American Journal of Emergency Medicine (2012) 30, 1569-1573

Brief Report

Bedside ultrasound performed by novices for the detection of abscess in ED patients with soft tissue infections^?

Tony Berger MD^a,?, Francisco Garrido MD^b, Jeffrey Green MD^a,

Penelope Chun Lema MD^c, Jay Gupta MD^c

^aDepartment of Emergency Medicine, University of California Davis Medical Center, Sacramento, CA, USA

^bDepartment of Emergency Medicine, Lenox Hill Hospital, New York, NY, USA

^cDepartment of Emergency Medicine, New York Hospital Queens, Flushing, NY, USA

Received 24 June 2011; revised 31 July 2011; accepted 1 August 2011

Abstract

Objective: The objective was to compare bedside ultrasound (US) to clinical examination for the detection of abscess.

Methods: This is a 24-month prospective, observational emergency department (ED) study. Adults with suspected nondraining abscess with planned incision and drainage (I&D) are included in the study. Exclusion criteria are Spontaneous drainage and perineal, perirectal, or intraoral location. Before I&D, a second ED physician conducts an US and records the presence or absence of findings suggestive of abscess. A positive I&D of the suspected abscess is the criterion standard. The treating practitioner is blinded to the US results. Ultrasound is performed by novice ED physicians. The findings of the US, the prediction of pus from the clinician and the ultrasonographer in 3 strata (low, indeterminate, definite), and the results of the I&D (pus/no pus) are recorded onto data sheets. Measures of association are reported and Fisher’s Exact test is used.

Results: Forty patients were enrolled. The sensitivity of novice sonographers to predict a positive I&D with US was 0.97 (0.83-1.00), the specificity was 0.67 (0.24-0.94), the positive likelihood ratio was 2.90, the negative likelihood ratio was 0.04, and the area under the receiver operating characteristic curve was 0.85 (0.66-1.00). Clinical examination yielded a sensitivity of 0.76 (0.58-0.89), specificity of

0.83 (0.36-0.99), positive likelihood ratio of 4.50, negative likelihood ratio of 0.29, and area under the receiver operating characteristic curve of 0.75 (0.50-1.00).

Conclusion: Novice ED sonographers can identify abscesses with only minimal US training. Identifi- cation of abscess on US may change management of Cutaneous abscesses.

(C) 2012

Introduction

^? Presentation: The abstract for this manuscript was presented as a poster form presentation at the American College of Emergency Physicians annual meeting, 2008.

* Corresponding author. Tel.: +1 916 734 5010; fax: +1 916 734 7950.

E-mail address: [email protected] (T. Berger).

Soft tissue skin infections (STSIs) are a common presenting complaint in the emergency department (ED). The rise of methicillin-resistant Staphylococcus aureus has led to significant increases in related ED visits since its detection [1]. The treatment of abscess is incision and drainage (I&D) [2-5]. Differentiating underlying abscess

0735-6757/$ – see front matter (C) 2012 doi:10.1016/j.ajem.2011.08.002

from other subcutaneous processes can be challenging because of overlying edematous and indurated skin associ- ated with STSIs [6].

The application of ultrasound (US) for the evaluation of STSIs in the ED has been described, and an improvement in accuracy for abscess detection has been demonstrated. The use of US in the detection of abscess in the ED improved the accuracy of detection of superficial skin abscesses when used in conjunction with clinical examination [7]. The change in the decision to perform I&D with application of bedside US performed by a core group of expert ultrasonographers also changed about half of the physician management [8].

There are well-described US findings for detecting abscess [9]. There is evidence that even novice ultrasonog- raphers (US novices) can differentiate forms of STSIs using standard bedside US machines [6]. Given the rise in methicillin-resistant S aureus infections, the management of STSIs using prudent I&D is essential [1,3]. In addition, previous studies have allowed bedside US results to effect management decisions, which could lead to selection bias. Evaluating US performed by novices and compared with the criterion standard diagnostic evaluation in patients present- ing with an STSI is a logical next step.

We hypothesized that US performed by US novices in the ED can predict the presence of an underlying abscess and that US performed by US novices could improve a clinician’s ability to predict the presence of abscess in patients with equivocal findings on physical examination. The goal of this study was to evaluate the test characteristics of US for the detection of abscess compared with the results of I&D. We also aimed to compare clinicians’ abilities to predict the likelihood of pus on I&D with and without the aid of US.

Methods

Study design

This was a prospective, observational study of a con- venience cohort of adult ED patients seen and treated for STSI suspicious for underlying abscess. All patients enrolled had an I&D performed because clinician suspicion for abscess was, at a minimum, high enough to warrant attempted I&D. The main study outcome was the presence of visible pus expressed during the procedure (positive I&D). This study was approved by the institutional review board of the institution.

Study setting and population

The study was conducted over 24 months at an urban, community ED with 95,000 patient visits annually and a 3-year emergency medicine residency. Inclusion criteria were as follows: (1) 18 years or older, (2) clinical suspicion of a

nondraining cutaneous abscess by the treating physician or physician assistant, and (3) clinician plan to I&D the sus- pected abscess. Exclusion criteria were spontaneous drainage and perineal, perianal, and intraoral suspected abscesses.

Study protocol

Participating US novices were resident and attending physicians who completed a 2-day US course taught by a fellowship-trained Registered Diagnostic Medical Sonogra- pher-certified physician. They also received a 15-minute didactic session focused on soft tissue and abscess US. Criteria for suspected abscess were defined as described in Table 2. Attending physicians, resident physicians, and phy- sician assistants (clinicians) identified and recruited a convenience sample of patients. Demographic data, the presence of a trauma-related complaint, history of immuno- suppression, abscess location, and physical examination findings were recorded by the clinicians using a standardized data collection form (DCF). The clinicians also recorded their prediction of the likelihood of a positive I&D (low, indeterminate, or definite) on the DCF before performance of I&D. A separate US novice then conducted an US of the suspected abscess. He or she interpreted the scan; recorded the findings; generated an US-based impression of the likelihood of the presence of a positive I&D as low, indeterminate, or definite (Table 2); and documented it on a separate DCF. The DCFs were completed before perform- ing I&D, and treating clinicians and US novices were blinded to each other’s interpretations. We recorded a choice of 3 strata of probability to reflect true clinical practice; however, all subjects enrolled were suspected to have an abscess that required attempted I&D. Written consent was waived because clinical care of the patient was not altered by the findings of the US examination (Fig. 1).

Two Sonosite (Bothell, WA) US machines were used: the Titan or MicroMaxx model, using the 10-MHz high- frequency Linear probe.

Outcome measures

The primary outcome was a positive I&D. The sensitivity of US for the prediction of a positive I&D and an analysis to establish if clinical suspicion alone is predictive of a positive I&D were performed. We sought to determine whether clinicians were more accurate in predicting a positive I&D than a separate US novice using US. Finally, we determined whether relevant physical examination findings (Table 1) were associated with agreement between clinical and sono- graphic impression of abscess and a positive I&D.

Data analysis

Data was collected on DCFs and transcribed into a Microsoft Excel spreadsheet (Redmond, WA). We calculated

Positive Abscess Negative Abscess

Table 1 Associated clinical factors and demographics

Fig. 1 Theoretical pathway for use of bedside US in the evaluation of STSI.

the sensitivity of ED US performed by US novices to detect an abscess in patients with a suspected abscess.

Table 2 Contingency tables for the detection of abscess

We compared the prediction of positive I&D at 3 strata (low, indeterminate, definite) between US novices and the clinician’s physical examination (Table 2).

To determine test characteristics for comparison of US- and clinical examination-derived suspicion of abscess, we collapsed the 3 strata of each into 2: “indeterminate” (previously: low) and “probable” (previously: indeterminate and definite) suspicion for abscess. We did this to allow for comparison of test characteristics.

Abscess No abscess (n = 34) (n = 6)

demographic factors Diabetic 6 (17.7%) 3 (50%) Immunocompromised 5 (14.7%) 1 (16.7%) minor trauma 12 (35.3%) 2 (33.3%) Physical findings Erythema 28 (83.3%) 6 (100%) Tenderness 30 (88.2%) 5 (83.3%) Duration of symptoms (d) 6.2 (4.9) 3.8 (1.8) Size of erythema (cm) 2.8 (1.3) 3.4 (1.6) Fluctuance 13 (38.2%) 2 (33.3%)

Dichotomous variables are presented as number (percentage). Continuous variables are presented as mean (standard deviation).

Our analysis included sensitivities, specificities, positive likelihood ratios (LRs+), negative likelihood ratios (LRs-), Area Under the Receiver Operating Characteristic Curve , and the Fisher’s exact test.

We created a variable that represented agreement between the clinical examination and US prediction of positive I&D to analyze associations between physical examination findings and the agreement between the 2 tests.

All statistical analyses were performed in STATA, version 10 (College Station, TX).

Results

A total of 40 patients were enrolled in the study. Clinical factors and demographics are presented in Table 1.

The US-derived sensitivity for positive I&D was 0.97 (0.83-1.00), the specificity was 0.67 (0.24-0.94), the LR+

was 2.90, the LR- was 0.04, and the AUC was 0.85 (0.66-

1.00) (Table 3). The association between US-derived pre-I&D suspicion for abscess and positive I&D was statistically significant (P = .001). The clinical examination yielded a sensitivity of 0.76 (0.58-0.89), specificity of

0.83 (0.36-0.99), LR+ of 4.50, LR- of 0.29, and AUC of

0.75 (0.50-1.00). The association between clinical examina- tion for abscess and positive I&D was statistically significant

Pus on I&D No pus on I&D

US detection of abscess Probable abscess 33 2 Indeterminate abscess 1 4 Clinical examination detection of abscess Probable abscess 26 1 Indeterminate abscess 8 5

		Low	Indeterminate	Definitive
Negative I&D	cases (n = 6)
Post-US	Low	4	0	0
prediction	Indeterminate	1	0	0
of pus on	Definitive	0	0	1
I&D	Total	5	0	1
Positive I&D cases (n = 34)
Post-US	Low	1	1	0
prediction	Indeterminate	3	3	0
of pus on	Definitive	4	14	8
I&D	Total	8	18	8

(P = .007). There was no statistically significant difference between the AUCs of these 2 tests (P = .06). Fig. 2 depicts the receiver operating characteristic curves for both tests.

Table 3 Clinical prediction of the presence of pus on I&D compared with US stratified by I&D results

Clinical prediction of pus on I&D

There were 8 (23.5%) of 34 subjects who had a positive I&D in whom clinical suspicion for abscess was low. Of these, 7 of8 had a “probable” abscess on US. There were 2 (5.8%) of 34 subjects who had a positive I&D and in whom US-derived suspicion for abscess was low. Therefore, more patients with a low clinical suspicion for abscess subsequently had a posi- tive I&D than patients with a low sonographic suspicion; and there was no statistically significant association between these groups (P = .36) (Table 4).

Of the 18 subjects with indeterminate suspicion based on clinical examination and a positive I&D, 14 (77.8%) had a definite probability on US. No subjects had an indeterminate clinical examination and a negative I&D.

There were 17 (42.5%) cases with pre-I&D agreement between the clinical examination and US. In independent tests of association between the physical findings (Table 1) and the agreement between clinical examination and US for

0.00

0.25

0.50

1-Specificity

0.75

1.00

Sensitivity 0.50

0.75

1.00

Bedside US, AUC 0.85

Exam, AUC 0.75

Fig. 2 Receiver operating curves and AUCs for the prediction of positive I&D.

0.00

0.25

the prediction of abscess, we found the following: fluctuance was the only physical finding with a statistically significant association with agreement between clinical examination and US testing for a positive I&D (odds ratio, 5.1; 95% confidence interval, 1.3-20.9). No physical finding had a statistically significant association with a positive I&D.

Table 4 Clinical interpretation and Sonographic findings associated with each of 3 prediction strata

Clinical Interpretation prediction

Low Indeterminate physical examination findings Intermediate Probable abscess based on physical examination High Definite abscess based on physical examination US prediction US findings

Low Hyperechoic soft tissue suggestive of edema or cellulitis with no definite areas of definable hypoechoic cavities Intermediate Isoechoic or hypoechoic heterogeneous regions or irregular lobulated cavities surrounded by hyperechoic soft tissue suggestive of abscess with surrounding edema or cellulitis High Hypoechoic or anechoic, well-defined, and circumscribed cavity with posterior wall hyperechoic acoustic enhancement and surrounding hyperechoic tissue strongly suggestive of abscess with surrounding edema or cellulitis

Limitations

This study had several important limitations. We enrolled a small number of subjects, which likely affected the statistical significance of our results. The requirement that clinicians could not use US in their management limited our enrollment. We did not evaluate the uniformity of skill level among the US novices who performed scans, although this could be accomplished by archiving images to evaluate imaging acquisition technique and image interpretation skills. They were not necessarily “proficient” according to the American College of Emergency Physicians guidelines that recommend that clinicians perform at least 25 docu- mented and reviewed scans in a Diagnostic modality [2]. The single setting limited applicability to multiple Clinical sites. We did not perform phone follow-up of patients. Thus, clinical outcomes were not available for interpretation of associations between ED care and final clinical outcomes. We did not enroll subjects with STSI that the clinicians did not intend to I&D. The performance of 7-day follow-up as a measured outcome would have made this possible and may have impacted our results. Furthermore, our “criterion standard” outcome was somewhat limited, as subjects with a negative I&D may have had clinical deterioration based

on inadequate surgical management in the ED. Finally, the criterion standard outcome we used was positive I&D, which may have been biased by clinician skill level.

Discussion

We determined that US performed by a novice can reliably identify an abscess. Our study adds to previous work by demonstrating that the use of US by novices can alter the prediction of positive I&D, especially for patients with a low clinical suspicion for abscess.

Among our 34 positive I&D subjects, 18 had an “indeterminate” Clinical impression documented. Of those, 14 (77.8%) had a “definite” US. Thus, US confirmed a high percentage of abscesses in patients for whom clinician expectation for positive I&D was not high. This suggests that the use of US might improve confidence in the per- formance of I&D for these patients. Among the 6 subjects with a negative I&D, only 1 case had disagreement between clinician and US novice, suggesting that the combination of low clinical suspicion and negative US reliably predicts a negative I&D.

As a test, US demonstrated a trend toward improved accuracy over the clinical examination. The comparison between the areas under the curve did not achieve statistical significance, with a P value of .06. However, if the study sample were larger, it is likely that we would have found a statistically significant difference between the performance of the clinical examination and bedside US.

We found that the only physical finding that had a statistically significant association with agreement between clinical examination and US was fluctuance. This makes sense clinically, as fluctuance is a strong predictor of the presence of abscess [10,11]. This finding should be interpreted with a degree of caution before applying it in clinical practice, given our sample size. Of note, fluctuance was not statistically significantly associated with a positive I&D in our sample.

Previous studies have evaluated the utility of US for detection of abscess, but not the ability of US to predict positive I&D in patients who all have I&D performed without the influence of the US results on clinical care [7]. Previous ED research has also evaluated the detection of abscess by a core of highly trained physicians with high

numbers of previously performed US scans [8]. Our study evaluated the performance of US by US novices, reflecting the reality of emergency medicine practice.

We conclude that US performed by US novices is adequate for evaluation of STSI and reliably predicts positive I&D. We determined that fluctuance may predict agreement between clinical examination and US. And finally, we conclude that the identification of abscess fluid on US may change clinical management of abscesses in the ED. In conjunction with clinical examination, US may provide a superior ability to detect the presence of a true abscess.

References

1. Pallin D, Egan D, Pelletler A, Espinola J, Hooper D, Camargo C. Increased US emergency department visits for skin and soft tissue infections, and changes in antibiotic choices, during the emergence of community-associated methicillin-resistant Staphylococcus aureus. Ann Emerg Med 2008;51:291-8.
2. emergency ultrasound guidelines. Policy statement of the American College of Emergency Physicians. Ann Emerg Med 2009;53: 550-70.
3. Gaspari R, Resop D, Mendoza M, Kang T, Blehar D. A randomized controlled trial of incision and drainage versus ultrasonographically guided Needle aspiration for skin abscesses and the effect of methicillin-resistant Staphylococcus aureus. Ann Emerg Med 2011;57(5):483-91.
4. Stevens D, Bisno A, Chambers H, Everett E, Dellinger P, Goldstein E, et al. Practice guidelines for the management of skin and soft-tissue infections. Clin Inf Dis 2005;41:1373-406.
5. Chambers H, Moellering R, Kamitsuka P. Management of skin and soft-tissue infection. N Engl J Med 2008;359:1063-7.
6. Chau CLF, Griffith JF. Musculoskeletal Infections: ultrasound appearances. Clin Radiol 2005;60:149-59.
7. Squire B, Fox J, Anderson C. ABSCESS: applied Bedside sonography for convenient evaluation of superficial soft tissue infections. Acad Emerg Med 2005;12(7):601-6.
8. Tayal VS, Hasan N, Norton HJ, Tomaszewski CA. The effect of soft- tissue ultrasound on the management of cellulitis in the emergency department. Acad Emerg Med 2006;13:384-8.
9. Yiengpruksawan A, Ganepola G, Freeman H. Acute soft tissue infec- tion in intravenous drug abusers: its differential diagnosis by ultra- sonography. J Natl Med Assoc 1986;78(12):1193-6.
10. Giovanni J, Dowd M, Kennedy C, Michael J. Interexaminer agreement in physical examination for children with suspected soft tissue abscesses. Pediatr Emerg Care 2011;27:475-8.
11. Sivitz A, Lam S, Ramirez-Schrempp D, Valente J, Nagdev A. Effect of bedside ultrasound on management of pediatric soft-tissue infection. J Emerg Med 2009;05:013.