Reliability of measuring lateral bowing angle of the femur in patients with atypical femur fractures

Abstract

Background:

Lateral femoral bowing causes a higher tensile mechanical load on the lateral side of the femur, which plays a critical role in the pathogenesis of atypical femoral fractures (AFFs). There are many ways to measure lateral femoral bowing on plain radiographs, and there are difficulties in finding a correspondence point between measurements among multiple measurers. The purpose of this study is to prove the best method of correspondence between the investigators by analyzing the reliability of various methods for measuring lateral femoral bowing.

Materials and Methods:

We retrospectively collected data from 85 patients (87 femurs) diagnosed with AFF who had plain radiographs of entire femur from October 2013 to March 2018. The femoral bowing was measured in coronal view of femur plain radiographs using five methods (Sasaki, Morin, Jang, Kim, and Yau) by three examiners, respectively. The intra- and interobserver reliability of each method was assessed using intraclass correlation coefficient (ICC) for continuous variables.

Results:

All methods showed excellent intra- and interobserver reliability with ICC of >0.8. Among five methods, the Yau’s method was the highest reliable method (ICC = 0.980, 95% confidence interval = 0.971–0.986).

Conclusions:

All methods of measuring lateral femoral bowing in the coronal plane of plain radiographs are reliable. And, we recommend Yau’s method, which has a clear reference point for measuring femoral bowing and is highest reproducible.

Keywords

atypical femoral fracture femoral bowing geometry intra- and interobserver reliability

Introduction

According to previous large database studies, the incidence of atypical femoral fractures (AFFs) is estimated to be 1.8–50 per 100,000 persons years.^1
–3 Despite the low incidence of AFFs, there is a considerable concern for AFFs as the aging society and increasing number of osteoporosis patients. Studies on femoral geometry as a risk factor for AFFs have been actively conducted recently.^4
–6 Anterolateral femoral bowing causes a higher tensile mechanical load on the lateral side of the femur, which plays a critical role in the pathogenesis of AFF.⁷ In addition, there have been recent reports comparing the location of AFFs according to the degree of femoral bowing.⁸

Predicting the risk of AFFs and managing for fracture prevention has recently become an important issue,⁹ so it is important to make reliable measurements of anterolateral femoral bowing, an important risk factor for AFFs. There are many ways to measure lateral femoral bowing on plain radiographs, and there are difficulties in finding a correspondence point between measurements among multiple measurers.^7,10

–14 For the accurate description and communication between the measurers, the measurement method should have high reliability and high agreement between the observers. However, there was no study to evaluate the reliability and agreement between measurements of lateral femoral bowing.

The purpose of this study is to determine the best method of correspondence between the investigators by analyzing the reliability of various methods for measuring lateral femoral bowing.

Materials and methods

To evaluate the reliability of measurement of anterolateral bowing of femur, we identified 118 patients who were treated for AFF at our hospital from October 2013 to March 2018.

The inclusion criteria were (1) patients who had radiographs before complete and (2) the availabilities of anteroposterior (AP) view of entire femur. We excluded 33 patients who did not have radiographs of the entire femur before complete fracture. Finally, we included 85 patients in this study. The mean age was 73.0 years (standard deviation, 8.5; range, 45.6–88.4 years). There were 2 men and 83 women. Two of them were treated with bilateral AFFs and the final 87 femurs with AP view of entire femur before complete fracture were included.

All radiographs were taken in the same radiology unit using a standardized protocol. AP radiographs of femur were obtained in supine position at a source to film distance of 110 cm. Hips were kept in extension, and knees were kept in extension and neutral rotation with patella facing anteriorly. In all cases, X-ray beam was centered on the middle of the thigh and the X-ray beam was perpendicular to the patients. All radiographic images were digitally acquired using a picture archiving and communication system (PACS; Impax: Agfa. Antwerp, Belgium), and all measurements on radiographs were subsequently carried out on 19-inch LCD monitor using PACS software.

Three observers who were orthopedic surgeons with 5 (HK), 4 (BS), and 4 (HJ) years of experience and were familiar with radiographs of femur. Consensus building sessions for each method were held before measuring lateral bowing of femur by three orthopedic surgeons and two senior surgeons (YK and KH).

Methods for measurement of lateral bowing of the femur on radiograph

Sasaki’s method

Two lines along the proximal and distal portions of the femoral shaft were drawn on AP X-rays. Lateral femoral shaft bowing angle is defined as the angle between the two lines on the AP view¹³ (Figure 1(a)).

Figure 1.

Assessment methods for measuring lateral femoral bowing on plain radiography. (a) Sasaki’s method, (b) Morin’s method, (c) Jang’s method, (d) Kim’s method, and (e) Yau’s method. (A = angle of lateral femoral bowing.).

Morin’s method

Lateral bowing is the angle between axes of 20% proximal and distal sections of femur diaphysis on AP view⁷ (Figure 1(b)).

Jang’s method

Femoral bows were defined as the angle formed by two lines which are parallel to the proximal and distal portions of femoral shaft. At this point, the proximal line goes through greater trochanter tip and the distal line through intercondylar notch¹⁰ (Figure 1(c)).

Kim’s method

Lateral bowing was measured as the angle formed at the bisection of the two lines drawn through the long axis of the proximal and distal parts of the femur.¹¹ Each line was drawn from the midpoint of the endosteal canal at the level of lesser trochanter for the proximal part and supracondylar ridge for the distal part (Figure 1(d)).

Yau’s method

To measure lateral femoral bowing, the diaphysis of the femur was divided into four equal parts and a line was then drawn in each quarter to highlight the midpoint of the endosteal canal. The degree of lateral femoral bowing was defined as the angle formed between the central line of proximal femoral diaphysis and the central line of distal femoral diaphysis¹⁴ (Figure 1(e)).

Assessment of reliability of measuring lateral bowing of femur

Reliability is defined as consistency of measured values. The femoral bowing was measured in AP view of femur plain radiographs using six methods (Sasaki, Morin, Jang, Kim, Yau, and Park) by three examiners (HK, BS, and HJ), respectively.

Each observer was blinded to other measurements and patients’ data. Orders of measurements were assigned randomly to each observer and all data were collected by a research assistant who did not participate in the reliability sessions. The intraobserver reliability of each method was assessed using the values measured by one examiner, who performed the reassessment 3 weeks later.¹⁵ The interobserver reliability of each radiographic measurement was evaluated with an intraclass correlation coefficient (ICC).

Statistical analysis

The ICC with 95% confidence interval (CI) was used to analyze interobserver reliability of the obtained continuous data among the three observers. The intra- and interobserver reliability was evaluated using the two-way random effects model assuming a single measurement and absolute agreement. And ICC of 1 means perfect reliability and 0 means the opposite. Values of ICC were characterized as poor (<0.20), fair (0.21–0.40), moderate (0.41–0.60), good (0.61–0.80), or excellent (0.81–1.00).¹⁶ Statistical analyses were performed using SPSS version 25.0 (SPSS Inc., Chicago, Illinois, USA).

This study was approved by the institutional review board at our hospital, who waived informed consent.

Results

In five radiological measurements which consisted of continuous data, the intra- and interobserver reliabilities were >0.80. Among five methods, the Yau’s method showed highest intra- and interobserver reliability (Table 1). And, Kim’s method tend to underestimate the lateral bowing angle of femur (Table 2).

Table 1.

Intra- and interobserver reliability of femoral bowing measurements on radiography.

	Intraobserver reliability		Interobserver reliability
	ICC	95% CI	ICC	95% CI
Sasaki	0.930	0.891–0.955	0.881	0.835–0.916
Morin	0.925	0.840–0.960	0.888	0.808–0.932
Jang	0.889	0.827–0.929	0.904	0.859–0.936
Kim	0.839	0.756–0.895	0.946	0.921–0.964
Yau	0.944	0.904–0.966	0.980	0.971–0.986

ICC: intraclass correlation coefficient; CI: confidence interval.

Table 2.

Measurement of each observer according to each method.

	Sasaki	Morin	Jang	Kim	Yau
Observer 1	6.17 ± 6.37	6.15 ± 5.99	5.60 ± 5.19	5.08 ± 4.08	7.49 ± 5.91
Observer 2	6.65 ± 4.36	7.73 ± 5.41	4.90 ± 4.35	4.68 ± 3.77	8.15 ± 5.68
Observer 3	5.98 ± 6.06	6.02 ± 5.89	5.79 ± 5.62	4.55 ± 3.69	7.49 ± 5.91

Discussion

The femoral shaft receives compressive and bending forces when loading is applied. At this time, buckling occurs in long bones with thin cortex.¹⁷ And buckling of the femoral shaft can proceed to femoral shaft bowing, which is the result of bone remodeling from the bending force. Excess lateral bowing of femur shaft can cause low energy fractures such as AFFs by increasing tensile stress on the lateral femoral cortex.¹⁸ Sasaki et al.¹³ reported that femoral curvature in low energy femoral shaft fractures group is statistically significantly greater than in the age- and gender-matched control group. And other studies have shown that anterolateral femoral bowing is most strongly associated with the occurrence of AFFs compared with other factors such as suppression of bone turnover.^7,10

Various methods have been developed to measure lateral femoral bowing, using radiographs, 2D–3D X-ray scanner (EOS™), or CT scans.^7,13,19 Because femoral bowing seen in plain radiography may be affected by the patient’s position or femoral torsion, femoral bowing on reconstructed CT is more accurate than plain radiographs.¹⁹ However, because of the high amount of radiation and the high cost of CT, plain radiographs are often used. The fact that there are many ways to measure femoral bowing on plain radiographs suggests that one ideal measurement was not revealed.

A good measurement method should have high reliability for proper communication between measurers. Reliability in statistics is the overall consistency of a measure and has to do with the quality of measurement. A measure that are highly reliable are accurate, reproducible, and consistent from one testing occasion to another. In its everyday sense, reliability is the “consistency” or “repeatability” of one’s measures.²⁰

In this study, all five methods consisted of continuous data showed a result of ICC 0.80 or higher. And these methods can be considered as a method of high agreement between measurers. The method with the lowest ICC was Sasaki’s method with ICC 0.881 (95% CI 0.835–0.916), and ICC was the highest in Yau’s method with ICC 0.980 (95% CI 0.971–0.986). The reason for the relatively low interobserver reliability in Sasaki’s method is that the reference points on the proximal and distal parts are ambiguous. On the other hand, the reason for the higher reliability in the Yau’s method is that the reference point for measuring the bowing is relatively accurately specified as the midpoint of the medullary canal in the proximal and distal quadrant. In addition, in the Morin’s method, the midpoint of the endosteal canal at 20% of the femoral shaft was specified as the reference point and showed good reliability with ICC 0.888 (95% CI = 0.808–0.932).

Recently, Park et al. presented their own classification. In their method, the reference line is drawn from the tip of the greater trochanter to the center of the intercondylar notch on AP radiograph. Grade A (nearly straight) indicates that the reference line is located in the middle one-third of the medullary canal at the apex of the curve. Grade B (mild) describes a reference line located in the medial one-third of the medullary canal. Grade C (moderate) refers to a reference line that begins outside of the medullary canal medially and passes through the medial cortex. Grade D (severe) describes a reference line that runs medial to the medial cortex.¹² We did not include the Park’s method, because their classification was categorical variables, unlike the included methods with continuous variables.

There were some limitations in this study. First, the validity value was not obtained because there was no gold standard for lateral bowing of femur. Second, considering the small differences of ICC between the five methods, positioning of the patients during obtaining radiographs might be more significant in measurement of lateral bowing of femur.

Despite these limitations, this is the first study on the reliability of lateral femoral bowing measurement. Among the methods of measuring lateral femoral bowing in the coronal plane of plain radiographs, five methods of continuous value (Sasaki, Morin, Jang, Kim, and Yau) are reliable. And, we recommend Yau’s method, which has a clear reference point for measuring femoral bowing and showed highest reliability.

Footnotes

Acknowledgement

We appreciated that HK, BS, and HJ measured the parameters for this study.

Authors’ contribution

Y-KL and JY contributed equally to this work as co-first authors.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Byung-Woong Jang

References

Schilcher

Michaelsson

Aspenberg

. Bisphosphonate use and atypical fractures of the femoral shaft. N Engl J Med 2011; 364: 1728–1737.

Dell

Adams

Greene

, et al. Incidence of atypical nontraumatic diaphyseal fractures of the femur. J Bone Miner Res 2012; 27: 2544–2550.

Meier

Perneger

Stern

, et al. Increasing occurrence of atypical femoral fractures associated with bisphosphonate use. Arch Intern Med 2012; 172: 930–936.

Mahjoub

Jean

Leclerc

, et al. Incidence and characteristics of atypical femoral fractures: clinical and geometrical data. J Bone Miner Res 2016; 31: 767–776.

Haider

Schneider

Michalski

, et al. Influence of geometry on proximal femoral shaft strains: implications for atypical femoral fracture. Bone 2018; 110: 295–303.

Taormina

Marcano

Karia

, et al. Symptomatic atypical femoral fractures are related to underlying hip geometry. Bone 2014; 63: 1–6.

Morin

Wall

Belzile

, et al. Assessment of femur geometrical parameters using EOS imaging technology in patients with atypical femur fractures; preliminary results. Bone 2016; 83: 184–189.

Chen

Chang

Huang

, et al. The correlation between lateral bowing angle of the femur and the location of atypical femur fractures. Calcif Tissue Int 2014; 95: 240–247.

Huntjens

van Geel

van den Bergh

, et al. Fracture liaison service: impact on subsequent nonvertebral fracture incidence and mortality. J Bone Joint Surg Am 2014; 96: e29.

10.

Jang

Yeo

, et al. Atypical femoral shaft fractures in female bisphosphonate users were associated with an increased anterolateral femoral bow and a thicker lateral cortex: a case–control study. Biomed Res Int 2017; 2017: 5932496.

11.

Kim

Byun

, et al. Factors affecting fracture location in atypical femoral fractures: a cross-sectional study with 147 patients. Injury 2017; 48: 1570–1574.

12.

Park

Song

Zheng

, et al. Intramedullary nailing for atypical femoral fracture with excessive anterolateral bowing. J Bone Joint Surg Am 2017; 99: 72635.

13.

Sasaki

Miyakoshi

Hongo

, et al. Low-energy diaphyseal femoral fractures associated with bisphosphonate use and severe curved femur: a case series. J Bone Miner Metab 2012; 30: 561–567.

14.

Yau

Chiu

Tang

, et al. Coronal bowing of the femur and tibia in Chinese: its incidence and effects on total knee arthroplasty planning. J Orthop Surg (Hong Kong) 2007; 15: 32–36.

15.

McKelvie

. Does memory contaminate test–retest reliability? J Gen Psychol 1992; 119: 59–72.

16.

Landis

Koch

. An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. Biometrics 1977; 33: 363–374.

17.

Demes

. In vivo bone strain and bone functional adaptation. Am J Phys Anthropol 2007; 133: 717–722.

18.

Koh

Goh

Png

, et al. Distribution of atypical fractures and cortical stress lesions in the femur: implications on pathophysiology. Singapore Med J 2011; 52: 77–80.

19.

Akamatsu

Kobayashi

Kusayama

, et al. Femoral shaft bowing in the coronal and sagittal planes on reconstructed computed tomography in women with medial compartment knee osteoarthritis: a comparison with radiograph and its predictive factors. Arch Orthop Trauma Surg 2016; 136: 1227–1232.

20.

Mokkink

Terwee

Patrick

, et al. The COSMIN study reached international consensus on taxonomy, terminology, and definitions of measurement properties for health-related patient-reported outcomes. J Clin Epidemiol 2010; 63: 737–745.