Five hundred and seventy-one patients who had pain in the anterior aspect of the knee that was believed to be secondary to lateral patellar tilting were entered into a rehabilitation protocol between July 1989 and January 1994. If physical examination revealed no associated abnormalities or internal derangements of the knee and if there was less lateral than medial patellofemoral cartilage space on the Merchant radiograph38, the patient was started on a home program of progressive strengthening of the quadriceps and adductor muscles and stretching of the hamstrings and the iliotibial band27,40. If there had been no improvement by six weeks, the limb was placed in a patellar stabilizing knee-sleeve and oral administration of anti-inflammatory agents was begun. If an additional six-week period of rest from the inciting activity and additional formal outpatient physical therapy, including patellar taping33, failed to relieve the symptoms, the patient became a candidate for operative treatment. Approximately one-third of the patients (thirty of ninety-one) who became candidates for operative intervention had had transient relief of the pain for two to three weeks after the patellar taping; the taping did not exacerbate the symptoms in any patient.
Patients who had had no improvement after six months of non-operative treatment, did not have a work-related injury or pending litigation, had unilateral pain in the anterior aspect of the knee, and had not had a previous patellar realignment or retinacular procedure were entered into the study. Eligible patients were told that both procedures had been found to relieve symptoms effectively in four of five patients and that there were advantages and disadvantages to both. Specifically, open lateral retinacular lengthening has the disadvantage of a less cosmetically appealing incision. Although arthroscopic lateral retinacular release has very little cosmetic effect on the tissues of the knee, postoperative hematoma, infection, weakness of the muscles about the knee, and continued pain (perhaps even more than the patient had had preoperatively) have been reported5,7,9,11,15,37. Patients who were randomized to treatment with open lateral retinacular lengthening were told that reported complications3,28 have included deep venous thrombosis, stitch abscess, and even neuroma caused by the cutting of nerves in the tissues around the knee. They also were informed that if they did not choose to participate in the study, they would be followed for six months postoperatively at six-week intervals and then on an annual basis, at no cost to them, for the duration of the study or for a minimum of two years. The patient then chose whether or not to be included in the study. After approval from the institutional review board had been obtained, patients were randomized to treatment with arthroscopic lateral retinacular release (Group I) if they had been born in an even-numbered year or to treatment with open lateral retinacular lengthening (Group II) if they had been born in an odd-numbered year.
One hundred and eight patients were eligible for the study, but seventeen refused to have the operation, leaving ninety-one patients (forty-four in Group I and forty-seven in Group II). One patient in each group was lost to follow-up, and three patients in Group II were later determined to have had previous lateral retinacular procedures and were excluded from this prospective analysis. Thus, the final study group consisted of eighty-six patients (eighty-six knees), with forty-three patients in each group. No comparison was made between these patients and the seventeen patients who had refused operative management.
Group I consisted of seventeen male patients (mean age, eighteen years; range, thirteen to fifty-six years) and twenty-six female patients (mean age, thirty-three years; range, fifteen to forty-seven years). Group II consisted of eighteen male patients (mean age, twenty-two years; range, thirteen to thirty-three years) and twenty-five female patients (mean age, thirty-four years; range, fourteen to fifty-nine years).
Group I comprised sixteen left knees and twenty-seven right knees, and Group II comprised twenty-three left knees and twenty right knees. The patients in Group I had had the symptoms for a mean of 29.8 months (range, six to 220 months), and the patients in Group II had had the symptoms for a mean of 26.4 months (range, six to ninety-six months). All of the patients were followed for two to six years. The mean duration of follow-up was 46.6 months (range, twenty-four to seventy-three months) for Group I and 43.1 months (range, twenty-four to seventy-two months) for Group II. With use of analysis of variance, no significant difference was found between the groups in terms of age (p = 0.57), gender (p = 0.83), or the preoperative duration of symptoms (p = 0.62).
Operative Technique
At the time of the index procedure, the knee was examined for associated abnormalities with the patient under anesthesia. An inability to rotate the lateral border of the patella anteriorly more than five millimeters was confirmed in all knees. An arthroscopic examination was then done, and any associated injuries were treated. I performed all of the procedures.
Arthroscopic Lateral Retinacular Release
Under arthroscopic visualization, the lateral retinaculum, the capsule, and the synovial tissue are incised longitudinally from the vastus lateralis muscle fibers to just beyond the tibial joint line (Fig. 1). The longitudinal full-thickness incision is carried out one centimeter lateral to the lateral border of the patella, with care being taken to reach but not to incise the arthroscopically visible fibers of the vastus lateralis muscle. The superior lateral geniculate artery is not protected and is cauterized when it is encountered. The release is considered complete when the lateral border of the patella can be rotated 90 degrees anteriorly, with the patella standing on its medial edge10.
Open Lateral Retinacular Lengthening
A six-centimeter longitudinal incision is made one centimeter lateral to the lateral border of the patella, beginning proximally at the level of the superior pole of the patella. The incision ends just a few millimeters proximal to the Gerdy tubercle. The dissection is carried out to the lateral retinaculum, which is incised longitudinally in line with the skin incision from the vastus lateralis muscle fibers down to the Gerdy tubercle. The deepest fibers of the retinaculum are left undisturbed with the joint capsule.
The dissection is then carried out laterally for two centimeters. A flap of lateral retinaculum superficial to the lateral aspect of the knee capsule and the deep retinaculum is raised from the fibers of the vastus lateralis superiorly to just beyond the tibial joint line. The deep fibers of the retinaculum and the superficial capsule then are incised two centimeters lateral to, and along the length of, the overlying retinacular incision (Fig. 2-A). This dissection can be difficult; any areas of penetration of the capsule or the synovial tissue are sutured with 2-0 absorbable sutures. The medial free edge of the capsule and the deep retinacular fibers then are sutured to the lateral free edge of the superficial lateral retinaculum with 2-0 absorbable sutures, thus effecting the lateral retinacular lengthening (Fig. 2-B).
Previous Operations
Eight patients (six female and two male) in Group I had had a total of eleven previous operations. Five patients had had one operation, and three patients (all female) had had two operations. The operations had included seven partial meniscectomies (four medial and three lateral) and three patellar shavings. Two patients had had excision of a medial synovial plica, one had had removal of a loose body, and one had had lysis of adhesions. (Some patients had had more than one procedure during an operation.)
Five patients (three male and two female) in Group II had had a total of six previous operations. The three male patients and one of the female patients had had one operation, and the other female patient had had two operations. The operations included two partial medial meniscectomies, two excisions of a medial synovial plica, one diagnostic arthroscopy, and one repair of the medial collateral ligament.
Concomitant Procedures
Twenty patients (47 per cent) in Group I had a total of twenty-five additional arthroscopic procedures concomitant with the lateral retinacular release. These procedures included seven excisions of a medial synovial plica, six procedures for the removal of a loose body, five chondroplasties of the medial femoral condyle, five partial meniscectomies (three medial and two lateral), and two excisions of osteophytes.
Twenty-one patients (49 per cent) in Group II had a total of twenty-seven additional procedures concomitant with the open lateral retinacular lengthening. All but one of these procedures was done arthroscopically. The one open procedure was a reconstruction of the medial collateral ligament, performed with use of a semitendinosus autogenous graft for the treatment of symptomatic, chronic laxity of the medial collateral ligament after a previous repair of the ligament had failed. The twenty-six arthroscopic procedures included eight chondroplasties of the medial femoral condyle, seven excisions of a medial synovial plica, five procedures for the removal of a loose body, and six partial meniscectomies (three medial and three lateral). With use of analysis of variance, no significant difference could be detected between the groups with regard to the percentage of patients who had had a previous operative proceedure (p = 0.37) or a concomitant operative procedure (p = 0.83).
Postoperative Regimen
Both groups of patients were discharged from the outpatient surgical facility on the day of the procedure. They were encouraged to not use crutches and to bear full weight immediately. They were instructed to keep the involved knee elevated and to apply ice at twenty-minute intervals for the first forty-eight hours after the operation. The patients were reinstructed with regard to straight-leg-raising quadriceps-strengthening exercises as well as stretching and strengthening exercises for the hamstrings, which were to begin the day after the procedure.
The patients were seen in my office two weeks postoperatively. At that time, progressive-resistance quadriceps-strengthening exercises were reemphasized and a running-and-cycling exercise regimen was begun. The patients were allowed to return to competitive athletic activity after the sixth postoperative week.
Orthopaedic evaluation was performed two, six, and twelve weeks postoperatively and annually thereafter. Merchant radiographs were made preoperatively and on an annual basis postoperatively. A frame was constructed so that the knee could be flexed reproducibly to 45 degrees and a tangential patellofemoral (Merchant) radiograph38 could be made 30 degrees distal to the horizontal plane. The congruence angle was measured on the Merchant radiograph. This angle is formed by the bisector of the femoral trochlear sulcus, which serves as the zero reference line, and the line between the most distal point of the femoral sulcus and the most distal point on the articular ridge of the patella. Values that are medial to a zero reference line are designated as negative and those that are lateral, as positive38.
Function was evaluated at yearly intervals, according to the system of Lysholm and Gillquist32 as modified by Tegner and Lysholm45. I (the treating physician) was not involved in the most recent follow-up evaluation in order to avoid biasing the patient's interpretation of the result. It has been shown that there are discrepancies between patients' and physicians' evaluations of the outcomes of orthopaedic procedures31. This discrepancy can increase when the patient is not satisfied with the outcome.
Open-chain testing of the extensors of the knee was performed with an isokinetic dynamometer (Lido, West Sacramento, California) at yearly intervals postoperatively. Peak torque and total work were measured in foot-pounds (one foot-pound equals 1.356 newton-meters). Concentric extension of the knee was tested between 90 and 0 degrees of flexion, at isokinetic velocities of 60, 180, and 240 degrees per second, with the patient seated and the waist and the distal aspect of the thigh strapped for stabilization. Deficits were recorded at all of the levels that were tested.
Closed-chain testing of the extensors was performed with a closed-chain isokinetic dynamometer (Lido Linea, West Sacramento, California), also at yearly intervals. Peak torque and total work were measured in foot-pounds. Bilateral reciprocal concentric leg-press was tested between 90 and 0 degrees of flexion of the knee, at isokinetic velocities of ten, twenty, and thirty inches (25.4, 50.8, and 76.2 centimeters) per second, with the patient seated and the waist strapped for stabilization. Deficits were recorded at all of the levels that were tested.
The range of motion of the knee and the circumference of the thigh were measured at annual intervals postoperatively, and these measurements were compared with those for the uninvolved limb. Any deficit in the range of motion compared with that on the contralateral side was recorded. The circumference was measured with a tape at the joint line and three and seven inches (7.6 and 17.8 centimeters) proximal to the joint line, and any deficit compared with the circumference on the contralateral side was recorded. The greatest recorded deficit was used for statistical analysis.
Statistical Analysis
Performance measures were assessed with analysis of variance, the Student t test of means, and the Wilcoxon rank-sum test, and the level of significance was set at p = 0.05. Analysis of variance and the Student t test assume that the measures are normally distributed; however, with small sample sizes, this assumption can be easily violated. The Wilcoxon rank-sum test is non-parametric, meaning that it makes no assumptions about the underlying distribution of the measures being tested. This test produces more conservative estimates of significance and therefore is relied on whenever there is a lack of agreement between its result and that obtained with analysis of variance.
Duration of Preoperative Symptoms
The patients in Group I had, on the average, less strength at the time of follow-up, as measured on open-chain testing at 60 and 180 degrees per second, when the symptoms had been present for a longer duration preoperatively. They also had less strength as measured on closed-chain testing at ten, twenty, and thirty inches (25.4, 50.8, and 76.2 centimeters) per second. In addition, the peak force was negatively associated with the duration of symptoms. These associations were all significant at the 95 per cent confidence level or better.
No significant association could be detected between the duration of symptoms and any of the outcome measures for the patients in Group II.
Chondromalacia Patellae
Twenty-two patients (51 per cent) in Group I had no arthroscopically demonstrable patellar chondromalacia41,42. Of the remaining twenty-one patients (49 per cent), one—a thirty-year-old woman—had grade-I changes25,41,42 in the midline raphe; six (14 per cent)—four male patients who were sixteen to forty-one years old and two female patients who were both seventeen years old—had grade-II chondromalacia of the midline raphe or the medial patellar facet; twelve (28 per cent)—five male patients who were fifteen to thirty-seven years old and seven female patients who were thirty to forty-seven years old—had grade-III chondromalacia of the facets, the midline raphe, or the entire patella; and two (5 per cent)—both female patients who were seventeen and forty-seven years old—had grade-IV chondromalacia of the entire patella.
Thirty-five patients (81 per cent) in Group II had no arthroscopically evident chondromalacia patellae. Of the remaining eight patients (19 per cent), two (5 per cent)—both female patients who were thirty-six and forty-two years old—had grade-II chondromalacia of the lateral patellar facet; five (12 per cent)—female patients whose ages ranged from forty-six to fifty-nine years old—had grade-III chondromalacia of either the medial or the lateral facet or the midline raphe; and one (2 per cent)—a boy who was thirteen years old—had grade-IV chondromalacia of the medial facet.
The rate of arthroscopically demonstrable chondromalacia patellae was significantly greater in Group I (p = 0.005, analysis of variance) (Table I).
Return to Pre-Injury Level of Activity
Twenty-three patients (thirteen female and ten male [53 per cent]) in Group I and twenty-two patients (twelve female and ten male [51 per cent]) in Group II had participated in sports activity that required cutting and pivoting before the onset of the symptoms. At the latest follow-up examination, forty (93 per cent) of the forty-three patients in Group I and all forty-three patients in Group II had resumed athletic activity at a level that they considered to be equal to or better than the level that they had achieved before the onset of the symptoms. Three patients (7 per cent) in Group I had given up athletic activity. With the numbers available for study, no significant difference with regard to return to sports activity was detected between the two groups (p = 0.08, analysis of variance) (Table I).
All three patients who had given up athletic activity had done so because of pain in the anterior aspect of the knee. One of these patients, a thirty-year-old woman who had given up league softball, had had grade-III chondromalacia of the midline raphe at the time of the operation. Postoperatively, she had a score of 84 points on functional testing32,45 and a 0-degree congruence angle as seen on the Merchant radiograph. At the latest follow-up evaluation, sixty-two months postoperatively, she had a 30 per cent quadriceps deficit on isokinetic testing and an eighteen-millimeter decrease in the circumference of the thigh. The second patient was a thirty-three-year-old woman who had given up jogging. She had had no demonstrable chondromalacia at the time of the operation. Postoperatively, she had a score of 89 points on functional testing, a -10-degree congruence angle, and no deficits on isokinetic testing despite a ten-millimeter decrease in the circumference of the thigh. At the latest follow-up evaluation, forty-six months postoperatively, she had lost 10 degrees of terminal flexion. The third patient was a forty-one-year-old man who had given up jogging. He had had grade-II chondromalacia of the midline raphe at the time of the operation. Postoperatively, he had a score of only 48 points on functional testing, a -3-degree congruence angle, and a 16 per cent quadriceps deficit on isokinetic testing, with no atrophy or loss of range of motion.
The seventeen-year-old girl in Group I who had had grade-IV chondromalacia resumed playing collegiate-level volleyball postoperatively, and the forty-seven-year-old woman who had had grade-IV chondromalacia, also in Group I, resumed jogging. The thirteen-year-old boy in Group II who had had grade-IV chondromalacia of the medial facet resumed playing junior-high-school basketball, but he was subsequently not selected for the high-school team.
Testing with the Dynamometer
At the latest follow-up evaluation, nineteen patients (44 per cent) in Group I had a quadriceps deficit on open-chain testing. These deficits reflect the greatest side-to-side differences in total work at 60, 180, and 240 degrees per second. The deficit was 10 per cent or less in fifteen patients, 11 to 20 per cent in three, and 30 per cent in one. Sixteen patients (37 per cent) in Group II had a quadriceps deficit on open-chain testing; all of these deficits were less than 10 per cent.
Sixteen patients (37 per cent) in Group I had a quadriceps deficit, ranging from 1 to 25 per cent, on closed-chain testing. These deficits reflect the greatest side-to-side differences in total work at ten, twenty, and thirty inches (25.4, 50.8, and 76.2 centimeters) per second. The deficit was 10 per cent or less in nine patients, 11 to 20 per cent in four, and 21 to 30 per cent in three. Thirteen patients (30 per cent) in Group II had a quadriceps deficit, ranging from 2 to 11 per cent, on closed-chain testing.
Sixteen patients (37 per cent) in Group I and nine patients (21 per cent) in Group II had a quadriceps deficit on open-chain testing of peak torque. In Group I, the deficit was 10 per cent or less in nine patients, 11 to 20 per cent in three, and 21 to 30 per cent in four. Seven patients (16 per cent) in Group II had a deficit of 10 per cent or less on open-chain testing, and two (5 per cent) had a deficit of 15 per cent.
There was a trend for the patients in Group II to perform better than those in Group I on open-chain testing at 60 and 180 degrees per second and on testing at peak torque (but not at the 95 per cent confidence level). At 60 degrees per second, the p value for the difference between the groups was 0.09, according to analysis of variance, and at 180 degrees per second, the p value was 0.10. The p value for the difference between the groups on testing of peak force was 0.10, according to analysis of variance (Table I).
No significant differences could be detected on closed-chain testing at ten, twenty, or thirty inches (25.4, 50.8, or 76.2 centimeters) per second.
Range of Motion
In Group I, four female patients (9 per cent), whose ages ranged from sixteen to forty-seven years old, had a loss of flexion ranging from 4 to 12 degrees. No patient in Group I had a loss of extension.
In Group II, five patients (12 per cent)—a fourteen-year-old boy and four female patients ranging in age from fourteen to forty-three years old—had a loss of flexion ranging from 2 to 23 degrees. No patient in Group II had a loss of extension.
Only one patient in each group had a loss of more than 10 degrees of flexion.
No significant difference could be detected between the two groups with regard to loss of motion (p = 0.75, analysis of variance) (Table I). The precision of the measurement was ±5 degrees.
Circumference of the Thigh
At the latest follow-up evaluation, twelve patients (five male and seven female; 28 per cent) in Group I had a decrease in the circumference of the thigh ranging from five to twenty-three millimeters compared with the uninvolved thigh.
In Group II, seven patients (four male and three female; 16 per cent) had a decrease in the circumference ranging from five to eighteen millimeters.
No significant difference could be detected, with regard to the change in circumference, between the two groups (p = 0.31, analysis of variance) (Table I). The precision of the measurement was ±5 millimeters.
Complications and Subsequent Operative Procedures
Four patients (9 per cent) in Group I had postoperative complications, including a contracture of the iliotibial band in a thirty-three-year-old woman, a hematoma in a forty-seven-year-old woman, and a superficial infection at the site of an arthroscopic portal in a forty-seven-year-old woman. The fourth patient, a sixteen-year-old boy, had a hematoma that he drained himself with a needle at home; an infection subsequently developed in the knee joint.
Three patients (7 per cent) in Group II had postoperative complications, including a stitch abscess in a sixteen-year-old girl, a keloid of the skin incision in a fourteen-year-old girl, and an inability to flex the knee more than 90 degrees four weeks postoperatively in a twenty-three-year-old man.
Two patients (5 per cent) in Group I had an additional operation. The patient who had the contracture of the iliotibial band had a release of the band thirty-two months postoperatively, and the patient who had the joint infection had arthroscopic irrigation and débridement three weeks postoperatively. Both of these patients were able to return to their previous level of activity after the additional procedure.
In Group II, only one patient (2 per cent), the man who had early loss of flexion, had an additional procedure. The knee was manipulated to full flexion with the patient under anesthesia, and the limb was placed in a splint with the knee in 125 degrees of flexion. The patient subsequently returned to his previous level of activity.
The difference in the numbers of patients in each group who had complications (p = 0.70) and additional procedures (p = 0.31) could not be shown to be significant with analysis of variance (Table I).
Findings on Merchant Radiographs
The patients in Group I had a mean congruence angle of +3 degrees (range, +48 to -23 degrees) preoperatively compared with 0 degrees (range, +36 to -24 degrees) postoperatively (Figs. 3-A and 3-B). The average correction, or medialization, was 4.8 degrees (mean, 2 degrees; range, 13 degrees of lateralization to 34 degrees of medialization).
The patients in Group II had a mean congruence angle of -2 degrees (range, +30 to -40 degrees) preoperatively compared with -7 degrees (range, +25 to -32 degrees) postoperatively (Figs. 4-A, 4-B, 5-A and 5-B). The average correction was 3.6 degrees (mean, 1 degree; range, 23 degrees of lateralization to 56 degrees of medialization).
There was a significant difference in the preoperative and postoperative congruence angles between the groups. The patients in Group I had less medialization (p = 0.02, analysis of variance) than did those in Group II; however, no significant difference could be detected with regard to the amount of correction (medialization) achieved with either procedure. No significant association could be detected between the amount of correction and the functional outcome (p = 0.1, Wilcoxon rank-sum test).
Functional Testing
According to the knee-rating system of Lysholm and Gillquist32 as modified by Tegner and Lysholm45, twenty-one knees (49 per cent) in Group I had a score of 90 points or more compared with twenty-five knees (58 per cent) in Group II. Twelve knees (28 per cent) in Group I had a score between 80 and 89 points compared with thirteen knees (30 per cent) in Group II. Four knees (9 per cent) in Group I had a score between 70 and 79 points compared with five knees (12 per cent) in Group II. In Group I, two knees (5 per cent) had a score between 60 and 69 points; one knee (2 per cent), a score of 52 points; and three knees (7 per cent), a score between 40 and 49 points. No knee in Group II had a score of less than 70 points.
The mean knee score in Group I was 89 points (average, 85 points; range, 40 to 100 points), and the mean score in Group II was 91 points (average, 91 points; range, 70 to 100 points). Thirty-three patients (77 per cent) in Group I had a good or excellent result (a score of 80 points or more) compared with thirty-eight patients (88 per cent) in Group II.
The results for the knee scores are inconclusive. According to analysis of variance, this self-reported score was significantly different between the two groups at the 97 per cent confidence level (p = 0.028) (Table I).
Duration of Follow-up
Although there seemed to be a definite trend toward improved function of the knee in association with longer durations of follow-up, no significant association between the duration of follow-up and improvement in the outcome measures could be detected in either group (p = 0.24).