Adult patients with chronic hip infections, who had each received a staging complete hip arthroplasty between January 2010 and January 2016 in the First Affiliated Hospital of Fujian Medical University, were retrospectively analyzed. These patients were initially diagnosed according to their disease history, clinical manifestations (local erythema, tenderness, exudation and pain) and imaging performances like disappearance of joint space and femoral head damage (Fig. 1A). All patients underwent a preoperative joint cavity puncture, as well as an intraoperative joint fluid and tissue culture. The inclusion criteria were as follows: i) A chronic sinus connected to the joint cavity; ii) the presence of pus in the joint puncture or pus and destruction of femoral head during the surgery; iii) increased (>20 mg/l) serum C-reactive protein (CRP) concentrations (detected using a Modular automated protein analyzer; IMMAGE800; Beckman Coulter, Inc., Brea, CA, USA) and positive frozen sections during the stage I surgery (numbers of white blood cells per high-power field >5); and iv) cultures of the joint punctures or intraoperative specimens yielding positive results. A positive result was considered when at least one culture yielded a strict pathogen (Staphylococcus aureus, Pseudomonas aeruginosa, Enterobacteriaceae and anaerobes) or when two cultures yielded a strain that was a skin commensal (including coagulase-negative staphylococci or Propionibacterium acnes) (15–17). The exclusion criteria were as follows: i) Patients with infections after a THA and ii) patients with incomplete clinical and laboratory information. Additionally, the following data of the patients were preoperatively collected: Cause of disease, causative organisms, comorbidities, CRP and erythrocyte sedimentation rate (ESR; detected using an Automatic ESR analyzer; Monitor-100; Vital, Puteaux, France).

All surgeries were performed using the same approaches and procedures as described below and by the same experienced surgeon. The stage I surgery was performed as follows: An incision was made at the posterolateral region of the hip joint. The purulent fluid was sucked out and sent for bacterial culture and drug sensitivity examinations. Subsequently, the articular capsule was cut open to fully expose the hip joint (the internal fixture was taken out, if present) and was followed by the removal of the destroyed head and neck of the femur. Subsequently, the proximal femur was unreamed to prevent inflammation from the reaming to the medullary cavity. The purulent pus, inflammatory tissue, detached articular surface, and scar tissue were removed. For the Girdlestone alone group, the incision was closed after the drainage tube was placed (Fig. 1B and C). For the cement spacer group, after debridement, 8 g vancomycin (Lilly, Indianapolis, IN, USA) was mixed with 80 g of bone cement (Refobacin; Zimmer Biomet, Warsaw, IN, USA), which was used to wrap two 4 mm Kirschner wire (Fig. 2B) or cement prostheses (Fig. 3) composed of a 28 mm femoral head, a 32 mm femoral neck and a 105 mm femoral stem (CM-CZ; AKMedical, Beijing, China) to make a spacer that matched the acetabular and femoral medullary cavities. The spacer was then implanted into the hip joint, which could help to maintain the normal eccentricity and tension of the gluteus medius. Finally, the incision was sutured after the drainage tube was fixed. After the surgery, the drainage tube was retained until a clear drainage fluid was observed and the drainage fluid volume was >20 ml in both groups. The selection of antibiotics was based on the preoperative or intraoperative cultures and the drug sensitivity results, while vancomycin was applied in patients without culture results. The courses of intravenous and oral administration of antibiotics depended on the clinical symptoms as well as the CRP decline. After the patients received intravenous antibiotics (Table II) for 2–4 weeks, they were switched to oral antibiotics, which included 0.6 g levofloxacin (Daiichi Sankyo Co., Ltd., Tokyo, Japan) once per day and 0.3 g rifampicin (Chengdu Tiantaishan Pharmaceutical Co., Ltd., Chengdu, China) twice per day for 1–2 months. The affected limbs were allowed to gradually perform flexion and extension activities, as well as partial loading.

When movement of the hip did not incur obvious pain and the soft tissue did not appear to be swollen, patients underwent a CRP and ESR review a total of three times, with each review occurring once every week and yielding normal results. The stage II surgery was conducted using the original approach. The antibiotic bone cement spacer was removed, followed by an intra-articular scar clearance and debridement. The patients in both groups underwent THA using bioprosthesis (Johnson & Johnson Services, Inc., New Brunswick, NJ, USA).

Patients were assessed in terms of Harris hip scores (18) before the stage I surgery, between the two surgeries, and within the follow-up period (12.0–62.0 months; mean duration, 24.2±3.8 months) after the last surgery. The complications after the stage I surgery, the time of antibiotic applications after the stage I surgery, the interval between the stage I and stage II surgeries, the operation times, and the operative hemorrhage of the stage II surgery were recorded and the lengths of the lower extremities were postoperatively compared. In addition, the patients were reviewed using the inflammatory indicators (ESR and CRP) once every 3 months within the follow-up period, in order to monitor the recurrences of disease.

All of the data are presented as the mean ± standard error of the mean. A Mann-Whitney U test was performed to evaluate the differences between the Girdlestone alone group and the Girdlestone + cement spacer group for all of the indicators analyzed. The statistical significance threshold was set at P=0.05.

According to the inclusion criteria, 16 patients were eligible for the present study. Among these patients, three patients with incomplete clinical data were excluded. Therefore, a total of 13 patients (14 hip joints, with one case of a bilateral infection) were enrolled in the study, including 8 men and 5 women, mean age 59.3±4.3 years; range 19–79 years. The present study included 7 cases (8 hip joints) of primary hip infections, 5 cases of infections after the internal fixation of the hip fracture (4 cases were femoral neck fractures and one case was an acetabular fracture), and one case of a chronic infection after the focal cleaning of osteonecrosis of the femoral head. All of the patients complained of symptoms for >2 months (Table I). Ten patients had pain and swelling at the affected site, and two patients had pain symptoms and a sinus connected with the articular cavity. The symptoms of these patients lasted for 60–125 days (mean: 82.7±7.5 days) before the first surgery. Nine patients had a history of antibiotic use before the first surgery, while certain patients underwent incision drainages (cases 1, 3, 5, and 7) with no improvement in clinical manifestations (Table I). Beside the cases that were negative for the presence of bacterial cultures, half of the bacterial cultures of the two groups were comprised of gram-negative bacteria, which were different from the commonly found staphylococci and other positive cocci in PJI cases. On the basis of the bacterial culture results, the patients in both groups were given antibiotics, out of which two cases of negative cultures were given a combination of vancomycin and quinolones (levofloxacin; cases 7 and 10; Table II).

For the stage I surgery, 4 patients (5 hips) underwent the resection of the head and neck (cases 1–4, hereafter referred to as the Girdlestone alone group), and 9 patients (9 hips) underwent the resection of the head and neck combined with the implantation of a bone cement spacer (cases 5–13, hereafter referred to as the spacer group). In the Girdlestone alone group, the highly pathogenic Burkholderia cepacia was cultured in one case (case 1, Fig. 1), which did not show significant complications after the stage I surgery but required a longer time duration for controlling the infection. One case had an incision sinus and infection recurrence after the surgery (case 3, Escherichia coli) and required a re-debridement and an extended anti-infective treatment. Another case had pneumonia, poor wound healing and other complications, and underwent a re-debridement (case 2, Stenotrophomonas maltophilia). The infection and complications in these three cases were under control and CRP levels were restored to a normal level after the stage I surgery in a mean of 8.4±1.5 weeks (range, 5–12 weeks). Among the patients undergoing a spacer implantation, two cases had complications. One case had a pneumonia and type I respiratory failure after the stage I surgery and was healed after treatment and underwent the stage II surgery successfully (case 6 with Corynebacterium). Another case had a spacer fracture after walking with weight loading at 3 months after the surgery and underwent the stage II surgery soon after (case 7). The mean time for CRP levels of the patients to decline to the normal level after the stage I surgery was 5.8±1.0 weeks (range 4–12 weeks), which was a shorter time period than that of the Girdlestone alone group. However, this difference was statistically insignificant. Additionally, two cases of the Girdlestone alone group who underwent a debridement after the stage I surgery suffered from poor wound healing again after the stage II surgery (cases 2 and 3) and were healed after receiving a debridement. However, all the patients in the spacer group were well healed. Signs of infection recurrence were not found in both groups until the end of the follow-up periods (Table II).

After the stage I surgery, the patients in the Girdlestone alone and spacer groups were given antibiotics for 120±3.2 weeks and 8.0±0.8 weeks, respectively, but this difference in antibiotic treatment times was statistically insignificant. The mean time interval between the two surgeries was 4.0±0.4 months in the spacer group, which was a significantly shorter time interval than that in the Girdlestone alone group (29.5±10.7 months, P<0.01, Table II). The stage II arthroplasty was more difficult to perform in the Girdlestone alone group. The mean operation time was longer (124.0±13.6 vs. 105.6±7.9 min) and the incidence of a hemorrhage was increased more in the Girdlestone alone group (501.0±100.8 vs. 322.2±55.4 ml), but these differences were statistically insignificant. The Girdlestone alone group showed a more marked leg-length discrepancy after the stage II THA than the spacer group (2.8±0.6 vs. 0.8±0.2 cm, P<0.01; Table III).

All patients were followed up for >12 months after the stage II surgery (12.0–62.0 months), with a mean of 24.2±3.8 months (Girdlestone alone group: 31.3±10.8 months; spacer group: 21.1±2.9 months). No cases had an infection recurrence or emerging infection surrounding the prosthesis, and no cases had prosthesis loosening, a dislocation, or a fracture around the prosthesis during the follow-up periods. The mean Harris scores before the surgery were comparable between the two groups (47.8±2.5 vs. 40.7±3.1; Table IV). The Harris scores after the stage I replacement were 70.4±2.1 vs. 85.2±2.2 in the Girdlestone alone group and spacer group, respectively, which showed significant differences (P<0.01). At the last follow-up, the Harris score in the spacer group was also improved than that in the Girdlestone alone group (81.6±1.1 vs. 88.9±1.7, P<0.01; Table IV).