Pancreatic injury is a relative enigma, even in modern medical practice with technology and advanced diagnostic methods. Although initially hard to diagnose, most minor pancreatic injuries are relatively easy to treat. However, a delayed diagnosis of pancreatic injury, mild or severe, is easy to diagnose but becomes a major therapeutic challenge to the medical team and a potentially disastrous situation for the patient.
The pancreas, sitting in a relatively protected position high in the retroperitoneum, is infrequently injured in typical blunt injuries (eg, from motor vehicle crashes) compared with its splenic and hepatic counterparts. Accordingly, many blunt pancreatic injuries are not immediately recognized and consequently end up causing higher morbidity and mortality rates than observed in injuries to other intraperitoneal organs.[1]
Conversely, penetrating abdominal trauma—by its very nature usually mandating emergency exploration—more frequently includes pancreatic injury. But even physical visualization and examination of the pancreas in the operating room may miss an isolated ductal injury to the pancreas without adjunctive tests.
This article summarizes the findings commonly associated with pancreatic injury, available diagnostic modalities and their sensitivities, and treatment issues and options.
The overall rate of blunt pancreatic injury observed in level 1 trauma centers is rather low compared with other injuries. The pancreas is estimated to be the 10th most injured organ compared to other organs (eg, brain, spleen, liver). To consider a pancreatic injury, a trauma that occurred from a significant force is usually required. The incidence of diagnosed pancreatic injury is expected to be higher at a trauma center specializing in serious injury than at a community hospital. Of 100 patients with blunt trauma, fewer than 10 will have a documented pancreatic injury.
The incidence of pancreatic injury in patients with a penetrating trauma is much higher. Gunshot wounds (GSWs); shotgun injuries; and stabbings to the back, flank, and abdomen (defined as nipple level to inguinal ligament) frequently include pancreatic injury, occurring in approximately 20-30% of all patients with penetrating traumas.[2] This occurrence elicits another key point in pancreatic trauma: Because the blunt force required to injure the pancreas is so significant and penetrating trauma usually injures multiple organs, a pancreatic injury is rarely a solitary injury. When the pancreas is injured, with the possible exception of child abuse or the well-placed stab in the back, the physician or surgeon can be confident that other organs are also affected. Therefore, multiple organ injury is a red flag suggesting the possibility of a pancreatic injury.
Because of its anatomic position (see Relevant Anatomy), an isolated pancreatic injury may occur with penetrating trauma to the mid back in the form of stab wounds or impalement. In a blunt trauma–induced isolated pancreatic injury, fracture over the spinal column is usually observed in smaller children and is caused by direct abdominal blows from malpositioned seat belts or intentional child abuse. Fortunately, both of these situations are relatively rare.
Usually, penetrating trauma caused by firearms results in the highest frequency of pancreatic injury and is almost always associated with concurrent injury to other intra-abdominal organs. This injury can result in a relatively simple isolated puncture of the body or tail of the pancreas (a highly complex and difficult injury) or an injury to the pancreatic head with involvement of the biliary and pancreatic ductal systems. In addition, the proximity of the larger vessels (eg, portal vein), the abdominal aorta, and the inferior vena cava (IVC) to the pancreatic head increases the risk of exsanguinating hemorrhage accompanying pancreatic penetrating injury. Exsanguinating hemorrhage due to concomitant vascular injury accounts for the greatest number of deaths in patients with pancreatic injury.
Unlike the spleen, few data suggest that preexisting primary or secondary diseases of the pancreas result in a higher risk of injury or a higher mortality rate when the pancreas is injured. Clearly, preexisting severe pancreatitis or diabetes mellitus negatively affects the overall mortality and morbidity rates in patients with pancreatic trauma, but few published data support this commonly held clinical view.
However, the postinjury development of pancreatitis or diabetes mellitus is a different issue. The development of either of these conditions after injury is associated with a significant increase in morbidity and overall mortality rates in patients who experience trauma.
The type of injury (ie, blunt vs penetrating) and information about the injuring agent (eg, GSW, knife) help focus the clinician on the possibility of pancreatic injury.
During the physical examination, seat belt marks, flank ecchymoses, or penetrating injuries should alert the physician to the potential for pancreatic injury. Pancreatic injury can be frighteningly symptom free early in the postinjury time frame and even silent in many cases. Rarely, a contained fracture of the spleen with retroperitoneal hematoma or leak manifests as dull epigastric pain or back pain, but the more common scenario is for patients to exhibit severe peritoneal irritation and a positive abdominal examination finding, usually caused by injury to other organs. Symptoms of injury to other structures commonly mask or supersede that of pancreatic injury, both early and late in the hospital course. Therefore, a high degree of clinical awareness is necessary to ensure that pancreatic injuries are not overlooked or missed, either early in the course of trauma or later in the ICU when the patient is not clinically improving as expected.
CT scan findings of pancreatic trauma are generally categorized as direct signs (eg, pancreatic laceration) that tend to be specific but lack sensitivity and indirect signs (eg, peripancreatic fluid) that tend to be sensitive but lack specificity.[3]
In patients experiencing blunt trauma, CT scans provide the best overall method for diagnosis and recognition of a pancreatic injury. Retroperitoneal hematoma, retroperitoneal fluid, free abdominal fluid, and pancreatic edema frequently accompany injuries to the pancreas.
In patients with penetrating trauma, visualization of perforation, hemorrhage or fluid leak (eg, bile, pancreatic fluid), or retroperitoneal hematoma around the pancreas suggests the need for further evaluation.
In cases in which the CT scan findings are indefinite or clinical suspicion of a pancreatic injury remains, further investigation with magnetic resonance cholangiopancreatography (MRCP) may be used. Although injury of the main pancreatic duct may be strongly suggested upon initial CT scanning, MRCP can demonstrate clear delineation of the duct and its integrity.[3]
Located in a relatively protected area of the abdominal cavity, the pancreas is high and posteriorly situated in a fixed retroperitoneal position. The rib cage provides a bony structural protection, in addition to the protection afforded by the thick dorsal muscle groups (paraspinous). Anteriorly, the mature adult rectus and abdominal muscles, combined with the energy-absorbing characteristics of the liver, colon, duodenum, stomach, and small bowel, provide physiologic padding that protects the pancreas from blunt injury. In severe blunt trauma, the anatomic position may result in injury (eg, fracture of the body over the spinal column and vertebral bodies posteriorly). However, the anatomic position of the pancreas neither protects nor increases the risk from penetrating injury.
The proximity of vascular structures to the head of the pancreas has a marked effect on the morbidity and mortality rates of patients who experience a pancreatic injury. The subhepatic IVC and the aorta sit just posterior to the pancreatic head to the patient's right side, and the superior mesenteric vein coalesces into the portal vein immediately behind the pancreas. Exsanguinating hemorrhage from concurrent injury to these vessels is a frequent cause of death in patients with a pancreatic injury.
The splenic artery (off the celiac trunk) and vein (draining into the portal vein) run superior and posterior to the body and tail of the pancreas and are relatively easier to expose and control compared to the IVC and portal vein. The vascular anatomy that causes such difficulty in repairing injuries to the head of the pancreas actually makes injuries to the body and tail easier to manage.
No absolute contraindications exist for pancreatic exploration or resection in patients who are experiencing trauma. The presence of hypothermia, dilutional coagulopathy, and other fatal or near-fatal injuries obviously influences the surgeon's decision to use damage-control techniques versus operative repair or resection. A trauma surgeon may explore and widely drain, perform a segmental resection, or even, very rarely, perform a trauma Whipple procedure (pancreatic duodenectomy) initially or in delayed fashion, depending on the presence of other injuries and the physiologic condition of the patient.
Blood work is notoriously unreliable when used to help diagnose blunt or penetrating trauma to the pancreas. Amylase and lipase levels may be within reference ranges, even in the presence of ductal disruption and pancreatic transection. Normal blood work findings do not guarantee or exclude pancreatic injury. Elevation in amylase levels is suggestive of pancreatic injury or inflammation but is not diagnostic. Elevated amylase levels in trauma may be from salivary glands, small bowel injury, ovarian injury, or a host of other sources. In some studies, amylase elevation was not observed until 3-4 hours after injury and, in some cases, not at all. Lipase levels are no more specific for pancreatic injury. This method has been used more frequently in children, but recent reports suggest that it may not be as cost-effective for screening pediatric pancreatic injuries as once thought.
Amylase detected in diagnostic peritoneal lavage (DPL) fluid is much more sensitive and specific for pancreatic injury than blood or serum amylase determinations. However, this study is not a standard or routine test in most institutions and may take significantly longer than anticipated to perform and to receive results unless previous arrangements or protocols are in effect with the diagnostic laboratory.
Plain films of the abdomen may show pancreatic calcification from previous episodes of pancreatitis but are rarely of any benefit in detecting blunt trauma. These films can be valuable in detecting penetrating trauma by visualizing and localizing foreign bodies such as bullet fragments and projectile-induced bony injury. Frequently, these films can be obtained simultaneously with the bony pelvis films advocated by advanced trauma life support (ATLS) protocols in trauma victims by use of a larger x-ray plate and widening of the field of x-ray exposure.
Kidney, ureter, bladder (KUB) film or upright abdominal films rarely provide useful information and only serve to delay the implementation of further care or diagnostic measures.
While not specifically useful in the detection of pancreatic trauma, the upright chest film may show free air under the diaphragm, which is suggestive of an associated gastric, duodenal, or small bowel injury and is frequently associated with pancreatic injury.
CT scan
In a patient who is hemodynamically stable, a CT scan provides the safest and most comprehensive means of diagnosis of pancreatic injury. Unfortunately, the sensitivity of this modality is reported only to be in the range of 40-68%, so patients must be monitored closely if pancreatic injury is suspected. Most reports have examined single detector or spiral scanners, but 64 detector helical scanners that are now available may provide a more accurate determination of injury. A CT scan may be augmented by the judicious use of ERCP in select cases. Laparotomy has a higher sensitivity but is not a reasonable screening test for all suspected cases of pancreatic injury.
A CT scan of the abdomen provides the simplest and least invasive diagnostic modality currently available to aid in the detection of a stable blunt pancreatic injury.[4] However, this study is only rarely useful in acute penetrating injury.
A workup for patients who are stable and have knife wounds to the back or flank may include a CT scan, but a patient who is unstable must never be placed in the CT scanner, whether the injury is blunt or penetrating trauma.
CT scan is contraindicated in patients who are hemodynamically unstable or who have a penetrating trauma in which the decision for operative intervention has been made.
A CT scan of the pancreas is also useful in the follow-up care of patients with a pancreatic injury and trauma. Traumatic pancreatic cysts, pseudocysts, delayed ductal injury, pancreatic transection, pancreatitis, abscess, pancreatic necrosis, and splenic artery aneurysms may be noted after surgery or after the patient is released from the hospital.[5]
Magnetic resonance cholangiopancreatography (MRCP) is being used more frequently in level 1 Trauma Centers to assess injury to the ductal components but has not been prospectively compared to CT or other modalities.
An evaluation of the utility of multidetector CT (MDCT) with MRI correlation in patients with blunt pancreatic trauma concluded that MDCT performs well in grading pancreatic injury and evaluating pancreatic ductal injury. The study further concluded that MRI is useful in evaluation of acute pancreatic trauma as it can increase diagnostic confidence and provide more qualitative information regarding the extent of injury.[6]
Intraoperative cholangiograms and pancreatic ductograms, especially with reflux into the pancreatic ducts (eg, Wirsung, Santorini), frequently provide information regarding the status of the injured pancreas when direct visualization is not helpful. Some authors recommend that these studies be performed during operative exploration, noting that they may help decrease complications due to missed pancreatic ductal injuries.
In the patient who is unstable, operative exploration provides the optimal diagnostic tool for pancreatic injury. As in blunt trauma, endoscopic retrograde cholangiopancreatography (ERCP) or intraoperative dye studies may provide more information in a select patient population.
ERCP is increasingly being used to help diagnose, both immediately and in delayed fashion, the presence of pancreatic ductal injuries. Some authors suggest early ERCP (ie, within 6-12 h of injury) to minimize delayed complications. While extremely helpful, this procedure has potential complications that can limit its usefulness in patients with pancreatic trauma. For it to be of benefit, the endoscopist must be skilled and experienced in its use in the injured and potentially severely ill trauma patient. This is especially true when used in the operating room in a patient with an open abdomen who is at risk for hypothermia with exposed abdominal contents.
Histologic examination of the resected pancreas documents the presence of hemorrhage and, frequently, of crush injuries to the tissue. Occasionally, this examination may reveal chronic preexisting pancreatic conditions such as pancreatitis, saponification, scarring, or tumors.
Visit the American Association for the Surgery of Trauma (AAST) Web site for a published injury scoring system for pancreatic trauma that correlates to morbidity and mortality.
In the early 1900s, observation of pancreatic injury was associated with a 100% mortality rate. However, more recently, the medical literature supports observation in select blunt injuries to the pancreas. The standard of care in penetrating injuries is still operative exploration.
Patients who have experienced blunt trauma and who have stable hemodynamics and CT scans showing no evidence of pancreatic parenchymal fracture, parenchymal hematoma, parenchymal edema, fluid in the lesser sac, or retroperitoneal hematoma may be observed but should not be considered to be cleared for pancreatic injury for at least 72 hours. Any patient with blunt trauma who continues to have abdominal pain or who develops symptoms of pancreatic injury should be thoroughly reassessed for pancreatic injury and operative intervention. Cuenca and Islam reported success with nonoperative treatment of low-grade injuries.[7]
Surgery is by far the most common therapeutic modality for patients with pancreatic trauma, especially in those with penetrating trauma, in whom exploratory laparotomy is both a diagnostic and therapeutic measure. Vasquez et al (2001) showed improved outcomes when penetrating pancreatic injury therapy was based on injury grade and location.
In most cases of blunt injury, ductal damage can be visualized directly. Most commonly, the damage is minor and such findings as capsular tears, superficial lacerations, bullet wounds of the body or tail, small contusions, or hematomas should be visualized and documented; however, they should not be surgically explored unless ductal injury is suspected. Soft closed suction drains (eg, Jackson-Pratt, Blake) should be used. The author usually uses 2 drains, but a single drain may suffice in some situations. Continued drainage with high amylase levels persisting beyond 48-72 hours is highly suggestive of a missed ductal injury. These problems must be treated with workup of the ductal integrity with ERCP or another modality and may require another operation. Occasionally, a trial of total parenteral nutrition (TPN) or elemental diet through a feeding jejunostomy may result in decreased drainage and closure of the leak.
Significant blunt trauma to the pancreas with parenchymal fracture is easily visualized on exploration by noting the retroperitoneal hematoma around the pancreas at the spinal column. Patients with ductal injury usually require resection of the tail and body distal to the vertebral column, and patients with documented intact ductal systems may be drained and observed.
More details regarding pancreatic surgical technique can be found in the Intraoperative details section below. Combined pancreatic and duodenal injuries are not reviewed in this article.
Adherence to ATLS standards and adequate preparation for emergency surgery in patients with pancreatic injury reduces morbidity and mortality rates. Early surgical intervention with identification of ductal injuries has been shown to reduce the incidence of early and late complications and death.
In most cases of blunt injury, surgical resection is not necessary. Small or superficial capsular tears, contusions, or hematomas are best managed without sutures. Wide drainage with soft closed suction drains suffices in 80-90% of patients with pancreatic injuries. Pancreatic parenchymal transection against the vertebral bodies may require resection of the body with oversewing of the distal duct with a nonabsorbable suture and drainage of the pancreatic bed. If possible, the distal parenchyma should also be oversewed, but contused, edematous pancreatic parenchyma is notoriously difficult to sew, and drainage of the bed may be all that is possible in this situation. Some surgeons use linear surgical staplers, which create a staple line that is 1.5 mm in depth.
Make sure to assess and suture ligate the pancreatic duct in these cases. Ligation of the duct has traditionally been performed with a nonabsorbable suture, but a few authors have had good results with the newer, long-lasting monofilament absorbable sutures.
Splenic preservation, although ideal, is frequently not possible with a fracture of the pancreatic body. The same anatomic orientation over the spinal column that created the parenchymal fracture and ductal injury has usually caused a splenic artery or venous injury, which results in thrombosis or aneurysmal formation and eventual splenic loss. Resection of the pancreas at the vertebral column usually leaves 40-50% of the glandular tissue, so permanent diabetes and exocrine insufficiency are unusual after resection.
Penetrating trauma
As with blunt trauma, examination and peripancreatic drainage is the most common intervention, but the range and severity of penetrating injuries to the pancreas are much more extreme. Resection of the tail or body is accomplished in a similar fashion and is technically simple, but injuries to the head and neck of the pancreas may require more creative and more difficult operative therapies.
Penetrating trauma to the head and neck of the pancreas without ductal injury can be managed with simple drainage. The appearance of bile from a penetrating injury should alert the surgeon to the possibility of a ductal injury, and a cholangiogram or ductogram is extremely valuable in helping the surgeon to plan the needed operative intervention. Isolated minor ductal damage can occasionally be stented operatively or by the interventional radiologist and should always be accompanied by an exploratory laparotomy with wide local drainage and close observation.
For patients in whom the duodenum is intact but injury to the pancreatic head and ductal system precludes repair, stenting, or local drainage, some authors suggest a Roux-en-Y pancreaticojejunostomy to preserve the pancreatic parenchyma. While theoretically feasible, the actual incidence of its use is rare (summarized in separate reviews by Graham et al and Jones). Asenio et al provide medical illustrations of many of the rarely used surgical options for trauma to the head and neck of the pancreas in the monograph on management of pancreatic injuries.
Fortunately, massive injury to the head of the pancreas, including the duodenum, rarely requires a trauma Whipple operation or pancreaticoduodenectomy. The mortality rate remains high in these cases, with even experienced centers reporting only a 50-64% survival rate. Unless the patient is actively exsanguinating, surgeons in smaller facilities with limited operating and intensive care unit (ICU) facilities should consider damage-control methods, stabilize injuries, staunch active hemorrhage, and then transfer the patient to the closest level 1 trauma center, where experience with this type of injury is more common.
The intimate anatomic position of the vena cava, portal vein, and mesenteric artery and vein with the pancreas also creates significant problems. Higher mortality rates in these cases are caused by uncontrolled hemorrhage rather than pancreatic injury; make every effort to repair these injuries first, directing attention to the pancreatic injury only after vascular integrity has been accomplished. Damage-control techniques may be necessary in these situations, addressing the pancreatic injury at a second operation, 12-48 hours later, when the patient has been warmed and stabilized.
In the recovery room, direct attention toward warming the patient; monitoring metabolic acidosis, especially in prolonged operations; and maintaining normal hemodynamic parameters. Adequate urine output, vigorous intravenous fluid replacement with crystalloid solution and blood products (as needed), and mechanical support of ventilation are necessary.
The second greatest cause of death related to pancreatic injury is noted in the ICU during the postoperative period. As might be expected, death is most common with massive injury of multiple organs and a history of significant blood loss. Acute respiratory distress syndrome (ARDS), multisystem organ failure, and infection are the most common causes of delayed death in these situations. In cases of more isolated pancreatic injury, common early complications include fistula formation, pancreatitis, and abscess formation.
Delayed presentation of complications of pancreatic injury are rare but should be considered during the follow-up period. Chronic pancreatitis, traumatic pseudocysts, and splenic artery aneurysms have been reported and should be considered in all patients.
Exocrine or endocrine insufficiency is also rare and usually occurs only in patients with a pancreatic resection greater than 80-90%. Relative insufficiency may also occur and should be considered if symptoms of altered glucose hemostasis or gastrointestinal abnormalities manifest after injury.
Complications of pancreatic injury are myriad and run the gamut from minor pancreatitis prolonging the hospital stay to death.
Fistula formation is the most frequently reported complication, but with wide local drainage and good nutrition and supportive care, fistulas usually resolve spontaneously within 2 weeks of injury. Prolonged output of greater than 250 mL/d for more than 2 weeks or outputs of 750 mL/d or more should prompt ERCP or other diagnostic evaluation of the ductal system. Somatostatin analogues have been reported anecdotally in multiple prospective randomized clinical trials to decrease fistula output and to facilitate closure, but they have not been proven to be of absolute benefit. The commercially available analogue is expensive and, if used, should be closely monitored for its effects in individual patients.
Delayed diagnosis and delayed surgery while under observation show a higher rate of pancreas-specific morbidity and mortality. Early diagnosis and therapy are associated with better overall outcomes in these patients.
More recent papers in the trauma literature have reported an increase in infectious complications in patients when pancreatic wounds occur in conjunction with hollow viscus injury (ie, duodenum, small bowel, colon). While the pathophysiology of this finding has not been well elucidated yet, experimental data in rats with pancreatic ductal ligation protected from hemorrhagic shock induced lung injury and, to a lesser degree, gut injury. Pancreatic proteases may activate digestive enzymes in ischemic intestine, which then travel via lymphatics to cause lung injury.
Delayed complications of recurrent pancreatitis, pancreatic pseudocysts, splenic artery aneurysm, and endocrine or exocrine insufficiency have also been reported.
Outcome from minor and isolated pancreatic injury is usually quite good. Complications are rare, and functional outcome is excellent.
Outcome from severe pancreatic injury is much poorer overall than with other organ injuries. This outcome is primarily due to the frequent presence of associated life-threatening injuries and the unforgiving nature of the pancreas, both for missed injury and after major emergency operative intervention.
The future will continue to bring better and faster diagnostic modalities (eg, faster and more precise CT technology, including CT cholangiopancreatography). As these new methods are examined and compared to current technology, the evolution of care of the pancreatic injury will improve and morbidity rates will decrease. Multidetector CT scanners and MRCP are emerging as more sensitive diagnostic tools, but further clinical trials will be necessary to prove their effectiveness in pancreatic injury.
H Scott Bjerke, MD, FACS, Clinical Associate Professor, Department of Surgery, University of Missouri-Kansas City School of Medicine; Medical Director of Trauma Services, Research Medical Center; Clinical Professor, Department of Surgery, Kansas City University of Medicine and Biosciences
Disclosure: Nothing to disclose.
Specialty Editors
Francisco Talavera, PharmD, PhD, Adjunct Assistant Professor, University of Nebraska Medical Center College of Pharmacy; Editor-in-Chief, Medscape Drug Reference
Disclosure: Received salary from Medscape for employment. for: Medscape.
Robert L Sheridan, MD, Assistant Chief of Staff, Chief of Burn Surgery, Shriners Burns Hospital; Associate Professor of Surgery, Department of Surgery, Division of Trauma and Burns, Massachusetts General Hospital and Harvard Medical School
Disclosure: Received research grant from: Shriners Hospitals for Children; Physical Sciences Inc, Mediwound.
Chief Editor
John Geibel, MD, MSc, DSc, AGAF, Vice Chair and Professor, Department of Surgery, Section of Gastrointestinal Medicine, Professor, Department of Cellular and Molecular Physiology, Yale University School of Medicine; Director of Surgical Research, Department of Surgery, Yale-New Haven Hospital; American Gastroenterological Association Fellow; Fellow of the Royal Society of Medicine
Disclosure: Nothing to disclose.
Additional Contributors
Ernest Dunn, MD, Program Director, Surgery Residency, Department of Surgery, Methodist Health System, Dallas
Adamson WT, Hebra A, Thomas PB, et al. Serum amylase and lipase alone are not cost-effective screening methods for pediatric pancreatic trauma. J Pediatr Surg. 2003 Mar. 38(3):354-7; discussion 354-7.
Cohen DB, Magnotti LJ, Lu Q, et al. Pancreatic duct ligation reduces lung injury following trauma and hemorrhagic shock. Ann Surg. 2004 Nov. 240(5):885-91.
Ilahi O, Bochicchio GV, Scalea TM. Efficacy of computed tomography in the diagnosis of pancreatic injury in adult blunt trauma patients: a single-institutional study. Am Surg. 2002 Aug. 68(8):704-7; discussion 707-8.
Mayer JM, Tomczak R, Rau B, et al. Pancreatic injury in severe trauma: early diagnosis and therapy improve the outcome. Dig Surg. 2002. 19(4):291-7; discussion 297-9.
Olah A, Issekutz A, Haulik L, Makay R. Pancreatic transection from blunt abdominal trauma: early versus delayed diagnosis and surgical management. Dig Surg. 2003. 20(5):408-14.
Tyburski JG, Dente CJ, Wilson RF, et al. Infectious complications following duodenal and/or pancreatic trauma. Am Surg. 2001 Mar. 67(3):227-30; discussion 230-1.
Vasquez JC, Coimbra R, Hoyt DB, Fortlage D. Management of penetrating pancreatic trauma: an 11-year experience of a level-1 trauma center. Injury. 2001 Dec. 32(10):753-9.
Wolf A, Bernhardt J, Patrzyk M, Heidecke CD. The value of endoscopic diagnosis and the treatment of pancreas injuries following blunt abdominal trauma. Surg Endosc. 2005 May. 19(5):665-9.
