Cellulitis is a generic term for an infection involving the dermis and subcutaneous connective tissue.1 It is a commonly diagnosed condition of horses and one that can cause heat, severe pain, reluctance to move, and substantial swelling of the affected limb.2 Because of these clinical signs, it is often difficult to determine whether there is an accompanying joint infection or synovitis. A confirmed diagnosis of synovitis or infection within a joint can only be made by assessing various characteristics of joint fluid obtained via arthrocentesis.
Arthrocentesis performed through cellulitic tissue is perceived to be risky in that needles advanced through infected tissue (skin and subepidermal layers) may deposit bacteria into an otherwise normal joint that may subsequently become infected.3 For this reason, arthrocentesis is often postponed2 and horses are aggressively treated with antimicrobials, NSAIDs, cold-water treatments, and wraps in an effort to reduce infective load and swelling (and thus risk) before arthrocentesis is performed. Unfortunately, the delay associated with attempting to reduce this risk may result in a failure to diagnose and treat an established joint infection in a timely manner, which may lead to loss of an animal.4
The most common pathogen in cellulitis2 or iatrogenic joint sepsis3 reported is Staphylococcus aureus. This organism has been used in models of joint sepsis at inoculums of between 1.5 × 105 CFUs and 2.16 × 106 CFUs.5–7 Although a dose as small as 1 × 102 CFUs has resulted in classical sepsis in 4 of 5 inoculated horses, the minimum infective dose has been determined as 33 CFUs of S aureus.8,9
The objective of the study reported here was to determine whether insertion of needles of various sizes through known concentrations of inoculated culture media would result in bacterial translocation sufficient to reach the minimum infective dose required to induce septic arthritis. There were 2 hypotheses. First, the number of bacteria translocated by a needle increases as the size of the needle or concentration of bacteria in the medium (or both) increases. Second, when needles are inserted through culture medium inoculated with a bacterial concentration equal to that of possibly infected tissue (ie, 105 CFUs/g of tissue),10 bacterial translocation does not exceed 33 CFUs of bacteria.8,9
Materials and Methods
Sample
A hot needle was used to melt 3 small holes, equidistant from the center and from each other, through the bottom of sterile empty Petri plates. The holes were immediately covered with tape,a and the plates were filled with agar (depth of agar, 3.2 mm) that contained a known concentration of S aureus (103, 104, 105, and 106 CFUs/mL). The agar was inoculated by use of a seed culture that consisted of 10 mL of brain-heart infusion broth and S aureus (attenuated cell culture strain 25923) incubated at 37°C for 18 hours.11 Bacterial concentration of the seed culture was standardized before inoculation into the agar. A spectrophotometerb set to a wavelength of 600 nm was used to determine OD of the seed culture. Serial dilutions of the seed culture were inoculated on sheep blood agar; concentrations were replicated and confirmed on the basis of the OD.
To maintain accuracy during manual plate counts, a plate must yield between 30 and 300 CFUs.12,13 Although the specific counts for this range differ slightly, the difference between the number of expected and observed CFUs diverges outside of this range.12,13 To achieve values within the range of 30 to 300 CFUs, several serial dilutions must be created, with each dilution outside of this range being discarded. Serial dilutions of a standard stock solution (1012 CFUs/mL) were inoculated into agar, which resulted in plate concentrations of 103, 104, 105, and 106 CFUs/mL. Four milliliters of each concentration were poured into a tube containing 16 mL of melted Mueller-Hinton agar (maintained at 45°C). Mixing of the diluted standard plate concentrations with the melted agar resulted in a 1:4 dilution, which was accounted for when plating. The suspension was homogenized, each tube was inverted 10 times, and the contents were immediately poured into the aforementioned sterile Petri dishes in which the holes had been covered with tape. The plates were placed on a flat surface in a refrigerator (4°C) to minimize thermal bacterial killing and allowed to cool.
Once the agar was completely solidified, plates were allowed to equilibrate to room temperature (approx 20°C); the bottom of each plate was aseptically prepared with three 30-second washes with chlorhexidine scrub.14 The bottom of each plate was wiped with alcohol after each wash. After the third wash, plates were allowed to dry.
Procedures
Noncoated, 1.5-inch, 19-, 20-, or 22-gauge needlesc were aseptically inserted through the tape and bacteria-containing agar (1 needle gauge for each hole of a Petri dish); all needle insertions were performed in triplicate and by the same investigator (TTS). After the needle emerged through the agar, a 3-mL syringe with the plunger extended was attached to the hub of the needle. The plunger was depressed, which thereby ejected any agar core that was present within the needle bore into a sterile, preweighed test tube. Sterile shears were used to transect the needle at the surface of the agar; that portion of the contaminated needle was placed into the same sterile test tube as the agar core. A scale sensitive to 4 decimal places was used to weigh each test tube and determine the weight of the needle fragment and any agar.
A 1:10 dilution of sterile saline solution (ie, 1 g of sample weight to 9 mL of saline solution) was used to rinse the needle fragment and for emulsification of the agar. The material was mixed in a vortex device, and the resulting solution was serially diluted and inoculated onto sheep blood agar; plates were incubated, and the colonies were manually counted. Only colonies containing the characteristic double-halo of S aureus were counted. Culturing was performed in accordance with strict standards of asepsis, and bacterial contamination was considered negligible.
To determine the number of bacteria lost throughout the process, a killing curve was created. A portion of the standard stock solution was serially diluted and inoculated onto blood agar; results were used to determine the bacterial concentration of the standard stock solution. Another portion of the standard stock solution was evaluated after use in the experimental procedures, from inoculation into melted agar to re-inoculation of the emulsified end product. The number of colonies on these plates was then manually counted, and the number of viable colonies was compared. Typically, 105 bacteria/mL were lost during the experimental procedures.
Three plates of each concentration of S aureus (no bacteria [control sample] and 103, 104, 105, and 106 CFUs/mL) were created. A sterile scalpel was used to divide each plate into 9 equal sections. Three randomly selected sections (as determined by use of a random number generator) from each of the plates were excised and emulsified. The emulsified agar of each concentration was serially diluted and inoculated onto plates containing sheep blood agar. Each plate of the same dilution contained a similar number of CFUs (as determined on the basis of the manual count), which established that there was uniformity throughout the agar. This also confirmed that each of the concentrations was as calculated and that all noninoculated plates failed to yield bacterial growth.
Statistical analysis
Results were compiled by use of a commercially available spreadsheet programd and transported into a commercial statistical packagee for analysis. Manual counts of CFUs were logarithmically (base 10) transformed to normalize the data. Samples with > 33 CFUs of S aureus (after controlling for dilution) were classified as septic; the remainder were considered nonseptic.
Generalized estimating equations controlling for repeated observations were used to examine the effect of factors (needle size and S aureus concentration in agar) on the number of bacterial CFUs translocated after needle insertion through the bacterial-containing agar. Logistic regression analysis was used to evaluate factors for needles being able to translocate < or > 33 CFUs of S aureus (data reported as ORs). Factors were initially screened by use of unconditional analysis. Variables with a value of P < 0.2 were then considered in building the final multivariable model.15 All final models were built by use of manual backward elimination. When both variables had significant (P < 0.05) effects, biologically plausible 2-way interactions were assessed, with interactions retained in the final model if the type 3 likelihood ratio test was significant (P < 0.05).
Results
Agar without bacteria (control sample) failed to yield bacterial growth, irrespective of needle gauge.
Univariable analysis of factors affecting the log10 number of CFUs and sepsis (> 33 CFUs of S aureus) indicated that both a larger needle gauge and increased bacterial concentration of the agar were significantly associated with higher CFU counts and number of septic versus nonseptic outcomes. For the multivariable models, needle gauge and bacterial concentration of the agar had significant effects. For both the univariable and multivariable models, interaction terms were significant (P = 0.01 and P < 0.001, respectively).
Pairwise comparisons of estimated marginal means of log10 CFUs by plate concentration and needle gauge indicated that for agar containing 106 S aureus/mL, the 19-gauge needle translocated a significantly greater number of bacteria than did the 20- or 22-gauge needles (Table 1). However, all needle gauges translocated > 33 CFUs (1.518 log10 CFUs) of S aureus. For agar containing 105 S aureus/mL, only the 19-gauge needle translocated > 33 CFUs of S aureus. The 20- and 22-gauge needles translocated 0.891 and 0.518 log10 CFUs of S aureus, respectively.
Pairwise comparison of estimated marginal means of log10 CFUs of Staphylococcus aureus translocated during needle insertion through Mueller-Hinton agar containing 4 concentrations of Staphylococcus aureus.
| Â | Â | Â | 95% confidence interval (mean log10 CFUs) | |
|---|---|---|---|---|
| Plate concentration | Needle gauge | Bacterial (mean log10 CFUs) | Lower | Upper |
| 106 | 22 | 3.526a | 3.405 | 3.647 |
| Â | 20 | 3.627a | 3.585 | 3.669 |
| Â | 19 | 3.868b | 3.701 | 4.034 |
| 105 | 22 | 0.518a | –0.056 | 1.091 |
| Â | 20 | 0.891b | 0.094 | 1.688 |
| Â | 19 | 2.672c | 2.180 | 3.163 |
| 104 | 22 | 0.537a | 0.065 | 1.008 |
| Â | 20 | 0.504a | 0.378 | 0.630 |
| Â | 19 | 1.145b | 0.612 | 1.678 |
| 103 | 22 | 0.231a | 0.046 | 0.417 |
| Â | 20 | 0.000b | 0.000 | 0.000 |
| Â | 19 | 0.411a | 0.309 | 0.512 |
Within a plate concentration, values with different superscript letters differ significantly (P < 0.05).
Evaluation of ORs for translocation of > 33 CFUs of S aureus revealed that sufficient numbers of bacteria to induce infection were translocated for all bacterial concentrations (Table 2). For agar containing 103 S aureus/mL, the ORs for translocating > 33 CFUs of S aureus were > 0 only when the 19-gauge needle was used (OR, 0.11). For agar containing 104 S aureus/mL, there was no difference in the ORs for translocation between 22- and 20-gauge needles or between 20- and 19-gauge needles, but the OR for a 19-gauge needle translocating > 33 CFUs of S aureus was significantly higher than the OR for the 22-gauge needle. For agar containing 105 S aureus/mL, there was no difference in the ORs for translocation of > 33 CFUs of S aureus between 22- and 20-gauge needles, but the ORs for translocating > 33 CFUs of S aureus were greater for both 20- and 22-gauge needles than for the 19-gauge needle. For agar containing 106 S aureus/mL, the OR for translocation of > 33 CFUs of S aureus was 1.00 for all 3 needle gauges.
Mean OR for translocating > 33 CFUs of S aureus on the basis of plate concentration and needle gauge.
| Â | Â | Â | 95% Wald confidence interval | |
|---|---|---|---|---|
| Plate concentration | Needle gauge | OR for > 33 CFUs | Lower | Upper |
| 106 | 22 | 1.00 | 1.00 | 1.00 |
| Â | 20 | 1.00 | 1.00 | 1.00 |
| Â | 19 | 1.00 | 1.00 | 1.00 |
| 105 | 22 | 0.22a | 0.04 | 0.69 |
| Â | 20 | 0.22a | 0.04 | 0.69 |
| Â | 19 | 0.89b | 0.57 | 0.98 |
| 104 | 22 | 0.22a | 0.09 | 0.44 |
| Â | 20 | 0.22a | 0.09 | 0.44 |
| Â | 19 | 0.44b | 0.16 | 0.77 |
| 103 | 22 | 0 | 0.00 | 0.00 |
| Â | 20 | 0 | 0.00 | 0.00 |
| Â | 19 | 0.11 | 0.02 | 0.43 |
Within a plate concentration, values with different superscript letters differ significantly (P < 0.05).
Discussion
The act of arthrocentesis poses a risk of sepsis, even when the needles are inserted through apparently normal tissue. Authors of a retrospective study3 of 192 horses reported that approximately 70% of joints with postinjection sepsis could be attributed to contamination with S aureus. In the study reported here, we determined that substantial bacterial translocation (> 33 CFUs of S aureus) during arthrocentesis was dependent on both needle gauge and bacterial concentration within the tissue. A larger needle diameter and higher concentration of bacteria would be significantly associated with bacterial translocation sufficient to cause iatrogenic infection of an equine joint in vivo. However, the amount of bacterial translocation was not linear among all needle sizes, with the 20- and 22-gauge needles associated with significantly less bacterial translocation than for 19-gauge needles. This pattern was supported by results of another study16 in which investigators found that 19-gauge needles contaminated significantly more joints than did 20-gauge needles. The authors of that study16 also established that the use of silicon-coated needles significantly reduced overall joint contamination, compared with contamination resulting from the use of noncoated needles. Noncoated needles were used in the present study because it was the consensus among the authors that noncoated needles were the ones most widely used by practitioners.
Aseptic preparation of equine skin can dramatically reduce bacterial numbers14,17 and thus the risk of iatrogenic infection following arthrocentesis.18 In cellulitic tissue, however, aseptic preparation of the epidermis reduces the superficial bacterial load but has no effect on the infected dermis and subcutaneous tissue, which potentially allows arthrocentesis to introduce bacteria of sufficient numbers to induce septic arthritis.
The effect of needle size on translocation of hair and tissue debris or tissue plugs during arthrocentesis has been previously reported.16,18 In one of those studies,18 researchers found that many of the tissue fragments from both 20- and 22-gauge needles were < 200 μm in length, whereas fragments from needles with a larger diameter were often > 500 μm in length. Analysis of results of the present study indicated that 19-, 20-, and 22-gauge needles would translocate sufficient bacterial numbers to establish infection only when tissue had a bacterial concentration of ≥ 105 S aureus/mL. If the model we used accurately reflected the situation in live animals and the size of the agar plug in the present study was similar to the plug found in another study,18 then it is possible that the larger core associated with 19-gauge needles caused the increased rate of bacterial translocation. If this were true, then use of a needle with a stylet may reduce the chance of contamination within a joint.18
Procedures were used to ensure accurate and homogenous bacterial concentrations of the test agar. Although we recognize that the distribution of bacteria in vivo is unlikely to be uniform in nature, the uniform density for the model of the present study was derived from the typical bacterial concentration reported for cases of naturally developing cellulitis, which thus served as a method of bacterial quantification in vitro. The authors are not aware of any published reports of the concentration of bacteria in equine cellulitic tissue; however, information has been providedf that concentrations of 106 bacteria/g of tissue have been cultured from 2 separate horses. In the study reported here, all samples of our model were contaminated with > 33 CFUs of S aureus for agar containing 106 S aureus/mL, irrespective of needle size, which indicated that performing arthrocentesis through cellulitic tissue with this concentration of bacteria would appear to represent a substantial risk of inducing joint sepsis.
The depth of agar used to simulate tissue in this study (3.2 mm) was similar to that of tissue found over equine joints.g Median thickness of equine skin was 3.26 mm (range, 1.96 to 4.57 mm) and became progressively thicker from the proximal to distal aspect of a limb.g The insertion of needles through tissue is likely to be different from that for the agar used in the present study. Tissue is denser than agar, and the change in density that occurs while inserting a needle through connective tissue and the joint capsule may further increase tissue drag, which ultimately reduces the number of bacteria that are translocated into a joint. The argument can be made that cellulitic tissue is thicker than normal tissue, but it can be reasoned that inflamed tissue would be less dense than normal tissue, which would result in fewer bacteria per gram of tissue. In the in vitro model described here, skin (which was represented by tape) did not contain hair follicles, sweat glands, or sebaceous glands, all of which have been found to harbor bacteria after surgical preparation of tissue.19 Skin removed from cadavers was not used in the present in vitro study because there was a concern that there would be too much variability introduced into the experiments such that the data would be skewed because of inability to control all variables.
Extrapolation of data from the present in vitro study to clinical cases should be made with caution because we examined a worst-case scenario (with the exception of inserting a needle through an abscess). Other biological variables need to be taken into account in living animals. The presence of inflammation (as a result of periarticular inflammation, trauma, or osteoarthritis) or use of immunosuppressive drugs (eg, corticosteroids) may compromise joint defenses, which could lead to smaller numbers of bacteria required to establish infection.9 The authors recognize that no in vitro model can completely replace or mimic the use of live tissues for this kind of study, but in view of the restricted access to such tissues (eg, animals arriving at a clinic with similar lesions) as well as restrictions for use of animals in which investigators can create similar infections and experimental challenges as those seen in field settings, the study was intended to mimic the outcome for such tissues. Nevertheless, analysis of the in vitro data for the present study suggested that arthrocentesis without subsequent joint sepsis might be performed in horses with tissue concentrations of < 105 bacteria/mL by use of 20- or 22-gauge needles, rather than 19-gauge needles, although the narrow internal diameter of 22-gauge needles may make it difficult to obtain a synovial fluid sample.
Acknowledgments
Supported by Drs. Carmalt and Chirino-Trejo.
The authors declare that there were no conflicts of interest.
ABBREVIATION
| OD | Optical density |
Footnotes
Scotch tape, 3M Canada, London, ON, Canada.
NanoDrop2000C, Thermo Scientific, Wilmington, Del.
Standard point hypodermic needle, Sigma-Aldrich, St Louis, Mo.
Microsoft Excel, Microsoft Canada Inc, Mississauga, ON, Canada.
SPSS, IBM Canada Ltd, Markham, ON, Canada.
Marais J, Department of Companion Animal Clinical Studies, Faculty of Veterinary Science, University of Pretoria, Onderstepoort, Republic of South Africa: Personal communication, 2009.
Volkering ME. Variation of skin thickness over the equine body and the correlation between skin fold measurement and actual skin thickness. Doctorate of Veterinary Medicine thesis, Faculty of Veterinary Medicine, University of Utrecht, Utrecht, The Netherlands, 2009. Available at: dspace.library.uu.nl/handle/1874/37084. Accessed May 13, 2014.
References
1. Swartz MN. Cellulitis. N Engl J Med 2004; 350: 904–912.
2. Fjordbakk CT, Arroyo LG, Hewson J. Retrospective study of the clinical features of limb cellulitis in 63 horses. Vet Rec 2008; 162: 233–236.
3. Schneider RK, Bramlage LR, Moore LM, et al. A retrospective study of 192 horses affected with septic arthritis tenosynovitis. Equine Vet J 1992; 24: 436–442.
4. Wright IM, Smith MR, Humphrey DJ, et al. Endoscopic surgery in the treatment of contaminated and infected synovial cavities. Equine Vet J 2003; 35: 613–619.
5. Bertone AL, Jones RL, McIlwraith CW. Serum and synovial fluid steady-state concentrations of trimethoprim and sulfadiazine in horses with experimentally induced infectious arthritis. Am J Vet Res 1988; 49: 1681–1687.
6. Bertone AL, McIlwraith CW, Jones RL, et al. Povidone-iodine lavage treatment of experimentally induced equine infectious arthritis. Am J Vet Res 1987; 48: 712–715.
7. Bertone AL, McIlwraith CW, Jones RL, et al. Comparison of various treatments for experimentally induced equine infectious arthritis. Am J Vet Res 1987; 48: 519–529.
8. Gustafson SB, McIlwraith CW, Jones RL. Comparison of the effect of polysulfated glycosaminoglycan, corticosteroids, and sodium hyaluronate in the potentiation of a subinfective dose of Staphylococcus aureus in the middle carpal joint of horses. Am J Vet Res 1989; 50: 2014–2017.
9. Gustafson SB, McIlwraith CW, Jones RL, et al. Further investigations into the potentiation of infection by intra-articular injection of polysulfated glycosaminoglycan and the effect of filtration and intra-articular injection of amikacin. Am J Vet Res 1989; 50: 2018–2022.
10. Robson MC. Infection in the surgical patient: an imbalance in the normal equilibrium. Clin Plast Surg 1979; 6: 493–503.
11. Mah RA, Fung DY, Morse SA. Nutritional requirements of Staphylococcus aureus S-6. Appl Microbiol 1967; 15: 866–870.
12. Breed RS, Dotterrer WD. The number of colonies allowable on satisfactory agar plates. J Bacteriol 1916; 1: 321–331.
13. Sutton S. Counting colonies. Available at: www.microbiol.org/resources/monographswhite-papers/counting-colonies/. Accessed May 13, 2014.
14. Zubrod CJ, Farnsworth KD, Oaks LJ. Evaluation of arthrocentesis site bacterial flora before and after 4 methods of preparation in horses with and without evidence of skin contamination. Vet Surg 2004; 33: 525–530.
15. Dohoo I, Martin W, Stryhn H. Model-building strategies. In: Dohoo I, Martin W, Stryhn H, eds. Veterinary epidemiologic research. 2nd ed. Charlottetown, PE, Canada: VER Inc, 2012365–394.
16. Waxman SJ, Adams SB, Moore GE. Effect of needle brand, needle bevel grind, and silicone lubrication on contamination of joints with tissue and hair debris after arthrocentesis. Vet Surg 2014; 43: 1–6.
17. Hague BA, Honnas CM, Simpson BR, et al. Evaluation of skin bacterial flora before and after aseptic preparation of clipped and nonclipped arthrocentesis sites in horses. Vet Surg 1997; 26: 121–125.
18. Wahl K, Adams SB, Moore GE. Contamination of joints with tissue debris and hair after arthrocentesis: the effect of needle insertion angle, spinal needle gauge, and insertion of spinal needles with and without a stylet. Vet Surg 2012; 41: 391–398.
19. Selwyn S, Ellis H. Skin bacteria and skin disinfection reconsidered. Br Med J 1972; 15: 136–140.
