Interlaboratory Reliability of Synovial Fluid Analysis in Prosthetic Knee Aspiration

Abstract

Background

Synovial fluid analysis is a critical step in the diagnosis of periprosthetic knee joint infection. We hypothesized that clinically meaningful differences between white blood cell (WBC) count, polymorphonuclear leukocyte (PMN) cell percentage, and culture results would exist between 2 independent laboratories when the same synovial fluid aspirate is analyzed.

Methods

A retrospective chart review was completed on total knee arthroplasty patients who underwent knee arthrocentesis between 2017 and 2025. The synovial fluid aspirate from a single knee arthrocentesis was divided into identical samples and analyzed by 2 independent laboratories. WBC count, PMN percentage, and culture results were recorded.

Results

In total, 202 consecutive prosthetic knee aspirates were analyzed. Agreement was found in 14 of 20 samples (70%) that tested ≥3000 WBC/μl (k = 0.81; 95% confidence inverval [CI], 0.66-0.96), 9 of 13 samples (69%) that tested ≥80% PMNs (k = 0.81; 95% CI, 0.62-0.99), 7 of 12 cases (58%) reported both ≥80% PMNs and ≥3000 WBC/μl, 8 of 9 cases (89%) that tested ≥90% PMNs (k = 0.94; 95% CI, 0.82-1.00), 3 of 23 cases (13%) with positive culture results (k = 0.18; 95% CI, −0.05-0.40).

Conclusions

Two independent laboratories displayed clinically meaningful variability in WBC count, PMN percentage, and culture results. The difficulty in diagnosing periprosthetic joint infection may at least in part be due to poor interlaboratory reliability concerning synovial fluid analysis. Refinement of interlaboratory agreement, further research into advanced diagnostic methods, and measures to improve the accuracy of synovial fluid analysis are all critical steps toward refining the diagnosis of periprosthetic joint infection.

Level III

Retrospective Cohort study, Diagnostic.

Introduction

Periprosthetic joint infection (PJI) is a devastating complication following total joint arthroplasty and is one of the most common indications for revision total knee replacement. [1] Unfortunately, diagnosing PJI continues to be a challenge for orthopaedic surgeons despite the many advancements that have been made. Synovial fluid biomarkers, including synovial white blood cell (WBC) count and polymorphonuclear leukocyte (PMN) percentage, and culture are commonly used in diagnostic algorithms for PJI. Notably, however, wide variations in WBC count and PMN percentage suggestive of PJI have been reported. [[2], [3], [4], [5]] A gold standard cutoff value or range has yet to be determined. [6]

The Musculoskeletal Infection Society’s 2011 criteria for PJI provided an algorithm for diagnosing hip and knee PJI, which proved to have excellent performance on external validation in a 2018 update. [2,7,8] Combining serum and synovial fluid biomarkers with histology has improved clinicians’ ability to accurately diagnose PJI. [2,7,8] The effective use of the updated Musculoskeletal Infection Society criteria, or any diagnostic algorithm dependent on synovial fluid analysis, relies on laboratories producing accurate, reproducible results from synovial and serum samples. While the quality of aspiration technique is known to influence sample purity and consequently affect synovial biomarker values, to our knowledge, the effect of interlaboratory variability or error when analyzing these samples is not currently known. [9]

Anecdotal observations of interlaboratory variation in synovial fluid results ultimately prompted our investigation. We hypothesized that analysis of the same synovial fluid sample by 2 independent laboratories would yield clinically meaningful differences in WBC count, PMN percentage, and culture results. If disagreement was found between synovial fluid biomarker results from the 2 laboratories, interlaboratory variability may represent a major contributing factor to the diagnostic dilemma of PJI. [7]

Material and methods

An institutional review board–approved, single-institution retrospective chart review was completed on patients who underwent knee arthrocentesis between 2017 and 2025. Patients were identified based on current procedural terminology and International Classification of Diseases codes using a search function in the institutional electronic medical record. Queried codes were “initial encounter for inflammatory reaction due to internal prosthesis” or “pain due to internal orthopedic device”. Specimens provided by patients undergoing knee arthrocentesis due to clinical concern for PJI were sent to 2 independent laboratories for analysis. [10,11] The inclusion criteria required a minimum arthrocentesis aspirate yield of approximately 5 mL, and that patients were at least 90 days remote from their most recent arthroplasty procedure. Proprietary laboratory analysis techniques were not disclosed by either industrial clinical laboratory.

Sample collection technique and laboratory analysis

Arthrocentesis was performed in a clinical setting under sterile technique by a fellowship-trained adult reconstructive orthopaedic surgeon or a trained physician assistant. The provider divided a single synovial fluid aspirate into 2 sets of samples to be analyzed by the 2 independent laboratories. The samples sent to lab 1 included one no-additive vacutainer, one K2 ethylenediaminetetraacetic acid (EDTA) vacutainer, and a sterile specimen cup for excess fluid. These samples were then placed in an on-site drop box for same-day pick-up and shipping by a lab 1 courier. Lab 2 provided a collection kit including 2 no-additive vacutainers, 1 K2 EDTA vacutainer, collection instructions, packaging, and shipping labels. Following collection and packaging per provided instructions, samples were shipped overnight directly to the laboratory after aspiration. No additive and K2 EDTA vacutainers each required a minimum sample volume of 1 mL, with a preferred minimum volume of 3 mL for the no additive vacutainers. A minimum total aspirate yield of approximately 5 mL was required for adequate sample volume to be sent to both labs. Labs 1 and 2 incubated anaerobic cultures for a total of 14 days and 7 days, respectively. However, if lab 2 identified Cutibacterium acnes, a slow-growing species, on its proprietary immunoassay-based identification test, then the incubation period would be extended to 14 days. The results from each laboratory were uploaded into the institution’s electronic medical record. Samples with missing data from either lab were excluded from the respective analysis.

Data analyses

Data analyses were performed with R 4.4.3. Estimates are reported as count (percentage) and mean (standard deviation). Cohen’s Kappa (k) was used to quantify the agreement between the 2 laboratories. Agreement was defined as both labs reporting the same result for a sample. For example, if both labs reported a cell count ≥3000 WBC/μl for a sample, then the labs were in agreement. Disagreement was defined as both labs reporting a different result for a sample. For example, if one lab reported a cell count ≥3000 WBC/μl and the other lab reported a cell count <3000 WBC/μl, then the labs were in disagreement. Variables, including automated WBC count, PMN percentage, and culture results, were recorded from each laboratory’s synovial fluid analysis report. Greater than 3000 WBC/μl and 80% PMNs were selected for use in the agreement analysis to indicate a high suspicion of infection. Values between 2000 and 2999 WBC/μl and 60 to 79% PMNs were selected to indicate a moderate suspicion of infection. Values below 2000 WBC/μl and 60% PMNs were selected to indicate a low suspicion of infection. These values were selected based on a review of diagnostic parameters found in the literature [6].

Results

Between 2017 and 2025, 202 consecutive arthrocenteses on prosthetic knee joints from 193 total patients were performed. Complete paired data were available for 199 WBC counts, 200 PMN percentages, and 194 culture results after excluding samples with missing data. The median time between the most recent surgery and aspiration was 22 months (range, 4 months to 22 years). Average (standard deviation) age was 65 (10) years, 45.0% (91 of 202) were male, 44.6% (90 of 202) were smokers, 23.3% (47 of 202) were diabetic, average (standard deviation) body mass index was 32.0 (6.4) kg/m², 29.2% (59 of 202) had total knee arthroplasty revision surgery before the arthrocentesis testing including 26 (12.9%) for infection irrigation and debridement.

WBC count

Twenty samples resulted in ≥3000 WBC/μl from one or both of the independent laboratories, 14 of 20 (70%) were in agreement between both laboratories (k = 0.81; 95% CI, 0.66-0.96) in the high suspicion of infection group (Table 1). Eleven samples resulted in 2000 to 2999 WBC/μl from one of the independent laboratories, 1 of 11 (9%) were in agreement between both laboratories in the moderate suspicion of infection group. One hundred eighty-one samples resulted in <2000 WBC/μl from one of the independent laboratories; 171 of 181 (94%) were in agreement between both laboratories in the low suspicion of infection group.

Table 1.

Results of synovial fluid WBC count.

Cell count (WBC/μl)	Lab 1	Lab 2	Total	Agreement
≥3000	17	17	20	14 of 20 (70%)
2000 to 2999	6	6	11	1 of 11 (9%)
<2000	176	176	181	171 of 181 (94%)
≥2400	21	19	25	15 of 25 (60%)
≥1700	25	24	30	19 of 30 (63%)

PMN percentage

Thirteen samples resulted in ≥80% PMNs from one of the independent laboratories, 9 of 13 (69%) were in agreement between both laboratories in the high suspicion of infection group (k = 0.81; 95% CI, 0.62-0.99) (Table 2). Twenty-three samples resulted in 60 to 79% PMNs from one of the independent laboratories, 4 of 23 (17%) were in agreement between both laboratories in the moderate suspicion of infection group. One hundred eighty-five samples resulted in <60% PMNs from one of the independent laboratories, 166 of 185 (90%) were in agreement between both laboratories in the low suspicion of infection group. Nine samples resulted in ≥90% PMNs from one of the independent laboratories, 8 of 9 (89%) were in agreement between both laboratories (k = 0.94; 95% CI, 0.82-1.00).

Table 2.

Results of PMN cell percentage.

Poly %	Lab 1	Lab 2	Total	Agreement
≥80	11	11	13	9 of 13 (69%)
60 to 79	15	12	23	4 of 23 (17%)
<60	174	177	185	166 of 185 (90%)
90 to 100	8	9	9	8 of 9 (89%)
80 to 89	3	2	4	1 of 4 (25%)
70 to 79	10	4	12	2 of 12 (17%)
60 to 69	5	8	12	1 of 12 (8%)

Cultures

Microorganisms were identified by culture in 23 of 194 analyzed samples at lab 1, lab 2, or both laboratories. Microorganisms were identified by only lab 1 in 12 samples, by only lab 2 in 8 samples, and by both laboratories in 3 samples. The 3 samples (13%) showed agreement regarding bacterial speciation (k = 0.18; 95% CI, −0.05-0.40). The identified microorganisms were Staphylococcus aureus, Escherichia coli, and Streptococcus mitis/oralis in the 3 samples. The Staphylococcus aureus sample had >28,000 WBC/μl and >80% PMNs reported by both labs. The Escherichia coli sample had > 37,000 WBC/μl and >90% PMNs reported by both labs. The Streptococcus mitis/oralis sample had >90% PMNs reported by both labs, but >20,000 WBC/μl in one lab compared with >2000 WBC/μl in the other.

Aggregates

Twelve samples resulted in both ≥80% PMNs and ≥3000 WBC/μl, 7 of 12 (58%) were in agreement between both laboratories in the aggregate high suspicion of infection group (Table 3). Eight samples resulted in ≥80% PMNs, ≥3000 WBC/μl, and a positive culture; 2 of 8 (25%) were in agreement between all 3 variables.

Table 3.

Cases with identified microorganisms and aggregated results.

Microorganism identified in culture			Lab 1	Lab 2	Total	Species agreement
Yes			15	11	23	3 of 23 (13%)
Cases with ≥80% poly and ≥3000 Cell count
Poly %	Cell count (WBC/μl)
≥80	≥3000		9	10	12	7 of 12 (58%)
60 to 79	2000 to 2999		1	2	3	0 of 3 (0%)
Cases with ≥80% Poly, ≥3000 Cell count, and positive culture
≥80	≥3000	Positive culture	3	7	8	2 of 8 (25%)
60 to 79	2000 to 2999		0	0	0	None

Discussion

Infection is the most common indication for early revision in modern total knee arthroplasty. [1,12] PJI is currently diagnosed utilizing synovial fluid analysis in conjunction with clinical exam and imaging modalities. We sought to evaluate the reliability of 2 independent laboratories to provide accurate analysis of the same synovial fluid aspirate. Due to concerns about the accuracy and reproducibility of synovial fluid analysis, based on anecdotal experiences, the principal investigator began sending each aspirate to 2 laboratories during the retrospectively analyzed period. This study affirmed the authors’ hypothesis that clinically meaningful variation in synovial fluid analysis results exists between laboratories regarding WBC count, PMN percentage, and culture results, despite the samples being identical.

Of the synovial aspirates included in our culture analysis, microorganisms were identified in 23 samples at one or both laboratories. The same bacterial species were identified in 3 of these 23 samples (13%). In 2 cases, WBC count and PMN percentage from each laboratory exceeded thresholds for high suspicion of infection. The third case had PMN percentages that exceeded thresholds for high suspicion of infection, with conflicting WBC count values. The laboratories failed to identify a matching bacterial species in 87% (20 of 23) of culture-positive samples. Given the central role of culture in PJI diagnosis, the inconsistent bacterial speciation between laboratories represents one of the most concerning findings of our study. Although poor sample quality or aspiration technique can lead to unreliable culture results, [13] we mitigated this by using a standardized technique to obtain a single knee aspirate, which was then divided and sent to 2 independent laboratories. The poor interlaboratory reliability in determining bacterial speciation we observed may significantly impact antibiotic selection and surgical decision-making in the management of PJI.

The diagnostic parameters for WBC count and PMN percentage for chronic PJI vary widely. In order to evaluate agreement, we stratified WBC count and PMN percentage into high, moderate, and low suspicion of infection groups. The cell count and percentage parameters of each of these groups were based on averages of diagnostic ranges found in the literature during an extensive review. [6,14,15] Given that clinical and surgical decision-making hinges upon synovial fluid analysis results, it is generally assumed that synovial fluid WBC count and PMN percentage results are accurate and comparable across different reputable laboratories. However, our results reveal that interlaboratory variability may be greater than previously recognized.

Some striking examples of WBC count variation between the laboratories when analyzing the same sample include the following (lab 1 and lab 2): 20,725 and 2017 WBC/μl, 1266 and 29,550 WBC/μl, and 6165 and 2718 WBC/μl. Less striking, but clinically meaningful, disparity can also be seen in the following (lab 1 and lab 2): 3625 and 968 WBC/μl, 2400 and 900 WBC/μl, and 863 and 2225 WBC/μl. These results indicate the presence of clinically meaningful, substantial interlaboratory variability in WBC count, despite both samples being derived from the same aspirate. Given the critical role of WBC count in PJI diagnosis, this variability is a concerning finding. [16]

Evidence in the literature has suggested that interlaboratory variability in synovial WBC count results may exist, leading some to propose that the PMN percentage value should take precedence in PJI diagnosis. [17] However, PMN percentage agreement was also relatively poor in our analysis. PMN percentage ≥80% (high suspicion of infection) and between 60 to 79% (moderate suspicion of infection) showed only 69% (9 of 13) and 17% (4 of 23) agreement, respectively. Samples below the threshold for low suspicion of infection were observed to have a much higher agreement at 90% (166 of 185). Notably, PMN ≥90% achieved 89% agreement (8 of 9). These findings indicate that while both biomarkers exhibit variable interlaboratory agreement, PMN ≥90% may be a more consistent, reliable diagnostic adjunct in PJI, despite laboratory variability.

One limitation of this study is its narrow scope. It is intended only to determine interlaboratory accuracy and reproducibility of synovial fluid analysis. Future research should evaluate the diagnostic accuracy of PJI parameters, the clinical impact of poor interlaboratory reliability, concordance between aspirated and intraoperative cultures, and patient demographic factors associated with PJI diagnosis. Additionally, we did not assess other validated biomarkers known to improve PJI diagnostic accuracy, including synovial alpha-defensin, serum inflammatory markers, and histopathology. [7] While these limitations prevent the study from having clinical applicability in diagnosing PJI, they strengthen its core objective: determining whether providers can reliably trust traditional synovial fluid results across laboratories.

Automated, as opposed to manual, cell counts were performed by both laboratories. Although once viewed as a limitation, modern automated systems have been validated as accurate and efficient alternatives to manual methods. [18] Additionally, transport time and handling variability may significantly affect cell counts and bacterial load recovery, as delays and inconsistent practices are known to reduce both. [19] It is difficult to quantify how differing transportation times between the 2 laboratories may have affected our results, but it remains a potential confounder. The lack of complete transparency regarding each laboratory’s proprietary analysis techniques remains a limitation. We, however, believe that the limited scope and previously disclosed limitations of our analysis strengthen its real-world clinical relevance: PJI diagnosis routinely involves different laboratories and transport protocols, yet few clinicians account for this variability.

Conclusions

Our findings raise the question: Is the challenge of diagnosing chronic PJI at least in part due to erroneous, irreproducible laboratory results? Moreover, is the variability in diagnostic parameters for PJI due to variation in laboratory analysis? The data we present underscores the importance of relying not only on laboratory analysis in diagnosing PJI, but also history, clinical examination, and imaging. The importance of emerging technologies, including advanced polymerase chain reaction techniques and next-generation sequencing, is essential in improving diagnostic accuracy. While culture and biomarker evaluation remain the primary pillars of diagnosis, this may change in the coming years. Continued research into alternative diagnostic modalities and refining interlaboratory agreement are critical steps to enhance our ability to accurately diagnose PJI and identify the offending organisms.

Conflicts of interest

Mark G. Freeman is a paid consultant for Smith & Nephew and Conformis; owns stock or stock options in Axis Research & Tech, Omnimed, and Irrimax Corp.; and receives research support from Smith & Nephew.

The other authors declare no potential conflicts of interest.

For full disclosure statements refer to https://doi.org/10.1016/j.artd.2025.101920.

CRediT authorship contribution statement

Richard D. Murray: Writing – review & editing, Writing – original draft. Charles W. Powell: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Mitchell S. Scull: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Andrew W. Wilson: Writing – review & editing, Writing – original draft, Validation, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Mark G. Freeman: Writing – review & editing, Writing – original draft, Supervision, Project administration, Methodology, Investigation, Data curation, Conceptualization.