The Classic: Revision of Aseptic Loose Total Hip Arthroplasties

Abstract

Harlan C. Amstutz was born and raised in Santa Monica, CA. After attending local schools, he enrolled in the University of California, Los Angeles where he received his bachelor degree in 1953 and his medical degree in 1956. After interning at the Los Angeles General Hospital and spending 1 year as a surgical resident at the UCLA Medical Center, he began his orthopaedic residency at the Hospital for Special Surgery in New York. On completion of his residency in 1961, he entered the Air Force as the orthopaedic surgeon at the Hospital for the 862nd SAC Division in Minot, ND. After serving in the military, Amstutz went to London where he spent 1 year at the Royal National Orthopaedic Hospital and another year at the Institute of Orthopaedics. He then returned to the Hospital for SpecialSurgery where he spent the next 5 years before returning to Los Angeles. In 1970, Amstutz was appointed Professor and Chief of the Division of Orthopaedic Surgery and Chief of the Joint Replacement Section at UCLA. Since his retirement in 1991, Amstutz became the Director of the Joint Replacement Institute at the Orthopaedic Hospital. A pioneer in the bioengineering of total joint replacements, Amstutz’s vast clinical experience and work in the laboratory have made him a popular lecturer and visiting professor. His contributions to the orthopaedic literature number more than 200. He has received many honors and awards for his work, including the John Charnley Award of the Hip Society and the Nicolas Andry Award. He has been elected president of the American Orthopaedic Association, the Hip Society, and the Orthopaedic Research Society. The following paper on aseptic loosening of total hip prostheses is an early discussion of the problem.Figure.: Dr. Harlan C.AmstutzLeonard F. Peltier, MD, PhD Aseptic loosening is the major cause of failure of conventional hip arthroplasty. The incidence is rising, and revision surgery constitutes 25% of all arthroplasties performed at the University of California at Los Angeles. Although there are few published reports of the results of revision surgery for loose total hip arthroplasties, there is almost universal agreement that the long-term results will be less satisfactory than those of primary index arthroplasty. With each failure there is loss of bone stock, both in quality and quantity, making revision increasingly difficult and results even more unsatisfactory. There are currently reports of Girdlestone procedures being performed for aseptic loosening.6,9 This study analyzes the results following revision surgery during a ten-year period, and relates them critically to those with conventional hip arthroplasty performed by the Joint Replacement Service at UCLA during the same period and using similar techniques. MATERIALS AND METHODS Eighty-eight patients underwent revision for aseptic loosening of total hip arthroplasty at UCLA during the period from October 1970 to October 1980. The criteria for inclusion in the study group were the following: (1) an acrylic fixed conventional hip metal/polyethylene arthroplasty; (2) revision surgery that was performed for suspected loosening (revisions for heterotopic bone, trochanteric problems, and recurrent dislocation were excluded); (3) no evidence of sepsis, including on pre- and intraoperative cultures, or after histologic examination of tissue obtained at surgery.11 Sixty-six of the 88 patients had a one- to nine-year follow-up period (average, 2.1 years). The hospital and surgical records, the UCLA hip evaluation forms, and the serial radiographs were reviewed in detail. Fifteen of the 66 patients underwent original operation at UCLA. The primary diagnoses in these 66 patients included a higher percentage of dysplasia and post-trauma as compared with the overall total hip arthroplasty series at UCLA3 (Table 1). The failure incidence was noticably lower in patients with rheumatoid arthritis.Table 1: Distribution of DiagnosesTwo patients had bilateral failures, but only one hip was included because of inadequate follow-up study on the contralateral hip. There were 31 men and 35 women. The average age at revision was 51.7 years. The average time to surgery was 4.0 years. For the 15 patients whose index surgery was performed at UCLA, the average time to operation was 4.8 years, and for those from “outside,” 3.7 years. Twenty-eight patients (42%) underwent hip operation prior to revision for failed hip arthroplasty: 17 had two operations, five patients had three operations, four had four operations, and two patients had more than five previous operations. Nine patients underwent prior osteotomy, shell procedure, pinning, or arthrodesis. Seventeen patients had had prior cup or hemiarthroplasty, and 13 had previously undergone failed total hip arthroplasty. The failed prosthesis types included 24 Mueller, 18 Charnley, 11 T-28, nine Bechtol, two Harris, and two Aufranc-Turner. At revision surgery, 14 patients (21%) had isolated acetabular loosening, and 25 (38%) had only femoral loosening (including 8 stem fractures). Twenty-three (35%) patients had loosening of both components, and four patients had no “gross” loosening of either component, although two had “obvious” radiographic evidence of loosening consisting of subsidence and migration or lucencies greater than 2 mm wide. The operation was performed or supervised by one of five attending surgeons. The senior author was involved in 47% of cases. The types of femoral prostheses inserted included 52 from the T-28 and TR-28 series. There were 15 wide or hemi conversion, 13 large, nine extralong, eight straight, four extrawide extralarge, two extralarge, and two custom prostheses. There were eight other exchanges, including three Charnley and five Mueller prostheses. The average operative time for revision surgery was 245 minutes. Revision of the isolated acetabular component averaged 216 minutes, while exchange of the femoral component averaged 248 minutes. Revision of both components averaged 285 minutes. Total blood loss, including postoperative suction catheter drainage, averaged 2610 cc (range, 830–8024 cc). The blood loss in eight patients who had acetabular revision only averaged 2500 cc, and in seven patients who had femoral component revision only, blood loss averaged 2316 cc. Those patients with both components revised (n = 40) averaged 2900 cc of blood loss. Hypotensive anesthetic agents were used in 28% of the patients who underwent revisions. Sixty-two per cent of the limbs lengthened, 28% remained unchanged, and 10% lost length. Prior to operation, the average limb-length discrepancy was 1.9 cm. The average gain in those who were lengthened was 1.65 cm (maximum, 5 cm). The maximum loss of length was 2 cm. RESULTS Patients’ pain, walking, and function were evaluated using the UCLA 10-point scale, where a rating of 10 denotes absolutely no pain and excellent walking and function. The average pain rating improved from 3.3 to 7.3 points, which was less than the overall index surgery improvement of 4.4 (4.0–8.4). Walking improved from 5.0 to 6.1 points, and function from 4.5 to 6.1 points, which was notably less than in the index surgery group: walking improved from 4.7 to 7.1 and function from 4.3 to 6.8. The range of motion for revision patients improved: flexion arc increased from 80.6° to 89.8°, abduction increased from 20.2° to 27.8°, and rotation are increased from 37.3° to 60.4°. These results were compared with those from the overall UCLA series, in which flexion arc increased from 76.3° to 107.6°, abduction increased from 17.5° to 28.4°, and rotation arc increased from 24.3° to 58.9°. The final flexion arc of the revision group was 8.8° less than that of the overall series. Although the overall series had less abduction and rotation arc prior to operation, there was no significant difference in the final abduction and rotation arc between the two groups. There was a high complication rate following revision surgery. There were five (7.5%) peroneal nerve palsies, with complete recovery in three and partial recovery in two. There were three nonfatal pulmonary emboli, three wound hematomas (one required evacuation surgically), and seven dislocations (10.6%), with one subsequently requiring revision surgery and component change. There were also seven subluxations, one of which subsequently loosened the acetabular component. Of the seven patients with dislocations, all had undergone at least one prior surgery, as had all but those with subluxation. Three of the patients who had dislocations had complications: two had undergone prior revision for failed total hip arthroplasty, and one had Charcot’s arthropathy. The seven patients with subluxation were similarly complicated by three failed total hip arthroplastics, one arthrodesis take-down, and one failed hemiarthroplasty. There were five trochanteric migrations (7.5%), three additional fibrous unions, and one iliac incisional hernia from a bone graft donor site. There were nine (13.6%) major complications involving the prosthesis. Six (9%) had undergone re-revision of their hip arthroplasties, including one who developed gonococcal septic total hip arthroplasty two years after revision. One patient previously mentioned underwent re-revision one year four months postrevision for recurrent dislocation, one for stem fracture three years postrevision, one for femoral fracture, one for a loose acetabular component in the chronic subluxator at 13 months postrevision, and another for loosening of both components six years postrevision. There were four femoral shaft fractures: two through cortical windows and two at the stem tip. The latter two did not have obvious cortical defects previously. One of these, although neither component was loose, was revised with a longer stem to bridge the fracture. The other was treated conservatively initially, but later required grafting and internal fixation. One patient was a 24-year-old woman with multiple epiphyseal dysplasia who had lateral hip pain and increasing limitation of function. She underwent bilateral total hip arthroplasties and has remained asymptomatic on the left side, despite the use of a small-stem stainless steel T-28 prosthesis for ten years. However, the right side was replaced with a thin-stem T-28, which fractured while she was practicing yoga three years eight months after operation. A cortical window was used to remove the femoral component and acrylic in 1974. A custom 18-cm prosthesis was too large because of the anterior femoral bow; a large-stem Trapezoidal-28 conventional 13-cm prosthesis was inserted but bulged slightly and, hence, the cortical window reattached imperfectly. Five months after operation, the patient fell and fractured her femur at the stem tip. This was treated conservatively by traction and, subsequently, a spica cast. The fracture healed with an offset, and at last follow-up examination, six years postfracture, the patient had returned to a fairly active life with the aid of crutches. The femoral component is loose, but the femur offset has prevented further subsidence and gross symptoms. She has severe arthropathy of the knees and ankles as well. Ten (15%) of the femoral shafts had bone stock deficiencies (cortical defects or windows, from prior surgery) or loosening at revision surgery. One of these ten patients had been treated conservatively for a femoral shaft fracture through the defect prior to revision for a loose stem. In six patients the cortical defects were created by the stem tip protruding through the cortex at the original surgery. All ten of these patients were noticed to have grossly loose femoral stems at revision. In addition, femoral cortical windows or defects were made intentionally or unintentionally in ten (15%) of the revision arthroplasties to extract the stem and/or acrylic. Eight of the ten occurred prior to 1976, when revision techniques were evolving. Two also had pre-existing defects. None of these patients as yet has required re-revision, although two patients fell four and seven months, respectively, postrevision, fracturing the femoral shaft at the cortical defect. The radiographs of patients who had femoral cortical defects demonstrated wider and more extensive bone-cement radiolucencies. Bone grafts were placed in six of the acetabular revisions (all had congenital dysplasia, except one who was undergoing a fourth hip surgery) and in two of the femoral revisions to close cortical defects. Protrusio and eccentric acetabular components were used in 16 (24%) revisions. The use of larger components to fill bone defects is illustrated. A 62-year-old woman with pain in her left hip fractured her femoral neck and was treated initially by pinning and subsequently by hemiarthroplasty. She underwent total hip arthroplasty ten years later. She did well for four years, but experienced increasing pain and shortening of her left leg, with a slightly restricted range of motion. Roentgenographic examination demonstrated an acetabular radiolucency of 2 mm, with proximal migration of 1.5 cm. The femoral component was also grossly loose. These findings were confirmed at operation, and the hip was revised using a 15-mm eccentric cup and an extrawide femoral component from the Trapezoidal-28 system. After operation, the patient is pain free and fully ambulatory. A 20-year-old man with generalized rheumatoid arthritis presented to UCLA in 1977 with increasing pain one year following revision of a right total hip arthroplasty. At 16 years of age, a conventional press fit ring type prosthesis was implanted, which became loose within three years. Roentgenograms show calcar resorption, acrylic fracture, and loosening of the femoral component. At operation, the femoral component was grossly loose. Both components were revised using a protrusio socket and a custom extralong, extrawide stem Trapezoidal-28 femoral component. The patient currently is 39 months postrevision and clinically doing well, with no radiographic progression of radiolucency evidenced in the last two years. He has been advised to use a single support at all times. RADIOGRAPHIC ANALYSIS Serial radiographs, made immediately, one year, and two years or longer after operation, were analyzed in 59 patients. The average radiographic follow-up period was two years six months (maximum, 8 years 6 months) (Tables 2–6). In the postoperative radiograph, 63.3% demonstrated no femoral bone-cement radiolucency, decreasing to 16.1% at last follow-up; 3.3% had 51%-99% involvement, which increased to 16.1% at last follow-up. Those who had 100% involvement increased from 3.3% to 25.8% at last follow-up (Table 2). The incidence of 100% involvement in the UCLA overall group was 0.3% after operation and progressed to only 4.0% at comparable follow-up.3Table 2: Percent Interface Femoral Bone-Cement Lucency: RevisionsTable 3: Maximum Width Femoral Bone–Cement Lucency: RevisionsTable 4: Percent Interface Acetabular Bone-Cement LucencyTable 5: Acetabular Zones with LucencyTable 6: Maximum Width of Acetabular LucencyIn the postoperative radiograph, no patients showed maximum femoral radiolucency greater than 1 mm, but after two years or more, 19.4% had 1.1–2.0 mm radiolucencies, and 22.6% had radiolucency greater than 2.0 mm maximum width (Table 3). In the overall group at two years’ follow-up, 16.1% had 1.1–2.0 mm lucencies, and 8.8% had lucency greater than 2.0 mm maximum width. Only 28.3% of the patients had no detectable radiolucency in any of the three zones on the acetabular side after operation (Table 4), compared with 36% in the overall conventional arthroplasty group. There were no hips at two years’ follow-up that were radiolucency-free, as compared with 5.7% in the overall group. After operation, 10% had complete radiolucency around the bone-cement interface, increasing to 71% at last follow-up. The involvement by zones is shown in Table 5. Fifty-five per cent of the patients had radiolucency after operation in zone 3 of the acetabulum, increasing to more than 90% at last follow-up. In addition, zone 3 averaged the widest radiolucency. The maximum width of acetabular radiolucency is noted in Table 6. After operation, 6.6% had maximum width of 1.1–2.0 mm, progressing at two years or more to 32.2%. In the overall group, 26.8% had radiolucency of 1.1–2.0 mm and 11%, 2 mm or greater at two years. The latter figure is higher than if only patients who had primary surgery had been included. DISCUSSION The reported clinical results following revision of aseptic loose conventional hip arthroplasties have varied. Eftekhar7 reviewed his personal experience with 30 patients in 1976, reporting that his clinical results were comparable with those for primary total hip arthroplasties. However, one-third of his revisions were press-fit “Ring-type” revisions. Krengel10 reviewed 64 revisions, including 20 metal-on-metal acrylic-fixed and three press-fitted prostheses operated on by a group of orthopedic surgeons in Seattle, and found that the clinical results were less than satisfactory. In the present group of 66 patients followed up for one to nine years, postoperative pain, walking, and function ratings as well as flexion arc were significantly lower as compared with the overall UCLA conventional hip arthroplasty series.3 All of the reported series in the literature, including the present series, have reported longer operative times, increased blood losses, and higher complication rates than with primary surgery, all of which reflect the complexity and magnitude of the revision operation. The literature reports an average blood loss of from 120010 to 2600 cc for this series, although apparently postoperative suction catheter drainage was not included in other series that were reported. There were several patients from the present series who had exceptional loss. One patient who was operated on early in the revision experience lost more than 8000 cc. The operative time also has varied in the literature (range, 168 minutes to more than 5 hours).10,13 Eftekhar reported an operative time of 210 minutes, twice that of his primary surgery. The present results and those of Pellicci et al13 demonstrated a significantly shorter operative time when only one component was exchanged. Also, the magnitude of each revision case apparently depends on the unique problems encountered, and those patients who had long stems (25 cm) were the most difficult, requiring more operative time and blood loss. As the authors’ experience with revision surgery increased, the ability to deal with these problems improved, with a consequent diminution in operative time and blood loss. When the patient’s condition permits, the authors use hypotensive anesthesia and try to obtain blood pressure levels below 70 mm systolic. When the patient has lost 5 units of blood, fresh frozen plasma and platelets are added to the blood replacement program, and compression is carefully applied using an ace bandage spica after operation. This is of vital importance in preventing massive blood losses. A thorough debridement, removing all scar tissue to the level of healthy tissue, assists in controlling blood loss. Despite the long operative time, high blood loss, and magnitude of the procedures, there have been no fatalities in this or other series reported. Systemic complications have not been greater than with conventional arthroplasties, and no myocardial infarctions or strokes have occurred.7,10,13,14 There was no correlation between those patients who developed nerve palsy and the amount of limb lengthening. The high incidence of nerve palsy may be a reflection of the intense scarring about the hip and the immobility of the nerves encountered at the time of surgery, which were not always dissected free, combined with excessive traction and poor limb support. The current policy is to dissect and protect the sciatic nerve carefully, as well as to support the limb during the procedure, thereby avoiding a traction injury. The literature emphasizes the importance of trochanteric osteotomy to obtain wide exposure, facilitating the revision surgery.7,10,13,14 The incidence of nonunion and migration of trochanteric reattachment has been high. Pellicci et al13 reported a 13% incidence (3 times that of primary surgery). With the standard two-wire interlocking technique previously described, the incidence was 7.6%.5 Trochanteric reattachment failure and migration have usually been associated with a small or fragmented trochanter placed against a poor bone base for union, either acrylic bone cement or the lateral cortical shaft. A rather large piece of trochanter should be carefully removed using a Gigli saw or a wide 2.5-inch osteotome. If the trochanter can be sectioned below the vastus ridge and the vastus lateralis subsequently repaired, attachment strength will be enhanced. The bone surfaces should be shaped and prepared for maximum contact using two vertical wires placed approximately 1 cm apart and held by the horizontal wire while the interlocking knots are completed following the usual protocol.2 If reattachment is not secure when tested manually at surgery, the patient should be restricted to bed rest for one or two weeks and ambulation progressed gradually. Consideration should also be given to using a Murray abduction-type brace. The high dislocation rate in this and other series is undoubtedly multifactorial.13 All of the patients had undergone at least one previous surgery prior to revision, with poor muscle stock and turgor. There has also been an association with trochanteric migration and/or failure to restore muscle biomechanics completely due to bone or muscle stock loss. The use of special revision stems and sockets (protrusio and eccentric) enables the surgeon to restore the biomechanics more efficiently. The authors generally mobilize their patients on the third day after operation, beginning ambulation with crutches. However, when trochanteric reattachment is insecure or there are other extenuating circumstances, e.g., extreme debility or poor muscle tone, extra precautions are recommended; patients undergoing revision surgery are at higher risk for dislocation. Loosening is the most frequent and serious complication. The failure rate with recurrent loosening has been high even in series with relatively short follow-up periods. Weber14 reported a group of 54 patients who had been followed up for six to eight years. Despite the fact that 80% underwent socket revision only, 13% required a second operation after three years and a further 18% required a second operation within three years and, sometimes, a third operation. The loosening rate of two other series was 6.25% and 14% in patients followed up for 2.0–8.7 years.10,13 In this series there was a high correlation of loosening to bone stock deficiencies, especially femoral. The loosening rate requiring revision was 6% with an average follow-up period of just more than two years (range, 1–9 years). What is alarming but not reported in the literature is the evidence of poor fixation, as judged by the width and extent of the radiolucencies. Pellicci et al13 concluded that fixation in the 86% that did not fail mechanically was as good as that found after primary surgery, although the incidence of progressive radiolucencies was 22%, and 14% failed. The authors’ studies are the first to report a detailed radiographic analysis, revealing a 29% incidence of serious femoral and acetabular radiolucencies (100% plus >2 mm on the acetabulum and 100% femoral plus >1 mm, with or without subsidence, or have been revised for loosening). Of additional concern is the youth of the group of aseptic loosening patients (average age, 50 years). Several patients had loose prostheses but could not be revised and, hence, were not included in the study; conversion to Girdlestones was performed. In a series from the University of Toronto Hospitals, 22% of patients were converted to Girdlestones, some of whom were, or became, septic. They reported a strikingly high (32%) sepsis rate, whereas infection in this series was 1.5% and in others, only slightly higher.7,10,13 Buchholtz,6 however, reported that 16% of his “aseptic loosening cases” proved to be septic at operation. It must be emphasized that careful preoperative evaluation is essential, with repeat aspirations prior to operation, cultures obtained intraoperatively, and inspection of tissue at surgery, with the most suspicious tissue obtained for culture to rule out infection.4,11 There is a considerably higher risk of femoral fracture during and following revision surgery. In this series there were four cases (6%), excluding one patient who had aseptic loosening and in whom revision surgery was planned; but, at surgery, a fracture occurred, which was fixed with an intramedullary nail, and a Girdlestone was performed. Subsequently, the patient formed massive heterotopic bone, and fusion occurred. There are widely ranging opinions on how to remove the acrylic bone cement and fractured stems. Weber14 described a technique of window removal and repair using a plate and silicone tube. Eftekhar7 utilized a window to facilitate removal, but removal without a window is strongly recommended, if possible, to preserve structural integrity of the cortical tube and thereby to contain the acrylic so that compaction can be optimized. The authors have continually improved their surgical revision techniques to minimize complications and to obtain more optimal fixation. The necessity of preliminary planning to determine optimal size and shape of both the femoral and socket components, with sufficient size to obtain an acrylic layer (ideally 2–4 mm in thickness) surrounding the components, should be emphasized. Size can be determined by using the magnification factors of the femoral head of the failed component, templates, and/or CAT scan. The authors recommend and frequently use protrusio or eccentric sockets, extrawide, extralarge femoral prostheses in either 13-, 18-, or 25-cm lengths, and, occasionally, other custom prostheses. The use of an extrawide, extralarge stem and a protrusio cup is illustrated in a 76-year-old man in whom a Mueller prosthesis loosened and was revised to a Bechtol-type arthroplasty three years two months after operation. Three years postrevision it again loosened, with extensive osteolysis of bone and femoral stem penetration. Re-revision to an extralarge, extrawide TR-28 prosthesis was performed to fill the large cortical tube. Because the follow-up period was short (6 months), he has not been included in this study, although he had good pain relief. His activity level, however, is commensurate with his age. Since even the best preliminary planning may not allow for extenuating circumstances, a wide variety of prostheses and instruments should be available. The stem should be long enough to bridge defects or weakened areas, and made from newer high-strength alloys. For dysplastic cases, the straight stems with a variety of neck lengths are useful; these are made of Tivanium (Zimmer, Warsaw, Indiana) (titanium, 6 aluminum, 4 vanadium), as are the extrawide, extralarge prostheses. A wide hemiconversion and other TR-28 stems are made from micrograin (hot isostatic pressed cobalt chrome). If the available shelf or catalogue lengths are not optimal, a custom length must be ordered or the prosthesis cut to an appropriate length before or during surgery. If the stem to be removed is rigidly fixed distally, or if there is compacted acrylic to be removed that may require the use of a high-speed burr, then it is wise to await thorough removal at surgery before cutting the prosthesis to length. If a defect is encountered that was unsuspected or created, then it can be appropriately bridged. Routine use of extralong stems is not recommended because they are often difficult to insert due to the femoral bow and because it is difficult to compact cement below the isthmus. They are even more difficult to revise should they become loose. Trochanteric osteotomy should be carefully planned if the limb is to be lengthened. The trochanter can be mobilized after lengthening by sharply excising all of the scar tissue proximal to the trochanter down to a fat or muscle layer. Thorough debridement is essential, because it is impossible to rule out low-grade sepsis. However, in a situation in which only one component is grossly loose, the quality of the fixation of the other component must be carefully evaluated to determine whether it should be revised. When the acetabular component is not grossly loose, the extent and width of the radiolucent lines, the age and activity level of the patient, and the type of component should be evaluated. If the component is not optimally fixed, positioned, or matched with the planned femoral component, it may be desirable to change the acetabular component. If, for example, the radiolucency is in the 1–2 mm range, there is good remaining bone stock, and the surgeon believes that a new component would be desirable and better fixation could be achieved, then there may be a possible advantage in revision. This assessment must also evaluate the medical condition of the patient, as hypotensive anaesthesia is most helpful in obtaining a dry field. There may be hyperemia, which would affect the quality of fixation that can be achieved. When the acetabular component is loose but the femoral component is tight within bone, once again, careful evaluation of the bone stock, quality of fixation, prosthetic type and material, and age and activity levels of the patient must be considered. If, for example, a patient had an early Mueller-type femoral stem and/or a prosthesis made from a non-superalloy and the prosthesis has already had a number of years of fatigue cycling, it may be preferable to exchange the component if there is good remaining bone stock. The decision will depend on the need and the surgeon’s ability to achieve good fixation. These are relatively new concepts regarding exchange of “tight” components and may be controversial. One 78-year-old woman presented in 1975 with progressive hip pain one year after a Mueller total hip arthroplasty had been performed for intracapsular fracture at another institution. She had good initial relief, but pain and obvious loosening of the femoral component developed. The acetabular component developed was not loose at the time of revision, although there was a radiolucency of about 1 mm at the interface. It was elected just to exchange the femoral component with a Mueller component. At that time (1975), information on the high loosening rates associated with this stem design were not known. Although good initial fixation was achieved and the patient returned to an active life-style for 5.5 years, there was increasing pain and subsequent femoral loosening. The acetabular component was also loose, and both components were changed using a wide stem TR-28. In retrospect, it would have been better to revise both components initially and to utilize a femoral component of better design made from superalloy. To minimize loss of bone stock, a “tight” socket should be removed by bisecting it with a high-speed scalpel-like cutter. Those components that are grossly loose can be removed with curved gouges, taking care not to lever against the acetabulum, which may fracture the walls. After the polyethylene has been removed, the acrylic should be divided with an osteotome and removed. The membrane, which is invariably present, should be removed using a sharp curette and high-speed burrs. All acrylic contained in the fixation holes is thoroughly removed. When acrylic has penetrated into the pelvis through perforations or defects, it must be stabilized, divided, and removed carefully. If infection can be definitely ruled out by frozen section, it may be advisable not to attempt to remove all intrapelvic acrylic and risk bladder or vessel injury. Preoperative veinography may be useful in preoperative evaluation. Once all of the contents have been removed, remaining bone stock should be assessed. Bone grafting of defects is preferred to the use of adjunctive protrusio rings or intramedullary metallic devices unless absolutely necessary because of the mismatch of materials, metal, screws, and acrylic bone, as well as the interference these objects present to obtaining good interdigitation of acrylic. There are also stress risors in the surrounding acrylic. If the femoral stem is intact, removal of the component, even if it is not grossly loose, is usually not difficult. The acrylic should be divided using an osteotome to a level below the curved portion of the stem. The Buchholz stem, however, is particularly difficult to remove because of the negative angle of the stem recess. Any stem type with which the surgeon is not thoroughly familiar should be obtained prior to operation and inspected to assess the problems with removal accurately. Fractured stem removal is facilitated by removing the acrylic that surrounds the fractured distal stem using a high-speed scalpel cutter to a depth of at least 1 cm. The authors recommend drilling a hole into the fractured stem and gripping the stem as described by Harris et al8 A fiberoptic light is essential to visualize the acrylic and canal contents. The authors strongly advise against the use of windows or gutters, which can make fixation more difficult and lead to femoral fracture. As demonstrated in this series, these defects may lead to femoral fractures. If there is a defect, it should be bypassed by a stem at least two shaft diameters to avoid a stress risor. Additional acrylic is removed by the divide-and-conquer method. Thick acrylic can be thinned from within toward the periphery using a high-speed burr, osteotomized, and removed. The image intensifier has been helpful for some surgeons, although the authors prefer to palpate the femur distally as the burring progresses. Once all of the acrylic has been removed, it is advisable to radiograph and check carefully for any defects. Consideration should be given to bone grafting for both central and lateral defects of the acetabulum, as well as the proximal femur. Preparation of the acetabulum should include maximal cancellous bone surface with four-quadrant fixation. Thin cortical bone may be grooved if quadrant fixation is poor, although creating a bleeding surface must be considered versus the advantage of grooving and increasing the surface area. Acrylic should be contained by filling defects, preferably with bone grafts and intramedullary plugs. Surfaces should be cleaned and dried with pulsating lavage, suction, and gauze. Where cancellous bone is present, acrylic is used in the low-viscosity state and injected. The blood pressure should be lowered maximally with hypotensive anesthesia at the time of insertion. Occasionally, a pulsatile bleeding vessel can be sealed off within the femoral canal by means of bone wax introduced on a spatula. When the bone is cortical and the surface moist, the acrylic should be introduced in late dough stage to tamponade the oozing surface. A sturdy syringe gun is necessary to fill the femoral canal with acrylic when it is already in the dough stage. There is no advantage to early acrylic insertion when all that remains is a cortical tube. The femoral component should be inserted in neutral or slightly valgus position. Team speed and coordination are essential. Finally, a reliable trochanteric reattachment method should be utilized; the authors recommend two vertical wires. These improved surgical techniques developed during the past ten years have assisted in revision of aseptic loose conventional hip arthroplasties and, it is hoped, will reduce the high complication rates and poor results. However, patients must be advised that they are at greater risk for complications, especially recurrent loosening. Patients should be carefully followed up, and if radiolucencies appear and progress, protected weight-bearing should be advised. SUMMARY Sixty-six patients were revised for aseptic loosening of their conventional hip arthroplasties: follow-up periods ranged from one to nine years. In comparing them with an overall conventional arthroplasty series, there was a higher failure rate with dysplasia and post-traumatic patients, and a lower incidence in osteoarthritic and rheumatoid patients. The average time to revision was four years. The patients were eight years younger than those in the overall UCLA conventional hip arthroplasty series. Forty-two per cent had undergone hip surgery prior to the original hip arthroplasty that failed. The average improvement, as well as the follow-up pain, walking, and function ratings, and the post-operative flexion arc were less than those in the overall conventional arthroplasty series. The quality of femoral and acetabular fixation obtained at revision was considerably inferior to that of the primary surgery. Six patients (9%) have already required re-revision of their hip arthroplasties. In a further 20%, the radiolucencies progressed substantially in extent and width, and are radiographically loose. Although these patients are relatively asymptomatic, prognosis is guarded. Forty-four per cent had no complications and are radiographically well fixed. Other complications included trochanteric migration (7.6%), dislocation (10.6%), and peroneal nerve palsy (7.6%), but there were no deaths or other serious medical complications and only one case (1.5%) of sepsis.