The demographic characteristics of the patients, the initial diagnoses, the criteria for inclusion and exclusion from the series, and the details of the operative techniques have been reported previously15. The average age at the time of the operation was reported to be sixty-one years (range, twenty-one to eighty-five years). One patient who was included in the earlier report, but should not have been, was excluded from the current study because the patient had had a conversion operation (meaning a total hip replacement done after a previous intra-articular operation on the hip, such as a cup arthroplasty or a hemiarthroplasty) rather than a primary total hip arthroplasty. That patient had been evaluated at 17.4 years after the total hip arthroplasty, at which time both components were in place and neither was loose.
Of the original 140 patients (161 hips), sixty-seven patients (seventy-seven hips) died within seventeen years after the index operation. The remaining seventy-three patients (eighty-four hips) were followed for an average of eighteen years (range, seventeen to twenty years). The average duration of follow-up for the entire series of 140 patients (161 hips) was 14.7 years (range, 2.3 to twenty years).
The seventy-three surviving patients filled out a questionnaire, which included data for determining the Harris hip score6. These patients were interviewed by the junior one of us (S. W. S.) rather than by the operating surgeon. Twenty patients returned to the office for the interview and a physical examination, and fifty-three were interviewed by telephone in order to supplement the questionnaire. The efficacy of the questionnaire combined with the telephone interview was recently verified13. No patient was lost to follow-up, and the outcome for all patients was determined. Current radiographs were available for seventy-eight (93 percent) of the eighty-four hips.
The femoral cement mantle was graded on all postoperative radiographs according to the criteria initially reported by Barrack et al.1 and modified by Mulroy et al.15. Definite evidence of loosening of the femoral component was defined as subsidence of the stem or the cement mantle, bending or breakage of the stem, fracture of the cement mantle, or debonding of the stem as indicated by a radiolucent line of any width at the cement-metal interface.
The presence of osteolysis in the pelvis or the femur, or both, was carefully assessed. Femoral osteolysis was defined as a focal area of scalloped, endosteal, intracortical, or cancellous bone resorption that was not linear and was wider than two millimeters15. Osteolysis in the pelvis was considered to be present not only when there was a so-called balloon lesion behind the acetabular component or an eccentric area of periprosthetic bone loss but also when there was a radiolucent zone at the cement-bone interface if the zone was linear and parallel to the surface of the bone cement and was more than three millimeters thick. Areas of femoral osteolysis were measured and recorded according to location with use of the zones described by Gruen et al.5, and areas of pelvic osteolysis, with use of the zones described by DeLee and Charnley4.
Kaplan-Meier10 survivorship curves were generated, with use of revision of the femoral component because of aseptic loosening and revision of the acetabular component because of aseptic loosening as the end points.
Over the twenty-year interval, from January 20, 1976, to March 25, 1996, eight (5 percent) of the 161 femoral components and twenty-eight (17 percent) of the 161 acetabular components were revised because of aseptic loosening. A reoperation for any reason was performed in thirty-two hips (20 percent; twenty-nine patients), and a revision because of aseptic loosening of either component was performed in twenty-nine hips (18 percent; twenty-six patients) (Table I). Nine (43 percent) of the twenty-one metal-backed acetabular components were revised because of aseptic loosening compared with nineteen (14 percent) of the 140 all-polyethylene acetabular components; this difference was significant (p = 0.0028, Yates corrected chi-square test).
In addition to the nine metal-backed acetabular components that were revised because of aseptic loosening, seven (33 percent) were loose according to radiographic criteria; thus, a total of sixteen (76 percent) of the twenty-one metal-backed cups were loose. In addition to the nineteen all-polyethylene acetabular components that were revised because of aseptic loosening, nineteen (14 percent) were loose according to radiographic criteria; thus, a total of thirty-eight (27 percent) of the 140 all-polyethylene cups were loose (p < 0.0001, Yates corrected chi-square test).
Of the twenty-nine revision operations that were performed because of aseptic loosening, twenty-one were performed (at an average of fourteen years [range, six to 18.6 years] postoperatively) because of loosening of the acetabular component only, one was performed because of loosening of the femoral component only, and seven were performed (at an average of 15.6 years [range, 10.8 to 19.6 years] postoperatively) because of loosening of both components.
Over the twenty-year span of this study, 145 (90 percent) of the 161 femoral components and 129 (80 percent) of the 161 acetabular components remained in place and were functioning either at the time of death or at the time of the latest follow-up evaluation (Figs. 1-A, 1-B, and 1-C). Of the thirty-two revisions of the socket, twenty-eight (88 percent) were performed because of aseptic loosening. The remaining four (13 percent) were performed because of aseptic loosening of the femoral component only, late infection, severe heterotopic ossification, or an acetabular fracture sustained in a motor-vehicle accident.
Of the eight femoral components that were revised because of aseptic loosening, seven were revised in conjunction with revision of a loose acetabular component and one was removed (at another institution) in conjunction with removal of a solidly fixed acetabular component as part of a resection arthroplasty that was performed, surprisingly, because of aseptic loosening of the femoral component only. No revisions of the femoral component were pending at the time of the most recent follow-up. Eight additional femoral components were removed for reasons other than aseptic loosening. One solidly fixed femoral stem was revised to a calcar-replacement stem because of limited motion due to heterotopic ossification, one stem was revised because of infection, and six solidly fixed stems were removed to gain exposure for revision of a loose acetabular component.
Eight (7 percent) of the 117 CAD (computer-assisted-design) stems (Howmedica, Rutherford, New Jersey) and none of the forty-four Harris Design-II stems (Howmedica) were revised because of aseptic loosening; with the numbers available for study, this difference was not found to be significant (p = 0.17, Yates corrected chi-square test). One CAD stem and one Harris Design-II stem were loose according to radiographic criteria. The acetabular component was revised because of aseptic loosening in twenty-one (18 percent) of the hips with a CAD stem and in seven (16 percent) of those with a Harris Design-II stem. There was no substantial difference between the patients who had the CAD stem and those who had the Harris Design-II stem with regard to gender, age, or weight.
The probability of survival of the femoral component was 95 percent at fifteen years, 92 percent at seventeen years, and 88 percent at twenty years (Fig. 2). The probability of survival of the acetabular component was 90 percent at fifteen years, 79 percent at seventeen years, and 69 percent at twenty years (Fig. 3).
Of the 145 femoral components that were in place at the time of death or at the time of the latest follow-up, two (1 percent) were loose according to radiographic criteria; both had debonding of less than one millimeter at the cement-metal interface. The overall prevalence of loosening of the femoral component for the entire series of 161 hips was 6 percent (ten hips).
On the basis of the evaluation of all of the postoperative radiographs, five (3 percent) of the 161 hips were rated as having a grade-A cement mantle; forty-nine (30 percent), a grade-B mantle; twenty-eight (17 percent), a grade-C1 mantle; and seventy-nine (49 percent), a grade-C2 mantle. No hip had a grade-D cement mantle. All eight femoral stems that were revised because of aseptic loosening had a grade-C2 cement mantle. One of the two stems that were considered loose according to radiographic criteria had a grade of C2, and the other had a grade of C1. No femoral stem in a hip with a grade-A or B cement mantle became loose. Eight of the seventy-nine hips that had a grade-C2 cement mantle were revised compared with none of the eighty-two that had another grade; this difference was significant (p = 0.0095, Yates corrected chi-square test).
The all-polyethylene acetabular components that had been revised because of aseptic loosening had an average rate of linear wear of 0.21 millimeter per year compared with an average rate of 0.12 millimeter per year for those that had not been revised (p = 0.0001).
The average age of the twenty-six patients (twenty-nine hips) in whom the acetabular or the femoral component was revised because of aseptic loosening was fifty-five years at the time of the index operation, whereas that of the remaining 114 patients (132 hips) was sixty-two years (p = 0.005).
Forty-nine (35 percent) of the 140 patients had a complication. There were nineteen cases of proximal deep venous thrombosis (12 percent), sixteen cases of distal deep venous thrombosis (10 percent), and four cases of nonfatal pulmonary embolus (2 percent). Five hips (3 percent) had a dislocation, which necessitated a reoperation in two (1 percent). Trochanteric wires that were causing pain were removed from three hips (2 percent). One hip (1 percent) had a late infection. One hip had a wound hematoma that necessitated evacuation, and another had an enlarging hematoma with active bleeding that necessitated arterial embolization. Seven other hips had a hematoma that did not necessitate intervention. There were two partial palsies of the sciatic nerve, both of which resolved. Two (2 percent) of 106 trochanteric osteotomy sites did not unite. One hip that had grade-IV heterotopic ossification2 needed a resection. One postoperative gastrointestinal bleed and one perioperative nonfatal myocardial infarction resolved satisfactorily. There were no perioperative deaths.
The sixty-seven patients (seventy-seven hips) who died less than seventeen years after the index operation had an average age of sixty-eight years (range, forty-six to eighty-five years) at the time of the operation and an average age of eighty years (range, forty-six to ninety-eight years) at the time of death. The average duration of survival after the index operation was twelve years (range, 2.3 to 16.3 years). Sixty-nine (90 percent) of the seventy-seven implants were in place at the time of death; the remaining eight had been revised, as already described.
The seventy-three patients (eighty-four hips) who were alive at least seventeen years after the index operation had an average age of fifty-three years (range, twenty-one to seventy-eight years) at the time of the operation. Seventy-two femoral components (86 percent) and sixty-one acetabular components (73 percent) were intact and functional at the time of the latest follow-up. Six femoral components (7 percent) had been revised because of aseptic loosening, at an average of 16.4 years (range, 14.2 to 19.6 years). Eight of the nineteen metal-backed acetabular components had been revised because of aseptic loosening compared with fifteen (23 per cent) of the sixty-five all-polyethylene components; with the numbers available, this difference was not found to be significant (p = 0.17, Yates corrected chi-square test). The overall rate of revision of the acetabular component because of aseptic loosening was 27 percent (twenty-three of eighty-four hips). The overall rate of loosening of the femoral component (with or without revision) was 10 percent (eight hips).
Of the fifty-six acetabular components that were in situ and for which current radiographs were available at the time of the latest follow-up, twenty-three components (41 percent), including six metal-backed cups and seventeen all-polyethylene cups, were loose. The overall prevalence of aseptic loosening of the acetabular component16 was forty-six (59 percent) of seventy-eight hips.
Of the nineteen metal-backed acetabular components that had been inserted in patients who were alive at least seventeen years postoperatively, fourteen were revised because of aseptic loosening or were loose according to radiographic criteria. Of the sixty-five all-polyethylene acetabular components that had been inserted in patients who were alive at least seventeen years postoperatively, thirty-two (49 percent) were revised because of aseptic loosening or were loose according to radiographic criteria.
With the numbers available, no significant difference was detected between the CAD and the Harris Design-II stems with regard to the rate of survival, loosening, or revision, nor was a significant difference found in the rates of revision of the acetabular component associated with the two types of stems.
The average preoperative Harris hip score for the sixty hips in the fifty-two surviving patients who had not had a revision operation was 51 points (range, 27 to 72 points) compared with 85 points (range, 54 to 100 points) at the time of the latest follow-up. Pain relief was excellent overall, with thirty-six hips (60 percent) causing no pain and sixteen (27 percent) causing only slight, occasional pain with no limitation in activity. The remaining eight hips (13 percent) were mildly painful, with no effect on activity, or were painful only after unusual activity. No patient had moderate or severe pain. The average pain score for this group was 41 points (range, 30 to 44 points) of a possible 44 points. All fifty-two patients were satisfied with the result at the time of the latest follow-up.
Twenty-eight (47 percent) of the sixty hips had an overall rating of excellent according to the Harris hip-scoring system; thirteen (22 percent), a rating of good; eleven (18 percent), a rating of fair; and eight (13 percent), a rating of poor. Disease unrelated to the hip contributed to all but one of the poor results. Three patients were bedridden secondary to a cerebrovascular accident, and one each had a severe neurological deficit after an operation on the spine, severe low-back pain secondary to metastatic prostate cancer, severe emphysema and spinal stenosis, and severe Alzheimer disease. The poor result in the eighth patient was secondary to aseptic loosening of the acetabular component, and a revision was recommended.
Twenty-five (48 percent) of the fifty-two patients were able to walk an unlimited distance, seven (13 percent) were able to walk six blocks, fifteen (29 percent) were able to walk two to three blocks, two (4 percent) were able to walk indoors only, and three (6 percent) were unable to walk. Twenty-four patients (46 percent) had no limp, fifteen (29 percent) had a slight limp, nine (17 percent) had a moderate limp, and four (8 percent) had a severe limp or were unable to walk. Thirty-three patients (63 percent) were able to walk without support, eight (15 percent) used a cane for long walks only, seven (13 percent) used a cane full-time, and four (8 percent) used two crutches or were unable to walk. The average physical-component score on the Short Form-36 was 36.9 points, and the average mental-component score was 51.4 points.
In the entire series of 161 hips, osteolysis was present in four of the eight hips that had had revision of the femoral component because of aseptic loosening and in an additional twenty-two hips (14 percent) in areas other than the proximal one centimeter of zones I, VII, VIII, or XIV5. Osteolysis was present in all twenty-eight hips in which the acetabular component had been revised because of aseptic loosening. Osteolysis about the acetabular component developed in sixty-three (39 percent) of the 161 hips, and osteolysis developed around twenty-nine (52 percent) of the fifty-six acetabular components that were in situ and for which current radiographs were available at an average of eighteen years postoperatively.
We assessed the clinical outcomes as well as the rates of revision, aseptic loosening, and periprosthetic osteolysis in a consecutive series of patients who had had a standard primary total hip arthroplasty with insertion of a bead-blasted, finished, monoblock, chromium-cobalt femoral stem with use of so-called second-generation cementing techniques. No patient was lost to follow-up. Current radiographs were available for seventy-eight (93 percent) of the eighty-four hips in patients who survived at least seventeen years after the index operation.
In order to focus on a homogeneous group of patients, those in whom the implant had been revised or converted from a preexisting replacement and those who had needed bulk bone graft or custom components were excluded from the study. The current series had one less patient (one hip) than the group that was reported on at fifteen years15 because, in the earlier study, a hip that had had the total hip replacement after a preexisting cup arthroplasty had inadvertently been included. Otherwise, the current report is based on the cohort of patients in the earlier report.
In the current study, over a twenty-year period the prevalence of revision because of aseptic loosening of the femoral component was eight (5 percent) of 161 hips and that of the acetabular component was twenty-eight (17 percent) of 161. At an average of fifteen years, the prevalence of revision of the femoral component because of aseptic loosening had been four (2 percent) of 162 hips and that of the acetabular component had been seventeen (10 percent) of 16215. At an average of eighteen years, the results continued to be excellent. In the patients who had survived seventeen to twenty years after the index operation, the prevalence of revision of the femoral component because of aseptic loosening was 7 percent (six of eighty-four hips). No additional Harris Design-II stems had become loose since the time of the fifteen-year study. The rate of revision of the acetabular component because of aseptic loosening was twenty-three (27 percent) of eighty-four, and the rate for the all-polyethylene acetabular components was fifteen (23 percent) of sixty-five. A slight increase in the average rate of polyethylene wear was seen in association with the acetabular components that had been revised because of aseptic loosening (0.21 millimeter per year compared with 0.12 millimeter per year for those that had not been revised).
The rates of revision because of aseptic loosening of either component were related to the age of the patient. A revision was performed in nineteen (31 percent) of the sixty-two hips in the patients who were fifty-nine years old or younger but in only ten (10 percent) of the ninety-nine hips in those who were sixty years old or older.
The clinical results continued to be good or excellent, at an average of eighteen years, for forty-one (68 percent) of the sixty hips in the surviving patients who had not had a revision; fifty-two (87 percent) of these sixty hips caused either no pain or slight, occasional pain. The poor results in all but one of the remaining eight hips (13 percent) could be attributed directly to severe, coexisting, unrelated medical conditions, as discussed earlier.
We believe that our refinement of the cement-grading process, which involves the assessment of all postoperative radiographs and not just those that were made immediately postoperatively, increases the accuracy of the process. Continuing experience with the grading system for evaluation of the femoral cement column has made it clear that this increase in accuracy is due to two factors. First, subsequent radiographs often are made at slightly different angles or rotations than are the initial postoperative radiographs and thus show different projections of the cement mantle. If new findings (such as voids or defects in the mantle that were not visible on the earlier radiographs) are revealed, the grade of the cement is changed to reflect this additional information. Second, the development of radiolucent lines between the cement and the cortex3,12 often reveals important features of the cement mantle that were not apparent on the original radiographs. For example, what may have appeared, on the original radiograph, to have been a so-called whiteout and thus would have been interpreted as grade-A cement can sometimes be shown, on later radiographs, to be an area in which the mantle is incomplete. In such instances, the whiteout represented an area in which the stem was directly against the cortex rather than one in which the cement mantle was intact between the stem and the cortex. As adaptive remodeling occurs over time, the radiolucent line delineates the interface between the cement and the cortex and may reveal a defect of the cement mantle. In such instances, the grade is adjusted to reflect the more accurate data. The longer duration of follow-up and the improvement in the process of grading the cement increase the usefulness of the data on these total hip replacements.
There were more hips with grade-C1 or C2 cement and fewer with grade-A or B cement in our study than there were in the study of Mulroy et al.15, who evaluated the same cohort at an average of fifteen years. This was because we graded hips on the basis of all postoperative radiographs, whereas Mulroy et al. evaluated only the early postoperative radiographs. The assessment of all postoperative radiographs regardless of when they had been made often revealed areas of thin cement that may not have been visible on the radiographs made immediately postoperatively. Thus, the number of C2 cement grades increased in the current study. A thin (less than one-millimeter-thick) mantle of cement at one location or more continued to be a predictor of radiographic loosening of the femoral component.
Similarly, different radiographic projections also allowed detection of some voids that had been obscured by the stem on the immediate postoperative radiographs. Therefore, the number of C1 grades increased and the number of A and B grades decreased.
We are aware of two similar, detailed reports of the results of total hip arthroplasty with use of cement with a duration of follow-up that was equal to or greater than the minimum duration in the current series. Schulte et al. reported on 330 hips in which a polished monoblock Charnley implant had been inserted18. Eighty-three patients (ninety-eight hips) were followed clinically, and sixty-three of them (seventy-six hips) were followed radiographically, for at least twenty years. Eight hips were lost to follow-up. The rate of revision because of aseptic loosening, for the 322 hips for which the outcome was known, was 6 percent for the acetabular components and 2 percent for the femoral components. For the ninety-four hips in the patients who survived at least twenty years and who had been followed radiographically for at least five years, the prevalence of debonding of the femoral component was 38 percent on the basis of evaluation of the anteroposterior radiograph only18. In the current series, we designated stems that had any debonding as loose, whereas Schulte et al. used the criterion of more than one millimeter of radiolucency. In the present series, there was debonding of two (1 percent) of the 145 femoral components that were in place and functioning at the time of death or at the time of the latest follow-up. Kavanagh et al. reported on 333 hips in which a smooth monoblock Charnley implant had been inserted11. One hundred and twelve patients (112 hips) were followed for at least twenty years, and thirty-one hips were lost to follow-up. The overall rate of revision of the femoral component because of aseptic loosening or breakage of the stem was twenty-three (7 percent) of 333, and the rate of subsidence of the femoral component in patients who had had long-term follow-up and for whom current radiographs were available was twenty-two (32 percent) of sixty-nine.
Hozack et al., in a series of 1157 hips, reported the probability of revision of a Charnley femoral stem at twenty years to be 13 percent9. The 3 percent rate of revision of the femoral component in the ninety-eight hips in the patients who were alive at least twenty years postoperatively in the series reported by Schulte et al.18 differed from the rate reported by Hozack et al. by a factor of four9. The difference between these rates may reflect other differences, such as whether all of the operations were performed by one surgeon (as in the series of Schulte et al.) as opposed to multiple surgeons (as in the series of Kavanagh et al. and Hozack et al.). Other, unspecified differences, including the operative technique, the criteria for selection of the patients, the age and activity level of the patients, and the indications for the revision, indicate the limitations of comparing different series that are not prospective and randomized. It is interesting that the rate of revision of the bead-blasted femoral components in the current series (six [7 percent] of eighty-four) is the same as that reported by Kavanagh et al.11 and is between those reported for smooth Charnley components by Schulte et al.18 (3 percent) and Hozack et al.9 (13 percent).
The bead-blasted femoral stems in the current series had a rate of revision because of aseptic loosening similar to the rates for the smooth stems in the series just mentioned. However, the rate of radiographic loosening of the stems in the present series (1 percent) was lower than that of the polished stems in the other three series9,11,18. We believe that the low rates of radiographic loosening and revision because of aseptic loosening in the current series were due to the improved cementing techniques and the improved design of the femoral component, and we recommend use of a bead-blasted monoblock femoral component with second-generation cementing techniques for standard primary total hip arthroplasties.
For the 140 all-polyethylene acetabular components, the rate of revision because of aseptic loosening increased from 7 percent (ten hips) at the fifteen-year follow-up evaluation15 to 14 percent (nineteen hips) in the current study.
Twenty-eight (88 percent) of the thirty-two revisions in the current study were necessitated by aseptic loosening of the acetabular component. This was the sole cause of approximately two-thirds (twenty-one) of the revisions, and it continues to be the primary cause of failure in these patients.
Because of the high rate of failure of the acetabular components that had been inserted with cement and the low rate of failure of the femoral components that also had been inserted with cement, in this series and others18, as well as on the basis of our experience with so-called hybrid total hip replacement (insertion of the acetabular component without cement and the femoral component with cement)17, we recommend that most primary total hip arthroplasties be performed with a porous ingrowth acetabular component and insertion of a bead-blasted monoblock femoral stem with cement.