Between July 1976 and June 1978, the senior one of us (R. C. J.) performed 357 total hip replacements in 320 patients (135 men [152 hips] and 185 women [205 hips]). The average age of the patients at the time of the index arthroplasty was sixty-nine years (range, twenty-four to eighty-eight years). For the 130 patients who were alive at least fifteen years postoperatively, the average age at the time of the index arthroplasty was sixty-two years (range, twenty-four to eighty-eight years).
Preoperatively, the clinical evaluation included documentation of the level of activity, range of motion of the hip, level of pain, gait pattern, and level of performance of activities of daily living. An anteroposterior radiograph of the pelvis was made. The diagnosis was osteoarthrosis for 265 (74 per cent) of the 357 hips, post-traumatic degenerative joint disease for fifty (14 per cent), degenerative osteoarthrosis secondary to congenital dysplasia of the hip for sixteen (4 per cent), rheumatoid arthritis for nine (3 per cent), septic arthritis for seven (2 per cent), slipped capital femoral epiphysis for three (less than 1 per cent), Legg-Calvé-Perthes disease for three (less than 1 per cent), osteonecrosis for two (less than 1 per cent), and poliomyelitis and Paget disease for one hip each (less than 1 per cent for each). One hundred and ninety-four arthroplasties were performed on the right hip and 163 were performed on the left hip.
A previous operation had been done on fifty-three hips (15 per cent) before the index arthroplasty. These procedures included fixation of a fracture (twenty-five hips), a cup arthroplasty (fourteen hips), a hemiarthroplasty without or with cement (eight and two hips, respectively), an osteotomy (two hips), and an arthrodesis because of a previous infection (two hips). Five of these hips had had an additional previous procedure: an osteotomy before a cup arthroplasty was done in two hips, and incision and drainage, pinning before an osteotomy, and a second osteotomy were done in one hip each.
The operative technique involved a lateral approach, an osteotomy of the greater trochanter, and a complete capsulectomy. The acetabulum was prepared by placement of three one-centimeter-diameter holes in the ilium, ischium, and pubis. Cement was introduced in the doughy stage, and the component was placed as inferiorly and medially as possible. After it became available (in 1978), a plunger was used to pressurize the acetabular cement. The femoral canal was prepared by removal of all loose cancellous bone and meticulous drying. A cement plug was placed distally, and the cement gun was used to introduce cement in a retrograde manner. No antibiotics were used prophylactically. The trochanter was reattached as far laterally as possible, with use of three stainless-steel wires (two vertical and one horizontal). A body-exhaust system was not used, and the procedures were performed in a contemporary (1975) operating room.
Postoperatively, the patients were managed with bed rest on the first day, followed by partial weight-bearing with crutches for six weeks and progressing to full weight-bearing as tolerated.
A Charnley hip prosthesis (Zimmer, Warsaw, Indiana) was used in all patients. A stainless-steel polished flatback femoral stem with a 22.25-millimeter-diameter head, and an ultra-high molecular weight polyethylene acetabular component with an outer diameter of either forty or forty-four millimeters, were inserted with Simplex-P cement (Howmedica, Rutherford, New Jersey).
The latest follow-up evaluation was conducted at a minimum of fifteen years after the index arthroplasty. We attempted to interview all living patients and the families of the patients who had died. Living patients either returned for clinical and radiographic follow-up or, if they were unable to return, were asked to send radiographs (made locally) to us for evaluation. All living patients were evaluated in person or were interviewed by telephone with use of a standard system of terminology for reporting results as described by Johnston et al.22. Relatives of patients who had died were asked about the function of the hip at the time of death. For 299 patients (334 hips), a complete functional assessment was possible through either a personal interview or an interview with a family member; for the remaining twenty-one patients (twenty-three hips), only the status of the hip with regard to whether or not it had been revised was known.
At a minimum of fifteen years after the operation, 130 patients (142 hips) were still alive, 189 patients (214 hips) had died, and one patient (one hip) had been lost to follow-up. Thus, the status of 356 hips (more than 99 per cent of the original 357 hips) was known. Of the 130 patients who were alive, 104 (116 [82 per cent] of the 142 hips) were evaluated clinically and with an anteroposterior radiograph of the pelvis that included the tip of the femoral stem. Seventy-four of these patients (eighty hips) returned for examination, and thirty (thirty-six hips) sent radiographs that had been made elsewhere. The remaining twenty-six patients (twenty-six hips) refused to have follow-up radiographs made and were evaluated on the basis of a telephone interview only. For these latter patients, the most recent radiographs had been made at a minimum of ten years postoperatively (fourteen hips), five years postoperatively (eleven hips), or less than five years postoperatively (one hip).
The prevalence of revision in the study was calculated on the basis of 356 hips (all except the one that had been lost to follow-up). The prevalence of radiographic loosening without infection was determined on the basis of 340 hips (all except those that had been revised because of infection or dislocation and the one that was lost to follow-up). For the prevalence of revision at least fifteen years after the index operation, 142 hips in 130 living patients were included. For the prevalence of radiographic loosening without infection at least fifteen years after the index operation, 116 hips for which radiographs were made at the latest follow-up visit were included while those that had been revised because of dislocation or infection were excluded. The clinical follow-up data at least fifteen years after the index operation were evaluated on the basis of 142 hips in 130 living patients.
Radiographic Evaluation
Observations and measurements were based on anteroposterior radiographs of the pelvis, made early in the postoperative period and at five-year intervals thereafter until the latest follow-up visit. Correction for magnification was obtained by standardization of all measurements against the magnification of the measured size of the femoral head as compared with the known size. Osteolysis was defined as any non-linear radiolucency at the bone-cement interface that was at least five millimeters long and was recorded according to the three acetabular zones described by DeLee and Charnley11 and the seven femoral zones delineated by Gruen et al.15. The position of the femoral stem was determined on the basis of the angle formed between the central axis of the prosthesis and the lateral endosteal cortex. Heterotopic ossification was graded according to the criteria of Brooker et al.5.
Loosening of the Femoral Component
Loosening of the femoral component was classified according to the criteria of Harris et al.17,18. Definite loosening was defined as subsidence of the femoral component, fracture of the cement or the stem, or a radiolucent line at the cement-prosthesis interface as seen on serial radiographs. (The last criterion was modified to include only radiolucent lines of more than one millimeter in width.) These radiolucent lines always occurred in zone I of Gruen et al.15. Probable loosening was characterized by a continuous radiolucent line along the entire bone-cement interface. Possible loosening was indicated by a radiolucent line at the bone-cement interface that encompassed more than 50 but less than 100 per cent of the circumference of the stem on the anteroposterior radiograph. Any radiolucency between the prosthesis and the cement in zone I, regardless of width, was recorded as debonding because this finding signified that, at least in that one area, the prosthesis was no longer in contact with the cement. However, many patients who have less than one millimeter of radiolucency between the cement and the prosthesis in zone I never have any additional evidence of loosening in the long term.
Subsidence of the femoral component, determined with use of the method of Loudon and Charnley26, was defined as a decrease of at least five millimeters (with magnification taken into account), between the initial postoperative radiographs and those made at the latest follow-up evaluation, in the distance from a line drawn perpendicular to the central axis of the femoral stem and intersecting the tip of the stem and a line drawn perpendicular to the central axis and intersecting the point where the trochanteric wire passes through the lesser trochanter.
Loosening of the Acetabular Component
Definite loosening of the acetabular component was defined as migration of the component or any new fracture in the cement mantle; probable loosening, as radiolucency around 100 per cent of the component at the bone-cement interface19; and possible loosening, as radiolucency around 50 to 99 per cent of the component at the bone-cement interface28. Migration of the acetabular component was evaluated with use of the criteria of Massin et al.27. On each radiograph, the vertical distance between the center of the cup and a line joining the two teardrops was measured. The horizontal distance between the center of the cup and a vertical line through the teardrop was also measured. The acetabular component was considered to have migrated if these distances varied more than five millimeters between the immediate postoperative radiographs and those made at the latest follow-up evaluation, after correction for magnification.
Wear
Linear wear was determined by measuring the change in the shortest distance between the center of the femoral head and the periphery of the acetabular component as seen on the immediate postoperative radiographs compared with that seen on those made at the latest follow-up evaluation25. Volumetric wear was calculated by multiplying pr2 by the amount of linear wear. Measurements were made with a digitizing stylus and tablet (Sigma Scan; Jandel Scientific, La Jolla, California) that had an accuracy of 0.025 millimeter. Magnification was standardized against the known circumference of the femoral head.
Statistical Analysis
The Kaplan-Meier23,30 method was used to evaluate survival of the implant with regard to revision or loosening, or both. Survivorship curves with corresponding confidence intervals were generated, with failure defined according to five end points: (1) revision for any reason; (2) revision because of aseptic loosening of the acetabular component; (3) revision because of aseptic loosening of the femoral component; (4) loosening of the acetabular component, defined as definite or probable radiographic loosening or revision because of aseptic loosening; and (5) loosening of the femoral component, defined as definite or probable radiographic loosening or revision because of aseptic loosening12,13,32.
The clinical and radiographic results were analyzed with use of the two-tailed Fisher exact test for categorical variables. The Wilcoxon rank-sum test was used to compare rates of wear according to categorical variables as these rates are not normally distributed. The Spearman correlation coefficient was used to analyze the association between age and wear as well as that between weight and wear.
Clinical Results
At the time of the minimum fifteen-year follow-up evaluation, the average age of the 130 patients who were still alive was seventy-seven years (range, thirty-nine to 103 years). For the 189 patients who had died, the average age at the time of death was eighty years (range, fifty-five to 102 years). Forty-nine patients had died in the first five years after the index arthroplasty; fifty-nine, between five and ten years; and eighty-one, between ten and fifteen years.
Revision of the Original Prosthesis
Of the 356 hips, thirty-five (10 per cent) had had a revision (thirty-three hips) or a Girdlestone resection arthroplasty (two hips) by the time of the latest follow-up evaluation. The revision was performed because of aseptic loosening in nineteen hips (5 per cent), because of loosening with infection in seven (2 per cent), and because of dislocation in seven (2 per cent). The resection arthroplasty was performed because of loosening with infection in both hips. Of the 142 hips in the patients who survived for at least fifteen years after the index operation, fifteen (11 per cent) had a revision because of aseptic loosening; three (2 per cent), because of loosening with infection; and four (3 per cent), because of dislocation. Neither patient who had a resection arthroplasty was followed for at least fifteen years.
Of the nine infections, eight developed early (less than five years postoperatively) and one, ten years after the index procedure. Six of the nine hips were successfully treated with a revision. One hip needed a second revision but was stable at the latest follow-up evaluation. All reimplantations were performed as a one-stage procedure. The remaining two hips had a Girdlestone resection arthroplasty, as mentioned.
Of the 340 hips for which the outcome was known at the latest follow-up evaluation and that had not been revised because of dislocation or infection, fifteen (4 per cent) had been revised because of aseptic loosening of the acetabular component; two (less than 1 per cent), because of aseptic loosening of the femoral component (both had a fracture of the stem); and two (less than 1 per cent), because of loosening of both components. Of the 135 hips in the patients who were still alive at least fifteen years after the initial arthroplasty and who had not had a revision because of infection or dislocation, twelve (9 per cent) had been revised because of aseptic loosening of the acetabular component; one (less than 1 per cent), because of aseptic loosening of the femoral component; and two (1 per cent), because of loosening of both components.
Other Complications
Twenty-three (6 per cent) of the entire series of 356 hips and eleven (8 per cent) of the 142 hips in the living patients (including those who had had a revision because of dislocation) had dislocated at the time of the latest follow-up evaluation. The trochanteric wires had been removed because of bursitis in twelve (3 per cent) of the 356 hips and in nine (6 per cent) of the 142 hips.
Relief of Pain
Preoperatively, nine (3 per cent) of the original 357 hips had caused no pain, forty-five (13 per cent) had caused mild pain, 138 (39 per cent) had caused moderate pain, and 165 (46 per cent) had caused severe pain. All patients who had had pain preoperatively had excellent relief after the total hip replacement, and this was maintained during the course of follow-up for all but one patient. Of the 142 hips in the patients who were still alive at least fifteen years after the index arthroplasty, 102 (72 per cent) caused no pain and thirty-six (25 per cent) caused only slight, occasional pain. Only three patients had moderate pain that caused them to modify their activities, and one patient had severe pain associated with recurrent dislocation of the hip after a revision for a fracture of the femoral stem. Of the 356 hips for which the outcome was known at the latest follow-up evaluation, 297 (83 per cent) caused no pain; fifty-one (14 per cent), mild pain; seven (2 per cent), moderate pain; and one (less than 1 per cent), severe pain.
Activities of Daily Living
Preoperatively, none of the 319 patients had performed strenuous labor and 312 (98 per cent) had performed either moderately strenuous labor (140 patients; 44 per cent) or light labor (172 patients; 54 per cent). Thirteen (4 per cent) of the 319 patients had no difficulty with activities of daily living, and 287 (90 per cent) had mild difficulty (eighty-three patients; 26 per cent) or moderate difficulty (204 patients; 64 per cent) with tying their shoes, ascending stairs, or getting in and out of a chair. The remaining nineteen (6 per cent) could not perform these activities. One hundred and twenty (38 per cent) of the 319 patients used some type of walking aid. All 356 hips were associated with a limp, which was mild (120; 34 per cent), moderate (191; 54 per cent), or severe (forty-five; 13 per cent). Of the 130 patients who were still alive at least fifteen years after the index arthroplasty, nine (7 per cent) performed strenuous labor, thirteen (10 per cent) performed moderately strenuous labor, fifty-seven (44 per cent) performed light labor, fifty (38 per cent) were semi-sedentary, and one was bedridden. In this same group, eighty patients (62 per cent) were able to walk without support, thirty-two (25 per cent) used a cane part time, thirteen (10 per cent) used some type of support full time, and five (4 per cent) were unable to walk. Seventy-three patients (56 per cent) were able to walk for an unlimited duration, twenty-one (16 per cent) were able to walk for thirty to sixty minutes, fifteen (12 per cent) were able to walk for ten to thirty minutes, nine (7 per cent) were able to walk for two to ten minutes, seven (5 per cent) were able to walk for less than two minutes, and five (4 per cent), as just mentioned, were unable to walk.
Range of Motion
For the eighty hips in the patients who were alive at least fifteen years after the index procedure, were physically examined, and had not had a revision, the range of motion had improved dramatically. Flexion contractures were measured with use of the Thomas test, which involves flexion of the contralateral hip to eliminate lumbar lordosis. Preoperatively, the average flexion contracture had been 20 degrees and postoperatively, it was 0 degrees. On the average, flexion had improved 30 degrees, from 100 to 130 degrees; abduction, 20 degrees, from 12 to 32 degrees; adduction, 16 degrees, from 11 to 27 degrees; and internal and external rotation, 20 degrees each, from -1 to 19 degrees and from 17 to 37 degrees, respectively.
Satisfaction
Of the 142 hips in the patients who survived at least fifteen years after the total hip arthroplasty, 141 (99 per cent) were considered by the patient to have better function; 139 (98 per cent), to be less painful; and 136 (96 per cent), to have a satisfactory result. Of the 339 hips for which all clinical information was available, either from the patient or from a relative if the patient had died, 330 (97 per cent) had a satisfactory result; 331 (98 per cent), a decrease in pain; and 322 (95 per cent), an improvement in function. Three patients were dissatisfied because of recurrent dislocation after a revision performed because of dislocation, one was dissatisfied because of a three-centimeter limb-length discrepancy after a revision performed because of aseptic loosening, and two were dissatisfied after a Girdlestone resection arthroplasty performed because of infection. Additionally, one patient was displeased that a revision had been needed because of aseptic loosening even though, after the revision, the patient was doing well. One patient's family could not give a reason for the dissatisfaction.
Radiographic Results
Radiographs were made at least fifteen years after the arthroplasty for 116 (82 per cent) of the 142 hips in the patients who were still alive. The average interval between the index arthroplasty and the most recent radiographs was 14.1 years. Radiographs made after a reasonably long (more than five-year) follow-up interval were unavailable for thirty-five of the 356 hips in the entire series.
Grading of the Cementing Technique
The immediate postoperative radiographs of the hips were used to grade the cementing technique according to the criteria of Schmalzried and Harris33. Of the 142 hips in the patients who were alive at least fifteen years after the index arthroplasty, twenty-seven (19 per cent) were grade A (Figs. 1-A, 1-B, and 1-C); seventy-one (50 per cent), grade B (Figs. 2-A, 2-B, and 2-C); eleven (8 per cent), grade C1; and thirty-three (23 per cent), grade C2. Of the 356 hips in the entire series, fifty-four (15 per cent) were grade A; 194 (54 per cent), grade B; twenty-four (7 per cent), grade C1; and eighty-four (24 per cent), grade C2. Of the ten hips that had aseptic loosening of the femoral component, one was classified as having had grade-A cementing technique; five, grade-B; one, grade-C1; and three, grade-C2. Of the two hips that had loosening of the femoral component as the result of a fracture of the stem, one had had grade-A cementing technique and one had had grade-B cementing technique.
We could detect no correlation, with use of the Fisher exact test, between the grade of the cementing technique and aseptic loosening of the femoral component (p = 0.32) or revision performed because of aseptic loosening of the femoral component (p = 0.19).
Wear and Osteolysis
Wear of the acetabular component was measured in the 116 hips that had been followed radiographically for at least fifteen years. The average amount of linear wear was 0.091 millimeter (range, 0.00 to 0.214 millimeter) a year. The calculated volumetric wear was thirty-eight cubic millimeters (range, 0.0 to 88.0 cubic millimeters) a year. Revision because of aseptic loosening of the acetabular component was associated with linear wear (p = 0.02) according to the Wilcoxon rank-sum test (average rate of wear, 0.17 millimeter a year for the revised components as compared with 0.09 millimeter a year for the stable components). Although the rate of wear was higher in the hips that had evidence of aseptic loosening of the acetabular component than in those in which the acetabular component was stable (0.12 compared with 0.09 millimeter a year), we could not detect a significant difference (p = 0.53). We also could not detect a correlation between linear wear and the level of activity (p = 0.12) or gender (p = 0.82). Increased linear wear of the acetabular component was correlated with decreasing age, with use of the Spearman correlation coefficient (p = 0.01); however, we could detect no correlation with weight (p = 0.94). Of the 135 hips in the patients who survived at least fifteen years and who had not had a revision because of infection or dislocation, fifty-six (41 per cent) had femoral osteolysis in zone VII. In this same group, osteolysis occurred in femoral zones I through VI in five hips (4 per cent), on the acetabular side alone in ten hips (7 per cent), and on both the acetabular and the femoral side in three hips (2 per cent). Osteolysis in femoral zone VII was correlated with wear (p = 0.02), revision because of aseptic loosening of the acetabular component (p < 0.0001), and aseptic loosening of the acetabular component (p < 0.0001). The prevalence of radiolucent lines (of any thickness) at the prosthesis-cement interface in femoral zone I (so-called debonding of the femoral component) was 26 per cent (thirty-seven of 142 hips).
Heterotopic Ossification
Heterotopic ossification was present in 119 (37 per cent) of the 321 hips for which at least two years of radiographic follow-up was available. According to the criteria of Brooker et al.5, eighty-seven hips (27 per cent) had grade-I ossification; thirteen (4 per cent), grade-II; and nineteen (6 per cent), grade-III. Grade-IV ossification was not found in any hip.
Aseptic Loosening
Of the 142 hips in the patients who survived at least fifteen years, 116 were not revised because of deep infection or dislocation and had at least fifteen years of radiographic follow-up. Of these 116 hips, twelve (10 per cent) had loosening of the acetabular component (definite in nine and probable in three), and three (3 per cent) had definite loosening of the femoral component. Eleven hips (9 per cent) had possible loosening of the acetabular component, and no hip had possible loosening of the femoral component. Of the 340 hips that had not had a revision because of infection or dislocation, twenty-four (7 per cent) had loosening of the acetabular component (definite in seventeen and probable in seven), and six (2 per cent) had definite and none had probable loosening of the femoral component. Forty-seven acetabular components (14 per cent) and one femoral component (less than 1 per cent) had possible loosening.
The combined prevalence of definite or probable radiographic loosening of the femoral component according to the modified criteria of Harris et al.17,18 and of aseptic loosening of the femoral component necessitating revision was 3 per cent (ten of 340 hips) over-all and 5 per cent (six of 116 hips) for the hips that had at least fifteen years of radiographic follow-up. The combined prevalence of definite or probable radiographic loosening of the acetabular component17,18 and of aseptic loosening of the acetabular component necessitating revision was 12 per cent (forty-one of 340 hips) over-all and 22 per cent (twenty-six of 116 hips) for the hips that had at least fifteen years of radiographic follow-up (Table I).
Loosening of both the acetabular and the femoral component was associated with the patient's age at the time of the replacement, according to the two-sample t test. The average age of the patients who had aseptic loosening of the femoral component was 62.6 years, compared with 69.1 years for those in whom the hip was stable (p = 0.02). The average age of the patients who had aseptic loosening of the acetabular component was 64.4 years, compared with 69.4 years for those in whom the hip was stable (p = 0.003). Revision performed because of aseptic loosening of the acetabular component was strongly associated with age (average, 56.5 years for the patients who had a revision, compared with 69.4 years for those who did not) (p < 0.0001). With the numbers available, there was no relationship between revision for aseptic loosening of the femoral component and age.
With the numbers available, loosening of the acetabular or the femoral component was not associated with gender, previous operations, or diagnosis.
The average weight of the patients who had aseptic loosening of the femoral component was 188 pounds (eighty-five kilograms), in comparison with 165 pounds (seventy-five kilograms) for those in whom the hip was stable. However, with the numbers available, we were unable to demonstrate a significant difference (p = 0.12).
We could not detect a difference, with regard to the results of fixation of the acetabular component, between the hips in which the plunger had been used and those in which it had not.
Survivorship Analysis
At the latest follow-up evaluation, 321 (90 per cent) of the 356 original prostheses were functioning or had been in place when the patient died. Of the 142 hips in the patients who were living at least fifteen years after the arthroplasty, 120 (85 per cent) were still functioning with the index prosthesis in place (Table II). The Kaplan-Meier survivorship analysis23,30 with 95 per cent confidence intervals demonstrated that, for all hips, the probability of retention of the prosthesis at the time of the latest follow-up evaluation was 86 ± 4 per cent (Fig. 3-A). The probability of survival was 88 ± 6 per cent with revision because of aseptic loosening as the end point Fig. 3-B), 98 ± 2 per cent with revision because of aseptic loosening of the femoral component as the end point (Fig. 3-C), and 89 ± 6 per cent with revision because of aseptic loosening of the acetabular component as the end point (Fig. 3-D). With radiographic evidence of definite or probable loosening or revision because of aseptic loosening as the end point, the probability was 93 ± 4 per cent for the femoral component (Fig. 3-E) and 69 ± 10 per cent for the acetabular component (Fig. 3-F).