1455. Impact of Preprocedural Ultrasound on Aspiration Approach and Yield in Patients with Suspected Infected Total Knee Arthroplasty
Authors * Denotes Presenting Author
  1. Michael Perry *; University of Virginia
  2. Joshua Knight; University of Virginia
  3. Nicholas Nacey; University of Virginia
Ultrasound is a frequently utilized modality to guide knee aspiration in patients with clinically suspected prosthetic joint infection. Preprocedural ultrasound can demonstrate an appropriate pocket of joint fluid to target, however in some cases little or no joint fluid is visualized. In these cases the needle can be guided to a questionable pocket of fluid or into the joint space posterior to the patella. The utility of performing aspiration in these patients is unclear in the absence of appreciable joint fluid.

Materials and Methods:
With IRB approval, a retrospective report search was performed to find patients status post total knee arthroplasty who underwent ultrasound guided aspiration in our department over a 5 year period from 2014 to 2019. Patients who underwent subsequent revision surgery within 100 days were included. Two board certified musculoskeletal fellowship trained radiologists independently reviewed saved ultrasound images in PACS and categorized expected joint fluid volume of <2 cc, 2-4 cc, or >4 cc. Subjective scan quality was recorded as high or low. Targeted location of aspiration attempt (suprapatellar, medial, or lateral recess) was noted. Preprocedural erthrycocyte sedimentation rate (ESR) and C reactive protein (CRP) levels were recorded if available, as were the results of the aspiration.

87 aspirations were included in the data set, with an average patient age of 65 and an average of 31 days between aspiration and surgery. 30 patients had infection at surgery and 57 did not. Substantial interobserver agreement was present as to the amount of fluid present with a kappa of 0.636. Disagreement between the two readers was higher for studies with low scan quality, although this was not statistically significant by a Chai square test (p=0.2987). There was a statistically significant difference between anticipated and actual volume of aspirated volume of fluid for both readers (p<<0.05), with a tendency for the volume of aspirated fluid to be underestimated by both readers. There was no significant association between anticipated fluid volume and infection for either reader 1 (p=0.6637) or reader 2 (p=0.08) using a Chai square test. Using anticipated fluid volume <2 cc as an indicator of lack of infection, the two readers had a negative predictive value of 40-45%. Among patients with anticipated fluid volume <2 cc by either reader, patients with a normal ESR and CRP had an 83% negative predictive value for infection.

While there is high interobserver agreement for anticipated volume of aspirate, the anticipated volume is significantly different than the actual volume with a tendency to be higher than expected. Low anticipated volume of aspiratable fluid had no association with infection and was not a reliable predictor for the absence of infection. Even among patients with normal inflammatory markers and low volume of fluid on preprocedural ultrasound, some may still have prosthetic joint infection.