a b s t r a c t

Objective: Knee arthrocentesis can be performed by landmark (LM) or ultrasound (US) guidance. The goal of per- forming knee arthrocentesis is to obtain synovial fluid, however, it is also important to consider the number of attempts required and accidental bone contacts that occur. This study evaluates procedural success without bone contact in knee arthrocentesis and compares both LM and US guided techniques in a cadaver model.

Methods: This was a randomized crossover study comparing US vs LM guidance for arthrocentesis in a single ac- ademic center. Volunteers were randomized to perform both LM and US guided knee arthrocentesis on cadavers. The primary outcome was procedural success, defined as first attempt aspiration of synovial fluid without bone contact. Secondary outcomes included number of attempts, number of bone contacts, time to aspiration, and confidence.

Results: Sixty-one participants completed the study with a total of 122 procedures performed. Procedural success without bone contact was greater in the US group (84% vs 64% p = 0.02). Time to aspiration was longer for US (38.75 s vs 25.54 s p = 0.004). Participants were more confident with US compared to LM both before the pro- cedure on a Visual Analog Scale from 1 to 100 (29 vs 21 p = 0.03) as well as after the procedure (83 vs 69 p = 0.0001). Participants had a greater median increase in confidence with US following training (44 vs 26 p = 0.01). Conclusions: Study participants had greater procedural success without bone contact when US guidance was used. The increase in confidence following training was greater for US guidance than the LM method. Use of US guidance may offer a benefit by allowing for better needle control and avoidance of sensitive structures for clinicians performing knee arthrocentesis.

(C) 2023

1. Introduction

Knee arthrocentesis is an important procedure for a number of med- ical specialties. The traditional technique, or Landmark (LM) method, involves palpating Anatomic landmarks to identify the best site for nee- dle entry and aspiration [1]. A common site is immediately superior and lateral to the patella with the needle directed toward the suprapatellar recess. Ultrasound (US) can also be used to identify the suprapatellar recess and then dynamically guide the needle into this space. Needle guidance with US offers several theoretical and practical benefits [2].

* Corresponding author at: MUSC Department of EM, 169 Ashley Avenue, MSC 300, Charleston, SC 29425, USA.

E-mail address: shahaa@musc.edu (A. Shah).

Prior literature has suggested that US guidance is associated with in- creased rates of fluid aspiration and may yield a larger volume of aspi- rate [3]. A prior systematic review evaluating US vs LM guidance showed that US use may also lead to more favorable patient pain scores both during and after the procedure [4]. This may be due to avoidance of contact with pain-sensitive structures within the joint [5]. Studies have shown increased user confidence when US is used compared to LM [3]. While the use of US may require a special skillset and additional train- ing, prior literature has shown that novices can quickly develop proce- dural competency in using US for procedural guidance [6,7]. A prior study showed it was feasible for Emergency Medicine residents to per- form US guided arthrocentesis of the hip, ankle, and wrist in cadavers following only a 30-min Instructional video [8], and a recent RCT of emergency department patients revealed improved rates of success of elbow, wrist and ankle aspiration with US compared to LM guidance [9].

https://doi.org/10.1016/j.ajem.2023.06.044

0735-6757/(C) 2023

Similarly, another study suggested that a lecture alone was adequate to teach intra-articular knee injection to trainees [10].

Despite the advantages of US guidance, many clinicians still rely on the LM method for knee arthrocentesis. US potentially allows for greater needle tip control, increasing aspiration success on the first attempt and reducing accidental bone contact. The objective of this study was to compare these variables in both landmark and ultrasound guided knee arthrocentesis with clinicians of varying training and Experience levels.

1. Methods

This was a single center, randomized, crossover study. Participants were voluntarily recruited through convenience sampling using univer- sity email. Medical students, emergency medicine (EM) residents, EM physician assistants, and EM attending physicians were included. The IRB approved the study under exempt status.

An a priori sample size/Power calculation was not conducted, and the sample size for this experiment was based on feasibility of available participants during the time the project was conducted. Emails were sent to medical students, EM residents, and EM physician assistants, and EM faculty inquiring about participation. Enrolled participants were randomized to initial technique (LM or US) and laterality for left and right knees (Fig. 1). Randomization was achieved using REDCap software (REDCap, Nashville, TN, USA). Neither participants nor study personnel were blinded to randomization results. Participants were di- vided into novice and experienced groups prior to data analysis based on years of clinical practice, with those with <2 years of clinical experi- ence in Emergency Medicine placed in the novice group.

Simulated knee Joint effusions were created by research personnel in 11 embalmed cadavers by inserting an IV catheter into the medial joint space and instilling saline, a method previously described by the authors [11]. The sonographic dimensions of each effusion were stan- dardized to within 0.05 cm of the desired size of 0.55 cm in the anterio-posterior maximal dimension. Cadavers were assessed between each participant with ultrasound by research personnel and refilled. Those cadavers that failed to maintain appropriate effusion size or dem- onstrated needle marks at insertion sites were excluded from the re- mainder of the study.

Participants completed a pre-study survey assessing demographics such as prior experience and Confidence level. This was followed by a 20-min training session which reviewed both US and LM guided knee arthrocentesis from the superior-lateral approach. The training session included a 5-min instructional video and opportunity to practice needle insertion with and without US guidance on a gel block (CAE Blue phan- tom, Sarasota, FL, USA).

In the US technique, the transducer (Sonosite Edge, Transducer HFL38, FUJIFILM WA, USA) was oriented transversely in the suprapatellar space. An 18G needle was guided using an in-plane technique (Fig. 2). In the LM

technique, participants were instructed to palpate for anatomical land- marks and presence of an effusion, and to insert the needle in the lateral suprapatellar space. Participants were instructed to self-report bone con- tacts verbally to research personnel who documented each occurrence of bone contact. All of the necessary equipment for LM or US guided arthrocentesis was laid out in a standard fashion beside the participant and US machines were prepared with sterile drapes and appropriate set- tings by research personnel prior to the procedure.

Participants were randomized with respect to the order in which LM and US arthrocentesis was performed. Each participant received up to 5 min to survey the joint with the prescribed technique, following which research personnel instructed participants to begin the proce- dure and a timer was started. The number of attempts (defined as needle entry through skin) and bone contacts was recorded. The timer was stopped upon fluid aspiration and the total time was doc- umented. Immediately following the completion of the first tech- nique, participants initiated the complementary technique on a different cadaver in the same sequence as above, beginning with up to a 5 min period to survey the joint with the prescribed tech- nique. Following the performance of both techniques, participants completed a post-study survey.

The primary outcome of this experiment was procedural success de- fined as aspiration of synovial fluid on the first attempt, without bone contact. Secondary outcomes included the number of attempts, number of bone contacts, time to obtain aspirate, as well as confidence both be- fore and after the procedure.

Data was recorded in REDCap and analyzed using SAS software (SAS for Windows version 9.4, SAS, Cary, NC, USA). Simple descriptive statis- tics including frequency, percentage, median, and interquartile range (IQR) were used to describe participants regarding previous training and experience. Differences in outcomes were compared via Fisher’s exact tests for binary outcomes. Confidence was examined as a contin- uous outcome separately at multiple time points: (1) baseline (pre),

(2) after the procedures (post), and (3) absolute change from baseline (post-pre). Confidence outcomes and time to aspirate were compared between groups via Wilcoxon Ranked Sum tests, due to relatively small sample sizes and skewness in the data. Within group (overall, and separately by novice and experienced), paired differences be- tween LM and US were compared via McNemar’s tests for binary outcomes and Wilcoxon Signed Rank Tests for continuous outcomes. Differences [delta] in outcomes between ultrasound and landmark procedures were calculated for each individual. The mean and 95% confidence intervals for these differences are included to demon- strate the effect size between ultrasound and landmark for each out- come. Participants recorded responses to visual analog scales (VAS) scores by directly using a computer-generated slider to correspond to scores ranging 0-100. Sensitivity analysis was performed and showed no impactful differences when excluding advanced practice providers (n = 2), so all participants were included in the results.

Fig. 1. Participant Randomization flow diagram.

Fig. 2. Image of procedure setup and needle guidance.

Left: Image of provider performing US guided procedure with in-plane technique.

Right: Image of US screen during procedure. Note appearance of hyperechoic needle (arrow) and anechoic fluid pocket (*).

Table 1 Demographics.
		N		%
Classification	Medical Student	27		44
	Resident Physician:
	1st year	6		10
	2nd year	6		10
	3rd year	6		10
	Attending Physician	14		23
	Advanced Practice Provider	2		3
		Median		IQR?
PrevioUS procedures
	LM procedures performed	2		[0, 15]
	LM knee arthrocentesis performed	0		[0, 4]
	US procedures performed	2		[0, 75]
	US knee arthrocentesis performed	0		[0,2]
		Novice(n = 33)		Experienced (n = 28)
		Median [IQR?]		Median [IQR?]
Previous Procedures	LM procedures performed	0 [0, 15]		17.5 [8.5, 50]
	LM knee arthrocentesis performed	0 [0, 5]		5 [2, 17.5]
	US procedures performed	0 [0, 75]		75 [47.5, 100]
	US knee arthrocentesis performed	0 [0, 2]		0 [0, 2]
* IQR = Interquartile Range.

1. Results

Sixty-one participants took part in the experiment, and each per- formed both US and LM techniques. There were no participants who were lost to follow up or did not complete the experiment. Baseline demographics obtained via the pre-study survey are depicted in Table 1.

Over all participants, the median number of baseline LM and US guided procedures was 2. In the novice group, the median [Inter- quartile Range] number of baseline LM and US guided procedures was 0 [0,15], and 0 [0,75] In the experienced group, the median num- ber of LM guided procedures was 17.5 [8.5,50] and US guided proce- dures was 75 [47.5, 100].

Table 2

Outcome measures.

N = 61 LM US

N % N % p-value

Success? 39 64 51 84 0.02

LM US Delta (between US and LM)

	Median	IQR+	[Min, Max]		Median	IQR+	[Min, Max]	p-value	Mean	95% CI
Time to aspirate in seconds	25.53	[17.56, 34.82]	[6.41, 300.00]		40.43	[31.28, 62.75]	[10.53, 138.42]	<0.01	9.44	(-8.70, 20.90)
Number of attempts++	1	[1]	[1, 17]		1	[1]	[1, 4]	0.1	-0.46	(-1.05, 0.13)
Number of bony contacts	0	[0,1]	[0, 7]		0	[0,0]	[0, 1]	<0.01	-0.44	(-0.74, -0.15)
Confidence (pre)	21	[2, 54]	[0,100]		29	[5, 62]	[0, 100]	0.03	4.67	(-2.19, 11.53)
Confidence (post)	69	[49, 84]	[3,100]		83	[67, 92]	[32,100]	<0.01	15.30	(9.06, 24.53)
Change in confidence (pre to post)	26	[10, 51]	[-27, 95]		44	[16, 64]	[-6, 98]	0.01	10.62	(2.16, 19.08)

LM = Landmark, US=Ultrasound.

• Fluid on first attempt without bony contact.

⁺ IQR = Interquartile Range.

⁺⁺ One person did not complete the LM procedure and was stopped after 5 min (17 attempts).

Table 3

Group analysis by novice and experienced.

Novice (N = 33) Experienced (N = 28)

LM US LM US

N % N % p-value N % N % p-value Success? 18 55 25 76 0.09 21 75 26 93 0.1

LM US Delta (between US and LM)

LM US Delta (between US

and LM)

A. Shah, R.M. Barnes, L.E. Rocco et al.

American Journal of Emergency Medicine 71 (2023) 157–162

160

	Median	IQR+	[Min, Max]		Median	IQR+	[Min, Max]	p-value	Mean	95% CI		Median	IQR+	[Min, Max]		Median	IQR+	[Min, Max]	p-value	Mean	95% CI
Time to aspirate in	30.47	[23.47,	[15.38,		55.84	[37.06,	[24.03,	0.2	1.17	(-25.12,		21.62	[16.25,	[6.41,		31.61	[24.57,	[10.53,	<0.01	23.64	(18.31, 28.98)
seconds		68.63]	300.0]			73.38]	138.42]			27.46)			29.15]	70.69]			38.68]	106.98]
Number of attempts++	1	[1, 2]	[1, 17]		1	[1]	[1, 2]	0.1	-0.85	(-1.93.		1	[1]	[1, 2]		1	[1]	[1, 4]	>0.90	0	(-0.26, 0.26)
										0.23)
Number of bony contacts	0	[0, 1]	[0,3]		0	[0, 0]	[0, 1]	<0.01	-0.48	(-0.82,		0	[0, 0]	[0, 7]		0	[0, 0]	[0,1]	0.1	-0.39	(-0.92, 0.14)
										-0.15)
Confidence (pre)	3	[0, 19]	[0, 54]		5	[0, 19]	[0, 52]	0.02	2.09	(-0.38,		57.5	[30,	[4, 100]		65	[45, 86]	[7, 100]	0.2	7.71	(-7.39,
										4.56)			80.50]								22.82)
Confidence (post)	63	[32, 74]	[3, 97]		74	[64, 86]	[32, 100]	<0.01	19.12	(8.89,		76.5	[57, 91]	[31,100]		89.5	[77, 93.5]	[55, 100]	<0.01	10.79	(4.19, 17.38)
										29.36)
Change in confidence (pre	41	[23, 69]	[-16, 95]		63	[44, 84]	[12, 98]	<0.01	17.03	(6.59,		15	[1, 29.50]	[-27,		16.5	[3.50, 36]	[-6, 70]	0.7	3.07	(-10.83,
to post)										27.47)				70]							16.98)

LM = Landmark, US=Ultrasound.

• Fluid on first attempt without bony contact.

⁺ IQR = Interquartile Range.

⁺⁺ One person (novice) did not complete the LM procedure and was stopped after 5 min (17 attempts).

For the primary outcome, across all participants, there was a higher rate of procedural success without bone contact, when participants used US compared to the LM method (84% vs. 64%, p = 0.02) (Table 2). How- ever, when evaluating within novice and experienced groups alone, we were unable to detect a meaningful difference between the two methods (Table 3). When no consideration was given to bone contact or attempts required, synovial fluid was aspirated 100% of the time with the US method (61/61) and 98% of the time with the LM method (60/61), with the single failure representing a novice participant.

Over all participants, we were unable to detect a difference between the US and LM method for number of attempts required to obtain fluid (median 1 [1] vs 1 [1] p = 0.10) (Table 2). This was also true for the novice and experienced groups (Table 3). Across all partici- pants, there was a lower number of bony contacts in the US method compared to LM (median 0 [0,0] vs 0 [0,1] p < 0.01) and this was also true for the novice group. We were unable to detect a difference in number of bony contacts between US and LM methods in the experi- enced group (Table 3).

Over all participants, time to aspirate was shorter in the LM method vs US method (median 25.54 [17.56, 34.82] seconds vs. 38.75 [31.28, 62.75] seconds, p < 0.01). This was also true in the experienced group (21.62 s vs 31.61 s, < 0.01) However, within the novice group, we were unable to detect a statistically significant difference in time to as- pirate between LM and US methods (median 30.47 [23.47, 68.63] sec- onds vs 55.84 [37.06, 73.38] seconds, p = 0.20).

Over all participants, clinicians had greater baseline and post- training confidence with the US procedure vs LM (median 29 [5, 62] vs 21[2, 54], p = 0.03 and 83 [67, 92] vs 69[49, 84], p < 0.01). There

was also a greater increase in confidence with US compared to that with the LM method following training (median 44 [16, 64] vs 26, [10, 51] p = 0.01). This larger improvement in confidence with the US method was also present when evaluating the novice group alone (me- dian 63 [44, 84] vs 41[23, 69], p < 0.01). In the experienced group alone, we were unable to detect a significant difference in magnitude of confi- dence change in favor of either method following training (Table 3).

1. Discussion

This study showed that use of US guidance for knee arthrocentesis improved participants’ ability to perform the technique in a single at- tempt without contacting bony structures. This result reflects an appar- ent improvement in needle control and precision of needle placement when using US as compared to the LM method.

Based on the results of this study as well as prior research on the sub- ject, it is likely possible for clinicians to reliably obtain synovial fluid re- gardless of use of US or LM guidance. However, this does not take into account important factors such as number of attempts required to ob- tain fluid or possible contact with and damage to intraarticular struc- tures due to poor technique. While US use may only lead to modest improvement in rates of successful fluid aspiration, prior research shows improved pain scores when using US for knee arthrocentesis compared to LM, though the etiology of this finding had not been pro- posed [5,12]. We believe that avoiding bony contact has serious implica- tions for procedural pain. As such, a greater emphasis should be placed on the concept of needle control and avoidance of intraarticular struc- tures. We believe the results of our study are of significant clinical rele- vance for clinicians deciding whether to utilize US guidance routinely as a preferred approach for knee arthrocentesis.

While procedural time was greater in the US group, this difference was small and unlikely to be clinically important as it is an expected finding under real life conditions. The US machines for this study were pre-prepared with sterile covers and thus this preparation time was not captured in our study. Had this been captured, it would add substan- tial time to the US method, making the procedural time difference

negligible in comparison to the time that would be spent in additional preparation of equipment.

Unsurprisingly, participants in our experiment reported higher con- fidence with both techniques following a training period. However, the overall confidence was greater in the US group, which reflects the par- ticipants’ own perception of being able to accurately guide the needle to the desired target. Additionally, the increase in confidence following training was greater in the US group, revealing that our 30-min training session was effective at improving provider confidence.

This experiment was unique in that we measured the frequency of inadvertent bony contact during knee arthrocentesis. This is a proce- dural skill not studied previously. As a benefit of using a cadaver model, we were able to provide a controlled environment, with control for size and variability of joint effusions. These were limitations of prior studies [3-5].

In performing this experiment, we identified several limitations. First, because we utilized a convenience sample, our sample size was relatively small. This may have resulted in the experiment not being ad- equately powered to detect a difference in procedural success without bone contact within the novice and experienced subgroups. Second, participants may have been subject to selection bias, on the basis of being volunteers to participate in the experiment. This bias was mini- mized somewhat as the experiment included all residents from PGY1-3 in an emergency medicine residency program, though it may be present in the selection of the other study demographic groups. Third, participants were placed into novice and experienced categories based on their level of training, which may not reflect their competency or experience with the procedure. However, the novice group did self-report less ultrasound and procedural experience than that of the experienced group. Fourth, although we trained the participants in rec- ognizing bony contacts, it is possible participants may have misreported the incidence of bony contact, either intentionally or unintentionally. Fifth, while the training that participants underwent was typical for an emergency medicine training program, this was not independently evaluated for efficacy. Sixth, though participants were randomized re- garding the order in which the procedures were performed, there may be some unaccounted order effects in which their initial procedure could influence performance of the subsequent procedure. Finally, the clinical generalizability of this experiment is limited due to the nature of a cadaver based study, although cadavers have previously been used successfully in procedural evaluation and training [13]. Similarly, this model did not allow us to measure directly patient centered out- comes. Prior studies have noted that needle insertion markings can be- come visible after 15-20 punctures [14]. Because we had >10 cadavers available and were able to rotate their use, we did not appreciate this to be a problem in our experience. The effusions were standardized to be

0.55 cm of fluid in an anterior to posterior measurement, which was based on consensus among 3 emergency physicians as representative of a typical joint effusion. In our experience, this corresponds to a moderate-to- large joint effusion. This may have led to overall higher rates of procedural success and limits generalizability of our findings when considering smaller effusions. While our data are promising, we did not measure competency and further study would be needed to spe- cifically assess achievement of procedural competency markers.

1. Conclusions

This study revealed that between US and LM guidance for knee arthrocentesis, US use was associated with greater procedural success, as defined by obtaining fluid on the first attempt while avoiding bony contact with articular structures. It appears a short training session was adequate to teach participants how to perform the procedure. Par- ticipants demonstrated increased confidence with US compared to LM. US use appears to improve participants needle control. There was no

difference in the number of attempts required between methods and procedural time was longer when using US. Further study is needed to determine if US guidance for knee arthrocentesis improves users’ proce- dural competency or patient-oriented outcomes.

CRediT authorship contribution statement

Aalap Shah: Writing – review & editing, Writing – original draft, Su- pervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Ryan M. Barnes: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Methodology, Formal analysis, Data curation, Conceptualization. Lauren E. Rocco: Writing – original draft, Project administration, Methodology, Formal analysis, Data curation, Conceptualization. Chris Robinson: Writing – original draft, Project administration, Conceptualization. Steven W. Kubalak: Writing – original draft, Resources, Project administration, Methodology. Amy E. Wahlquist: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Data curation. Bradley C. Presley: Writing – review & editing, Writing – original draft, Super- vision, Project administration, Methodology, Investigation, Data curation, Conceptualization.

Declaration of Competing Interest

[AS, RMB, LER, CR, SWK, AEW, BCP] report no conflict of interest.

Acknowledgements

This publication [or project] was supported, in part, by the National Center for Advancing Translational Sciences of the National Institutes of Health under Grant Number UL1 TR001450. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

References

1. Thomsen TW, Shen S, Shaffer RW, Setnik GS. Arthrocentesis of the knee. N Engl J Med. 2006;354(19). https://doi.org/10.1056/nejmvcm051914.
2. Akbarnia H, Zahn E. Knee arthrocentesis. January 23, 2020, Retrieved from: https:// www.statpearls.com/ArticleLibrary/viewarticle/17862; 2019.
3. Wiler JL, Costantino TG, Filippone L, Satz W. Comparison of ultrasound-guided and standard landmark techniques for knee arthrocentesis. J Emerg Med. 2010;39(1): 76-82.
4. Wu T, Dong Y, Hx Song, Fu Y, Jh Li. Ultrasound-guided versus landmark in knee arthrocentesis: a systematic review. Semin Arthritis Rheum. 2016;45(5):627-32.
5. Sibbitt W, Kettwich L, Band P, et al. Does ultrasound guidance improve the outcomes of arthrocentesis and corticosteroid injection of the knee? Scand J Rheumatol. 2011; 41(1):66-72.
6. McGaghie WC, Issenberg SB, Cohen ER, Barsuk JH, Wayne DB. Does simulation-based medical education with deliberate practice yield better results than traditional clin- ical education? A meta-analytic comparative review of the evidence. Acad Med. 2011;86(6):706-11.
7. Amini R, Stolz LA, Breshears E, et al. Assessment of ultrasound-guided procedures in preclinical years. Intern Emerg Med. 2016;12(7):1025-31.
8. Berona K, Abdi A, Menchine M, et al. Success of ultrasound-guided versus landmark- guided arthrocentesis of hip, ankle, and wrist in a cadaver model. Am J Emerg Med. 2017;35(2):240-4.
9. Gibbons RC, Zanaboni A, Genninger J, Costantino TG. Ultrasound-versus landmark- guided medium-sized joint arthrocentesis: a randomized clinical trial. Acad Emerg Med. 2022 Feb;29(2):159-63.
10. Michels NR, Vanhomwegen E. An educational study to investigate the efficacy of three training methods for infiltration techniques on self-efficacy and skills of trainees in General practice. BMC Fam Pract. 2019;20:133.
11. Ross G, Ashenburg N, Wynn D, et al. Intra-articular catheter placement: a novel ap- proach for simulating ankle effusions in cadaver models. WestJEM. 2018;20(1): 92-3.
12. Sibbitt WL, Peisajovich A, Michaekl, et al. Does sonographic needle guidance affect the clinical outcome of intraarticular injections? J Rheumatol. 2009;36(9): 1892-902.
13. Hoyer R, Means R, Robertson J, et al. Ultrasound-guided procedures in medical edu- cation: a fresh look at cadavers. Intern Emerg Med. 2015;11(3):431-6.
14. Kay R, Manoharan A, Nematollahi S, et al. A novel fresh cadaver model for education and assessment of joint aspiration. J Orthop. 2016;13(4):419-24.