|Year : 2020 | Volume
| Issue : 2 | Page : 147-152
Evaluating the accuracy of the orthopedic eye
Gerhard Thiart, Marilize Burger, Nando Ferreira
Department of Surgical Sciences, Division of Orthopaedic Surgery, Faculty of Medicine and Health Sciences, Stellenbosch University, Stellenbosch, South Africa
|Date of Submission||17-Jul-2020|
|Date of Decision||05-Nov-2020|
|Date of Acceptance||22-Sep-2020|
|Date of Web Publication||31-Dec-2020|
Dr. Gerhard Thiart
Department of Surgical Sciences, Division of Orthopaedic Surgery, Faculty of Medicine and Health Sciences, Stellenbosch University, Cape Town, 7505, Stellenbosch
Source of Support: None, Conflict of Interest: None
Purpose: Orthopedic surgeries are still dependent on the human factor and more specifically the human eye to gauge the end result. Thus, the planned result and the final surgical result may differ at times. Our hypothesis was that the human eye would not be able to distinguish any angulational difference <3° from the planned trajectory. Materials and Methods: A cross-sectional study in the form of an online survey was conducted. Five clinical scenarios (each with seven variations) that require judgment of angles were recreated. Thirty-five yes or no statements were tested. Results were collected and analyzed statistically. Results: Seventy-four respondents completed the survey. The mean responded age was 42.8 years (standard deviation [SD]: 11.2, range: 28–80). The average of years in practice as a doctor (but not yet a specialist) was 8.3 years (SD: 2.1, range: 4–11), and the median number of years after qualifying as an orthopedic surgeon was 9.0 years (interquartile range, 5.0–17.5). The highest frequencies of inaccuracies occurred around the 1° error margin for all scenarios. Although specialists appeared to score higher (66.6 ± 7.3%) than trainees (62.9 ± 11.4%), this difference was not significantly different (P = 0.107). Conclusion: Orthopedic surgeons can distinguish alignment differences of <3° if the reference framework is perfectly orthogonal and parallel. If the reference framework is rhomboid (skewed) then gauging angles of <3 degrees off it, especially if the angle being gauged is on the same side as one of the acute angles formed by the reference framework.
Keywords: Angle measurement, gauging, judging
|How to cite this article:|
Thiart G, Burger M, Ferreira N. Evaluating the accuracy of the orthopedic eye. J Limb Lengthen Reconstr 2020;6:147-52
| Introduction|| |
Orthopedic operations frequently involve the manual manipulation of two objects in space relative to each other. Examples range from the simple realignment of a transverse humerus fracture, recreating a straight bone, progressing through corrective high tibial osteotomies to complex reconstruction of a malunited tibia fracture that is displaced and angulated in all three anatomical planes. With the advent of digital planning software, surgical correction is planned to within a fraction of degree and millimeter. This precise plan is then executed with the surgical equivalent of a hacksaw, and the final alignment is judged by the surgeon's “gut feel” to lead to the oft-stated remark “this looks right.” The old adage of “measure with a micrometer, mark with chalk, and cut with an axe” comes to mind. Computer-assisted surgery is advancing rapidly in certain fields, but its pickup has been slow and possible scope of application is not widespread. Most surgeries are still dependent on the human factor and more specifically the human eye (in this case, the “orthopedic eye”) to gauge the end result. Thus, the planned result (the science of surgery) and the final surgical result (the art of surgery) may differ at times.
But how good is the orthopedic eye really in perceiving accurate positioning? A “few degrees off” might be acceptable in certain cases, but Paley has demonstrated in his coronal plane deformity planning how much the mechanical axis deviation can increase if the medial proximal tibia angle falls outside the population range of 85-90 degrees. The orthogonal positioning of implants in certain situations is important to the surgical outcome. If the reference ring during the application of a Taylor spatial frame hexapod (Smith and Nephew, Memphis, Tennessee) is not placed orthogonally to the axis of the bone in both the sagittal and coronal planes, the process of postoperative measurement and planning is influenced. Another example is that the malpositioning of reference transverse wires on the ends of an Ilizarov frame might result in an angular deformity of the bone. This is also true for the insertion angle of a tibial intramedullary nail, which must be parallel to the anatomic axis of the tibia.
Studies investigating this were based on either comparing the ability to measure angles manually versus digitally or the ability to measure the same angle consistently using manual or digital methods. There is a paucity in the literature on the accuracy of gauging angles without the aid of calibrated measuring tools.
The aim of this study was to describe how well surgeons (trainees and specialists) could judge angles and deviations of angles, commonly used during orthopedic surgical procedures. Secondary objectives were to (i) investigate whether there are specific differences between trainee orthopedic surgeons and qualified specialists in the ability to judge accurately and (ii) to determine whether there are any relationships between experience and the ability to judge angles accurately.
| Materials and Methods|| |
Five clinical scenarios that require judgment of intra-operative angles were recreated. Radiographs of a left tibia as well as a tibia Sawbones model (Malmö, Sweden) were used. An anteroposterior (AP) radiograph at the level of the knee was used to ensure that accurate measurements were possible (Scenarios 1, 2, and 4). The lateral radiograph was used for the same left tibia in Scenario 3. AP radiographs of a Sawbones model was used to create Scenario 5. Scenarios using radiographs were created using the analysis/measurement tools of the Philips (Koninklijke Philips, Amsterdam, Netherlands) PACS software while the Sawbones scenario was created using the AFGA Healthcare (Mortsel, Belgium) XERO PACS software analysis/measurement tools. All the radiographic pictures used for this study were taken off the PACS system. A cross-sectional study in the form of an online survey (created using Google Forms) was conducted. The survey was distributed via the South African Orthopaedic Association to all its members. Institutional ethics committee approval was obtained prior to collection of any data. As part of the survey, demographic data regarding age, orthopedic subspecialty, years in practice after qualifying as a doctor as well as years after qualifying as an orthopedic surgeon were collected.
Scenario 1 [Figure 1] entailed a straight line drawn superimposed over the proximal tibial joint line (the tibial plateau) on the AP radiograph, with a second line parallel to it at the level of the fibula head. The second line simulated a smooth wire that would normally be placed close to the proximal tibial joint line when treating a tibial plateau fracture with a fine wire circular fixator. This second line was then tilted 1°, 3°, and 5° to either side of neutral (described as “medial side up” or “medial side down”), thus creating in total seven variations of Scenario 1. The same test statement for each of the seven variations of Scenario 1 was used, stating that the second line was parallel to the proximal tibial joint line with “yes” or “no” being the potential answers.
|Figure 1: A line depicting a smooth wire placed parallel with the proximal tibial joint with the different 1°, 3°, and 5° tested variations|
Click here to view
Scenario 2 [Figure 2] entailed the mechanical axis drawn on the AP view of a tibia (connecting the midpoints of the proximal and distal tibial joints). A second line was then drawn, in the region of the proximal third tibial shaft, orthogonal to the drawn mechanical axis. This second line was to simulate smooth wire placement as would routinely be used whilst applying as standard IEF to any tibia. This second line was also tilted 1°, 3°, and 5° (also described as “medial side up” or “medial side down”) to either side of the orthogonal (”neutral”), thus creating in total seven variations of Scenario 2. The same test statement for each of the seven variations of Scenario 2 was used, stating that the second line was orthogonal to the mechanical axis with “yes” or “no” being the potential answers.
|Figure 2: A line depicting a smooth wire placed orthogonal to the anatomic axis of the proximal third tibia shaft with the different 1°, 3°, and 5° tested variations|
Click here to view
Scenario 3 [Figure 3] entailed a line drawn orthogonal to the posterior cortex of the tibial shaft on a lateral X-ray view. This was to simulate half pin placement as during the application of a hexapod circular external fixator. The drawn line was then deviated 1°, 3°, and 5° “anterior up” and “anterior down” from the orthogonal position to create in total seven variations of Scenario 3. The same test statement was used for each of the seven variations of Scenario 3, stating that the line simulating the half pin was orthogonal to the posterior tibial cortex with “yes” or “no” being the potential answers.
|Figure 3: A line depicting a half pin being placed orthogonal to the posterior tibial cortex with the 1°, 3°, and 5° tested variations|
Click here to view
Scenario 4 [Figure 4] attempted to simulate the placement of the entry guide pin in the AP plane of the proximal tibia, as would be done when inserting a tibial nail. A line was drawn parallel to the anatomic axis of the tibia in the AP plane. This line was then tilted 1°, 3°, and 5° to either side of the parallel entry point (described as “hand medial” and “hand lateral”), thus creating in total seven variations of Scenario 4. Again, the same statement was made for all seven variations, with only a “yes” or “no” answer being possible: “Is the line simulating the entry pin parallel with the anatomic axis of the tibia?”
|Figure 4: A line depicting an entry wire placed parallel with the anatomic axis of the proximal tibia shaft with the 1°, 3°, and 5° tested variations|
Click here to view
The last scenario, 5 [Figure 5], was created to see if a deformity of the proximal tibia could be gauged to be accurately corrected after a medial opening wedge osteotomy was performed. A medial sided wedge was cut, and the proximal tibia was correctly positioned to the shaft to the population average medial proximal tibial angle (MPTA) of 87°. The wedge was then closed (undercorrected) by 1°, 3°, and 5° and then opened up (overcorrected) more by 1°, 3°, and 5°. Again, seven variations of a scenario were created, and AP radiographs were taken of each. The same question was asked, with only a “yes” or “no” answer being possible: “Is MPTA restored back to 87° following a medial sided opening wedge osteotomy?”
|Figure 5: A medial sided opening wedge osteotomy restoring the medial proximal tibial angle to 87° with the 1°, 3°, and 5° tested variations|
Click here to view
A sample size calculation, using StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC, was performed using an estimate of 75% of participants being able to accurately gauge the angles, with a precision of 10%. A sample size of 73 participants was considered sufficient in order to have 95% confidence in the results being reflective of the total population. Demographic data were summarized using means ± standard deviations or medians (interquartile ranges, [IQR]), depending on distribution. Categorical data were reported as frequencies and counts. A total of 35 “yes” or “no” statements were tested for each participant. The participants' answers were purely based on observation of the PACS images in the survey. No measurement aids were provided or available. Participants could achieve a maximum correct score of 35. The total score obtained by each participant, as a percentage, was evaluated between different subgroups using an independent t-test. The Pearson's correlation or the Spearman rank correlation was used, depending on distribution, to evaluate whether age or number of years in practice affected accuracy of evaluations. No cutoffs for scores to be regarded as good, fair, or bad were established as there is none to be found in the literature; subsequently, the scores were visually represented using a heat map with quintiles using different shadings to illustrate the scoring patterns.
| Results|| |
A total of 74 respondents completed the survey. The mean responded age was 42.8 ± 11.2 years (range: 28.0–80.0). The mean number of years being in practice as a doctor (but not yet a specialist) was 8.3 ± 2.1 years (range: 4.0–11.0), and the median number of years after qualifying as an orthopedic surgeon was 9.0 (IQR: 5.0–17.5) (range: 1.0–38.0) [Table 1].
The highest frequencies of inaccuracies occurred around the 1° error margin for all scenarios [Table 2]. As the error increased toward the 5° margin, less inaccuracy was found in all scenarios except for Scenario 2, medial side up, and Scenario 5, overcorrection.
Although specialists appeared to score higher (66.6% ± 7.3%) than trainees (62.9% ± 11.4%), this difference was not significantly different (P = 0.107) [Table 3]. Similarly, no difference between the scores of the different subspecialties was observed (P = 0.567).
There was no statistical difference observed between Years Qualified as a Specialist and Total Score (P = 0.995) [Figure 6]a. A weak relationship between the number of years since graduating from medical school and the total score for trainee orthopedic surgeons was observed (R2 = 0.214, P = 0.035) [Figure 6]b.
|Figure 6: Relationship between (a) number of years a specialist and total score and (b) the number of years since graduating with a MBChB degree and total score|
Click here to view
| Discussion|| |
The aim of this study was to describe how well surgeons (trainees and specialists) could judge, using their eyes only, angles, and deviations of angles, commonly used during orthopedic surgical procedures. It is known that inaccuracies can be checked intraoperatively on the image intensifier (II) by either using the built-in digital measurement software (if the system allows for it) or just by manually placing a goniometer on the II screen. Both methods are not foul proof, and there is still a possibility of error by 3.3° (2.5°–4.5°) if using the manual method and of 2.6° (2.3°–33°) if using a digital method. Therefore, it remains paramount that surgeons still aim for perfection in the execution of their work using all the tools at their disposal.
The first important finding from this study was that there was a large amount of inaccuracy and a large variance in answers when the evaluated angle was <3° from the tested “zero” (or neutral). The answers were more consistently correct when the evaluated angle differed by more than 3° from the ideal. When an angle needed to be evaluated, is gauged against a perfectly orthogonal reference framework (perfectly parallel, as in scenario 1, or perpendicular, as in scenario 3), most respondents were able to do this with 77% (57/74) and 83.8% (62/74) of accuracy. In both instances, the variation in accuracy was within 1° of the true “zero” value which would not be clinically significant and therefore negligible. To further illustrate this [Figure 7], the reference framework is square, and angles can accurately be judges against all sides of the square. Thus it is easy to gauge the middle gray line (of the 5 lines) as the one that is perpendicular to the vertical black line.
When the reference framework becomes rhomboid [Figure 8] as in the proximal tibia where the joint line is not perfectly orthogonal to the anatomical axis (MPTA = 87°), the ability to judge angles is compromised – an angular perception “blind spot.” The respondents had difficulty gauging angles correctly when the target angle did not equal a 0° parallel line or a 90° orthogonal line, with only 44.6% (33/74), 37.8% (27/74), and 58.1% (43/74) reporting correctly on Scenarios 2, 4, and 5, respectively. This was especially apparent if both the target angle and the angles being gauged were on the varus side of the tibia, but only if this difference was 3° or less. At 5°, the error was accurately identified. The authors' postulation was that this gauging error was forced by the shape of the tibia.
|Figure 8: Rhomboid reference framework: Gauging becomes difficult, especially on the side of the acute angle. In this example, line C is perfectly orthogonal to line AB|
Click here to view
Evidence suggests that two factors are responsible for this inability to accurately perceive angles. The first is the anatomical properties of the eye itself: the ability of the eye to correctly “capture” graphical information is dependent on anatomic factors such as pupil size and condition of the retina and cornea. The distance and size of an object also interfere with its perception. It is practically impossible to perceive a 1° difference between two 1 cm lines if they are 1 cm apart. As these two lines become longer, the angulation between them becomes more obvious. This would seem to indicate that we require additional information, like distance between the ends of perceived parallel lines, to assist with gauging angles. The second error occurs with our perception (cognition) of reality itself. The eye is not collecting the information incorrectly, but the brain is processing it incorrectly and seen with popular straight-line illusions like the Zollner, Hering, and Poggendorf illusions. An understanding of our angular perception blind spot is important if we are to mitigate its effects when the operative scenario requires it.
A second interesting finding of this study was a weak correlation to support improvement in ability to gauge angles over the progress of a trainee's career, but this “improvement” trend disappeared once qualified as an orthopedic surgeon. A possible explanation for this observation was that trainees' ability to accurately gauge angles improves during their training and reaches its peak potential by the time they qualify as orthopedic surgeons. This finding would therefore support the previously mentioned anatomic and cognitive limitations: humans are unable to become more proficient in gauging angles when the anatomic and cognitive perception limitations have been reached.
The findings of this study did not support any difference in ability to correctly gauge angles between subspecialties, likely because everyone is hampered by the same limitations. In addition, the small sample size of the subgroups did not lend itself for detecting small differences between groups.
To our knowledge, this is the first study conducted on the accuracy of orthopedic surgeons' perception of angles during simulated operative scenarios. The study is limited by the fact that it was only conducted on one anatomic area (the left proximal tibia) and that it was performed on computer screens which might not replicate the true clinical scenario. As this study has potentially identified an angular perception blind spot, similar studies on gauging accuracy around the distal tibia and distal femur, where the anatomical joint congruency angles are 89° and 81°, respectively, should be performed. Larger sample sizes would also be advisable to detect differences between subgroups.
| Conclusion|| |
Orthopedic surgeons can distinguish alignment differences of <3° if the reference framework is perfectly orthogonal and parallel. If the reference framework is rhomboid (skewed) then gauging angles of <3 degrees off it, especially if the angle being gauged is on the same side as one of the acute angles formed by the reference framework.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Paley, Dror. Principles of Deformity Correction. Berlin, Germany. Springer-Verlag Berlin and Heidelberg GmbH & Co. KG. 2014
Ferreira N, Mare P, Marais L. Circular external fixator application for midshaft tibial fractures: Surgical technique. SA Orthopaedic J 2012;11:39-42.
Zelle BA, Boni G. Safe surgical technique: Intramedullary nail fixation of tibial shaft fractures. Patient Saf Surg 2015;9:40.
Shea KG, Stevens PM, Nelson M, Smith JT, Masters KS, Yandow S. A comparison of manual versus computer-assisted radiographic measurement: Intraobserver measurement variability for Cobb angles. Spine 1998;23:551-5.
Morrissy R, Goldsmith G, Hall E, Kehl D, Cowie G. Measurement of the Cobb angle on radiographs of patients who have. J Bone Joint Surg Am 1990;2:320-7.
Bould M, Barnard S, Learmonth ID, Cunningham JL, Hardy JR. Digital image analysis: improving accuracy and reproducibility of radiographic measurement. Clin Biomech (Bristol, Avon) 1999;14:434-7.
Cline D, Hofstetter HW, Griffin JR. Dictionary of Visual Science. Butterworth-Heinemann; 1997.
Judd CH, Courten HC. The Zollner Illusion. The Psychological Review: Monograph Supplements; 1905.
Coren S. Lateral inhibition and the Wundt-Hering illusion. Psychonomic Sci 1970;8:341.
Gillam B. A depth processing theory of the Poggendorff illusion. Percept Psychophysics 1971;10:211-6.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6], [Figure 7], [Figure 8]
[Table 1], [Table 2], [Table 3]