Adult patients in the NU Hospital Group diagnosed with a PJI between 1 January 2012 and 31 December 2018 were included in the study. Information from three official database registers were collected. First the operative record's database of performed surgical interventions as well as the in-patient database in the NU Hospital Group were both searched for the ICD-10 code “T84.5”, indicating a deep periprosthetic joint infection. All discovered procedures were then controlled against our surgical data reported to and collected in the Swedish National Hip Registry. Our primary THAs between 2012 and 2014 were additionally matched with the Swedish Prescription Drug register in a follow-up study of infection registration based on the National Registry (Lindgren et al., 2014). The patient files for all registered infections were then collected and scrutinized by the authors to both confirm the diagnosis and match it according to the guidelines published by Musculoskeletal Infection Society (Parvizi et al., 2011) and the Infectious Diseases Society of America (Osmon et al., 2013) to ensure conformity. Patients with semi-total arthroplasties without an acetabular component were excluded, since these implants were only used in treatment of cervical hip fractures in old and fragile patients. As previously described these patients have considerable comorbidity, are often institutionalized, and both treatment and follow-up is unreliable (Guren et al., 2017).

A PJI was considered present with the major criterium of two or more positive cultures with growth of the same microbiological organism in joint fluid or collected tissue samples. Additionally, at least two of the following criteria were present: wound discharge, sinuses, fever, local inflammation signs, raised CRP (> 10), or clinical findings of joint distress and particularly load pain.

Routinely, five tissue cultures were collected at implant exchange surgery from the interface between both the cup and the stem implant as well as cancellous bone when available, or otherwise including at least five samples from the joint capsule. The samples were harvested before perioperative intravenous antibiotic treatment was administrated. Antibiotic treatment was given according to established protocols (Osmon et al., 2013; Zimmerli et al., 2004) and national guidelines (Tevell et al., 2019) depending on antibiotic susceptibility and started with intravenous antibiotic therapy that was continued for 7–10 d followed by oral medication that was scheduled to terminate after 3 months.

The laboratory handling of microbiological tissues was changed in early 2015. Before, 10 samples were collected in broth and transported to the laboratory, where the material was plated on agar plates. From 2015, at least five tissue samples are routinely harvested at surgery and collected in empty sterile test tubes. Any further treatment of samples is performed in the laboratory. Transportation time is thereby considered less crucial, and the laboratory cultivates a fresh tissue section with imprint on agar substrate primarily. Afterwards, residual tissue pieces are continuously chopped for prolonged enrichment, which is routinely performed in all samples labeled “prosthesis”. This reduces the risk of rapidly and strongly growing species taking over and hiding slow-growing pathogens such as Cutibacterium acnes.

An infection was classified as either postoperative or hematogenous. Cases with a sudden onset of inflammatory symptoms well beyond the postoperative period with a previously well-functioning arthroplasty were considered hematogenous (here ranging from 5 months to 25 years; median 6 years). In postoperative infections, we have refrained from distinguishing between acute and chronic PJI in accordance with the recommendation at the Second International Consensus Meeting on Musculoskeletal Infections (Elkins et al., 2019). In most of our postoperative PJIs there had been wound complications and a superficial site infection following the insertion of the implant without a continuous period with an optimally functioning joint. However, there were also infections with a probable origin in the prosthetic procedure with well healed surgical wounds and no septic episodes but a rehabilitation phase with continuous discomfort and loading pain. These infections were first evident after repeated visits, laboratory investigations, and finally progressive radiographic changes (Fig. 1).

A PJI was considered eradicated when there were no inflammatory signs including normalized ESR and/or CRP or symptoms from the hip joint for a minimum of 2 years after termination of antibiotic treatment, which is considered the shortest time in published literature to establish infection eradication (Sendi et al., 2017). Unrelated death (n=2) or hip revisions for other reasons than infection with negative perioperative cultures (n=3) occurring during follow-up within 2 years were labeled eradicated. Final failure was defined as an ongoing PJI with no planned future surgery but suppressive life-long antibiotic treatment. Girdlestone resection arthroplasties were reported separately since they were successful regarding infection eradication, but not from a prosthetic point of view.

In all primary surgery from 2016, the patient arrived at the hospital on the day of the operation. Preoperative cleansing at home with sponges containing chlorhexidine gluconate (CHG) were advocated twice on the day before surgery as well as in the morning before leaving home. Prior to surgery the affected leg and groin was cleaned once more with CHG in the theater and then with chlorhexidine alcohol 50 % before draping. Between 15 and 30 min before skin incision, 2 g of flucloxacillin was administered intravenously followed by two additional doses 2 and 6 h after the first dose according to Swedish tradition and guidelines (Tevell et al., 2019). In 5 to 10 % of the patients another antibiotic was chosen due to allergy. The theater had laminar air flow and all persons in the room were dressed in reusable non-permeable polyester coveralls including helmets. The surgical staff used an additional reusable impenetrable gown and two pair of gloves.

All patients with at least two culture-positive tissue samples and/or joint aspiration with growth of the same microorganism were included in the study (Fig. 2). Four patients undergoing an outdated treatment with only drainage and limited irrigation due to high age and poor health status were excluded. In accordance with inclusion criteria, four patients who were diagnosed with a suspect joint infection by only one joint aspiration culture or a wound cultivation were excluded as well as one patient with a suspected but culture-negative PJI. The remaining 89 PJIs were divided in three groups according to surgical treatment (Fig. 2).

Debridement, antibiotics, and implant retention in accordance with published DAIR protocols (Tsang et al., 2017). The joint was dislocated, and aggressive surgical debridement was performed by experienced revision surgeons followed by irrigation with 1–3 L of sterile sodium chloride before exchangeable modular heads and liners were changed. Resorbable, equine collagen fleeces containing between 70 and 280 mg of gentamicin were implanted in the joint before wound closure. The size of the fleece was usually not listed in the operative records. The procedure was planned to be performed within 6 weeks postoperatively but sometimes occurred later. Time from onset of symptoms was usually not specified in files since initial surgical wound care and agraffe removal was performed by district nurses and general practitioners. Hematogenous PJI was treated with DAIR within 2 weeks of presumed acute onset of symptoms.

Descriptive statistics were used to summarize demographic data, time, antibiotic treatment, and involved microorganisms. Calculations of group comparisons using 2×2 contingency tables were performed by Fisher's exact test. Non-parametric group comparisons used the Mann–Whitney U test. A p value of < 0.05 was considered statistically significant.

A total of 87 patients with 89 PJI episodes were included in the study. One patient with bilateral arthroplasties had a simultaneous bacterial precipitation in both hips and bilateral DAIR treatment, while another was initially successfully treated but infected with a new pathogen after further revision surgery. A total of 18 primary arthroplasties had been performed in other hospitals at a median of 2 years (23 d–24 years) before the patients attended care at our institution. Median age at the time of diagnosis was 68 years (40–92); 31 of the patients were female (34 %), and 12 patients died from causes not related to the PJI at an average of 59 months (16–90) after revision surgery and excision arthroplasties. No deaths directly attributable to the PJI were identified.

Distribution of bacteria in collected tissue samples is presented in Table 2.

Coagulase negative staphylococci (CoNS), Staphylococcus aureus, and Cutibacterium acnes were the three most prevalent bacteria and were detected in 30 %, 20 %, and 16 % of the cultures, respectively. Multiple organisms were found in 24 % of the cases while 76 % of the infections were monomicrobial with no difference in failure rate (2 out of 19 vs. 10 out of 58; p=0.7). Final surgical failure resulting in life-long antibiotic suppression (n=6) or Girdlestone (n=6) was highly associated with Staphylococcus aureus infection (10 out of 12; 83 %; p=0.0001) (Table 2).

Cutibacterium acnes was isolated in 18 of 89 PJIs (20 %), in 14 out of 18 cases (78 %) as the single growing pathogen. A total of 15 cases (83 %) were diagnosed after 2015. Patient characteristics in Cutibacterium infections compared with other infectious agents are presented in Table 3. DAIR (n=11) or revision implant exchange (n=7) was successful in all cases. Cutibacterium acnes infections were seen in 15 out of 18 of the uncemented implants (83 %) as compared to 30 out of 71 (42 %) of all other PJIs (P=0.003) (Table 3).

The flow chart for the patients in the study is summarized in Fig. 2. Treatment with DAIR was performed in 53 out of 89 PJI episodes (60 %). Failure was noted in 7 (13 %) of these procedures. Two chronic failures ended up with retained prostheses and lifelong antibiotic suppression; three of the patients suffered extended implant exchanges (revisions) and two Girdlestones which all resulted in infection eradication. Two patients with wound complications after DAIR were treated with vacuum-assisted closure (VAC) and received antibiotics for an additional 3 and 8 months, respectively. There was no sign of infection recurrence 2 and 3 years after antibiotic termination.

A total of 36 primary revision implant exchanges were performed, resulting in 28 out of 36 (78 %) one-stage revisions and 8 out of 36 (22 %) two-stage revisions. The outcome of all implant exchange surgery is presented in Table 4. Peri- and postoperative antibiotic treatment was given in accordance with preoperative aspiration and wound cultures when available. Follow-up peroral antibiotics were given following the results of tissue cultures and in collaboration with an infectious diseases specialist. Changes of antibiotic therapy were made in approximately 5 % of patients according to allergy or other toxic effects.

A total of 34 infections (38 %) were classified as hematogenous while 55 (62 %) were postoperative. The postoperative infections also included three failed DAIR treatments (5 %) which were later successfully revised with infection eradication. A statistically inferior final outcome of hematogenously infected patients was not proven (8 out of 34 vs. 4 out of 55; p=0.052).

The time of occurrence of the postoperative infections is presented in Table 5. In patients with revision surgery (DAIR or implant exchange, excluding Girdlestones) performed within 30 d after THA, infection eradication was seen in 88 % (21 out of 24) as compared to 95 % (20 out of 21) in patients revised between 31 and 180 d after primary surgery. Among patients with a later time point for revision (> 180 d, n=12) three patients had previously undergone a DAIR. The causative pathogens in treatment failures of postoperative infections within 180 d (n=4) including one primary Girdlestone were monobacterial CoNS in two cases and Staphylococcus aureus and Enterococci in one case each respectively.