Through a search of operative records, we identified 126 patients (139 hips) who had revision of an alumina-alumina hip prosthesis between January 1977 and December 2005 for mechanical failure. Of these 139 hips, 110 (99 patients) had revision only of the acetabulum with a retained femoral component. The indication for revision was isolated aseptic loosening of the acetabular component in 91 hips, unexplained pain in seven, fracture of a ceramic implant in eight (one head, seven sockets), and recurrent dislocation in four (Table ).
Indications for revision surgery
We excluded six patients who had revision for sepsis. The time elapsed between the most recent arthroplasty and revision averaged 76 months (range, 2–222 months). There were 68 women (69%) and 31 men (31%), with an average age at the time of revision of 64 years (range, 27–92 years). Mean (±SD) patient weight was 67 ± 11 kg (range, 40–104 kg), and height was 163 ± 13 cm (range, 148–180 cm). The predominant underlying condition leading to THA was osteoarthritis in 54% of the cases (Table ). The revision surgery was the first revision for 100 hips and the second for 10 hips. Of the original 99 patients, 26 patients (29 hips) died, two patients (two hips) were living abroad and could not be contacted, and seven (eight hips) were lost to followup before 5 years, leaving 64 patients (71 hips—64%) for review. In the group of 26 patients who died, 18 patients were older than 70 years at the time of revision surgery (age, 72 ± 10 years), and 19 were born before 1920 and would have been older than 90 years at the time of this review. Minimum followup was 60 months (mean, 112 months; range, 60–319 months). No patient was recalled specifically for this study; all data were obtained from records. Our review had prior approval of our institutional review board.
Bone stock deficiencies were classified according to the American Academy of Orthopaedic Surgeons (AAOS) system [10
]. The classification was based on the analysis of preoperative plain radiographs and on the description of bone defects in operative notes. Acetabular bone stock losses were classified as follows: 16 hips had no acetabular deficiency, two had a segmental Type I defect, 62 had a cavitary Type II defect, 30 had a combined Type III defect, and none had a Type IV pelvic discontinuity. On the femoral side, bone defects were limited to the calcar in 90% of the hips. According to the AAOS classification system [9
], the defects were classified as cavitary Type II defects in 108 hips and combined Type III defects in two hips. Bone grafting was only necessary on the acetabular side. Contained acetabular bone defects were packed with morselized allograft, whereas relevant segmental bone defects were reconstructed with bulk allografts supported with either the Kerboull acetabular reinforcement device (Stryker Howmedica, Herouville Saint Clair, France) (39 hips, 35.4%) or a Müller reinforcement ring (Zimmer, Inc, Warsaw, IN) (three hips, 2.7%).
Different designs and modes of fixation had been explored on the socket side during this 25-year period. The acetabular component was inserted with cement in 54 hips (49%) and without cement in 56 hips (51%). Fifty-four were cemented plain alumina sockets, 34 were titanium alloy smooth screw-in rings with an alumina insert, 16 were fiber-mesh metal backs with an alumina insert, and six were press-fit bulk alumina sockets. All revised prostheses were manufactured by Ceraver® (Roissy, France). On the femoral side, a collared smooth stem made of anodized titanium alloy was cemented in 103 cases with the so-called second-generation technique, and a fully hydroxyapatite-coated tapered cementless stem was inserted in the remaining seven hips. Our experience with the cementless stem began in 1997.
The revision arthroplasty was performed through a posterior approach in 73 hips (66.3%), a transtrochanteric approach in 26 hips (23.7%), and a Hardinge transgluteal approach in 11 hips (10%). Only the acetabular component was changed in all 110 hips. The alumina head was left in place in 33 hips, the same alumina head was removed for better exposure and reimplanted in seven, a new alumina head was implanted on the same stem in 45, a metallic head in 16, and a zirconia head in nine. Overall, an alumina-alumina combination was implanted in 31 hips (20 press-fit metal backed, five threaded titanium, five cemented sockets, one press-fit bulk alumina socket), an alumina-polyethylene in 55, a metal-polyethylene in 15, and a zirconia-polyethylene in nine. Polyethylene cups were cemented in 91% of the cases (Table ).
Fixation status of the socket after the revision surgery
The postoperative regimen included early mobilization and strengthening exercises begun on the first postoperative day. All patients received inpatient physical therapy for a mean of 6 weeks under the supervision of qualified therapists and physicians in a rehabilitation center. Patients were restricted to 50% body weightbearing with two crutches until the sixth postoperative week, at which time they were encouraged to weightbear fully as tolerated. When acetabular bone grafting was performed, the protocol was slightly modified, allowing walking without weightbearing during the first 6 weeks and 50% body weightbearing from 6 to 12 weeks.
All patients underwent postoperative followup evaluation with a complete physical examination in the outpatient clinic at 6 weeks, 3 months, 6 months, 1 year, and annually thereafter. The Merle d’Aubigné-Postel rating system was used to quantify hip pain, mobility, and stability [11
Two of us (JD, DH) independently reviewed standardized AP pelvic radiographs taken preoperatively and at the latest followup. On the socket side, the parameters assessed were the inclination angle of the cup, vertical and horizontal migration of the hip center, the presence of radiolucent lines in six zones of the socket, and when a polyethylene cup was implanted, measurement of wear with the method of Livermore et al. [25
]. The interobserver variability for the measurement of cup inclination angle is estimated to be within 2°, and the accuracy of the Livermore method has a median error of less than 0.6 mm [12
]. When a Kerboull acetabular device was implanted, parameters analyzed included the positioning of the hook under the inferior margin of the acetabulum and the inclination angle of the acetabular device. Acetabular loosening was defined as a change in the opening angle of the socket of more than 4°, a cup migration of more than 4 mm, or a failure of the acetabular reinforcement device. On the femoral side, several parameters were investigated: the distance between the collar of the prosthesis and the lesser trochanter, the presence of radiolucent lines in seven zones of the femur according to Gruen et al. [16
], and the presence of cement fracture lines.
Qualitative data were expressed as counts and percentages within groups, and quantitative data by mean ± SD. Qualitative data (gender, preoperative diagnosis) were compared using the chi square test or Fisher’s exact test. For quantitative data (PMA score, inclination angle of the acetabulum), we determined differences between pre- and postoperative values using Student’s t test and the Mann-Whitney test depending on the distribution of the data. We attempted to determine whether there was a relationship between iterative aseptic loosening, if present, and gender, age of more than 50 years, body mass index of 30 or greater, high level of activity, type of prosthesis implanted at revision surgery, and amount of preoperative osteolysis. To test differences among at least three groups, we performed ANOVA or the Kruskal-Wallis test when the data did not meet the assumptions of normality. We performed actuarial survivorship analysis [15
] using mechanical failure and revision for any mechanical failure of either component at the time of followup as the end points. The survival curve was derived from the cumulative survival rate over time as calculated from the actuarial life table. The standard error, given as a percentage, and the 95% confidence intervals (CIs) were calculated from the data in the life table. Comparative survivorship analysis was performed using the log-rank test. We used StatView®
5.0 software (SAS Institute, Inc, Cary, NC) for all analyses.