana-
tomical differences between men and women, and only two
studies look at ethnic diversity,11,44 where both showed that
there are significant differences between Europeans and
Arabs, Africans and Chinese, with the non-European popula-
tions having a much higher rate of patella alta. Leung et al11
developed the ‘patella alta index’ as the ratio of the sum of
patellar length and patellar tendon length, to the length of the
patellar articular surface, with a much higher range of normal
values to better suit the Southern Chinese population.
Accuracy also seems to be a problem, with few studies
looking at the inter- and intra-observer errors of any method,
even the four most popular ones. Only three publications can
be identified that compare the Insall-Salvati, the modified
Insall-Salvati, the Blackburne-Peel and Caton-Deshamps
methods, and even then abnormal knees were used.15,16,45
Blackburne-Peel and Caton-Deschamps seem to have better
accuracy and reproducibility, but in the absence of any studies
using normal knees, these cannot be deemed as fully reliable.
As Biedert34 has noted, no study where a new technique
has been described has looked at how quadriceps tension,
the angle of flexion and weight-bearing influences patellar
height. These issues have been looked at separately, but not
in the same study.6,7,10,18,46 The only way to solve this prob-
lem is to use dynamic MRI, but at present this is only practi-
cal for research purposes to define what is normal.
Yiannakopoulos et al46 have recently shown that in normal
knees the height of the patella changes significantly on
weight-bearing through the effect of quadriceps contraction,
regardless of which of the four most common ratios is used
to calculate this. Although they gave no absolute measure-
ments, there was a significant mean difference between the
ratios calculated for the non-weight-bearing and fully
weight-bearing groups of 0.2 (p < 0.05).46
Having reviewed the extensive literature on this contro-
versial subject, it appears that only the passing of time has led
to acceptance of the most common methods for measuring
patellar height. Not only is there little concrete evidence to
support the accuracy and validity of these methods, but none
has proved to be suitable for universal application, although
both the Blackburne-Peel and Caton-Deschamps methods
may be the most reliable indirect plain-film radiographic
techniques. Direct methods have either proved too complex
or too novel to be adopted regularly. MRI is currently under-
used in the assessment of patellar height and undoubtedly
has great potential for use in further research. It would
require a very large study with a truly population comprising
all ages, genders and ethnicities to determine a truly normal
range of values for any technique. However, before this vast
undertaking can be considered, a method that accurately
reflects the patellofemoral relationship and can also be used
routinely and repeatedly in a clinical situation needs to be
established. Ideally this should be able to demonstrate the
patellofemoral relationship under physiological loading con-
ditions in the symptomatic knee. Nonetheless, until dynamic
imaging techniques become standardised and routinely
available, we need to identify the best method of imaging and
measuring patellar height from those currently available.
From the limitations described above, there is clearly signifi-
cant scope for improving the methods designed to date, inso-
far as to say that it is still very much an ‘evolving’ project,
even after nearly eight decades.
No benefits in any form have been received or will be received from a commer-
cial party related directly or indirectly to the subject of this article.
References
1. Blumensaat C. Die Lageabweichungen und Verrenkungen der Kniescheibe. Ergebn
Chir Orthop 1938;31:149-223 (in German).
2. Amis AA, Farahmand F. Biomechanics masterclass: extensor mechanism of the
knee. Curr Orthop 1996;10:102-9.
3. Singerman R, Davy DT, Goldberg VM. Effects of patella alta and patella infera on
patellofemoral contact forces. J Biomech 1994;27:1059-65.
4. Yamaguchi GT, Zajac FE. A planar model of the knee joint to characterize the knee
extensor mechanism. J Biomech 1989;22:1-10.
THE JOURNAL OF BONE AND JOINT SURGERY
 THE MEASUREMENT OF PATELLAR HEIGHT
5. Kannus PA. Long patellar tendon: radiographic sign of patellofemoral pain syn-
drome: a prospective study. Radiology 1992;185:859-63.
6. Ward SR, Powers CM. The influence of patella alta on patellofemoral joint stress
during normal and fast walking. Clin Biomech (Bristol, Avon) 2004;19:1040-7.
7. Ward SR, Terk MR, Powers CM. Patella alta: association with patellofemoral
alignment and changes in contact area during weight-bearing. J Bone Joint Surg [Am]
2007;89-A:1749-55.
8. Insall J, Goldberg V, Salvati E. Recurrent dislocation and the high-riding patella.
Clin Orthop 1972;88:67-9.
9. Micheli LJ, Slater JA, Woods E, Gerbino PG. Patella alta and the adolescent
growth spurt. Clin Orthop 1986;213:159-62.
10. Koshino T, Sugimoto K. New measurement of patellar height in the knees of chil-
dren using the epiphyseal line midpoint. J Paediatr Orthop 1989;9:216-18.
11. Leung YF, Wai YL, Leung YC. Patella alta in souther china: a new method of mea-
surement. Int Orthop 1996;20:305-10.
12. Insall J, Salvati E. Patella position in the normal knee. Radiology 1971;101:101-4.
13. Grelsamer RP, Proctor CS, Bazos AN. Evaluation of patellar shape in the sagittal
plane: a clinical analysis. Am J Sports Med 1994;22:61-6.
14. Burgess RC. A new method of determining patellar position. J Sports Med Phys Fit-
ness 1989;29:398-9.
15. Seil R, Muller B, Georg T, Kohn D, Rupp S. Reliability and interobserver variability
in radiological patellar height ratios. Knee Surg Sports Traumatol Arthrosc
2000;8:231-6.
16. Berg EE, Mason SL, Lucas MJ. Patellar height ratios: a comparison of four mea-
surement methods. Am J Sports Med 1996;24:218-21.
17. Schlenzka D, Schwesinger G. The height of the patella: an anatomical study. Eur
J Radiol 1990;11:19-21.
18. Miller TT, Staron RB, Feldman F. Patellar height on sagittal MR imaging of the
knee. AJR Am J Roentgenol 1996;167:339-41.
19. Grelsamer RP, Meadows S. The modified Insall-Salvati ratio for assessment of
patellar height. Clin Orthop 1992;282:170-6.
20. Kadakia NR, Ilahi OA. Interobserver variability of the Insall-Salvati ratio. Orthope-
dics 2003;26:321-3.
21. Blackburne JS, Peel TE. A new method of measuring patellar height. J Bone Joint
Surg [Br] 1977;59-B:241-2.
22. Caton J, Deschamps G, Chambat P, Lerat JL, Dejour H. Patella infera: a propos
of 128 cases. Rev Chir Orthop Reparatrice Appar Mot 1982;68:317-25 (in French).
23. Hepp WR. 2 new methods for determination of the height of patella. Z Orthop Ihre
Grenzgeb 1984;122:159-66 (in German).
24. Defrate LE, Nha KW, Papannagari R, et al. The biomechanical function of the
patella tendon during in-vivo weight-bearing flexion. J Biomech 2007;40:1716-22.
25. de Carvalho A, Holst Andersen A, Topp S, Jurik AG. A method for assessing the
height of the patella. Int Orthop 1985;9:195-7.
26. Egund N, Lundin A, Wallengren NO. The vertical position of the patella: a new
radiographic method for routine use. Acta Radiol 1988;29:555-8.
VOL. 92-B, No. 8, AUGUST 2010
1053
27. Boon-Itt S. The normal position of the patella. AJR Am J Roentgenol 1930;24:389-94.
28. Brattström H. Patella alta in non-dislocating knee joints. Acta Orthop Scand
1970;41:578-88.
29. Andersen P Thestrup-. Congenital deformities of the knee joint in dislocation of the
patella and achondroplasia. Acta Orthop 1964;28:31-50.
30. Seyahi A, Atalar AC, Koyuncu LO, Cinar BM, Demirhan M. Blumensaat line and
patellar height. Acta Orthop Scand Traumatol Turc 2006;40:240-7 (in Turkish).
31. Labelle H, Peides JP, Lévesque HP, Fauteux P, Laurin CA. Evaluation of patellar
position by tangential x-ray visualisation. Union Med Can 1976;105:870-3 (in French).
32. Laurin CA. The investigation of the patellofemoral joint. J Bone Joint Surg [Br]
1977;59-B(Suppl):107.
33. Norman O, Egund N, Ekelund L, Rünow A. The vertical position of the patella.
Acta Orthop Scand 1983;54:908-13.
34. Biedert RM, Albrecht S. The patellotrochlear index: a new index for assessing
patellar height. Knee Surg Sports Traumatol Arthrosc 2006;14:707-12.
35. Staeubli HU, Bosshard C, Porcellini P, Rauschning W. Magnetic resonance
imaging for articular cartilage: cartilage-bone mismatch. Clin Sports Med
2002;21:417-33.
36. van Huyssteen AL, Hendrix MR, Barnett AJ, Wakeley CJ, Eldridge JD. Carti-
lage-bone mismatch in the dysplastic trochlea: an MRI study. J Bone Joint Surg [Br]
2006;88-B:688-91.
37. Bernageau J, Goutallier D, Debeyre J, Ferrané J. New exploration technic of the
patellofemoral joint: relaxed axial quadriceps and contracted quadriceps. Rev Chir
Orthop Reparatrice Appar Mot 1975;61(Suppl 2):286-90 (in French).
38. Caton J. Diagnostic et traitement des rotules hautes et des rotules basses. E-
mémoires de l’Académie Nationale de Chirurgie 2008;7:33-9 (in French).
39. Barnett AJ, Prentice M, Mandalia V, Wakeley CJ, Eldridge JD. Patellar height
measurement in trochlear dysplasia. Knee Surg Sports Traumatol Arthrosc
2009;17:1412-15.
40. Walker P, Harris I, Leicester A. Patellar tendon-to-patella ratio in children. J Pae-
diatr Orthop 1998;18:129-31.
41. Aparicio G, Abril JC, Albiñana J, Rodriguez-Salvanés F. Patellar height ratios in
children: an interobserver study of three methods. J Paediatr Orthop 1999;8:29-32.
42. Ficat P, Hungerford D. Disorders of the patellofemoral joint. Baltimore: Williams &
Wilkins, 1977.
43. Iwano T, Kurosawa H, Tokuyama H, Hishikawa Y. Roentgenographic and clinical
findings of patellofemoral osteoarthrosis: with special reference to its relationship to
femorotibial osteoarthrosis and etiologic factors. Clin Orthop 1990;252:190-7.
44. Karadimas JE, Piscopakis N, Syrmalis L. Patella alta and chondromalacia. Int
Orthop 1981;5:247-9.
45. Rogers BA, Thornton-Bott P, Cannon SR, Briggs TW. Interobserver variation in
the measurement of patellar height after total knee arthroplasty. J Bone Joint Surg
[Br] 2006;88-B:484-8.
46. Yiannakopoulos CK, Mataragas E, Antonogiannakis E. The effect of quadriceps
contraction during weight-bearing on four patellar height indices. J Bone Joint Surg
[Br] 2008;90-B:870-3.