Comparison of equine synovial sepsis rate following intrasynovial injection in ambulatory versus hospital settings

Summary

Background:

Frequency of synovial sepsis in horses following intrasynovial injection has been reported, but not compared with respect to environment in which the injection was performed.

Objectives:

To describe occurrence of synovial sepsis following intrasynovial injections performed in ambulatory versus hospital settings.

Study design:

Retrospective cohort study.

Methods:

Records from Colorado State University were evaluated (2014-2018) and horses receiving intrasynovial injections identified. Patients presenting for septic synovial structures were excluded. Patient signalment, primary supervising service, medications injected, location (field/hospital), whether synovial sepsis resulted, and at what time sepsis was recognised were recorded. Logistic regression was used to estimate the contributions of covariates to the occurrence of synovial sepsis following injection.

Results:

During the study period, 3866 intrasynovial injections were performed in 1112 horses during 1623 sessions, with 643/1623 sessions performed in the field. The most frequently used medications were hyaluronate (846/1623, 52.1%), triamcinolone acetonide (780 /1623, 48.1%), and amikacin sulfate (684/1623, 42.1%). Four horses developed synovial sepsis (0.2% sessions, 0.1% synovial structures); 3/4 were injected in the field, 2/4 received antibiotics with the injection. The frequency of septic synovitis was 10.4 cases per 10,000 injections, or 1 in 967 injections. All horses recovered following synovial lavage and antibiotic therapy. Performing injections in the field (p=0.2) or without antibiotics (p=0.7) did not alter risk of synovial sepsis.

Main limitations:

Limitations include the retrospective nature of data collection and low rate of infection overall, which prohibited evaluation of individual medication regimes as factors associated with resultant infection.

Conclusions:

The frequency of synovial sepsis in this population of horses was not higher when injections were performed in the field or without concurrent antibiotic administration. These data may help to inform practitioners and clients regarding the relative potential risk of complications following intrasynovial medication in different environmental settings.

Introduction

Synovial sepsis represents a potentially life-threatening complication following intrasynovial medication in equine practice.¹ Bacterial species most frequently cultured depend on the etiology of infection, with Staphylococci and Streptococci spp. being more commonly isolated following injection or surgery, and Enterobacteriaceae spp. with traumatic injury.^2,3 Current treatments include arthroscopic or needle lavage with debridement of purulent material and antibiotic therapy administered systemically, intrasynovially and by regional limb perfusion.³ Survival rates have been reported to be as high as 85 to 90% with aggressive therapy,^3–5 but are reduced when multidrug resistant and Gram-negative isolates are cultured.⁶ The increasing incidence of infections with multidrug antibiotic resistance in equine practice has been recently documented in multiple reports.^7–17 Although further documentation of resistance patterns in synovial sepsis specifically are warranted, these findings indicate that an infection resulting from joint injection may have even more severe consequences for patient morbidity and mortality as well as increase cost of treatment to the client. These findings prompt re-evaluation of clinical practices towards identifying factors that may reduce the likelihood of infection resulting following injection.

Aspects of case management that have been previously associated with increased risk of septic synovitis include lack of aseptic technique with sterile gloves, needles and syringes, having another individual prepare the injection site, experience level of practitioner, use of multi-dose medication vials, removal of hair from the injection site, angle of needle insertion, needle gauge, and not using the stylet when inserting spinal needles.^18–24 Smith et al. recently reported the synovial infection rate in ambulatory practice alone (0.04%),²⁵ which was not found to be higher than previous studies in horses (0.02 to 0.08% of injections) or people (0.001 to 0.04%).^{18,19,25–30} Prophylactic intrasynovial antibiotic administration has not been demonstrated in any retrospective study to reduce the risk of synovial infection following injection, with the exception of polysulfated glycosaminoglycans (PSGAGs).^18,19,25,31 However, in a survey of equine practitioners in 2009, 46.2% reported using antibiotics such as amikacin sulphate in all synovial injections, with 78.6% of equine veterinarians including antibiotics in some situations, most commonly with corticosteroids or PSGAGs.³² Gillespie et al. reported similar findings five years later in a large retrospective survey of equine veterinarians performing 319,760 joint injections, where 46.5% described using antibiotics prophylactically.¹⁸ One rationale proposed for the inclusion of antibiotics is the potential for greater environmental contamination when performing injections in ambulatory practice, despite evidence that aseptic preparation by multiple techniques reduces bacterial flora on equine skin to similar levels regardless of the initial degree of contamination.²² The increasing incidence of antibiotic resistance in equine practice should encourage reassessment of prophylactic antibiotic use in many situations and makes the practice of routine use in synovial injections controversial due to the low risk of infection even without concurrent antibiotic administration.

Previous studies have not compared rates of infection directly between injections performed in ambulatory versus hospital settings or in regard to concurrent antibiotic usage in either environmental setting. This information is of clinical importance to veterinarians performing personal audits of complication rates within their practice and in communicating with clients on the relative risk of complications following synovial injections in different conditions. Therefore, the objectives of this study were to retrospectively compare rates of synovial sepsis following intrasynovial injection in field versus hospital settings, and to determine whether concurrent use of antibiotics reduced that risk in either situation. We hypothesised that the rate of synovial infection following injection would not be higher in ambulatory practice and would not be reduced by prophylactic intrasynovial antibiotic administration.

Materials and Methods

Data collection and inclusion criteria -

Medical records (2014 to 2018, inclusive) of all horses presenting to the Surgery, Sports Medicine and Ambulatory services at Colorado State University were retrospectively reviewed. Horses that received injections of synovial structures, including joints, bursas and tendon sheaths, either for diagnostic anaesthesia during lameness evaluation or medication to treat osteoarthritis, bursitis and tenosynovitis were recorded. Horses presenting for treatment of septic synovitis or receiving injections of the sacroiliac joints or interspinous regions were excluded. Data retrieved from the medical records included patient signalment, clinician, medications injected, joint and limb involved, concurrent use of antimicrobials, location of horse at the time of injection (field/hospital), whether synovial sepsis developed following injection, time following injection at which the septic synovial structure was recognised and treatment for sepsis if performed. Data were recorded via computer-based spreadsheet (Microsoft Office Excel, Microsoft Corporation, Redmond, Washington, USA) and subsequently transferred to statistical program (R version R.0.0 and 4.0.3, R Foundation, Vienna, Austria) for analysis.

The criteria used to determine whether sepsis of the injected synovial structure occurred included any one abnormal clinical parameter (lameness of the affected limb, peri-synovial soft tissue swelling, synovial effusion, and/or fever) in addition to abnormal synovial fluid analysis (at least two of the following parameters: elevated synovial fluid nucleated cell count >10 **x 10⁹/L, total protein >25 g/L, neutrophil percentage >80%, and/or positive bacterial culture) and failure to resolve with conservative therapy such as nonsteroidal anti-inflammatory medications, as previously described.²⁵

Joint injection technique –

General hospital protocols for intrasynovial injections include implementation of aseptic technique using sterile gloves, needles and syringes, and use of new medication vials in all cases. Clipping of hair over the injection site was not routinely performed but was not recorded in the medical record in the majority of cases and therefore not included in the final analysis. The injection site was aseptically prepared using betadine or chlorhexidine gluconate followed by alcohol for a minimum of seven minutes as previously reported.¹⁸ The concurrent use of antibiotics such as amikacin sulphate was based on faculty clinician preference. It is general hospital protocol to bandage the medication site until the next day following medication of synovial structures distal to the carpus or tarsus; however, as this was not included in the medical record in some cases, this variable was not included in the final analysis.

Data analysis -

The contribution of environmental setting, use of antibiotics, and primary service to the incidence of synovial sepsis was analysed using logistic regression with the binary outcome of sepsis (yes/no). Due to the few recorded cases of sepsis and presence of separated data, a penalised maximum likelihood (PML) method was used for estimation of the parameters.³³ The R package logistf was used to produce parameter estimates for the models with the resultant odds ratios and their 95% confidence intervals calculated by hand as exp(coef).³⁴ Statistical analysis was performed using R version 4.0.0 “Arbor Day” and 4.0.3 “Bunny-Wunnies Freak Out” (R core team).³⁵

Models evaluated included: 1 - service (surgery, sports medicine, or ambulatory), location (field/hospital), antibiotic use (yes/no) and an interaction term between location and antibiotic use, 2 – the same model without the interaction term, and 3 – a model with only location and antibiotic use. Models were compared with Akaike Information Criterion (AIC) as well as significance of factors. Due to the lack of significance the interaction term was dropped, and as service as a predictor was also not significant, this parameter was removed from the final model as well, leaving location and antibiotic use as predictors. This final model also had the lowest AIC value of the 3 models evaluated.

Results

Over the time period evaluated, 1112 horses received 3866 intrasynovial injections in 1623 sessions (Figure 1). Of the 1623 sessions identified, 643 were performed in the field while 980 were performed inside in the hospital environment. Of the sessions performed in the field, 278/643 (43.27% received concurrent antibiotic administration, one of which resulted in synovial sepsis (0.36%, CI = 0 – 1.32%). Of the remaining 365/643 (56.72%) field sessions where antibiotics were not administered prophylactically, 2 horses developed synovial sepsis (0.55%, CI = 0.02 - 2.11%). Of the sessions performed in the hospital, 406/980 (41.4%) received concurrent antibiotic administration, one of which resulted in synovial sepsis (0.2%, CI = 0 - 0.90%). Of the remaining 574/980 (58.6%) synovial injection sessions performed in the hospital where antibiotics were not administered, no horses developed sepsis.

Figure 1:

Flow diagram demonstrating the number of horses receiving intrasynovial medications in injection sessions in either field or hospital settings and those injections that resulted in infection. Over the time period evaluated (2014-2018), 3866 intrasynovial injections were performed in 1112 horses in 1623 medication sessions, resulting in four cases of synovial sepsis following injection (0.1% injections, 0.2% sessions).

The risk of septic arthritis was 10.4 cases per 10,000 injections, or 1 in 967 injections. Four instances of post-injection synovial sepsis occurred in four different horses, indicating a proportion of total medication sessions of 0.2% and total synovial injections of 0.1%. Three of the four cases that resulted in synovial sepsis were performed in the field (3/643 or 0.5 (0.09 - 1.43)% sessions performed in the field; 1/980 or 0.1 (0-0.37)% sessions performed in the hospital). Two of the cases that resulted in synovial sepsis (one in field and one in the hospital) received concurrent prophylactic intra-synovial antibiotics (125 mg amikacin sulfate in both cases). Medications injected and synovial fluid parameters in cases diagnosed with sepsis are summarised in Table 1. Post-injection synovial sepsis occurred in the tarsocrural joint of a 12-year-old Shire gelding injected with autologous conditioned serum, the distal interphalangeal joint of a 12-year-old Quarter Horse gelding injected with triamcinolone acetonide, hyaluronate, and amikacin sulfate, the proximal interphalangeal joint of a 13-year-old Quarter Horse gelding injected with triamcinolone acetonide, hyaluronate, and amikacin sulfate, and the digital flexor tendon sheath of a 12-year-old Quarter Horse gelding injected with platelet rich plasma. In the four horses developing synovial sepsis, two occurred after a single synovial structure was injected and two occurred after multiple structures were injected (two or six structures). The four horses that developed sepsis received treatment which consisted of arthroscopic (n=3) or needle lavage (n=1), intrasynovial antibiotic administration (n=4), regional limb perfusion (n=4), systemic intravenous antibiotics (n=3) and/or oral antibiotics (n=4) and nonsteroidal anti-inflammatories (n=4). Three of the four cases were referred to the hospital for management and were hospitalised 7, 10 or 11 days for treatment. All horses recovered and returned to athletic activity following treatment.

Table 1:

Summary of medications injected, location, and synovial fluid parameters in cases diagnosed with synovial sepsis following injection. Abbreviations: Total Nucleated Cell Count (TNCC); Total Protein (TP).

		Synovial Fluid Parameters
Patient	Medications injected	Location injected	TNCC (cells x 109/L	TP (g/L)	Neutrophils (%)	Culture Submitted (Y/N)
1	Autologous conditioned serum	Field	57	45	97	No
2	Triamcinolone acetate, hyaluronate, amikacin sulfate	Field	61.3	50	82	No
3	Triamcinolone acetate, hyaluronate, amikacin sulfate	Hospital	45	46	85	No
4	Platelet rich plasma	Field	37.4	47	94	Yes; negative

A total of eleven faculty clinicians on the Surgery, Sports Medicine and Ambulatory services working with seventeen house officers (residents, interns) over the time period investigated administered the joint medications. Four different senior clinician and house officer pairs were each associated with a single sepsis case each. Synovial sepsis was detected on day eight (n=1) or day nine (n=3) post-medication. The effect of clinician on frequency of synovial sepsis was not assessed as records did not indicate which clinician performed the actual injection in most cases. The most frequently administered medications were hyaluronate (846/1623, 52.1%), triamcinolone acetonide (780 /1623, 48.1%), and amikacin sulfate (684/1623, 42.1%). Distribution of frequency of medications administration in either field or hospital settings and their relative proportion in injections that resulted in infection are summarised in Table 2.

Table 2:

Summary of intrasynovial medications administered in 1623 medication sessions (643 field, 980 hospital) aggregated at level of session in 1112 horses between 2014 to 2018. Confidence intervals were calculated using the Agresti – Coull modification of the Wald formula or the Ott & Longnecker recommended adjustment to the confidence interval.

Intrasynovial Medication	Location	Total # Sessions	Percentage Sessions (%)	Total # Sessions with Antibiotics	Percentage Sessions with Antibiotics (%)	# Septic	Percent Septic (%)	Population Extrapolated CI (%)
Triamcinolone acetonide	Field	280	17.25	177	63.21	1	0.36	0 – 1.31
	Hospital	500	30.81	319	63.80	1	0.20	0 – 0.74
Methylprednisolone acetate	Field	91	5.6	20	21.98	0	0	0 – 3.97
	Hospital	40	2.5	21	52.50	0	0	0 – 8.81
Dexamethasone phosphate	Field	2	0.1	0	0	0	0	0 – 84.19
	Hospital	0	0	0	0	0	0	NA
Betamethasone	Field	22	1.4	0	0	0	0	0 – 15.44
	Hospital	0	0	0	0	0	0	NA
Stanazolol	Field	1	0.06	1	100.00	0	0	0 – 97.50
	Hospital	2	0.1	1	50.00	0	0	0 – 84.19
Autologous conditioned serum	Field	114	7	10	8.77	1	0.88	0 – 3.18
	Hospital	125	7.7	9	7.20	0	0	0 – 2.91
Platelet-rich plasma	Field	1	0.06	0	0	1	100.00	2.50 – 100
	Hospital	9	0.6	0	0	0	0	0 – 33.63
Mesenchymal stromal cells	Field	8	0.5	0	0	0	0	0 – 36.94
	Hospital	56	3.5	3	5.36	0	0	0 – 6.37
Autologous protein solution	Field	0	0	0	0	0	0	NA
	Hospital	39	2.4	7	17.95	0	0	0 – 9.03
Hyaluranate	Field	351	21.6	182	51.85	1	0.28	0 – 1.05
	Hospital	495	30.5	318	64.24	1	0.20	0 – 0.74
Amikacin sulphate	Field	278	17.1	278	100.00	1	0.36	0 – 1.32
	Hospital	406	25	406	100.00	1	0.25	0 – 0.90
Polysulphated glycosaminoglycan	Field	1	0.06	1	100.00	0	0	0 – 97.50
	Hospital	2	0.1	2	100.00	0	0	0 – 84.19
Polyacrylamide hydrogel	Field	1	0.06	1	100.00	0	0	0 – 97.50
	Hospital	4	0.2	2	50.00	0	0	0 – 60.24
Mepivacaine	Field	146	9.0	55	37.67	0	0	0 – 2.49
	Hospital	184	11.3	21	11.41	0	0	0 – 1.98

There were no statistically significant differences with regards to development of synovial sepsis in field versus hospital settings (OR 1.3, 95% CI 0.6-33.6, p = 0.2) or with or without the inclusion of antibiotics in the injectate (OR 3.5, 95% CI 0.2 – 8.6, p = 0.7). Model coefficients for the final model are shown in Table S1

Discussion

This study described the rate of synovial sepsis in equine practice in field versus hospital environments and the relative risk for development of infection with or without antibiotic administration in either scenario. The frequency of synovial sepsis overall in the population was low (0.1%) and similar to previous reports in horses (0.02-0.08%)^18,19,25,31 and human subjects (0.04%).²⁶ Differences in injection technique, case inclusion, reporting, experience level of individuals performing injections, or record keeping given the retrospective nature of study design may account for differences to previous reports and range in rates published. The risk of infection was no different in sessions performed in the field or hospital and was not reduced by concurrent antibiotic administration in either setting. As this study was performed in a single clinic where patients are evaluated in both environments, other factors that may affect the risk of infection such as aseptic technique were maintained relatively consistent between clinicians and injection sessions.

Synovial injections were aggregated at the level of medication sessions, rather than individual joint, as all structures injected in each session were performed using the same aseptic equipment and technique and were injected concurrently, as previously described.²⁵ Two horses developed infection when a single synovial structure was injected, and two developed infection when multiple structures were injected, so there did not appear to be a linear effect of increased risk of infection when more structures were entered. Time to detection of sepsis was eight (n=1) or nine days (n=3), corroborating previous reports that the anti-inflammatory effects of some commonly used intrasynovial medications can delay detection.^25,36–38 The overall rate of infection reported here was similar to but slightly higher than that by reported by Gillespie et al. (2.1 out of 10,000)¹⁸ or Steel et al. (7.8 out of 10,000 injections)¹⁹, which may be attributed to a number of factors including the inclusion of bursas and tendon sheaths in this retrospective analysis compared to others¹⁸, the inexperience of individuals, frequently trainees, actually performing the injections in most cases, and time period evaluated.

Antibiotics were administered concurrently in 42.1% of injection sessions overall, similar to the frequency previously reported in two large surveys of equine practitioners (46.2 and 46.5%).^18,32 An association between polysulphated glycosaminoglycans (PSGAG) injection and synovial sepsis was previously reported in experimental studies and recently corroborated in a retrospective series of intrasynovial injections performed in ambulatory situations, leading to the recommendation that antibiotics should be administered concurrently with PSGAG medications.^25,39,40 Antibiotics were administered concurrently in all three injections in this study where PSGAGs were used. Medication to environment interactions for infrequently injected medications such as PSGAGs (n=3) or PRP in the field (n=1) could not be assessed from this study population given the low frequency of use. Local anaesthetics have recently been demonstrated in vitro to have antimicrobial activity against common equine pathogens at concentrations clinically applied in joint injections, further calling into question whether the prophylactic use of antimicrobials in local anaesthetic injections is justified.⁴¹ However, it was acknowledged that one of the most common bacterial pathogens in iatrogenic septic arthritis, S. aureus, was also least susceptible to the bactericidal action of local anaesthetics, and that further investigation in vivo is warranted.⁴¹

Injection of synovial structures in the field was not found to increase the risk of infection in this case population over the time period evaluated, corroborating one recent study reporting a similar rate of synovial infection in ambulatory practice (0.04%) compared to that reported overall.²⁵ While it is assumed that the field environment presents a greater risk for contamination with higher room fallout of bacterial flora than the hospital (<20 CFU/h in surgical suites), this has not been compared quantitatively and was not performed in this study due to the retrospective nature of study design.^22,42 Environmental sampling to determine bacterial air fallout in situations where synovial infections have resulted following injection versus hospital or operating room settings may help to guide whether higher air bacterial load actually poses a greater risk in which to perform injections. However, Gillespie et al. previously reported prepping the injection site for at least seven minutes, preparing the site themselves, not removing hair from the site and wearing gloves to be factors associated with a decreased risk of infection by univariate analysis.¹⁸ Zubrod et al. expanded on this concept to demonstrate that aseptic preparation by multiple methods was sufficient to reduce bacterial load at the injection site regardless of initial degree of contamination.²² These studies support the concept that aspects of synovial injections within the veterinarian’s control, such as appropriate site preparation and aseptic technique, are likely more important to minimise risk of infection than environment in which the joint is injected.

Limitations of this study include the retrospective nature of data collection and low rate of infection overall, which prohibited evaluation of individual medication regimes as factors associated with resultant infection. As owners were not directly contacted, it is possible that additional follow-up may have allowed discovery of more complete information regarding complications and outcomes, including whether owners elected treatment elsewhere following development of synovial sepsis. The definition of synovial sepsis is controversial and may represent a limitation in that some horses deemed to have inflammatory reactions or ‘flares’ may have resolved with more conservative management or cases considered septic in this retrospective study may have actually been non-septic reactions as previously discussed.^19,43 However, the criteria used here to define ‘sepsis’ has been previously reported and used for retrospective designation of synovial infection, and all cases described as septic met those criteria for total nucleated cell count, total protein and percentage of neutrophils in synovial fluid.²⁵

Finally, the relatively rare occurrence of synovial infection following injection, particularly regarding an increased risk with specific medications, necessitates caution in interpretation or extrapolation to populations. Due to the limited number of injections examined combined with the low prevalence of post-injection infection, it is possible that potential effects may not be appreciated with statistical analysis. Due to the few recorded cases of sepsis and presence of separated data, a penalised maximum likelihood (PML) method was used for estimation of the parameters.³³ This method of analysis has been shown to give 95% confidence intervals (CI) that closely approximate the true rates of occurrence in actual and simulated data sets, with estimated CI coverage of 93.4 – 97.7% in one study.³³ The PML method is also far superior to ordinary multivariate logistic regression (OLR) in data where near-complete or complete separation is present, as OLR often overestimates the odds ratios and produces infinite confidence intervals in these situations. In addition, the sample used for analysis represents a convenience sample of all cases seen over a period of time. It is always possible that extending out the time period for case collection and data analysis might allow for different results, however the odds ratios and their resultant confidence intervals are wide enough that the authors do not believe additional cases would significantly alter the results presented here. These findings add to the body of literature regarding relative risk of intrasynovial injection in equine practice in different circumstances and with or without the addition of antibiotics.

In conclusion, the frequency of synovial sepsis following intrasynovial injection in this population of horses was not higher when injections were performed in the field or without antibiotics. These data may be useful to equine practitioners in examining their own practice’s complication rate and medication administration practices and informing clients about factors associated with risk of synovial infection following injection.