This is an observational, single-center study which is based on prospectively collected data of a consecutive series of patients.
Implants and patients
The Virtec straight stem (Zimmer, Winterthur, Switzerland) is a variant of the Müller straight stem. The stem is double-tapered, has an oval cross section and is made of CoNiCrMo . The design rationale was, analogous to the Kerboull stem , to provide better filling of the proximal femur when compared to the original Müller straight stem . Various sizes in standard and lateral versions are available. In the first period of this study, the PF stem (Cedior, Montbéliard, France) with the same shape, however, manufactured out of stainless steel, was used. As results did not differ between the two implants, no differentiation was made for further evaluation (data not shown).
All 172 patients (178 hips) who underwent revision with the PF or Virtec straight stem between 01/1994 and 08/2006 were included. Ethics committee approval was obtained prior to study commencement. The mean age (standard deviation) of the patients at the time of the surgery was 68.4 (9.3) years (range 36–90 years); 126 (71%) cases were male. Indication for revision was aseptic loosening in 137 cases (77%), periprosthetic joint infection (PJI) in 32 cases (18%), malposition of the stem in 6 cases (3.4%), and a broken implant in 3 cases (1.7%). The type of explanted stems are listed and femoral defects prior to replantation classified according to Della Valle and Paprosky  shown in Table 1. 116 cases received their primary implantation in our hospital, 62 cases were referred to us from another hospital. A total of 19 cases had at least one previous revision.
A revision of the acetabular component was performed in 82 out of 178 cases. In 61 cases, a cemented Mueller acetabular reinforcement ring was implanted, in 11 cases, a non-cemented SL-cup, and in 10 cases, a Burch-Schneider anti-protrusio cage (all Zimmer, Winterthur, Switzerland).
Every revision surgery was standardized using a lateral transgluteal approach with the patient in a supine position. After stem removal, the medullary canal was cleaned using drills and chisels to remove all cement remnants; granulation tissue was completely removed using specially designed curettes (Fig. 1). Thereafter, the medullary canal was repeatedly rinsed with a 0.2% polyhexanide solution (Lavasept®, B. Braun, Melsungen, Germany) using a bulb syringe. Subsequently, the largest implant fitting into the medullary canal was implanted, aiming for a primary press-fit fixation in the anterior–posterior (ap) plane (Table 2). Implants were cemented line-to-line with the final broach . All stems where implanted with a second-generation cementing technique (no vacuum mixing, distal plug, retrograde filling) using high-viscosity Palacos R + G (Heraeus, Hanau, Germany) cement . In case of septic loosening, patients were treated according to our established algorithm either with a 1-stage (14 cases) or 2-stage revision (18 cases) . No additional antibiotics were added to the cement in any cases. The postoperative mobilization was the same for all patients with initial full weight bearing as tolerated starting the first postoperative day.
Follow-up examinations were scheduled at 4 months, 1 year, 2 years, and 5 years, and all 5 years thereafter. Patients were rated as lost to follow-up when last contact (outpatient clinic or telephone call) was 5 years overdue. For radiological analysis, the first postoperative radiograph and the most recent radiograph of all unrevised patients with a minimum radiological follow-up of > 10 years or in case of a revision, the last radiograph prior to revision, were analyzed. Standardized AP radiographs were taken centered on the symphysis, showing the entire implant. Findings were stratified according to the Gruen zone system . Osteolysis was defined as any newly developed progressive endosteal bone loss at the cement–bone interface with a diameter > 3 mm and categorized as either scallop- or bead-shaped . Axial subsidence of the stem was measured as an increase of any radiolucency in the proximal cement in Gruen zone 1, created due to distal migration of the shoulder of the prosthesis. It was considered relevant if it was more than 2 mm . Debonding was defined as present, if a radiolucent line at the prosthesis–cement interface, not visible on the first postoperative radiograph, was observed . The stem was rated as being radiologically loose if circumferential radiolucency in all Gruen zones  and/or excessive subsidence of more than 10 mm  was present. The radiological analysis was carried out by one of the first authors at the end of the study; ambiguous findings were discussed with the senior investigator and agreed upon.
A survival analysis with death as a competing risk was performed with various endpoints: (1) aseptic loosening of the stem, (2) worst-case scenario with all cases lost to follow-up judged as aseptic loosening, and (3) re-revision of the stem and/or cup (including exchange of the liner) for any reason. Patients without any re-revision were censored at the date of last contact. For each endpoint, cumulative incidence functions were used. Furthermore, time to aseptic loosening of the stem was analyzed using an absolute risk regression model with death as a competing risk. The included variables were age, sex, preoperative defect, stem offset, and stem size.
A sensitivity analysis was performed using only the first hip in patients with bilateral stem revision. Since it resulted in similar estimates, no joint frailty model was estimated. Data of six patients with bilateral stem revisions were analyzed as independent.
A part of the results from a previously published study using the Virtec straight stem in primary THA (same stem, same bone cement) were compared to the results from the current study . Cumulative incidence for aseptic loosening of the stem was compared according to Gray .
Continuous data were presented as mean (standard deviation) with range. Categorical data were presented as frequencies (percentages). The analysis was performed using a significance level of α = 0.05. R statistical package was used for all analyses.