OBJECTIVES
Accurate diagnosis of peri-prosthetic joint infection is critical to allow adequate treatment. Currently, the criteria of the Musculo-Skeletal Infection Society (MSIS) serve as a validated reference tool. More recently, these criteria have been modified for better accuracy. The goal of this study was to compare retrospectively the diagnostic accuracy of these two different tools in cases of known peri-prosthetic total knee arthroplasty (TKA) infection or in aseptic cases and to analyze one additional criterion: presence of an early loosening (prior to 2 years after implantation).

Methods

All cases of knee prosthesis exchange operated on at our department during the year 2017 have been selected. There were 130 cases in 127 patients: 67 men and 60 women, with a mean age of 69 years. 74 cases were septic and 53 cases were aseptic.
All criteria included in both classifications were collected: presence of a fistula, results of bacteriological samples, ESR and CRP levels, analysis of the joint fluid, histological analysis. Additionally, the presence of an early loosening was recorded.
The diagnosis accuracy of the classical MSIS classification and of the 2018 modification were assessed and compared with a Chi-square test at a 0.05 level of significance.

Results

The conventional MSIS classification correctly discriminated between infected and non-infected cases in 128/130 cases (98%). There were two failures by infected cases: one case was considered infected with no major criteria and only three minor criteria; one case was considered infected with no major criteria and only two minor criteria. There was no failure by non-infected cases.
The new MSIS classification correctly discriminated between infected and non-infected cases in 129/130 cases (99%). There was one single failure by infected cases: one case was considered infected despite a score of 4 points.
There was no significant difference between the diagnostic accuracy of both classifications.
The presence of an early loosening had a high specificity (85%) but a low sensitivity (22%).

Conclusion

The conventional MSIS classification had a high diagnostic accuracy. The new MSIS classification offered only minor, non significant increase of this accuracy. As the new classification involves several additional biological assays, these results might question the cost-effectiveness of the new classification. The presence of an early loosening might be an interesting additional criterion at no additional cost.