Session 1: Hip / Knee

Onsdag den 21. oktober
09:00 – 10:30
Lokale: Reykjavik
Chairmen: Per Kjærsgaard / Thomas Jakobsen

1. Risk factors for early periprosthetic fractures following primary cementless THA
Ann Buhl Bersang, Anders Troelsen, Kalemose Thomas , Husted Henrik, Gromov Kirill
Orthopaedic department, Hvidovre Hospital

Background: The use of cementless fixation for primary THA has increased world wide, which can be considered paradoxal as registry data suggest inferior survival of cementless THA. Periprosthetic fracture is one of the main complications following cementless THA.
Purpose / Aim of Study: In this retrospective cohort study we 1) investigate risk and 2) identify pre-and intraoperative risk factors for early periprosthetic fracture following primary cementless THA.
Materials and Methods: 1,255 unselected and consecutive patients undergoing primary THA between July 2010 and May 2014 were identified. Patients with missing data or patients operated with other implants than a standard cementless THA were excluded, leaving 1,183 patients for analysis. Patient demographics were recorded. Level of preoperative arthritis, Dorr type and cortical index were recorded on preoperative xrays. Varus/Valgus placement of the stem and canal fill were recorded postoperatively. Early periprosthetic fractures (< 1 year) were identified. Regression analysis was performed to identify risk factors for early periprosthetic fracture.
Findings / Results: 37 (3.1%) early periprosthetic fractures were identified. 28 (76%) occurred in females. Mean age was 72.7 years. Median time until fracture was 16 days (1-102). 10 (37%) of all fractures occurred in patients >80 years old. 8 fractures occurred in Dorr type C femurs (21.6%). Patients with Dorr type C femurs had a 7.5 OR (95%CI (1.6-35.0)) for early periprosthetic fractures. Varus/Valgus placement of the stem (1.1 OR (95%CI(0.9-1.2)) and canal fill (1.1 OR (95%CI(1.0-1.2)) were not identified as predictors of early periprosthetic fracture.
Conclusions: Dorr type C is an independent risk factor for early periprosthetic fracture. Surgeons should be aware of this and considerer cemented arthroplasty in those patients.

2. Is patient-reported outcome after total hip arthroplasty influenced by type of bearings?
Claus Varnum, Alma B. Pedersen, Per Kjærsgaard-Andersen, Søren Overgaard
Department of orthopaedic surgery, Vejle Hospital; Department of clinical epidemiology, Aarhus University Hospital; Department of orthopedic surgery and traumatology, Odense University Hospital

Background: The outcome of total hip arthroplasty (THA) has traditionally been assessed in prosthetic survivorship and complications. Patient- reported outcome (PRO) is recognized as a very important tool for evaluating the outcome after THA.
Purpose / Aim of Study: We aimed to compare PRO scores from patients having ceramic-on-ceramic (CoC) and metal-on-metal (MoM) to scores from patients with metal-on-polyethylene (MoP) THA.
Materials and Methods: We identified 4,212 patients matched on sex, age, and year of surgery operated during 2002 to 2009 from the Danish Hip Arthroplasty Registry with no revision. 450 patients were excluded due to death, protection against inquiry from researchers, unknown address, hip resurfacing arthroplasty, and dual mobility cup. In total, 3,762 patients received the following PROs: The hip disability and osteoarthritis outcome score (HOOS), EQ-5D, and UCLA activity score. Of these, further 145 patients were excluded due to confirmed revision surgery. 3,082 patients responded (response rate 85%). Univeriate linear regression was used to compare mean age between the bearings groups. Multivariate linear regression was used to compare mean values of PRO subscale scores within the three groups.
Findings / Results: Among responders, 1,393 (45%) had CoC, 512 (17%) MoM, and 1,177 (38%) MoP THA. There was a similar distribution of sex within the three bearing groups. Patients with CoC and MoM bearings were significantly younger (mean age 65.7 and 65.5 years, respectively) than patients with MoP bearings (mean age 67.4 years). No significant difference in mean scores was found in the 5 HOOS subscales, EQ-5D index, EQ VAS, or UCLA activity score between patients with CoC, MoM, and MoP THA.
Conclusions: We found no association of CoC, MoM, and MoP bearings on PRO after THA.

3. Promising migration Pattern in a 2 year RSA follow-up of the short Primoris Femoral Stem
Janus Duus Christiansen, Michael Ulrich Jensen, Ashir Ejaz, Poul Torben Nielsen, Mogens Berg Laursen
Northern Ortopaedic Division, Aalborg University Hospital

Background: In order to save proximal bone stock in primary total hip arthroplasty (THA), short femoral stems are introduced.
Purpose / Aim of Study: Designed for perfect fit within the femoral neck, the Primoris® femoral stem has been released for clinical studies in coherence with stepwise introduction of new implants. This is the preliminary report of the Radio Stereometric Analysis (RSA) results after 2 year follow-up (FU).
Materials and Methods: We carried out a prospective cohort study of 52 patients scheduled for surgery with the femoral neck-preserving Primoris® stem. Migration was analyzed by RSA, and clinical results recorded using the Harris hip score, UCLA activity score, WOMAC, EQ5D health questionnaire and Oxford Hip scores.
Findings / Results: 2 patients were excluded intra-operatively and 1 patient was revised due to aseptic loosening 3 months postoperatively. 4 patients were excluded from the RSA analysis due to technical issues, leaving 44 patients for analysis. RSA showed minor micro motions of the stem: Mean subsidence was 0.35 mm (precision: 0.10) and mean rotation around the longitudinal axis was 0.10º (precision: 0.64) after 2 years.
Conclusions: The stems showed migrations characteristic for stable uncemented implants. Long term follow up will be performed in future years.

4. Surgical Selection Bias may influence results presented in current literature after two-stage revision of chronic peri-prosthetic hip joint infection.
Jeppe Lange, Alma B. Pedersen, Anders Troelsen, Kjeld Søballe
Orthopaedic Surgery, Regional Hospital Silkeborg; Clinical Epidemiology, Aarhus University Hospital

Background: Currently, the gold-standard treatment for hip periprosthetic joint infection (PJI) is two-stage revision. However, advocates for the one-stage revision strategy argue that current literature is biased in the selection of two-stage revision.
Purpose / Aim of Study: We wanted to investigate the potential presence of surgical selection bias in patients undergoing two-stage revision in chronic hip PJI.
Materials and Methods: We identified patients in the National Patient Register from 2003-2008 at 11 orthopaedic departments. A manual review of the medical records verified 130 patients with a treatment performed for a chronic hip PJI. We divided the patients, into two groups based on the revision strategy chosen. 82 patients constituted one group and was characterized by having a re-implantation performed in a two-stage revision. The remaining 48 were not treated using a two-stage revision
Findings / Results: Patients in the two-stage group were younger (mean age 68, 95% CI: 66-71 vs. 76, 95% CI: 72-80; p-value <0.001) and had better overall health as indicated by the surrogate health markers ASA score (median score 2, IQR: 0 vs. median 2, IQR: 1; p-value <0.001) and CCS Score (median score 0, IQR 1 vs. median 1, IQR 1; p-value 0.005). Patients not re-implanted had a crude 68% higher risk of dying in the follow-up period compared to patients undergoing two-stage revision.After adjusting for selected confounding variables the risk of dying remained 25% higher, although not found to be statistically significant
Conclusions: When discussing the optimal treatment strategy, surgeons must keep in mind that 1-in-3 of all chronic hip PJI are not treated with a two-stage revision. We also found a significant difference between our established groups indicating that patients undergoing re-implantation are a selected group of patients

5. The association between gender and the familial prevalence of hip dysplasia in Danish patients with hip dysplasia
Rima El Jashi, Maria Biehl Gustafson, Mette Bjørn Nielsen, Jens Michael Hertz, Kjeld Søballe, Inger Mechlenburg
Department of Orthopaedic Surgery , Aarhus University Hospital, Denmark ; Department of Clinical Genetics, Odense University Hospital; Department of Orthopaedic Surgery, Aarhus University Hospital, Denmark ; Department of Orthopaedic Surgery, Centre of Research in Rehabilitation (CORIR), Department of Clini, Aarhus University Hospital and Aarhus University, Denmark

Background: The development of hip dysplasia is associated with several risk factors. One of the risk factors is gender and 80% of patients with symptomatic hip dysplasia are females. Another risk factor is familial prevalence of hip dysplasia. Studies indicate that the risk of hip dysplasia is increased with familial prevalence of hip dysplasia. However, little is known about the association between the familial prevalence and gender and development of hip dysplasia.
Purpose / Aim of Study: The aim of the study was to estimate the prevalence and genealogy degree of family history in Danish patients with hip dysplasia operated with periacetabular osteotomy (PAO) and furthermore, to assess the association of gender and family history in the same group of patients.
Materials and Methods: A cross-sectional study of 676 patients drawn from a clinical database of patients operated with a PAO at Aarhus University hospital from 1998 to 2014. Information on gender operated hip side and age was collected from the clinical PAO database, and information on familial prevalence was collected through questionnaires. The association between gender and familial prevalence of hip dysplasia was presented as the prevalence proportions ratio (PPR), tested by χ2 test. Stratification was conducted for the variables age and operated hip side, and tested for statistical significance.
Findings / Results: Familial prevalence of hip dysplasia was 30% (95% CI 27; 34), with 73% reporting 1st degree relatives. Females had 32% increased risk of familial prevalence of hip dysplasia compared to men, but this difference in risk was not statistically significant (p=0.10).
Conclusions: Females had increased familial prevalence of hip dysplasia compared to men, but the increased prevalence was not statistically significant probably due to the low power of the study.

6. Validation of the prosthetic joint infection diagnosis in the danish hip arthroplasty register
Per Hviid Gundtoft, Alma Becic Pedersen, Henrik Carl Schønheyder, Søren Overgaard
1. Ortopædkirurgisk 2: Ortopædkirurgisk 3: Klinisk Institut, 1: Kolding Sygehus 2: Odense University Hospital 3: Sydddansk Universitet; Klinisk Epidemiologisk, Aarhus Universitetshospital; Klinisk Mikrobiologisk, Aalborg Universitetshospital; 1: Ortopædkirurgisk 2: Klinisk Institut, 1: Odense Universitetshospital 2: Sydddansk Universitet

Background: The Danish Hip Arthroplasty Register (DHR) is a national database on total hip arthroplasties (THAs) with a high completeness and validity of registration for primary procedures.
Purpose / Aim of Study: The aim was to validate the registration in DHR for revisions due to Prosthetic Joint Infection (PJI).
Materials and Methods: We identified a cohort of patients in the DHR who underwent primary THA from January 1, 2005 to December 31, 2012 and followed these patients until first-time revision, death, emigration or December 31, 2012. The PJI-diagnosis registered was tested against a gold standard encompassing information from microbiology, prescription, clinical biochemistry registries and clinical findings retrieved from medical records. We estimated the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) with 95% confidence interval for PJI in DHR alone and in combination with microbiology registries.
Findings / Results: Out of 37,828 primary THAs, 1,382 were registered with any revision, 232 of which were due to PJI. For PJI revisions in DHR, the sensitivity was 67.0% (CI: 61.0 - 72.6), specificity 95.2% (CI: 93.8 - 96.4), PPV 77.2% (CI: 71.2 - 82.4), and NPV 92.3% (CI: 90.7 - 93.8). Combining DHR with microbiology registries led to a notable increase in the sensitivity for PJI revision to 90.3% (CI: 86.1 - 93.5) and likewise for specificity 99.6% (CI: 99.1 - 99.9), PPV 98.4% (CI: 95.9 - 99.6) and NPV 98.5% (CI: 97.6 - 99.1).
Conclusions: Only two thirds of PJI revisions were captured in DHR and the PPV was moderate (77%). However, combining DHR with microbiology registries improved the accuracy remarkably.

7. The suitability of dynamic RSA for evaluation of dual-mobility liner motion
Peter Bo Jørgensen, Bart L. Kaptein, Maiken Stilling
Orthopaedic Research, Aarhus University Hospital; Dept. of Orthopaedic Surgery, Leiden University Medical Center

Background: Dual-mobility hip prostheses have potential benefits in terms of increased hip joint mobility, decreased dislocation risk, and less polyethylene wear. The shape, material and surroundings of the polyethylene component, complicates recording of the essential in-vivo liner motion.
Purpose / Aim of Study: To assess the feasibility of in-vivo liner motion recording using dynamic RSA in a phantom set-up.
Materials and Methods: The phantom set-up consisted of a Sawbones® pelvis and extremity, and was prepared with a Restoration ADM® cup (Stryker Inc.) size 56. A matching liner was prepared with 12 1-mm tantalum beads in four unique groups and a 28 mm ceramic femoral head (Biolox) and a CoCr femoral stem. The phantom was positioned supine with 45 degrees flexion in the hip for dynamic stereometric recordings at 5 fps through a 10 cm acrylic layer, with a set-up (tubes and calibration box) tilted in a 45 degree cranial/caudal angle during external and internal rotation in the hip joint. A marker-model of the polyethylene tantalum-beads was combined from multiple RSA-recordings using visual validation and computed clustering. A CAD model of the ADM cup and stem was registered to the contour of the implant in the stereoradiographs and relative angles of the liner and femoral component were evaluated with respect to the ADM cup.
Findings / Results: Of 12 polyethylene bead-markers we obtained a model of 10 markers (83%). The relative angle varied between polyethylene liner and cup (>8 deg.) and between polyethylene liner and the femoral component (>20 deg.) during internal-external rotation.
Conclusions: Liner- and stem motion in a dual-mobility hip implant was quantified in a phantom set-up by use of dynamic RSA. In-vivo measurement of polyethylene motions in the dual-mobility implant is expected to work well.

8. Improved diagnostic accuracy of prosthetic joint infection (PJI) by combining culture of sonication fluid and tissue samples from revision total hip (THA) and knee arthroplasty (TKA)
Christen Ravn, Michael Kemp, Per Kjærsgaard-Andersen, Søren Overgaard
Dep. of Orthopaedic Surgery and Traumatology, Institute of Clinical Research, Odense University Hospital and University of Southern Denmark; Dep. of Clinical Microbiology and Institute of Clinical Research, Odense University Hospital and University of Southern Denmark; Dep. of Orthopaedic Surgery, Vejle Hospital

Background: Culture-negative PJI is challenging and occurs in 60- 80 % of THA and TKA revisions suspected for deep infection. Culture of dislodged biofilm bacteria in sonication fluid (cSF) from explanted implant materials has been reported to increase the diagnostic sensitivity in comparison to conventional culture of periprosthetic tissue samples (cPT) (Trampuz, 2007. NEJM).
Purpose / Aim of Study: To compare the diagnostic accuracy of cPT alone with a combination of cPT and cSF in revision of THA and TKA.
Materials and Methods: We prospectively analyzed explanted prosthetic materials from consecutive THA and TKA revision surgery performed at Odense University Hospital and Vejle Hospital during one year. The prostheses were sonicated for one minute due to the Trampuz- protocol (NEJM, 2007). cSF was considered positive with >20 colonies/ml. Conventional cPT was performed due to best practice and considered positive with identical bacterial findings in 3/5 tissue samples. We defined PJI by different criteria incl. microbiology, and/or a combination of biochemistry and clinical findings.
Findings / Results: Revision of 214 THA (131) and TKA (83) were indicated by aseptic loosening (73), deep infection (57) and ‘other indications’ (84). PJI was defined in 56 cases and microbial diagnosis was achieved in 41 and 48 cases by either cPT or cSF. Bacterial findings were similar in all concordant cases. Sensitivity and specificity of cPT alone was 0.73 and 1.00 respectively. The combination of cPT and cSF gave a microbial diagnosis in 53 PJI-cases; sensitivity 0.93 and specificity 0.95 (p<0.05).
Conclusions: The combination of cPT and cSF significantly increased the sensitivity of culture-based methods for PJI-diagnosis leaving only 3/56 PJI-cases culture-negative. Sonication is useful for detecting surface-attached bacteria living as biofilm on prosthetic implants.

9. Laminar airflow reduces microbial air contamination in comparison to turbulent airflow during simulated total hip arthroplasty (THA)
Christen Ravn, Anders Overgaard, Niels B. Knudsen, Jørn Toftum, Lars Henrik Frich, Søren Overgaard
Dep. of Orthopaedic Surgery and Traumatology, Institute of Clinical Research, Odense University Hospital and University of Southern Denmark; Dep. of Orthopaedic Surgery, Herlev University Hospital; Fournais Energi Aps, ; Dep. Of Civil Engineering, Technical University of Denmark

Background: It has been debated whether type of ventilation system LAF versus TAP can influence the risk of deep infection following THA and TKA. Moreover, choice of ventilation system is of debate in construction of hospitals
Purpose / Aim of Study: To compare the number of airborne bacteria and particles under laminar airflow (LAF) vs turbulent airflow (TAF) with 100% or 50% reduced fresh air exchange in simulated THA
Materials and Methods: We conducted 32 simulated operations under 4 conditions: LAF or TAF with either full (n=8+8) or 50% reduced (n=8+8) air exchange. Microbial contamination was determined by Microbiological Active Sampler (MAS-100), and colony-forming units (CFU) per m3 air were counted after 2 days incubation. Airborne particulate (0.5-10 µm) was sampled with light scattering particle analyzer (MET- 1). Large particle sizes (>5 µm) must not exceed a 2.900/m3-threshold for cleanroom operations
Findings / Results: Microbial air concentration (mean CFU/m3 ±SD) with LAF under full and 50% reduced air exchange were 0.4±0.8 and 0.4±0.4 respectively, whereas air contamination under TAF conditions had significantly higher values of 7.6±2.0 and 10.3±8.1, respectively (p<0.05) Large (>5 µm) airborne particulate (mean no./m3 ±SD) with LAF under full and 50% reduced air exchange were 1.581±2.841 and 1.018±1.084 respectively, whereas particulate under TAF conditions were 7.923±5.151 and 6.157±2.439 (p>0.05). The number of particles measured under TAF conditions exceeded the threshold for cleanroom operations in 12/16 simulated operations
Conclusions: Microbial air contamination was significantly lower under LAF ventilation compared to TAF during simulated THA under both full and 50% reduced air exchange in modern operating theatres. These findings indicate that LAF may reduce the airborne microbial risk factor of surgical site infection in comparison to TAF

10. How accurate are common biochemical and microbiological tests towards a multi-criteria definition of prosthetic joint infection (PJI) for revision total hip (THA) and knee arthroplasty (TKA)?
Christen Ravn, Michael Kemp, Per Kjærsgaard-Andersen, Søren Overgaard
Dep. of Orthopaedic Surgery and Traumatology, Institute of Clinical Research, Odense University Hospital and University of Southern Denmark; Dep. of Clinical Microbiology and Institute of Clinical Research, Odense University Hospital and University of Southern Denmark; Dep. of Orthopaedic Surgery, Vejle Hospital

Background: A certain diagnosis of PJI is sometimes difficult to achieve when clinical, biochemical and microbiological findings are ambiguous. Seemingly there is no Danish consensus for a PJI definition as proposed by the International Consensus Meeting (Parvizi, 2013. BJJ)
Purpose / Aim of Study: We suggest a PJI definition incl. pre-operative C- reactive protein (CRP) as well as culture of intra- operative joint fluid (JF) and periprosthetic tissue samples (PT) from patients undergoing revision surgery in order to determine the diagnostic accuracy of each parameter
Materials and Methods: Revision of 214 THA (131) and TKA (83) were indicated by aseptic loosening (73), deep infection (57) and other (84). We analyzed CRP, JF- and PT-culture and defined PJI by either 1-4: 1. Presence of a sinus tract 2. Culture of identical microorganisms in ≥3/5 PT 3. Culture of identical or microorganisms in 2/5 PT in combination with either -a JF cultured with an identical organism OR -b Elevated CRP (threshold >100mg/l at 0-90 postoperative days; >10mg/l after 90 days) 4. Minor criteria (required 3/4): -a Culture-positive PT (≥1/5) -b Culture-positive JF -c Purulence of JF -d Elevated CRP Results are presented as number of test-positive PJI-cases (sensitivity / specificity)
Findings / Results: PJI was found in 56 cases according to our definition criteria 1-4. CRP was elevated in 32 cases (0.60 / 0.85). JF-aspiration was dry in 35 and culture- negative in 136 cases. JF-culture gave a microbial diagnosis in 37 PJI-cases (0.70 / 0.96). PT-culture with a diagnostic threshold at 3/5 positive samples gave a microbial diagnosis in 41 PJI-cases (0.73 / 1.00)
Conclusions: Diagnostic accuracy of each parameter showed no tests exceed 0.75 leaving every fourth PJI-cases without a bacterial diagnosis. Thus in order to increase sensitivity we suggest introducing new validated diagnostic tools.

11. Serious renal and urological complications in fast-track primary total hip and knee arthroplasty
Lars Bjerregaard, Christoffer Jørgensen, Henrik Kehlet
Section for Surgical Pathophysiology and the Lundbeck Foundation Centre for Fast-track Hip and Knee , Rigshospitalet, Copenhagen University, Denmark.

Background: The fast-track methodology has reduced overall medical complications after total hip (THA) and knee arthroplasty (TKA). However, limited data exists on renal and urological (RU) complications.
Purpose / Aim of Study: To study the incidence and consequences of serious RU complications in otherwise uncomplicated elective fast-track THA and TKA, resulting in length of stay (LOS) > 4 days or a 30-day readmission.
Materials and Methods: A detailed, observational study of prospectively collected pre-operative data with complete 30-day follow-up based on the Danish National Patient Registry and review of medical charts in a large cohort of 8,804 fast-track THA and TKA from 7 Danish high-volume orthopaedic departments, with similar standardised fast-track protocols.
Findings / Results: Of 8,804 procedures, 0.6 % developed serious RU complications resulting in 38 LOS >4 days and 17 readmissions. Acute kidney injury (AKI), defined as an increase in serum creatinine by ¡Ý 0.3 mg/dl or ¡Ý 1.5 times baseline, accounted for 43 cases (0.5 %), and was most frequently explained by dehydration. Of the 43 patients with AKI, 25 (58 %) had a preoperative estimated glomerular filtration rate < 60 mmol/ml. and 16 of these (64 %) had received a NSAID as postoperative pain treatment. 7 cases (0.1 %) were due to urological complications, mainly haematuria after bladder catheterisation, and 5 patients (0.1 %) developed postoperative urosepsis or pyelonephritis.
Conclusions: The overall incidence of serious RU complications after fast-track THA and TKA was low (0.6 %). Of these, 78 % were AKI, most often due to dehydration, pre-existing kidney disease and NSAID treatment, calling for an increased focus on perioperative fluid management and optimisation of the perioperative care of patients with pre- existing kidney disease.

12. When to catheterise after fast-track total hip (THA) and knee arthroplasty (TKA) – a RCT on bladder volumes of 500 vs. 800 ml
Lars Stryhn Bjerregaard, Ulla Hornum, Charlotte Troldborg, Bogø Stina, Per Bagi, Henrik Kehlet
Section for Surgical Pathophysiology and the Lundbeck Foundation Centre for Fast-track Hip and Knee , Rigshospitalet, Copenhagen university; Department of Orthopaedics, Aalborg University Hospital, Farsø; Department of Urology, Rigshospitalet

Background: A bladder volume of 500 ml is a commonly used catheterisation threshold for treating POUR after THA and TKA, but this threshold is without evidential support.(1,2)
Purpose / Aim of Study: The purpose was, in a RCT to compare two bladder volumes (500 vs. 800 ml) as catheterisation threshold.
Materials and Methods: We included 800 adult patients (2x400), planned for fast track THA or TKA and who had given their informed written consent. 721 patients completed the study. Bladder scans were performed at 2 hours intervals until the first voluntary micturition. All patients completed a preoperative questionnaire on voiding difficulties, which were repeated at 30 days follow-up. Main outcome was the proportion of patients needing postoperative catheterisation. Secondary outcomes included the incidences of UTI and voiding difficulties within the 30 days follow-up period.
Findings / Results: 114 of 355 patients (32.1 %) were catheterised in the 500 mL group compared to 49 of 366 patients (13.4 %) in the 800 mL group (p < 0.0001). The need for repeated catheterisation was 4.5 % vs. 0.8 % respectively. Follow-up data on the incidences of UTI and PO voiding difficulties will be presented at the conference.
Conclusions: Our results show that increasing the catheterisation threshold to a bladder volume of 800 mL reduces the need for postoperative catheterisation. This first large RCT on the topic may serve as basis for future evidence based guidelines on perioperative urinary bladder management in fast track THA and TKA. References 1. Bjerregaard LS, Bagi P, Kehlet H. Postoperative urinary retention (POUR) in fast-track total hip and knee arthroplasty. Acta Orthop 2014;85:8-10. 2. Bjerregaard LS, Bogo S, Raaschou S et al. Incidence of and risk factors for postoperative urinary retention in fast- track hip and knee arthroplasty. Acta Orthop 2015;86:183-8.