Accuracy of Physical Examination, Ankle-Brachial Index, and Ultrasonography in the Diagnosis of Arterial Injury in Patients With Penetrating Extremity Trauma: A Systematic Review and Meta-analysis.

Abstract

Penetrating Extremity Trauma (PET) may result in arterial injury, a rare but limb- and life-threatening surgical emergency. Timely, accurate diagnosis is essential for potential intervention in order to prevent significant morbidity.

Using a systematic review/meta-analytic approach, we determined the utility of physical examination, Ankle-Brachial Index (ABI), and Ultrasonography (US) in the diagnosis of arterial injury in emergency department (ED) patients who have sustained PET. We applied a test-treatment threshold model to determine which evaluations may obviate CT Angiography (CTA).

We searched PubMed, Embase, and Scopus from inception to November 2016 for studies of ED patients with PET. We included studies on adult and pediatric subjects. We defined the reference standard to include CTA, catheter angiography, or surgical exploration. When low-risk patients did not undergo the reference standard, trials must have specified that patients were observed for at least 24 hours. We used the Quality Assessment Tool for Diagnostic Accuracy Studies (QUADAS-2) to evaluate bias and applicability of the included studies. We calculated positive and negative likelihood ratios (LR+ and LR-) of physical examination ("hard signs" of vascular injury), US, and ABI. Using established CTA test characteristics (sensitivity = 96.2%, specificity = 99.2%) and applying the Pauker-Kassirer method, we developed a test-treatment threshold model (testing threshold = 0.14%, treatment threshold = 72.9%).

We included eight studies (n = 2,161, arterial injury prevalence = 15.5%). Studies had variable quality with most at high risk for partial and double verification bias. Some studies investigated multiple index tests: physical examination (hard signs) in three studies (n = 1,170), ABI in five studies (n = 1,040), and US in four studies (n = 173). Due to high heterogeneity (I > 75%) of the results, we could not calculate LR+ or LR- for hard signs or LR+ for ABI. The weighted prevalence of arterial injury for ABI was 14.3% and LR- was 0.59 (95% confidence interval [CI] = 0.48-0.71) resulting in a posttest probability of 9% for arterial injury. Ultrasonography had weighted prevalence of 18.9%, LR+ of 35.4 (95% CI = 8.3-151), and LR- of 0.24 (95% CI = 0.08-0.72); posttest probabilities for arterial injury were 89% and 5% after positive or negative US, respectively. The posttest probability of arterial injury with positive US (89%) exceeded the CTA treatment threshold (72.9%). The posttest probabilities of arterial injury with negative US (5%) and normal ABI (9%) exceeded the CTA testing threshold (0.14%). Normal examination (no hard or soft signs) with normal ABI in combination had LR- of 0.01 (95% CI = 0.0-0.10) resulting in an arterial injury posttest probability of 0%.

In PET patients, positive US may obviate CTA. In patients with a normal examination (no hard or soft signs) and a normal ABI, arterial injury can be ruled out. However, a normal ABI or negative US cannot independently exclude arterial injury. Due to high study heterogeneity, we cannot make recommendations when hard signs are present or absent or when ABI is abnormal. In these situations, one should use clinical judgment to determine the need for further observation, CTA or catheter angiography, or surgical exploration.