Utilization and Short-Term Outcomes of Computer Navigation in Unicompartmental Knee Arthroplasty

Abstract

Background:

The use of navigation remains a controversial topic in knee arthroplasty. The purpose of this study is to evaluate current rates of utilization of navigation in unicompartmental knee arthroplasty (UKA) in the United States, as well as the incidence of short-term complications and operative times between navigated and non-navigated UKA.

Methods:

A query of the National Surgical Quality Improvement Project (NSQIP) database was used to identify cases of primary UKA during years 2006-2017. Additional common procedural terminology (CPT) codes were used to identify cases in which navigation was utilized. Operative time, length of stay, and short-term outcomes were compared. Propensity score matching was used to minimize differences in demographics and comorbidities between the navigation and non-navigation cohorts.

Results:

A total of 10,586 cases of UKA were identified; 343 of these cases (3.2%) utilized navigation. The unadjusted rate of any complication for the entire cohort was 3.6%. Navigated UKA had mean operative times 8 minutes longer than non-navigated UKA (92.1 min vs. 84.3 min; p<0.001). There was no difference in overall complication rates between the matched navigated (3.5%) and non-navigated (3.2%) cohorts (p=0.65). There was no difference in rates of readmission (0.31% vs. 0.58%; p=0.31), reoperation (0.29% vs. 0.29%; p=1.00), and mean length of stay (1.3 ± 1.6 days vs. 1.2 ± 1.9 days; p=0.15).

Conclusion:

UKA utilizing navigation had a mean operative time 8 minutes longer than non-navigated UKA. We found no difference in rates of short-term complications, readmission, reoperation, or mean length of stay between navigated and non-navigated UKA.

Level of Evidence: III

Introduction

The use of console-based and hand-held navigation remains a controversial topic in knee reconstruction.^1-4 Since the technology became available for use in the 1990s, navigation has failed to gain widespread traction amongst surgeons that perform total knee arthroplasty (TKA) and unicompartmental knee arthroplasty (UKA).^1,3,5 From 2010-2014, utilization of imageless navigation in TKA decreased by 38.3% in the United States; current national rates of utilization of navigation in UKA are not known.⁵

Proponents of navigation cite improved implant positioning and improved limb alignment relative to non-navigated procedures, decreased risk of perioperative transfusion, and high rates of implant survivorship^6-11. The clinical significance of improved limb and implant alignment in terms of patient-reported outcomes and implant survival remains controversial^1,11-14. Additionally, the use of navigation may come at a higher overall cost per surgery, especially in low volume centers, and longer operative times.^7,9,15,16

The purpose of this study is to evaluate current rates of utilization of navigation in UKA in the United States. Additionally, we compare the incidence of short term complications, perioperative blood transfusions, and operative times between navigated and non-navigated UKA. We hypothesized that the rate of use of navigation in UKA will be similar to rates in TKA, approximately 3-5% of all cases,⁵ and that there would be no difference in rate of short term complications, perioperative blood transfusions, and operative times between navigated and non-navigated UKA.

Methods

This study was granted exemption status from the institutional review board. A query of the American College of Surgeons National Surgical Quality Improvement Project (ACS-NSQIP) database was performed for patients who had undergone primary UKA (Common Procedural Terminology [CPT] code 27446) during years 2006-2017. Additional CPT codes of 0055T (CT/MRI navigation), and 20985 (image-less navigation) were used to identify cases in which navigation was utilized. There were three cases with associated CPT code 0054T (fluoroscopic navigation); these cases were excluded from analysis. Patients undergoing revision surgery and emergency surgery were excluded from this study.

NSQIP database utilizes data submitted by participating institutions.¹⁷ Methods for data collection and curation of the NSQIP database, as well as specific inclusion and exclusion criteria for the database have been described previously.^5,18,19 Studies aimed at auditing and assessing the quality of data within national databases have demonstrated the NSQIP database to be a reliable source of data with high rates of inter-rater reliability.^20,21 Data in the NSQIP database is reviewed and audited semiannually.¹⁸

Patient Demographics

Patient-specific and case-specific variables utilized for this study are demonstrated in Table 1. Exhaustive descriptions of all variables in the NSQIP database are located within the NSQIP user guide.²²

Table 1.

Preoperative Patient Demographics

Variable	Non-navigated UKA (n=10,243)	Navigated UKA (n=343)
Demographics
Age, mean (SD)	64.1 (10.6)	64.9 (9.8)
Gender, % female	51.7%	52.2
Race, %
White	77.7	93.6
Black	3.6	3.2
Other	18.7	3.2
Preoperative conditions
BMI (kg/m2), mean (SD)	31.6 (6.3)	31.6 (5.7)
Functional status, %
Independent	98.7	99.7
Dependent	1.3	.3
Diabetes mellitus, %	15.2	15.7
Smoking, %	9.9	9.9
Dyspnea, %	4.6	5.8
COPD, %	2.8	2.6
Congestive heart failure (CHF), %	.1	0
Prior MI, %	0	0
Hypertension, %	56.8	58.6
Chronic kidney disease (CKD), %	0	0
End-stage renal disease (ESRD), %	.1	0
Metastatic cancer, %	.1	0
Prior wound infection, %	.2	.3
Steroid use, %	1.7	.6
Bleeding disorder, %	1.8	2
Preoperative blood transfusion, %	0	0
Sepsis, %	0	0
Prior operation within 30 days, %	.1	0
ASA Class, %
1 – no disturbance	3.7	3.2
2 – mild disturbance	57.6	58.6
3 – severe disturbance	37.8	37.3
4 – life threatening disturbance	.9	.9
Preoperative laboratory values
Creatinine, mean (SD)	.8	.8
Albumin, mean (SD)	2	2
WBC, mean (SD)	6.4	6.2
Hematocrit, mean (SD)	38.5	38.1
Platelets, mean (SD)	218.9	216.2
Anesthesia type
General, %	46.6	30.3
Spinal, %	38.1	36.7
Epidural, %	.7	.9
Regional (non-spinal), %	10.3	19.5
Other, %	4.3	12.6

UKA – unicompartmental knee arthroplasty; BMI – body mass index; SD – standard deviation; WBC – white blood cell count; ASA – American Society of Anesthesiologists.

Outcomes

ACS-NSQIP collects patient morbidity and mortality outcomes to 30 days after the index surgery. Per the ACS-NSQIP user guide, reporting of all outcomes (either positive or negative) is mandatory for all participating centers and records entered into NSQIP,²² as such, no outcomes are omitted from analysis. In the present study, we report 15 clinical outcome variables that we believe to be most clinically relevant to patients undergoing UKA; these variables are listed in Table 2. Additional outcome variables, including peripheral nerve injury, were not included in the study, as the methods of collection of this variable were deemed erroneous by the NSQIP PUF.²²

Table 2.

Complications

	Non-navigated UKA		Navigated UKA	p value
Complications, %	Unadjusted	Matched		Unadjusted	Matched
Any	3.55	3.21	3.55	.36	.65
Superficial wound infection	.61	.29	.29	.72	1
Deep wound infection	.21	0	0	1
Organ space infection	.21	.29	0	.53	1
Wound dehiscence	.13	0	0	1
Pneumonia	.18	0	.29	.47
Reintubation	.06	0	0	1
Pulmonary embolism	.21	0	.29	.53
Renal failure	.05	0	0	1
UTI	.59	.58	.29	.72	1
Stroke	.03	.29	0	1	1
MI	.12	0	0	1	1
Blood transfusion	.53	.58	.29	1	1
DVT	.37	.29	0	1	1
Sepsis	.17	.29	0	1	1
Shock	.01	0	0	1
Operative time, min, mean (SD)	87.5 (36.4)	84.3 (29)	92.1 (29.3)	<.001	<.001
LOS, days, mean (SD)	1.8 (2.2)	1.3 (1.6)	1.2 (1.9)	<0.001	.15
Readmission, %	.31	0	.58	.31	1
Reoperation, %	.74	.29	.29	.52	1

UTI – urinary tract infection; MI – myocardial infarction; DVT – deep vein thrombosis; SD – standard deviation; LOS – length of stay.

Statistical Analysis

Propensity score matching is a powerful statistical procedure for reducing selection bias due to confounding factors between study groups.²³ Previous studies have used these methods in orthopedics cohorts.^24-26 In this study, we conducted a one-to-one propensity score matching for patients in which navigation was utilized against cases in which no navigation was used. Multivariate logistic regression was built to predict the probability of everyone to select an equal number of subjects with comparable characteristics for both groups, controlling baseline difference in patient characteristics for age, race, BMI, ASA class, postoperative disposition, and diabetes status. Hosmer and Lemeshow goodness-of-fit test was used to detect the fitness of the model.

Univariate analysis was used to detect the difference between additional procedure group and none additional procedure group on demographic, comorbidities, and 30 days complication outcomes before and after matching procedure, using Chi-square statistics for categorical variables and Wilcoxon sum rank test for continuous variables. Linear regression analysis was used to examine temporal trends. Complications were defined by variables listed in Table 2. All statistical analysis was performed using the SAS 9.4 (Cary, NC). Statistical significance was set at p<0.05.

Results

A total of 10,586 cases of UKA were identified; 343 of these cases (3.2%) utilized navigation. Comparing the non-navigated and navigated cohorts, there was no difference in age (64.1 ± 10.6 years vs. 64.9 ± 9.8 years; p=0.12), body mass index (BMI) (31.6 ± 6.3 kg/ m2 vs. 31.6 ± 5.7 kg/ m2; p=0.75), gender (51.7% female vs. 52.2% female; p=0.85), incidence of diabetes (15.2% vs. 15.3%; p=0.78), smoking (9.9% vs. 9.9%; p=0.98), and chronic obstructive pulmonary disease (COPD) (2.9% vs. 2.6%) (Table 1). The only difference in patient demographics and comorbidities between the cohorts was a higher preoperative white blood cell count in the non-navigated cohort relative to the navigated cohort (6.36 k/mm3 vs. 6.17k/mm3; p=0.04).

The unadjusted rate of any complication for the entire cohort was 3.6%. Common 30-day complications in the entire cohort included superficial wound dehiscence (0.61%), urinary tract infection (UTI) (0.59%), requiring a blood transfusion (0.53%), and deep vein thrombosis (DVT) (0.37%) (Table 1).

Following propensity score matching, there was no difference in overall complication rates between the matched non-navigated and navigated cohorts (3.2% vs. 3.6%; p=0.65) (Table 2). Rates of specific complications within each respective cohort are present in Table 2. There was no difference in rates of readmission between the matched non-navigated and navigated cohorts (0.58% vs. 0.31%; p=0.31); there was also no difference in rates of reoperation (0.29% vs. 0.29%; p=1.00). Mean length of stay was no different between cohorts 1.3 ± 1.6 days vs. 1.2 ± 1.9 days; p=0.15). The navigated UKA cohort had mean operative times 8 minutes longer than the non-navigated UKA cohort (92.1 min vs. 84.3 min; p<0.001). (Table 2).

Evaluating the cohort by year, no navigated UKA were recorded in years 2006-2009 (Table 3, Figure 1). From 2010-2017, rates of navigation utilization ranged from 1.5% to 5.7% (Table 3, Figure 1). There was a statistically significant increase in the rate of navigation from 2006 to 2017 (p<0.001). However, when excluding years 2006 from 2009, the increase in navigation utilization from 2.1% in 2010 to 4.5% in 2017 was not statistically significant (p=0.18).

Table 3.

Rates of Navigation During UKA By Year

Year	No. Nav UKA	No. total UKA	% Nav
2006	0	17	0
2007	0	77	0
2008	0	147	0
2009	0	677	0
2010	10	484	2.1
2011	16	601	2.7
2012	26	745	3.5
2013	14	787	1.8
2014	19	1293	1.5
2015	46	1620	2.8
2016	119	2088	5.7
2017	93	2050	4.5
Total	343	10,586	3.2

No. – number; Nav – navigation

Figure 1.

Utilization of navigation in UKA by year.

Discussion

Navigation has been shown to improve limb and implant alignment following primary UKA.^7,8,13,27,28 However, the clinical implications of this improved alignment remain unclear.⁹ Short and medium-term outcome studies suggest that the survival benefit may be marginal¹² and that navigation may offer better clinical outcomes only in select patient populations.²⁹ The impact of these potential benefits must be weighed against the costs of navigation use, namely, the potential for longer operative times^7,9 and increased rates of complication.⁹ While the present study did not examine measures of patient pain, function, or long term outcome, we did evaluate potential effects of navigation use on operative time and short term outcomes, including wound complications and deep infections.

In the present study, we evaluated the rate of utilization of navigation in primary UKA. From 2006-2017, the mean rate of navigation utilization within the NSQIP database was 3.2%. The rate of navigation utilization increased significantly over the entire study period (Table 3; Figure 1); however when evaluating only years with ≥1 UKA, the trend was no longer statistically significant. In a retrospective review of primary TKA in the ACS-NSQIP database, Gholson et al.⁵ demonstrated a navigation utilization rate of 4.96% in 2010 and 3.06% in 2014, a 38% decrease. In contrast, Antonios et al.³⁰ found an increase in the rate of utilization of navigation in primary TKA from 1.2% in 2005 to 6.3% in 2014. In their study, Antonios et al.³⁰ noted the presence of significant regional variation within navigation utilization, with higher rates of computer utilization in the Western US.

A common detraction regarding the use of computer navigation is the concern that it increases operative time without an increase in clinical benefit. In a cohort of 296 patients undergoing primary UKA, Jenny et al.⁷ found operative times to be 20 minutes longer in the navigation group. A 2013 meta-analysis by Weber et al.³¹ noted operative times to be 15.4 minutes longer in the navigated cohort relative to the non-navigated cohort. Nair et al.,⁹ in a 2014 systematic review, also noted longer operative times in navigated UKAs. In the present study, we found a statistically significant difference in operative times, with navigated UKA, on average, taking 5-8 minutes longer than non-navigated UKA. With respect to operative times in primary TKA, the literature is widely variable, with studies demonstrating no difference in operative times between navigated and non-navigated groups,⁵ others demonstrating faster operative times in navigated cohorts^32,33 and others still with faster operative times in non-navigated cohorts.^34,35

Overall rates of short-term (≤90 days postoperatively) complication following UKA are estimated to be 3.2-5.6%.^36-40 Complication profiles between UKA performed on an inpatient versus outcome basis are thought to be similar.^40,41 In the present study, we noted an overall complication rate of 3.6% in the first 30 days following UKA (Table 2). There was no significant difference in rate of complication between navigated and non-navigated UKA (Table 2). Further, there was no difference in rates of readmission or reoperation. In a cohort of over 10,000 patients from a Medicare claims database, Chona et al.³⁶ found no difference in rates of revision operation or need for repeat arthrotomy within 2 years between navigated and non-navigated UKA. They also demonstrated no difference in rates of deep venous thrombosis (DVT) between navigated and non-navigated cohorts.³⁶

The present study has several limitations. First, this study is a retrospective review of a prospectively collected database. While the ACS-NSQIP is vetted for accuracy multiple times a year, it is still subject to errors in data collection, collation, and transcription.^20,21 The study is also subject to coding inaccuracies at the time of surgery. While this study utilizes propensity score matching, we are unable to account for all patient specific variables. Further, we are unable to account for surgeon-specific variables that may influence patient outcomes: surgeon experience, operative technique, and postoperative protocols. It is important to reiterate that this study does not examine measures of patient pain, function, or clinical outcomes beyond 30 days, where the use of navigation may or may not translate into clinical benefit. Finally, it is important to note that we, the authors, have no way of validating the data within the NSQIP database outside of the internal measures NSQIP already uses, and to this end, we have no way of knowing how many cases of UKA were coded as not using navigation that did use navigation, or vice versa.

In conclusion, the rate of utilization of computer navigation during UKA appears to be increasing; however, overall rates of utilization remain low around 3-5%. Based on data from a large clinical registry, we found no difference in rates of short term complications, reoperation, or readmission following navigated versus non-navigated UKA. Mean operative times in the navigated UKA cohort were eight minutes longer than the non-navigated cohort. Further studies elucidating the clinical benefit and cost effectiveness of navigation are needed.