July 2020: Orthopedic ID
Early “aseptic revision” for TKA and THA is a strong predictor for PJI in the following year. This retrospective study queried a large private payer database (PearlDiver Patient Records) with de-identified patient data from medicare (2005-2014) and Humana (2007-2017). The authors identified patients with ICD9 codes indicating they underwent primary total knee (TKA) or total hip (THA) arthroplasty for any reason, then compared those who had a revision of the same joint coded as being for a reason other than infection (i.e. instability, loosening, wear, mechanical complication, broken/failed hardware, or periprosthetic fracture) within 90 days of the index surgery to those who did not have such a revision. Cases were matched to controls at a rate of 1:10. The primary endpoint was prosthetic joint infection (again identified by ICD9 code) within 1 and 2 years of the revision procedure. I am only presenting the 1 year outcome here because the results were very similar at the 1 and 2 year timepoints.
The two data sets yielded a total of 267 patients who had THA with early aseptic revision (n=182 from the Medicare data and n=85 from the Humana data). Controls were identified for each group within the same dataset. In the medicare group, the rate of THA PJI at 1 year was 5.5% in those undergoing revision vs 0.8%, for OR of 8.7 with p<0.001. In the Humana group, the rate of THA PJI at 1 year was 11.8% in those undergoing revision vs 1.3% in those not, for an OR of 14.2 and p<0.001. The data sets yielded 416 patients who had TKA with early aseptic revision (n=190 from the medicare group and n=226 from the Humana group). Again, early revision was associated with TKA PJI at 1 year in both the medicare group (6.8% vs 1.2%, OR 7.7 with p<0.001) and the Humana group (6.2% vs 1.3%, OR 4.9 with p<0.001).
I think there are two main interpretations of this data. First is that it reflects a critical weakness of the study design, which is extrapolating what the orthopedic surgeons thought about why their patients needed a revision surgery by looking at ICD9 codes. Many of my consults are for periprosthetic fracture – due to orthopedic hardware infection – and so classifying a revision for that as “aseptic” does not make sense to me. Perhaps if the authors had some third filter on the cases, like whether antibiotics were prescribed after the revision surgery, we would see different results. The other interpretation is that there are some overly optimistic orthopedic surgeons out there diagnosing a bunch of infected arthroplasties as aseptic failures. This has not been my experience working with our orthopedists, but perhaps it’s possible elsewhere. 32712119
Platelet count/Lymphocyte ratio (PLR) and platelet count/platelet volume ratio (PVR) may have value as markers of prosthetic joint infection after TKA. What’s the premise behind these ratios being an indicator of infection? First, they’re an acute phase reactant, meaning inflammatory states are often associated with thrombocytosis (though paradoxically thrombocytopenia can also be a marker of sepsis). Second, platelets, when stimulated by bacteria or thrombin, excrete a number of their contents stored as alpha granules and dense bodies. Third, bacterial infection is often associated with lymphopenia. Hence, the investigators examined whether these data, which any lab can provide as part of a routine complete blood count, might have diagnostic value for PJI.
The authors reviewed the records of 538 patients admitted for revision TKA who had a CBC up to 7 days prior to surgery. Of these, 206 had a pre-operative diagnosis of PJI and 332 were classified as aseptic revisions. Within the PJI cohort, patients with chronic PJI based on IDSA’s diagnostic criteria were included, and others were excluded. The authors also excluded patients with systemic autoimmune diseases (e.g. SLE, IBD, RA) those with pre-existing thrombocytopenia from liver disease or other conditions, and those with prior history of TKA revision or I&D. The authors calculated PLR and PVR for each case as well as recording ESR, CRP, and synovial WBC and neutrophil percentage (PMN) values when obtained within 4 weeks of the revision surgery. Finally, they calculated the diagnostically optimal cutoffs for PVR and PLR and how their addition to current MSIS cutoffs for PJI would affect the sensitivity and specificity of those criteria.
For the diagnosis of PJI in patients undergoing revision for TKA, PLR had an AUC of 0.84 (95% CI 0.82-0.86) with an optimal cutoff of 234 yielding sensitivity/specificity of 78% and 83%, respectively. For PVR, the AUC was 0.85 (95% CI 0.81-0.89) and at an optimal cutoff of 30.8 yielded sensitivity/specificity of 90% and 76%, respectively. For reference, CRP at a cutoff of 10mg/L was 82% sensitive and 77% specific; ESR at a cutoff of 30mm/hr was 67% sensitive and 78% specific. Compared to the standard collection of biomarkers for PJI (ESR and CRP plus synovial WBC and PMN; sensitivity and specificity 95% and 94%), addition of the PVR increased sensitivity to 99% and specificity to 98%, and addition of the PLR increased both sensitivity and specificity to 99%.
I think this is very interesting – primarily because these new data points require a cheap, widely accessible test that we’re often already obtaining (as opposed to, say, an expensive send-off test like alpha defensin). In contrast to the authors, we have not found the ESR to add much diagnostic value to the CRP in the UNMC Ortho ID group, but I am interested to look through our data and see whether PVR and PLR might have identified patients with PJI not identified by synovial fluid analysis or CRP alone. 32773272
One year follow-up is adequate for reporting outcomes of knee and hip PJI. This was a retrospective review of 792 patients who underwent DAIR or 2-stage exchange of hip or knee PJI between 2000 and 2017. Treatment failure was identified using criteria a Delphi method, and the primary outcome was time from surgery to treatment failure using a Kaplan-Meier curve. There were no differences between survivorship of knee and hip PJI, or between DAIR and 2-stage exchange. The majority of failures occurred between 3-6 months after surgery, and rates began to plateau at 1.1 years. Long story short, research on PJI using a minimum of 1 year follow-up (as is standard practice) is probably reasonable. 32229149