CLINICAL ORTHOPAEDICS AND RELATED RESEARCH

Number 423, pp. 166–171

© 2004 Lippincott Williams & Wilkins

The Optimum Knee Flexion Angle for Skyline

Radiography Is Thirty Degrees
A. P. Davies, MRCS; J. Bayer, MRCS; S. Owen-Johnson FRCS (Orth); L. Shepstone, PhD;
C. Darrah, RGN Dip Sci; M. M. Glasgow, FRCS; and S. T. Donell, FRCS (Orth)

There is wide variation in practice among orthopaedic sur- phy. Numerous techniques for taking the skyline radio-
geons regarding the use of skyline tangential patellar radio- graph have been suggested by various authors. Settegast10
graphs of the patellofemoral joint in patients with anterior proposed positioning the patient prone with the knee
knee pain. Various techniques are available for taking such flexed beyond 90° (Fig 1A) and this technique was advo-
radiographs and numerous radiologic parameters can be
cated in a review article on patellofemoral imaging.9 Mer-
measured from them. There is no information as to which
knee flexion angle is superior when taking skyline radio-
chant et al8 originally described a technique with the pa-
graphs. The purpose of the current study was to compare the tient supine and the knee flexed to 45° (Fig 1B). In later
results of skyline radiographs taken at 30°, 50°, and 90° knee work however, Merchant7 reported cases in which skyline
flexion. Sixty-seven knees were studied prospectively. All ra- radiographs obtained at 45° knee flexion seemed normal
diographs were made anonymous and then were studied in whereas radiographs taken at 30° showed patellar subluxa-
random order by two observers. Five recognized patellofem- tion. Laurin et al6 proposed that the patient be positioned
oral parameters were recorded. All the parameters studied supine with the knee flexed to 20° (Fig 1C). Ficat et al4
were measured most reproducibly when measured from the required a series of three radiographs be taken with the
radiograph taken at 30° knee flexion. The parameters were knee in 30°, 60° and 90° flexion (Fig 1D). Ficat et al4 also
least reproducible when measured from the radiograph reported cases in which skyline radiographs obtained at
taken at 90° knee flexion. In terms of detecting abnormality
60° knee flexion seemed normal whereas radiographs
of the parameters measured, the radiographs taken at 30°
and 50° were similar. The radiograph taken at 90° knee
taken at 30° showed patellar subluxation. One study of
flexion detected the fewest abnormalities. Skyline radio- clinical practice in the United Kingdom11 showed that
graphic examination of the patellofemoral joint should be a more than 19% of orthopaedic surgeons would not request
mandatory part of the investigation of all knee problems. a skyline image whereas 21% would request more than
One radiograph, ideally taken at 30° knee flexion, offers the one image at various knee flexion angles. Eight knee flex-
best means of assessing the patellofemoral joint. There is no ion angles for skyline radiography are in current clinical
benefit to requesting multiple images at different angles of use.
knee flexion. Various radiologic parameters have been defined to al-
low measurement of numerous patellofemoral disorders.
Brattstrom2 defined the sulcus angle as the angle formed
Imaging of the patellofemoral joint is controversial. An- by the trochlear groove (Fig 2A). The sulcus angle has a
terior knee pain is a common source of referral of young normal value of 138° with a standard deviation of 6°. The
patients to the orthopaedic clinic. The first imaging mo- congruence angle was defined by Merchant et al8 to mea-
dality available to most surgeons will be plain radiogra- sure the degree of patellar subluxation (Fig 2B). It mea-
sures the angular distance between the articular ridge of
Received: November 27, 2002 the patella and a reference line that bisects the sulcus
Revised: April 17, 2003; August 6, 2003 angle. If the apex of the patellar articular ridge is lateral to
Accepted: September 16, 2003 the zero line, the congruence angle is designated positive.
From the Department of Orthopaedics, Norfolk and Norwich University
Hospital, Norwich, UK. If it is medial, the congruence angle is negative. The nor-
Correspondence to: A.P. Davies, MRCS, Department of Orthopaedics, Nor- mal congruence angle was defined at 45° knee flexion as
folk and Norwich University Hospital, Colney Lane, Norwich, NR4 7UY, -6° with a standard deviation of 11° (Fig 2B). The lateral
UK. Phone: 01603 286706; Fax: 0117 9595936; E-mail:
andypdavies@hotmail.com. patellofemoral angle was described by Laurin et al6 at 30°
DOI: 10.1097/01.blo.0000129160.07965.e7 knee flexion as being the angle between the intercondylar

166
Number 423
June 2004 Skyline Radiographs and the Patellofemoral Joint 167

MATERIALS AND METHODS

Ethical committee approval was granted before the start of the
study. Written informed consent was obtained from all subjects
before participation in the study. Between September 1999 and
June 2000, 72 new patients between 12 and 30 years presented

Fig 1A–D. (A) Using the radiographic technique of Sette-

gast,10 the patient is prone with the knee flexed to 120° and the
xray beam is vertical. (B) For the technique of Merchant et al,8
the knee is flexed to 45° and the xray beam is proximal to
distal. (C) The technique of Laurin et al6 has the patient’s knee
flexed 20° to 30° with the xray beam distal to proximal. (D)
Ficat et al4 had the patient flex the knee 30° and the xray beam
was distal to proximal.

line and the lateral patellar facet (Fig 2C). This angle is
normal if it opens laterally. It is a measure of patellar tilt.
Pure lateral patellar displacement was measured at 30°
knee flexion by Laurin et al6 by calculating the distance
between a perpendicular to the intercondylar line at the
level of the medial femoral condyle and the medial edge of
the patella (Fig 2D). It is abnormal if the patella lies lateral
to this line. The facet angle was defined by Wiberg12 as the
angle formed between the medial and lateral facets of the
patella (Fig 2E). It is a measure of the depth of the patella
and therefore its stability in the trochlear groove. The pur-
pose of the current prospective study was to determine the Fig 2A–E. (A) The sulcus angle2 with a mean of 138° and a
intraobserver and interobserver reliability of the five pa- standard deviation of 6° measures trochlear dysplasia. (B) The
rameters at each of the three knee flexion angles studied. congruence angle8 with a mean of -6° and a standard devia-
We studied the changes in these parameters as knee flex- tion of 11° measures patellar subluxation. (C) Lateral displace-
ion increased and determined at which knee flexion angle ment6 is abnormal if the medial border of the patella is lateral
to the medial condyle. This angle measures pure patellar
the greatest number of abnormal parameters could be de- translocation. (D) The lateral patellofemoral angle6 normally
tected. Using this method we determined the optimum opens laterally and measures pure patellar tilt. (E) The facet
knee flexion angle for skyline patellofemoral radiographs. angle12 measures the depth and fit of the patella in the trochlea.
 Clinical Orthopaedics
168 Davies et al and Related Research

to our institution with symptoms of anterior knee pain. After The sulcus angle was measured most reproducibly at
informed consent was obtained, these patients had a series of 30° knee flexion and least reproducibly at 90° knee flexion
three skyline (tangential patellar) radiographs of their patello- (Table 1). At all angles of knee flexion studied, the repro-
femoral joints taken at 30°, 50°, and 90° knee flexion. The
ducibility of the sulcus angle measurement was high (ICC,
method of Laurin et al6 was used to obtain the skyline patello-
femoral images. A set of knee rests was made at the three angles
0.865–0.930). This confirms findings from an earlier
of knee flexion, to ensure consistency of knee flexion angle and study.3
to support the knee, which allowed the quadriceps to relax at the The measured sulcus angle showed a statistically sig-
time that the images were taken. As is standard practice in our nificant progression of narrowing of the sulcus as the knee
department, AP radiographs of the knee with the patient standing flexion angle increased. The mean sulcus angle at 30° knee
and lateral radiographs with the patient supine also were taken. flexion was 143.3° which narrowed to 142.5° at 50° knee
The skyline radiographs were evaluated by two observers flexion and additionally to 140.6° at 90° knee flexion. This
who were blinded to the study. The sulcus angle, lateral patel- progression was highly statistically significant p < 0.001
lofemoral angle, facet angle, lateral displacement, and congru-
(Table 2). Lateral displacement was measured most repro-
ence angle were calculated on each image obtained. One ob-
server (APD) evaluated every radiograph on two occasions sepa- ducibly at 30° knee flexion and least reproducibly at 90°
rated by several days to calculate intraobserver correlation knee flexion (Table 1). The degree of patellar translocation
coefficients. as measured by lateral displacement also was highly re-
producible at all knee flexion angles studied (ICC, 0.715–
Statistical Analysis 0.796). The measured distance between the medial con-
All data were entered into a MicrosoftTM Excel spreadsheet file dyle and the medial border of the patella did not change
(Microsoft Corporation, Seattle, WA) and submitted for inde- progressively as the knee flexion angle increased (Table
pendent statistical analysis. The intraclass correlation coefficient 2). Measurement of the congruence angle was most repro-
was calculated to assess the reliability of measurements scored ducible at 30° knee flexion and least reproducible at 90°
on a continuous scale (sulcus angle, congruence angle, lateral knee flexion (Table 1). The extent of patellar subluxation
displacement, facet angle). The kappa statistic was used to assess as measured by the congruence angle showed good repro-
agreement with respect to categorical diagnoses (abnormal pa- ducibility at 30° knee flexion (ICC, 0.657) but this de-
tellofemoral angle). Statistical significance was set at the 5% creased to only poor reproducibility at 90° knee flexion
level. (ICC, 0.383).
There was a trend toward progressively more negative
RESULTS congruence angles as the knee flexion angle increased.
The mean congruence angle was -4.34° at 30° knee flex-
Fifty-eight patients (69 knees) fulfilled the entry criteria ion, -5.61° at 50° knee flexion, and -6.66° at 90° knee
for the study. Two sets of radiographs (two knees) could flexion. This did not reach statistical significance (Table 2).
not be retrieved for repeat analysis and so the measure- Measurement of the lateral patellofemoral angle was
ments from the remaining 56 patients (67 knees) form the most reproducible at 30° knee flexion and least reproduc-
data set for this study. There were 33 male and 23 female ible at 90° knee flexion. The measurement of patellar tilt
patients. Pain was bilateral in eight male and three female by the lateral patellofemoral angle showed fair reproduc-
patients, which resulted in 41 knees in males and 26 knees ibility at 30° knee flexion (ICC, 0.589) but decreased to
from females for the study. The mean age of the patients poor reproducibility at 90° knee flexion (ICC, 0.150)
was 18.3 years (range, 12.5–29.7 years). (Table 1).

TABLE 1. Intraclass Correlation Coefficients for Reliability of Different Measures between and within
Raters for Different Degrees of Flexion
Between Raters Within Rater
Parameter 30° Flexion 50° Flexion 90° Flexion 30° Flexion 50° Flexion 90° Flexion
Sulcus angle 0.930 0.908 0.865 0.915 0.885 0.919
Facet angle 0.597 0.591 0.597 0.757 0.718 0.724
Lateral displacement 0.796 0.761 0.715 0.894 0.838 0.733
Lateral PF angle 0.589 0.538 0.150 0.659 0.480 0.114
Congruence angle 0.657 0.544 0.383 0.850 0.663 0.591

PF = patellofemoral
Number 423
June 2004 Skyline Radiographs and the Patellofemoral Joint 169

TABLE 2. Mean Values1 (Standard Deviations) of the Five Morphometrics

by Rater, Degree of Flexion, and Identification of Subluxation
Degree of Flexion
Parameter N2 30° 50° 90° p Value3
Sulcus angle 45 143.3 (7.06) 142.5 (6.86) 140.6 (6.48) < 0.001
Facet angle 53 131.2 (7.62) 131.5 (7.49) 132.2 (7.17) 0.588
Lateral displacement 47 0.58 (3.96) 0.30 (3.66) 0.51 (2.75) 0.966
Lateral PF angle 46 9.29 (6.81) 9.73 (5.55) 11.83 (9.31) 0.040
Congruence angle 44 −4.34 (16.23) −5.61 (13.73) −6.66 (13.60) 0.279
1
Taken from the average readings of the three measures
2
Number of observations used in the repeated measures ANOVA (subjects with measurements from all 3 raters at all
3 angles of flexions)
3
p Values resulting from a repeated measures ANOVA (with subject repeat measurements over rate and angle of
flexion) using the Huynh-Feldt adjustment
PF = patellofemoral

The measured lateral patellofemoral angle became pro- The optimum knee flexion angle for skyline radiogra-
gressively more positive as the knee flexion angle in- phy is 30°. Measurements of radiographic patellofemoral
creased. The mean lateral patellofemoral angle was 9.29° parameters are most reproducible from radiographs ob-
at 30° knee flexion and increased to 9.73° at 50° knee tained at this angle and the greatest number of abnormali-
flexion and to 11.83° at 90° knee flexion. This trend was ties is detected.
statistically significant (p ⳱ 0.04) (Table 2).
Measurement of the facet angle of the patella was most
reproducible at 30° knee flexion and least reproducible at DISCUSSION
90° knee flexion. The facet angle showed fair reproduci-
bility and varied little (ICC, 0.597–0.591)(Table 1). The Plain radiography is most likely to remain the primary
measured facet angle also varied little as the knee flexion examination for patients with anterior knee pain attribut-
angle increased. The mean facet angle was 131.2° at 30° able to the patellofemoral joint. These images are increas-
knee flexion, 131.5° at 50° knee flexion, and 132.2° at 90° ingly being supplemented by CT and MRI scanning, how-
knee flexion (Table 2). This was not statistically signifi- ever such examinations are costly and time consuming and
cant. In every case, the reproducibility of measurement are not available to all patients. When ordering any ex-
was best for the radiograph taken at 30° flexion, interme- amination involving radiation exposure to the patient it is
diate for the film taken at 50° flexion, and worst for the imperative that the optimum amount of clinically useful
film taken at 90° flexion. In terms of reproducibility of information is obtained for the least amount of radiation
measurement, the radiograph taken at 30° flexion was su- exposure.
perior to the other two angles (Table 1). We compared skyline radiographs of knees obtained at
In terms of the frequency of detecting abnormal param- various flexion angles using a collection of well-
eters, the radiographs taken at 30° and 50° flexion were established radiologic parameters to measure different as-
similarly able to define abnormal patellar tilt and subluxa- pects of the patellofemoral joint. These radiologic param-
tion. The radiograph taken at 90° flexion was the least eters have defined normal ranges which allowed us to
sensitive at detecting abnormality. There were 47 knees determine the reproducibility of measurements and the fre-
with at least one radiologic abnormality. The measure-
ments of the remaining 20 knees were within normal limits
for all parameters at all angles of knee flexion. The 30°-
radiograph detected 11 abnormal sulcus angles, eight ab- TABLE 3. Performance of the Three Skyline
Radiographs in Detecting Abnormal Parameters
normal facet angles, four abnormal lateral patellofemoral
angles, and 15 abnormal congruence angles (Table 3). The 30° Film 50° Film 90° Film
50° radiograph detected 11 abnormal sulcus angles, 10 Parameter Abnormal Abnormal Abnormal
abnormal facet angles, five abnormal lateral patellofemor- Sulcus angle 11 11 8
al angles and 15 abnormal congruence angles. The 90°- Facet angle 8 10 10
radiograph detected eight abnormal sulcus angles, 10 ab- Lateral patellofemoral
angle 4 5 1
normal facet angles, one abnormal lateral patellofemoral Congruence angle 15 15 14
angle, and 14 abnormal congruence angles.
 Clinical Orthopaedics
170 Davies et al and Related Research

quency of abnormal parameters being detected. The con- showed the lines parallel and in six knees (20%) the angle
sistency with which these radiologic features can be mea- opened medially.5 In the current study, four of the 67
sured from such radiographs is defined and the efficiency knees (6%) had a lateral patellofemoral angle that opened
of the radiographs taken at the different knee flexion medially when measured at 30° knee flexion. This figure
angles is compared. decreased to one knee (1.5%) with an abnormal lateral
The limitations of this study include the number of patellofemoral angle when measured at 90° knee flexion,
subjects examined and the choices of the radiologic pa- emphasizing the importance of minimal knee flexion
rameters studied. Numerous radiologic parameters have angles. Merchant7 and Ficat et al4 published cases in
been described for the patellofemoral joint. The param- which subluxation was present at 30° and absent at 45° or
eters chosen for this study were those used in our clinical greater knee flexion. Despite this, the technique of Sette-
practice. gast10 was used to show patellofemoral imaging in one
Merchant et al8 defined the congruence angle in 100 review.9 We included the 90° radiograph in the current
normal knees as -6° with a standard deviation of +16° study to measure the parameters of the patellofemoral joint
(Table 4). In a population of 25 patients with known re- across various knee flexion angles and because of the con-
current patellar dislocation, Merchant et al8 found the tinuing clinical use of this technique.11
mean congruence angle to be +23°. Aglietti et al1 found The results of our study show that the radiologic pa-
the mean congruence angle in a group of 150 healthy rameters studied are most reproducibly measured from
subjects to be minus 8° with a standard deviation of +6° skyline tangential patellar images of the patellofemoral
(Table 4). In 37 subjects with patellar subluxation, Aglietti joint taken at 30° flexion. With respect to detection of
et al1 found the mean congruence angle was +16°. In the abnormality, the radiographs taken at 30° and 50° flexion
current study, the mean congruence angle was minus 4.34° were similar. There is no value in obtaining multiple sky-
with a standard deviation of 16.23° when measured at 30° line images from a patient because this increases exposure
knee flexion, which is in keeping with the range of values to radiation with no increase in clinical information.
from healthy subjects in the studies of Merchant et al8 and Changes were seen in the radiologic parameters measured
with a standard deviation of 6° (Table 4). Aglietti at al1 as knee flexion angle increases. For most parameters, there
found the sulcus angle in their series of normal knees to be was a trend toward more normal measurements as knee
137° with a standard deviation of 6°. In a previous study3 flexion angle increased (Table 2). This was in accordance
the mean sulcus angle was 140.4° with a standard devia- with the observations of Merchant7 and Ficat et al4 that
tion of 5.2° in knees with no evidence of trochlear dys- knees could appear radiologically normal at knee flexion
plasia. In the current study, the mean sulcus angle was angles of 45° or greater when there was clear patellar
143.3° with a standard deviation of 7.06° when measured subluxation seen at 30° knee flexion.
at 30° knee flexion (Table 4). At 90° knee flexion, the One skyline patellofemoral radiograph should be ob-
mean sulcus angle was 140.6° with a standard deviation of tained for every patient who presents with anterior knee
6.48°. These findings indicate that the sulcus angles in the pain. The minimum achievable angle of knee flexion
population of symptomatic patients in the current study should be advised when making this image and every ef-
were widened in keeping with some degree of trochlear fort should be made to take a radiograph with the knee at
dysplasia. Laurin et al5 defined the lateral patellofemoral 30° flexion. This is the optimal flexion angle if reproduc-
angle as normal if it opened laterally. In a group of 100 ible measurements are to be made from the images ob-
normal knees, the lateral patellofemoral angle was open tained.
laterally in 97 knees and the lines were parallel in three
knees.5 In 30 knees with patellar subluxation, 24 (80%) References
1. Aglietti P, Insall JN, Cerrulli G: Patellar pain and incongruence.
Clin Orthop 176:217–224, 1983.
2. Brattstrom H: Shape of the intercondylar groove normally and in
TABLE 4. Previous Series Reporting Normal recurrent dislocation of the patella: A clinical and x-ray anatomical
Values for Measured Parameters investigation. Acta Orthop Scand 68(Suppl):1–148, 1964.
3. Davies AP, Costa ML, Shepstone L, et al: The sulcus angle and
Sulcus Congruence extensor mechanism malalignment of the knee. J Bone Joint Surg
Author Angle Angle 82B:1162–1166, 2000.
4. Ficat RP, Philippe J, Hungerford DS: Chondromalacia patellae: A
Merchant et al8 −6° (SD 16°) system of classification. Clin Orthop 144:55–62, 1979.
Aglietti et al1 137° (SD 6°) −8° (SD 6°) 5. Laurin CA, Levesque HP, Dussault R: The abnormal lateral patel-
Brattstrom2 138 (SD 6°) lofemoral angle: A diagnostic roentgenographic sign of recurrent
Davies et al3 140 (SD 5°) patellar subluxation. J Bone Joint Surg 60A:55–60, 1978.
Current study 143 (SD 7°) −4° (SD 16°) 6. Laurin CA, Dussault R, Levesque HP: The tangential xray investi-
gation of the patellofemoral joint. Clin Orthop 144:16–26, 1979.
Number 423
June 2004 Skyline Radiographs and the Patellofemoral Joint 171

7. Merchant AC: Radiographic Evaluation of the Patellofemoral Joint. 10. Settegast A: Typische Roentgenbilder von normalen Menschen.
In Aichroth PM, Cannon Jr WD (eds). Knee Surgery: Current Prac- Lehmanns Med Atlantan 5:211, 1921.
tice. London, Martin Dunitz Ltd 380–388, 1992. 11. Vince AS, Singhania AK, Glasgow MMS: What xrays do we need?
8. Merchant AC, Mercer RL, Jacobsen RH, et al: Roentgenographic A survey of orthopaedic surgeons in the United Kingdom. Knee
analysis of patellofemoral congruence. J Bone Joint Surg 7:101–104, 2000.
56A:1391–1396, 1974. 12. Wiberg G: Roentgenographic and anatomic studies on the femoro-
9. Peterfy CG, Howard DS: Imaging the patellofemoral joint: Current patellar joint with specific reference to chondromalacia patellae.
status and future directions. Am J Knee Surg 10:110–120, 1997. Acta Orthop Scand 12:319–409, 1941.