The accuracy of focused assessment with sonography in trauma (FAST) in blunt trauma patients: Experience of an Australian major trauma service

Introduction 

Focused assessment with sonography for trauma (FAST) has been used, world-wide, for the detection of haemoperitoneum/free fluid in the trauma patient's initial workup. The accuracy of FAST performed by non-radiologists has been well documented in the published literature.

Although these focused examinations are becoming an accepted practice in many trauma centres, the published experience utilising FAST within Australasia is limited.35, 18

This report examines an Australasian institution's experience with FAST performed by non-radiologists in the Emergency Department.

Method 

The study site is a 600 bed tertiary referral teaching hospital, which is designated as the Major Trauma Service for a population of approximately 800,000 people.

Between September 1999 and December 2004, FAST were prospectively analysed for the evaluation of patients with potential truncal injuries. Any trauma patient who came through the emergency department underwent a FAST study if the appropriate personnel were available. Those patients, who did not have the ultrasound results confirmed by either computed tomographic scanning (CT-scan) or laparotomy, were excluded from the study.

Information was collected on a pre-determined data sheet.

Haemodynamic instability was defined as a systolic blood pressure<90mmHg and/or a persistent tachycardia>120 beats per minute, unresponsive to fluid administration as per ATLS guidelines.

The study was approved by the hospital's Research and Ethics Committee.

Results 

A total of 463 FAST examinations were carried out on 463 patients. All of the patients sustained blunt truncal trauma. About 53 scans were excluded due to lack of a corresponding gold-standard investigation. Average age was 37.1±23 years old. Only five patients were under the age of 16. 71% were males and 29% females. The results of the FAST examinations are shown in Table 1.

Table 1. Results of FAST examinations

	True positive	78
	False negative	22
	True negative	302
	False positive	8
	Total scans	410
	Sensitivity	78% (95% CI: 70–86)
	Specificity	97% (95% CI: 96–99)
	Positive predictive value	91% (95%CI: 85–94)
	Negative predictive value	93% (95%CI: 90–96)
	Accuracy	93% (95% CI: 90–96)
	Error	7% (95% CI: 4–10)

Further analysis was carried out on those scans performed by the two credentialed physicians, as shown in Table 2.

Table 2. Results of FAST examinations by credentialed physicians (n=2)

	True positive	33
	False negative	8
	True negative	116
	False positive	0
	Total scans	157
	Sensitivity	80% (95% CI: 73–88)
	Specificity	100%
	Positive predictive value	100%
	Negative predictive value	94% (95% CI: 91–96)
	Accuracy	95% (95% CI: 93–97)
	Error	5% (95% CI: 3–7)

All of the false negative examinations were confirmed by CT scan. Trace free fluid, predominantly in the pelvis, was the main finding on CT scans of those patients with false negative FAST examinations. Three patients with negative FAST examinations were found to have a low grade splenic laceration with free fluid, but none required a laparotomy.

30 patients had their positive ultrasound findings confirmed by laparotomy. Of these 30, 18 patients were haemodynamically unstable and taken directly from the resuscitation room to the operating theatre. The mechanism and operative findings on these unstable patients are shown in Table 3.

Table 3. Operative findings of haemodynamically unstable patients with a positive FAST

	Age	Sex	Mechanism	Operative findings
	29	Male	MVC	Grade II splenic injury
	46	Male	Boating	Grade IV splenic injury
	24	Male	Explosion	Grade V splenic injury
	30	Male	Pedestrian	Grade IV splenic injury
	41	Female	MVC	Grade IV splenic injury
	49	Male	MBC	Grade IV splenic injury
	25	Male	MBC	Grade IV splenic injury
	22	Male	MVC	Small bowel and sigmoid mesenteric tear
	28	Male	MVC	Grade III liver injury, pancreatic contusion
	62	Male	Crush	Right hepatic artery transection, common bile duct transection
	23	Male	MVC	Grade IV splenic injury
	39	Male	MVC	Diaphragm laceration
	28	Male	MBC	Grade III liver injury, ruptured duodenum, pancreas laceration
	84	Female	MVC	Grade III splenic injury, ruptured diaphragm, ruptured left ventricle
	29	Male	MVC	Grade III liver injury, small bowel perforation
	30	Male	MVC	Grade IV splenic injury
	23	Male	MBC	Small bowel mesenteric tear
	49	Male	Fall	Small bowel mesenteric tear

MVC, motor vehicle crash and MBC, motorbike crash.

No laparotomies were performed for the false positive examinations.

Discussion 

This study represents the largest series to date of ultrasound examinations performed by non-radiologists for trauma in Australasia.

When compared to those studies which examine non-radiologist performed ultrasounds, the sensitivity and specificity in our series (sens 78%, spec 97%) compares favourably (sens 69–98%, spec 85–99%).35, 26, 27, 28, 29, 15, 16, 17, 18, 19, 34, 21, 12, 8, 9, 32, 23

The values are also comparable to those studies which utilise sonographer/radiology resident/radiologist performed/interpreted scans (sens 69–93%, spec 98–100%).7, 22, 14, 11, 25

It is very important to note that definitions vary somewhat in the existing literature. All papers define a positive ultrasound result, be it true or false, and a false negative result, by means of a confirmatory gold-standard investigation, namely CT, Diagnostic Peritoneal Lavage (DPL) or laparotomy. However, the vast majority of studies define those patients with a negative FAST and a benign course of clinical observation, as a true negative examination. Some studies do not specify exactly how many true negative examinations were determined by clinical observation, while those which do, have between 34% and 96% of their total true negative examinations confirmed by clinical observation. This has been identified as a weakness by many authors, given the lack of long-term follow-up.

We chose only to include those examinations which had a corresponding gold-standard investigation, as we believe this to be a more accurate assessment of FAST. An international consensus conference on FAST also agreed that in the pure statistical sense, all true negatives should be confirmed by a standard test, however it may be impractical to confirm all negative results with an additional study.30 One can only speculate as to how many patients with a negative FAST and negative clinical observation period would have had intraperitoneal free fluid if a gold-standard confirmatory test been carried out. Thus, these studies involving clinical observation may be not be a strictly accurate assessment of FAST, in the true methodological sense. When looking at studies which include only those examinations with a confirmatory gold-standard investigation, the range of sensitivities and specificities are much narrower (sens 70–81%, spec 97–98%).35, 18, 8, 9

The false negative examinations in our series were largely clinically insignificant, as all but three examinations demonstrated only a trace of free fluid in the pelvis on CT scanning, with no subsequent laparotomy required. In this study, the subdiaphragmatic space above the spleen was not imaged on a regular basis, and thus was probably a significant contributory factor in the false negative examinations with splenic injuries. Also, there is the possibility that fluid may have accumulated during the time between the initial ultrasound examination and the subsequent CT scan. There does appear to be a minimal amount of free fluid required for ultrasound detection, however, the quantitative amount has not been accurately described in the literature. Two North American studies have used a model infusing fluid into trauma patients via a DPL catheter. The mean amount of fluid able to be detected was 619 and 668ml.10, 1

These studies have limitations as they only examined a single area of the abdomen, namely Morrison's pouch. Lentz and McKenney have made two comments relevant to these studies. Firstly, the models reflect the amount of instilled fluid required to overflow from the pelvis into Morrison's pouch, not the minimal detectable amount of fluid. Secondly, during a trauma ultrasound examination, multiple areas must be scanned, since the flow of intraperitoneal fluid is determined by peritoneal anatomic compartmentalisation, and that various factors will influence where fluid may accumulate including gravity, changes in intra-abdominal pressure and the injured organ of origin.20

In the Branney study, the minimal volume detectable was 225ml.10 Further support for a minimal ultrasound detectable amount of fluid is evidenced in a series by Miller et al., where 20 out of 22 false negative FAST studies had ≤300ml of fluid seen on CT scanning.24

With the knowledge that small amounts of free fluid may be undetectable by ultrasound, it has been recommended that, at a minimum, a negative FAST result in a haemodynamically stable patient be followed by a period of observation and at least one follow-up FAST examination.30 Currently there is minimal quantitative evidence to support the use of repeat FAST examinations.6 However, this is an area which requires further investigation.

It has been suggested that standard credentialing criteria should be set forth by a nationally recognised organisation.30 Within Australasia, the Australasian College of Emergency Medicine (ACEM) and the Australasian Trauma Society (ATS) have published credentialing policies. Both specify that an ultrasound workshop is essential, however only the ATS guidelines specify a minimum of 16 workshop hours. Both organisations require a minimum of 25 proctored scans with at least five positive scans.4, 5

Recommendations from other countries are more variable. In Germany, ultrasound use is incorporated into surgical residency training. The requirements are 15h in theory and 15h in practice, with a minimum of 400 examinations under supervision.15 In the USA, most published guidelines by national organisations, such as the Society for Academic Emergency Medicine and the American Institute of Ultrasound in Medicine refer to broader ultrasound examinations and are therefore not applicable to FAST credentialing alone.2, 3 Otherwise, training and credentialing appears to be institute specific.

Several studies have attempted to elucidate the learning curve involved with FAST examinations. Thomas et al. demonstrated that the accuracy of ultrasound examinations plateaued after an average of 50 examinations per trauma fellow.33 Their results were consistent with the published literature (sens 81%, spec 99%), after a single 8h trauma ultrasound course and a minimum of 50 proctored examinations. Gracias et al. hypothesised that more experienced providers would perform FAST with greater accuracy, by utilising peritoneal dialysis patients as models for injured patients with free fluid. They concluded that the learning curve for FAST starts to flatten out at 30–100 examinations.13

The prevalence of free fluid in our series was 24% with a pooled error rate of 7%. This is consistent with the arbitrary error rate of 8% suggested by Shackford et al.31 Their conclusion for reasonable credentialing criteria was a didactic course of 4h and a minimum of 30 examinations to determine the error rate. Jones et al. similarly described a pooled error rate of 10% (prevalence 22%) in their series assessing the ACEM/ATS credentialing criteria.18 When looking at the results from the two prior credentialed physicians in our series, the ACEM/ATS criteria appear to be acceptable to ensure accurate credentialing.