The total joint registry at our institution was used to identify all patients who were found to have a pelvic discontinuity during the course of revision of an acetabular component of a total hip prosthesis. The registry contains a specific code for acetabular fracture, making identification of this population of patients possible. Pelvic discontinuity was defined as a complete separation of the superior aspect of the pelvis from the inferior aspect due to fracture or bone loss through the acetabulum2. Clinical data on all patients were collected prospectively before and at regular intervals (at one, two, and five years and every five years thereafter) after the revision arthroplasty. These data were compiled retrospectively by one of us (D. J. B.) and were analyzed at the time of the study.
An anteroposterior radiograph of the pelvis and an anteroposterior and a true lateral radiograph of the affected hip were made before the index revision and after the operation at regular intervals, with use of a standardized method of positioning. Judet radiographs were made for some patients at the discretion of the treating surgeon. The preoperative and postoperative radiographs were reviewed by the same one of us who compiled the data. The appearance of the cemented acetabular components on the postoperative radiographs was evaluated with use of the method of Hodgkinson et al.8. The component was considered to be probably or definitely loose if a complete radiolucent line of any width was present at the bone-cement interface, if there was measurable migration of the component, or if there was a measurable change in the position of the component. An acetabular component that had been inserted without cement was considered to be probably or definitely loose if migration or a change in the position of the component had occurred or broken fixation screws were present. Changes in the position of the component were measured with use of the method of Mohler et al.10. According to this method, the position and inclination of the socket are determined relative to anatomical landmarks (radiographic teardrops). In hips that had an antiprotrusio cage, the implant was considered loose if there was measurable migration or change in the position of either the cage or the socket or if fixation screws had backed out or broken1.
The pelvic discontinuity was considered to be definitely healed if bridging callus or trabecular bone was visible across the site of the discontinuity, and it was considered to be possibly healed if there were no signs of nonunion (such as failure of the hardware or displacement of the fracture) and the site of the fracture was too poorly visualized (because of the presence of bone grafts or the implant) to allow a definite determination of healing to be made. This was the case in five hips, all of which had been treated with an antiprotrusio cage. The discontinuity was considered to be unhealed if discontinuity was still visible or if there were other signs of failure of fracture-healing, such as failure of the hardware or displacement of the fracture.
We subclassified the pelvic discontinuity with use of the American Academy of Orthopaedic Surgeons' classification of acetabular bone deficiencies, which defines pelvic discontinuity as a distinct entity and classifies it as type-IV bone loss2. We subclassified the pelvic discontinuity as type IVa if it was associated only with cavitary bone loss (type II, according to the American Academy of Orthopaedic Surgeons' classification) or mild-to-moderate segmental bone loss (type I, according to the American Academy of Orthopaedic Surgeons' classification), as type IVb if it was associated with segmental bone loss (type I) or combined bone loss (type III, according to the American Academy of Orthopaedic Surgeons' classification), and as type IVc if it was associated with previous irradiation of the pelvis with or without cavitary or segmental bone loss. Three hips were classified as type IVa; twenty-two, as type IVb; and six, as type IVc.
The method of treating the pelvic discontinuity and performing the acetabular reconstruction was chosen by the operating surgeon and was based on his preference for the reconstructive technique as well as on the unique circumstances of each case (Table I). In thirteen hips, the discontinuity was treated with a Burch-Schneider antiprotrusio cage (Protek, Bern, Switzerland); the cage alone was used for fixation in twelve hips, and the cage was augmented with a plate in one. In seven hips, a single posterior column plate and a socket inserted without cement was used; the socket was hemispherical in six hips and of the oblong, bilobed type in one. Dual plates for fixation of the anterior and posterior columns were used in seven hips, five of which had insertion of a socket with cement and two, without cement. Two hips had insertion of a cup only, without ancillary fixation of the discontinuity; one of these hips had insertion of the socket with cement, and the other had insertion of the socket without cement. The remaining two hips were treated with a resection arthroplasty.
All hips had placement of particulate fresh-frozen bone allograft at the site of the discontinuity. Twelve hips also had insertion of autogenous particulate bone graft, and nine hips also had insertion of structural fresh-frozen bone allografts. Two femoral heads were used in two of the nine hips; three femoral heads, in one; the distal part of the femur, in four; the femoral head and the distal part of the femur, in one; and a whole acetabulum, in one. The structural grafts were used in combination with a socket inserted without cement and with plates in two hips, with an antiprotrusio cage in two, and with a socket inserted with cement and with plates in five.
Demographic risk factors for pelvic discontinuity were tested for statistical significance with use of Fisher's exact test, with a p value of less than 0.05 considered significant.
The thirty-one hips (twenty-seven patients) that had a pelvic discontinuity were identified from a total of 3505 revisions of the acetabular component that were performed between 1969 and 1995. All thirty-one pelvic discontinuities were identified after 1983, and twenty-five were identified after 1990. Twenty-nine of the pelvic discontinuities were present before the revision arthroplasty at which they were identified, and two developed during the revision procedure. Two patients had bilateral discontinuity. Two additional patients were managed for pelvic discontinuity on the same side on two separate occasions because the first treatment had failed. (Both operations are included in the study.) One of these patients initially was managed with insertion of a socket without cement and one plate and then had revision with insertion of a socket without cement and two plates; the patient died within two years after the second operation. The other patient initially was managed with insertion of a socket with cement and two plates and then had revision with insertion of an antiprotrusio cage and one plate.
The mean age of the twenty-seven patients at the time of the operation for the treatment of the pelvic discontinuity was sixty-one years (range, thirty-eight to eighty years). Seventeen right and fourteen left hips were involved. Twenty-eight hips were in women, and three were in men. Twenty-seven of the twenty-nine established discontinuities (that is, those that did not occur intraoperatively) were in women. The underlying diagnosis leading to the primary hip arthroplasty was osteoarthritis in thirteen hips, rheumatoid arthritis in eight, congenital dysplasia in five, osteonecrosis of the femoral head in two, posttraumatic arthritis due to a previous acetabular fracture in two, and a tumor of the proximal part of the femur in one. Six hips had a history of treatment with high doses of radiation to the pelvis. The involved hips had had a mean of 2.1 previous arthroplasties; nine hips had had one previous arthroplasty, twelve had had two, eight had had three, and two had had four. The failed acetabular component that was removed during the index procedure had been inserted with cement in twenty-three hips and without cement in three; the remaining five hips had a bipolar implant.
Women (p < 0.001) and patients who had an underlying diagnosis of rheumatoid arthritis (p = 0.003) were identified as being at significantly increased risk for pelvic discontinuity. The demographic risk factors were identified by comparing the frequency of the factors in the study group with the frequency of the same factors in the database of all 3505 acetabular revisions that had been performed at our institution during the same time-period. Whether or not the patient had a history of treatment with irradiation of the pelvis was not recorded in our database of other acetabular revisions; thus, statistical testing for this as a risk factor for discontinuity was not possible.
The radiographic findings associated with pelvic discontinuity in the twenty-nine hips that had the discontinuity before the revision included a visible fracture line through the pelvis and translation or rotation of the inferior aspect of the hemipelvis relative to the superior aspect (Fig. 1). The fracture line was clearly visible on the anteroposterior radiographs of twenty-two pelves. Preoperative Judet radiographs of only seven pelves were available, and they clearly showed the discontinuity in five. Kohler's line was broken and the inferior aspect of the hemipelvis was translated medially through the discontinuity in twenty-three hips. Asymmetry of the obturator rings, suggesting rotation of the inferior aspect of the hemipelvis relative to the superior aspect through the discontinuity, was present in ten hips. In the twenty-seven hips in which the socket was in place (the other two had had a previous resection), the center of rotation of the failed hip replacement was a mean of forty-two millimeters (range, eighteen to seventy-five millimeters) superior to a line drawn between the radiographic teardrops. The failed socket had protruded medial to Kohler's line in seventeen of these hips.
Intraoperatively, the discontinuity appeared to be an unhealed transverse acetabular fracture that had occurred through areas of acetabular bone loss in all twenty-nine hips in which the discontinuity had been present before the operation. The inferior aspect of the hemipelvis, including the pubis and the ischium, could be moved relative to the superior aspect of the hemipelvis through the unhealed fracture or bone defect by stressing the inferior aspect in an anterior-posterior direction.
Two patients (two hips) were managed with a resection arthroplasty when the discontinuity was identified. Two additional patients (two hips) died within two years after the operation for the treatment of the pelvic discontinuity. One of these patients had been managed with a socket inserted without cement and with two plates, and the other had been managed with a socket with cement and without plates. Neither of these patients had had additional operations on the ipsilateral hip before they died. Thus, twenty-seven hips were treated with revision and were followed for a minimum of two years after the operation for the treatment of the pelvic discontinuity or until they had a repeat revision. No patient was lost to follow-up. The mean duration of follow-up was three years (range, 0.2 [a repeat resection] to seven years).
Nine of the twenty-seven acetabular reconstructions failed and necessitated revision or removal of the component. Four of the failures were due to aseptic acetabular loosening. In all four hips, a socket resting primarily on multiple structural bone grafts became loose and migrated into the grafts (Figs. 2-A, 2-B, and 2-C). In three of the four hips, the construct consisted of a socket inserted with cement and two plates, and in one it consisted of a socket inserted without cement and one plate. Four other failures were due to recurrent dislocation, necessitating a reoperation two months to 2.5 years after the index procedure. At the time of the reoperation, the discontinuity had healed in two hips and had not healed in two. The socket was stable but was removed from three hips because of its position. In the fourth hip, the polyethylene liner of a socket that had been inserted without cement was exchanged; the socket itself was radiographically loose at the time of the most recent follow-up 2.5 years later, but it had not been revised. The etiology of the recurrent dislocation in three of the four hips was thought to be, in part, a suboptimum position of the socket; although no cup was retroverted, all three were abducted between 50 and 70 degrees. All three hips also had some degree of abductor weakness, but none had a trochanteric nonunion. In the remaining unstable hip, a tumor prosthesis was used on the femoral side; the components were well oriented, and the dislocation was thought to be due to abductor weakness. The ninth failure was due to deep infection at 1.3 years. In that hip, the socket had been inserted without cement and was supported mostly by allograft. A resection arthroplasty was performed because of the infection, and at that time the socket was seen to have migrated intrapelvically and the discontinuity was unhealed.
A satisfactory result was defined as no additional operations on the acetabulum, a stable acetabular component, definite or possible healing of the discontinuity, and the absence of severe pain. According to these criteria, the result was satisfactory for sixteen (59 percent) of the twenty-seven hips. All three of the hips that had type-IVa bone loss (Figs. 3-A and 3-B), ten of the nineteen that had type-IVb bone loss (Figs. 4-A and 4-B), and three of the five that had type-IVc bone loss had a satisfactory result.
A Burch-Schneider antiprotrusio cage was used to reconstruct thirteen hips, and the socket remained stable in all thirteen. In eleven of these hips, the discontinuity definitely (six hips) or possibly (five hips) healed. Eleven of the thirteen hips had a satisfactory result. Of the two that had an unsatisfactory result, one had a reoperation because of recurrent instability at two months and the other had an unhealed discontinuity in association with a stable socket that had not been treated with a reoperation. Two of the thirteen hips had a single structural allograft, and the discontinuity healed in both; none of the thirteen hips had multiple structural allografts. The antiprotrusio cage was used in nine hips with type-IVb bone loss and in four with type-IVc bone loss. The one mechanical failure, due to an unhealed discontinuity, was in a hip that had type-IVc bone loss; however, with the numbers available, no significant difference in outcome was detected between the hips with type-IVb bone loss and the hips with type-IVc bone loss that had been treated with an antiprotrusio cage.
Eight hips were reconstructed with use of a socket inserted without cement and with one posterior plate (seven hips) or two plates (one hip). Of these eight hips, five had a stable socket at the time of the latest follow-up and six had definite healing of the discontinuity. Four of the eight hips had a satisfactory result. Of the four failures, one was due to acetabular loosening in association with an unhealed discontinuity; one, to infection (the socket was also loose, and the discontinuity was unhealed); and two, to recurrent dislocation. One of the hips with recurrent dislocation needed an acetabular revision; the cup was stable and the discontinuity was healed at the time of the revision. The other hip with recurrent dislocation was treated with an exchange of the acetabular liner; at the time of the latest follow-up, the discontinuity was healed but the socket was radiographically loose.
Two of the sockets that were inserted without cement were placed against structural bone graft; the distal part of the femur was used in one, and two femoral heads were used in the other. The amount of the socket that was covered by the bone grafts was estimated to be 50 and 70 percent, respectively. Neither socket was stable at the time of the latest follow-up, accounting for two of the three loose sockets in this treatment group. Three of the eight hips in this group had type-IVa bone loss, and five had type-IVb. No hip with type-IVa bone loss needed structural bone grafts, and all three had a satisfactory result. In contrast, four of the five hips that had type-IVb bone loss (two of which had structural grafts) had an unsatisfactory result, and only two of the five had a stable socket and a healed discontinuity at the time of the latest follow-up.
Five hips had insertion of a socket with cement and two plates without use of an antiprotrusio cage. None of these hips had a satisfactory result. Four hips had failure because the socket loosened. In two of these hips the discontinuity was healed, and in two it was not. The fifth hip had a reoperation because of recurrent dislocation at three months, and the hip subsequently became infected. All five hips in this treatment group had structural bone grafts (a whole acetabular allograft, a distal femoral allograft, three femoral heads, two femoral heads, and the distal part of the femur with a femoral head in one hip each). The three hips that were treated with multiple structural grafts and the one hip that was treated with a distal femoral graft had mechanical failure, whereas the one hip that was treated with a whole acetabular allograft failed because of recurrent dislocation and subsequent infection.
One of the two hips that had an acute intraoperative discontinuity was treated with a socket inserted without cement, no plates, and no bulk structural allograft. That hip had a mechanically stable construct and a satisfactory result.
Definite or possible healing of the discontinuity was associated with the likelihood of a satisfactory clinical result. Eleven of the fifteen hips that had definite healing had a satisfactory result. The reasons for the unsatisfactory results in the other four hips were acetabular loosening in three hips and a reoperation because of recurrent dislocation in one hip. All five hips that had possible healing also had a satisfactory result. When healing is excluded as a criterion of success, six of the seven hips that had an unhealed discontinuity had an unsatisfactory result. In two of these hips, the acetabular component had to be removed, to treat recurrent dislocation, before the discontinuity could heal. One hip became infected. In three hips, the cup was loose; two of these cups were revised, and one was loose radiographically. Only one hip (a hip that had type-IVc bone loss and was treated with an antiprotrusio cage) with an unhealed discontinuity had an intact, stable socket and caused little pain.
Healing of the discontinuity and stability of the socket were assessed separately as a function of the classification of the discontinuity and the type of treatment. For this analysis, we excluded three hips from which the implant had been removed within two years after the operation for reasons other than loosening. (Two implants were removed because of recurrent dislocation and one, because of infection.) Of the remaining twenty-four hips, fifteen had definite healing of the pelvic discontinuity, five had possible healing, and four were unhealed. Definite healing occurred in all three hips that had type-IVa bone loss, in ten of the sixteen that had type-IVb bone loss, and in only two of the five that had type-IVc bone loss. Definite healing was most frequent in the hips in which a socket had been inserted without cement and with a pelvic plate or plates; six of these seven hips had definite healing. Selection factors may have partially accounted for the higher rates of definite healing in this treatment group: nearly half of these hips had had type-IVa bone loss, whereas all hips that had been treated with an antiprotrusio cage and all that had been treated with a socket inserted with cement and two plates had had type-IVb or IVc bone loss. The discontinuity was definitely healed in seven of the eleven hips that had been treated with autogenous bone graft and in eight of the thirteen that had not been so treated.
Stability of the socket was assessed as a function of the classification of the discontinuity and the type of treatment in the same group of twenty-four hips. The socket was stable in eighteen hips (all three that had type-IVa, eleven of the sixteen that had type-IVb, and four of the five that had type-IVc bone loss), and it had loosened in six. The highest rate of stability of the socket was noted in hips that had been treated with an antiprotrusio cage (twelve of twelve) and in those that had been treated with a socket inserted without cement and with a pelvic plate or plates (five of seven). At the latest evaluation, seventeen of the twenty-four hips had a mechanically stable construct (defined as a stable socket and a possibly or definitely healed discontinuity). A mechanically stable construct was most frequently obtained in hips that had been treated with an antiprotrusio cage (eleven of twelve) or a socket inserted without cement and with a pelvic plate or plates (five of seven), and it was noted least frequently in hips that had been treated with a socket inserted with cement and two plates (zero of four). None of the four hips that were reconstructed with use of multiple structural grafts were mechanically stable.
Complications that were related to the index operation included recurrent dislocation in four hips; a partial sciatic-nerve palsy in three (all of which ultimately had partial recovery); and a deep infection, a hematoma necessitating evacuation, a retained drain necessitating removal, a urinary tract infection, and intraoperative fluid overload in one hip each.
Pelvic discontinuity is recognized as a unique form of acetabular bone deficiency encountered during revision total hip arthroplasty. The literature on this subject that we are aware of is confined to reports of a few cases, most of which are mentioned as subgroups of larger series of revision hip arthroplasties4,9,11-13,15,20,21. The purpose of the current study was to review a large enough experience to allow us to identify factors associated with the development of this difficult condition and to evaluate the results of treatment.
Our findings show that pelvic discontinuity is uncommon: it was identified during only thirty-one of 3505 consecutive acetabular revisions. All discontinuities were identified after 1983, although patients who had acetabular revision before 1983, in an era when pelvic discontinuity was not commonly discussed, may have had discontinuity that was not recognized or identified as such by the operating surgeon. Nevertheless, the fact that twenty-five of the thirty-one discontinuities were identified after 1990 suggests that the prevalence of the problem may be increasing as patients who have had hip arthroplasty live longer, have bone loss, and become at risk for this complication.
Most of the discontinuities appeared to be unhealed transverse acetabular fractures that occurred through areas of pelvic bone loss. It may be theorized that discontinuities develop as stress fractures through areas of weak or deficient pelvic bone and subsequently—for mechanical or biological reasons, or both—fail to heal and thus become nonunions. In 1972, Miller reported on nine patients who had a late acetabular fracture through the ischiopubic region after a total hip arthroplasty; many of the fractures appeared to have been pelvic discontinuities, and none healed spontaneously9. Miller believed that these discontinuities represented pelvic stress fractures. Many of the patients in our series had reduced quality or quantity of bone or reduced healing capacity, which put them at increased risk for discontinuity. Pelvic discontinuity was more common in women and in patients who had rheumatoid arthritis, possibly because of poorer bone quality. All nine patients in Miller's report were women9. Most of the hips with discontinuity in our series had extensive associated bone loss, suggesting that this predisposes to discontinuity. Previous treatment with irradiation of the pelvis also appeared to be a risk factor, but statistical testing was not possible because this parameter was not recorded in our database of all acetabular revisions. Nevertheless, the frequency of previous irradiation in the current series (six of thirty-one hips) and its relative infrequency in the general population suggest that these patients are also at risk for discontinuity.
The current report provides results of different methods for the treatment of pelvic discontinuity. Differences with regard to the patients, the severity of the bone loss, and the duration of follow-up all preclude definitive comparison of the treatment methods; however, some clear-cut trends emerge from these data. Patients who had good remaining pelvic bone stock (type-IVa bone loss) had a higher likelihood of successful treatment than did those who had severe segmental bone loss (type IVb) or those who had had previous treatment with irradiation of the pelvis (type IVc). Insertion of a socket without cement in combination with stabilization of the discontinuity with a plate was successful in all three patients who had type-IVa bone loss. In contrast, patients who had type-IVb or IVc bone loss had the best results with reconstruction in combination with use of a Burch-Schneider antiprotrusio cage.
Eleven of the twelve hips that were followed for at least two years after treatment with an antiprotrusio cage had a mechanically stable construct. In those that had a type-IVb or IVc defect, the cage appeared to function more effectively than did the other methods of reconstruction. However, the cage has been used as a treatment method most recently and the duration of follow-up therefore was the shortest for these patients; thus, more time will be needed to determine if this technique is successful in the long term. The use of Burch-Schneider cages for the treatment of pelvic discontinuity also has been reported by other authors. Rosson and Schatzker16 and Possai et al.14 reported on one patient each. Garbuz et al. reported on a series of hips that had been revised with use of massive pelvic bone grafts; an unspecified number of these hips had pelvic discontinuity5. Those authors found that protection of the construct with a Burch-Schneider cage was associated with a successful reconstruction. In the current study, use of the Burch-Schneider cage was not associated with as high a rate of definite healing of the discontinuity as was use of a socket inserted without cement and with a pelvic plate or plates. This may be due in part to the fact that the cage often obscured the site of the discontinuity radiographically, so more patients who had been managed with a cage were categorized as having a defect that was possibly but not definitely healed. It is also possible that, in some patients who had been managed with a cage, pelvic healing did not occur but the presence of the cage led to effective bypass of the discontinuity and satisfactory function of the hip.
The use of a pelvic plate or plates and a socket inserted without cement produced a mechanically stable construct in five of seven hips that were followed for at least two years and was associated with the highest rate of definite healing. This method was particularly successful in hips that had type-IVa bone loss (all three were treated successfully), and it was somewhat successful in those that had type-IVb bone loss (two of four hips were mechanically stable). We prefer to use a socket inserted without cement in combination with a pelvic plate or plates when sufficient contact between the socket and the native bone is possible to provide a high likelihood of bone ingrowth. When much more than 50 percent of the socket is covered by bone graft, we usually consider protecting the graft with an antiprotrusio cage and inserting the socket with cement.
Insertion of a socket with cement and use of double pelvic plates produced the poorest mechanical results in this study; none of four hips that were followed for at least two years after such treatment were mechanically stable. Part of the reason for these failures may be that the technique was used exclusively in hips that had type-IVb or IVc bone loss. Furthermore, in three of the four hips, the construct was used in combination with more than one femoral head graft or a femoral head and a distal part of a femur rather than in combination with a single piece of structural graft. In comparison, in a report on twelve hips that had a large pelvic bone defect, Stiehl identified eight hips with pelvic discontinuity19. All twelve hips were treated with structural bone grafts (five whole acetabular grafts, two partial acetabular grafts, and five femoral head grafts were used) in combination with an extensile exposure and anterior and posterior pelvic plates19. Importantly, he reported healing of the graft and the pelvic discontinuity in all hips, attesting to the mechanical advantages of rigid plate fixation and use of a single massive structural bone graft. In Stiehl's series, three hips in which a socket had been inserted with cement into a whole acetabular allograft were stable. Two patients in whom a socket had been inserted without cement and placed largely on bone graft needed a reoperation because of acetabular loosening, two had a deep infection, and six had dislocation. Thus, the mechanical advantages of use of double pelvic plates and massive bone grafts must be weighed against the complications, such as infection and dislocation, that are associated with such extensive operations and against the long-term problems, such as fracture or collapse, that are associated with unprotected massive structural pelvic bone grafts7.
We believe that the guiding principles for the treatment of pelvic discontinuity should be (1) identification of the problem, (2) stabilization or effective bypass of the discontinuity, (3) bone-grafting at the site of the discontinuity, (4) treatment of any associated bone loss, and (5) placement of a stable acetabular implant. On the preoperative radiographs, a visible fracture line and a break in Kohler's line should raise the suspicion of pelvic discontinuity. Judet radiographs may be helpful for confirmation, although implants or bone cement may preclude visualization of the acetabular columns. Intraoperatively, a pelvic discontinuity can be difficult to identify: the fracture line may pass through areas of bone loss, may be filled with fibrous tissue, and may not be very mobile. Stressing the inferior aspect of the pelvis in an anterior-posterior direction may demonstrate motion at the site of the discontinuity. The reconstruction should be individualized on the basis of the severity of the bone loss. If satisfactory bone is present to provide mechanical stability and biological fixation, we stabilize the pelvis with a plate or plates and then insert a porous-coated socket without cement and with screw fixation, a technique applicable to most hips with type-IVa lesions and some with type-IVb lesions. When large bone defects (including many type-IVb defects) or poor bone biological activity (type-IVc defects) precludes insertion of a porous-coated socket without cement, we prefer to use an antiprotrusio cage (in combination with a single piece of structural bone graft when needed) fixed to the ilium with screws and to the ischium with screws or with the inferior flange of the cage alone. The use of autogenous bone graft at the site of the discontinuity also may be beneficial.
For the successful treatment of pelvic discontinuity, the surgeon must obtain initial stability of the socket, establish conditions for long-term stability of the socket, stabilize the pelvic discontinuity, and establish conditions favorable for healing. It is not always possible to achieve all of these goals; when it is not possible, resection arthroplasty (as was performed in two patients in the current report) or two-stage reconstruction—first with pelvic stabilization and bone-grafting and later with acetabular reconstruction—may be considered6.
Serious complications related to these large-scale, difficult reconstructions included recurrent dislocations, injuries of the sciatic nerve, and deep infections. Four of the twenty-nine revised hips in the current study needed a second revision because of recurrent dislocation; thus, these patients should be considered at high risk for instability of the hip. A partial sciatic-nerve palsy developed in three of the twenty-nine hips. The sciatic nerve is at risk during the extensive dissection that is needed to stabilize the discontinuity and to treat associated bone defects. There also may be scar formation around the nerve from multiple previous operations, increasing the risk of nerve injury. Special efforts may be warranted to protect the sciatic nerve with methods such as intraoperative nerve-monitoring.
In summary, the current study provides information on the risk factors associated with the development of pelvic discontinuity and on the identification of the radiographic features of the condition. A classification scheme was also presented. We found a relationship between the results of treatment and the severity of associated pelvic bone loss and previous treatment with irradiation. The rate of complications may be high, but with improved treatment techniques and a better understanding of this difficult problem a successful reconstruction is possible for many patients.