Surgical Management of Osteoarthritis of the Knee
Endorsed by: The Knee Society, SOMOS, AAHKS, ACR, AGS, AANA
BMI as A Risk Factor
Strong evidence supports that obese patients have less improvement in outcomes with total knee arthroplasty (TKA).
Strong Evidence Strong Evidence
There were four high quality papers extracted that addressed complication rates after total knee arthroplasty for obese patients. Two (Bordini 2009, Judge 2012) demonstrated no higher complication rates in obese patients, whereas the other two (Jamsen, 2013, Amin,2006) did show higher rates of  complications.  The conflicting high quality papers negate each other and did not allow for a recommendation regarding complications. There were two high quality papers that demonstrated less improvement in functional outcomes in obese patients after total knee arthroplasty (Judge 2012, Amin 2006).  As such the recommendation was made that strong evidence supports the risk for less good outcomes after total knee arthroplasty.
 
Diabetes as a Risk Factor
Moderate evidence supports that patients with diabetes are at higher risk for complications with total knee arthroplasty (TKA).
Moderate Evidence Moderate Evidence
There was one high quality paper (Jamsen 2013) that showed a higher rate of complications and an increased risk of revision surgery for diabetics after total knee arthroplasty. Since it was the only high quality paper extracted, the recommendation strength is moderate. 
 
Chronic Pain as a Risk Factor
Moderate evidence supports that patients with select chronic pain conditions have less improvement in patient reported outcomes with TKA.
Moderate Evidence Moderate Evidence
One moderate quality paper (Boyle 2014) used low back pain as one form of chronic pain and demonstrated less good outcomes. Another moderate quality paper (Perruccio 2012) showed less good outcomes after total knee arthroplasty for patients with multiple joint and/or spine pain. The two retain a moderate quality of evidence leading to the recommendation that moderate evidence supports that patients with select chronic pain conditions have less improvement in patient reported outcomes with total knee arthroplasty.
 
Depression/Anxiety as a Risk Factor
Limited evidence supports that patients with depression and/or anxiety symptoms have less improvement in patient reported outcomes with total knee arthroplasty (TKA).
Limited Evidence Limited Evidence
One moderate quality study (Duiven 2013) and one low quality paper (Singh 2010) demonstrated less good outcomes in patients with anxiety/depression. There only being one moderate quality paper, the recommendation made was that limited evidence supports that patients with depression and/or anxiety symptoms have less improvement in patient reported outcomes with total knee arthroplasty.
 
Cirrhosis/Hepatitis C as a Risk Factor
Limited evidence supports that patients with cirrhosis or hepatitis C are at higher risk for complications with total knee arthroplasty (TKA).
Limited Evidence Limited Evidence
Given that the liver is the target organ for hepatitis C, these two risk factors were grouped together. Shih (2004) demonstrated higher complication rates after total knee arthroplasty in patients with cirrhosis and was assigned moderate quality. Pour (2011) was lower quality paper that demonstrated the same in patients with hepatitis C virus status alone and without liver damage. Given the one moderate study and one low quality study, the recommendation was made that limited evidence supports that patients with cirrhosis or hepatitis C are at higher risk for complications with total knee arthroplasty.
 
Preoperative Physical Therapy
Limited evidence supports that supervised exercise before total knee arthroplasty (TKA) might improve pain and physical function after surgery.
Limited Evidence Limited Evidence
Four high quality studies (Villadsen 2013, Gstoettner 2011, McKay 2012, D’Lima 1996) and four moderate quality studies (Rooks 2006, Topp 2009, Weidenhielm 1993, Brown 2012) compared pre-operative structured exercise program to groups receiving no-exercise, placebo exercise, or education.
 
One study of high quality (Villadsen 2013) and two studies of moderate quality (Topp 2009, Brown 2012) investigated the effects of exercise programs that combined primarily functional training, resistance training, and flexibility exercises compared to not receiving such exercise programs. Villadsen et al compared an exercise program of  eight week duration (1 hour twice a week) supervised by physical therapists that combined warm-up, core stability, postural orientation, resistance training, and functional exercises, to a group who received education on exercise. They reported significantly improved physical function and pain six weeks after surgery, but the differences were no longer significant 3 months after total knee arthroplasty. Topp et al compared an experimental group who received supervised exercise program of four week duration (3 times per week) that combined flexibility exercises, resistance training, and step training, to a group who did not exercise. They reported conflicting results for physical function and pain. At 3 months after total knee arthroplasty the exercise group performed more sit-to-stand repetitions than the control group but the control group ascended stairs faster than the exercise group. The exercise group has less pain during stairs descend but more pain during sit-to-stand task as compared to the control group. Brown et al compared a 8-week (3 session per week) supervised exercise program comprised of warm up, resistance training at moderate intensity, flexibility exercises, and step training, to a control group who did not exercise. They reported better physical function in the exercise group.
 
Two studies of high quality (McKay 2012, D’Lima 1996) and one study of moderate quality (Weidenhielm 1993) evaluated the effects of resistance training primarily. McKay et al compared a group who performed 6 weeks of moderate-intensity strength training of the lower body to a group who did upper body resistance training (placebo). D’Lima designed a three-group study to compare strength training of lower and upper body, aerobic training, and routine care (no exercise). D’Lima was the only study on pre-rehabilitation that had an exercise group who did aerobic training only. Weidenhielm et al compared a 5-week exercise program of knee range of motion and lower body strength training to a group who did not exercise. These studies found no significant differences in outcome between groups. One study of moderate quality (Rooks 2006) et al compared a 6-week exercise program with cardiovascular, strength, and flexibility training to an attention-control group who received education on total knee arthroplasty. Amongst the outcomes evaluated at 8 and 26 weeks after total knee arthroplasty, only bodily pain at 26-week was significantly less in the exercise group. One study of high quality (Gstoettner 2012) demonstrated that 6-week of stretching and balance training was not effective on physical function and pain.
 
Delayed Total Knee Arthroplasty
Moderate evidence supports that an eight month delay to total knee arthroplasty (TKA) does not worsen outcomes.
Moderate Evidence Moderate Evidence
There was one high quality study (Tuominen, U., 2010) that addressed the question of worsening of outcomes or an increase in complications on delayed cases of KA among adult patients with osteoarthritis, compared to cases without delay after having failed non-surgical management.
 
This study evaluated the effects of waiting time on health related quality of life, knee pain and physical function. The study also addressed the use and costs of medication of patients awaiting TKA. The mean waiting time was 94 days among those patients short waiting times versus 239 days (mean of 8 months) among those with non-fixed waiting times groups, respectively. Those in the short waiting time group had higher weekly costs of medication at admission, and reached better quality of life 3 months earlier than those in the other group, but the latter had better quality of life after operation. 
 
Peri-Articular Local Anesthetic Infiltration
Strong evidence supports that the use of peri-articular local anesthetic infiltration in total knee arthroplasty (TKA) decreases pain and opioid use compared to placebo.
Strong Evidence Strong Evidence
Five high quality studies (Nakai 2013, Koh 2011, Klasen 1999, Busch 2006, Chen 2012) compared peri-articular infiltration (PAI) to placebo (normal saline or no infiltration) for total knee arthroplasty. Improved function (Chen 2012), lower opioid consumption Busch 2006, Chen 2012, improved patient satisfaction (Busch 2006), and lower visual analog scale (VAS) pain scores (Nakai 2013, Koh 2011, Busch 2006, Chen 2012) all favored peri-articular injection.
 
Twenty-seven high quality studies originally met the selection criteria. Comparisons between PAI and placebo, PAI and peripheral nerve blocks (femoral and/or sciatic nerve blocks), and PAI and epidural blocks were attempted. However, due to the heterogeneity of the studies, PAI could only be compared to placebo. The heterogeneity of the studies included differences in infiltration solution (long-acting local anesthetics, plus or minus ketorolac, plus or minus opioid, plus or minus corticosteroid), varying concentrations of infiltration solution and injections, single-injection or catheter peripheral nerve blocks, peripheral nerve blocks (femoral and/or sciatic), and epidural catheter infusions (local anesthetic, opioid, and rates). 
 
Peripheral Nerve Blockade (PNB)
Strong evidence supports that peripheral nerve blockade for total knee arthroplasty (TKA) decreases postoperative pain and opioid requirements.
Strong Evidence Strong Evidence
There were seven high-quality (McNamee 2001, Good 2007, Kadic 2009, Xie 2012, Chan 2012, Moghtadaei 2014, Liu 2014) and three low-quality (Lau 1998, Beaupre 2012, Kim 2012) studies evaluating whether the use of peripheral nerve blockade reduces complications or improves outcomes in adult patients undergoing knee arthroplasty compared to no peripheral nerve block use.
 
Three high-quality studies (Chan 2012, Moghtadaei 2014, Liu 2014) demonstrated significantly lower VAS pain scores and opioid requirements during the postoperative period when peripheral nerve blockade was compared to parenteral opioids alone. 
 
One high-quality study (Chan 2012) demonstrated improvement in overall range-of-motion and a reduction in opioid-related side effects with the use of peripheral nerve blockade when compared to no peripheral nerve block use.  Another high-quality study (Liu 2014) demonstrated that peripheral nerve block use improved the Quality of Recovery (e.g., Emotive, Nociceptive and Cognitive domains) during the immediate postoperative period.
 
Neuraxial Anesthesia
Moderate evidence supports that neuraxial anesthesia could be used in total knee arthroplasty (TKA) to improve select perioperative outcomes and complication rates compared to general anesthesia.
Moderate Evidence Moderate Evidence
There were six high-quality (Nielsen PT 1990, Nielson WR 1990, Mitchell 1991, Jorgensen 1991, Williams-Russo P 1995, Williams-Russo P 1996) and three low-quality (Sharrock 1991, Stundner 2012, Memtsoudis 2013) studies evaluating whether neuraxial anesthesia (“spinal or epidural”) reduces complications or improves outcomes in adult patients undergoing knee arthroplasty compared to general anesthesia.
 
Two high-quality studies (Nielsen PT 1990, Jorgensen 1991) and one low-quality (Sharrock 1991) study demonstrated significantly lower rates of deep venous thrombosis (DVT) compared to general anesthesia.  Of note, the two high-quality studies did not utilize any form of perioperative prophylactic anticoagulation; and the low-quality study utilized postoperative aspirin therapy only. Neither study used warfarin or low-molecular weight heparin as part of their postoperative DVT prophylactic regimen.  Four additional low- (Stundner 2012, Memtsoudis 2013) quality studies demonstrated significant reductions in overall postoperative complications with neuraxial anesthesia; including reductions in blood transfusion rates, pulmonary compromise, pulmonary embolism, pneumonia, mechanical ventilation rates, acute renal failure and composite infectious complications.
 
Two high-quality studies demonstrated improved short-term functional outcomes after neuraxial anesthesia.  Specifically, Williams-Russo (1996) demonstrated improved short-term range-of-motion (flexion) and short-term ambulation (days until unassisted stair climbing) compared to general anesthesia.  Nielson WR (1990) demonstrated improved short-term cognitive function (Wechsler Memory Scale; Controlled Oral Word Association) compared to general anesthesia.
 
One low-quality study (Memtsoudis) demonstrated a significant reduction in 30-day mortality in patients undergoing neuraxial anesthesia compared to general anesthesia.
 
Tourniquet: Blood Loss Reduction
Moderate evidence supports that the use of a tourniquet in total knee arthroplasty (TKA) decreases intraoperative blood loss.
Moderate Evidence Moderate Evidence
With regard to pain, two high quality studies (Liu 2014 and Ledin 2012) and another moderate quality study (Ejaz 2014) showed decreased pain in the no tourniquet group in the very early postoperative period that was not significant after four days (Ledin 2012 and Liu 2014) and eight weeks (Ejaz 2014) respectively.
 
One high quality study (Ledin 2012) and one moderate quality (Aglietti 2000) found increased intraoperative blood loss in the no tourniquet patients. However, Ledin 2012 found no increased total bleeding when hemoglobin dilution was measured and Aglietti 2000 found no difference when overall total blood loss was tabulated.
 
One high quality study (Liu 2014) showed better quadriceps function in the no tourniquet group but equivalent Oxford Knee Scores and range of motion. One moderate quality (Ejaz 2014) study demonstrated better Knee Injury and Osteoarthritis Outcomes (KOOS) subscores and early range of motion to week eight postoperatively in the no tourniquet group, where differences then became statistically insignificant.
 
Tourniquet: Postoperative Pain Reduction
Strong evidence supports that tourniquet use in total knee arthroplasty (TKA) increases short term post-operative pain.
Strong Evidence Strong Evidence
With regard to pain, two high quality studies (Liu 2014 and Ledin 2012) and another moderate quality study (Ejaz 2014) showed decreased pain in the no tourniquet group in the very early postoperative period that was not significant after four days (Ledin 2012 and Liu 2014) and eight weeks (Ejaz 2014) respectively.
 
One high quality study (Ledin 2012) and one moderate quality (Aglietti 2000) found increased intraoperative blood loss in the no tourniquet patients. However, Ledin 2012 found no increased total bleeding when hemoglobin dilution was measured and Aglietti 2000 found no difference when overall total blood loss was tabulated.
 
One high quality study (Liu 2014) showed better quadriceps function in the no tourniquet group but equivalent Oxford Knee Scores and range of motion. One moderate quality (Ejaz 2014) study demonstrated better Knee Injury and Osteoarthritis Outcomes (KOOS) subscores and early range of motion to week eight postoperatively in the no tourniquet group, where differences then became statistically insignificant.
 
Tourniquet: Postoperative Function
Limited evidence supports that tourniquet use in total knee arthroplasty (TKA) decreases short term post-operative function.
Limited Evidence Limited Evidence
With regard to pain, two high quality studies (Liu 2014 and Ledin 2012) and another moderate quality study (Ejaz 2014) showed decreased pain in the no tourniquet group in the very early postoperative period that was not significant after four days (Ledin 2012 and Liu 2014) and eight weeks (Ejaz 2014) respectively.
 
One high quality study (Ledin 2012) and one moderate quality (Aglietti 2000) found increased intraoperative blood loss in the no tourniquet patients. However, Ledin 2012 found no increased total bleeding when hemoglobin dilution was measured and Aglietti 2000 found no difference when overall total blood loss was tabulated.
 
One high quality study (Liu 2014) showed better quadriceps function in the no tourniquet group but equivalent Oxford Knee Scores and range of motion. One moderate quality (Ejaz 2014) study demonstrated better Knee Injury and Osteoarthritis Outcomes (KOOS) subscores and early range of motion to week eight postoperatively in the no tourniquet group, where differences then became statistically insignificant.
 
Tranexamic Acid
Strong evidence supports that, in patients with no known contraindications, treatment with tranexamic acid decreases postoperative blood loss and reduces the necessity of postoperative transfusions following total knee arthroplasty (TKA).
Strong Evidence Strong Evidence
Eight high quality studies (Antinolfi, 2013, Charoencholvanich, 2011, Gautam, 2011, Good, 2003,Ishida, 2011,Roy, 2012, Sa-Ngasoongsong, 2013, Sarzaeem, 2014, Pachauri, 2014) were reviewed to assess the impact of tranexamic acid administration on blood loss and transfusion rates post total knee arthroplasty. There was significant variability in dosing, route of administration, and timing of administration; when assessed collectively, however, the use of tranexamic acid did show improvement in blood loss related outcomes.
 
Using six of the high quality studies a meta-analysis was performed on rate of blood transfusions which demonstrated a 52% reduction in patients receiving tranexamic acid (Figure 1). Three high quality studies demonstrated an improvement in Hgb. One high quality study (Ishida 2011) demonstrated a reduction in swelling and one high quality (Sa-Ngasoongsong 2013) study found an improvement in WOMAC function scores out to 1 year.
 
Antibiotic Bone Cement
Limited evidence does not support the routine use of antibiotics in the cement for primary total knee arthroplasty (TKA).
Limited Evidence Limited Evidence
Two moderate quality studies and one low quality registry review were considered.  One moderate quality randomized study demonstrated a reduction in total knee arthroplasty infection in diabetic patients from 13.5 % to 0% when cefuroxime was added to the cement.  This study was performed in operating rooms without modern features (Chiu 2001).  One moderate quality, randomized, prospective study demonstrated a reduction in revision total knee arthroplasty infection rates when vancomycin was added to the cement (Chiu 2009).  A large Canadian registry study reviewing more than 36000 patients found no difference in revision rates for infection between those patients treated with or without antibiotics in the cement. Given two moderate quality studies that are not widely applicable to patients with osteoarthritis undergoing primary total knee arthroplasty and one low quality, although large, registry review demonstrating no benefit from routinely adding antibiotics to cement for primary total knee arthroplasty, it is the conclusion of the work group that limited evidence does not support the routine use of antibiotics in the cement for primary total knee arthroplasty. One study did provide some suggestion that antibiotics added to the cement may be of benefit in diabetic patients. (Chiu 2001).

Of note, the FDA approved indications for antibiotic loaded cement in total knee arthroplasty are limited to revision scenarios and do not include primary applications (http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpcd/classification.cfm?id=mbb).
 
Cruciate Retaining Arthroplasty
Strong evidence supports no difference in outcomes or complications between posterior stabilized and posterior cruciate retaining arthroplasty designs.
Strong Evidence Strong Evidence
Meta-analysis of included literature was unable to show a difference between the cruciate retaining and posterior stabilized designs with regard to complications, pain, function or patient reported outcomes.
 
There is one high quality prospective comparative study (Maruyama 2004) evaluating outcomes and ROM in consecutive patients having bilateral total knee arthroplasty who had one posterior stabilized (PS) implant and one posterior cruciate retaining (CR) implant. They found equivalent Knee Society Scores, but statistically improved ROM in the PS group. Another high quality study (Roh 2013) failed to show improved kinematics or improved clinical outcome with PCL retention in highly conforming mobile bearing total knee arthroplasty. A third high quality study (Cankaya 2014) investigated blood loss with CR and PS designs in a prospective randomized study of 100 patients. They found no difference in either perioperative blood loss or postoperative transfusion rates between the two types of designs.
 
A moderate quality study (Clark 2001) in patients without extreme pre-operative deformities showed no notable differences between PS and CR designs with regard to knee scores, ROM or patient reported outcomes instruments SF-12 and WOMAC.  Likewise, four other moderate quality studies (Tanzer 2002, Catani 2004, Molt 2014, Ishii 2011) showed no differences between the CR and PS designs. Tanzer 2002 controlled for surgical technique by having a single surgeon perform a similar surgical technique for each design. Catani 2004 and Molt 2014 showed no statistical difference between designs with regard to tibial migration. Ishii 2011 found no difference between designs in range of motion. 
 
Polyethylene Tibial Components
Strong evidence supports use of either all-polyethylene or modular tibial components in knee arthroplasty (KA) because of no difference in outcomes.
Strong Evidence Strong Evidence
Three high (Kalisvaart 2012, Murray 2014, Hyldahl 2001) and five moderate quality (Adalberth 2001, Gioe 2000, Muller 2006, Norgren 2004, Adalberth 2000) studies evaluated the use of all-polyethylene versus modular (metal baseplate and polyethylene insert) tibial components in knee arthroplasty. 
 
One high quality randomized controlled trial (Kalisvaart 2012) of cemented posterior-stabilized total knee arthroplasty demonstrated no differences in range of motion, functional outcomes, stair climbing, or revisions across three tibial designs (all-polyethylene fixed-bearing, modular metal-backed fixed-bearing, rotating-platform) at two and five years post-operatively.
 
In a high quality multicenter trial (Knee Arthroplasty Trial; Murray 2014) randomizing the use of all-polyethylene and modular metal-backed tibia components in the United Kingdom, 89% of patients received the allocated procedure.  There were no differences in Oxford Knee Scores or rates of complications, reoperations, and revisions at ten years post-operatively.  There was a trend towards the metal-backed group having better EQ-5D and Short Form-12 scores based on marginal estimates over the entire ten-year follow up period.
 
A third high quality randomized trial (Hyldahl 2001) in unicompartmental knee arthroplasty, with a focus on radiostereometric analysis (RSA) of component fixation, found no differences with respect to clinical results (Hospital for Special Surgery score) or migration of the comparative tibial components over a two-year follow-up period.
 
Five moderate quality (Adalberth 2001, Gioe 2000, Muller 2006, Norgren 2004, Adalberth 2000) randomized controlled trials with minimum two years of follow up demonstrated no differences with respect to clinical results (all studies used the Knee Society Score, except for Short Form-12 and Oxford Knee Score used in the study by Muller 2006) and range of motion between all-polyethylene and modular tibial components in total knee arthroplasty.  Likewise, complications and reoperations were similar between groups in all studies, and equivalent component migration was measured in four studies utilizing RSA techniques (Adalberth 2001, Muller 2006, Norgren 2004, Adalberth 2000).
 
The practitioner must be aware that results in the literature may be implant specific, and that surgical technique and surgeon experience with particular methods are important factors in achieving durable results.  The decision to use modularity versus a monolithic tibial design may be influenced by particular patient situations, such as metal hypersensitivity and severe bone loss.  The practitioner should be aware of the advantages and disadvantages of the two treatments methods.
 
Patellar Resurfacing: Pain and Function
Strong evidence supports no difference in pain or function with or without patellar resurfacing in total knee arthroplasty.
Strong Evidence Strong Evidence
Strong evidence from high quality studies show very similar outcomes and complications with both patella resurfacing and no resurfacing. Unresurfaced categories often included a variety of limited debridements or releases such as circumferential patella osteophyte debridement or electrocautery. A meta-analysis showed that only reoperation rate (all reoperations, although a significant number were patella-related) was statistically increased in knees without patella resurfacing. This was only significant when enough reoperation data was aggregated to include reoperation after five years.
 
The high quality KAT trial (Breeman 2011 and Murray 2014) favors resurfacing for reasons of decreased reoperation. Four moderate quality studies also favored resurfacing for different reasons. Waters 2003 demonstrated higher anterior knee pain following total knee arthroplasty without resurfacing. Wood 2002 showed higher incidence of anterior knee pain in the knees that had not been resurfaced. One moderate quality study (Barrack 2001) showed anterior knee pain was same for overall KSS, and pain and function subscores, but reoperation significantly more common without resurfacing. Schroeder-Boersch 1998 showed better task knee function scores with resurfacing. Newman 2000 showed increased need for secondary surgery in the unresurfaced group. Partio 1995 showed decreased anterior knee pain in the resurfaced knees.
 
On the other hand, two high quality study (Bourne 1995) showed improved total Knee Society Scores (KSS) and KSS function scores in patients without patellar resurfacing. Liu 2012 chose to reshape the patella (osteophyte debridement) and found no difference in total KSS and in pain and function subgroups, arguing to keep patella bone stock. Campbell 2006 was unable to recommend resurfacing because of no significant differences in outcomes or complications. The KAT trial (Breeman 2011 and Murray 2014) found no statistically significant differences in EQ-5D score, SF-12 physical component scores and SF12 mental component scores.
 
Patellar Resurfacing: Reoperations
Moderate evidence supports that patellar resurfacing in total knee arthroplasty (TKA) could decrease cumulative reoperations after 5 years when compared to no patellar resurfacing in total knee arthroplasty (TKA).
Moderate Evidence Moderate Evidence
Strong evidence from high quality studies show very similar outcomes and complications with both patella resurfacing and no resurfacing. Unresurfaced categories often included a variety of limited debridements or releases such as circumferential patella osteophyte debridement or electrocautery. A meta-analysis showed that only reoperation rate (all reoperations, although a significant number were patella-related) was statistically increased in knees without patella resurfacing. This was only significant when enough reoperation data was aggregated to include reoperation after five years.
 
The high quality KAT trial (Breeman 2011 and Murray 2014) favors resurfacing for reasons of decreased reoperation. Four moderate quality studies also favored resurfacing for different reasons. Waters 2003 demonstrated higher anterior knee pain following total knee arthroplasty without resurfacing. Wood 2002 showed higher incidence of anterior knee pain in the knees that had not been resurfaced. One moderate quality study (Barrack 2001) showed anterior knee pain was same for overall KSS, and pain and function subscores, but reoperation significantly more common without resurfacing. Schroeder-Boersch 1998 showed better task knee function scores with resurfacing. Newman 2000 showed increased need for secondary surgery in the unresurfaced group. Partio 1995 showed decreased anterior knee pain in the resurfaced knees.
 
On the other hand, two high quality study (Bourne 1995) showed improved total Knee Society Scores (KSS) and KSS function scores in patients without patellar resurfacing. Liu 2012 chose to reshape the patella (osteophyte debridement) and found no difference in total KSS and in pain and function subgroups, arguing to keep patella bone stock. Campbell 2006 was unable to recommend resurfacing because of no significant differences in outcomes or complications. The KAT trial (Breeman 2011 and Murray 2014) found no statistically significant differences in EQ-5D score, SF-12 physical component scores and SF12 mental component scores.
 
Bone Cement: Cemented Tibial Components VERSUS Cementless Tibial Components
Strong evidence supports the use of tibial component fixation that is cemented or cementless in total knee arthroplasty due to similar functional outcomes and rates of complications and reoperations.
Strong Evidence Strong Evidence
There were five high (Lizaur-Utrilla 2014, Kim 2014, Beaupre 2007, Demey 2011, Fernandez-Fairen 2013) and seven moderate quality (Park 2011, Khaw 2002, Carlsson 2005, Baker 2007, Pandit 2013, Parker 2001, Pulido 2015) studies evaluating the use of various combinations of cemented versus cementless component (tibia, femur, patella) fixation in knee arthroplasty.

The overall body of evidence was notable for heterogeneity in study design and comparative study groups (including cementless, hybrid, and cemented fixation).   Nevertheless, across comparative groups, no major differences existed between cemented and cementless fixation with respect to rates of complications and re-operations, including studies with longer follow up (Khaw 2002, Baker 2007, Kim 2014).  

Only small differences were seen with respect to outcome measures, depending on the particular study comparative groups, length of follow up, and scoring instruments.  Lizaur-Utrilla found no significant differences in WOMAC scores at follow-up time points of two years or less when comparing cemented and cementless tibial fixation (with cementless femoral fixation and selective patellar resurfacing in both arms). WOMAC scores were significantly better in the uncemented (porous) tibial group (-5[-9.49,-0.51]) at final follow up (average 7 years), but this difference was not clinically significant.  Knee Society function scores were significantly better in the uncemented tibial group only at the 2 year follow up (-4[-7.62,-0.38]). Knee Society pain scores were significantly better at 2 years (-3 [-5.58, -0.42]) and at final follow up (-3 [-5.68, -0.32]), but not at 6 months or one year.  In a study comparing cemented and cementless tibial fixation (with cementless femoral fixation and selective patellar resurfacing in both arms), 
Beaupré  reported that WOMAC pain and RAND SF-36 bodily pain scores were significantly worse in the group with cementless hydroxyapatite-coated tibial components (9.1[2.79,15.41] versus 18.1[9.66,26.54] for cemented fixation)at 6 months. The differences in pain did not remain statistically significant at 1 or 5 years post-operatively. Fernandez-Fairen found that WOMAC scores were worse in the cemented tibial fixation group compared to scores in the cementless tibial fixation group (cementless femoral fixation and no patellar resurfacing in both arms), with a difference of 4 points (CI 0.13, 7.87) that was not clinically significant.  When comparing non-modular cemented tibial components with non-modular cementless porous tibial components, Pulido demonstrated more improvement in Knee Society pain scores (5 [0.08, 9.92]) in the cemented tibial group, but this difference was not clinically significant.  In a study of unicompartmental knee arthroplasty patients implanted with either cemented or cementless femoral/tibial fixation, Pandit reported significantly worse Knee Society function scores at 5 years (-12.2[-20.26,-4.14]), but not at 1 or 2 years, for the cemented group.  Tegner Activity Scores in the cemented group were significantly worse at 2 years (-0.6[-1.10,-0.10]), but not at 1 or 5 years.

More data is needed in particular patient subgroups, such as young and active patients, or those patients with poor bone quality.  Of note, two of the included controlled trials enrolled patients only 55 years of age and younger (Lizaur-Utrilla, Kim).  Likewise, the management of the patella varied across studies, and often between groups within a single study, including use of cemented fixation, selective resurfacing, and unresurfaced patellae.  More historical studies highlighting cementless tibia failure modes were largely excluded either due to poor study quality or date of publication.  The study by Parker noted worse Kaplan-Meier survivorship in patients with a metal-backed patella, and rates of metal-backed patella failure were particularly worse in the cementless fixation (tibial/ femoral) group when compared to patients with cemented fixation.

The practitioner should be aware that results in the literature may be implant- and design-specific, and that surgical technique and surgeon experience with particular fixation methods is important in achieving durable results.  Study quality was adjusted for those studies in whom authors had conflicts of interest with implant manufacturers.  
Bone Cement: Cemented Femoral & Tibial Components Versus Cementless Femoral & Tibial Components
Moderate evidence supports the use of either cemented femoral and tibial components or cementless femoral and tibial components in knee arthroplasty due to similar rates of complications and reoperations.
Moderate Evidence Moderate Evidence
There were five high (Lizaur-Utrilla 2014, Kim 2014, Beaupre 2007, Demey 2011, Fernandez-Fairen 2013) and seven moderate quality (Park 2011, Khaw 2002, Carlsson 2005, Baker 2007, Pandit 2013, Parker 2001, Pulido 2015) studies evaluating the use of various combinations of cemented versus cementless component (tibia, femur, patella) fixation in knee arthroplasty.

The overall body of evidence was notable for heterogeneity in study design and comparative study groups (including cementless, hybrid, and cemented fixation).   Nevertheless, across comparative groups, no major differences existed between cemented and cementless fixation with respect to rates of complications and re-operations, including studies with longer follow up (Khaw 2002, Baker 2007, Kim 2014).  

Only small differences were seen with respect to outcome measures, depending on the particular study comparative groups, length of follow up, and scoring instruments.  Lizaur-Utrilla found no significant differences in WOMAC scores at follow-up time points of two years or less when comparing cemented and cementless tibial fixation (with cementless femoral fixation and selective patellar resurfacing in both arms). WOMAC scores were significantly better in the uncemented (porous) tibial group (-5[-9.49,-0.51]) at final follow up (average 7 years), but this difference was not clinically significant.  Knee Society function scores were significantly better in the uncemented tibial group only at the 2 year follow up (-4[-7.62,-0.38]). Knee Society pain scores were significantly better at 2 years (-3 [-5.58, -0.42]) and at final follow up (-3 [-5.68, -0.32]), but not at 6 months or one year.  In a study comparing cemented and cementless tibial fixation (with cementless femoral fixation and selective patellar resurfacing in both arms), 
Beaupré reported that WOMAC pain and RAND SF-36 bodily pain scores were significantly worse in the group with cementless hydroxyapatite-coated tibial components (9.1[2.79,15.41] versus 18.1[9.66,26.54] for cemented fixation)at 6 months. The differences in pain did not remain statistically significant at 1 or 5 years post-operatively. Fernandez-Fairen found that WOMAC scores were worse in the cemented tibial fixation group compared to scores in the cementless tibial fixation group (cementless femoral fixation and no patellar resurfacing in both arms), with a difference of 4 points (CI 0.13, 7.87) that was not clinically significant.  When comparing non-modular cemented tibial components with non-modular cementless porous tibial components, Pulido demonstrated more improvement in Knee Society pain scores (5 [0.08, 9.92]) in the cemented tibial group, but this difference was not clinically significant.  In a study of unicompartmental knee arthroplasty patients implanted with either cemented or cementless femoral/tibial fixation, Pandit reported significantly worse Knee Society function scores at 5 years (-12.2[-20.26,-4.14]), but not at 1 or 2 years, for the cemented group.  Tegner Activity Scores in the cemented group were significantly worse at 2 years (-0.6[-1.10,-0.10]), but not at 1 or 5 years.

More data is needed in particular patient subgroups, such as young and active patients, or those patients with poor bone quality.  Of note, two of the included controlled trials enrolled patients only 55 years of age and younger (Lizaur-Utrilla 2014, Kim 2014).  Likewise, the management of the patella varied across studies, and often between groups within a single study, including use of cemented fixation, selective resurfacing, and unresurfaced patellae.  More historical studies highlighting cementless tibia failure modes were largely excluded either due to poor study quality or date of publication.  The study by Parker noted worse Kaplan-Meier survivorship in patients with a metal-backed patella, and rates of metal-backed patella failure were particularly worse in the cementless fixation (tibial/ femoral) group when compared to patients with cemented fixation.

The practitioner should be aware that results in the literature may be implant- and design-specific, and that surgical technique and surgeon experience with particular fixation methods is important in achieving durable results.  Study quality was adjusted for those studies in whom authors had conflicts of interest with implant manufacturers.
Bone Cement: All Cemented Components versus Hybrid Fixation (Cementless Femoral Component)
Moderate evidence supports the use of either cementing all components or hybrid fixation (cementless femur) in total knee arthroplasty due to similar functional outcomes and rates of complications and reoperations.
Moderate Evidence Moderate Evidence
There were five high (Lizaur-Utrilla 2014, Kim 2014, Beaupre 2007, Demey 2011, Fernandez-Fairen 2013) and seven moderate quality (Park 2011, Khaw 2002, Carlsson 2005, Baker 2007, Pandit 2013, Parker 2001, Pulido 2015) studies evaluating the use of various combinations of cemented versus cementless component (tibia, femur, patella) fixation in knee arthroplasty.

The overall body of evidence was notable for heterogeneity in study design and comparative study groups (including cementless, hybrid, and cemented fixation).   Nevertheless, across comparative groups, no major differences existed between cemented and cementless fixation with respect to rates of complications and re-operations, including studies with longer follow up (Khaw 2002, Baker 2007, Kim 2014).  

Only small differences were seen with respect to outcome measures, depending on the particular study comparative groups, length of follow up, and scoring instruments.  Lizaur-Utrilla found no significant differences in WOMAC scores at follow-up time points of two years or less when comparing cemented and cementless tibial fixation (with cementless femoral fixation and selective patellar resurfacing in both arms). WOMAC scores were significantly better in the uncemented (porous) tibial group (-5[-9.49,-0.51]) at final follow up (average 7 years), but this difference was not clinically significant.  Knee Society function scores were significantly better in the uncemented tibial group only at the 2 year follow up (-4[-7.62,-0.38]). Knee Society pain scores were significantly better at 2 years (-3 [-5.58, -0.42]) and at final follow up (-3 [-5.68, -0.32]), but not at 6 months or one year.  In a study comparing cemented and cementless tibial fixation (with cementless femoral fixation and selective patellar resurfacing in both arms), 
Beaupré reported that WOMAC pain and RAND SF-36 bodily pain scores were significantly worse in the group with cementless hydroxyapatite-coated tibial components (9.1[2.79,15.41] versus 18.1[9.66,26.54] for cemented fixation)at 6 months. The differences in pain did not remain statistically significant at 1 or 5 years post-operatively. Fernandez-Fairen found that WOMAC scores were worse in the cemented tibial fixation group compared to scores in the cementless tibial fixation group (cementless femoral fixation and no patellar resurfacing in both arms), with a difference of 4 points (CI 0.13, 7.87) that was not clinically significant.  When comparing non-modular cemented tibial components with non-modular cementless porous tibial components, Pulido demonstrated more improvement in Knee Society pain scores (5 [0.08, 9.92]) in the cemented tibial group, but this difference was not clinically significant.  In a study of unicompartmental knee arthroplasty patients implanted with either cemented or cementless femoral/tibial fixation, Pandit reported significantly worse Knee Society function scores at 5 years (-12.2[-20.26,-4.14]), but not at 1 or 2 years, for the cemented group.  Tegner Activity Scores in the cemented group were significantly worse at 2 years (-0.6[-1.10,-0.10]), but not at 1 or 5 years.

More data is needed in particular patient subgroups, such as young and active patients, or those patients with poor bone quality.  Of note, two of the included controlled trials enrolled patients only 55 years of age and younger (Lizaur-Utrilla 2014, Kim 2014).  Likewise, the management of the patella varied across studies, and often between groups within a single study, including use of cemented fixation, selective resurfacing, and unresurfaced patellae.  More historical studies highlighting cementless tibia failure modes were largely excluded either due to poor study quality or date of publication.  The study by Parker noted worse Kaplan-Meier survivorship in patients with a metal-backed patella, and rates of metal-backed patella failure were particularly worse in the cementless fixation (tibial/ femoral) group when compared to patients with cemented fixation.

The practitioner should be aware that results in the literature may be implant- and design-specific, and that surgical technique and surgeon experience with particular fixation methods is important in achieving durable results.  Study quality was adjusted for those studies in whom authors had conflicts of interest with implant manufacturers.  
Bone Cement: All Cementless Components versus Hybrid Fixation (Cementless Femoral Component)
Limited evidence supports the use of either all cementless components or hybrid fixation (cementless femur) in total knee arthroplasty due to similar rates of complications and reoperations.
Limited Evidence Limited Evidence
There were five high (Lizaur-Utrilla 2014, Kim 2014, Beaupre 2007, Demey 2011, Fernandez-Fairen 2013) and seven moderate quality (Park 2011, Khaw 2002, Carlsson 2005, Baker 2007, Pandit 2013, Parker 2001, Pulido 2015) studies evaluating the use of various combinations of cemented versus cementless component (tibia, femur, patella) fixation in knee arthroplasty.

The overall body of evidence was notable for heterogeneity in study design and comparative study groups (including cementless, hybrid, and cemented fixation).   Nevertheless, across comparative groups, no major differences existed between cemented and cementless fixation with respect to rates of complications and re-operations, including studies with longer follow up (Khaw 2002, Baker 2007, Kim 2014).  

Only small differences were seen with respect to outcome measures, depending on the particular study comparative groups, length of follow up, and scoring instruments.  Lizaur-Utrilla et al found no significant differences in WOMAC scores at follow-up time points of two years or less when comparing cemented and cementless tibial fixation (with cementless femoral fixation and selective patellar resurfacing in both arms). WOMAC scores were significantly better in the uncemented (porous) tibial group (-5[-9.49,-0.51]) at final follow up (average 7 years), but this difference was not clinically significant.  Knee Society function scores were significantly better in the uncemented tibial group only at the 2 year follow up (-4[-7.62,-0.38]). Knee Society pain scores were significantly better at 2 years (-3 [-5.58, -0.42]) and at final follow up (-3 [-5.68, -0.32]), but not at 6 months or one year.  In a study comparing cemented and cementless tibial fixation (with cementless femoral fixation and selective patellar resurfacing in both arms), 
Beaupré et al reported that WOMAC pain and RAND SF-36 bodily pain scores were significantly worse in the group with cementless hydroxyapatite-coated tibial components (9.1[2.79,15.41] versus 18.1[9.66,26.54] for cemented fixation)at 6 months. The differences in pain did not remain statistically significant at 1 or 5 years post-operatively. Fernandez-Fairen et al found that WOMAC scores were worse in the cemented tibial fixation group compared to scores in the cementless tibial fixation group (cementless femoral fixation and no patellar resurfacing in both arms), with a difference of 4 points (CI 0.13, 7.87) that was not clinically significant.  When comparing non-modular cemented tibial components with non-modular cementless porous tibial components, Pulido et al demonstrated more improvement in Knee Society pain scores (5 [0.08, 9.92]) in the cemented tibial group, but this difference was not clinically significant.  In a study of unicompartmental knee arthroplasty patients implanted with either cemented or cementless femoral/tibial fixation, Pandit et al reported significantly worse Knee Society function scores at 5 years (-12.2[-20.26,-4.14]), but not at 1 or 2 years, for the cemented group.  Tegner Activity Scores in the cemented group were significantly worse at 2 years (-0.6[-1.10,-0.10]), but not at 1 or 5 years.

More data is needed in particular patient subgroups, such as young and active patients, or those patients with poor bone quality.  Of note, two of the included controlled trials enrolled patients only 55 years of age and younger (Lizaur-Utrilla 2012, Kim 2014).  Likewise, the management of the patella varied across studies, and often between groups within a single study, including use of cemented fixation, selective resurfacing, and unresurfaced patellae.  More historical studies highlighting cementless tibia failure modes were largely excluded either due to poor study quality or date of publication.  The study by Parker et al noted worse Kaplan-Meier survivorship in patients with a metal-backed patella, and rates of metal-backed patella failure were particularly worse in the cementless fixation (tibial/ femoral) group when compared to patients with cemented fixation.

The practitioner should be aware that results in the literature may be implant- and design-specific, and that surgical technique and surgeon experience with particular fixation methods is important in achieving durable results.  Study quality was adjusted for those studies in whom authors had conflicts of interest with implant manufacturers.  
Bilateral TKA
Limited evidence supports simultaneous bilateral total knee arthroplasty for patients aged 70 or younger or ASA status 1-2, because there are no increased complications.
Limited Evidence Limited Evidence
There is one low quality retrospective comparative study (Yoon 2010) evaluating systemic complications in consecutive patients who had bilateral simultaneous total knee arthroplasty that met criteria for inclusion. They found equivalent complications among patients who were not elderly (defined as less than 71 years old) or not high risk (defined as ASA 1 and 2). Analysis showed patients aged 71 and older or ASA 3-4 were at higher risk of having systemic complications.

More data was not available for inclusion because many of the relevant studies included a mixture of patients with osteoarthritis and rheumatoid arthritis and the outcomes data was not split out. 
UKA: Revisions
Moderate evidence supports that total knee arthroplasty (TKA) could be used to decrease revision surgery risk compared to unicompartmental knee arthroplasty (UKA) for medial compartment osteoarthritis.
Moderate Evidence Moderate Evidence
One moderate quality study (Sun 2012) and our meta-analysis of two moderate quality (Sun 2012, Newman 1998) and one low quality (Cameron 1988) studies demonstrated that the rate of revision surgery was significantly higher for those patients with medial compartment OA of the knee treated with unicompartmental arthroplasty, when compared to total knee arthroplasty.

Comparing the data of two moderate quality studies (Newman 1998, Murray 2014) and one low quality study (Cameron 1988) for early complications there were fewer thromboembolic events and manipulations in the unicompartmental when compared to total knee arthroplasty.

One high quality (Stukenborg-Colsman 2001) and two moderate studies (Weidenhielm 1993 and Borjesson 2005) compared the outcomes of UKA and HTO in patients with predominantly medial compartment osteoarthritis. There were no statistically significant differences in complications or outcomes.

There was no data comparing tibial tubercle osteotomy to patellofemoral arthroplasty or total knee arthroplasty. Likewise, there was no data comparing distal femoral osteotomy to lateral compartment unicompartmental arthroplasty or total knee arthroplasty.
UKA: DVT & Manipulation Under Anesthesia
Limited evidence supports that unicompartmental knee arthroplasty might be used to decrease the risk of deep vein thrombosis (DVT) and manipulation under anesthesia compared to total knee arthroplasty (TKA) for medial compartment osteoarthritis.
Limited Evidence Limited Evidence
One moderate quality study (Sun 2012) and our meta-analysis of two moderate quality (Sun 2012, Newman 1998) and one low quality (Cameron 1988) studies demonstrated that the rate of revision surgery was significantly higher for those patients with medial compartment OA of the knee treated with unicompartmental arthroplasty, when compared to total knee arthroplasty.

Comparing the data of three moderate quality studies (Newman 1998, Murray 2014, Cameron 1988) for early complications there were fewer thromboembolic events and manipulations in the unicompartmental when compared to total knee arthroplasty.

One high quality (Stukenborg-Colsman 2001) and two moderate studies (Weidenhielm 1993 and Borjesson 2005) compared the outcomes of UKA and HTO in patients with predominantly medial compartment osteoarthritis. There were no statistically significant differences in complications or outcomes.

There was no data comparing tibial tubercle osteotomy to patellofemoral arthroplasty or total knee arthroplasty. Likewise, there was no data comparing distal femoral osteotomy to lateral compartment unicompartmental arthroplasty or total knee arthroplasty.
UKA Versus Osteotomy
Moderate evidence supports no difference between unicompartmental knee arthroplasty (UKA) or valgus-producing proximal tibial osteotomy in outcomes and complications in patients with medial compartment knee osteoarthritis.
Moderate Evidence Moderate Evidence
One moderate quality study (Sun 2012) and our meta-analysis of two moderate quality (Sun 2012, Newman 1998) and one low quality (Cameron 1988) studies demonstrated that the rate of revision surgery was significantly higher for those patients with medial compartment OA of the knee treated with unicompartmental arthroplasty, when compared to total knee arthroplasty.

Comparing the data of three moderate quality studies (Newman 1998, Murray 2014, Cameron 1988) for early complications there were fewer thromboembolic events and manipulations in the unicompartmental when compared to total knee arthroplasty.

One high quality (Stukenborg-Colsman 2001) and two moderate studies (Weidenhielm 1993 and Borjesson 2005) compared the outcomes of UKA and HTO in patients with predominantly medial compartment osteoarthritis. There were no statistically significant differences in complications or outcomes.

There was no data comparing tibial tubercle osteotomy to patellofemoral arthroplasty or total knee arthroplasty. Likewise, there was no data comparing distal femoral osteotomy to lateral compartment unicompartmental arthroplasty or total knee arthroplasty.
Surgical Navigation
Strong evidence supports not using intraoperative navigation in total knee arthroplasty (TKA) because there is no difference in outcomes or complications.
Strong Evidence Strong Evidence
Three high quality studies (Thiengwittayaporn 2013, Seon 2009, Kiss 2012) and two moderate quality studies (Lutzner 2010, Dutton 2008) compared surgical navigation to conventional instrumentation for total knee arthroplasty. At follow-up greater than 90 days, there were no differences in patient reported quality of life outcomes (EQ-5D, SF-36 Mental Component Summary), patient reported knee function (Oxford Knee Score, Knee Society Score, and WOMAC), and pain (WOMAC). 

Four high quality studies (Lutzner 2008, Church 2007, Chin 2005, Blakeney 2011) and one moderate quality study (Kalairajah 2005) were all consistent in their findings that length of surgery favored no surgical navigation. A meta-analysis on infection found no difference in infection risk comparing surgical navigation to conventional instrumentation for total knee arthroplasty.

The work group recognizes that there are scenarios where computer navigation theoretically could be considered, such as malunions, intramedullary implants, or in training scenarios, but the evidence is insufficient to make a recommendation.
Patient Specific Instrumentation: Pain and Function
Strong evidence supports not using patient specific instrumentation compared to conventional instrumentation for total knee arthroplasty (TKA) because there is no difference in pain or functional outcomes.
Strong Evidence Strong Evidence
Two high quality studies (Pfitzner 2014 and Pietsch 2013) demonstrated no difference in knee society and visual analog pain scores between patient specific cutting guides and conventional instrumentation. Both studies also found no difference in knee society function score and one study (Pfitzner 2014) additionally found no difference in the overall WOMAC score. There was heterogeneity in outcomes regarding perioperative blood loss with Pietsch 2013 demonstrating a clinically significant difference in perioperative blood loss, however, no impact in the drop in hgb and need for transfusion. This indicates that while there is some improvement in blood loss in patients who receive patient specific instrumentation there was no significant clinical impact when compared to conventional instrumentation. Three moderate quality studies and one low quality study found similar results with no significant difference between both treatment groups.

There was no evidence to demonstrate a clinically relevant difference in surgical time or shorter length of stay when comparing both groups. One moderate quality study (Boonen 2013) found no difference in hospital length of stay between both groups.

The work group recognizes that there are scenarios where patient specific instrumentation theoretically could be considered but the evidence is insufficient to make a recommendation.
Patient Specific Instrumentation: Transfusions and Complications
Moderate evidence supports not using patient specific instrumentation compared to conventional instrumentation for total knee arthroplasty (TKA) because there is no difference in transfusions or complications.
Moderate Evidence Moderate Evidence
Two high quality studies (Pfitzner 2014 and Pietsch 2013) demonstrated no difference in knee society and visual analog pain scores between patient specific cutting guides and conventional instrumentation. Both studies also found no difference in knee society function score and one study (Pfitzner 2014) additionally found no difference in the overall WOMAC score. There was heterogeneity in outcomes regarding perioperative blood loss with Pietsch 2013 demonstrating a clinically significant difference in perioperative blood loss, however, no impact in the drop in hgb and need for transfusion. This indicates that while there is some improvement in blood loss in patients who receive patient specific instrumentation there was no significant clinical impact when compared to conventional instrumentation. Three moderate quality studies and one low quality study found similar results with no significant difference between both treatment groups.

There was no evidence to demonstrate a clinically relevant difference in surgical time or shorter length of stay when comparing both groups. One moderate quality study (Boonen 2013) found no difference in hospital length of stay between both groups.

The work group recognizes that there are scenarios where patient specific instrumentation theoretically could be considered but the evidence is insufficient to make a recommendation.
Drains
Strong evidence supports not using a drain with total knee arthroplasty (TKA) because there is no difference in complications or outcomes.
Strong Evidence Strong Evidence
Four high-quality studies and three moderate-quality studies were reviewed.  These studies showed no difference in multiple measures including VTE, infection, swelling, blood transfusions, hematoma formation, range of motion, length of stay, pain or reoperation between the treatment groups.  One high quality study demonstrated a significantly greater need for manipulation (8 % vs. 0 %, P-value<0.05) in patients who did not receive a drain (Esler, 2003). Two high-quality studies reported significantly higher transfusion rates in patients who received a drain (Esler, 2003 and Li 2011).  Two high-quality studies found no difference in transfusion rates in the presence or absence of a drain (Ritter 1994 and Jenny 2001).  Meta-analysis of the included studies did not show significant differences in infection or flexion range of motion in the presence or absence of a drain.  One study (Niskanen 2000) suggested that there may be more wound drainage in patients without a drain.  All studies were relatively small ranging from 20 – 50 patients per treatment group with the exception of one high-quality study with 138 patients per treatment group (Ritter 1994).
Cryotherapy Devices
Moderate evidence supports that cryotherapy devices after knee arthroplasty (KA) do not improve outcomes.
Moderate Evidence Moderate Evidence
The literature extraction and review revealed one high quality study, two moderate quality studies, and one low quality study. 

The high quality study (Ivey 1994) used a cryotherapy sleeve on all of the patients and randomized the target temperatures up to including 70°F.  There was no evidence for dose dependent differences in need for pain medication including the control of 70°F. There is some issue with this control, in that it does have a cooling effect. As the only high level study it falls to a moderate level of evidence for this guideline.

Of the two moderate quality studies that used cryotherapy, one (Holmström 2005) consisted of postoperative unicompartmental knees that were randomized between epidural anesthesia, cryotherapy, and a third arm that does not document the use of simple cold packs/ice. It reported less pain medication consumption in the two treatment arms.

The second of the moderate quality studies (Su 2012) compared cryotherapy/compression to cryotherapy alone, including the early post-discharge period, and showed no significant outcome differences other than less overall narcotics used over the broad period of the first two weeks and higher levels of patient satisfaction.  The study involved 11 sites and the patients could not be blinded to treatment.

One low quality study (Thienpoint, 2014) demonstrated less flexion in the cryotherapy group at an intermediate time period; this was attributed to the patient having less freedom to bend their knee while in the device.

The lack of dose effect in reducing narcotic consumption in the high level study contradicts the findings in the two relevant moderate level studies, both of which were not internally supported by significant differences in visual analogue pain scales. There were no other significant differences in other outcomes in the two relevant moderate studies. 
Continuous Passive Motion (CPM)
Strong evidence supports that CPM after knee arthroplasty (KA) does not improve outcomes.
Strong Evidence Strong Evidence
Two high quality studies (Beaupre 2001, Denis 2006) and five moderate quality studies (Can 2003, Chen 2013, Herbold 2014, MacDonald 2000, Montgomery 1996) compared the utilization of continuous passive motion during hospital stay to no utilization of continuous passive motion. The combined results provide strong evidence that the surgical outcomes for those who used continuous passive motion are not better than for those who did not use continuous passive motion.   

Five of the seven studies measured outcomes of physical function and quality of life. Beaupre, Denis, Herbold, and MacDonald found no significant differences in a gamut of outcomes (WOMAC, SF-36, Timed “up + go” [TUG], functional independence measure [FIM], and Knee Society Score). Chen reported better quality of life in the group that did not use continuous passive motion. Knee range of motion was investigated by Beaupre, Denis, and Chen. Meta-analysis showed no differences in knee range of motion. Complications were evaluated by Beaupre and Denis and were not statistically different between groups. Beaupra, Can, Chen, MacDonald, and Montgomery demonstrated that pain and stiffness were not decreased by CPM, whereas Denis reported significantly less pain in the continuous passive motion group (12 points difference in VAS ranging from 0-100). Meta-analysis from Denis, Herbold, and Montgomery showed no differences in length of hospital stay. 

One high quality study (Lenssen 2008) demonstrated no statistically significant benefits in functional outcome scores or range of motion with the use of continuous passive motion in conjunction with physical therapy compared to physical therapy alone. The continuous passive motion was used for 17 consecutive days after surgery (about 2 weeks after discharge).
Postoperative Mobilization: Length of Stay
Strong evidence supports that rehabilitation started on the day of the total knee arthroplasty (TKA) reduces length of hospital stay.
Strong Evidence Strong Evidence
Two high quality studies (Labraca 2011; Larsen 2008) investigated the effects of starting rehabilitation on the day of surgery compared to delayed rehabilitation (start on the day after surgery or later). Labraca compared a group who initiated rehabilitation within the first 24 hours post-surgery to a control group who remained at rest during the first 24 hours and started rehabilitation after that. They found that the group who started rehabilitation within 24 hours had fewer days of hospital stay, reduced pain, and improved physical function (balance, muscle strength and range of knee motion). Larsen compared an intervention group who received a new accelerated peri-operative protocol compared to a control group who received conventional perioperative procedure. The accelerated protocol aimed to mobilize the patient in bed and out of bed in the day of surgery and progressed to four hours out of bed (combination of physical and occupational therapy) on the first postoperative day, and eight hours of mobilization for the rest of the hospital stay. The control group started mobilization in and out of bed on the day after surgery and increased mobilization according to patient’s state. The accelerated protocol also included education, pain relief, nausea control, nutrition, and elimination. The study found that the accelerated group had less length of stay as compared to the control group. Quality of life was not different between the groups. 
Postoperative Mobilization: Pain and Function
Moderate evidence supports that rehabilitation started on day of total knee arthroplasty (TKA) compared to rehabilitation started on postop day 1 reduces pain and improves function.
Moderate Evidence Moderate Evidence
Two high quality studies (Labraca 2011; Larsen 2008) investigated the effects of starting rehabilitation on the day of surgery compared to delayed rehabilitation (start on the day after surgery or later). Labraca compared a group who initiated rehabilitation within the first 24 hours post-surgery to a control group who remained at rest during the first 24 hours and started rehabilitation after that. They found that the group who started rehabilitation within 24 hours had fewer days of hospital stay, reduced pain, and improved physical function (balance, muscle strength and range of knee motion). Larsen compared an intervention group who received a new accelerated peri-operative protocol compared to a control group who received conventional perioperative procedure. The accelerated protocol aimed to mobilize the patient in bed and out of bed in the day of surgery and progressed to four hours out of bed (combination of physical and occupational therapy) on the first postoperative day, and eight hours of mobilization for the rest of the hospital stay. The control group started mobilization in and out of bed on the day after surgery and increased mobilization according to patient’s state. The accelerated protocol also included education, pain relief, nausea control, nutrition, and elimination. The study found that the accelerated group had less length of stay as compared to the control group. Quality of life was not different between the groups. 
Early Stage Supervised Exercise Program: Function
Moderate evidence supports that a supervised exercise program during the first two months after total knee arthroplasty (TKA) improves physical function.
Moderate Evidence Moderate Evidence
One high quality study (Evgeniadis 2008) and one moderate quality study (Akbaba 2014) investigated supervised exercise programs started after hospital discharge compared to no exercise or minimal exercise during the first two months after surgery. Evgeniadis et al compared a group of patients post total knee arthroplasty who received a home exercise program of eight weeks (three times a week) that consisted of lower extremity strength training, to a group who did not receive supervised exercises. The exercise group had significantly better physical function and knee flexion and extension range of motion. Akbaba et al compared a group of patients with bilateral total knee arthroplasty who received a month of intensive supervised rehabilitation (two times a week for one hour) to a control group who received supervised rehabilitation once every 15 days. The intense supervised group had less pain and stiffness, and better balance and physical function than the control group.

Two high quality studies (Liao 2013, Moffet 2004) and one moderate quality study (Valtonen 2010) investigated supervised intensive exercise programs started two or more months after surgery (late stage post total knee arthroplasty) compared to no or less exercise. Liao et al compared a group who performed functional exercise supplemented with balance training to a group who performed functional training only. The exercise programs lasted eight weeks and started two months post-surgery. The group who received a combination of functional and balance exercises had better patient reported and performance-based outcomes of physical function. Moffet et al compared a group who received intensive functional training during eight weeks to a standard care group who received minimal rehabilitative care.  Pain and emotional health was significantly better in the intensive functional training group at 4 and 6 months, but the effects were no longer significant at the 12 months’ time point.  Valtonen et al compared a group who performed a high-intensity progressive aquatic resistance training of six week duration that started at least four months after surgery to a control group who did not exercise. The outcomes of both groups were similar. 
Late Stage Postoperative Supervised Exercise Program: Function
Limited evidence supports that selected patients might be referred to an intensive supervised exercise program during late stage post total knee arthroplasty (TKA) to improve physical function.
Limited Evidence Limited Evidence
One high quality study (Evgeniadis 2008) and one moderate quality study (Akbaba 2014) investigated supervised exercise programs started after hospital discharge compared to no exercise or minimal exercise during the first two months after surgery. Evgeniadis et al compared a group of patients post total knee arthroplasty who received a home exercise program of eight weeks (three times a week) that consisted of lower extremity strength training, to a group who did not receive supervised exercises. The exercise group had significantly better physical function and knee flexion and extension range of motion. Akbaba et al compared a group of patients with bilateral total knee arthroplasty who received a month of intensive supervised rehabilitation (two times a week for one hour) to a control group who received supervised rehabilitation once every 15 days. The intense supervised group had less pain and stiffness, and better balance and physical function than the control group. 

Two high quality studies (Liao 2013, Moffet 2004) and one moderate quality study (Valtonen 2010) investigated supervised intensive exercise programs started two or more months after surgery (late stage post total knee arthroplasty) compared to no or less exercise. Liao et al compared a group who performed functional exercise supplemented with balance training to a group who performed functional training only. The exercise programs lasted eight weeks and started two months post-surgery. The group who received a combination of functional and balance exercises had better patient reported and performance-based outcomes of physical function. Moffet et al compared a group who received intensive functional training during eight weeks to a standard care group who received minimal rehabilitative care.  Pain and emotional health was significantly better in the intensive functional training group at 4 and 6 months, but the effects were no longer significant at the 12 months’ time point.  Valtonen et al compared a group who performed a high-intensity progressive aquatic resistance training of six week duration that started at least four months after surgery to a control group who did not exercise. The outcomes of both groups were similar. 
Early Stage Supervised Exercise Program: Pain
Limited evidence supports that a supervised exercise program during the first two months after total knee arthroplasty (TKA) decreases pain.
Limited Evidence Limited Evidence
One high quality study (Evgeniadis 2008) and one moderate quality study (Akbaba 2014) investigated supervised exercise programs started after hospital discharge compared to no exercise or minimal exercise during the first two months after surgery. Evgeniadis et al compared a group of patients post total knee arthroplasty who received a home exercise program of eight weeks (three times a week) that consisted of lower extremity strength training, to a group who did not receive supervised exercises. The exercise group had significantly better physical function and knee flexion and extension range of motion. Akbaba et al compared a group of patients with bilateral total knee arthroplasty who received a month of intensive supervised rehabilitation (two times a week for one hour) to a control group who received supervised rehabilitation once every 15 days. The intense supervised group had less pain and stiffness, and better balance and physical function than the control group.

Two high quality studies (Liao 2013, Moffet 2004) and one moderate quality study (Valtonen 2010) investigated supervised intensive exercise programs started two or more months after surgery (late stage post total knee arthroplasty) compared to no or less exercise. Liao et al compared a group who performed functional exercise supplemented with balance training to a group who performed functional training only. The exercise programs lasted eight weeks and started two months post-surgery. The group who received a combination of functional and balance exercises had better patient reported and performance-based outcomes of physical function. Moffet et al compared a group who received intensive functional training during eight weeks to a standard care group who received minimal rehabilitative care.  Pain and emotional health was significantly better in the intensive functional training group at 4 and 6 months, but the effects were no longer significant at the 12 months’ time point.  Valtonen et al compared a group who performed a high-intensity progressive aquatic resistance training of six week duration that started at least four months after surgery to a control group who did not exercise. The outcomes of both groups were similar. 

ACKNOWLEDGEMENTS

Guidelines Work Group:
Brian McGrory, M.D., M.S., Chair
Kristy Weber, M.D., Co-Chair
John A. Lynott, M.D. 
Vinod Dasa, M.D.
John C. Richmond, M.D. 
Charles Moore Davis III, M.D., PhD
Adolph Yates, Jr., M.D.
Atul F. Kamath, M.D.
Gregory Alexander Brown, MD, PhD
Tad L. Gerlinger, M.D. 
Sara Piva, PT, PhD
Tomas Villanueva, D.O., M.B.A., FACPE, SFHM
James Hebl, M.D. 


AAOS Guidelines Oversight Chair:
David Jevsevar, MD, MBA

AAOS Clinical Practice Guidelines Section Leader
Kevin Shea, MD

AAOS Council on Research and Quality Chair
Kevin J. Bozic, MD, MBA

Additional Contributing Members

James Keeney, M.D. 

AAOS Staff:
William Shaffer, MD, AAOS Medical Director
Deborah Cummins, PhD, Director of Research & Scientific Affairs
Jayson N. Murray, MA, Manager, Evidence-Based Medicine Unit
Patrick Donnelly, MA, Research Analyst, Evidence-Based Medicine Analyst
Nilay Patel, MA, Research Analyst, Evidence-Based Medicine Analyst
Peter Shores, MPH, Statistician, Evidence-Based Medicine 
Anne Woznica, MLS, Medical Librarian
Erica Linskey, Administrative Assistant, Evidence-Based Medicine Unit 
Kaitlyn Sevarino, MBA, Evidence-Based Quality and Value Coordinator

Former Staff
Ben Brenton, MSAE, 
Research Analyst, Evidence-Based Medicine Analyst

LEARN MORE ABOUT AAOS

The Future of OrthoGuidelines

FIND OUT MORE

The OrthoGuidelines Process