Diagnosis of septic arthritis in the acute care setting: the value of routine intra-operative sample culture

Abstract

Objective

Diagnosing septic arthritis can be challenging and frequently involves clinical assessment, laboratory investigations and synovial fluid analysis. We sought to determine the utility of synovial aspiration and intra-operative synovial fluid and tissue culture for the accurate diagnosis of septic arthritis.

Methods

We carried out a retrospective review of the records of patients referred to a tertiary orthopaedic unit with possible septic arthritis between 2015 and 2019 inclusive, including clinical and laboratory data for this cohort study. Performance characteristics were determined for synovial aspiration, intra-operative synovial fluid and tissue culture in diagnosing expert review-determined true septic arthritis. Concordance between discharge diagnosis, antibiotic prescribing and true septic arthritis was determined.

Results

Of 268 patients identified with suspected septic arthritis, 143 underwent both synovial fluid aspiration and intra-operative synovial fluid and tissue biopsy culture. True septic arthritis was not differentiated significantly by laboratory parameters including serum white cell count (WCC), CRP or synovial WCC. Considering only patients with negative pre-operative synovial aspirate cultures, intra-operative samples led to diagnosis of true septic arthritis in 6 of 63 patients [number needed to treat (NNT) 10.5]. For all patients sampled in theatre, positive synovial tissue biopsy was the only evidence of true septic arthritis in six (NNT 23.9). Despite insufficient microbiological evidence, 27 of the 59 patients who did not have septic arthritis received a discharge diagnosis of septic arthritis, 26 of whom were discharged with antibiotics.

Conclusion

Intra-operative sample collection, particularly tissue biopsy, increases the likelihood of a true septic arthritis diagnosis. Such measures might help to reduce diagnostic ambiguity in clinical practice and might therefore reduce overtreatment.

Key messages.

• Peripheral blood parameters and synovial white cell count are unreliable in diagnosing septic arthritis.

• Synovial tissue biopsy aids in diagnosing septic arthritis above pre-operative and intra-operative synovial fluid culture.

• There is considerable overdiagnosis and overtreatment of septic arthritis without sufficient evidence.

Introduction

The diagnosis of septic arthritis in the acute care setting is usually made using a combination of history, examination and initial laboratory investigations [1]. Delays in the diagnosis and thereby treatment incur a significant risk of morbidity, including rapid and irreversible joint damage [2]. Despite this, the optimal approach for diagnosis has yet to be established. History, examination and initial laboratory investigations have all been shown to be insufficient to establish a diagnosis of septic arthritis reliably, and a positive synovial fluid specimen via aspiration or intra-operative specimen collection is usually required [3]. The sensitivity and specificity of these synovial fluid and intraoperative specimens have also been debated, with sensitivity values for the former ranging from 75 to 95%, depending on a variety of parameters including the degree of bacterial inoculation in the joint, bacterial species involved, prior exposure to antibiotics and handling of the synovial fluid sample after collection [4, 5]. In addition, the majority of studies in the literature are retrospective and without control groups [6].

The utility of pre-operative synovial aspiration or intra-operative synovial fluid and tissue biopsy culture in reliably diagnosing true septic arthritis therefore remains unclear. Although synovial aspiration is a relatively inexpensive and quick method of detection of early septic arthritis, whether its performance is comparable to intraoperative sample collection remains unclear. A number of studies have suggested that intra-operative synovial tissue biopsy might increase the likelihood of a culture-confirmed septic arthritis diagnosis, even with negative synovial fluid analysis [7, 8]. A systematic review conducted in 2011, however, concluded that there was a need for further studies to evaluate routine synovial tissue biopsy definitively for the diagnosis of septic arthritis [7].

Providing an accurate discharge diagnosis of septic arthritis is of great importance given that it dictates both the choice and duration of antibiotics [9]. In Australia, this information is sent to primary care physicians in the form of a hospital discharge summary and serves as the primary form of communication between the hospital and treating general practitioners. Concordance between an accurate discharge diagnosis and antibiotic prescribing is therefore important in ensuring safe patient management. It also avoids unnecessary antibiotic exposure and guards against the misdiagnosis of septic arthritis, thereby overlooking the true diagnosis (e.g. crystal arthropathy), with the consequent loss of opportunity to institute appropriate preventative management [9, 10].

In this study, we set out to ascertain the utility of pre-operative synovial aspiration and intra-operative synovial fluid culture and tissue biopsy for determining the true diagnosis of septic arthritis, in addition to the concordance between such a diagnosis and antibiotic prescription upon discharge.

Methods

All patients referred with a suspected diagnosis of septic arthritis by the emergency department (ED) to the orthopaedic surgery service at a tertiary referral centre in Melbourne, VIC, Australia are routinely entered prospectively into a structured electronic database. A suspected diagnosis of septic arthritis is generally made based on clinical features of a warm, swollen, erythematous joint with compromised function and history of fever, where an alternative diagnosis is not apparent. Factors influencing the likelihood of the diagnosis, including which joint is affected, the duration of symptoms before presentation, the presence of risk factors for septic arthritis [i.v. drug use (IVDU), trauma to the joint or immunosuppression] are also considered, and all these characteristics are captured in the database. Unless contraindicated, all such patients proceed to the operating theatre for joint arthrotomy or arthroscopy with washout, during which synovial fluid and tissue biopsy for culture are collected. This form of drainage is used to reduce bacterial load and aid in source control.

For the purposes of this study, the aforementioned database was interrogated for suspected septic arthritis cases between January 2014 and December 2019 (Fig. 1). Additional clinical features were also collected from chart review, including the joint affected, presence of a prosthesis, antibiotic administration before sample collection and use of US guidance for joint aspiration. Laboratory parameters collected for patient on presentation included serum white cell count (WCC <12.0 × 10⁹ g/l), CRP <5 mg/l, ESR <12 mm/h and blood culture results. Pre-operative aspirate results recorded included synovial fluid WCC, Gram stain, presence of crystals and culture results. Results of intra-operative synovial fluid and tissue biopsy culture collected were also recorded. These samples were collected routinely in the operating theatre and placed in a sterile approved receptacle (alone or in a saline-moistened gauze) and transported to the laboratory for analysis. The study was approved by the Austin Health Human Research Ethics Committee before commencement (LNR/16/Austin/408). Ethical approval was given for a research audit, and the institutional ethics committee waived the requirement for individual patient consent.

Figure 1.

Study flowchart

True septic arthritis was classified on the basis of growth of an organism from the pre-operative synovial aspirate and intra-operative specimens, including both synovial fluid and synovial tissue biopsy (Table 1). In the absence of a gold-standard diagnostic tool to detect contamination, discordant results were adjudicated by an independent expert microbiologist/infectious diseases physician (M.L.L.) to determine the true nature of the case by using the full clinical history, laboratory investigations and microbiological samples.

Table 1.

Study definitions for true septic arthritis and not septic arthritis

Test	Pre-operative synovial aspirate positive	Pre-operative synovial aspirate negative
Intra-operative synovial fluid or synovial tissue biopsy positive	Classified as true septic arthritis	Classification dependent on expert assessment of microbiological and clinical data
Intra-operative synovial fluid or synovial tissue biopsy negative	Classification dependent on expert assessment of microbiological and clinical data	Classified as not septic arthritis

Results

Over the 6 years of the study period, there were 268 patients comprising 274 ED presentations with a suspected diagnosis of septic arthritis. For patients with multiple presentations and recurrent septic arthritis, only data from the first ED presentation were used for analysis.

Of the 268 patients identified, 143 patients had both pre-operative synovial fluid aspiration and intraoperative specimens taken. A further 31 patients had pre-operative joint aspiration alone, and 88 had only intra-operative specimen collection. There were six patients with neither pre-operative aspirate nor intra-operative specimen collection. The median age of the 268 patients was 67 years [interquartile range (IQR) 51–78 years], and 151 (59%) were men. Patient demographics and the joints involved are listed in Table 2. The majority of patients (73%) had pre-existing joint disease (i.e. documented degenerative or inflammatory disease affecting that joint), and 68 patients (27%) had joint trauma before their presentation. The median CRP was 145 mg/l and median ESR 48.5 mm/h. Although only 70% of patients had peripheral blood cultures obtained during their admissions, these cultures were positive in 39% (70 patients). The median synovial WCC in those who had a joint aspiration was 126 109 × 10⁶/l (IQR 41 000 × 10⁶/l to 152 100 × 10⁶/l).

Table 2.

Demographic data on all patients and subgroups based upon specimens collected

Parameter	Total	Pre-operative synovial aspirate collected; intra-operative specimen collected	No pre-operative synovial aspirate collected; intra-operative specimen collected	Pre-operative synovial aspirate collected; no intra-operative specimen collected	No pre-operative synovial aspirate collected; no intra-operative specimen collected
Patients, n	268 (289 joints)	143	88	31	6
Age, years, median (IQR)	67 ( 51–78)	64 (50–79)	70 (54–75)	65 (51–80)	67 (65–73)
Joint affected, n (%)	Knee 107 (37)	Knee 80 (56)	Knee 19 (22)	Knee 7 (23)	Knee 1 (17)
	Hip 47 (18)	Hip 20 (14)	Hip 25 (28)	Hip 2 (7)	Hip 0
	Shoulder 36 (12)	Shoulder 19 (13)	Shoulder 11 (13)	Shoulder 4 (13)	Shoulder 2 (33)
	Elbow 22 (8)	Elbow 13 (9)	Elbow 6 (7)	Elbow 1 (3)	Elbow 0
	Ankle/foot 37 (13)	Ankle/foot 11 (8)	Ankle/foot 18 (20)	Ankle/foot 6 (19)	Ankle/foot 2 (33)
	Wrist 12 (4)	Wrist 9 (6)	Wrist 0	Wrist 2 (6)	Wrist 0
	Other 28 (10)	Other 7 (5)	Other 11 (13)	Other 9 (29)	Other 1 (17)
CRP, mg/l, median (IQR)	145 (44–248)	160.7 (50.2–263.3)	136.6 (34.8–250.3)	83.9 (35.8–111)	59.7 (19.7–59.3)
WCC, ×109 g/l, median (IQR)	11.6 (8.7–13.8)	12.4 (9.6–14.4)	10.5 (7.9–12.5)	10.5 (8.4–12.3)	10.4 (9.5–11.5)
ESR, mm/h, median (IQR)	48.5 (25.5–75)	55 (25–79.5)	44 (25–68)	35 (30.0– 56.0)	43 (36.5–49.5)
Trauma to joint, n	68	33	31	7	3
Joint prosthesis, n	77	15	56	3	3
Known IVDU, n	16	9	5	2	0
Pre-existing joint disease, n	197	95	81	16	5
Blood cultures during admission, n	219	142	57	18	2
Blood culture positive, n	73	44	24	4	1
Synovial WCC from aspirate, ×106/l, median (IQR)	126 109 (41 000–152 100)	133 030 (51 750–153 500)	N/A	42 979 (10 530–42 750)	N/A
Aspirate Gram stain positive, n	50	49	N/A	1	N/A

IQR: interquartile range; IVDU: i.v. drug use; N/A: not assessed; WCC: white cell count.

Some differences were apparent when patients with both pre-operative synovial aspirate and intra-operative specimen (fluid and/or tissue biopsy) results were compared with those who had only an aspirate or an intra-operative specimen (Table 2). The knee joint was the most commonly affected joint overall, but in the subgroup who proceeded straight to an intra-operative specimen the most commonly affected joint was the hip. This is possibly owing to this joint being difficult to access for bedside aspiration. The majority of patients with risk factors for septic arthritis, such as pre-existing joint disease, joint trauma and known IVDU, were also more likely to proceed to the operating theatre than not.

Using the results obtained from pre-operative synovial fluid aspirate and intra-operative specimens, the patients were divided into four subgroups, as outlined in Table 2. Median CRP and synovial fluid WCC were higher in the group with positive cultures in both pre-operative aspirate and intra-operative specimens when compared with the group with only pre-operative synovial aspirate or intra-operative specimen results alone (133 030 × 10⁶/l compared with 42 970 × 10⁶/l in the aspirate-only group). Patients with both pre-operative aspirate and intra-operative culture-positive specimens also included 142 (99%) patients with blood cultures collected during admission compared with 57 (65%) in the intra-operative alone and 18 (58%) in the aspirate-alone subgroups. When comparing the four subgroups, the group with the lowest median CRP, ESR and peripheral WCC was the group in which neither pre-operative synovial aspirate nor intra-operative specimens were collected.

The subgroups were examined further to evaluate what extra benefit was obtained from culturing an intra-operative sample. Of the 143 patients who had both a pre-operative synovial fluid aspirate and intra-operative synovial specimen collected, 85 (59%) had at least one positive culture and were thus defined as having true septic arthritis, whereas 58 did not. In cases with a disconcordance of pre-operative and intra-operative specimens, assessment was subsequently undertaken. On three occasions, septic arthritis was deemed unlikely even when there was a positive synovial tissue or fluid culture, after expert adjudication determining that the bacteria cultured was likely to be a contaminant, based on microbiological characteristics including species (Table 3).

Table 3.

Patients excluded from true septic arthritis group after expert adjudication

Age (years)	Joint	Previous joint disease	Aspirate WCC	Aspirate crystals	Blood culture	Location of positive organism	Intra-operative findings
77	Knee	Gout and OA	73 000	Nil	Staphylococcus epidermidis	Intra-operative swab	Pseudomonas aeruginosa with mixed skin and enteric flora
58	Knee	Gout	12 000	Urate crystals	Negative	Intra-operative tissue biopsy	Staphylococcus lugdunensis and Staphylococcus epidermidis
90	Knee	Gout and OA	Not done	Nil	Not done	Intra-operative synovial fluid	Staphylococcus epidermidis and urate crystals

The median age was similar in both groups (65 and 64 years in the septic arthritis and non-septic arthritis groups, respectively). Although the most commonly affected joint in both groups was the knee, those with large joint involvement (such as the hip) were unlikely to receive a pre-operative joint aspiration and were more likely to proceed to intra-operative specimen collection. The median blood WCC, CRP and ESR were also all higher in the septic arthritis group, as were the presence of clinical risk factors, including IVDU [relative risk (RR) 6] and trauma to the joint (RR 1.4). Table 4 outlines the demographic and risk factors associated with both true septic arthritis and not septic joint groups.

Table 4.

Demographic data comparing patients with culture-confirmed septic arthritis and non-septic arthritis

Parameter	True septic arthritis	Not septic arthritis	P-value
Patients	85	58
Age, years	65 (50–78)	64 (47–79)	0.727
Joint affected	Knee 48	Knee 32	0.077
	Hip 13	Hip 7
	Ankle/Foot 5	Ankle/Foot 6
	Shoulder 15	Shoulder 4
	Elbow 4	Elbow 9
	Wrist 7	Wrist 2
	Other 6	Other 1
CRP, mg/l	159 (67–303)	96 (43–151)	0.001
WCC, ×109 g/l	11.8 (9.1–15.4)	11.4 (9.8–14.2)	0.903
ESR, mm/h	70 (40–88)	31 (15–56)	0.001
Trauma to joint	21 (22)	12 (20)	0.843
Known IVDU	7 (8)	2 (3)	0.630
Pre-existing joint disease	59 (69)	36 (62)	0.373
Blood cultures during admission	71 (85)	34 (59)	0.001
Blood culture positive	41 (48)	3 (5)	<0.001
Synovial WCC from aspirate, ×106/l	78 300 (38 750–150 075)	103 500 (57 375–166 750)	0.148
Discharge summary diagnosis of septic arthritis	83 (98)	26 (45)	<0.001

Summary statistics are presented as the median (interquartile range) for continuous values and total (percentage) for counts. Statistical significance testing was performed using the Wilcoxon rank sum test for comparison of continuous values and Fisher's exact test for count values.

IVDU: i.v. drug use; WCC: white cell count.

Of the 85 patients with true septic arthritis, 70 (82%) had peripheral blood cultures taken, with a 48% positivity rate, whereas 34 (59%) of the not septic arthritis group had peripheral blood cultures taken, with only a 5% positivity rate. Of the three patients with false-positive peripheral blood cultures, all three had Staphylococcus species in their blood cultures. One of these patients had cellulitis as the source of their infection, and the other two might not have had definite diagnoses in practice, because they were treated as a septic joint with no other source of infection identified. In examining whether synovial WCC is a useful differentiator of septic arthritis from non-septic arthritis, the median synovial WCC was >70 000 × 10⁶/l in both the true septic arthritis group and the group with non-septic arthritis, and there was a wide range of values attributable to both groups (Supplementary Fig. S1, available at Rheumatology Advances in Practice online).

Diagnostic yield of proceeding to intra-operative specimen after a negative pre-operative aspirate

Of the 143 patients with both pre-operative aspiration and intra-operative specimen collections, 63 had a negative pre-operative synovial aspirate. Of these 63, 6 had a positive intraoperative specimen confirming the diagnosis of septic arthritis, resulting in a number needed to treat (NNT) of 10.5 for patients with a negative pre-operative synovial fluid aspirate required to undergo intra-operative sampling for one culture-positive intra-operative synovial specimen confirming septic arthritis. All six of these cases were confirmed on intra-operative synovial biopsy culture, and none with true septic arthritis had a positive intra-operative synovial fluid culture.

Concordance of intra-operative specimen collection with a positive pre-operative joint aspiration

Of the 81 patients with a positive pre-operative synovial aspirate, 10 (12%) had negative intra-operative synovial specimens. However, this discordance might be accounted for by numerous factors, including antibiotic use before intraoperative specimen collection in nine patients (90%), time between pre-operative and intra-operative sampling, and variability in the degree of bacterial inoculation.

Utility of proceeding to synovial tissue biopsy intraoperatively

Of the 143 patients who had an intra-operative sample collected, 6 patients had both a negative pre-operative synovial aspirate and negative intra-operative synovial fluid culture but a positive culture from the synovial tissue biopsy specimen. This correlates with a NNT of 23.9 patients undergoing intra-operative investigation (agnostic of pre-operative synovial fluid culture results) to have an intra-operative synovial tissue biopsy in order to diagnose an additional case of true septic arthritis.

Comparison of discharge diagnosis of septic arthritis with culture results

Of the 85 patients with true septic arthritis, 83 of these had this as a discharge diagnosis, with 1 documented with a diagnosis of reactive arthritis (Table 5). However, 26 of the remaining 58 (45%) patients classified in this study as not septic arthritis were also discharged with a diagnosis of septic arthritis, with all but 1 being prescribed antibiotics (Table 5).

Table 5.

Discharge diagnosis and antibiotic usage in patients with septic arthritis and not septic arthritis

	True septic arthritis	Not septic arthritis
Discharge diagnosis
Discharge diagnosis septic arthritis	83	26
Discharge diagnosis not septic arthritis	2	32
Antibiotic usage
Choice of antibiotics documented	73	24
Choice of antibiotics not documented	12	34

Discussion

In this large retrospective study of 268 patients with suspected septic arthritis, 85 had a confirmed diagnosis of true septic arthritis, although few clinical factors at the time of presentation, if any, were helpful in predicting this. As a consequence, relying solely on the initial clinical and laboratory parameters can make it difficult to diagnose septic arthritis in a safe and timely manner [11–13]. Although conventional laboratory parameters, such as WCC, CRP and ESR, were higher in the true septic arthritis group, they were broadly distributed. There was also little difference in the pre-operative synovial fluid WCC, indicating its poor ability to discriminate septic arthritis from other pathology.

This study quantifies the crucial role of pursuing an intra-operative synovial tissue biopsy specimen to establish a diagnosis of septic arthritis, particularly if pre-operative synovial fluid is culture negative. Amongst 63 patients who had a negative synovial aspirate but proceeded to intra-operative specimen collection, an additional six true septic arthritis diagnoses were made, all of which were based on biopsy specimens. This would suggest that, for patients returning a negative pre-operative synovial aspirate, 10.5 patients would have to be sampled intra-operatively to diagnose one additional case of true septic arthritis. For this to be prospectively applicable to such patients, it must be known that pre-operative synovial fluid cultures of such patients are negative when the decision to sample intra-operatively occurs, and therefore it is reliant on the likelihood of intra-operative culture in true septic arthritis patients remaining constant over time throughout the period of investigation in this cohort. Although it is uncertain from these data whether this is the case, and the effect that intervening antibiotic treatment might have, these results would suggest that, in patients considered for a diagnosis of septic arthritis but with a negative synovial aspirate and no intra-operative sampling, that the latter should be undertaken.

Furthermore, this study has demonstrated the utility of synovial tissue biopsy when in theatre, given that all six of the patients confirmed with true septic arthritis intra-operatively were confirmed with tissue biopsy. If operative intervention has been decided upon without knowing pre-operative synovial fluid culture status, 23.9 patients would have to undergo biopsy in order to diagnose one case of true septic arthritis. Given the limited morbidity associated with this, we feel this is highly worthwhile, particularly given that no patients were diagnosed with true septic arthritis from intra-operative synovial fluid. Finally, significant overdiagnosis and overtreatment for presumed septic arthritis was observed in this cohort, further emphasizing the importance of accuracy in diagnosis to avoid treatment-related morbidity and missing other important underlying diagnoses.

The majority of previously reported studies in this field are both retrospective and without control groups [9, 14–16]. In these, Gram staining of the synovial fluid has been found to be relatively insensitive, and there is general agreement in the literature that the diagnosis of septic arthritis should be based upon a positive culture of synovial fluid [9]. However, it is not always possible to obtain a positive culture from aspirated synovial fluid in patients with septic arthritis [3, 9, 14].

Although there are no definitive clinical risk factors nor routine laboratory investigations that can differentiate patients reliably, our study has confirmed, similar to previous studies, that some risk factors might be of utility in increasing the pre-test probability of septic arthritis [13, 17, 18]. These included a history of IVDU, preceding trauma to the affected joint and positive blood cultures during admission [13]. Laboratory investigations, including median serum WCC, CRP and ESR, were numerically higher in the septic arthritis group, but the results were widely distributed and not statistically different between the groups [11–13].

Some studies have suggested that an intra-operative synovial tissue biopsy might improve the likelihood of obtaining a positive culture from the affected joint [13]. A smaller study of 30 samples in 2012 found that synovial biopsy alone can guide eventual diagnosis of numerous joint pathologies in 12% of cases [14]. In that study, when patients were taken to theatre and prospectively cultured from both synovial tissue and fluid, there was a definite benefit [14]. In the majority of patients, there was a definite diagnosis when patients were taken to theatre [14]. Therefore, obtaining a positive culture from the joint concerned remains pivotal in making the diagnosis of septic arthritis.

A positive culture is key to identifying the causative organism, guiding antibiotic choice and duration and preventing morbidity [15]. With an NNT of 10.5, the intra-operative specimen plays a crucial role in diagnosis in those where the clinician has sufficient suspicion of septic arthritis. When an intra-operative specimen is collected, it is worth pursuing a synovial tissue biopsy because there is benefit to increasing the yield of a positive specimen (NNT 23.9). This is paramount for patients, because the consequences of undiagnosed and untreated septic arthritis are significant morbidity and even mortality [15]. Conversely, overdiagnosis without sufficient evidence can lead to unnecessary antibiotic usage, inadequate treatment of the true underlying joint pathology (e.g. crystal arthropathy) and inaccuracies in communication between hospital and primary care physicians.

There are a number of potential limitations to this retrospective single-centre study that can affect the generalizability of the results. The patients were selected from the orthopaedic referral database in a large tertiary referral centre, which is likely to includes patients with a higher likelihood of septic arthritis than the average ED presentation. This analysis is therefore reliant on the orthopaedic team to identify patients to incorporate into the database; consequently, it is possible that this cohort has been selected for more serious or complicated cases. Also, given that patients with joint prostheses require specialized expertise for safe pre-operative synovial aspiration, which might be available only in a tertiary centre, analyses based on joint prostheses might have been disproportionately subject to residual confounding. Given this, they might not necessarily be valid, and we did not undertake them. It is particularly possible that synovial WCCs might have been lower in patients with prostheses, given the relative absence of synovial tissue, and might have diluted the prognostic capabilities of this parameter. Although we were able to review antibiotic administration that occurred during inpatient hospital admissions, and we considered this in expert assessment of the septic arthritis diagnosis, antibiotics administered before hospital presentation were incompletely captured. This might have impacted aspirate and theatre specimen culture results, although given the inconsistency in this documentation, we sought not to assess this systematically, and prospective study should examine this further. The study definition of septic arthritis is also the best estimate of septic arthritis that could be made, based on both the investigations performed at the time of diagnosis and the independent retrospective review of the samples by an experienced microbiologist. In this study, we examined the subset of patients who had both a synovial aspirate and intraoperative specimen collected, which is an inherently different subgroup from those who have had only one or none of these investigations performed. This might have been for a multitude of reasons, including the clinical condition of the patient, clinician availability and patient preference. However, not only was this necessary to assess the utility of these investigations, but also it is perhaps in this subgroup (in the middle ground of suspicion) where examining the role of these investigations is of the greatest utility. Those who are taken directly for intra-operative joint washout are often more clear-cut cases of septic arthritis, hence less reliant on these investigations for their diagnosis and management.

Despite the clear value of intra-operative sampling for culture demonstrated in an unselected patient population with possible septic arthritis, there remains a clear need for better prognostication before samples are taken. It remains possible that pre-operative clinical and laboratory features might better predict patients who might have septic arthritis, including the use of new biomarkers not routinely in practice, such as calprotectin [16]. Recently, machine learning has been used to distinguish such patients better [17], but this should be externally validated and refined in larger collaborative data sets. Such further efforts might help to refine appropriate candidates for intra-operative intervention, but until they are appropriately regulated there remains clear value in routine clinical practice in ensuring that all patients undergo intra-operative specimen collection, in order to minimize overdiagnosis and its sequelae. Especially with the availability of arthroscopic intervention to minimize patient trauma, endeavouring to collect an intra-operative specimen is valuable.

In conclusion, this retrospective review shows that the addition of culture from intra-operative synovial tissue biopsy increases the certainty of making the diagnosis of septic arthritis. Diagnostic accuracy can be improved by taking a detailed history of relevant risk factors, collecting peripheral blood cultures, performing a timely joint aspiration before antibiotic exposure and having both an intra-operative synovial fluid and tissue biopsy collected. Future studies should include ED-based prospective analyses of current and emerging investigative strategies to improve accuracy.

Supplementary material

Supplementary material is available at Rheumatology Advances in Practice online.

Data availability

The data that support the findings of this study are available on request from the corresponding author, P.S. The data are not publicly available because they contain information that could compromise the privacy of research participants.

Funding

This paper has been published as part of the Rheumatology Advances in Practice Trainee Publishing Programme, which is supported by a grant from Biogen. No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: C.M. has received payment for advisory board participation from Abbvie. C.E.O. has received speaking honoraria from Abbvie, Janssen, Roche and Novartis and financial support from Roche for attendance at ACR 2019. The remaining authors have declared no conflicts of interest.