Since its introduction in the early 1970s, numerous surgical approaches have been described for doing a condylar knee replacement. Approaches such as the midvastus,2,3 subvastus,6 and lateral parapatellar9 have had their advocates; however, the incision most frequently used in the United States has been the median parapatellar,1,13 with separation of the rectus femoris from the vastus medialis muscle and eversion of the patella. The results of knee replacement using this type of incision have been excellent. However, recuperation from knee replacement has been arduous and often painful. Attaining flexion after surgery has been a daunting task for the patient and surgeon.
Minimally invasive approaches have been used to treat a variety of disparate surgical problems, including anterior cruciate ligament (ACL) repair, gall bladder disease, and gynecologic disorders. The premise has been that by limiting the amount of disruption of the soft tissues during these procedures, the time for recuperation could be sped up, the blood loss could be diminished, and the postoperative pain could be decreased. These were the premises that were used in this study of minimally invasive total knee replacement. Could we do the surgery through an incision that did not extensively disrupt the quadriceps and suprapatellar pouch, and still get adequate exposure for proper component position and ligament balancing, and would the result be less pain after surgery, a faster recuperation, and possibly less blood loss? To evaluate these premises we did this study with a retrospective control group. Because the parameters we wanted to measure were those attendant to the perioperative period, we accumulated data beginning before and extending for 3 months after surgery.
MATERIALS AND METHODS
Beginning in mid-2002, prospectively, a minimally invasive midvastus approach was proposed for all patients having a primary knee replacement. Patients who had a previous open procedure on their knee were excluded, however, there were no exclusions based on body mass index (BMI), gender, or deformity. In three patients, all heavily muscled males, the surgical procedure had to be modified because of inability to obtain adequate exposure. Thirty-two patients who had knee replacement using this minimally invasive incision were included in the minimally invasive surgery (MIS) group.
The immediate 32 primary knee replacements done before the use of the mini-midvastus incision were evaluated. In six patients, a modified midline incision had been used and these patients were subsequently excluded. The resultant 26 knee replacements were included in the standard group.
The average age was 70 years (range, 62–78 years) for the MIS group and 68 years (range, 58–78 years) for the standard group. There were 12 men and 20 women in the MIS group and 11 men and 15 women in the standard group. The diagnosis was primary osteoarthritis (OA) in all but one patient in each group. The average BMI was 29 (range, 24–34) in the MIS group and 30 (range, 26–34) in the standard group.
Preoperative Knee Society scores were 70 points (SD ± 19 points) for the MIS group and 76 points (SD ± 26 points) for the standard group. Lahey Clinical scores5 were 17 points (SD ± 5 points) and 15 points (SD ± 5 points) for the MIS group and the standard group, respectively. The flexion was 112° (SD ± 7°) in the MIS group, compared with the 107° (SD ± 10°) in the standard group. The average preoperative standing alignment measured on a standing AP radiograph was 6° degrees varus (range, 20° valgus–15° varus) in the MIS group, and 6° varus (range, 19° valgus–15° varus) in the standard group. The preoperative comparison between the two groups is shown in Table 1.
For the MIS group, after tourniquet inflation, and after the knee was flexed, a straight skin incision was made. It began 2 cm distal to the joint line (just medial to the tibial crest), crossed the medial ⅓ of the patella, and extended for 2 cm proximal to the superior pole of the patella (Fig 1). The exact length of the skin incision was dependent on the superior-inferior dimension of the patella and varied between 9 and 13 cm.
The median parapatellar retinaculum was incised 1 cm medial to the patella. Proximally, the deep incision extended 2 cm proximal to the superior pole of the patella and then medially into the fibers of the vastus medialis obliquus muscle for 2 cm (Fig 2). With the knee kept in maximum flexion, the medial capsular structures were elevated from the tibia as the tibia was externally rotated. The exposure continued to the level of the midcoronal plane.
Next, the knee was placed in approximately 45° flexion and a portion of the fat pad was removed. The patella was displaced laterally, but not everted, and was held in position with a PCL retractor placed laterally around the distal portion of the lateral femoral condyle and lateral tibial plateau. The knee was again maximally flexed and the tibia was subluxed forward. Remnants of the menisci were removed, and the ACL, if present.
In the standard group, a median parapatellar incision was made extending for approximately 10 cm in the interval just central to the junction of the rectus femoris and vastus medialis muscles. The patella was everted before flexing the knee.
For both groups, intramedullary guides and anterior referencing cutting blocks were used for positioning of the femoral resections. The goal was to place the femoral component in 5° valgus as referenced to the femoral intramedullary axis in the coronal plane and at 180° as referenced to the femoral intramedullary axis in the lateral plane. An intramedullary guide was used for placement of the tibial resection guide. The goal was to resect the tibia at 90° to its anatomical axis in the coronal plane, and with a 3° downslope in the lateral plane. The placement of the inset patellar implant was done using a reamer clamp and guide with an attempt to remove as much patellar cartilage and bone as the thickness of the implant being used.
The instruments that were used in the patients in the standard group were modified for use in the patients in the MIS group (Fig 3).
In both groups, soft tissue balancing was done in a standard manner. The PCL was released from the posterior aspect of the proximal tibia for 2 to 3 cm below the joint line. Patellar tracking was ascertained using the no touch test with the retinaculum tensed proximally. The implant used for the arthroplasties was the Genesis II (Smith and Nephew, Memphis, TN) with a deep-dish conforming PE.11 All the implants were cemented. No lateral retinacular releases were required in this series.
All the operations were done by the senior author in a vertical laminar flow environment, with surgeons using body exhaust suits. Cephalosporin (1 gm every 8 hours for 24 hours total) was used as a perioperative antibiotic in all patients, except for patients with a documented cephalosporin allergy.
A combined epidural and femoral nerve block was used for anesthesia in all patients. Postoperatively, a closed suction drainage was used for 24 hours.
The epidural block was continued for 2 days. Patients received a basal level of analgesics and were allowed to increase this amount by using a self-administered patient- controlled analgesia (PCA) pump (CADD-PRIZM, Simos Delta, St. Paul, MN). The patients were started on knee flexion exercises using a continuous passive motion (CPM) machine within the first few hours after surgery, beginning at 60° flexion. Patients were allowed to walk, bearing weight on the affected leg, the day after surgery, using a knee immobilizer for support and a walker for balance. On the third postoperative day most patients were advanced from a walker to a cane in the contralateral hand and the knee immobilizer was discontinued.
We evaluated the patients preoperatively, using the Knee Society rating system, both clinically and radiographically4,7 and the data were entered into the computer database. During the hospitalization, a physical therapist measured the patient’s passive range of flexion and the time required to achieve specific rehabilitation milestones. The amount of pain medication used in the epidural catheter, and subsequently used orally, was recorded. Closed suction blood drainage amounts were recorded as well. The pain level was assessed by the visual analog scale (VAS) on an hourly basis while the patients were awake. The patients were re-evaluated at the first and second postoperative visits, 6 weeks and 3 months postoperatively, using the Knee Society rating system both clinically and radiographically, and a patient satisfaction scoring system. The radiographs were evaluated by an independent radiologist who was blinded to the type of skin incision that had been used.
The data distributions of all variables were examined. The relationship of each demographic and clinical variable to the outcome variables was examined. T-tests were usedfor group comparisons of normally distributed continuous data. Nonparametric tests, such as the Mann-Whitney test, were used for nonparametric data and contingency table analysis was used for categorical data. Relationships between independent variables were also assessed at the bivariate level.
Statistically, general linear models were used to test variables that showed statistically significant relationships to the outcomes in bivariate models. The alpha level was set at 0.05. This study, with a comparison to a retrospective control group, was submitted to and approved by the Institutional Review Board of our institution.
The tourniquet times were 58 minutes (SD ± 11 minutes) in the MIS group and 51 minutes (SD ± 8 minutes) for the standard group (p = 0.01). No lateral retinacular release was done in either group.
The average VAS per hour was significantly lower in the MIS group, especially at the day of surgery (p = 0.03) and the first postoperative day (p = 0.01). The total PCA volume used by the patients in the MIS group (255 mL, SD ± 119 mL) was lower than the volume used in the standard group (301 mL, SD ± 119 mL). In addition to the narcotic pain medication used from the epidural catheter, the total dose of orally administered narcotic pain medication was standardized to morphine sulfate equivalents (9) in mg and recorded daily. The daily dose of morphine sulfate equivalents used was always lower in the MIS group, being significantly (p = 0.008) lower on postoperative Day 2, at which point the epidural catheter was discontinued. The total dose of morphine sulfate equivalents used by the MIS and standard groups were 55 mg (SD ± 70 mg) and 118 mg (SD ± 120 mg), respectively (p = 0.01, Table 2) (Fig. 4).
The estimated total blood loss (from the drains) was 713 mL (SD ± 289 mL) for the MIS group and 573 mL (SD ± 171 mL) for the standard group (p = 0.04).
The range of flexion of each group was assessed on a daily basis. The flexion in the MIS group was always greater than in the standard group (Table 3) (Fig. 5). Only one patient in the standard group reached flexion greater than 80° by postoperative Day 3, whereas 20 patients of the MIS group reached 80° or greater flexion by postoperative Day 2.7 ±1.1 days (p < 0.001). Analysis of variance (ANOVA), done through postoperative Day 3, showed a significantly (p = 0.0001) higher knee flexion in the MIS group. To eliminate the bias that the MIS group had preoperatively, ANOVA with preoperative passive flexion as a covariate, the MIS and standard arthrotomy as the between factors, and postoperative day flexion as the dependent variable was done. The flexion in the MIS group was higher throughout the hospital stay and significantly higher through postoperative Day 4 (Table 4). Survival analysis for the time to reach 80° flexion was 4.4 days for the MIS group and longer than 6 days for the standard group; this difference was significant (p < 0.0001).
At the 6-week postoperative followup, the average passive flexion was 115° (SD ± 9°) for the MIS group and 100° (SD ± 10°) for the standard group (p = 0.02). The change in the knee scores was significant (p = 0.03), favoring the MIS incision. There were no significant differences between the groups regarding pain level, extension, and function scores at the postoperative 6-week followup (Table 5). By the 3-month followup the groups had equalized with each other (Table 6).
Femoral and tibial implant alignments were measured in both sagittal and frontal planes. There were no significant differences in component alignments between the groups, with excellent position of the components in both groups (Table 7). Likewise, excellent stability to varus and valgus and AP stresses were found in both groups, with no significant difference between them.
During the last 30 years, there have been multiple surgical approaches suggested for doing a total knee replacement. These approaches have included the median parapatellar, lateral parapatellar, midvastus, and subvastus. All of these have entailed either disruption of the quadriceps mechanism or eversion of the patella, or, in most cases, both. During these 30 years, the instruments used for doing these knee replacements have been large, necessitating the use of large incisions for their proper placement.
Using these techniques, the ultimate results of knee replacement have been excellent. However, difficulties with regaining motion, suprapatellar adhesions, and pain have made the recuperation arduous and often painful. The underlying premise of the minimally invasive surgical exposure used in this study was that by not everting the patella and disrupting the suprapatellar pouch, recuperation might be facilitated. This was not a new concept, having been espoused by surgeons who do unicompartmental replacements.14 The question was, however, whether this approach could be applied to patients having total knee replacements.
Minimally invasive has become a general use term to describe operations done through smaller incisions with less disruption of the deep tissues. Minimally, however, is an improper word to describe this surgical approach. Minimally, as defined by Webster’s dictionary,15 is an adverb meaning the least possible, or barely adequate. The surgical incision used in this study is neither of these definitions. A more descriptive and accurate term might be less invasive surgery. Likewise, the lessening of the invasiveness of the surgery is not primarily related to the skin. Although a smaller skin incision often is a secondary benefit, the primary advantage of this procedure is the less invasive handling of the quadriceps muscle and deep tissues.
Using a smaller incision required some modification of the surgical instrumentation and an understanding of how the soft tissues about the knee move with flexion and extension. When the knee was extended, the soft tissue window moved proximally, whereas when the knee was flexed, it moved somewhat distally. Therefore, when doing portions of the surgery requiring exposure of the supracondylar aspect of the femur (such as resecting the anterior surface of the femoral condyles), the knee was extended. For portions of the surgery that were related to resection and exposure of the tibial plateau, the knee was flexed; in this manner the soft window moved distally.
A standard PCL retractor held the patella laterally displaced, but not everted. This decreased the tension on the patellar tendon and decreased the change for its avulsion or damage.
The mini-midvastus split was normally only 2 cm long. This placed it at a distance from the branch of the femoral nerve to the vastus medialis obliquus, which could be damaged in a longer midvastus incision.
Obesity was not a contraindication to using this incision, although it was not attempted in morbidly obese patients. On occasion, in the heavily muscled patient this incision may be inadequate and it will be extremely difficult to displace the quadriceps mechanism and displace the patella laterally. In our study, in those cases the incision was extended another 2 to 3 cm into the muscle and the skin incision was lengthened proximally as well.
There was an average increase of 7 minutes in the time that the tourniquet remained inflated in the MIS group, compared with the standard group. This difference did not appear to represent a learning curve because it existed in both the earlier and later patients in this study. This increase probably represents the extra time required to change positions of the knee during surgery and to place the retractors. There were no complications associated with this increased time.
We had hypothesized that by using a less invasive surgical approach, the amount of blood lost postoperatively would be decreased. We did not find that to be the case, however. Accepting the limitations attributable to using absolute amounts of drainage as the measure of blood loss, patients in the MIS group showed a slight trend towards increased output, however this was not significant. The patients in the MIS group had a higher rate of return of flexion, and this might have caused the small increase in wound drainage in that group.10
A reduction in pain was a major salutary effect of the MIS approach. Visual analog scores were significantly lower, as were the amounts of narcotic medications required in the epidural catheter for the patients with the MIS incision. This salutary effect on pain relief was carried forth after the epidural catheter was removed. It was possible to convert the amount of analgesics used orally after removal of the epidural catheter to morphine sulfate equivalents.8 The MIS group used lower amounts of morphine sulfate equivalents of oral medication than the standard group, and the difference was highly significant (p = 0.008).
A second beneficial effect of the minimally invasive approach was related to the rate of regaining flexion after surgery. By the second postoperative day, more than 70% of the patients in the MIS group had achieved more than 80° flexion, whereas in the standard group this degree of flexion was usually not obtainable until the fourth postoperative day. Because the MIS group had a slightly greater flexion arc preoperatively, an ANOVA evaluation using the preoperative passive flexion as a covariate and the postoperative flexion as the dependent variable was done. The MIS continued to provide a more rapid return of flexion. This translated into the patients in the MIS group reaching their functional milestones, which would have permitted discharge from the hospital, approximately 20% faster than the cohort of patients who had a standard incision.
The patients in the MIS group continued to have greater flexion than the patients who had a standard incision at the 6-week evaluation, however, the amount of passive flexion appeared to merge between the two groups at 3 months.
All these beneficial effects would be obviated if it would have been shown that using a small incision resulted in implants that were not placed in accurate positions. This was not the case, however. We found no significant difference in component alignment when studying either means or outliers between the MIS and standard incision groups, with almost all patients in both groups having implant positions within the limits of what was sought using the instrumentation systems. Likewise, excellent ligamentous balancing in both the sagittal and coronal planes was found in both the MIS and Standard groups, with no significant difference between the two groups.
The ability to do a straight leg raising (SLR) maneuver postoperatively might be used as a marker for return of quadriceps function. The patients who had surgery in this study, however, received a femoral nerve block as part of their anesthesia. The depth and duration of that block made any comparison of SLR times unreliable.
There were no skin complications in the MIS group. However, one patient, for whom we used a minimally invasive midvastus incision just after the study enrollment was ended, developed a minor skin necrosis distally measuring 1 × 2 cm. This healed with local skin care without any evidence of infection. The skin incision must be made long enough so that there is not undue tension on it. The area of most concern is distally, where there is little soft tissue underlying the skin. The surgeon’s threshold for lengthening the skin incision in this area should be very low. To the contrary, it has not been necessary in any patient to extend the incision proximally, rather relying on the movable window concept to adequately expose the anterior femur.
A prospective study, in which patients are randomized into either a MIS or standard incision group, would theoretically be the most optimum way of obtaining data. However, the reticence of patients in the primary surgeon’s practice to participate in such a randomized evaluation led to the evaluation that was used in the current study. Likewise, the parameters measured in this study are those related to the immediate postoperative period. For that reason, the authors think that data collected for the first three peri-operative months was appropriate. The numbers of patients chosen for the study were based on an independent statistician’s evaluation, so that we could achieve a power of 0.81 with an alpha of 0.01.
1. Dalury DF, Jiranek WA: A comparison of the midvastus and paramedian approaches for total knee arthroplasty. J Arthroplasty 14:33–37, 1999.
2. Engh GA, Holt BT, Parks NL: A midvastus muscle-splitting approach for total knee arthroplasty. J Arthroplasty 12:322–331, 1997.
3. Engh GA, Parks NL: Surgical technique of midvastus arthrotomy. Clin Orthop 351:270–274, 1998.
4. Ewald FC: The Knee Society total Knee arthroplasty roentgenographic evaluation and scoring system. Clin Orthop 248:9–12, 1989.
5. Healy WL, Iorio R, Ko J, Appleby D, Lemos DW: Impact of cost reduction programs on short-term patient outcome and hospital cost of total knee arthroplasty. J Bone Joint Surg 84A:348–353, 2002.
6. Hofmann AA, Plaster RL, Murdock LE: Subvastus (Southern) approach for primary total knee arthroplasty. Clin Orthop 269:70–77, 1991.
7. Insall JN, Dorr LD, Scott R: Rationale of the Knee Society clinical rating system. Clin Orthop 248:13–14, 1987.
8. Kanamiya T, Whiteside LA, Nakamura T, et al: Ranawat Award paper. Effect of selective lateral release on stability in knee arthroplasty. Clin Orthop 404:24–31, 2002.
9. Keblish PA: The lateral approach for total knee arthroplasty. J Knee Surg 16:62–68, 2003.
10. Knoben JE, Anderson PO: Handbook of Clinical Drug Data. Ed 7. Drug Intelligence Publishing Co, Hamilton, IL 598-600, 1993.
11. Lachiewicz PF: The role of continuous passive motion after total knee arthroplasty. Clin Orthop 380:144–150, 2000.
12. Laskin RS, Maruyama Y, Villaneuva M, Bourne R: Deep-dish congruent tibial component use in total knee arthroplasty. A randomized prospective study. Clin Orthop 380:36–44, 2000.
13. Parentis MA, Rumi MN, Deol GS, et al: A comparison of the vastus splitting and median parapatellar approaches in total knee arthroplasty. Clin Orthop 367:107–116, 1999.
14. Price AJ, Webb J, Topf H, et al: Rapid recovery after oxford unicompartmental arthroplasty through a short incision. J Arthroplasty 16:970–976, 2001.
15. Webster’s Ninth New Collegiate Dictionary: Merriam-Webster INC, Springfield, MA, 756, 2002.