WIXLIB

SICOT-J 2018, 4, 18 The Authors, published by EDP Sciences, e online at:www.sicot-j.orgRESEARCHImageless navigation total hip arthroplasty – an evaluationof operative timeEpaminondas Markos Valsamis1,*, David Ricketts1, Adnan Hussain2, and Amir-Reza Jenabzadeh212Brighton and Sussex University Hospitals, Brighton BN2 5BE, UKHinchingbrooke Hospital NHS Trust, Huntingdon PE29 6NT, UKReceived 11 February 2017, Accepted 15 April 2018, Published online 23 May 2018Abstract -- Introduction: Imageless navigation has been successfully integrated in knee arthroplasty but itseffectiveness in total hip arthroplasty (THA) has been debated. It has consistently been shown that navigationadds significant time and cost to the operation. Further, the relative success of traditional hip replacements hasimpeded the adoption of new techniques.Methods: We compared the operative time between fifty total hip replacements with and without the use ofimageless navigation by a single senior surgeon in a retrospective study. We employed standard statistical toolsto compare the two methods. A correlation-based analysis was used to delimit the “learned” phase of imagelessnavigation to make comparisons meaningful.Results: Contrary to what has previously been reported, there was no significant difference between operativetime in navigated, when compared to traditional operations (p 0.498). Only fourteen operations were requiredto delimit the learning phase of this operation.Discussion: This is the first study that demonstrates no added operative time when using imageless navigationin THA, achieved with an improved workflow. The results also demonstrate a very reasonable learning curve.Key words: Imageless navigation, Hip, Arthroplasty, Operative time.IntroductionTotal hip arthroplasty (THA) is becoming an increasingly popular operation as the prevalence of osteoarthritisis on the rise. Inaccurate implant positioning, particularlyplacement beyond the Lewinnek “safe zone” (outliers), hasbeen implicated in poor long-term postoperative outcome,higher dislocation and revision rates [1–6]. Moderntechnology has developed various operative adjunctshoping to strengthen the long-term efficacy of theseprocedures including imaging-based and imageless navigation systems. Although imageless systems provide lessinformation about skeletal anatomy, their lower cost,radiation hazard, technical effort and time requirementhave made them increasingly favourable when comparedto image-based alternatives [7].Imageless navigation is relatively new technologywhich has been shown to increase accuracy in theplacement of the acetabular component [7–11], and ameta-analysis has demonstrated a significant reduction incup outliers [12]. The Australian joint registry [13] hasshown navigation reduces long term revision rates in THA.*Corresponding author: markosvalsamis@gmail.comDespite these advantages, imageless navigation in hiparthroplasty has been only slowly adopted into orthopaedic practice, potential reasons being its higher cost,prolonged operative time, poor workflow and the generallysatisfactory outcomes that traditional hip replacementsachieve. Moreover, there is a lack of evidence to support animprovement in functional outcome in THA when usingimageless navigation.Studies [7,8,14] have demonstrated up to 58 min addedoperative time when using imageless navigation, thoughless so when compared to CT-navigation. The wide rangeof reported increases in operative time when usingimageless navigation could be the result of variation inoperative speed between surgeons as well as variation intechnique. Furthermore, although navigation can be usedexclusively for acetabular component positioning, othersurgeons may also use it to place the femoral stem andmeasure leg length and offset which would inadvertentlyincrease total operative time and make comparisonsunfair.Previous studies [8,9,15] concluding on operative timeand the learning curve for imageless navigation THA havenot proposed a rigorous method that delimits learningfrom expert phases and therefore the definitions betweenThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0),which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

2E.M. Valsamis et al.: SICOT-J 2018, 4, 18Table 1. Patient demographics. Percentages and ranges are shown for categorical and continuous data respectively in brackets.FactorAge (years)GenderMale (%)Female (%)ASABMIAll (n 72)71.4 (54.5–86.3)–24 (36.1)48 (63.9)2 (1.00–4.00)29.9 (19.0–42.0)Traditional (n 36)70.2 (54.5–86.3)–9 (25.0)27 (75)2 (1.00–3.00)30.0 (19.0–42.0)inexperienced and experienced surgeons (in using navigation) are somewhat vague. Indeed, in evaluating andcomparing any new method researchers should carefullyidentify the onset of expert performance in any statisticalanalysis to make comparisons fair. To this end we areutilising a novel two line learning curve model whichallows learning to be delimited from the learned phase.In this study, we compared the operative times in THAwith and without the use of imageless navigation performedby the same surgeon. A key feature of this study is that wedelineated the learning curve for navigation in terms of theoperative time and determined the point at which thesurgeon had, for all intents and purposes, learned theprocedure. This allowed us to restrict the comparison of thetwo methods to the expert stages of the methods making thecomparison valid and meaningful.Materials and methodsAll operations were undertaken by the consultantsurgeon (ARJ) who was proficient in traditional THA andhad already undertaken 130 traditional hip replacementsas a consultant. He had prior experience in the use ofnavigation for total knee replacements.We analysed the surgeon’s first fifty consecutiveimageless navigation-assisted THAs and determined thenumber of operations required to delimit the learningstage; subsequent operations were defined as “learned”.We additionally reviewed the latest consecutive traditional THA operations by the same surgeon and comparedthe operative time to his “learned” navigated operations.All operations studied were primary hip replacements.No revisions were included.Operative records were used to source informationregarding operative time, age, gender, BMI and ASAgrade. We defined operative time as knife-to-skin untilwound closure.Table 1 shows the patient demographics.Surgical techniqueThe patient is placed in the lateral decubitus position.They have a standard posterior support and an anteriorabdominal support placed between the anterior superioriliac spine and the pubic symphysis pubis. This allowspalpation of the bony point to register the Anterior PelvicPlane. The Aesculap Orthopilot THA 1.0 NavigatedNavigated (n 36)72.6 (60.5–85.8)–15 (41.7)21 (58.3)2 (1.00–4.00)29.2 ical technique was used with a mobile tracker fixed tothe iliac crest. Pelvic orientation was defaulted to theAnterior Pelvic Plane, however in high BMI patientswhere the ASIS is not palpable the Acetabular Centre Axiswas utilised. A cup first technique with a posteriorapproach is utilised as per the surgical technique manualfor THA Universal 1.0. The following implants were used:Aesculap plasmafit cup, Excia T stem, polyethylene insertwith either a ceramic or cobalt- chromium head dependingon patient age.Apart from the use of navigation, the surgicalapproach was unchanged between navigated and traditional operations.Statistical analysisAnalysis of the data was undertaken using XLSTATadd-in software [16] on Microsoft Excel as well as SPSSstatistical software [17]. T-tests were used to comparecontinuous, normally distributed data (age, BMI, ASA)whereas chi-squared tests were used to compare categorical data (gender). A p-value less than 0.05 was consideredstatistically significant.Modelling the learning curveRather than attempting to devise a model thatdescribes a general learning scenario, we applied a simplemethod merely to delimit the learning phase from learnedphase in the time series of operative times. This modellingtechnique was developed and adapted by one of theauthors (EMV) [18].Assuming that the first k points in a series of operationsbelong to the learning phase simply represented by astraight line of the form y mx c while the remainingpoints belong to a plateau y j, we aimed to find k thatminimises a simple convex function of the deviations of thecorrelation coefficients of the two phases from their idealvalues and used this as the ordinal value that delimitslearning from learned. This offers a simple and computationally effective method of identifying the ordinal value inthe time series which separates learning from learned.More specifically we considered points (i, xi) i (1–n),where i is the ordinal value in a series and xi is the operativetime of the ith operation. An optimal way of fitting thesedata to a set of two straight lines (a learning line y m.i c and a plateau y j) is to choose p as the value of i that

E.M. Valsamis et al.: SICOT-J 2018, 4, 183Figure 1. Operative time for the first fifty navigated THAs. Double straight line model superimposed.minimises s(p) where:sðpÞ ¼ ½fðpÞ þ 1 2 þ ½gðpÞ 2 :f(p) is the correlation coefficient for the first p datapoints and g(p) the correlation coefficient of the remainingn–p data points.The rationale behind this method is that an ideallearning phase would produce a correlation coefficient of 1 and an ideal plateau would give a correlationcoefficient equal to zero. Any deviation from these idealvalues contributes to an increase in s(p) and by finding thepoint p which minimizes s(p) we conjointly find a “best”compromise between learning and learned.ResultsFigure 2. Comparison of operative time between imagelessnavigation and traditional operations. The means are notsignificantly different, p 0.498.Operative timeFourteen navigated operations were required to reachthe learning plateau for navigated THAs (Figure 1). Wehence compared operative time between the 36 learnednavigated operations and the 36 latest traditionaloperations. There was no significant difference betweenthe operative time for learned navigated (n 36, mean72.6 min) and traditional operations (n 36, mean74.2 min): t-test, p 0.498 (Figure 2).Age, BMI and ASA gradeAge, BMI and ASA grade did not correlate stronglywith operative time, and p values were not significant(Table 2).DiscussionIn this study, we primarily set out to compare theoperative time between navigated and traditional operations. To do this we delimited the learning phase byconsidering the learning curve for navigated THAs by asingle consultant surgeon to allow meaningful comparisonbetween “learned” operations. Our analysis shows thatfourteen operations were required for the surgeon to reacha plateau in terms of operative time (72.6 min).We did not find a significantly different operative timebetween navigated and traditional THAs. Althoughnavigation improved efficiency by avoiding multiplesequential acetabular reamings and repeated trials in