Multi-factorial analysis of time efficiency in total knee arthroplasty

Abstract

The objective of this retrospective review was to determine whether time efficiency could be gained by optimizing the navigation protocol based on a surgeon's specific technique and work flow. Three groups of 30 consecutive patients operated on by the same surgeon were studied. The groups were from three distinct periods between 2002 and 2008. The first group consisted of patients in which no navigation was performed (the control group); the second group consisted of patients in which navigation was performed using a standard protocol; and the third group consisted of patients in which navigation was performed using a customized protocol that eliminated certain steps. Statistical analysis considered analysis of variance for covariates of total time in the operating room, duration of the procedure, and tourniquet time. Chi-square analysis considered categorical variables of age, gender, body mass index (BMI), Knee Society score, and patellar resurfacing against the surgical group. Multiple linear regression analysis evaluated surgical time adjusted for preoperative deformity, BMI, and patellar resurfacing. In the adjusted model, the customized navigation protocol significantly reduced the surgical time by 10 minutes compared to the non-customized navigation. Not resurfacing the patella significantly reduced the surgical time by 9 minutes. Variables of age, gender, BMI, preoperative deformity and Knee Society score were not related to differences in operating time. Time efficiency may be gained with the use of improved computer navigation protocols and patellar non-resurfacing. [Level of evidence: Level 3.]

• Computer navigation
• patellar resurfacing
• total knee arthroplasy
• tourniquet
• time-motion

Introduction

In modern surgery, the duration of the operative procedure has been an important consideration for the outcome of the surgical intervention. Longer surgical times have been associated with increased surgical morbidity, complications and costs. While surgical expertise, the method used, and the desired outcome are all significant issues, numerous confounding factors remain beyond the control of the surgeon, such as a patient's chronic obesity, prior surgery or trauma, and chronic untreated deformities.

Computer navigation has evolved as a method to improve surgical outcomes by increasing the precision and accuracy of the surgical technique, with the possibility of eliminating outliers that may lead to earlier failure. While navigated mechanical axis alignment in total knee arthroplasty (TKA) is improved by the use of navigation, most studies have shown additional time requirements, with longer tourniquet and total surgery times [1–13]. As computer navigation requires the fixation of tracking arrays on the femur and tibia using bone pins and a complex referencing protocol performed by the surgeon, it is not surprising that additional time is required. However, one study demonstrated that, after an initial learning curve, the surgeons were able to improve their surgical technique to achieve a “time neutral” procedure, despite the additional navigation steps [14]. The surgeons were more confident with their ligament releases, and felt that the need for additional bone cuts was diminished.

A recent meta-analysis of computer navigation (13) in TKA in a large number of cases indicated a 23% increase in surgical time with the use of navigation (13). However, the level of the surgeons’ skill or experience and the overall level of sophistication of the software protocol were not known. For example, many early computer systems offered complex capabilities such as surface morphing and CAD model implant sizing. While these methods appear to be helpful for improving surgical precision, they may not be the most important elements in assisting a particular surgeon with his technique: The steps of the computer protocol may represent a “one size fits all” approach that provides unnecessary information.

Two questions need to be addressed: (1) Can the computer navigation surgical protocol be streamlined for a particular surgeon to meet his specific needs?; and (2) Given the other clinical problems typically encountered with a cohort of TKA patients, can this modified protocol be shown to reduce the overall surgical time required? To investigate these questions, a preliminary time-motion study was undertaken of three distinct periods in a single surgeon's practice. These comprised a pre-navigation period; a period of initial experience with a universal navigation system designed for the standard TKA procedure; and a period of experience with a custom navigation system optimized to support just those steps required by that particular surgeon in the exact sequence that he specified. Other factors were also assessed, including the patient demographics, knee deformity, patellar resurfacing, and the preoperative Knee Society pain and function scores.

Methods

This retrospective case review considers three distinct time periods from 2002 to 2008 in the practice of a single surgeon. For each period, thirty consecutive patients were selected who had received the same implant, namely the NexGen LPS High Flex prosthesis (Zimmer, Inc., Warsaw, IN). The first group of patients was collected from January to June 2002 and consisted of patients in which no navigation was performed during the TKA procedure (the Pre-navigation control group); the second group was collected from October 2003 to April 2004 and consisted of patients in which navigation was performed using a standard protocol (by Medtronic); and the third group was collected from August 2007 to January 2008 and consisted of patients in which navigation was performed using a customized protocol (by PRAXIM) that eliminated certain steps. The surgical approach was medial parapatellar for all varus knees and lateral for all valgus knees, and the tibial cut first technique was used in all cases. For each patient, demographic data included age, sex, BMI, preoperative deformity, and preoperative Knee Society pain and function scores. The specific prosthetic size was recorded, along with whether the patella was resurfaced or not. In the first and second groups, 33% had non-resurfaced patellas, while in the third group 80% were non-resurfaced, reflecting the changing technique of the surgeon over time. In the last group, patellar resurfacing was not performed if satisfactory cartilage remained, the patient was younger, or the diagnosis was osteoarthritis.

For the time-motion study, time in and out of the operating room was recorded. Surgery time was the time from the initial skin incision until placement of the final dressing, at which point sterility of the surgical field was broken. Total tourniquet time was measured from when the tourniquet was first elevated to the time that it was finally deflated. For the technique used, the deflation time coincided with the complete insertion of the implant with bone cement and the final navigation of the prosthetic insertion.

The navigation systems used in this study were the Medtronic Universal Total Knee application (Medtronic, Inc., Louisville, CO) and the PRAXIM Universal Total Knee application (PRAXIM medivision S.A., La Tronche, France). Both systems required placement of femoral and tibial tracking arrays. The Medtronic software protocol was the earliest version, requiring registration with the hip kinematic reference protocol and 15 touch points including the femoral center, lateral epicondyle, medial epicondyle, distal lateral condyle, distal medial condyle, posterior lateral condyle, posterior medial condyle, anterior-posterior axis of Whiteside, tibial center, tibial anterior-posterior axis, medial tibial plateau, lateral tibial plateau, tibial tubercle, medial malleolus and lateral malleolus. The order of cuts for this protocol was (1) distal femoral, (2) anterior-posterior distal femoral, and (3) proximal tibial. When using this system, the surgeon was required to have an assistant select from the screen the desired steps, as they were “out of order” for the surgeon's technique. Specifically, the order for this surgeon was (1) proximal tibial, (2) anterior-posterior femoral, and (3) distal femoral.

The PRAXIM software protocol was a “custom”, surgeon-specific scheme that required registration with the hip kinematic reference protocol and 11 touch points including the femoral center, anterior-posterior axis of Whiteside, distal medial femoral condlyle, distal lateral femoral condyle, anterior distal femoral “morphing” zone, tibial center, tibial anterior-posterior axis, medial tibial plateau, lateral tibial plateau, medial malleolus and lateral malleolus. The order of cuts was (1) proximal tibial, (2) anterior-posterior distal femoral, and (3) distal femoral. The surgeon advanced the screen steps with a floor pedal and they were in the appropriate order for this surgeon's technique.

Statistical analysis

Descriptive statistics were performed for each group (Table I). A Chi-square test was performed to assess the relationship between categorical variables (gender and patellar resurfacing) and the surgical navigation type, while ANOVA was used for the continuous variables (Knee Society score, age and BMI) (Table III). Analysis of Variance was done to compare total surgical time, total operating room time, and tourniquet time for each type of navigation. For variables with a significant F-test, pairwise comparisons were done using a Tukey's HSD adjustment to compare differences (Table II). A multiple linear regression model was created with predictors of navigation type, patellar resurfacing, BMI, and deformity against surgical time as the outcome (Table IV). A regression analysis of chronological changes was performed within each navigation group, looking for a learning-curve effect.

Table I.  Comparison of means for total time in the operating room, surgical procedure time, and tourniquet time.

Navigation group	Total time (min)	Surgery time (min)	Tourniquet time (min)
Pre-navigation	131 ± 21	94 ± 17	64 ± 12
Medtronic	139 ± 21	101 ± 15	75 ± 12
PRAXIM custom	126 ± 15	86 ± 12	60 ± 9

Table II.  Results of ANOVA and pair-wise comparison using Tukey's adjustment for significant differences between surgical navigation groups: Difference in means (95% simultaneous confidence interval; ^**denotes significant result).

	p-value for equality among groups	Medtronic vs. pre-navigation	Medtronic vs. PRAXIM	Pre-navigation vs. PRAXIM
Total time	0.031	8.43 (−3.57, 20.44)	13.44^** (1.40, 25.40)	4.97 (−7.04, 16.97)
Surgical time	0.002	7.63 (−1.84, 17.10)	14.87^** (5.40, 24.34)	7.23 (−2.24, 16.70)
Tourniquet time	<0.001	11.00^** (4.00, 18.00)	14.60^** (7.61, 21.60)	3.60 (−3.40, 10.60)

Table III.  Chi-square analysis is performed to test associations between categorical variables and surgical navigation type, and the F-test is used to compare mean values for the three navigation groups for continuous variables.

Variable	p-value	Pre-navigation mean (SD)	PRAXIM mean (SD)	Medtronic mean (SD)
Age	F2.87 = 1.20, p = 0.31	64.5 (9.9)	67.2 (8.6)	68.1 (9.6)
BMI	F2.87 = 1.23, p = 0.30	31.1 (6.3)	32.9 (8.0)	30.2 (5.9)
Knee score	F2.86 = 0.48, p = 0.62	41.5 (7.9)	42.3 (17.3)	39.3 (8.8)
Gender	X = 0.0, p = 1.00	17/30 female	17/30 female	17/30 female
Patellar resurfacing	X = 24.3, p < 0.01	9/30 non-resurfaced	26/30 non-resurfaced	10/30 non-resurfaced

Table IV.  Multiple linear regression analysis with surgical time as outcome, adjusting for patellar resurfacing, BMI, deformity category and navigation type.

Variable	Estimate (Std. error)	F-value	p-value
Patellar resurfacing:		16.53	<0.001
No	−9.13 (3.79)
Yes	0.00 (reference)
BMI	0.14 (0.25)	0.13	0.72
Deformity group:		0.82	0.54
Valgus, 0–4°	2.45 (9.66)
Varus, 0–4°	−5.96 (4.69)
Valgus, 5–9°	2.51 (7.24)
Varus, 5–9°	0.00 (reference)
Valgus, 10+°	−0.18 (5.84)
Varus, 10+°	−6.26 (6.28)
Navigation type:		3.58	0.54
Medtronic	8.09 (3.98)
PRAXIM	−2.71 (4.65)
Pre-navigation	0.00 (reference)

Results

The study found that modification and streamlining of a surgical navigation protocol (as was done with the PRAXIM protocol) could significantly reduce total operating room time, surgical time and tourniquet time, as compared to a standard navigation protocol (in this case the Medtronic protocol) (Table II). The three navigation groups were similar with respect to age, BMI, gender, and Knee Society score distribution; however, the incidence of patellar resurfacing varied (Table III). The estimated differences shown in Table II could be biased if patellar resurfacing affects surgical time. A linear regression analysis adjusting for all the potential confounding factors found that the mean difference between the PRAXIM and Medtronic navigated groups was still significant, though the estimated difference was reduced to 10 minutes (Table IV). This would confirm the hypothesis that improvements in the navigation software protocol could improve the surgical flow and efficiency of the standard surgical procedure

Cases in which patellar resurfacing was not performed were shown to have significantly shorter surgical time compared to those including resurfacing by a mean of 9 minutes after controlling for all other variables (Table IV). With the linear regression model, neither BMI nor degree of preoperative deformity were shown to have a significant effect on surgical time. In other words, severe obesity or deformity did not increase surgical time. Severity of deformity was not a factor, as the most severe varus knees with over 10° of deformity took six minutes less than the reference group, and severe valgus knees were time-neutral compared to the reference group (Table IV). We found no time changes within either navigation group that might suggest a learning-curve effect.

Discussion

Computer navigation in total knee arthroplasty has been shown to significantly improve overall mechanical axis alignment and reduce outliers. Most studies to date have demonstrated that computer navigation adds significant time to the procedure. The goal of this retrospective review was to determine whether that extra surgical time could be reduced by streamlining the navigation protocol while accomplishing the same objectives. An efficient custom protocol was developed that used only the basic steps required to satisfactorily balance ligaments and make basic bone cuts for restoration of anatomical alignment.

The current study is limited and must be considered preliminary as a number of factors were not controlled for by the study design or protocol randomization. Based on a review of the literature, it was not known at the outset whether some of these factors, such as BMI, degree of deformity, patellar resurfacing or preoperative Knee Society score, would also be shown to influence the efficiency of the procedure. Importantly, the two navigated groups in this study represent different periods in the surgeon's experience: The Medtronic protocol group could be considered to have been operated during the learning curve, while the PRAXIM group was only initiated after the surgeon had already performed over 300 navigated cases. Inter-observer reproducibility was not considered in this study, as only one surgeon's cases were analyzed.

We found that the surgical time required to perform TKA with surgical navigation could be reduced by 10 minutes using a software protocol that was predetermined to match a surgeon's specific surgical technique and optimized to eliminate steps that were considered non-essential by that surgeon. There did not appear to be a learning curve with either navigation group, as chronological assessment revealed no significant improvement in times. After reviewing the literature, one must presume that most reported studies were conducted using standard protocols for the navigation systems employed, which would involve a prescribed sequence of steps that the surgeon must follow. A summary of recent papers presenting surgical durations indicates that this leads to an increase in total operating time on the order of 14 minutes (Table V). Most authors identify this factor as one of the liabilities of the navigation technology.

Table V.  Published studies evaluating mean computer navigation and conventional surgery times in total knee arthroplasty.

	Mean surgery time (min)
Study	Conventional	Computer navigated	SD
Bäthis et al., 2004 [2]	64	78	P < 0.01
Victor et al., 2004 [13]	73.5	93	P < 0.0001
Chauhan et al., 2004 [4]	67	80	P < 0.001
Duckling et al., 2005 [5]	79	92	P < 0.001
Jenny et al., 2005 [8]	99	108	P < 0.01
Seon et al., 2006 [11]	92	97	NSD
Kim et al., 2007 [9]	82	97	P < 0.001
Bauwens et al., 2007 [15]	73	90	P < 0.001
Dutton et al., 2008 [6]	83	107	P < 0.001
Summary	81 ± 12	95 ± 11

From statistical analysis, we found patellar resurfacing or non-resurfacing to be the only other factor influencing time considerations. The patellar resurfacing step clearly increased the surgical time on the order of nine minutes. This would include the steps for cutting the surface, drilling holes and trial fitting of the patellar prosthesis, as well as those few moments needed to insert the final implant. While patellar resurfacing remains a controversial topic, with the majority of North American authors favoring resurfacing using contemporary techniques, the authors have followed the recommendations of others who believe that non-resurfacing of the patella is a viable option for patients with preserved cartilage on the patellar articular surface and a prosthesis considered to be anatomical and favorable for articulation of the native patella [16–20].

No significant trends could be identified with respect to patient obesity, as measured by BMI, or preoperative deformity. In fact, the highest deformity groups with greater than 10° of varus or valgus had nominally lower mean surgical times than those with average deformity. Morbid obesity has been suggested to cause wound-healing problems, reduced postoperative range of motion, and slower recovery, but no other study has evaluated the effect on surgical operating time. This study could find no trend toward longer surgical time or tourniquet times with patients noted to be morbidly obese.

For computer assisted navigation of TKA, the obvious recommendation would include software refinements that allow each surgeon to easily customize his surgical technique for efficiency and avoid certain steps in the protocol that may have limited benefit for the needed precision. For example, determining the femoral component rotation lacks precision with most systems, and this is related to the challenges of point referencing with image-free methods [19]. While this problem may be obviated with more sophisticated imaging modalities such as computed tomography, the added complexity and expense then become greater determining factors.

In conclusion, this preliminary study has shown that surgical and tourniquet time are not adversely affected by computer navigation if the method is time-efficient and the surgeon has confidence in the computer guidance. Complicating factors such as patients with morbid obesity or severe angular deformities do not necessarily require more surgical time in experienced hands; however, routine resurfacing of the patella will add time for the standard arthroplasty procedure.