Infection after total joint replacement of the hip and knee: research programme including the INFORM RCT

Article history

The research reported in this issue of the journal was funded by PGfAR as project number RP-PG-1210-12005. The contractual start date was in February 2014. The final report began editorial review in June 2020 and was accepted for publication in May 2021. As the funder, the PGfAR programme agreed the research questions and study designs in advance with the investigators. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The PGfAR editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2022 Blom et al. This work was produced by Blom et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2022 Blom et al.

Background to the INFORM programme

Programme development

To support our programme application, we conducted a systematic review comparing single- and two-stage revision procedures in patients with hip prosthetic joint infection. 47 We found 2-year rates of reinfection or development of a new infection after single-stage revision of 8.6% [95% confidence interval (CI) 4.5% to 13.9%] and after two-stage revision of 10.2% (95% CI 7.7% to 12.9%). We limited bias by including series of consecutive patients with hip prosthetic joint infection treated exclusively by single- or two-stage revision. We concluded that randomised trials were needed to establish optimum management strategies. Furthermore, the evidence base was highly limited regarding health service, patient and economic significance, risk factors, patient and surgeon experiences, diagnosis, and patient preferences for treatments.

Programme development: patient and public involvement

Collaboration with patient-partners took place through meetings with our dedicated patient and public involvement (PPI) group, the Patient Experience Partnership in Research (PEP-R), which comprised nine people who had musculoskeletal conditions. 48 The group met every 6–8 weeks, providing input into project design and conduct. Through ongoing training and support from research staff and our patient involvement co-ordinator, group members were familiar with research designs, conduct and some of the barriers to successful research. We used National Institute for Health and Care Research (NIHR) INVOLVE guidance to ensure that PPI was appropriately organised49 and we were part of the South West ‘People in Research’ consortium.

In December 2010, February 2011 and September 2011 we discussed the planned research with PEP-R. The group believed that, although relatively few people developed an infection, the research was important. They noted the need for feasibility work to assess the acceptability of randomisation. They offered their full support for future work and we considered their ongoing involvement crucial to ensure that the project reflected the priorities of the public and was interwoven throughout the programme.

The INFORM programme

The overall aim of the INFORM (INFection ORthopaedic Management) programme was to identify methods that may improve treatments and outcomes for patients with deep prosthetic joint infection after total hip or knee replacement. Studies were grouped by research methodology into seven work packages, all supported by PPI (Figure 1):

• Work package 1. We aimed to conduct systematic literature reviews of treatments for prosthetic joint infection after hip and knee replacement and conduct an individual patient data (IPD) meta-analysis to compare reinfection outcomes after single- and two-stage revision surgery. We also aimed to review risk factors, methods of diagnosis and costs.

• Work package 2. We aimed to analyse the National Joint Registry for England, Wales, Northern Ireland and the Isle of Man (NJR) to identify the predictors of prosthetic joint infection after hip and knee replacement and to compare outcomes according to health-care characteristics.

• Work package 3. Through qualitative interviews, we aimed to assess the impact on patients of prosthetic joint infection and treatment strategies, surgeons’ views on treatment, and patients’ and surgeons’ views on a randomised trial.

• Work package 4. In a randomised controlled trial (RCT) with embedded qualitative interviews, we aimed to investigate whether or not treating hip prosthetic joint infection with a single-stage revision rather than the traditionally used two-stage revision improved patients’ quality of life and was cost-effective.

• Work package 5. In analyses of the NJR we aimed to assess the economic implications of prosthetic joint infection and treatment strategies. We also aimed to conduct a comprehensive economic evaluation within the RCT comparing single- and two-stage revision for treatment of hip prosthetic joint infection.

• Work package 6. Through a discrete choice questionnaire, we aimed to assess the trade-offs that patients are willing to make between patient-reported and clinical outcomes and explore the degree to which treatment strategies change preferences for those outcomes.

• Work package 7. Finally, we aimed to disseminate findings to patients, the public, clinicians and stakeholders.

FIGURE 1.

The INFORM programme. WP, work package.

Work packages were conducted in parallel. An exception was the work package 3 qualitative study looking at surgeon decision-making, which directly influenced the design of the INFORM RCT in work package 4. The study of patient preferences in work package 6 followed on from the INFORM trial but focused on patient experiences of surgical treatments independently of trial participation.

Changes to the programme and additional research

Planned research is published, has been submitted to a journal or is being written up. We have also received a NIHR Programme Development Grant to develop recommendations based on the INFORM findings and explore implementation.

An additional published systematic review of outcomes after DAIR is summarised. Further studies included systematic reviews of risk prediction tools, health-care needs and support, and treatment comparisons in ankle and shoulder infection; and a UK survey of care pathways.

We had originally planned to examine the accuracy, costs and cost-effectiveness of a broad range of methods for diagnosing and monitoring infection. Ultimately, in the programme, we focused on new promising biomarkers for the diagnosis of infection.

In NJR analyses, we planned to compare outcomes in ‘specialist’ and other centres. We were unable to define ‘specialist’ centre and have compared care according to different health-care characteristics.

The NJR and Hospital Episode Statistics (HES) linked analysis of economic burden of hip prosthetic joint infection was completed as planned. The study of knee prosthetic joint infection has been delayed as the COVID-19 pandemic prevented on-site access to NJR data.

Programme Steering Committee

The programme was overseen by a Programme Steering Committee chaired by Professor Rod Taylor, NIHR Senior Investigator and Professor of Health Services Research, University of Exeter. Other independent committee members were Martyn Porter, the past president of the British Orthopaedic Association and Medical Director of the NJR; Ali Heawood, Senior Research Fellow in Qualitative Primary Care Research; and Julie Chappell, a solicitor specialising in medico-legal negligence. The Programme Steering Committee met on nine occasions (every 6 months), with at least three independent members present, to monitor the progress of the programme and help the research team to meet agreed milestones. They expressed their satisfaction with progress in all work packages but noted the slow recruitment to the RCT and supported measures to enhance recruitment, in particular the inclusion of more sites.

Pan-programme group

The pan-programme group (research team, co-applicants, site principal investigators and sponsor representative) met on 24 occasions (every 3 months) to discuss progress, outputs and dissemination from each of the work packages.

Work packages

Each work package lead arranged an internal working group that met regularly to ensure the delivery of agreed outputs. Work package 4 was run by the Trial Management Group (chief investigator, site principal investigators, co-applicants, trial manager, statistician, health economist, qualitative researcher and methodologist from Bristol Randomised Trials Collaboration), which met monthly to design and deliver the randomised trial.

Data monitoring committee

For work package 4, an independent data monitoring committee was established, consisting of a trial methodologist and two orthopaedic surgeons. This committee met on four occasions and reported directly to the Programme Steering Committee.

Microbiology

A meeting was held in 2014 with the lead consultant clinical microbiologists from four of the trial sites (Bristol, Exeter, Oxford and Cardiff). The group assisted with trial design and participant recruitment processes and advised on the collection of microbiological, antibiotic and blood biomarker data.

Study sponsor

Good Clinical Practice, local research governance and finance were overseen by the study sponsor (North Bristol NHS Trust Research and Innovation Department) and the University of Bristol. NHS Research Ethics and Research and Development approvals were obtained for work packages 3, 4 and 6.

INFORM randomised controlled trial

Publication committee

To achieve high publication standards, an INFORM publication policy was adhered to. Manuscripts and conference articles were approved by the committee prior to submission. The committee consisted of Professor Ashley Blom (chief investigator), Professor Rachael Gooberman-Hill (qualitative lead) and Mr Stephen Jones (orthopaedic surgeon).

Aims

The aims of our PPI were to gain input into the design of patient recruitment and information literature, research processes and questionnaires; identification of outcomes of importance to patients; and dissemination strategies. PPI also helped to ensure that milestones were met and that input was provided into the interpretation of data and findings:

It feels that between us all we may make a difference to patients who may have an infection.

INFORM forum group member

Patient and public involvement

Patients have been crucial to the success of this research in the following ways:

Patient and public involvement meetings

Members of the PEP-R group worked on the grant application and were involved throughout the programme. A meeting in March 2020 to share results was postponed due to the COVID-19 pandemic. We will continue working with the group remotely. PEP-R group members helped to prepare the Plain English summary for this report.

The INFORM patient forum has met 33 times and provided input into all work packages. We are continuing to work with the forum remotely to complete work on writing an information leaflet for newly diagnosed patients, reviewing information on prosthetic joint infection on the internet, writing a summary of results for trial participants, creating infographics, finishing their conference presentation, and working with Versus Arthritis to support and inform patients who have prosthetic joint infection. The PPI activity is summarised in Table 1.

Contributions to INFORM and impact on programme

The impact of the INFORM patient forum on programme conduct and dissemination is summarised in Tables 2 and 3. In Box 1 we summarise INFORM researcher comments on PPI impact as reported in evaluations in 2014 and 2019.

Public dissemination

Dissemination of the results from INFORM is ongoing. A 1-day conference to share results with study participants, patient-partners, carers and the general public was planned for May 2020 but has been delayed because of the COVID pandemic.

Our film about the INFORM randomised trial was featured on University of Bristol and NIHR websites in June 2018. By 22 February 2020 it had been viewed 378 times (www.youtube.com/watch?v=_AH1Nu3tWVch).

Our film about the PPI evaluation was tweeted about by the University of Bristol and featured in the INVOLVE September 2018 newsletter. By 22 February 2021 it had been viewed 231 times (www.youtube.com/watch?v=TrZfHfVaMZE).

The Musculoskeletal Research Unit website is at www.bristol.ac.uk/translational-health-sciences/research/musculoskeletal/orthopaedic/research/inform.html and INFORM Twitter (Twitter, Inc., San Francisco, CA, USA; www.twitter.com) account is at https://twitter.com/BristolINFORM.

Aims

We aimed to describe the risk of revision due to prosthetic joint infection for patients undergoing primary and revision hip or knee replacement, the changes in risk over time and the overall significance of prosthetic joint infection to the NHS. Supported by a systematic review, we aimed to estimate the economic significance of revision for hip prosthetic joint infection.

Methods

Results

Surgeon, patient and health-care implications of hip prosthetic joint infection

The NJR cohort analysis included 623,253 primary hip replacements and 63,222 aseptic revision hip replacements. A total of 7040 patients required one or more courses of revision surgery to treat prosthetic joint infection (n = 7585 procedures). Of these, 3338 patients (n = 3546 procedures) had no first surgery recorded in the NJR. Thus, our analyses were limited to 3702 patients (n = 4039 procedures), of whom 2705 had an index primary hip replacement (n = 2926 procedures) and 997 had a revision hip replacement (n = 1113 procedures).

The cumulative incidence functions of revision for prosthetic joint infection following index primary and aseptic revision hip replacement are shown in Figure 2. The probability of revision for prosthetic joint infection after a primary hip replacement was 0.15% (95% CI 0.14% to 0.16%) at 1 year and 0.62% (95% CI 0.59% to 0.65%) at 10 years. The probability of revision for prosthetic joint infection following an aseptic revision was higher, with rates of 0.69% (95% CI 0.63% to 0.76%) at 1 year and 2.25% (95% CI 2.08% to 2.43%) at 10 years.

FIGURE 2.

Cumulative incidence function of revision for prosthetic joint infection following index primary and aseptic revision hip replacement.

The prevalence of revision due to prosthetic joint infection in the 3 months following primary hip replacement more than doubled between 2005 and 2013 (rate ratio 2.29, 95% CI 1.28 to 4.08; time trend likelihood ratio test p < 0.0001). No time trends for revision for the time periods 3–6 months, 6 months to 1 year and 1–2 years were found. However, later rates of revision prosthetic joint infection decreased over time (between 2 and 3 years rate ratio 0.52, 95% CI 0.32 to 0.84, time trend p < 0.0001; between 3 and 4 years rate ratio 0.49, 95% CI 0.31 to 0.80, time trend p = 0.001; and between 4 and 5 years rate ratio 0.44, 95% CI 0.27 to 0.73, time trend p = 0.028).

The absolute number of procedures performed annually to treat infection increased from 384 in 2005 to 1002 in 2014, a 2.6-fold increase. This was greater than the twofold increase in primary procedures during this period.

Overall, 70% of revisions for prosthetic joint infection after hip replacement were two-stage procedures. However, in 2014, 60.7% of revision surgeries were carried out with a two-stage procedure and 29.7% were carried out with a single-stage procedure. The use of DAIR rose from < 1% of revision surgeries in 2005 to 7.6% in 2014.

Surgeon, patient and health-care implications of knee prosthetic joint infection

Between 2003 and 2014, 679,010 index primary knee replacements and 33,920 index revision knee replacements were registered in the NJR. A total of 8247 revision total knee replacements were performed as a result of a diagnosis of prosthetic joint infection. Our analyses of prevalence were limited to 3659 patients with a primary knee replacement (4004 procedures) and 717 patients with a revision knee replacement (785 procedures) as 3458 index operations had been conducted before the establishment of the NJR.

The cumulative incidence functions of revision for prosthetic joint infection following index primary and aseptic revision knee replacement are shown in Figure 3. The probability of revision knee replacement due to prosthetic joint infection was 0.17% (95% CI 0.16% to 0.18%) at 1 year and 0.75% (95% CI 0.72% to 0.78%) at 10 years. The probability of revision for prosthetic joint infection following an aseptic revision was higher, with rates of 0.76% (95% CI 0.68% to 0.86%) at 1 year and 3.13% (95% CI 2.81% to 3.49%) at 10 years.

FIGURE 3.

Cumulative incidence function of revision for prosthetic joint infection following index primary and aseptic revision knee replacement.

Revision rates within 3 months of the index knee replacement increased over time, with the prevalence rate in 2013 over twice that in 2005 (rate ratio 2.46, 95% CI 1.15 to 5.25; p < 0.0001). No time trends for revision for other time periods were apparent with the exception 5–6 years, which decreased over time (rate ratio 0.53, 95% CI 0.33 to 0.86; time trend p = 0.018).

In 2014, 1048 revision procedures were performed to treat knee prosthetic joint infection compared with 378 in 2005, a 2.8-fold increase. Overall, 75% of revision operations were conducted as a two-stage procedure, but there was an increase in use of single-stage revision from 7.9% in 2005 to 18.8% in 2014. The use of DAIR also increased, from 2.1% in 2005 to 9.1% in 2014.

Conclusions

The 10-year rates of revision for prosthetic joint infection after hip and knee replacement were 0.62% and 0.75%, which are lower than those in Scandinavian registries. 30,31 People receiving aseptic revision were at four times greater risk of hip or knee prosthetic joint infection.

There was a suggestion that prosthetic joint infection rates rose between 2005 and 2013. This was limited to the first 3 months after primary surgery and may reflect earlier identification and treatment of infection. The use of single-stage revision and DAIR has increased, and although the use of two-stage revision has decreased it is still the most widely used treatment.

Registry analyses showed fivefold higher health-care costs in the 5 years after primary surgery for people with prosthetic joint infection than for people with no infection. This was consistent with studies from Canada, France, Germany and the USA that reported costs of revision for infection and subsequent treatments ranging from 3.7 to 5.3 times that of primary hip or knee replacement. Comparative costs of treatment strategies are dependent on the success rate of treatments, which varied markedly in studies comparing single- and two-stage revision.

Limitations

In NJR analyses, we do not know how many people had an infection that was not treated surgically. However, the number treated without surgery is probably small as biofilm involvement means that antibiotic treatment alone can only suppress infection. Furthermore, arthroscopic management is not recommended. 53 In a Swedish cohort, 91% of people treated for prosthetic joint infection, and who subsequently received continuous outpatient antibiotic treatment, underwent a reoperation. 28 The UK Bone and Joint Infection Registry (BAJIR), established in 2018, collects information on all patients diagnosed with prosthetic joint infection in the UK and will provide accurate estimations of infection rates. 54

In the systematic review, the resources included in cost calculations varied considerably. In a comparison of costs and cost-effectiveness, future larger-scale and longer-term studies should assess costs to health service providers, social care and society, including those attributable to further treatments and revision operations.

In our NJR analysis, we estimated the burden of hip prosthetic joint infection with respect to inpatient and day-case admissions but did not consider costs relating to outpatient, primary and community care, and prescribed medications. Including these would have increased the overall costs in both patients who were and patients who were not treated for hip prosthetic joint infection and would probably have increased the cost difference between the two groups. For patients revised for hip prosthetic joint infection, the average cost of inpatient and day-case admissions in the 5 years following primary total hip replacement (including the cost of revision for prosthetic joint infection) was £41,633 using 2014–15 costs. Vanhegan et al. 55 reported that the cost of revision for hip prosthetic joint infection alone was £21,937 at a single UK hospital using costs from 2007–8. The authors found that this cost did not reflect NHS tariffs, which were considerably lower. Assuming that this underestimate is true and also holds for NHS reference costs (which were used in this study), the financial burden of hip prosthetic joint infection may have been underestimated.

As HES include hospital admissions at NHS hospitals in England only, the cost of admissions funded by the NHS outside England or in private facilities was not included. An estimated 83% of hip revision procedures are conducted in NHS hospitals. 56 With the need for involvement of an arthroplasty surgeon experienced in treating prosthetic joint infection and a well-co-ordinated multidisciplinary team, complex cases requiring revision surgery are conducted almost exclusively in NHS orthopaedic centres.

In the INFORM randomised trial, health service, social care and participant resource use data were collected and used in the cost-effectiveness analysis.

Aims

We aimed to describe patients’ experiences and the impact of revision treatment for prosthetic joint infection after hip and knee replacement, and to compare patients’ experiences of single-stage and two-stage revision surgery.

Methods

We conducted qualitative semistructured interviews with patients who had received surgical revision for prosthetic joint infection after hip or knee replacement. Thirty-five patients (19 hip replacement and 16 knee replacement) from five NHS orthopaedic departments in England and Wales were interviewed between 2 weeks and 12 months after they had been discharged from hospital. Interview topic guides were developed in collaboration with the INFORM PPI group. The use of the topic guide was flexible to ensure that key topics were covered but that participants were able to discuss issues they considered important. The questions addressed included those on the experience of prosthetic joint infection, revision surgery and care, the impact of infection and treatment, and thoughts about recovery and the future. We asked participants who had received two-stage revision about their experiences of the time between operations. Data were audio-recorded, transcribed, anonymised and analysed thematically59 using the qualitative data management software NVivo 10 (QSR International, Warrington, UK).

Results

Conclusions

Prosthetic joint infection is life-changing and has an impact on all aspects of a patient’s life. Among patients who had undergone revision surgery for prosthetic joint infection, a two-stage procedure had greater impact on participants’ well-being than a single-stage procedure because the time between revision procedures meant long periods of immobility, pain and related psychological distress. Participants expressed a need for more psychological and rehabilitative support during treatment and long-term recovery.

Aims

We aimed to explore decision-making by consultant orthopaedic surgeons about the use of single- or two-stage revision surgery for patients with hip prosthetic joint infection to inform our assessment of the feasibility of a randomised trial.

Methods

To guide the development of the INFORM multicentre RCT, we conducted semistructured interviews with 12 consultant surgeons performing revision surgery for hip prosthetic joint infection at five high-volume NHS orthopaedic departments in England and Wales. We analysed the data thematically.

Results

When choosing between single- and two-stage revision for hip prosthetic joint infection, surgeons considered multiple factors, including the patient context (age, ability to cope, carer responsibilities); patient preferences; primary prosthesis fixation; patient physiology (bone integrity and tissue damage, comorbidities, frailty); surgeon’s knowledge, training and peer influence; local infrastructure (availability of microbiology, resources, costs); the infecting organism (clinicians’ ability to identify, sensitivity to antibiotics); and duration of infection (acute or chronic). With evidence accruing on similar outcomes between surgical techniques, and observations of colleagues’ successful use of single-stage revision, surgeons questioned whether or not revision in two stages remained the best treatment for hip prosthetic joint infection, and some were increasingly willing to consider more revisions in a single stage. Some surgeons managed uncertainty about the choice of surgical technique by using a CUMARS. Although single-stage revision was considered to be, potentially, the best strategy to treat hip prosthetic joint infection, surgeons thought that a change in practice was not yet justified. To inform any change in practice, surgeons noted the need for evidence from randomised trials. If there was no clear best strategy based on patient factors, their own knowledge and expertise, the available infrastructure and the infecting organism, then surgeons believed that the patients they treated would be eligible for randomisation.

Conclusions

With growing evidence of the success of treatment of hip prosthetic joint infection in a single stage, the willingness of some surgeons to change practice has increased over time. By using a CUMARS, surgeons were able to manage uncertainty about the choice between single- and two-stage revision. To guide treatment, surgeons identified the need for high-quality evidence to support their choice of revision strategy. Surgeons believed that a RCT comparing single- and two-stage revision for hip prosthetic joint infection is needed. While recognising that patient, infection, surgeon and infrastructure factors may indicate the need for a particular strategy, surgeons considered that randomisation would be feasible and acceptable.

Aims

Although prosthetic joint infection has been described as ‘an orthopaedic surgeon’s worst nightmare’,62 relatively little is known about how prosthetic joint infection affects surgeons professionally and personally. We explored the impact of cases of knee prosthetic joint infection on surgeons’ personal and professional well-being. The identification and acknowledgement of the emotional impact of prosthetic joint infection may help in developing support strategies and maintaining surgeons’ well-being,63 and a deeper understanding of the personal and professional impact of adverse events is needed. 64

Methods

We conducted qualitative telephone interviews with consultant orthopaedic surgeons who treated patients for prosthetic knee infection in one of six high-volume NHS orthopaedic departments. The interviews were audio-recorded, transcribed, anonymised and imported into the qualitative data management software NVivo 10 for thematic analysis. 59

Results

We interviewed 11 surgeons who had a mean of 9.5 years’ experience as an orthopaedic consultant (range 1 month to 20 years). Surgeons perceived that being required to treat knee prosthetic joint infection was inevitable at some time in their career and this was a major concern, irrespective of their years of experience.

In our analyses we identified three themes that characterised the views and experiences of surgeons treating knee prosthetic joint infection: at some point, infection is inevitable but surgeons still feel accountable; the profound emotional impact; and supporting each other.

Participating surgeons described dealing with a diagnosis of prosthetic joint infection as ‘devastating’, ‘soul-destroying’ and ‘deeply unpleasant’. Although they expected a diagnosis of prosthetic joint infection at some point in their career, its occurrence still made them question their practice, despite taking ‘every measure’, and left them feeling that they had let the patient down. Surgeons felt that empathy and honesty were important to patients who had received a diagnosis of prosthetic joint infection. They described how they needed to reflect carefully, questioning their practice, performance and surgical processes. To ensure that they maintained their confidence, surgeons felt that it was important to continue to perform surgery even after one of their patients had been diagnosed with prosthetic joint infection. Surgeons described their feelings of responsibility for patients’ well-being and how they were motivated by the potential for relieving a patient’s pain and improving quality of life, but that prosthetic joint infection could have serious consequences for a patient’s quality of life. When prosthetic joint infection does occur, surgeons highlighted the importance of acknowledging the diagnosis and being honest with patients. Our findings suggest that, in some departments, decisions about treatment may be made by surgeons individually and not always discussed with colleagues. Although the participants in the study felt supported by colleagues, one suggested that the occurrence of prosthetic joint infection could potentially be ‘isolating’ for a surgeon where such support did not exist.

Conclusion

Prosthetic joint infection has a considerable emotional impact on surgeons, who report a sense of devastation and personal ownership, particularly as they are largely unable to control its occurrence. Surgeons stressed the importance of a supportive multidisciplinary team in the management of prosthetic joint infection.

Background

The risk of developing prosthetic joint infection may be influenced by patient characteristics, the surgical intervention and postoperative care.

Aims

We aimed to identify risk factors for prosthetic joint infection after hip or knee replacement in new systematic reviews of published research and joint registry analyses.

Methods

Results: systematic reviews

General risk factors

Searches identified 66 observational studies, including 512,508 participants with, predominantly, hip and knee replacement. The mean follow-up was 3.7 years (range 1–17 years). The risk of prosthetic joint infection was greater in men than in women (RR 1.36, 95% CI 1.18 to 1.57) and in smokers than in non-smokers (RR 1.83, 95% CI 1.24 to 2.70) (Figure 4). There were no associations between age or alcohol intake and the risk of prosthetic joint infection. One study reported a lower risk of prosthetic joint infection in patients living in rural locations than in those living in non-rural locations. The results of meta-analyses were consistent in higher-quality studies.

FIGURE 4.

Sociodemographic characteristics and risk of prosthetic joint infection.

For BMI, there were consistent positive associations for comparisons with cut-off points of ≥ 30 kg/m² (Figure 5). Comparing BMI of ≥ 40 kg/m² with < 40 kg/m², the pooled RR was 3.68 (95% CI 2.25 to 6.01). In one study, people with BMI < 18.5 kg/m² had a greater risk of prosthetic joint infection than those with BMI of 18.5–30.0 kg/m².

FIGURE 5.

Body mass index and risk of prosthetic joint infection.

Medical and surgical risk factors for prosthetic joint infection were diabetes, rheumatoid arthritis, depression, history of steroid administration and previous joint surgery. There was no evidence of important associations with osteoarthritis, osteonecrosis, post-traumatic arthritis, cardiovascular disease, hypertension, cancer, high-risk dental procedures or intra-articular steroid injections.

Hip implant fixation

Searches identified four RCTs (945 hip replacements) and 11 observational cohorts (2,260,428 hip replacements).

All RCTs were rated as being at a low risk of bias for random sequence generation and incomplete outcome data but at an unclear risk of bias in ≥ 1 other area. There were no clear differences in the risk of prosthetic joint infection when cemented fixations were compared with uncemented or reverse hybrid fixations (Figure 6). In one RCT,75 the risk of prosthetic joint infection was lower in patients receiving implant fixation with antibiotic-loaded cement than in those receiving it with plain cement.

FIGURE 6.

Fixation types and risk of hip prosthetic joint infection in observational studies and RCTs.

In a pooled analysis of observational studies, any fixation with cement was associated with an increased risk of prosthetic joint infection compared with uncemented fixation. In studies with higher quality scores, the difference favouring uncemented over cemented fixation was consistent (RR 1.09, 95% CI 1.03 to 1.16).

Compared with antibiotic-loaded cemented fixations, plain cemented fixations were associated with an increased risk of prosthetic joint infection (RR 1.52, 95% CI 1.36 to 1.70).

Knee implant fixation

Searches identified eight RCTs (4029 knee replacements) and 24 observational studies (1,161,292 knee replacements).

Randomised controlled trials were rated as being at low risk of bias for incomplete outcome data and selective reporting. Four were rated as being at an unclear risk of bias for allocation concealment. There was no difference in the risk of prosthetic joint infection when uncemented fixation was compared with cemented or hybrid fixation and when hybrid fixation was compared with cemented fixation (Figure 7). In one trial76 that randomised 2948 patients, there was no difference in the risk of prosthetic joint infection when antibiotic-loaded cemented fixation was compared with plain cemented fixation.

FIGURE 7.

Fixation types and risk of knee prosthetic joint infection in observational studies and RCTs.

In a pooled analysis of observational studies, uncemented fixation was associated with a decreased overall risk of prosthetic joint infection compared with cemented fixation (RR 0.76, 95% CI 0.64 to 0.89). The results of meta-analyses were consistent in higher-quality studies. There was no difference in overall risk of prosthetic joint infection for antibiotic-loaded cemented fixation compared with plain cement.

Results: National Joint Registry analyses

Risk factors for revision of hip replacement for prosthetic joint infection

Of 623,253 primary hip replacements carried out between 2003 and 2013 and with ≥ 1 year of follow-up [median 4.6 years, interquartile range (IQR) 2.6–7.0 years], 2705 were revised for prosthetic joint infection. The incidence rate ratios for revision are summarised by patient, surgical and health system characteristics in Appendices 6–8.

Patient characteristics

People aged ≥ 80 years were at lower risk of revision for hip prosthetic joint infection than people aged < 60 years (RR 0.66, 95% CI 0.56 to 0.76) (Figure 8). Men were at higher risk of revision of hip prosthetic joint infection than women (RR 1.68, 95% CI 1.56 to 1.81). The numbers of revisions in ethnic minority groups were small, and we were unable to investigate ethnicity as a possible risk factor. People with a BMI of ≥ 30 kg/m² had a higher risk of revision for hip prosthetic joint infection than those with a BMI of < 25 kg/m² (RR 1.92, 95% CI 1.72 to 2.15). People with diabetes, chronic pulmonary disease, congestive heart failure, connective tissue-rheumatic disease, previous septic arthritis or fractured neck of femur had a higher risk of revision for hip prosthetic joint infection. More generally, people with greater medical comorbidity had a higher risk of revision for hip prosthetic joint infection (ASA grades 3–5 vs. ASA grade 1: RR 1.63, 95% CI 1.42 to 1.87). In the first 3 months after primary hip replacement, the risk of revision for hip prosthetic joint infection was higher for people with dementia than for those without (RR 3.78, 95% CI 1.21 to 7.81). People with liver disease had an increased risk of revision for hip prosthetic joint infection but only at ≥ 2 years after primary hip replacement (RR 2.43, 95% CI 1.35 to 3.82). There were no clear associations between cardiovascular disease, cerebrovascular disease, myocardial infarction, peripheral vascular disease, cancer (except metastatic disease) and paraplegia or hemiplegia and the risk of revision for hip prosthetic joint infection.

FIGURE 8.

Patient characteristics and risk of revision for hip prosthetic joint infection during the whole postoperative period: NJR analysis. Reference category in parentheses. a, Adjusted p-value < 0.05.

Surgical factors

Compared with unaffected people, those with osteoarthritis or congenital hip dysplasia were at lower risk, while people treated for fractured neck of femur or osteonecrosis were at higher risk of revision for hip prosthetic joint infection (Figure 9). There was a particularly high risk of further infection in those who had experienced a previous hip infection (RR 6.69, 95% CI 4.18 to 9.80). The use of the lateral surgical approach was associated with an increased risk of revision for hip prosthetic joint infection compared with the posterior approach (RR 1.32, 95% CI 1.21 to 1.43). Up until 2 years after hip replacement, the risk with metal-on-metal bearing combinations was lower than or similar to that with metal-on-polyethylene, but at longer follow-up the risk was higher with metal-on-metal. Ceramic-on-ceramic and ceramic-on-polyethylene bearing combinations were associated with a lower risk of revision after 24 months than metal-on-polyethylene bearings. Little or no difference in the risk of revision for hip prosthetic joint infection was found for anaesthetic technique, thromboprophylaxis regime, the use of acetabular bone graft or intraoperative complication. Patients with a femoral bone graft during primary hip replacement were at higher risk of revision for hip prosthetic joint infection.

FIGURE 9.

Surgical factors and risk of revision for hip prosthetic joint infection during the whole postoperative period: NJR analysis. Reference category in parentheses. a, Adjusted p-value < 0.05. CoC, ceramic-on-ceramic; CoM, ceramic-on-metal; CoP, ceramic-on-polyethylene; MoM, metal-on-metal; MoP, metal-on-polyethylene; THR, total hip replacement.

Health system factors

The risk of revision for hip prosthetic joint infection was similar in Wales and England and did not differ according to funding source, operating surgeon grade or consultant surgeon presence during surgery (Figure 10). Operating surgeons who performed more hip replacements annually had a lower risk of revision for hip prosthetic joint infection than surgeons with a lower volume. The volume of all hip procedures carried out by the surgeon in charge of the surgery did not affect the risk of revision. The risk of revision for hip prosthetic joint infection was higher in the first 3 months after primary surgery in hospitals that had performed over 255 hip procedures in the year before primary surgery than in hospitals with a small volume of activity. No specific difference in the rate ratios were found beyond this period or for units with lower volumes of hip procedures.

FIGURE 10.

Health system factors and risk of revision for hip prosthetic joint infection during the whole postoperative period: NJR analysis. Reference category in parentheses. a, Volume is the total number of hip replacements performed in the previous 12 months; b, adjusted p-value < 0.05.

Risk factors for revision of knee replacement for prosthetic joint infection

Between 2003 and 2015, 679,010 primary knee replacements were recorded in the NJR, of which 3659 were revised for prosthetic joint infection at ≥ 1 year follow-up (median 4.6 years, IQR 2.6–6.9 years). After primary knee replacement, incidence rate ratios of revision for prosthetic joint infection are summarised for different patient, surgical and health system characteristics in Appendices 9–11.

Patient characteristics

People aged ≥ 80 years were less likely to have a revision for knee prosthetic joint infection than those aged < 60 years (RR 0.50, 95% CI 0.43 to 0.57) (Figure 11). Men were more likely than women to have a revision for knee prosthetic joint infection (RR 1.83, 95% CI 1.71 to 1.96). There were too few revisions for knee prosthetic joint infection in ethnic minority groups to investigate ethnicity as a possible risk factor. The likelihood of revision for knee prosthetic joint infection was higher in people with a BMI of ≥ 30 kg/m² than in those with a BMI of < 25 kg/m² (RR 1.46, 95% CI 1.29 to 1.63). The risk of revision for knee prosthetic joint infection was higher in people with more medical comorbidities, and specifically in people with diabetes, chronic pulmonary disease, connective tissue and rheumatic diseases and peripheral vascular disease. An elevated risk in people with liver disease was specific to the period after 2 years.

FIGURE 11.

Patient characteristics and risk of revision for knee prosthetic joint infection during the whole postoperative period: NJR analysis. a, Adjusted p-value < 0.05.

Surgical factors

Aspects of surgery associated with an increased risk of revision for prosthetic joint infection after knee replacement were surgery for trauma, previous knee infection, inflammatory arthropathy, operation under general anaesthesia, requirement for tibial bone graft and use of posterior stabilised fixed-bearing prostheses compared with unconstrained fixed-bearing prostheses and constrained condylar prostheses (Figure 12). Compared with cemented total knee replacement, lower rates of prosthetic joint infection were seen in people who received uncemented, unicondylar or patellofemoral knee replacement.

FIGURE 12.

Surgical factors and risk of revision for knee prosthetic joint infection during the whole postoperative period: NJR analysis. a, Adjusted p-value < 0.05.

Health system factors

The risk of revision for prosthetic joint infection after knee replacement was similar in Wales and England (Figure 13). Privately funded procedures had a lower risk of revision than procedures funded by the NHS. Revision for knee prosthetic joint infection was not affected by the operating surgeon’s grade, the presence of a consultant surgeon during surgery, or the volume of knee procedures carried out by the operating surgeon or the surgeon in charge. The risk of revision for knee prosthetic joint infection was higher in high-volume hospitals than in low-volume hospitals. In hospitals that had carried out more than 440 knee procedures in the year preceding the index surgery, the risk of revision for prosthetic joint infection was higher in the first 3 months after surgery than in hospitals with a small volume of activity.

FIGURE 13.

Health service factors and risk of revision for knee prosthetic joint infection during the whole postoperative period: NJR analysis. a, Adjusted p-value < 0.05; b, volume is the total number of knee replacements performed in the previous 12 months.

Discussion

The two research approaches complemented each other as cohort studies included in systematic reviews and the NJR reported associations between a diverse range of potential risk factors and risk of prosthetic joint infection and need for revision.

Risk of hip and knee prosthetic joint infection and need for revision is higher in men, people with a high BMI, people with diabetes and those receiving implants with cemented fixation. In the NJR an association was noted between dementia and increased early risk of revision for hip prosthetic joint infection.

Meta-analysis of cohort studies showed no association with age, but in the NJR, younger patients were at higher risk of revision for hip or knee prosthetic joint infection. This may be because the NJR had a longer follow-up with more time for joint infections to occur in younger patients. Older people at high risk of adverse outcomes and death from surgery may have received non-operative management. NJR data show not the risk of infection, but the risk of revision for infection. It may be that clinicians are more likely to treat older patients with suppressive antibiotics than with surgery. In NJR analyses, people with more medical comorbidities had an increased risk of revision for hip and knee prosthetic joint infection, and this was noted for specific conditions in meta-analyses and NJR analysis.

Analyses provide reassurance that many surgical and health system factors do not increase the risk of prosthetic joint infection. There was a suggestion that the posterior approach and use of ceramic-on-ceramic and ceramic-on-plastic bearings had a lower risk of infection after hip replacement.

Limitations

In meta-analyses, definitions of prosthetic joint infection and the adjustment for confounding factors differed between studies. The results of meta-analyses were consistent in higher-quality studies and we noted no differences in studies reporting univariable or multivariable analyses.

The NJR analyses relied on subjective diagnosis of prosthetic joint infection by individual surgeons. As with our meta-analyses, there may have been residual confounding factors and associations may vary with different causative pathogens. For example, older people may be at higher risk of infection with organisms that are difficult to treat. 77

With observational data such as those collected in the NJR and cohort studies included in our systematic reviews, we cannot establish for certain whether or not the relationships between risk factors and revision for prosthetic joint infection are causal.

Aims

We aimed to assess the diagnostic accuracy of promising synovial biomarkers for prosthetic joint infection: the alpha-defensin immunoassay and leucocyte esterase colorimetric strip test.

Methods

The systematic review was registered with PROSPERO (CRD42015023704) and conducted in accordance with PRISMA and diagnostic test accuracy guidelines. 68,79 We searched MEDLINE and EMBASE on Ovid from inception until 30 May 2015 (see Appendix 12).

Inclusion criteria were:

• people with hip or knee replacement suspected of having a prosthetic joint infection but with true diagnostic uncertainty

• measurements of synovial fluid alpha-defensin or leucocyte esterase

• prosthetic joint infection according to Musculoskeletal Infection Society (MSIS) diagnostic criteria or confirmation during subsequent surgery

• reports of diagnostic test accuracy.

Screening and data extraction were performed independently by two reviewers. Extracted data related to patient characteristics, hip or knee joint, diagnostic test, test cut-off points, reference standards and results (sensitivity, specificity and likelihood ratios). Sensitivity and specificity values from each evaluation were pooled using a bivariate meta-analysis framework. 80 A quality assessment of each study was performed using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool. 81

Results

We included 11 eligible studies. The mean QUADAS-2 score was 13 out of 14 (range 11–14), suggesting that the studies were of good quality. The pooled diagnostic sensitivity and specificity of synovial fluid alpha-defensin (six studies) for prosthetic joint infection were 1.00 (95% CI 0.82 to 1.00) and 0.96 (95% CI 0.89 to 0.99), respectively. The area under the curve (AUC) was 0.99 (95% CI 0.98 to 1.00).

The pooled diagnostic sensitivity and specificity of leucocyte esterase (five studies) for prosthetic joint infection were 0.81 (95% CI 0.49 to 0.95) and 0.97 (95% CI 0.82 to 0.99), respectively. The AUC was 0.97 (95% CI 0.95 to 0.98).

There was substantial heterogeneity among studies for both diagnostic tests. No studies reported cost-effectiveness.

Discussion

Synovial fluid alpha-defensin and leucocyte esterase showed high diagnostic accuracy for prosthetic joint infection. Alpha-defensin was extremely sensitive and specific in the identification of prosthetic joint infection. Leucocyte esterase was slightly less sensitive but was extremely specific in the identification of prosthetic joint infection.

A systematic review published in 2019 considered a broad range potential serum, synovial and tissue diagnostic tests for prosthetic joint infection. 82 Of 83 tests identified, 17 had enough data from studies to allow meta-analysis. The authors reached similar conclusions to ours in identifying synovial fluid alpha-defensin and leucocyte esterase as the best-performing diagnostic tests for prosthetic joint infection.

Cost implications of the tests vary considerably. In a UK report, the lateral flow tests for alpha-defensin cost £300 each when purchased in a pack of five. 83 The laboratory-based immunoassay costed within a suite of tests was £450. This compares with a cost of £0.11 for a leucocyte esterase test. 78

Since the publication of our study, the MSIS criteria have reported a new definition of prosthetic joint infection citing our systematic review (Figure 14). 84 These include a definition that a hip or knee is infected if synovial leucocyte esterase is elevated and alpha-defensin is positive.

FIGURE 14.

The 2018 MSIS criteria for prosthetic joint infection. a, For patients with inconclusive minor criteria, operative criteria can also be used to fulfill definition for PJI; b, consider further molecular diagnostics such as next-generation sequencing. Note: proceed with caution in adverse local tissue reaction, crystal deposition disease and slow-growing organisms. CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; LE, leukocyte esterase; PMN, polymorphonuclear; WBC, white blood cell. Reprinted from The Journal of Arthroplasty, Vol. 33, Parvizi J, Tan TL, Goswami K, Higuera C, Della Valle C, Chen AF, Shohat N. The 2018 definition of periprosthetic hip and knee infection: an evidence-based and validated criteria. pp. 1309–14. Copyright 2018, with permission from Elsevier.

Reprinted from The Journal of Arthroplasty, Vol. 33, Parvizi J, Tan TL, Goswami K, Higuera C, Della Valle C, Chen AF, Shohat N. The 2018 definition of periprosthetic hip and knee infection: an evidence-based and validated criteria. pp. 1309–14. Copyright 2018, with permission from Elsevier.

Limitations

Our study was limited by the small number of studies in the systematic review. Thus, we were not able to assess publication bias, and four out of six studies of alpha-defensin were conducted by a research group who held patents relating to the test.

Aims

In systematic reviews and meta-analyses, we aimed to compare reinfection rates after single- and two-stage treatments of hip and knee prosthetic joint infection. In an IPD meta-analysis we aimed to compare reinfection outcomes after single- and two-stage treatment of hip prosthetic joint infection. In a meta-analysis we aimed to assess the risk of reinfection after treatment of prosthetic joint infection with DAIR.

Analyses of linked NJR/ONS data aimed to compare rates of reinfection and mortality for people with prosthetic joint infection treated with single- or two-stage revision.

Methods: systematic reviews

The systematic reviews were registered prospectively with PROSPERO (CRD42017057513, CRD42015016559 and CRD42015017327) and conducted according to PRISMA68 and MOOSE69 guidelines. We searched MEDLINE and EMBASE on Ovid and Web of Science from inception (see Appendices 14–16), and reference lists.

Eligible studies included:

• unselected patients with hip or knee prosthetic joint infection

• treatment with DAIR, or single- or two-stage revision

• reinfection or new infection outcome within 2 years of single- or two-stage revision, or any time after DAIR

• longitudinal design.

We excluded studies that reported case series of selected patients, for example those with a specific infection; studies with fewer than 10 patients; and, for studies of single- and two-stage revision, follow-up of < 2 years.

Searches were carried out on 1 May 2017 (DAIR), 1 March 2015 (hip single- or two-stage revision) and 1 August 2015 (knee single- or two-stage revision).

Methodological quality was assessed using MINORS (methodological index for non-randomized studies). 89 The rate of reinfection within 2 years was the primary outcome across studies. Pooled rates were calculated using the Freeman–Tukey variance-stabilising double arcsine transformation. 90 Publication bias was assessed using Egger’s regression symmetry tests. 91

Methods: individual patient data meta-analysis

The IPD meta-analysis protocol was registered on PROSPERO (CRD42015016664) and conducted as recommended by the Cochrane Collaboration,92 the guidance of Riley et al. 93 and the PRISMA-IPD guidelines. 94 The study was limited to studies reporting outcomes after single- and two-stage revision for hip prosthetic joint infection. Searches86 were updated in August 2016. Authors of eligible studies and well-known investigators in the field were contacted, provided with the study protocol and invited to join the Global INFORM collaboration. 95 Collaborators were provided with standardised spreadsheets that were combined into a single database. Data in published articles were also included.

The risks of reinfection were compared using Cox proportional shared frailty models. 96 Hazard ratios (HRs) with 95% CIs were calculated, with progressive adjustment for age, sex, Charlson Comorbidity Index, previous hip surgery and type of infecting organism (‘difficult to treat’ vs. ‘not difficult to treat’).

Methods: National Joint Registry analysis

As described in Appendix 13, we analysed NJR data for England and Wales collected between 2003 and 2014 linked with ONS data. We included patients with a primary hip or knee replacement that had subsequently been revised for prosthetic joint infection with either a single- or a two-stage procedure. Reinfection and mortality outcomes were compared between treatment of hip and knee prosthetic joint infection using single- and two-stage revision strategies, and for mortality also with primary hip and knee replacements, and aseptic revisions. Analyses were adjusted for age, sex and ASA grade.

Results: systematic reviews

Results: individual patient data meta-analysis

Eighty-five articles were identified reporting 98 cohorts of patients with hip prosthetic joint infection published up to August 2016. Contact with authors of studies led to the Global INFORM collaboration, comprising 15 cohorts (n = 1383 patients) with author-provided data. Data from a further 29 studies (n = 473 patients) were extracted from published articles.

After a median follow-up of 3.7 years, 222 reinfections were recorded. Reinfection rates per 1000 person-years of follow-up were 16.8 (95% CI 13.6 to 20.7) and 32.3 (95% CI 27.3 to 38.3) for single- and two-stage procedures, respectively. Among individuals with available data, the age-adjusted HR for reinfection comparing two- with single-stage revision was 1.69 (95% CI 0.58 to 4.98; p = 0.338). Progressive adjustment for sex, previous hip surgery and comorbidities did not change the outcome.

Results: National Joint Registry analysis

Conclusion

DAIR may be effective for > 60% of patients, particularly if it is performed early. An analysis of the NJR showed a higher risk of re-revision following single-stage revision for hip prosthetic joint infection than following two-stage revision, but this was limited to the 3 months after revision surgery. Patients who received a a single-stage revision had a smaller number of operations. When considered alongside the results of meta-analyses, the single-stage revision strategy for hip prosthetic joint infection is a reasonable option, with acceptable rates of infection control. Re-infection rates after single- and two-stage revision for knee prosthetic joint infection were similar, and patients who received single-stage revision underwent considerably fewer operations.

Limitations

An important limitation is that people may have been selected for a revision strategy depending on patient factors, including health status. An IPD meta-analysis was limited by a lack of available data for clinically relevant subgroups, including BMI and outcomes such as early or late infection. Our comparison of outcomes in cohorts of knee prosthetic joint infection was limited because few studies have reported outcomes after single-stage revision. Few studies collected patient-reported outcomes.

In the NJR studies there were small differences in patient characteristics between groups, which may indicate a degree of selection bias, and this may differ between the hip and knee cohorts. Furthermore, there is no linkage with microbiological data, and patients may have received a revision strategy based on the virulence of the infecting organism.

Aims

Our aim was to determine whether or not there is a difference in patient-reported outcome 18 months after randomisation to single- or two-stage revision surgery for the treatment of hip prosthetic joint infection. Eighteen months was chosen because the median interval between the stages of a two-stage revision is about 105 days (IQR 70–173 days), and, therefore, most people receiving a two-stage surgery should have achieved maximum recovery. A longer follow-up may have limited recruitment and patient retention. Although the trial was not powered to compare rates of reinfection or occurrence of a new infection, adverse events were monitored.

Methods

INFORM was a multicentre, two-arm, parallel-group, participant and observer unblinded, randomised superiority trial (ISRCTN Registry number ISRCTN10956306). Full details are in the published protocol97 and at https://doi.org/10.5523/bris.2323q1mtahy6y2dy5ua3opivqf. Patients deemed by their treating surgeons and multidisciplinary prosthetic joint infection teams to have infected hip replacements were randomised to either a single- or a two-stage revision.

Results

A CONSORT flow diagram is shown in Figure 15. Of 186 patients eligible for inclusion in the trial, 65 were randomised to single-stage revision and 75 were randomised to two-stage revision. The baseline characteristics of participants in the randomised groups are shown in Appendix 17, Table 13, and these were balanced between the groups, except that patients in the single-stage group were more likely to be male, with ASA grade 1 or 2 and have received previous non-surgical infection management. The groups were similar in the number and type of organisms cultured, the rates of culture-negative infection and the presence of a sinus tract. A total of 126 (90%) patients had the primary total WOMAC score outcome at 18 months post randomisation and 133 (95%) patients had at least one postoperative total WOMAC score and were included in the main analysis.

FIGURE 15.

The INFORM RCT CONSORT flow diagram. PJI, prosthetic joint infection.

There was no evidence of a difference in the mean total WOMAC score at 18 months post randomisation between single- and two-stage management of hip prosthetic joint infection (mean difference favouring two-stage 0.13, 95% CI –8.20 to 8.46; p = 0.98) (Figure 16).

FIGURE 16.

Mean (95% CI) global WOMAC score by revision surgery for prosthetic hip infection groups. A linear mixed regression model adjusted for place of surgery and accounting for participant repeated measures was used to produce the mean score at each time point.

At 3 months post randomisation, participants in the single-stage procedure group had a better total WOMAC score (mean difference favouring single-stage 11.53, 95% CI 3.89 to 19.17; p = 0.003). This is greater than the minimal clinically important improvement of 7 (95% CI 4 to 10) reported by Bellamy et al. 100 in a before-and-after study in a general osteoarthritis population. Sensitivity analyses, with imputation of the 14 missing primary outcome assessments and accommodation of the possible range of WOMAC scores, adjusted for further baseline variables and imputation plus adjustment, and limited to centres where a majority of two-stage procedures did not use a CUMARS, supported the conclusions of the main analysis.

From 6 months post randomisation onwards, no difference was found between the two surgical procedures. Similar findings were found for all patient outcomes.

A higher rate of complications of surgery was seen among patients receiving a two-stage revision than among those receiving a single-stage revision (57.3% vs. 41.5%; p = 0.04) and this difference was marked for intraoperative events (26.7% vs. 7.7%; p = 0.01), predominantly calcar cracks and femoral shaft fractures (18 vs. 3). There was no evidence of a difference between the groups in length of hospital stay; other complications; rehospitalisation or reoperation due to prosthetic joint infection, the surgery or another cause; or serious adverse events or deaths (5 vs. 2). At 15–18 months post randomisation, 9 out of 59 (15%) patients who received single-stage revision had signs of infection in the operated hip, compared with 8 out of 67 (12%) who received two-stage revision (p = 0.59).

Conclusion

As shown by the similarity in WOMAC total scores at 18 months after randomisation, the INFORM trial demonstrated that single- and two-stage revision are equally efficacious in the treatment of hip prosthetic joint infection. There was, however, an early improvement in WOMAC total scores in the single- compared with the two-stage group, which was clinically important. Given that patients state a strong preference for a quick return to normal activity, this should be considered when selecting the intervention. Further research should explore the value of earlier improvement to patients and health care. Single-stage surgery accounted for 30% of revision procedures for the treatment of prosthetic joint infection in 2014, so there is scope for increased utilisation.

Limitations

The INFORM study was powered for a patient-reported outcome and the sample size precluded study of infection clearance. The numbers of patients in the sensitivity analysis were large, except those restricted to centres where a majority of two-stage procedures did not use a CUMARS, which was limited to 90 patients.

Aim

To compare the cost-effectiveness of single- and two-stage revision surgery in the treatment of hip prosthetic joint infection.

Methods

The primary economic analysis was from an NHS and Personal Social Services (PSS) perspective, with a secondary analysis from a societal perspective. Resource use data relating to the initial surgery and subsequent treatment for revision of hip prosthetic joint infection were collected by research staff at each hospital at 3, 6, 9, 12, 15 and 18 months post randomisation. Admissions for revision surgery were microcosted and valued using information from the finance department of a treating hospital and the NJR. All subsequent inpatient stays, day cases and outpatient procedures were allocated an appropriate HRG code. NHS reference costs were used to value these stays, subsequent outpatient appointments and emergency department attendances. A self-reported resource use questionnaire was posted to participants at 6, 12 and 18 months post randomisation to collect information relating to community-based health-care use, prescribed medications and emergency department attendances; PSS provision received; personal expenses, including travel and home changes; and time off work, usual activities and informal care. Resources were valued using 2018–19 UK reference costs. 101–103

The economic evaluation outcome measure was the quality-adjusted life-year (QALY), derived from the EuroQol-5 Dimensions, five-level version (EQ-5D-5L) questionnaire at baseline and at 3, 6, 12 and 18 months post randomisation. These were converted to utility scores using the validated mapping function to the existing EuroQol-5 Dimensions, three-level version (EQ-5D-3L) societal UK utility tariffs. 104 Total QALYs were calculated by linear interpolation using the AUC approach, taking into account any deaths that occurred during the study period by allocating those who died a zero utility in the time periods following their death. 105 A 3.5% discount rate was applied to all costs and outcomes occurring after 1 year.

Multiple imputation by chained equations using predictive mean matching was used to address missing data. Simple imputation methods assuming zero resource use in a returned participant questionnaire when no responses were given for an item, and using mean imputation where a participant had indicated the use of a resource but did not provide further details, were used prior to multiple imputation. Missing values for baseline utility were also imputed using the mean baseline utility value106 prior to multiple imputation. The covariates in the imputation model were baseline utility, trial group, age, sex, ethnicity, education status, work status, hospital, ASA grade at first revision surgery and the latest surgical treatment for the management of prosthetic joint infection prior to study participation. Rubin’s rules were used to combine the 62 individual imputations, and a randomisation seed was used to enable reproducible imputations. 99

The adjusted mean costs and QALYs by trial group, the differences in adjusted mean costs and QALYs and the incremental net monetary benefit (iNMB) were estimated using the seemingly unrelated regression method, which accounts for the correlation between costs and QALYs. 107 Costs and QALYs were adjusted for study centre. QALYs were also adjusted for baseline utility. National Institute for Health and Care Excellence (NICE) willingness-to-pay thresholds of £20,000 and £30,000 per QALY were used. Sample uncertainty within the cost-effectiveness estimates was explored using cost-effectiveness acceptability curves (CEACs) derived from the estimates of the seemingly unrelated regression. Sensitivity analyses (see Appendix 18, Table 20) were undertaken to account for uncertainty in the parameters used in the cost-effectiveness analyses.

Results

Owing to differences in the collection of resource use data for UK and Swedish participants, the primary economic analysis was conducted on 128 participants from England and Wales using a multiply imputed data set.

Total theatre time was longer for those randomised to two-stage revision, and these patients spent longer in intensive care and recovery and had a longer overall hospital stay. They also had a greater number of subsequent inpatient stays and emergency department attendances. There were a greater number of primary care practice nurse visits for those randomised to a single-stage procedure, whereas those randomised to a two-stage procedure had a greater number of district nurse home visits, stays in residential homes and home care worker visits.

The EQ-5D-5L utility values at all time points are shown in Table 5. These illustrate the different pathways of the two procedures, with patients randomised to the single-stage procedure seeing a gradual improvement in their utility values from 3 months onwards, whereas for those randomised to the two-stage procedure the improvement begins only at 6 months.

Table 6 shows that the mean costs from the treating hospital were higher in the first year of follow-up in the single-stage group; however, in the last 6 months of the trial these costs were similar in the trial groups, indicating that the trial follow-up is sufficiently long to capture differences in costs.

The total adjusted mean costs in the single-stage group (£36,256) from the NHS/PSS perspective were lower than in the two-stage group (£46,312), a cost difference of –£10,055 (95% CI –£19,568 to –£542). The cost difference reduced slightly from the societal perspective (–£9450, 95% CI –£22,855 to £3956). Participants in the single-stage group had a greater number of adjusted mean QALYs (0.75) than those in the two-stage group (0.69), a difference of 0.06 (95% CI –0.07 to 0.18).

From the NHS/PSS perspective, the iNMB of single-stage revision compared with two-stage revision was £11,167 (95% CI £638 to £21,696) at £20,000 per QALY willingness-to-pay threshold, and at this threshold there was a 98% probability that the single-stage procedure was the cost-effective option. This reduced to 92% from the societal perspective. All sensitivity analyses conducted from the NHS/PSS perspective showed that the probability of the single-stage procedure being cost-effective was at least 98%.

Conclusion

The two hospital stays in a two-stage procedure led to the higher cost in this group. The EQ-5D-5L scores illustrate that while patients were waiting for their second-stage operation they had a poorer quality of life, which was reflected in the overall QALY score. The greater use of district nurse home visits and home care worker visits indicates that these patients were also less able to self-care and leave their home.

The within-trial economic evaluation has shown, with a great degree of certainty, that the single-stage procedure is the cost-effective option for patients with hip prosthetic joint infection.

Limitations

The analysis has limitations in relation to missing data, particularly for resource use as obtained from patient-completed questionnaires. A complete-case analysis could only be conducted from a treating hospital perspective, and indicated higher QALY values in both groups, indicating that those who did not complete all the questionnaires were likely to have poorer quality of life than those who did. Multiple imputation was used to account for the missing data, using data at each time point in the model, meaning that all available data were used. This meant that the multiple imputation could not be conducted by trial group. However, sensitivity analyses on different specifications of the multiple imputation model indicate that the model specification is unlikely to have affected the conclusions of the analysis.

Cost differences may extend beyond the 18-month trial follow-up. However, the randomised groups had similar hospital-based costs accruing between 12 and 18 months’ follow-up and had similar mean WOMAC and EQ-5D-5L scores at 18 months. There was also little to suggest a difference in reinfection rates between groups. Together, these suggest that patient outcomes and hospital costs may not differ in the longer term.

Aims

We aimed to explore surgeons’ experience and acceptance of the randomisation process, their understanding of equipoise, and their views on information about the trial and interventions.

Methods

We conducted telephone interviews with 12 surgeons from eight sites participating in the trial. We used a topic guide developed in collaboration with INFORM surgeons. Topics included surgeons’ experiences and acceptability of the randomisation process and their views on surgical equipoise, and information about the trial and interventions. Data were analysed using a descriptive qualitative approach and reported to the trial management group to inform the recruitment process and identify areas for improvement.

Results

Surgeons were happy with the information they received about the trial and the randomisation process. We identified a number of potential barriers to recruitment and areas for improvement. The complexity of cases often precluded patients from randomisation, and often this was because the surgeon lacked equipoise in that particular case. Some surgeons suggested that there was a lack of engagement at their centres and that their colleagues lacked equipoise about the trial. Recruitment of patients was difficult as prosthetic joint infection is a rare condition and we did not have the resources and staff to find eligible patients opportunistically at routine outpatient or emergency appointments. At one centre, a change in referral patterns meant that the number of patients being referred had reduced. We addressed these barriers in the following ways:

• We originally planned to recruit patients from four centres but increased this number in the early stages of the trial when we realised that recruitment would be a challenge. The funding of additional sites was met from the existing budget and with UK Clinical Research Network portfolio support.

• We continued to collect data on patients deemed ineligible.

• We communicated anonymised case studies from two recruiting centres to surgeons in the other centres to demonstrate the range of patients randomised and the characteristics of individual cases. This resulted in debate and dialogue between surgeons regarding equipoise.

• We communicated directly with all surgeons operating on patients with prosthetic joint infection within collaborating trusts to ensure that they remained engaged.

• We communicated findings of other work packages to surgeons through newsletters to enhance understanding of the broad context of INFORM.

• We held workshops with recruiting nurses to provide them with training and support on ways to identify and recruit patients and to ensure that potential participants were brought to the attention of participating surgeons.

• The chief investigator and the clinical lead of work package 4 hosted information events at national conferences and visited collaborating centres to explore local service and recruitment pathways to ensure that eligible patients were identified and that surgeons remained engaged.

Conclusion

Data from interviews with surgeons in the early phase of the trial enabled us to identify barriers to recruitment and implement remedial strategies to improve recruitment.

Aims

We aimed to explore the early and long-term experiences of patients receiving single- and two-stage revision. In patients with two-stage revision we also aimed to explore experiences of patients with either a temporary cement spacer or CUMARS.

Methods

On two occasions, we interviewed 32 people participating in the INFORM RCT who had received single- or two-stage revision surgery for hip prosthetic joint infection. Patients were interviewed at 2–4 months and 18 months after their first revision operation.

Interview topic guides were developed with the INFORM PPI group but were flexible to allow participants to discuss areas that they felt were important. Themes were generated from the data inductively using the constant comparative method.

Results

Thirty people were interviewed 2–4 months after their first revision operation, and 17 were interviewed at the 18-month follow-up. The overall sample of 32 patients comprised 15 women and 17 men and had a mean age of 68.9 years (range 51 to 89 years). Eleven had received a single-stage and 21 had received a two-stage revision. Of those who had received a two-stage revision, seven had a cement spacer, 10 had a CUMARS and four had an excised hip with no spacer.

During their early recovery period, people who had received single- or two-stage revision for hip prosthetic joint infection experienced prolonged hospital stays of 2–4 weeks, difficult antibiotic regimens and only brief physiotherapy. People described the practical challenges and gains that they had experienced during day-to-day life while recovering from revision at home and identified their needs for information and support during recovery. At 18 months post revision, all patients described improvements in mobility and independence, but they also reported ongoing restrictions to walking and some functional limitations. People also described their emotional resilience and benefits from participating in the INFORM RCT.

The experiences of people who received a two-stage revision with a cement spacer or excised hip differed from those receiving single-stage revision or a two-stage revision with a CUMARS. Patients who received a single-stage revision or a two-stage revision with a CUMARS had greater mobility and independence in the early stages of recovery. However, patients who received a cement spacer or CUMARS perceived that their recovery was slow, and those with a CUMARS reported dislocations and uncertainty about the need for further surgery.

Conclusion

After treatment for prosthetic joint infection, important aspects of patient recovery relate to mobility, function, pain and independence. Patients who receive single- and two-stage revision experience different patterns of recovery and these are also influenced by use of a spacer and CUMARS. To be supported during treatment and recovery, patients require appropriate information and opportunities to talk and share experiences.

Limitations

The comparison of experiences of recovery at both the postoperative and the 18-month time points in the same patients was limited as we were not able to reinterview all 30 of the original participants. We interviewed other participants at 18 months to balance the characteristics of the sample and to ensure saturation. The participants who took part in the qualitative study may have been those who were most positive about trial participation.

Aims

Single- and two-stage revision options for treatment of prosthetic joint infection each have implications for the patient time course and experience of recovery. We aimed to assess the surgical preferences of patients who had received revision surgery for hip prosthetic joint infection.

Methods

We undertook a discrete choice experiment to quantify the surgical preferences of patients who had received a single- or two-stage revision for hip prosthetic joint infection. In a discrete choice experiment, participants choose between features of interventions or approaches to care. These are attributes that can have different levels. A checklist based on the guidance of Lancsar and Louviere110 is provided in Appendix 19.

Results

Questionnaires were sent to 80 patients at nine orthopaedic centres, and 57 were fully completed (71%). The mean age of patients was 70 years (range 51–90 years); 21 (37%) were female, 26 (46%) had received single-stage revision, 14 (25%) lived alone and 41 (72%) were retired. The questionnaire took between 20 and 40 minutes to complete.

Regression coefficients and results are shown in Table 8. Participants had the strongest preference for a surgical option that resulted in the least restrictions on engagement in valued activities after the new hip was fitted, illustrated by the largest preference weight. Less valued but important preferences were for a surgical treatment that would result in a shorter time to return to normal activities, few or no side effects from antibiotics, and only one operation. The results also suggest that the least restrictions on engaging in valued activities and the shortest time taken to return to normal activity are the individual attributes most valued by patients. This is indicated by the larger spread of coefficients. The most acceptable option was a time period of between 3 and 6 months to return to normal activity. There was no clear preference up to 12 months, but 18 months was considered unacceptable.

Conclusion

The most valued factors in decisions about revision surgery for hip prosthetic joint infection were the ability to engage in valued activities and the time taken to return to normal activity.

Limitations

Feedback from early participants suggested that completing the questionnaire was difficult. However, after the instructions and format were altered and when nurse support was provided, participants were able to understand and complete the questionnaire. Our chosen analysis method of conditional logistic regression assumes preference homogeneity and it may be that participants exhibit a level of heterogeneity that cannot be captured using this method. The use of latent class analysis could further explore the patterns in choice data that suggest preference heterogeneity, but our sample size was too small for these explorations to be conducted.

Publications

All our research is published or submitted, or analyses are in progress. Several articles have become highly cited in their area, including those relating to risk factors, the patient experience, diagnostic tests and treatments.

British Hip Society

The significance of the INFORM programme is recognised by the British Hip Society and an afternoon symposium was given over to presentation and discussion of results at the 2020 annual meeting in South Wales (see Appendix 21).

International Consensus Meeting on Periprosthetic Joint Infection

To address the wide international variation in prevention and treatment of prosthetic joint infection, the International Consensus Meeting on Periprosthetic Joint Infection was organised by Thorsten Gehrke and Javad Parvizi. The meeting took place in July 2018 and the INFORM programme was represented by Setor Kunutsor.

Conferences

The results from the INFORM programme have been presented at European Federation of National Associations of Orthopaedics and Traumatology (EFORT), International Society of Arthroplasty Registries (ISAR), European Orthopaedic Research Society (EORS), British Sociological Association (BSA), Combined Orthopaedic Associations (COMOC), British Pain Society (BPS), British Society for Rheumatology (BSR), British Orthopaedic Association (BOA) and British Hip Society (BHS) conferences.

M Shed

A 1-day conference was planned for May 2020 to present results to patient-partners, INFORM trial participants and representatives of recruitment centres. This will be rescheduled when feasible.

Overview

In the INFORM programme, we critically reviewed existing research and conducted new high-quality studies with the aim of improving treatment and outcomes for people with hip and knee prosthetic joint infection. Work packages either informed each other or triangulated results. Our studies were supported with extensive PPI.

Our NJR analyses showed that, in the UK, rates of revision surgery within 10 years of primary hip and knee replacement are 0.62% and 0.75%, respectively, but are about four times higher after aseptic revision. Further NJR analyses and meta-analyses of previous research identified multiple modifiable and non-modifiable patient and surgical risk factors, potentially allowing for better counselling of patients and the suggestion of preventative strategies.

Qualitative studies highlighted that infection and revision treatments have a devastating and long-lasting impact on patients and their families, who report that more psychosocial support is required to address unmet needs. Patients and surgeons described prolonged treatments and periods of uncertainty. Important attributes of a patient being considered for revision surgery for hip prosthetic joint infection were identified. In surgeon interviews specifically focusing on the INFORM trial, barriers to recruitment were identified and remedial strategies implemented.

The use of DAIR is effective in > 60% of cases and is more efficacious if performed early. Surgeons should be encouraged to act promptly when infection has occurred.

The INFORM randomised trial shows that revision for hip prosthetic joint infection is equally efficacious with a single- and a two-stage procedure, and good long-term patient outcomes are achieved in a wide range of patients. Two-stage revision is the more widely used treatment, but the use of single-stage revision has increased, with about 30% of revisions for hip and 19% for knee prosthetic joint infection having been performed in single operations in 2014. Single-stage revision gives better patient-reported outcomes in the first 3 months, but not thereafter.

In the INFORM trial, single-stage surgery was cost-effective in terms of QALYs compared with two-stage surgery for treatment of hip prosthetic joint infection, but infection is very expensive to treat.

Any greater risk of reinfection or new infection after a single-stage procedure than after a two-stage procedure seen in NJR analyses may be limited to the first few months after surgery. In this period, patients receiving a two-stage revision may also have infections that have not fully cleared, leading to delay in the second stage reimplantation. Thus, as with a two-stage revision, close monitoring of patients with a single-stage revision is warranted. Patients with a single-stage revision for treatment of hip prosthetic joint infection have a smaller number of operations overall. In meta-analysis and IPD meta-analysis, the single-stage revision strategy for hip prosthetic joint infection is a reasonable option with acceptable rates of infection control. Reinfection rates after single- and two-stage revision for knee prosthetic joint infection were similar and those patients receiving single-stage revision received considerably fewer operations.

Complementing the results of the INFORM trial, responses to the discrete choice experiment suggested that patients identify the ability to engage in valued activities and the time taken to return to normal activity as the most valued characteristics in decisions about revision surgery. Other preferences are less severe side effects from antibiotics and revision with only one operation. Recognising the different outcome trajectories of single- and two-stage revision, the choice of intervention should be based on the patient’s preference for the recovery profile that is important to them.

Limitations

Our research studies have limitations that may affect their generalisability.

Research recommendations

• In collaboration with patients and health-care professionals, develop clear information for people receiving treatments for prosthetic joint infection.

• Develop, implement and evaluate enhanced care pathways for people with hip and knee prosthetic joint infection.

• Develop counselling, peer support and supportive interventions in the revision surgery pathway and improve physiotherapy provision for patients with prosthetic joint infection.

• Explore whether or not patient education and supportive care can lead to earlier recognition of signs and symptoms of infections.

• Future work could look at preparedness for adverse outcomes, help-seeking in impactful situations, and information for health-care professionals about early signs and care for prosthetic joint infection.

• Develop targeted preventative strategies for high-risk patients, for example better control of comorbidities such as diabetes and effective weight loss strategies in patients with very high BMI.

• Explore the effectiveness of counselling, monitoring and targeted preventative strategies based on risk factors for prosthetic joint infection.

• Use of the CUMARS has increased and research into their long-term survival is required.

• The role of spacers in two-stage revisions needs to be appraised in view of the high prevalence of complications associated with their use.

• The relative efficacy and cost-effectiveness of single- compared with two-stage revision surgery should be explored in certain high-risk subgroups, in particular those with difficult-to-treat organisms and culture-negative infections.

• The use of a single-stage revision strategy in knee prosthetic joint infection has increased and a randomised evaluation is needed.

• Further independent comparisons in UK representative cohorts should be made between synovial fluid alpha-defensin and leucocyte esterase, and traditional diagnostic tests to assess their relative clinical effectiveness and cost-effectiveness.

• The majority of the minor criteria in the MSIS diagnostic criteria for prosthetic joint infection are not routinely available or used in the diagnosis of prosthetic joint infection in the NHS setting, and a set of diagnostic criteria relevant to contemporary NHS practice needs to be established and assessed.

Reflections on work packages

The involvement of our PPI group was valuable in all work packages. The group gave support to the INFORM RCT, which helped with the recruitment, retention and engagement of patients. The model of taking research to the patients was welcomed by patients, who found that inviting clinicians and researchers to meetings ‘owned’ and run by the PPI group was preferable to and less intimidating than patients attending research and management meetings. Patients felt happy to set the agenda and direct the meetings over which they had ownership. Researchers and clinicians found this refreshing and informative.

Members of the INFORM group carrying out secondary research felt that discussing the methods and results of the evidence synthesis with the PPI group helped them appreciate the significance of prosthetic joint infection and its treatment to patients. In the field of prosthetic joint infection, systematic reviews included many published studies that had small sample sizes. There is a need for more collaborative analyses. Investigators were very willing to share data for the IPD meta-analysis and we wish that we had done more IPD analysis on related topics. Further work using this approach is planned. Hopefully, in future, there will be more data from clinical trials that will benefit from meta-analysis.

Something that stands out is the value of the qualitative work in giving a small patient population a voice. The work was very important in the programme and we wish that we had done more qualitative work to further understand the experience of patients, their significant others and health-care workers treating infection. Although prosthetic joint infection affects only a small proportion of patients who undergo joint replacement, our work has shown just how devastating and life-changing its impact can be and has identified an important need for further supportive care pathways. It has also given surgeons a voice. While prosthetic joint infection has been identified as every orthopaedic surgeon’s ‘worst nightmare’,62 until now its impact on surgeons’ professional and personal well-being had not been addressed in the literature. We hope that this work in some ways helps surgeons to broach the topic with colleagues in a supportive fashion, rather than, as one surgeon suggested, ‘burying your head in the sand’.

Across all study designs, the significance of patient outcomes and not just reinfection was reinforced in interviews with patients and surgeons and meetings with the PPI group.

We enjoyed undertaking a discrete choice experiment in orthopaedics, even though it proved challenging for patients and, ultimately, quite labour intensive. The results gave novel insights into what was important to patients. In future studies, we will undertake training and methodological work to make this research easier for participants.

The NJR proved a rich resource for exploring the significance of prosthetic joint infection to the NHS. The sheer number and breadth of data afforded by NJR/HES linkage allowed an exploration of secular trends, risk factors and treatments that is not possible with other observational data sets. We are encouraged to see the establishment and development in the UK of similar large national audits for other conditions.

To guide decisions on appropriate care, the top level of evidence in the hierarchy of primary research is the RCT, which is acknowledged to be difficult when comparing surgical procedures. The INFORM multicentre trial was completed successfully and showed that equipoise and random allocation of treatments for prosthetic joint infection was acceptable to patients and surgeons. The trial was a difficult undertaking. For anyone replicating it, we recommend opening many sites early, spending time encouraging surgeons, and getting surgeons at different centres to talk to each other about cases to set their minds at ease. The use of a per-patient fee at research centres was critical in a trial focusing on a rare condition.

Challenges and successes

Having been involved in a number of NIHR Programme Grants, we are particularly pleased with the large volume of clear results of direct relevance to patients, clinicians and health-care providers that this programme has delivered. We now have new important knowledge on impact, risk factors and treatment that will inform care.

The qualitative work, PPI and evidence synthesis were particularly productive. The IPD analysis allowed us to internationalise the programme and it was pleasing to have such willing international co-operation leading to an important scientific publication. The evidence synthesis could, therefore, expand beyond the original aims and cover infection in a large range of joints.

Despite our extensive experience with NJR/HES, this work was initially more complex than we had anticipated, with lengthy data cleaning and the careful unpicking of the complex patient journey. For example, many patients do not simply received a single- or two-stage revision but have multiple first stages before the second stage. Despite this, the work package delivered excellent outputs, which are already influencing care and are highly cited.

Certainly, the most challenging aspect of the study was the randomised trial. Possibly quite predictably, we overestimated recruitment rates at some centres and had to open more centres than initially planned. This necessitated extending the programme to 6 years. Costs, however, were contained by introducing a per-patient fee with payments tied to milestones. We would encourage researchers to adopt this funding model, which incentivises centres and controls costs. It may be argued that a RCT focusing on the treatment of infection should be powered to study infection clearance. Our previous research, patient interviews and PPI indicated that patients, rather than being concerned with biomedical outcomes relating to prosthesis survival and further treatments, are concerned about pain, function and well-being after surgery. When data accrue from further RCTs, meta-analysis will allow a more thorough analysis of infection clearance after single- and two-stage revision.

In the programme, it was excellent to be able to approach important questions using a variety of methods and resources. When answers concurred, this was very reassuring.

Service developments

We have shown that infection is devastating and that patients lack the physical and psychological support that they have identified as necessary. We have identified the patients at highest risk, allowing adequate preoperative counselling and targeted measures to decease risks. We have delineated the efficacy of DAIR and shown that early intervention markedly improves the success rate of this treatment. Single- and two-stage revision appear equally efficacious, and both achieve good results; however, single-stage surgery gives better early results and is cost-effective in terms of QALYs. Surgeons can, therefore, feel confident to adopt this treatment.

Trial Steering Committee

We thank the members of the Trial Steering Committee for their support in the development, conduct and delivery of the INFORM RCT:

• Professor Rod Taylor, University of Exeter

• Martyn L Porter, Centre for Hip Surgery, Wrightington Hospital

• Ali Heawood, Bristol Medical School: Population Health Sciences

• Julie Chappell, DAC Beachcroft, Bristol.

INFORM group

The INFORM group comprises the following.

Contributions of authors

Ashley W Blom (https://orcid.org/0000-0002-9940-1095) (Professor of Orthopaedic Surgery) was chief investigator with responsibility for the delivery of the overall programme and lead for work package 4. He is the corresponding author.

Andrew D Beswick (https://orcid.org/0000-0002-7032-7514) (Research Fellow in Evidence Synthesis) was a co-applicant and work package 1 lead. He provided support to the programme systematic reviewer and was responsible for drafting this report.

Amanda Burston (https://orcid.org/0000-0003-3480-4210) (Senior Research Associate in Public and Patient Involvement) was a co-applicant and led and co-ordinated PPI in the programme.

Fran E Carroll (https://orcid.org/0000-0002-7124-7719) (Research Fellow) developed and analysed the discrete choice questionnaire in work package 6.

Kirsty Garfield (https://orcid.org/0000-0002-8301-3602) (Research Associate in Health Economics) performed the health economic analysis of the NJR in work package 2.

Rachael Gooberman-Hill (https://orcid.org/0000-0003-3353-2882) (Professor of Health and Anthropology) was a co-applicant and overall lead for qualitative studies.

Shaun Harris (https://orcid.org/0000-0001-7724-6621) (Senior Research Associate in Health Economics) performed the health economic analysis of the INFORM RCT in work package 4.

Setor K Kunutsor (https://orcid.org/0000-0002-2625-0273) (Research Fellow in Evidence Synthesis) conducted the systematic reviews and meta-analyses in work package 1. He established and led the IPD meta-analysis.

Athene Lane (https://orcid.org/0000-0002-7578-4925) (Professor of Trials Research) provided expertise in randomised trial methods to support conduct of work package 4.

Erik Lenguerrand (https://orcid.org/0000-0002-0371-731X) (Senior Research Fellow in Medical Statistics) was a co-applicant and lead for work package 2. He was lead statistician in work package 4.

Alasdair MacGowan (https://orcid.org/0000-0002-6720-5268) (Professor of Antimicrobial Therapeutics) was a co-applicant and microbiology lead for the programme.

Charlotte Mallon (https://orcid.org/0000-0003-0179-0131) (Research Associate in Qualitative Research) carried out qualitative studies in work package 3.

Andrew J Moore (https://orcid.org/0000-0003-3185-1599) (Research Fellow in Qualitative Health Research) was qualitative lead for work packages 3 and 4 and provided support for work package 6.

Sian Noble (https://orcid.org/0000-0002-8011-0722) (Senior Lecturer in Health Economics) was a co-applicant and lead for work package 5.

Cecily K Palmer (https://orcid.org/0000-0002-2096-008X) (Senior Research Associate in Health Economics) carried out qualitative studies in work package 4.

Ola Rolfson (https://orcid.org/0000-0001-6534-1242) (Professor) was principal investigator for the Swedish work package 4 centres.

Simon Strange (https://orcid.org/0000-0002-4306-4640) was the programme manager and co-ordinated recruitment and data collection in work packages 4 and 6.

Michael R Whitehouse (https://orcid.org/0000-0003-2436-9024) (Reader and Consultant in Trauma and Orthopaedics) was work package 2 deputy lead and work package 4 co-lead.

Publications

Kunutsor SK, Whitehouse MR, Blom AW, Beswick AD, INFORM Team. Re-infection outcomes following one- and two-stage surgical revision of infected hip prosthesis: a systematic review and meta-analysis. PLOS ONE 2015;10:e0139166. https://doi.org/10.1371/journal.pone.0139166

Kunutsor SK, Whitehouse MR, Webb J, Toms A, Stockley I, Taylor A, et al. Re-infection outcomes following one- and two-stage surgical revision of infected hip prosthesis in unselected patients: protocol for a systematic review and an individual participant data meta-analysis. Syst Rev 2015;4:58. https://doi.org/10.1186/s13643-015-0044-0

Moore AJ, Blom AW, Whitehouse MR, Gooberman-Hill R. Deep prosthetic joint infection: a qualitative study of the impact on patients and their experiences of revision surgery. BMJ Open 2015;5:e009495. https://doi.org/10.1136/bmjopen-2015-009495

Kunutsor SK, Whitehouse MR, Blom AW, Beswick AD, INFORM Team. Patient-related risk factors for periprosthetic joint infection after total joint arthroplasty: a systematic review and meta-analysis. PLOS ONE 2016;11:e0150866. https://doi.org/10.1371/journal.pone.0150866

Kunutsor SK, Whitehouse MR, Lenguerrand E, Blom AW, Beswick AD, INFORM Team. Re-infection outcomes following one- and two-stage surgical revision of infected knee prosthesis: a systematic review and meta-analysis. PLOS ONE 2016;11:e0151537. https://doi.org/10.1371/journal.pone.0151537

Strange S, Whitehouse MR, Beswick AD, Board T, Burston A, Burston B, et al. One-stage or two-stage revision surgery for prosthetic hip joint infection – the INFORM trial: a study protocol for a randomised controlled trial. Trials 2016;17:90. https://doi.org/10.1186/s13063-016-1213-8

Wyatt MC, Beswick AD, Kunutsor SK, Wilson MJ, Whitehouse MR, Blom AW. The alpha-defensin immunoassay and leucocyte esterase colormetric strip test as diagnostic tests in periprosthetic infection: a systematic review and meta-analysis. J Bone Joint Surg Am 2016;98:992–1000. https://doi.org/10.2106/JBJS.15.01142

Kunutsor SK, Beswick AD, Peters TJ, Gooberman-Hill R, Whitehouse MR, Blom AW, Moore AJ. Health care needs and support for patients undergoing treatment for prosthetic joint infection following hip or knee arthroplasty: a systematic review. PLOS ONE 2017;12:e0169068. https://doi.org/10.1371/journal.pone.0169068

Kunutsor SK, Whitehouse MR, Blom AW, Beswick AD. Systematic review of risk prediction scores for surgical site infection or periprosthetic joint infection following joint arthroplasty. Epidemiol Infect 2017;145:1738–49. https://doi.org/10.1017/S0950268817000486

Lenguerrand E, Whitehouse MR, Beswick AD, Jones SA, Porter ML, Blom AW. Revision for prosthetic joint infection following hip arthroplasty: evidence from the National Joint Registry. Bone Joint Res 2017;6:391–8. https://doi.org/10.1302/2046-3758.66.BJR-2017-0003.R1

Lenguerrand E, Whitehouse MR, Beswick AD, Toms AD, Porter ML, Blom AW, National Joint Registry for England, Wales, Northern Ireland and the Isle of Man. Description of the rates, trends and surgical burden associated with revision for prosthetic joint infection following primary and revision knee replacements in England and Wales: an analysis of the National Joint Registry for England, Wales, Northern Ireland and the Isle of Man. BMJ Open 2017;7:e014056. https://doi.org/10.1136/bmjopen-2016-014056

Moore AJ, Blom AW, Whitehouse MR, Gooberman-Hill R. Managing uncertainty – a qualitative study of surgeons’ decision-making for one-stage and two-stage revision surgery for prosthetic hip joint infection. BMC Musculoskelet Disord 2017;18:154. https://doi.org/10.1186/s12891-017-1499-z

Moore AJ, Whitehouse MR, Gooberman-Hill R, Heddington J, Beswick AD, Blom AW, Peters TJ. A UK national survey of care pathways and support offered to patients receiving revision surgery for prosthetic joint infection in the highest volume NHS orthopaedic centres. Musculoskeletal Care 2017;15:379–85. https://doi.org/10.1002/msc.1186

Kunutsor SK, Beswick AD, Whitehouse MR, Wylde V, Blom AW. Debridement, antibiotics and implant retention for periprosthetic joint infections: a systematic review and meta-analysis of treatment outcomes. J Infect 2018;77:479–88. https://doi.org/10.1016/j.jinf.2018.08.017

Kunutsor SK, Whitehouse MR, Blom AW, Board T, Kay P, Wroblewski BM, et al. One- and two-stage surgical revision of peri-prosthetic joint infection of the hip: a pooled individual participant data analysis of 44 cohort studies. Eur J Epidemiol 2018;33:933–46. https://doi.org/10.1007/s10654-018-0377-9

Lenguerrand E, Whitehouse MR, Beswick AD, Kunutsor SK, Burston B, Porter M, Blom AW. Risk factors associated with revision for prosthetic joint infection after hip replacement: a prospective observational cohort study. Lancet Infect Dis 2018;18:1004–14. https://doi.org/10.1016/S1473-3099(18)30345-1

Mallon C, Gooberman-Hill R, Blom A, Whitehouse M, Moore A. Surgeons are deeply affected when patients are diagnosed with prosthetic joint infection. PLOS ONE 2018;13:e0207260. https://doi.org/10.1371/journal.pone.0207260

Mallon CM, Gooberman-Hill R, Moore AJ. Infection after knee replacement: a qualitative study of impact of periprosthetic knee infection. BMC Musculoskelet Disord 2018;19:352. https://doi.org/10.1186/s12891-018-2264-7

Kunutsor SK, Beswick AD, Whitehouse MR, Blom AW. One- and two-stage surgical revision of infected elbow prostheses following total joint replacement: a systematic review. BMC Musculoskelet Disord 2019;20:467. https://doi.org/10.1186/s12891-019-2848-x

Kunutsor SK, Beswick AD, Whitehouse MR, Blom AW, Lenguerrand E. Implant fixation and risk of prosthetic joint infection following primary total hip replacement: meta-analysis of observational cohort and randomised intervention studies. J Clin Med 2019;8:722. https://doi.org/10.3390/jcm8050722

Kunutsor SK, Wylde V, Whitehouse MR, Beswick AD, Lenguerrand E, Blom AW. Influence of fixation methods on prosthetic joint infection following primary total knee replacement: meta-analysis of observational cohort and randomised intervention studies. J Clin Med 2019;8:828. https://doi.org/10.3390/jcm8060828

Lenguerrand E, Whitehouse MR, Beswick AD, Kunutsor SK, Foguet P, Porter M, et al. Risk factors associated with revision for prosthetic joint infection following knee replacement: an observational cohort study from England and Wales. Lancet Infect Dis 2019;19:589–600. https://doi.org/10.1016/S1473-3099(18)30755-2

Carroll FE, Gooberman-Hill R, Strange S, Blom AW, Moore AJ. What are patients’ preferences for revision surgery after periprosthetic joint infection? A discrete choice experiment. BMJ Open 2020;10:e031645. https://doi.org/10.1136/bmjopen-2019-031645

Garfield K, Noble S, Lenguerrand E, Whitehouse MR, Sayers A, Reed MR, Blom AW. What are the inpatient and day case costs following primary total hip replacement of patients treated for prosthetic joint infection: a matched cohort study using linked data from the National Joint Registry and Hospital Episode Statistics. BMC Med 2020;18:335. https://doi.org/10.1186/s12916-020-01803-7

Kunutsor SK, Barrett MC, Whitehouse MR, Blom AW. Clinical effectiveness of treatment strategies for prosthetic joint infection following total ankle replacement: a systematic review and meta-analysis. J Foot Ankle Surg 2020;59:367–72. https://doi.org/10.1053/j.jfas.2019.04.016

Kunutsor SK, Barrett MC, Whitehouse MR, Craig RS, Lenguerrand E, Beswick AD, Blom AW. Incidence, temporal trends and potential risk factors for prosthetic joint infection after primary total shoulder and elbow replacement: systematic review and meta-analysis. J Infect 2020;80:426–36. https://doi.org/10.1016/j.jinf.2020.01.008

Palmer CK, Gooberman-Hill R, Blom AW, Whitehouse MR, Moore AJ. Post-surgery and recovery experiences following one- and two-stage revision for prosthetic joint infection – a qualitative study of patients’ experiences. PLOS ONE 2020;15:e0237047. https://doi.org/10.1371/journal.pone.0237047

Blom AW, Lenguerrand E, Strange S, Noble SM, Beswick AD, Burston A, et al. Clinical and cost effectiveness of single stage compared with two stage revision for hip prosthetic joint infection (INFORM): pragmatic, parallel group, open label, randomised controlled trial. BMJ 2022;379:e071281.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Following review, access to available anonymised data may be granted.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it is important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data is used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, CCF, PGfAR or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, the PGfAR programme or the Department of Health and Social Care.

Background

Different surgical methods to treat prosthetic joint infection have economic implications for health-care providers and patients. Health services and decision-makers need robust information identifying interventions that provide greatest value for money. Our aim was to identify and critically evaluate the available economic evidence to guide the surgical revision of prosthetic joint infection after hip and knee replacement.

Methods

The systematic review protocol was registered as PROSPERO CRD42017069526.

Results

Review progress is summarised in Figure 17. After detailed evaluation, 19 studies were included in the review. Fifteen studies reported cost comparisons of revision procedures for prosthetic joint infection with primary joint replacement or aseptic revision. One of these and two further studies presented different costs associated with single-stage and two-stage revision procedures. In two studies, costs associated with different spacers were compared. Quality assessment is summarised in Table 9.

FIGURE 17.

Economic evaluations of revision surgery for prosthetic hip or knee infection: systematic review flow diagram. PJI, prosthetic joint infection.

Discussion

The costs of planned revision of infection are, on average, approximately four times those of primary hip or knee replacement and 2.4 times those of aseptic revision. The costs of further treatments associated with treatment failure are highly dependent on the success rate of revision treatments in different studies, and this varied markedly in the studies we identified. Thus, in two studies with low rates of reinfection, two-stage revision was about 60–70% more costly than single-stage revision. However, in one study with a highly unfavourable reinfection rate after single-stage revision, a two-stage strategy was associated with 40% lower costs.

The studies we identified highlight the importance of knowing the clinical effectiveness of treatment strategies when assessing the costs associated with them. In the absence of randomised evaluations, in systematic reviews of effectiveness of revision strategies for treatment of prosthetic joint infection, an attempt to limit this has been to select and compare pure case-series in which all patients with infection have received a specific treatment. 47,86,87,133 The inclusion of studies with patients selected for a treatment strategy based on infection severity or their health may give misleading results. However, as surgeons must to some degree tailor surgical management to individual patients, this may not always be avoidable. Only through the collection of detailed cost data in randomised evaluations and treatment comparisons in large unselected cohorts can definitive information on the costs and cost-effectiveness of different treatment strategies be generated.

The treatment of infection after hip and knee replacement with a two-stage revision is costly, and the single-stage revision strategy and other alternatives potentially address the need of patients and health-care providers for a less invasive, time-consuming and expensive treatment for infection. Much emphasis has been placed on the use of DAIR as an alternative to single- or two-stage revision. This is less invasive, but overall infection clearance rates may be only 61.4%, with a wide variation in rates reported in case series (range of clearance rates 11.1–100%). 85 Another option for hip infection relies on the use of a CUMARS, with the option of postponing the second stage and retaining the CUMARS prosthesis. 46 As with DAIR, this is not the final treatment for many patients, but in a case series 45% of patients receiving this treatment chose to keep the temporary prosthesis rather than receive a new prosthesis in a second-stage operation. 46 The initial costs of DAIR and a CUMARS are lower than those of a two-stage revision but, with the need for further treatments, costs may, ultimately, be higher.

In the three studies we identified comparing single- and two-stage revision and considering the financial consequences of treatment failure, the rates of successful treatment differed markedly between studies. This led to contradictory conclusions that both the single-stage and the two-stage strategy are the lower-cost option. However, in these studies the numbers of patients receiving treatment were small.

With the reporting of ongoing RCTs comparing single- and two-stage revision of joints affected by prosthetic joint infection,97,134,135 associated economic analyses will provide both robust standalone data and information to populate decision models. Such analyses should be from a patient and societal perspective in addition to a health-care provider perspective. Until then, it is important that health-care providers are aware that the costs of treatment of prosthetic joint infection are high and vary considerably between individual patients, and that prior to the INFORM RCT, insufficient high-quality evidence was available to advise on choice of revision strategy.

Introduction

We aimed to compare risk of re-revision and mortality after single- or two-stage revision procedures for treatment of hip or knee prosthetic joint infection in England and Wales.

Methods

Results: hip revision procedures

Between 2003 and 2014, 2140 primary hip replacements were revised for prosthetic joint infection: 535 with a single-stage procedure and 1605 with a two-stage procedure. Patients revised with single-stage procedure were on average older (68 years vs. 66 years) and less likely to be male (51% vs. 55%) and to have an ASA grade of > 2 (26% vs. 29%) than patients revised with a two-stage procedure.

Results: knee revision procedures

Between 2003 and 2014, 3369 primary knee replacements were revised for prosthetic joint infection. Of these, 489 were treated with a single-stage procedure and 2377 with both stages of a two-stage procedure. A further 503 patients received only the first stage of a planned two-stage procedure.

Discussion

Our study of over 2000 revisions for the management of infected primary hip replacements shows that single-stage revision is associated with a higher risk of unplanned re-revision for both all-cause and specifically for further prosthetic joint infection when compared with two-stage revision. This is particularly marked in the earlier post-operative period.

In the management of knee prosthetic joint infection with over 3000 patients studied, rates of re-revision for all causes and specifically for reinfection were similar after single- and two-stage revision procedures.

Compared with patients receiving two-stage revision, those with single-stage revision for the treatment of hip and knee prosthetic joint infection received an average of 41% and 45% fewer operations respectively.

Although mortality was higher than after primary hip replacement or aseptic revision, mortality rates were comparable between single- and two-stage revision treatments of hip prosthetic joint infection. After revision of knee prosthetic joint infection, there was a higher mortality in patients who received a two-stage revision compared with those receiving a primary hip replacement or aseptic revision. There was no difference in mortality after single-stage revision compared with primary hip replacement or aseptic revision. Overall, mortality after single- and two-stage revision for knee prosthetic joint infection was similar.

To our knowledge, this is the largest study to compare the incidence of re-revision after single-stage and two-stage revision for prosthetic joint infection. We used a standardised data collection process and adjustment approach, examining component-level data to precisely define and group comparable procedures. Only procedures in which an implant is added, removed or modified are recorded in the NJR. Thus, we were unable to explore the risks for prosthetic joint infection treated with antibiotics or incision and drainage alone, but the reoperation outcomes are substantially worse for this strategy. The NJR does not capture data on the presence of a sinus or the microorganism causing the prosthetic joint infection and there may be selection of patients with easier to treat infections for a particular revision strategy.

Our research shows some advantage for a two-stage strategy over the single-stage revision with regard to preventing re-revision. However, with a two-stage strategy, the treatment burden for patients and families due to the greater number of surgeries, complications associated with the interim period and prolonged periods of immobility is considerable.

Aims

Our aim was to determine whether or not there is a difference in patient-reported outcome measures 18 months after randomisation to single- or two-stage revision surgery for the treatment of hip prosthetic joint infection. Patients deemed to have infected hip replacements by their treating surgeons and multidisciplinary prosthetic joint infection teams were randomised to either single- or two-stage revision. Although the study was not powered to compare rates of reinfection or occurrence of new infection, adverse events including infection were monitored.

Methods

The INFORM trial was a multicentre, two-arm, parallel-group, participant and observer unblinded, randomised, superiority trial with 1 : 1 treatment allocation (ISRCTN10956306). Full details of the study design, methods and statistical analysis, along with the associated cost-effectiveness and qualitative studies are available in the published protocol. 97

Results

A CONSORT flow diagram is shown in Figure 15. Of 186 patients eligible for inclusion in the trial, 65 were randomised to single-stage revision and 75 were randomised to two-stage revision. The baseline characteristics of participants in the randomised groups, shown in Table 13, were balanced between the groups, except that patients in the single-stage group were more likely to be male, to have ASA grade one or two and to have received previous non-surgical management for their infection. The groups were similar in the number and type of organisms cultured, rates of culture-negative infection and presence of a sinus tract. A total of 126 (90%) patients had the primary total WOMAC score outcome at 18 months post randomisation, and 133 (95%) patients had at least one postoperative total WOMAC score and were included in the main analysis.

In the single-stage group, 55 (84.6%) patients received their assigned intervention, eight received a two-stage procedure and one did not receive a revision. Four patients withdrew during follow-up and two died. In the two-stage group, 68 (90.7%) patients received their assigned intervention, five received an alternative revision procedure and one did not receive a revision. Three patients withdrew during follow-up and five died. The median time between stages in the two-stage group was 3.7 months (IQR 2.6–6.1 months).

Conclusion

The INFORM trial demonstrated that single- and two-stage revisions are equally efficacious for the treatment of prosthetic joint infection. However, an early functional benefit was seen in the single-stage group and, given that patients stated a strong preference for a quick return to normal activity, this should be considered when selecting the intervention. Single-stage surgery accounted for 30% of revision procedures for the treatment of prosthetic joint infection in 2014, so there is scope for increased utilisation.

Limitation

The INFORM sample size precluded study of reinfection outcome.

Background

In the INFORM RCT we incorporated an economic evaluation to determine the cost-effectiveness of a single-stage compared with a two-stage revision strategy in patients with hip prosthetic joint infection.

Methods