Article, Radiology

Ultrasound for the evaluation of soft tissue foreign bodies before and after the addition of fluid to the surrounding interstitial space in a cadaveric model

a b s t r a c t

Background: Point-of-care ultrasound may be used to facilitate foreign body (FB) localization and removal. We hypothesized that injection of normal saline adjacent to an FB may make it easier to detect.

Methods: The study was performed on one embalmed human cadaver. Potential FB sites were created of wood (24), metal (24), and null (24). Two sonographers evaluated each of the 72 sites both before and after a 25- gauge needle was inserted into each incision and 3 cc of normal saline was injected. Accuracy, sensitivity, and specificity were calculated both before and after injection of normal saline. Binomial tests were used to determine the statistical significance of FB detection before and after injection. A 2-tailed Student’s t test was used to deter- mine if there was a statistically significant difference between the 2 methods.

Results: Preinjection, 116 (81%) of the 144 interpretations (P <= .001) were correct in their assessment of whether or not an FB was present, with a sensitivity of 81% (95% confidence interval [CI], 72%-88%) and a specificity of 79% (95% CI, 65%-90%). Postinjection, 119 (83%) of these 144 interpretations (P <= .001) were correct in their assess- ment of whether or not an FB was present, with a sensitivity of 85% (95% CI, 77%-92%) and a specificity of 77% (95% CI, 63%-88%). This difference was not significant (P = .08; 95% CI, -0.04 to 0.01).

Discussion: Ultrasound was reasonably accurate, sensitive, and specific in identifying 1-cm metal and wood FBs. Although accuracy and sensitivity did improve after normal saline injection, this difference was not significant.

(C) 2016

Introduction

Subcutaneous foreign bodies (FBs) are a common chief complaint in the emergency department. Some FBs are not easily visualized during physical examination, and if left undetected, an FB can cause delayed wound healing and infection. When occult FB is suspected, the area may require evaluation with diagnostic imaging, which can include plain radiograph, computed tomography (CT), magnetic resonance, or ultrasound. Organic material such as wood, however, is not radio- opaque and may not be detected on plain radiographs or CT [1,2]. Point-of-care ultrasound has several potential advantages over these

? There is no grant support or involvement.

?? There are no conflicts of interest.

* Corresponding Author at: Department of Emergency Medicine, Mount Sinai West Hospital, Mount Sinai St Luke’s Hospital., 1000 10th Ave, New York, NY 10019. Tel.: +1 212 523 3981; fax: +1 212 523 2186.

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

other imaging modalities: it does not use ionizing radiation, it is capable of visualizing FBs of varying densities [3-4], it has a lower cost than CT or magnetic resonance, and it may be used in real time at the bedside to directly facilitate FB localization and removal.

Various foreign materials differ in how easily they are visualized sonographically. For example, metal FBs are associated with acoustic shadowing and a characteristic “ring down” artifact caused by the blockage and reverberation of the ultrasound waves. Organic matter such as wood, however, has a similar sonographic appearance to soft tis- sue and can thus be more difficult to detect. Therefore, we sought to identify a way to enhance the sonographic appearance of organic FBs that would allow for more accurate sonographic detection.

We hypothesized that fluid in the soft tissue might have this effect. Fluid is an excellent medium for the transmission of ultrasound waves. In the soft tissue, interstitial fluid can outline or create anechoic or hypoechoic halos around structures. In addition, fluid can cause the tissue to swell, increasing the space between the surface of the probe and structures of interest deep to the skin surface. This can move the area of interest further out of the dead zone (distance be- tween transducer and the first identifiable echo) and closer to the

http://dx.doi.org/10.1016/j.ajem.2016.06.004

0735-6757/(C) 2016

1780 T. Saul et al. / American Journal of Emergency Medicine 34 (2016) 17791782

focal zone (the most focused area of the ultrasound beam resulting in the highest image resolution).

Objectives

The objectives of this study were to determine the test characteris- tics of sonographic detection of FBs and to determine if injecting fluid into the soft tissue in the area of a suspected FB changed these test char- acteristics. Because organic material does not have as many artifactual clues as metal, we hypothesized that this difference, if present, would be more pronounced with wood FBs than with metal FBs.

Methods

This was a prospective observational study performed in a gross anatomy laboratory on the lower extremities of a single embalmed human cadaver. This study was exempt from institutional review board approval. The FBs were created from toothpicks (wood) and 21- gauge needles (metal). Each FB was approximately 1 cm in length and 1 to 3 mm in width. A pilot experiment of FB detection in a similar set- ting (FB inserted into porcine tissue, data not presented) showed an overall accuracy of FB detection of 73.9%. A Power calculation based on this result determined that 48 FB test sites and 24 null sites were re- quired in order to achieve a 95% confidence interval (CI) for sensitivity and specificity. Because it is not known whether metal and wood would be equally detectable, we included 24 pieces of each of the 2 materials.

The cadaveric specimen was positioned supine and the FB test sites were marked on the lower extremities by outlining 72 sectors with a skin-marking pen. Seventy-two incisions, each 2 cm in length, were made in the skin, and the FBs were inserted with a needle driver to a depth of approximately 1 cm. Each was inserted at an angle of approx- imately 45? with the intent to replicate how the FB might enter the skin in a real-world scenario (Fig. 1). For each of the null sites, a needle was inserted into the incision 1 cm into the deeper tissues, manipulated in a manner similar to those in which the FBs were inserted, and then removed.

The pattern of FB placement was determined using a computerized random sequence generator (www.random.org). Half of the FBs were

Fig. 1. Foreign body insertion at a 45? angle using a needle driver.

placed by the primary investigator (T.S.), an emergency ultrasound fellowship-trained attending physician, and half were placed by an emergency ultrasound fellow (A.B.D.).

Measures

Two emergency ultrasound fellowship-trained physician sonographers (G.R., S.D.S.) performed the scan protocol. Sonographers had each previously performed a minimum number of 175 Ultrasound studies, meeting the recommendations of the American College of Emergency Physicians 2008 policy statement [5]. This included at least 10 scans in soft tissue/musculoskeletal ultrasound but did not include any specific requirements for the identification of FBs. Two Sonosite Edge (Bothell, WA) ultrasound machines were used with L38 10-5 MHz linear transducers. The machines were set to the MSK setting and the multifrequency probes were set to resolution. The sonographers were permitted to adjust the depth and gain to their preference and to use as much conducting gel as desired. The sonographers were blinded to specimen preparation as well as to the frequency of FBs, the frequen- cy of type of FBs, and the pattern of the FBs. Sonographers documented the presence or absence of an FB on a data sheet. Each sonographer worked independently and was blinded to the other sonographer’s data sheet.

After the 2 sonographers completed the evaluation of each of the 72 FB sites, T.S. and A.B.D. inserted a small 25-gauge needle to a depth of 1 cm into each incision in the same direction as the FB had been inserted, and injected 3 cc of normal saline into the tissue. The sonographers then evaluated all 72 sites again and recorded their find- ings on the data sheet.

Data analysis

Accuracy, sensitivity, and specificity were calculated for the FBs both before and after injection of normal saline. These test characteristics were also calculated after stratifying by FB type. Diagnostic test evalua- tion was performed using MedCalc online statistical software (MedCalc Software bvba, Ostend, Belgium). Binomial tests were used to deter- mine the statistical significance of FB identification both before and after injection. A 2-tailed student’s t test was used to determine if there was a significant difference between the 2 methods.

Results

Preinjection, the 72 sites were evaluated by the 2 sonographers, for a total of 144 interpretations. Of the 144 (81%) interpretations (P <= .001), 116 were correct in their assessment of whether or not an FB was pres- ent, with a sensitivity of 81% (95% CI, 72%-88%) and a specificity of 79% (95% CI, 65%-90%; Figs. 2 and 3). After the injection of 3 cc of normal sa- line into each site, the 2 sonographers evaluated the 72 sites a second time. One hundred nineteen (83%) of these 144 interpretations (P <=

.001) were correct in their assessment of whether or not an FB was pres- ent, with a sensitivity of 85% (95% CI, 77%-92%) and a specificity of 77% (95% CI, 63%-88%; Figs. 4 and 5). Accuracies, sensitivities, and specific- ities for each FB type both before and after injection can be found in the Table.

There was no statistically significant difference found between the preinjection and postinjection sonographic evaluations (P = .08; 95% CI, -0.04 to 0.01).

Discussion

Previous studies of point-of-care sonographic detection of FBs have been in deceased animal tissue (e.g., chicken legs) [6] or human cadav- eric tissue [7,8] with varying results. Orlinsky et al [6] implanted wood- en toothpicks in 104 chicken thighs. Three emergency medicine physicians, 2 ultrasound technologists, and 1 radiologist performed

T. Saul et al. / American Journal of Emergency Medicine 34 (2016) 17791782 1781

Fig. 2. Metallic FB (arrow) with posterior reverberation artifact. Fig. 4. Metallic FB (arrow) surrounded by more hypoechoic tissue after injection.

sonographic evaluation of the sites and attained an overall accuracy of 82%, a sensitivity of 79%, and a specificity of 86%. Crystal et al. [7] im- planted 120 FBs in 150 test sites in the limbs of an unembalmed human cadaver. Six emergency medicine physicians detected the less than 2.5-mm3 FBs (wood, metal, plastic, glass) placed to a depth of up to 3 cm with a sensitivity of 53% and a specificity of 47%. Hill et al. [8] placed 53 FBs, each 2 cm in length (1.2-2 mm diameter), to a depth of

0.5 to 1.5 cm in 80 test sites in recently amputated human limbs. Two emergency physicians were able to detect wood with a sensitivity of 93% and plastic with a sensitivity of 73%, for an overall sensitivity of 83%. It is possible that in the study by Hill et al., the test characteristics were improved compared with the study by Crystal et al. because fresh- ly amputated tissue was used. The authors note that there may have been changes that occur in deceased tissue over time that affected their sonographic appearance.

Our group previously evaluated the test characteristics of FBs in a live anesthetized porcine model in an attempt to more closely mimic clinical conditions.[9] Foreign bodies were left in place for 2 hours, and the evaluation was repeated to determine if resultant surrounding edema or hematoma that developed with time increased sonographic detection. Sensitivity, specificity, and accuracy were similar to previous studies at both points in time. It is possible that these secondary findings may require longer than 2 hours for development and/or visualization. In this human cadaveric model, we demonstrated reasonably high accuracy, sensitivity, and specificity for FB detection. In both preinjection and postinjection, the probability of identification of FBs was statistically significant. These findings support the use of point-of- care ultrasound in the clinical setting when there is a question of a

Fig. 3. Wood FB (arrow) with echogenicity more closely resembling soft tissue. Posterior shadowing is visualized.

soft tissue FB, particularly with radiolucent objects that are not well vi- sualized by radiography. In addition, ultrasound can be performed quickly and at the bedside. Sonographers were able to interpret 72 sites for an FB in approximately 60 minutes.

Sonographers were able to detect metal FBs more accurately than wood FBs both before and after local normal saline injection; however, the sensitivities for both types of FB were still higher than that reported for radiography [3,10]. The difference is likely due to the significant re- verberation and shadowing artifacts generated by the interaction of ul- trasound waves with metal. Wood has an acoustic impedance similar to that of soft tissue, which is likely why there were more false negatives with this FB type.

Sixteen (17%) of 96 FB site evaluations before injection and 14 (15%) of 96 FB site evaluations after injection were false negatives in which sonographers did not detect the FB. The sonographers carefully per- formed the same scan protocol on each site, imaging in 2 orthogonal planes. It is possible that some of the false-negative results were due to the orientation of the FB compared with the skin surface. Ultrasound best visualizes structures that are perpendicular to the direction of the ultrasound beam. Foreign bodies that were positioned more perpendic- ular to the beam may have been easier to visualize than those oriented more parallel to the beam. Although the FBs were inserted in a consis- tent manner, there was no practical way to verify or standardize the final position and orientation of the FB in the tissue.

A possible explanation for false-positive cases, where an FB was

suspected but not present, is artifact from air introduced during test site preparation. To create the null sites, a needle was inserted and ma- nipulated in a manner similar to that of FB insertion. The sonographers may have mistaken irregular shadowing artifact created by air as an FB.

Fig. 5. Wood FB (arrow) surrounded by more hypoechoic tissue and anechoic fluid after injection.

1782 T. Saul et al. / American Journal of Emergency Medicine 34 (2016) 17791782

Table

Test characteristics for metal and wood FB before and after injection of 3 cc of normal saline

Preinjection Postinjection

Metal 90% (95% CI, 77%-97%) 88% (95% CI, 75%-95%)

Third, the very nature of the dense embalmed tissue itself may have impeded the distribution of the injected saline around the FB, thereby preventing the expected effect of increasing FB acoustic enhancement for better detection.

Fourth, although the data collection sheets were collected from the

Sensitivity

Wood 77% (95% CI, 63%-88%) 83% (95% CI, 70%-93%)

sonographers after the preinjection evaluation, sonographers may

Specificity Metal 79% (95% CI, 65%-90%) 77% (95% CI, 63%-88%)

Wood

Accuracy Metal 43/48 (90%) 42/48 (88%)

Wood 37/48 (77%) 40/48 (83%)

However, this situation may also occur in the clinical setting, if an FB en- ters and exits the Soft tissues intact and leaves subcutaneous air behind. After injection of the normal saline, the accuracy of detection of wood improved slightly, from 77% to 83%; however, for metal, it stayed approximately the same (90%-88%). These findings were not statistical- ly significant. We suspect that that although theoretically the injected fluid would surround the FB and make it easier to detect, this may not have occurred due to the higher density or other unique properties of embalmed cadaveric tissue. We are planning a future study of this tech- nique in live anesthetized tissue. Also, because the final location of the FB could not be verified exactly, it is possible that the injected saline

did not always directly surround the FB. It is important to note that many types of fluid can have a similar ap-

pearance on ultrasound. It is a known pitfall, for example, that an acute

hemothorax can appear identical to a transudative pleural effusion or that acute hemoperitoneum can appear identical to ascites. We expect that local anesthetics such as lidocaine and bupivacaine would be simi- lar in sonographic appearance to normal saline when injected in small volumes and examined within a short time interval.

Limitations

There are a number of limitations to this study. First, the study was performed in cadaveric human tissue. Sonographic echogenicity can vary significantly based on water content and density differences be- tween tissues. There are physiologic changes that occur after death and during the embalming process that likely alter the sonographic ap- pearance of the tissue. In addition, because of the denser nature of ca- daveric skin and soft tissue, a skin incision had to be created first to facilitate FB insertion. This may not accurately simulate a clinical scenar- io, in which an FB enters the tissues directly, with likely less air intro- duction and less disruption of surrounding tissue.

Second, when the postinjection evaluations were performed, the FB had been in place for a longer period, approximately 1 hour. It is possible that the difference in accuracy detected between the preinjection and postinjection evaluations was due to changes that occurred in the tissue over time independent of the normal saline injected. The absence of perfusion and an inflammatory response in cadaveric tissue, however, makes any such effect less likely.

have recalled their own previous assessments, possibly allowing their first sonographic evaluation to influence their second evaluation. Be- cause they performed 144 FB evaluations over the course of the exper- iment, the sonographers may also have improved in their FB detection skills as they gained experience; this could have contributed to their trend toward improved test characteristics in the postinjection evalua- tion. In addition, there was a high percentage of FBs (66% of sites), which may not replicate the frequency found in clinical practice. Sonographers may have increased their pretest probability of an FB after several sites were examined.

Finally, although our 2 sonographers had differing levels of training, both had met the minimal American College of Emergency Physician re- quirements for soft tissue ultrasonography and both had emergency ul- trasound Fellowship training. Their skills and, therefore, the results of this study may not be applicable to more novice or more experienced sonographers.

Conclusion

Ultrasound was reasonably accurate (81%), sensitive (81%), and spe- cific (79%) in identifying 1-cm metal and wood FBs in cadaveric human tissue. Accuracy (83%) and sensitivity (85%) were slightly higher after the injection of 3 cc of normal saline into each of the possible FB sites, but results were not statistically significant. The density of embalmed tissue was a significant impediment to local injection of fluid. We plan on a future study of this technique in a live anesthetized tissue model.

References

  1. Levine MR, Gorman SM, Young CF, Courtney DM. Clinical characteristics and man- agement of wound foreign bodies in the ED. Am J Emerg Med 2008;26(8):918-22.
  2. Blankenship RB, Baker T. Imaging modalities in wounds and superficial skin infec- tions. Emerg Med Clin North Am 2007;25(1):223-34.
  3. Graham Jr DD. Ultrasound in the emergency department: detection of wooden for- eign bodies in the soft tissues. J Emerg Med 2002;22:75-9.
  4. Wang R, Frazee B. Visual stimulus: splinter localization with ultrasound. J Emerg Med 2011;41(3):294-5.
  5. American College of Emergency Physicians. emergency ultrasound guidelines. Ann Emerg Med 2009;53(4):550-70.
  6. Orlinsky M, Knittel P, Feit T. The comparative accuracy of radiolucent foreign body detection using ultrasonography. Am J Emerg Med 2000;18(4):401-3.
  7. Crystal C, Masneri D, Hellums J. Bedside ultrasound for the detection of soft tissue foreign bodies: a cadaveric study. J Emerg Med 2009;36(4):377-80.
  8. Hill R, Conron R, Greissinger P. Ultrasound for the detection of foreign bodies in human tissue. Ann Emerg Med 1997;29(3):353-6.
  9. Saul T, Siadecki SD, Rose G, Berkowitz R, Matilsky D, Godbold J, et al. Ultrasound ac- curately identifies soft tissue foreign bodies in a liver anesthetized porcine model. Acad Emerg Med 2015;22(8):950-4.
  10. Ginsburg MJ, Ellis GL, Flom LL. Detection of soft-tissue foreign bodies by plain radi- ography, xerography, computed tomography, and ultrasonography. Ann Emerg Med 1990 Jun;19(6):701-3.

Leave a Reply

Your email address will not be published. Required fields are marked *