- Email: [email protected]
Inter- and intra-observer reliability of a smartphone application for measuring hallux valgus angles
Foot and Ankle Surgery 19 (2013) 18–21
Contents lists available at SciVerse ScienceDirect
Foot and Ankle Surgery journal homepage: www.elsevier.com/locate/fas
Inter- and intra-observer reliability of a smartphone application for measuring hallux valgus angles Richard Walter FRCS(Orth)a, Jonathan D. Kosy MRCSa,*, Richard Cove FRCS(Orth)b a b
Department of Trauma and Orthopaedics, Torbay Hospital, Lawes Bridge, Torquay TQ2 7AA, UK North Devon District Hospital, Raleigh Park, Barnstaple EX31 4JB, UK
A R T I C L E I N F O
A B S T R A C T
Article history: Received 2 June 2012 Received in revised form 25 July 2012 Accepted 8 August 2012
Background: Measurement of radiological angles can be useful in the planning of the management of patients with hallux valgus. A smartphone application offers an alternative way of measuring these angles in a clinic setting. We compared the reliability (inter- and intra-observer) of this method to the use of PACS. Methods: Radiographs of 30 feet from new patients referred with hallux valgus were examined and angles (HVA, IMA, and DMAA) recorded using the smartphone application and PACS. Results: The smartphone application provided good inter-observer reliability for HVA and IMA (r = 0.93 and r = 0.79 respectively). Intra-observer reliability for HVA and IMA was also found to be good (r = 0.93– 0.97 r = 0.82–0.93 respectively). The inter- and intra-observer reliability for using this method to measure DMAA fell below useful levels (r < 0.60 in each case). Conclusions: This smartphone application provides a reliable method to measure HVA and IMA but we would not recommend it to measure DMAA. ß 2012 European Foot and Ankle Society. Published by Elsevier Ltd. All rights reserved.
Keywords: Radiographic angles Hallux valgus Inter- and intra-observer reliability Accelerometer Smartphone
1. Introduction Various angles can be measured from foot radiographs. In the setting of hallux valgus, these angles may be used to guide appropriate management decisions and to measure postoperative outcomes. Commonly measured angles are the hallux valgus angle (HVA), intermetatarsal angle (IMA) and distal metatarsal articular angle (DMAA). Traditionally, angles were assessed using a goniometer to measure incident angles between axes drawn on radiographs in marker pen. In the past few years, the use of computer workstations to view digital radiographs using picture archive and communication systems (PACS) has become widespread. Consequently, standard practice currently involves measuring angles using digital tools within the PACS software. Previous studies have validated digital measurements to be of equivalent or improved reliability to the traditional goniometer technique [1–4]. More recently, an accelerometer-based angle measuring application (Hallux Angles App Version 1.0, Ockendon.net/WG Healthcare, October 2010) has been developed for the smartphone device. This software uses the inbuilt camera, accelerometer and display unit of the smartphone to allow the user to measure angles from either
traditional or digitally displayed radiographs. It may represent a useful alternative to PACS software angle measurement, for situations in which PACS angle measuring tools are not available, or when the object radiographs are in hard copy rather than digitally displayed format. This study assesses the reliability of the commonly used PACS software angle measuring lines and the newer accelerometerbased smartphone application for the measurement of angles from foot radiographs.
2. Objectives The objective of this study was to compare the inter- and intraobserver reliability between computer-assisted measurements using PACS software, and an accelerometer-based smartphone application, for radiographic angles in hallux valgus (HVA, IMA and DMAA).
3. Methods 3.1. Study design
* Corresponding author. Tel.: +44 01803 414567; fax: +44 01803 654519. E-mail address: [email protected] (J.D. Kosy).
This observational study was conducted at South Devon Healthcare NHS Foundation Trust.
1268-7731/$ – see front matter ß 2012 European Foot and Ankle Society. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.fas.2012.08.004
R. Walter et al. / Foot and Ankle Surgery 19 (2013) 18–21
3.2. Selection of radiographs Radiographs of thirty consecutive patients referred to our institute between January and March 2011 with a provisional diagnosis of hallux valgus were selected. Weight-bearing dorsoplantar foot views were taken using a standardised technique, with the beam centred on the midfoot and a 158 tilt of the X-ray tube towards the ankle. All patient identifying data was removed from radiographs, and radiographs were assessed by raters using Centricity WebPACS (GE Healthcare, Version 3.0) software. 3.3. Angles measured The following three angles were measured (Fig. 1): HVA: The angle between the longitudinal axes of the first metatarsal and proximal phalanx of the first toe. IMA: The angle between the longitudinal axes of the first and second metatarsals. DMAA: The angle between the longitudinal axis of the first metatarsal and a line through the distal articular surface of the first metatarsal.
3.4. Observers Radiographs were assessed by three observers: One orthopaedic surgeon with a specialist interest in foot and ankle surgery
19
Two senior specialty trainees in Trauma and Orthopaedics, both of whom had completed six month training placements in foot and ankle surgery specialist firms.
3.5. Group A: computer-assisted measurements using PACS software The ‘‘Cobb angle’’ measuring tool was used to measure angles, using a standardised technique for determining reference points on the radiographs as outlined by the ad hoc committee of the American Orthopaedic Foot and Ankle Society (AOFAS) on angular measurements [5]. One line of the angle-measuring tool was placed along the axis of the first metatarsal. The other was sequentially placed along the axis of the second metatarsal, the axis of the hallux proximal phalanx, and through the medial and lateral limits of the distal articular surface of the first metatarsal surface, recording the values displayed on screen as IMA, HVA and DMAA respectively. 3.6. Group B: smartphone accelerometer measurements The iPhone Hallux Angles App (Version 1.0, Ockendon.net/WG Healthcare, October 2010) was used with the iPhone 3G (Apple Inc., Cupertino, USA) according to its standard operating instructions to assess IMA, HVA and DMAA (Fig. 2). The software uses the inbuilt camera to project a real time moving image onto the iPhone screen. Observers angle the device until the required axis (such as longitudinal axis of the first metatarsal) is seen to align with a reference line on the iPhone screen. Upon pressing a button, the software measures the angle at which the smartphone is being held using an inbuilt accelerometer, and this information is stored. Once accelerometer data for axes of the first metatarsal, second metatarsal, hallux proximal phalanx and first metatarsal distal articular line has been measured, the software calculates and displays the IMA, HVA and DMAA. Observers used the application to assess a selection of radiographs prior to data collection, to avoid a learning curve effect during the study. 3.7. Collection of data All 3 raters assessed the thirty radiographs with both techniques, each on two separate occasions over a 10-week period: Week 1 – Computer-assisted measurements Week 4 – Smartphone accelerometer measurements
Fig. 1. HVA, IMA and DMAA angles illustrated.
Fig. 2. Image of iPhone showing screenshots of hallux valgus application in use. Produced with permission from WG Healthcare.
R. Walter et al. / Foot and Ankle Surgery 19 (2013) 18–21
20
Table 1 Interobserver reliability measurements (Shrout and Fleiss intraclass correlation coefficients type [1,2] with 95% confidence intervals).
HVA IMA DMAA
PACS measurements
Smartphone measurements
0.97 (0.95–0.99) 0.89 (0.81–0.94) 0.74 (0.58–0.85)
0.93 (0.80–0.97) 0.79 (0.59–0.89) 0.56 (0.36–0.74)
Week 7 – Computer-assisted measurements Week 10 – Smartphone accelerometer measurements
None of the observers had access to their previous measurements or measurements of the other observers. All the measurements were recorded electronically on spread sheets. 3.8. Analysis of data SPSS statistics (Version 20.0.0) was used to calculate Shrout and Fleiss ‘‘case 2’’ single measures type intraclass correlation coefficients with 95% confidence intervals for inter- and intraobserver agreement. 4. Results 4.1. Interobserver reliability Intraclass correlation coefficients and 95% confidence intervals are shown in Table 1. For HVA and IMA measurements, excellent interobserver agreement was seen with both methods. Interobserver agreement was lower for measurement of DMAA using both techniques. For all angles there was a trend towards lower interobserver reliability when using the smartphone accelerometer angle measuring tool. 4.2. Intraobserver reliability Intraclass correlation coefficients (ICC) and 95% confidence intervals are shown in Table 2. For HVA and IMA measurements, excellent intraobserver reliability was seen for both measurement techniques for all observers. Good intraobserver reliability was also seen for all observers measuring DMAA with the PACS angle measuring tool, and for two out of three observers using the smartphone application. 5. Discussion Measuring a three-dimensional structure using two-dimensional images has inherent problems and these exist regardless of Table 2 Intraobserver reliability of measurements (Shrout and Fleiss intraclass correlation coefficients type [1,2] with 95% confidence intervals).
Rater 1 HVA IMA DMAA Rater 2 HVA IMA DMAA Rater 3 HVA IMA DMAA
PACS measurements
Smartphone measurements
0.98 (0.96–0.99) 0.92 (0.85–0.96) 0.90 (0.80–0.95)
0.97 (0.90–0.99) 0.90 (0.79–0.95) 0.83 (0.67–0.91)
0.98 (0.95–0.99) 0.91 (0.82–0.96) 0.75 (0.55–0.87)
0.93 (0.85–0.97) 0.82 (0.66–0.91) 0.70 (0.46–0.85)
0.99 (0.98–0.99) 0.94 (0.87–0.97) 0.80 (0.62–0.90)
0.96 (0.92–0.98) 0.93 (0.86–0.97) 0.56 (0.27–0.76)
the tool used for measurement. In addition, error may be further compounded by poor image quality and image exposure. Despite this, radiographic angles are commonly used to guide decision making regarding appropriate management of hallux valgus. For example, a large HVA is unlikely to be fully corrected by a distal metatarsal osteotomy procedure. Therefore, a more proximal metatarsal osteotomy or arthrodesis procedure may be chosen. Similarly, IMA indicates the severity of the metatarsus primus varus component of deformity. It gives information on the appropriateness of performing a first metatarsal osteotomy, and the level at which this may be required in order to correct deformity. Raised DMAA is often found in congruent hallux valgus deformity for which a first metatarsal procedure combining translation, to decrease IMA, and rotation, to correct DMAA, may be chosen. In addition to these indices, various other radiographic findings may be considered. These include the interphalangeal angle, extent of sesamoid coverage by the metatarsal, presence of degenerative joint disease, and presence of increased joint space seen at the plantar aspect of the first tarsometatarsal joint. Several previous studies have examined inter- and intraobserver reliability for radiographic angle measurements in hallux valgus. Coughlin et al. [6] assessed inter- and intra-observer reliability of radiographic angle measurements using traditional techniques with hard copy radiographs. They found good reliability for HVA and IMA, but poor inter- and intra-observer agreement when measuring DMAA. Other authors have since confirmed poor reliability of DMAA measurements [7,8]. Studies assessing the reliability of using angle-measuring software to determine HVA, IMA and DMAA from digitised foot radiographs have shown equivalent [1] or improved inter- and intra-observer agreement compared to goniometer measurements from traditional radiographs [2–4]. However, such angle measuring software may not always be available, or radiographs may be available in traditional hard copy rather than digitally displayed format. In these conditions there may be a place for a reliable, portable and user-friendly device for measuring angles from foot radiographs. As many surgeons carry a smartphone, use of this application may prevent the need to carry additional equipment (A goniometer). To our knowledge inter- and intra-observer reliability of such a device has never been tested. This study tested the hypothesis that the Hallux Angles application for a smartphone has comparable inter- and intraobserver reliability to PACS software angle measuring tools. To analyse reliability we used intra-class correlation coefficients of the two way mixed single measures type in accordance with published guidance by Shrout and Fleiss [9]. Whilst there is no clear value that represents ‘‘acceptable reliability’’, Chinn [10] recommended ‘‘to be useful a measurement should have an intraclass correlation coefficient of at least 0.6.’’ All of our data for inter- and intraobserver reliability of both HVA and IMA measurements comfortably exceeds this threshold, suggesting that for these two angle measurements the smartphone accelerometer application is reliable. In addition, we have shown these data to be comparable to the current standard (computer assisted measurements using digitally-displayed radiographs). The trend towards slightly inferior inter- and intra-observer reliability, compared to those seen with the use of the PACS software, may be explained by the need to transfer the data onto the smartphone. Logic dictates that any such transfer will inevitably affect accuracy but this effect was seen to be very small. For the measurement of DMAA, the smartphone application was less reliable in terms of both inter- and intra-observer reliability. This is consistent with the findings of previous studies which have questioned the clinical usefulness of DMAA [1–5,7,8]. In addition to concerns regarding the reliability of DMAA measurements, Robinson et al. [11] also
R. Walter et al. / Foot and Ankle Surgery 19 (2013) 18–21
raised concerns over the accuracy of assessing this angle on dorsoplantar radiographs. Using a cadaveric model they demonstrated that both axial rotation into pronation and increasing angle of inclination of the first metatarsal independently influenced the observed DMAA. Changes of both axial rotation and angle of inclination can occur in patients with hallux valgus deformity. Therefore, even if reliable DMAA measurements could be achieved, they might not accurately represent the actual clinical deformity. 6. Conclusions This study provides independent non-manufacturer data that validate the smartphone hallux valgus angles application as a reliable tool for the measurement of HVA and IMA. DMAA measurements are less reliable and therefore not recommended. Use of the PACS measuring tool, where available, seems to be superior but this application provides comparable results and may be used in situations where hard-copies or imported images need to be assessed. Conflict of interests The authors have no conflicts to declare. None of the authors has any association with the designers of the smartphone application or WG Healthcare. No outside funding or support was provided for this work.
21
References [1] Panchbhavi VK, Trevino S. Comparison between manual and computer assisted measurements of hallux valgus parameters. Foot and Ankle International 2004;25:709–11. [2] Farber DC, DeOrio JK, Steel 3rd MW. Goniometric versus computerised angle measurement in assessing hallux valgus. Foot and Ankle International 2005;26:234–8. [3] Pique´-Vidal C, Maled-Garcia I, Arabi-Moreno J, Vila J. Radiographic angles in hallux valgus: differences between measurements made manually and with a computerised program. Foot and Ankle International 2006;27:175–80. [4] Srivastava S, Chockalingam N, El Fakhri T. Radiographic angles in hallux valgus: comparison between manual and computer-assisted measurements. Journal of Foot and Ankle Surgery 2010;49:523–8. [5] Coughlin MJ, Saltzman CL, Nunley JA. Angular measurements in the evaluation of hallux valgus deformities: a report of the ad hoc committee of the American Orthopaedic Foot and Ankle Society on angular measurements. Foot and Ankle International 2002;23:68–74. [6] Coughlin MJ, Freund E, Roger A. The reliability of angular measurements in hallux valgus deformities. Foot and Ankle International 2001;22:369–79. [7] Vittetoe D, Saltzman C, Krieg J, Brown T. Reliability and validity of distal metatarsal articular angle. Foot and Ankle International 1994;15:541–7. [8] Chi TD, Davitt J, Younger A, Holt S, Sangeorzan BJ. Intra- and inter-observer reliability of the distal metatarsal articular angle in hallux valgus. Foot and Ankle International 2002;22:722–6. [9] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychological Bulletin 1979;86:420–8. [10] Chinn S. Repeatability and method comparison. Thorax 1991;46:454–6. [11] Robinson AH, Cullen NP, Chaya NC, Sri-Ram K, Lynch A. Variation of the distal metatarsal articular angle with axial rotation and inclination of the first metatarsal. Foot and Ankle International 2006;27:1036–40.
Contents lists available at SciVerse ScienceDirect
Foot and Ankle Surgery journal homepage: www.elsevier.com/locate/fas
Inter- and intra-observer reliability of a smartphone application for measuring hallux valgus angles Richard Walter FRCS(Orth)a, Jonathan D. Kosy MRCSa,*, Richard Cove FRCS(Orth)b a b
Department of Trauma and Orthopaedics, Torbay Hospital, Lawes Bridge, Torquay TQ2 7AA, UK North Devon District Hospital, Raleigh Park, Barnstaple EX31 4JB, UK
A R T I C L E I N F O
A B S T R A C T
Article history: Received 2 June 2012 Received in revised form 25 July 2012 Accepted 8 August 2012
Background: Measurement of radiological angles can be useful in the planning of the management of patients with hallux valgus. A smartphone application offers an alternative way of measuring these angles in a clinic setting. We compared the reliability (inter- and intra-observer) of this method to the use of PACS. Methods: Radiographs of 30 feet from new patients referred with hallux valgus were examined and angles (HVA, IMA, and DMAA) recorded using the smartphone application and PACS. Results: The smartphone application provided good inter-observer reliability for HVA and IMA (r = 0.93 and r = 0.79 respectively). Intra-observer reliability for HVA and IMA was also found to be good (r = 0.93– 0.97 r = 0.82–0.93 respectively). The inter- and intra-observer reliability for using this method to measure DMAA fell below useful levels (r < 0.60 in each case). Conclusions: This smartphone application provides a reliable method to measure HVA and IMA but we would not recommend it to measure DMAA. ß 2012 European Foot and Ankle Society. Published by Elsevier Ltd. All rights reserved.
Keywords: Radiographic angles Hallux valgus Inter- and intra-observer reliability Accelerometer Smartphone
1. Introduction Various angles can be measured from foot radiographs. In the setting of hallux valgus, these angles may be used to guide appropriate management decisions and to measure postoperative outcomes. Commonly measured angles are the hallux valgus angle (HVA), intermetatarsal angle (IMA) and distal metatarsal articular angle (DMAA). Traditionally, angles were assessed using a goniometer to measure incident angles between axes drawn on radiographs in marker pen. In the past few years, the use of computer workstations to view digital radiographs using picture archive and communication systems (PACS) has become widespread. Consequently, standard practice currently involves measuring angles using digital tools within the PACS software. Previous studies have validated digital measurements to be of equivalent or improved reliability to the traditional goniometer technique [1–4]. More recently, an accelerometer-based angle measuring application (Hallux Angles App Version 1.0, Ockendon.net/WG Healthcare, October 2010) has been developed for the smartphone device. This software uses the inbuilt camera, accelerometer and display unit of the smartphone to allow the user to measure angles from either
traditional or digitally displayed radiographs. It may represent a useful alternative to PACS software angle measurement, for situations in which PACS angle measuring tools are not available, or when the object radiographs are in hard copy rather than digitally displayed format. This study assesses the reliability of the commonly used PACS software angle measuring lines and the newer accelerometerbased smartphone application for the measurement of angles from foot radiographs.
2. Objectives The objective of this study was to compare the inter- and intraobserver reliability between computer-assisted measurements using PACS software, and an accelerometer-based smartphone application, for radiographic angles in hallux valgus (HVA, IMA and DMAA).
3. Methods 3.1. Study design
* Corresponding author. Tel.: +44 01803 414567; fax: +44 01803 654519. E-mail address: [email protected] (J.D. Kosy).
This observational study was conducted at South Devon Healthcare NHS Foundation Trust.
1268-7731/$ – see front matter ß 2012 European Foot and Ankle Society. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.fas.2012.08.004
R. Walter et al. / Foot and Ankle Surgery 19 (2013) 18–21
3.2. Selection of radiographs Radiographs of thirty consecutive patients referred to our institute between January and March 2011 with a provisional diagnosis of hallux valgus were selected. Weight-bearing dorsoplantar foot views were taken using a standardised technique, with the beam centred on the midfoot and a 158 tilt of the X-ray tube towards the ankle. All patient identifying data was removed from radiographs, and radiographs were assessed by raters using Centricity WebPACS (GE Healthcare, Version 3.0) software. 3.3. Angles measured The following three angles were measured (Fig. 1): HVA: The angle between the longitudinal axes of the first metatarsal and proximal phalanx of the first toe. IMA: The angle between the longitudinal axes of the first and second metatarsals. DMAA: The angle between the longitudinal axis of the first metatarsal and a line through the distal articular surface of the first metatarsal.
3.4. Observers Radiographs were assessed by three observers: One orthopaedic surgeon with a specialist interest in foot and ankle surgery
19
Two senior specialty trainees in Trauma and Orthopaedics, both of whom had completed six month training placements in foot and ankle surgery specialist firms.
3.5. Group A: computer-assisted measurements using PACS software The ‘‘Cobb angle’’ measuring tool was used to measure angles, using a standardised technique for determining reference points on the radiographs as outlined by the ad hoc committee of the American Orthopaedic Foot and Ankle Society (AOFAS) on angular measurements [5]. One line of the angle-measuring tool was placed along the axis of the first metatarsal. The other was sequentially placed along the axis of the second metatarsal, the axis of the hallux proximal phalanx, and through the medial and lateral limits of the distal articular surface of the first metatarsal surface, recording the values displayed on screen as IMA, HVA and DMAA respectively. 3.6. Group B: smartphone accelerometer measurements The iPhone Hallux Angles App (Version 1.0, Ockendon.net/WG Healthcare, October 2010) was used with the iPhone 3G (Apple Inc., Cupertino, USA) according to its standard operating instructions to assess IMA, HVA and DMAA (Fig. 2). The software uses the inbuilt camera to project a real time moving image onto the iPhone screen. Observers angle the device until the required axis (such as longitudinal axis of the first metatarsal) is seen to align with a reference line on the iPhone screen. Upon pressing a button, the software measures the angle at which the smartphone is being held using an inbuilt accelerometer, and this information is stored. Once accelerometer data for axes of the first metatarsal, second metatarsal, hallux proximal phalanx and first metatarsal distal articular line has been measured, the software calculates and displays the IMA, HVA and DMAA. Observers used the application to assess a selection of radiographs prior to data collection, to avoid a learning curve effect during the study. 3.7. Collection of data All 3 raters assessed the thirty radiographs with both techniques, each on two separate occasions over a 10-week period: Week 1 – Computer-assisted measurements Week 4 – Smartphone accelerometer measurements
Fig. 1. HVA, IMA and DMAA angles illustrated.
Fig. 2. Image of iPhone showing screenshots of hallux valgus application in use. Produced with permission from WG Healthcare.
R. Walter et al. / Foot and Ankle Surgery 19 (2013) 18–21
20
Table 1 Interobserver reliability measurements (Shrout and Fleiss intraclass correlation coefficients type [1,2] with 95% confidence intervals).
HVA IMA DMAA
PACS measurements
Smartphone measurements
0.97 (0.95–0.99) 0.89 (0.81–0.94) 0.74 (0.58–0.85)
0.93 (0.80–0.97) 0.79 (0.59–0.89) 0.56 (0.36–0.74)
Week 7 – Computer-assisted measurements Week 10 – Smartphone accelerometer measurements
None of the observers had access to their previous measurements or measurements of the other observers. All the measurements were recorded electronically on spread sheets. 3.8. Analysis of data SPSS statistics (Version 20.0.0) was used to calculate Shrout and Fleiss ‘‘case 2’’ single measures type intraclass correlation coefficients with 95% confidence intervals for inter- and intraobserver agreement. 4. Results 4.1. Interobserver reliability Intraclass correlation coefficients and 95% confidence intervals are shown in Table 1. For HVA and IMA measurements, excellent interobserver agreement was seen with both methods. Interobserver agreement was lower for measurement of DMAA using both techniques. For all angles there was a trend towards lower interobserver reliability when using the smartphone accelerometer angle measuring tool. 4.2. Intraobserver reliability Intraclass correlation coefficients (ICC) and 95% confidence intervals are shown in Table 2. For HVA and IMA measurements, excellent intraobserver reliability was seen for both measurement techniques for all observers. Good intraobserver reliability was also seen for all observers measuring DMAA with the PACS angle measuring tool, and for two out of three observers using the smartphone application. 5. Discussion Measuring a three-dimensional structure using two-dimensional images has inherent problems and these exist regardless of Table 2 Intraobserver reliability of measurements (Shrout and Fleiss intraclass correlation coefficients type [1,2] with 95% confidence intervals).
Rater 1 HVA IMA DMAA Rater 2 HVA IMA DMAA Rater 3 HVA IMA DMAA
PACS measurements
Smartphone measurements
0.98 (0.96–0.99) 0.92 (0.85–0.96) 0.90 (0.80–0.95)
0.97 (0.90–0.99) 0.90 (0.79–0.95) 0.83 (0.67–0.91)
0.98 (0.95–0.99) 0.91 (0.82–0.96) 0.75 (0.55–0.87)
0.93 (0.85–0.97) 0.82 (0.66–0.91) 0.70 (0.46–0.85)
0.99 (0.98–0.99) 0.94 (0.87–0.97) 0.80 (0.62–0.90)
0.96 (0.92–0.98) 0.93 (0.86–0.97) 0.56 (0.27–0.76)
the tool used for measurement. In addition, error may be further compounded by poor image quality and image exposure. Despite this, radiographic angles are commonly used to guide decision making regarding appropriate management of hallux valgus. For example, a large HVA is unlikely to be fully corrected by a distal metatarsal osteotomy procedure. Therefore, a more proximal metatarsal osteotomy or arthrodesis procedure may be chosen. Similarly, IMA indicates the severity of the metatarsus primus varus component of deformity. It gives information on the appropriateness of performing a first metatarsal osteotomy, and the level at which this may be required in order to correct deformity. Raised DMAA is often found in congruent hallux valgus deformity for which a first metatarsal procedure combining translation, to decrease IMA, and rotation, to correct DMAA, may be chosen. In addition to these indices, various other radiographic findings may be considered. These include the interphalangeal angle, extent of sesamoid coverage by the metatarsal, presence of degenerative joint disease, and presence of increased joint space seen at the plantar aspect of the first tarsometatarsal joint. Several previous studies have examined inter- and intraobserver reliability for radiographic angle measurements in hallux valgus. Coughlin et al. [6] assessed inter- and intra-observer reliability of radiographic angle measurements using traditional techniques with hard copy radiographs. They found good reliability for HVA and IMA, but poor inter- and intra-observer agreement when measuring DMAA. Other authors have since confirmed poor reliability of DMAA measurements [7,8]. Studies assessing the reliability of using angle-measuring software to determine HVA, IMA and DMAA from digitised foot radiographs have shown equivalent [1] or improved inter- and intra-observer agreement compared to goniometer measurements from traditional radiographs [2–4]. However, such angle measuring software may not always be available, or radiographs may be available in traditional hard copy rather than digitally displayed format. In these conditions there may be a place for a reliable, portable and user-friendly device for measuring angles from foot radiographs. As many surgeons carry a smartphone, use of this application may prevent the need to carry additional equipment (A goniometer). To our knowledge inter- and intra-observer reliability of such a device has never been tested. This study tested the hypothesis that the Hallux Angles application for a smartphone has comparable inter- and intraobserver reliability to PACS software angle measuring tools. To analyse reliability we used intra-class correlation coefficients of the two way mixed single measures type in accordance with published guidance by Shrout and Fleiss [9]. Whilst there is no clear value that represents ‘‘acceptable reliability’’, Chinn [10] recommended ‘‘to be useful a measurement should have an intraclass correlation coefficient of at least 0.6.’’ All of our data for inter- and intraobserver reliability of both HVA and IMA measurements comfortably exceeds this threshold, suggesting that for these two angle measurements the smartphone accelerometer application is reliable. In addition, we have shown these data to be comparable to the current standard (computer assisted measurements using digitally-displayed radiographs). The trend towards slightly inferior inter- and intra-observer reliability, compared to those seen with the use of the PACS software, may be explained by the need to transfer the data onto the smartphone. Logic dictates that any such transfer will inevitably affect accuracy but this effect was seen to be very small. For the measurement of DMAA, the smartphone application was less reliable in terms of both inter- and intra-observer reliability. This is consistent with the findings of previous studies which have questioned the clinical usefulness of DMAA [1–5,7,8]. In addition to concerns regarding the reliability of DMAA measurements, Robinson et al. [11] also
R. Walter et al. / Foot and Ankle Surgery 19 (2013) 18–21
raised concerns over the accuracy of assessing this angle on dorsoplantar radiographs. Using a cadaveric model they demonstrated that both axial rotation into pronation and increasing angle of inclination of the first metatarsal independently influenced the observed DMAA. Changes of both axial rotation and angle of inclination can occur in patients with hallux valgus deformity. Therefore, even if reliable DMAA measurements could be achieved, they might not accurately represent the actual clinical deformity. 6. Conclusions This study provides independent non-manufacturer data that validate the smartphone hallux valgus angles application as a reliable tool for the measurement of HVA and IMA. DMAA measurements are less reliable and therefore not recommended. Use of the PACS measuring tool, where available, seems to be superior but this application provides comparable results and may be used in situations where hard-copies or imported images need to be assessed. Conflict of interests The authors have no conflicts to declare. None of the authors has any association with the designers of the smartphone application or WG Healthcare. No outside funding or support was provided for this work.
21
References [1] Panchbhavi VK, Trevino S. Comparison between manual and computer assisted measurements of hallux valgus parameters. Foot and Ankle International 2004;25:709–11. [2] Farber DC, DeOrio JK, Steel 3rd MW. Goniometric versus computerised angle measurement in assessing hallux valgus. Foot and Ankle International 2005;26:234–8. [3] Pique´-Vidal C, Maled-Garcia I, Arabi-Moreno J, Vila J. Radiographic angles in hallux valgus: differences between measurements made manually and with a computerised program. Foot and Ankle International 2006;27:175–80. [4] Srivastava S, Chockalingam N, El Fakhri T. Radiographic angles in hallux valgus: comparison between manual and computer-assisted measurements. Journal of Foot and Ankle Surgery 2010;49:523–8. [5] Coughlin MJ, Saltzman CL, Nunley JA. Angular measurements in the evaluation of hallux valgus deformities: a report of the ad hoc committee of the American Orthopaedic Foot and Ankle Society on angular measurements. Foot and Ankle International 2002;23:68–74. [6] Coughlin MJ, Freund E, Roger A. The reliability of angular measurements in hallux valgus deformities. Foot and Ankle International 2001;22:369–79. [7] Vittetoe D, Saltzman C, Krieg J, Brown T. Reliability and validity of distal metatarsal articular angle. Foot and Ankle International 1994;15:541–7. [8] Chi TD, Davitt J, Younger A, Holt S, Sangeorzan BJ. Intra- and inter-observer reliability of the distal metatarsal articular angle in hallux valgus. Foot and Ankle International 2002;22:722–6. [9] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychological Bulletin 1979;86:420–8. [10] Chinn S. Repeatability and method comparison. Thorax 1991;46:454–6. [11] Robinson AH, Cullen NP, Chaya NC, Sri-Ram K, Lynch A. Variation of the distal metatarsal articular angle with axial rotation and inclination of the first metatarsal. Foot and Ankle International 2006;27:1036–40.