Tag Archives: CT

Contrast-Induced Nephropathy: Confounding Causation

Author: Richard Sinert, DO (Professor of Emergency Medicine / Vice Chair of Research, SUNY Downstate Medical Center) // Edited by: Alex Koyfman, MD (@EMHighAK, EM Attending Physician, UTSW / Parkland Memorial Hospital) and Brit Long, MD (@long_brit)

I would like to applaud the study “Risk of Acute Kidney Injury After Intravenous Contrast Media Administration” by Hinson et al[1] in the February 2017 issue of Annals of Emergency Medicine.  Before discussing the details of this study, I would like to give a historical perspective on how the study of CIN has evolved.

Since the first observation by Bartels et al[2] of the association between between contrast administration and Acute Kidney Injury (AKI), multiple studies gave further added weight to the association between intravenous contrast and AKI[3-7].

Although CIN is defined by a relatively small change in serum creatinine (SCr) (25% from baseline or an absolute increase of 0.5 mg/dl 48-72 hours post infusion), the consequences for CIN patients at least on the surface seemed dire.  Among cardiac catheterization patients, CIN increased the mortality rates[3] from 6% to 16% in one study[7] and from 0.6% to 31% in another[6].  Higher composite mortality rates and need for renal replacement (relative risk = 36) were also observed in patients who met the definition for CIN following CT-PA, after intravenous contrast in pulmonary embolism patients who developed CIN[4].

At this point an iatrogenic injury (CIN) was linked to an easily measured disease marker (timed changes in SCr) that seemed to be associated with adverse outcomes.  Not surprisingly, the medical community with the best of intentions studied the risks[5],[6] and a wide-range[7],[8],[9] of potential measures to prevent CIN.

Yet all these studies documenting CIN incidence, risks, outcomes, and prophylactic strategies suffer a bias common to many observational studies— confounding bias[10],[11].   Confounding bias occurs when an exposure is inappropriately causally linked to an outcome, when a separate exposure (confounding variable) other than the one of interest better explains the observed outcome.  Since the definition of CIN requires a second timed measurement of SCr, these studies must select for a relatively ill group of hospitalized patients undergoing repeated laboratory testing; selection bias must be considered.  Decrements in kidney function signaled by a rise of SCr could have occurred from the incident disease before or after contrast administration.  In addition, intercurrent hemodynamic instability (eg., sepsis, hemorrhage, diuresis) and a multitude of nephrotoxins (eg., NSAID’s, ACE-Inhibitors, antibiotics) are common complications during hospitalization, which may also explain an increased SCr and associated higher mortality rates.  Newhouse et al[12] found that among 32,161 hospitalized patients not exposed to contrast, 19% of patients had a 25% increase in SCr, which would have fulfilled diagnostic criteria for CIN had they been exposed to IV contrast.

Lipsitch et al[13] stated that non-causal associations between outcomes and exposures are the result of either mismeasurement (recall bias), confounding bias, or selection bias.  To prevent confounding, Lipsitch et al[13] suggests designing a negative control experiment where the observation is repeated under conditions that are not expected to produce the outcome of interest.  If the outcome is encountered without the exposure, then a confounding bias may exist. This form of negative control experiment in which the incidence of AKI is compared across patients exposed and unexposed to contrast has been studied by multiple investigators[14], [15], [16], [17] , all failed to find a statistically significant difference in AKI rate (using CIN definition) between those exposed to contrast and controls.

These studies[18-21] that compared the incidence of CIN between contrast- exposed and unexposed groups also posed methodological issues related to the differences in the baseline risks of AKI between the two study groups.  It is not surprising that the patients hospitalized after requiring a contrast-enhanced CT may be inherently different that those not requiring a similar study.  To account for this potential selection bias, multiple studies have compared the incidence of AKI between contrast exposed and unexposed patients utilized propensity-scoring matching.  Propensity-score matching is a methodology that balances the baseline outcome risks between the study groups[18].  Even utilizing propensity-score matching for AKI, multiple studies[19], [20], [21], [22], [23] again failed to find a statistically higher incidence of AKI in the contrast- exposed compared to unexposed group of hospitalized patients.  In addition, the increases in risks of higher mortality rate in the CIN patients were not found when propensity-scoring matching accounted for the baseline risk of mortality of the contrast-exposed and non-contrast exposed patients.

The most recent CIN study by Hinson et al[1] in this recent issue of Annals of Emergency Medicine represents the latest in the line of investigations into the causal relationship between contrast and AKI.  Hinson et al[1] conducted a retrospective study over a 5-year period comparing the incidence of AKI among three groups, including contrast-enhanced CT (n=7,201), non-contrast enhanced CT (n=5,499), and those in whom CT was not performed (n=5,234).  These three groups were propensity-scoring matched for AKI risks.  AKI was defined both using the common CIN definition and definitions of AKI as reported in the Acute Kidney Injury Network/Kidney Disease Improving Global Outcomes (KDIGO) guidelines[24].   Applying the traditional definition of CIN, AKI was found in 10.6%, 10.2%, and 10.9% in contrast-enhanced CT, non-contrast CT, and non-CT groups, respectively.  Utilizing the KDIGO AKI definitions, AKI occurred in 6.7%, 8.9%, and 8.1% in contrast-enhanced CT, non-contrast CT, and non-CT groups, respectively.

Compared to previous propensity-scoring matched studies mentioned above, Hinson et al[1] went a step further by conducting a multiple logistic regression analysis, including in their model known predictors of AKI and contrast administration. From the multiple logistic regression model, contrast administration produced a non-significant odds-ratio for AKI as defined by both the CIN (0.96 [95% CI, 0.85-1.08]) and KDIGO criteria (1.00 [95% CI, 0.87-1.1.6]).  Moreover, the authors found no differences among the three study groups for the development of chronic kidney disease, need for dialysis, or renal transplantation in the following 6 months post-contrast exposure.

Although patients with elevated SCr (> 4.0 mg/dl) were excluded from their primary analysis, multiple logistic regression analysis of patients with elevated baseline SCr found no independent risk of AKI for contrast administration.

In conclusion, comparing the methodological rigor of more recent CIN studies to those in the past, it seems clear that earlier studies purporting a causal relationship between AKI and contrast administration were only identifying an association but not a true clinical entity. Older CIN studies were biased by confounding variables (e.g., hemodynamic instability, nephrotoxins), with well-established links to AKI providing a sufficient cause for AKI without implicating contrast as an additional AKI risk.

The history of the study of CIN is just another example of evidence-based medicine successfully applied to the debunking of a common belief in a clinical syndrome.  As ED physicians are faced with the challenge of rapidly diagnosing life-threatening conditions (i.e. aortic dissection/aneurysmal rupture, pulmonary embolism, occlusion or aneurysmal rupture of cerebral vessels, traumatic vascular injury), we should not delay emergent contrast-enhanced CT scans waiting for SCr.


References / Further Reading

[1] Hinson JS, Ehmann MR, Fine DM, et al. Risk of Acute Kidney Injury After Intravenous Contrast Media Administration. Ann Emerg Med 2017.

[2] Bartels ED, Brun GC, Gammeltoft A, Gjorup PA. Acute anuria following intravenous pyelography in a patient with myelomatosis. Acta Med Scand 1954;150:297-302.

[3] Pickering JW, Blunt IR, Than MP. Acute Kidney Injury and mortality prognosis in Acute Coronary Syndrome patients: A meta-analysis. Nephrology (Carlton, Vic) 2016.

[4] Mitchell AM, Jones AE, Tumlin JA, Kline JA. Prospective study of the incidence of contrast-induced nephropathy among patients evaluated for pulmonary embolism by contrast-enhanced computed tomography. Acad Emerg Med 2012;19:618-25.

[5] Mehran R, Aymong ED, Nikolsky E, et al. A simple risk score for prediction of contrast-induced nephropathy after percutaneous coronary intervention: development and initial validation. J Am Coll Cardiol 2004;44:1393-9.

[6] Lin KY, Zheng WP, Bei WJ, et al. A novel risk score model for prediction of contrast-induced nephropathy after emergent percutaneous coronary intervention. International journal of cardiology 2017;230:402-12.

[7] Li H, Wang C, Liu C, Li R, Zou M, Cheng G. Efficacy of Short-Term Statin Treatment for the Prevention of Contrast-Induced Acute Kidney Injury in Patients Undergoing Coronary Angiography/Percutaneous Coronary Intervention: A Meta-Analysis of 21 Randomized Controlled Trials. American journal of cardiovascular drugs: drugs, devices, and other interventions 2016;16:201-19.

[8] Wang N, Qian P, Kumar S, Yan TD, Phan K. The effect of N-acetylcysteine on the incidence of contrast-induced kidney injury: A systematic review and trial sequential analysis. International journal of cardiology 2016;209:319-27.

[9] Subramaniam RM, Suarez-Cuervo C, Wilson RF, et al. Effectiveness of Prevention Strategies for Contrast-Induced Nephropathy: A Systematic Review and Meta-analysis. Ann Intern Med 2016;164:406-16.

[10] Grimes DA, Schulz KF. Bias and causal associations in observational research. Lancet 2002;359:248-52.

[11] McNamee R. Confounding and confounders. Occup Environ Med 2003;60:227-34; quiz 164, 234.

[12] Newhouse JH, Kho D, Rao QA, Starren J. Frequency of serum creatinine changes in the absence of iodinated contrast material: implications for studies of contrast nephrotoxicity. AJR Am J Roentgenol 2008;191:376-82.

[13] Lipsitch M, Tchetgen Tchetgen E, Cohen T. Negative controls: a tool for detecting confounding and bias in observational studies. Epidemiology 2010;21:383-8.

[14] Cramer BC, Parfrey PS, Hutchinson TA, et al. Renal function following infusion of radiologic contrast material. A prospective controlled study. Arch Intern Med 1985;145:87-9

[15] Heller CA, Knapp J, Halliday J, O’Connell D, Heller RF. Failure to demonstrate contrast nephrotoxicity. Med J Aust 1991;155:329-32.

[16] Bruce RJ, Djamali A, Shinki K, Michel SJ, Fine JP, Pozniak MA. Background fluctuation of kidney function versus contrast-induced nephrotoxicity. AJR Am J Roentgenol 2009;192:711-8.

[17] Sinert R, Brandler E, Subramanian RA, Miller AC. Does the current definition of contrast-induced acute kidney injury reflect a true clinical entity? Acad Emerg Med 2012;19:1261-7.

[18] Haukoos JS, Lewis RJ. The Propensity Score. JAMA 2015;314:1637-8.

[19] Davenport MS, Khalatbari S, Dillman JR, Cohan RH, Caoili EM, Ellis JH. Contrast material-induced nephrotoxicity and intravenous low-osmolality iodinated contrast material. Radiology 2013;267:94-105.

[20] McDonald RJ, McDonald JS, Carter RE, et al. Intravenous contrast material exposure is not an independent risk factor for dialysis or mortality. Radiology 2014;273:714-25.

[21] Hsieh MS, Chiu CS, How CK, et al. Contrast Medium Exposure During Computed Tomography and Risk of Development of End-Stage Renal Disease in Patients With Chronic Kidney Disease: A Nationwide Population-Based, Propensity Score-Matched, Longitudinal Follow-Up Study. Medicine 2016;95:e3388.

[22] Tremblay LN, Tien H, Hamilton P, et al. Risk and benefit of intravenous contrast in trauma patients with an elevated serum creatinine. J Trauma 2005;59:1162-6; discussion 6-7.

[23] Cely CM, Schein RM, Quartin AA. Risk of contrast induced nephropathy in the critically ill: a prospective, case matched study. Critical care (London, England) 2012;16:R67.

[24] Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group. KDIGO clinical practice guideline for acute kidney injury. Kidney Int Suppl. 2012;2:1-138.

Commonly Missed Findings on CT Abdomen/Pelvis

Author: Emily Thompson, MD (EM Resident Physician, North Shore University Hospital) // Edited by: Alex Koyfman, MD (@EMHighAK, EM Attending Physician, UTSW / Parkland Memorial Hospital) and Brit Long, MD (@long_brit)

Computed Tomography.  The donut of truth.  If you are like me (even if you aren’t an intern), you probably dread the patient with non-specific abdominal pain.  I certainly breathe a little easier sending a patient home with a negative CT abdomen/pelvis.  However, the power of x-ray vision doesn’t allow us to turn off our brains.  Certain pathologies may have only subtle findings on CT, and others may lend themselves better to other imaging modalities, such as ultrasound.  By being aware of these pathologies and how to identify them, we can better recognize patients at risk of a missed diagnosis.

Case 1

A 45-year-old male is a restrained driver in a single car, head-on collision with a telephone pole.  He presents to a level-one trauma center by BLS ambulance in a c-collar.  His heart rate is 120, blood pressure is 100/60, respiratory rate is 24, and O2 saturation is 96% on 100% non-rebreather.  His eyes are closed, and he is moaning and making gurgling sounds.  He withdraws from pain, and he smells of alcohol.  He has shortening of his left leg and a seatbelt sign. The patient is put on the monitor, 2 large-bore IVs are placed running normal saline, and the patient is intubated.  The patient has bilateral breath sounds, a soft, obese abdomen, and intact distal pulses.  FAST exam is negative and the patient’s blood pressure responds to 2L normal saline, so the pelvis is stabilized and the patient is transported to CT.  The images are reviewed by the on-call radiology resident, as well as the ED resident and trauma surgeon, who found a left hip dislocation, an acetabular fracture, and two right lower rib fractures, with no other obvious injuries.  On day 2 in the surgical ICU he develops a fever and peritonitis.  Repeat CT shows free air and fluid.


Missed Injuries in Trauma

Bowel Rupture

Trauma is the most common cause of small bowel perforations.  These are the third most common traumatic perforations after liver and spleen and are usually associated with other injuries.  The most common CT finding is free air, but the leaks are less dramatic than with gastroduodenal perforation.  Other findings include a visible bowel wall lesion, “misty” mesentery, intestinal pneumatosis, free fluid, and extraluminal fecal matter.  CT is more sensitive than plain x-rays and extremely specific (95.4%).  However, free air may be as little as 50% sensitive for detecting bowel perforation.11

The best way to manage these patients is to be aware of the possibility of bowel injury, re-examine them frequently, and re-image them or even send them for ex-lap if their condition changes. One study reviewed trauma victims who underwent exploratory laparotomy after repeat CT.  While the initial CT was only 30% sensitive for bowel perforations, the second was 82% sensitive.13  These patients likely developed new symptoms that led to their repeat scans, which introduces some bias into the study.  However, it is important to remember that a single CT is only a “snapshot,” and patients with serious blunt trauma may have occult injuries that destabilize later.

 Bladder Trauma

Bladder injuries in trauma are relatively rare.  They are most commonly caused by blunt trauma to the pelvis.  They can be due to a direct blow to a full bladder, causing a dome rupture, or they can be associated with pelvic fractures, particularly disruption to the pelvic ring.  Bladder rupture is usually not immediately life-threatening, and the more common extraperitoneal ruptures are managed expectantly with catheter drainage.  Intraperitoneal ruptures, on the other hand, require surgical repair.  Extravasation of urine into the peritoneal space can result in chemical peritonitis and sepsis.

While a CT may be able to directly visualize breaks in the bladder wall, findings are usually more non-specific: free fluid, ascites, or bladder wall enhancement.  Suspicion should be high for GU injuries in pelvic trauma.  Every patient with gross hematuria or inability to void requires evaluation for GU tract damage.  Retrograde cystography is nearly 100% sensitive for bladder injury.  CT cystography is being used more often as part of the standard trauma series and is 85-100% sensitive for bladder rupture.9

Diaphragmatic Injury

The diaphragm is in the shadowy borderlands between the thoracic and abdominal cavity and is caught at the top of the CT abdomen/pelvis.  It can be damaged by penetrating trauma anywhere between the fourth rib or tip of the scapula and the inferior rib margin.  It can also be ruptured by blunt trauma that increases intra-abdominal pressure.   Injuries occur most commonly on the left side, as the liver shields the right.  The diaphragm heals poorly and delay in diagnosis may result in expansion of the injury or herniation of abdominal organs into the chest cavity.  These complications may present months to years after the injury and are very difficult to repair.1

CT diagnosis of a diaphragmatic injury is usually made by direct visualization of the irregularity or discontinuity in the diaphragm.  Unfortunately, the diaphragm is thin, mobile, and difficult to image.  CT has fairly poor sensitivity for diaphragmatic injuries.  One study found CT was only 82% sensitive and 88% specific for a diaphragm injury.15  For those watching at home, that means that nearly a fifth of stable patients with no diaphragm injury identified by CT went on to have one diagnosed by laparoscopy at 48 hours.  Another study, a retrospective review of blunt diaphragm injuries identified by laparotomy, found that only 57% of pre-op CT scans had signs of diaphragmatic damage.12

So how do we identify these small, slippery, and potentially serious lesions?  Clinically correlate, of course.  Be aware of injuries that have the potential to cross between the thoracic and abdominal cavities (for example, penetrating trauma as high as the fourth rib).  Know that CT will be negative for a significant percentage of patients with diaphragmatic injury.  If your suspicion is high enough, discuss laparoscopy with your friendly local trauma surgeon.1


Case 2

A 76-year-old woman with a past medical history of hypertension, diabetes, hyperlipidemia, hypothyroidism, dementia, and atrial fibrillation presents to the emergency department with abdominal pain.  She is mildly tachycardic, tachypneic, and normotensive, with an irregular heartbeat.  She is moaning in pain and holding her abdomen, but is unable to characterize or localize the pain.  Her abdomen is non-distended, soft, and diffusely tender with voluntary guarding.  Her home health aid has her medication list, which includes an 81 mg aspirin, but is unsure who her primary doctor is.  She has never had a colonoscopy.  CT abdomen/pelvis shows a small amount of pelvic free fluid and thickening of the small bowel wall.


Non-traumatic Abdominal Pathology

Mesenteric Ischemia

The classic presentation of mesenteric ischemia is abdominal pain out-of-proportion to exam.  Patients may lose blood supply to their bowels by a variety of mechanisms: generally embolic, thrombotic, or hypovolemic (dissection and vasculitis are less common causes).5  Each of these patients will have a different history, for example, atrial fibrillation in the patient with embolic ischemia.

While CT is generally the best imaging modality for mesenteric ischemia, CT findings are relatively non-specific.  The most common finding is bowel wall thickening.  This may be associated with the target sign, or an alternating high and low attenuation pattern due to submucosal hemorrhage or edema.  Other findings include free fluid and bowel dilation.  More specific signs include bowel pneumatosis and ischemia to other abdominal organs such as the liver and spleen (indicating that clots have been showered from a source like the heart).  Finally, CT angiography may offer visualization of the clot itself, filling defects, or gas in the intestinal vessels.5,7

For further discussion, go here: http://www.emdocs.net/mesenteric-ischemia-power-review/


Torsion is the twisting of an organ around its blood supply.  While virtually any organ can torse, the ones that will be missed by CT are ovaries and testicles.  Ovarian torsion presents with sharp lower abdominal pain/tenderness and adnexal tenderness on bimanual exam.  There may be a palpable mass either from the torsed ovary itself, the twisted vascular pedicle, or a mass that caused the torsion in the first place.  CT may show displacement of the ovary toward the midline, enlargement, surrounding inflammatory changes, and uterine deviation toward the affected side.

Ultrasound is the imaging modality of choice.  The most common finding is an enlarged ovary with heterogeneous echotexture.  The follicles may also move peripherally giving the “string of pearls” sign.  Unlike testicular ultrasound, Doppler plays virtually no role.  The ovary has a dual blood supply, so a torsed ovary may have flow.  A normal ovary can also have no flow, and flow asymmetry is common based on menstrual cycle phase.  The whirlpool sign may be present, which is color Doppler of the actual blood vessels in the pedicle twisted around each other.  See further discussion here: http://www.emdocs.net/ovarian-torsion-pearls-and-pitfalls/

Testicular torsion occurs when the testicle spins on its spermatic cord causing acute scrotal pain, pain with palpation, and a high-riding testicle in transverse lie.  Unlike ovarian torsion, Doppler is key.  The money is in the “buddy shot” of the testicles side-by-side showing asymmetric or absent flow to one.  Grey scale abnormalities usually don’t appear until it is too late.  Remember, a testicle is 100% salvageable at 6 hours, 20% at 12 hours, and approaches 0% at 24 hours, so don’t forget to check below the waist in a male patient with abdominal pain.2  See further discussion here: http://www.emdocs.net/testicular-torsion-pearls-and-pitfalls/


CT scan is only about 75% sensitive for gallstones.  Gallstones often have the same density as bile, which makes them invisible on CT.  Ultrasound is the best imaging modality for gallstones, with 96% accuracy, and should be used first to assess right upper quadrant pain.3

Contrary to popular belief, CT does fairly well identifying acute cholecystitis: 91.7% sensitive and 99.1% specific in one paper.8  Research directly comparing CT and ultrasound is limited, and most studies are small and showed their effectiveness to be similar.  In 1981, a combination of major and minor criteria was found to be 100% sensitive and 96% specific for cholecystitis.14  However, in a 2012 study of over 5,000 patients, ultrasound is only 81% sensitive and 83% specific for cholecystitis.4

Ultrasound is still the preferred modality for diagnosing cholecystitis due to easy access, speed, and lack of radiation.  The findings are the same for CT and ultrasound: hydrops, wall thickening or edema, and pericholecystic fluid.  CT will likely not pick up a stone in the neck of the gall bladder.  CT may be helpful for atypical presentations of acute cholecystitis, acalculous cholecystitis, and to identify complications such as gallbladder perforation.  However most stable patients who are highly suspicious for biliary disease should undergo ultrasound and follow up with MRCP.  For a fantastic review of pathologies and imaging modalities check out “Evaluating Patients with Right Upper Quadrant Pain” by Genevieve Bennett in the resources section.3


Resources/Further Reading:

  1. Arora, Sanjay; Menchine, Mike. “Cracking the Chest: Paper Chase 2 – CT for Diaphragm Injury.”  EM:RAP. 2016 Jan.  Accessed 4 April 2016.
  2. Ayoob, Andres R; Lee, James T. “Imaging of Common Solid Organ and Bowel Torsion in the Emergency Department.” AJR.  2014 Nov;203:W470-81.
  3. Bennett, Genevieve L. “Evaluating Patients with Right Upper Quadrant Pain.” Radiol Clin N Am. 2015;53:1093-30.
  4. Cartwright, Sarah L; Knudson, Mark P. “Diagnostic Imaging of Acute Abdominal Pain in Adults.” American Family Physician. 2015 April 1;91(7): 452-60.
  5. Gray-Eurom, Kelly; Deitte, Lori. “Imaging the Adult Patient with Nontraumatic Abdominal Pain.”  EB Medicine.  Published: February 2007.  Accessed: 2 April 2016. http://www.ebmedicine.net/topics.php?paction=showTopicSeg&topic_id=12&seg_id=84
  6. Fagenholz, Peter J, et al. “Computed Tomography is more sensitive than Ultrasound for the Diagnosis of Cholecystitis.” Surgical Infections. 2015, Oct 5;16(5): 509-12.
  7. Firetto, Maria Cristina; Lemos, Alessandro A. et al. “Acute bowel ischemia: analysis of diagnostic findings at MDCT angiography.”  Emerg Radiol. 2013;20:139-147.  DOI 10.1007/s10140-012-1078-4.
  8. Harvey, Robert T.; Miller, Wallace T. “Acute Biliary Disease: Initial CT and Follow up US versus Initial US and follow-up CT” Radiology.  1999 Dec;213(3).
  9. Hass, Christopher, et al. “Limitations of Routine Spiral Computerized Tomography in the evaluation of bladder trauma.” The Journal of Urology.  1999 July;162: 51-2
  10. Hefny, et al. “Usefulness of free intraperitoneal air detected by CT scan in diagnosing bowel perforation in blunt trauma: experience from a community-based hospital.” Injury. 2005 Jan;46(1):100-4
  11. Lo Re, Guiseppe; et al. “Small Bowel Perforations: What the Radiologist Needs to Know.” Semin Ultrasound CT MRI.  2006;37:23-30.
  12. Sprunt, Julie M, et al. “Computed Tomography to Diagnosed Blunt Diaphragm Injuries: Not Ready for Prime Time.”  The American Surgeon. 2014 Nov 11;80(11): 1124-7.
  13. Walker, Mark L.; et al. “The Role of Repeat Computed Tomography in the Evaluation of Blunt Bowel Injury.”  The American Surgeon. 2012 September;78(9): 979-85.
  14. Worthen, Nancy J, et al. “Cholecystitis: Prospective Evaluation of Sonography and 99mTc-HIDA Cholescintigraphy.” AJR. 1981;137:973-78.
  15. Yucel, Metin. “Evaluation of the diaphragm in penetrating left thoracoabdominal stab injuries: The role of multislice computed tomography.”    2015 Sep;46(9): 1734-7. doi:10.1016/j.injury.2015.06.022