Of 960 first-revision total hip replacements (THR) because of deep infection identified in the Swedish Hip Arthroplasty Registry, 16.9% were treated with a permanent implant extraction, while a staged or direct reconstruction revision protocol were employed in 56.2% and 26.9% respectively. The majority of the interventions were performed more than one year after index THR, and the dominating pathogen was coagulase negative staphylococci (CNS). We found a significant shift in types of bacteria over the years (Chi-square test, p smaller than 0.001): an increase in the CNS group and a decrease in Gram-negative aerobes. Patients treated with a permanent resection were generally older (p< 0.001), had more often a previous ipsilateral hip fracture (p< 0.001), and had more frequently Gram-negative infections (p=0.02). No systematic differences in patient characteristics or pathogens were detected between one-stage or two-stage procedures, of which the latter had a median re-implantation time of 2 (range: 0.2–62) months. Of 798 (one- or two-stage) revisions, 60 (7.5%) were revised again due to recurrent infection, with no difference between the two methods, and implying a 10-year survival of 90%; 95% confidence interval (CI95%) 88.2–93.0. Previous surgery for soft-tissue problems (RR 3.2 (CI95% 1.3–7.2)) predicted a worse outcome for one-stage procedures. The prognosis of two-staged revisions improved with increasing re-implantation interval (RR 0.8 (CI95% 0.7–1.0)) per month, and a 6 month interval carried the lowest risk of repeat revision due to infection; RR 0.1 (CI95% 0.0–0.9). Staged revisions in female patients (RR 2.3, (CI95% 0.9–5.7)) and with Staphylococcus aureus infections (RR 2.3 (CI95% 0.9–5.5)) predicted a worse outcome. Ten-year survival with repeat revision for aseptic loosening as end-point was 89% (95%CI 85.7–92.0), but decreased to 79% when all reasons for revision were taken into account (95%CI 75.0–82.3) mainly because of revision for peri-prosthetic fractures.

The results suggest that direct and staged revision protocols can have a good prognosis on a national level, but efforts must be made to counteract periprosthetic fractures and the high incidence of permanent implant extraction in elderly patients.

Our aim was to define the minimum set of patient-reported outcome measures which are required to assess health status after total hip replacement (THR). In 114 patients, we compared the pre-operative characteristics and sensitivity to change of the Oxford hip score (OHS), the Western Ontario and McMaster Universities osteoarthritis index (WOMAC), the SF-36, the SF-12 (derived from the SF-36), and the Euroqol questionnaire (EQ-5D).

At one year after operation, very large effect sizes were found for the disease-specific measures, the physical domains of the SF-12, SF-36 and the EQ-5D_index (1.3 to 3.0). Patients in Charnley class A showed more change in the OHS, WOMAC pain and function, the physical domains of the SF-36 and the EQ-5D_vas (p < 0.05) compared with those in the Charnley B and C group. In this group, the effect size for the OHS more than doubled the effect sizes of WOMAC pain and physical function. We found high correlations and correlations of change between the OHS, the WOMAC, the physical domains of the SF-12 and the SF-36 and EQ-5D_index. The SF-36 and EQ-5D scores at one year after operation approached those of the general population. Furthermore, we found a binomial distribution of the pre-operative EQ-5D_index score and a pre-operative discrepancy and post-operative agreement between the EQ-5D_vas and EQ-5D_index.

We recommend the use of the OHS and SF-12 in the assessment of THR. The SF-36 may be used in circumstances when smaller changes in health status are investigated, for example in the follow-up of THR. The EQ-5D is useful in situations in which utility values are needed in order to calculate cost-effectiveness or quality-adjusted life years (QALYs), such as in the assessment of new techniques in THR.