Living patients were followed clinically with the use of a standard-terminology questionnaire6 as well as Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores at the time of final follow-up7. Radiographs were evaluated for loosening and osteolysis with use of the same criteria employed in previous reports5,7-11.
Kaplan-Meier survivorship curve analysis12 was performed with the following end points: (1) revision for any reason, (2) revision of any component because of aseptic loosening, (3) revision of the acetabular component because of aseptic loosening, (4) revision of the femoral component because of aseptic loosening, (5) definite or probable radiographic evidence of loosening of the acetabular component, and (6) definite or probable radiographic evidence of loosening of the femoral component. All living patients had a radiograph made at a minimum of thirty-five years postoperatively, except for one patient who had radiographic follow-up for only twenty-one years.
When a revision had not been done, the length of time from implantation to death was recorded for the patients who died, and thirty-five years was considered to be the duration of prosthetic survival for the seven patients who were still living with their original implant in place at the time of the study. When a patient had required a revision before he or she had died, the time from the original surgery to the first revision was recorded as was the length of time that the patient had lived following the revision. When a patient who was still living at thirty-five years had required at least one revision (six patients with seven involved hips), the interval from the first revision to the thirty-five-year follow-up visit was calculated. Our hope was that these calculations would provide the reader with the longevity needs of the implants used in this cohort.
Source of Funding
No outside funding was used for the completion of the present study.
Of the 262 patients (330 hips) in the original cohort, twenty-seven (thirty-four hips) were living, 234 (295 hips) had died, and one (one hip) had been lost to follow-up at the time of our previous thirty-year study of this series5. In the current study, at a minimum of thirty-five years following the procedure, twelve patients (fifteen hips) were living, 249 (314 hips) had died, and one (one hip) had been lost to follow-up. Of the living patients, five (six hips) were men and seven (nine hips) were women. The average age of the living patients had been fifty-one years (range, thirty-six to sixty-four years) at the time of surgery and was eighty-eight years (range, seventy-two to 100 years) at the time of follow-up. A radiograph made at a minimum of thirty-five years postoperatively was available for eleven patients with fourteen involved hips. The time of the last radiographic follow-up of the remaining patient was twenty-one years.
Revision
Only one additional revision had been performed since the previous study5. It was done in a hip that had been previously revised because of aseptic acetabular loosening at thirteen years after the index procedure. The additional revision was also due to acetabular loosening and was done almost thirty-one years following the index procedure. Thus, in the entire cohort of 329 hips, there had been no change, since the time of the previous (thirty-year) follow-up, in the overall rate of revision for any reason (12%; thirty-nine hips), revision because of aseptic acetabular loosening (7%; twenty-three hips), or revision because of aseptic femoral loosening (3%; ten hips) (Table I and Appendix).
Of the twelve living patients (fifteen hips), six (seven hips) required at least one revision. These six patients included five men and one woman with an average age of forty-eight years (range, thirty-six to sixty-one years) at the time of surgery, sixty-one years (range, forty-four to seventy-nine years) at the time of the first revision, and eighty-four years (range, seventy-two to ninety-seven years) at the time of final follow-up. The original diagnosis for the seven revised hips was osteoarthritis in three, femoral neck fracture in two, slipped capital femoral epiphysis in one, and congenital dysplasia of the hip in one.
The seven patients (eight hips) who retained the initial arthroplasty construct throughout the thirty-five-year follow-up interval had an average age of fifty-five years (range, forty-three to sixty-four years) at the time of surgery and ninety-one years (range, seventy-eight to 100 years) at the time of follow-up. Thus, the average age at the time of the original surgery in the revision group was seven years younger than that in the group without a revision. All seven of the patients who had retained the initial arthroplasty construct were women. The diagnosis for the eight unrevised hips was osteoarthritis in two and congenital dysplasia of the hip in six. A thirty-five-year radiograph was obtained for seven of the eight hips, and the duration of the radiographic follow-up was only twenty-one years for the remaining hip. The apparent discrepancy in the numbers of living patients (twelve in total, with seven unrevised cases and six revised cases) results from the fact that, in one living patient who had both hips included in the study, one hip had never been revised and the other hip had been revised.
These trends for the living patients seemed to match those for the entire cohort. In the entire cohort, the average age at the time of surgery was fifty-seven years old for those with a revision compared with sixty-seven years old for those without a revision. The most common diagnosis in both the hips with and those without a revision was osteoarthritis (59% and 74%, respectively), which reflects the fact that 74% of the entire cohort had this diagnosis. In addition, approximately two-thirds of the hips in the revision group were in men.
The results of the revisions and the final outcomes of the hips are summarized in Table I and the Appendix. Additional information on the revisions can be obtained from the previous reports on this cohort at follow-up intervals of twenty, twenty-five, and thirty years3-5.
Radiographic Results
Two additional hips demonstrated radiographic evidence of loosening (of one acetabular and one femoral component) since the thirty-year study5. Thus, in the entire cohort, the overall rates of aseptic loosening of the acetabular and femoral components, including cases revised because of aseptic loosening, were 16% (fifty-three hips) and 8% (twenty-five hips), respectively.
Of the seven hips that had never been revised and for which thirty-five-year radiographs were available, three demonstrated definite or probable radiographic evidence of acetabular loosening. In addition, six hips in living patients had been revised because of aseptic loosening of the acetabular component. Of the seven hips with the original implant in place, one demonstrated definite or probable radiographic evidence of femoral loosening. In addition, one hip had had a revision because of aseptic femoral loosening (see Appendix). An area of acetabular osteolysis of >1 cm2 was noted in two of the unrevised hips (one of which had acetabular loosening), and an area of femoral osteolysis of >1 cm2—all in zone VII9—was noted in three of the unrevised hips (none of which had femoral loosening).
Clinical Results
The final follow-up WOMAC scores averaged 27, 24, and 25 points, respectively, for the living patients without a revision, the living patients with a revision, and all of the living patients. The WOMAC score was 29 points for the three living patients who had a loose implant but not a revision.
Survivorship Analysis
Survivorship (and 95% confidence interval) at thirty-five years was 78% ± 8% when the end point was revision for any reason, 81% ± 8% when it was revision because of aseptic loosening, 85% ± 7% when it was revision because of aseptic acetabular loosening, 93% ± 5% when it was revision because of aseptic femoral loosening, 27% ± 19% when it was radiographic evidence of loosening of the acetabular component, and 53% ± 22% when it was radiographic evidence of loosening of the femoral component (Fig. 1 and Appendix).
In this cohort of patients who were operated on for the treatment of disabling arthritis between 1970 and 1972 and at an average age of sixty-five years, this implant was required to function for an average of seventeen years, as calculated on the basis of when they died or the duration of follow-up of the still living patients. In the 290 unrevised hips (88% of the total), the prosthesis functioned until the patients' death or to the end of the follow-up period (average, sixteen years; range, less than one to thirty-five years). In the thirty-nine revised hips (12% of the entire group), the original implant had functioned for an average of thirteen years before the first revision. The patients in the revised group lived an average of twelve additional years after the first revision (until death, or until the time of final follow-up at thirty-five years for the six living patients who required a revision).
The Charnley total hip arthroplasty was introduced in the United States in the late 1960s. We have had the opportunity to closely follow the initial 330 consecutive Charnley total hip arthroplasties performed by a single surgeon (R.C.J.) between 1970 and 19721-5. With the exception of a group of hips followed for thirty-eight years at Wrightington Hospital13 (where the original procedure was developed), to our knowledge, this is the only study with a minimum duration of follow-up of thirty-five years. As all but twelve patients died, we consider this to be an end-result study. Although it is unclear how many hips were lost to follow-up in the Wrightington study13, the authors reported a 70% hip survival rate at a minimum of thirty-one years8. This compares with the 78% hip survival rate that we reported at thirty-five years. The strengths of the present study include the minimal number of patients lost to follow-up, knowledge of the date of each patient's death, and the large number of patients with serial long-term radiographs.
The end-result nature of this study allowed us to calculate the duration of the patients' remaining life span during which they functioned with their original implant. In 290 hips (88%), the original implant remained in place for the remainder of the patient's life, requiring an average of sixteen years of function. In the thirty-nine hips (12%) that required a revision, the original implant had been in place for an average thirteen years. This group lived, on the average, an additional twelve years from the time of the first revision. It is of interest that the average life expectancy of a sixty-five-year-old person in 1970 was an additional fifteen years. The average age of the patients at the time of the original surgery in this cohort was sixty-five years (see Appendix)14, and the average time that the prosthesis functioned in these patients was approximately fifteen years.
There is no question that the longer that the patient lived, the greater was the chance that a revision would be required, as demonstrated by our reported findings that 15% of the hips in the living patients required a revision by twenty years3; 23%, by twenty-five years4; 32%, by thirty years5; and 47%, by thirty-five years. Although we recognize that some of the unrevised hips had radiographic evidence of loosening with associated pain and that this could have decreased the patient's activity so that he or she did not become disabled enough to want additional surgery (the WOMAC scores for these patients were only slightly higher than those for the patients with hip implants that appeared stable radiographically), we think that these calculations may provide some idea as to the longevity required of a total hip replacement performed in patients during that era. As evidenced by the increase in life expectancy in the United States (see Appendix), hip replacements today will be required to function longer than they did in the early 1970s. This expectation of longer function is compounded by the fact that, today, hip replacements are being performed more often in younger patients.
This long-term report, in addition to demonstrating a 78% rate of survival of the Charnley total hip prosthesis at thirty-five years, should shed light on the reporting of long-term data following any type of total hip arthroplasty. Issues related to the long-term activity of patients with extended life spans need further study to discern the relationships between decreasing activity with age and implant durability. This may be especially important when determining differences in long-term durability among individual designs. However, we hope that this study will provide a benchmark for comparisons with newer designs with the caveats of increasing life expectancy and the fact that modern-design implants are being used in younger patients.