The study identified a consecutive series of patients who underwent acetabular revision using trabecular metal augments and the bone impaction grafting technique during the period from May 2011 to May 2014. This time period was chosen to obtain a minimum follow-up of 5 years. During the study period, seven patients were operated on with this technique, but only five were included in the study, all of whom were under 50 years of age at the time of the intervention. The two excluded patients were 64 and 60 years old at the time of surgery.
All patients were operated on by the author of the study, and none of them were lost to follow-up. The mean age of the patients was 46.8 years (45–50) at the time of the intervention, with a mean follow-up of 79 months (60–101). There were three women and two men. The initial diagnosis in two patients was osteoarthritis secondary to hip dysplasia, and in the other three, avascular necrosis of the femoral head (Table 1). In the two patients with a history of hip dysplasia, reconstruction was performed in two stages because both had septic acetabular loosening (Fig. 1). The only comorbidity was rheumatoid arthritis in a 47-year-old male (case 3) under treatment with biologics.
Bone defects were classified according to the Paprosky classification  because it is the most widely used in the literature and the Saleh classification  because it is a simple, reliable, valid system for the classification of bone defects .
Patients were operated on in the lateral decubitus position using a posterolateral approach to the hip. In all cases, infection was ruled out before surgery through analytical studies. Arthrocentesis was also performed before capsulotomy for collection of a joint fluid sample. At this time, antibiotic prophylaxis was initiated according to our hospital protocol. Samples were taken from the acetabular fundus for both anatomopathological and microbiological studies. Once the pathologist reported the absence or ≤ 5% per field of polymorphonuclear leukocytes, the acetabular cavity reconstruction was started. Debridement of fibrous tissue was initiated by milling the acetabular remnant, and the bone defect was identified, with special attention to the posterior column, acetabular fundus and posterosuperior area. When the defect affected almost the entire posterior wall, we decided to perform another reconstruction method with acetabular augments and components made of trabecular metal, associated or not with a cage. In these cases, it was difficult to place the screws in the posterior column that supported the mesh in the posterosuperior portion to reconstruct the defect and achieve rigid fixation, and this is one of the most important aspects of the bone impaction grafting surgical technique. If the defect was in the fundus of the acetabulum (central defects), we placed a mesh in it (central mesh) that was affixed to the pelvis with at least three 3.5-mm screws. The ischiopubic notch (teardrop), when present, was used as a reference to reconstruct the centre of rotation to its anatomical position. In case of a defect in this area, we used the ischium as a reference for reconstruction. A trial acetabular reamer was placed in the anatomical position, and we again checked the defect in the posterosuperior zone. At this time, the trial TM augments were sized in the defect to allow sufficient space for impaction to take place. When we found the correct size, a trial reamer was placed in anatomical position and held to offer inferior support to the augment to avoid displacement during the fixation of the augment to the iliac bone with al least two 6.5-mm screws. Next, we placed the mesh in position to convert the uncontained defect into a contained defect. In these large defects, it was essential to affix the first screw to the ischial area and then the rest of the screws to the anterior column. Correct fixation of the anterior and posterior corners is essential. We checked to ensure rigid fixation of the mesh to the iliac bone.
Bone chips were prepared (fresh-frozen bone chips) by hand with a gouge clamp, obtaining sizes of 7–10 mm3 to provide graft stability. We pressure-washed the graft with a spray gun, and the allograft was then packed into the defects and into the TM spaces. Further layers of graft were impacted using impactors of different sizes. We started with the smallest impactor and continued with impactors of greater diameter, reconstructing the socket to the desired position. The last impactor used had a diameter 4 mm greater than the outer diameter of the polyethylene component to allow for an adequate cement mantle. At this moment, the bed of impacted graft should feel like a cortical bone. An all-polyethylene flanged cup was inserted into viscous cement, held in position, and held with pressure until the cement had polymerised. In one case (case 2) we used a Trident® constrained acetabular insert because we used a structural allograft to reconstruct the femoral side. This component was cemented in a position with lesser abduction angle.
Clinical outcome was assessed according to preoperative questionnaires and the last follow-up. Hip status was assessed using the Western Ontario and McMaster Universities (WOMAC) questionnaire . Its usefulness comes from its ability to assess clinical changes patients have perceived in their state of health, and it has been validated in Spanish . The numeric rating scale is scored from 0 (best) to 10 (worst). Therefore, the maximal aggregate score for pain, stiffness, and function is 50, 20, and 170, respectively. Generic health was assessed using the SF-36, with its eight scales ranging from 0 to 100 (100 being best). The SF-36 also includes a transition item that asks about the change in general health status from the previous year. Lastly, every patient was asked whether they would undergo the operation again.
Radiological analysis was based on an anteroposterior (AP) radiograph of the pelvis and an axial projection taken preoperatively, immediately postoperatively, and at the last follow-up examination. Preoperative images gave an idea of the acetabular defect, but in all cases the defect was classified according to what was seen during surgery. To determine whether there was migration of the acetabular component, we established references on AP radiographs of the pelvis, following the study by Borland et al.. A line was marked that joined the two acetabular teardrops (x-axis), and another line was perpendicular to the first lateral to the teardrop (y-axis). The horizontal and vertical distances to the polyethylene cup were measured at its lowest and medial points. We also measured the abduction angle of the polyethylene cup relative to the x-axis. In the radiological controls of the last follow-up examination, the presence of radiolucent lines around the trabecular metal augments as well as the incorporation of the graft were evaluated  (Fig. 2), although the latter was complicated by the placement of the mesh.
Radiographic loosening was defined by a change in the abduction angle of more than 10° or a change in the vertical or horizontal position of the acetabular component of more than 5 mm.
Data were analysed in SPSS version 18 for Windows. The data were grouped according to frequency and to measures of central tendency and dispersion: percentages, means, and standard deviation. To compare the variables under study with respect to time (before and after surgery), the nonparametric Wilcoxon test of ranks and signs for two related samples was used. Significance was accepted when the probability p associated with the test statistic was < 0.05.
For the WOMAC scale, the results were considered clinically relevant when the difference in the before–after evaluation exceeded predetermined values for each considered sphere, taking as reference 50% of the initial value .