Penetrating Cardiac Injuries Safuh Attar, MD, Charles M. Suter, PhD, John R. Hankins, MD, Alejandro Sequeira, MD, and Joseph S. McLaughlin, MD Division of Thoracic and Cardiovascular Surgery, Department of Surgery, University of Maryland School of Medicine and Hospital, Baltimore, Maryland
One hundred nine penetrating cardiac injuries were reviewed: 49 gunshot wounds and 60 stab wounds. They were classified into four groups: group 1 (lifeless), 38; group 2 (agonal), 16; group 3 (shock), 33; and group 4 (stable), 22. Thirty-six patients in group 1 (94%)and 8 of 16 patients in group 2 (50%)underwent emergency room thoracotomy; 24 of 33 in group 3 (73%)and 20 of 22 (90%) underwent thoracotomy in the operating room. Twentyone (38%)of 55 patients undergoing emergency room thoracotomy survived, whereas 47 (87%)of 54 patients undergoing operating room thoracotomy survived. Sur-
viva1 was 12 of 38 (31%)in group 1, 11 of 16 (69%)in group 2,26 of 33 (79%)in group 3, and 18 of 22 (82%)in group 4 with an overall survival of 67 of 109 (61%). Gunshot wounds of the heart portend a worse prognosis than stab wounds. Survival of gunshot wounds was 20 of 49 (40%)compared with 47 survivors of 60 stab wounds (78%).Aggressive treatment, including emergency room thoracotomy, is justified for lifeless and deteriorating cardiac injury victims.
P
area of the emergency room. The cardiac tamponade was relieved by pericardiotomy, and bleeding from the cardiac wound was controlled by finger pressure and pledgeted sutures. If the condition of the patient stabilized, then he or she would be moved to the operating room for further repair of the heart wound, if necessary, and control of other thoracic and abdominal injuries. Patients who arrived in stable condition were taken to the operating room for repair of the cardiac injury. The patients ranged in age from 12 to 84 years (mean age, 32.6 years). There were 97 male and 12 female patients. They were classified into four groups according to their clinical status on admission, as suggested by Ivatury and associates [l ] (Table 1). Group 1 comprised 38 patients who arrived unconscious and without vital signs. Group 2 comprised 16 patients who were semiconscious with gasping respiration, a thready pulse, and no measurable blood pressure. Group 3 comprised 33 patients who were hypotensive with a blood pressure of 80 mm Hg or less. Group 4 comprised 22 patients who were in stable condition with a systolic blood pressure of 90 mm Hg or greater. There were 49 (45%)gunshot wounds and 60 (55%)stab wounds; the distribution in each group is shown in Figure 1. There were 55 emergency room thoracotomies, compared with 54 operating room thoracotomies. Actually, the operating room approach was through a left anterolateral thoracotomy in 31 cases and through a median sternotomy in 23 cases. Pericardiocentesis was performed before thoracotomy in 4 patients early in the series, and pericardial window was performed in 10 patients; two of the latter procedures proved to give false-negative results. The right ventricle was the most commonly involved cardiac chamber in 51 patients, followed by the left ventricle in 30 patients, the right atrium in 11 patients, and the left atrium in 4 patients (Fig 2). Multiple cardiac
enetrating cardiac injuries continue to increase in proportion to the steady rise in violence in our society. The method of assault has changed from knives and icepicks to more lethal low-velocity hand guns. With the widespread use of rapid medical transportation systems, many of the victims who would have been declared dead at the scene are reaching the trauma centers in urban areas “lifeless” or in extremis. With aggressive resuscitative therapy and emergency room thoracotomy, the salvage rate of these patients can reach up to 35%. This study reviews our experience with penetrating cardiac injuries over the past 23 years.
For editorial comment, see page 701.
Material and Methods One hundred nine patients with penetrating wounds of the heart were operated on at the University of Maryland Hospital between 1967 and 1990. Most of the patients were brought to the emergency room area within 30 minutes of the time of injury. The directional and anatomical location of the site of injury suggested a presumptive diagnosis of cardiac injury. Resuscitative measures including endotracheal intubation, volume replacement, and placement of chest tubes were carried out soon after admission to the emergency room. If the patient was “lifeless” or in extremis and did not respond to resuscitative measures, an emergency anterolateral thoracotomy was performed in a designated Presented at the Thirty-seventh Annual Meeting of the Southern Thoracic Surgical Association, Dorado, Puerto Rico, Nov 6-10, 1990. Address reprint requests to Dr Attar, Division of Thoracic and Cardiovascular Surgery, University of Maryland Medical Systems, 22 South Greene St, Baltimore, MD 21201.
0 1991 by The Society of Thoracic Surgeons
(Ann Thoruc Surg 1991;51:711-6)
0003-4975/91/$3.50
712
ATTAR ET AL PENETRATING CARDIAC INJURIES
Ann Thorac Surg 1991;51:711-6
Table 1. Classification According to Clinical Status on Admission Group
GS S Total No. Wounds Wounds of Patients
1
27
11
38
2
4
12
16
3
10
23
33
4
8
14
22
Total 49 (45%) 60 (55%) BP
=
blood pressure;
Clinical Status Unconscious, no vital signs, some sign of life in transit to hospital Semiconscious, thready pulse, gasping respiration, no measurable BP Hypotensive (BP, 330 mm Hg), conscious Stable
109
GS = gunshot;
S = stab.
chamber involvement was noted in 13 patients: the right and left ventricles in 7, the right ventricle and right atrium in 3, and the right ventricle and left atrium in 1. Involvement of the ventricular septum was noted in 4 patients, transection of the left anterior descending coronary artery proximally in 6 patients, and circumflex coronary artery transection in 2. Mitral valve disruption resulting in mitral regurgitation occurred in a stab wound of the left ventricle. The superior and inferior venae cavae were both disrupted in 2 patients, the inferior vena cava in 4 patients, the superior vena cava in l patient, the internal jugular vein in 1 patient, and aorta in 2 patients. A left anterior descending-right coronary artery-ventricular fistula developed in 1 patient. Most of the cardiac wounds were treated by simple suture repair over Teflon pledgets. Emergency coronary artery bypass grafting was necessitated in 3 patients, with reconstruction of the anterior walls of the right and left ventricles using a Dacron patch in 1 of them. Associated injuries were present in 77 patients. The
Fig I . Distribution of patients according to clinical group and mode of penetrating cardiac inju y.IGSW = gunshot wound; SW = stab wound.)
lungs were involved in 24 patients, liver in 15, stomach and intestines in 13, colon in 2, diaphragm in 4, spleen in 5, gallbladder in 3, pancreas in 2, kidney in 3, extremity in 6, and spine in 1. Quantification of the anatomical extent of injury to the heart was done according to the method suggested by Ivatury and associates [l].The method consists of assigning a risk factor for each thoracic organ based om the reported mortality and morbidity rates from injury to that organ. For example, the heart and major vessels of the thorax receive a score of 5, whereas the lung is given a risk factor of 4. The extent of injury to each organ is graded on a scale of 1 to 5, 1 being minor and 5 being the most severe. The penetrating cardiac trauma index is organ risk factor (5) x injury severity estimate. The product of the risk factor and the estimate of the severity of injury equals the organ injury score. The sum of the thoracic organ injury scores in a patient constitutes the penetrating thoracic trauma index (PTTI). In patients with combined thoracoabdominal injury, the total extent of trauma represented by the penetrating trauma index (PTI) can be expressed by the sum of thoracic and abdominal indices (PTTI + penetrating abdominal trauma index [PATI]). Penetrating abdominal trauma index is measured b y the method of Moore and co-workers (21. The clinical status of the patient is represented by a graded physiologic index (PI)of increasing severity from 5 to 20. Patients who are in stable condition (group 4) have a PI of 5, patients in shock (group 3) have a PI of 10, group 2 patients have a PI of 15, and group 1 patients have a PI of 20. All data were analyzed using Statistical Analysis System, version 5 programs (SAS Institute, Cary, NC). Nominal data were analyzed using a x2 test or Fisher exact probability test for two-by-two tables. Numeric data were analyzed using the Wilcoxon rank-sum test.
Results The survival rate of penetrating cardiac injuries was related to the time elapsed between injury and initiation of resuscitation, the clinical status on arrival to the emergency room area, the mechanism of injury (whether by
Group I GSE
Stab Wounds 60
Group Ill
sw
GSW
Gun Shot Wounds 49
ATTARETAL PENETRATING CARDIAC INJURIES
Ann Thorac Surg 1991;51:711-6
Single Wound &Patients
Multiple Wounds 13 Patients
\
RV+LV
RV 51 LV 30 RA 11 L A 4
7
713
Fig 2. Cardiac chamber involvement in penetrating cardiac injuries. (A0 = aorta; CIRC = circumflex artery; LA = left atrium; LAD = left anterior descending coronary artery; LV = left ventricle; RA = right atrium; RV = right ventricle; svc = superior vena cava.)
Include 6 with transection LAD
1
stab wound or gunshot wound), the extent of injury, the need for emergency room thoracotomy for resuscitation, and the presence of cardiac tamponade. The time interval was not a factor in this series of patients, because they were all transported to the emergency room center within 30 minutes of injury. No resuscitative measures were undertaken in the field other than the application of MAST trousers or cardiopulmonary resuscitation when indicated. There were 12 survivors (31%)among 38 patients who were lifeless on initial presentation (group 1). There were 16 patients in group 2, who were semiconscious but without measurable blood pressure, with 11 survivors (69%). There were 33 patients in group 3, who were in shock, with 26 survivors (79%).Twenty-two patients were stable clinically (group 4) with 18 survivors (82%). As to the mechanism of injury, there were 60 stab wounds and 49 gunshot wounds with 20 surviving the gunshot wounds (a survival rate of 40%)and 47 surviving the stab wounds (a survival rate of 78%).The need for emergency room thoracotomy was very significant, because it reflected the critical clinical condition of the patient. There were 55 patients who required emergency room thoracotomy with 21 survivors (a survival rate of 38%) as compared with 54 patients undergoing operating room thoracotomy with 47 survivors (a survival rate of 87%).When
the clinical condition of the patient was correlated with the need for emergency room thoracotomy, there were 36 patients in group 1 with 10 survivors (a survival rate of 27%)and 8 patients in group 2 with 4 survivors (a survival rate of 50%), as compared with the 24 relatively stable patients in group 3 and 20 patients in group 4 who underwent operating room thoracotomy with 19 survivors in group 3 (a survival rate of 79%)and 18 survivors in group 4 (a survival rate of 90%) (Table 2). The PI representing the clinical status of the patient on admission to the emergency room was very significantly correlated with survival or death ( p < 0.001). The effect of the thoracic injury superimposed on the cardiac injury was not significant in affecting survival. As to the extrathoracic injuries represented by PATI, they did not affect survival. Quantification of the injuries by the Ivatury method did separate the 49 cardiac injury patients with a mean PATI of 14 from the 60 patients with multiple injuries, who had a mean PATI of 32. The difference was significant ( p < 0.0001) (Tables 3, 4). The prognostic score, which represents the sum of the PI and the PTI, was also significantly different for the cardiac injury group alone versus the cardiac injury group with extrathoracic injuries. The mean prognostic score was 27 for the cardiac group, compared with 45 in the group with multiple injuries ( p < 0.0001). Although the
Table 2. Penetrating Cardiac Injuries: Efect of Clinical Presentations and Treatment on Survival
Group 1 2 3 4
38 16 33 22
Total
109
Survival a
Gunshot Wounds
No. of
Patients
ERT 27 2 3 2
ORT
(7)a (0) (2) (0)
0 2 (2) 7 (4) 6 (5)
15 73%
34 26 %
Stab Wounds
Subtotal 27 4 10 8 49 40%
Numbers in parentheses indicate survivors.
ERT
=
emergency room thoracotomy;
ORT
=
operating room thoracotomy
Subtotal
Overall
Survival
ERT
ORT
9 (3) 6 (4) 6 (5)
2 (2) 6 (5) 17 (15) 14 (13)
11 12 23 14
31% 69% 79% 82%
39 89%
60 78%
61%
0
21 61 %
714
AlTARETAL PENETRATING CARDIAC INJURIES
Ann Thorac Surg 1991;51:7114
Table 3. Effect of Age and Indices on Survival
Table 5. Complications in Resuscitated Patients ~
Variable
No.
Mean
Standard Error
p Value NS
Age (Y) D S
42 67
30.8 34.0
21.976 21.941
42 67
84.0 81.0
20.664 20.727
PI
0.001
D S
PTTI D S PATI D S PTI D S
22.0 19.0
21.776 k1.421
NS 42 67
3.66 3.38
k1.234
-co.841 NS
42 67
26.0 23.0
Penetrating cardiac injuries have shown a progressive increase over the past two decades. Although the literature indicates the increase is mostly due to gunshot
Table 4. Factors Afecting Suwival
GSW
sw
ERT
ORT TAM+ TAM-
SCH MCH
Survivors
Survivors
49 60 55 54 60 49 96 13
20 47 21 46 40 27 60 7
40 78 38 85 67 55 62 54
9 (3)" 5 (3)
Cardiopulmonary Myocardial infarction Postcardiogenic shock Pulmonary embolus Respiratory insufficiency Cerebral Anoxic encephalopathy
Paraplegia
Comment
No. of
Coagulopathy Reexploration for bleeding Sepsis Wound infection Wound dehiscence
Total
prognostic score was different for the two groups, the survival rate was 63% (31/49) for the cardiac group and 60% (36/60) for the multiple injury group, which was not significant (see Table 3). The most frequently associated injuries were injuries to the lungs and liver. Complications encountered in resuscitated patients are summarized in Table 5. Forty patients had complications, most of which were nonfatal. The most serious complications were cerebral damage in 5 patients, all of whom died, and coagulopathy in 9 patients, 3 of whom survived.
No. of Patients
~~~
Hemorrhagic
k2.422 ?1.648
NS = not significant; D = died; PATI = penetrating abdominal PI = physiologic index; PTI = penetrating trauma trauma index; index; PTTI = penetrating thoracic trauma index.
Factor
~
No. of Patients
Empyema NS
42 67
Complication
%
p Value 0.001 0.001
NS NS
GSW = gunshot wound; ERT = emergency room thoracotomy; ORT = operating NS = not significant; MCH = multiple chambers; room thoracotomy; SCH = single chamber; SW = stab wound; TAM- = no tamponade. TAM+ = tamponade;
a
Numbers in parentheses indicate survivors.
wounds [3, 41 the current series showed a concomitant increase in both gunshot and stab wounds, the implication being that gunshot wounds carry a worse prognosis [4]. They are associated with larger defects in the pericardium and more destruction of myocardial tissue than stab wounds. Subsequently, they are more likely to produce cardiac hemorrhage that results in exsanguination, in contradiction to stab wounds, which produce a small1 rent in the pericardium that seals off and produces cardiac tamponade. Eighty to 90% of stab wounds demonstrate pericardial tamponade, but only 20% of gunshot wounds of the heart have cardiac tamponade when first seen [4]. Moreno and associates [5] observed that the presence of tamponade improved survival for patients with stab and gunshot wounds of the right as well as the left ventricle. They reported a survival rate of 73% (2433) in patients with tamponade, compared with 11% (5/44) without its protective effect. Our series does not confirm this observation, because the survival rate was not affected by tamponade. The relative frequency of involvement of the cardiac chambers depends on their anatomical location. The right ventricle is most frequently involved, considering its anterior position, followed by the left ventricle. Karrel and co-workers [4] reported the distribution of involvement of cardiac chambers in a review of 1,802 penetrating cardiac wounds. The right ventricle was involved in 765 cases (42.5%), the left ventricle in 594 cases (33%), the right atrium in 277 cases (15.4%), and the left atrium in 105 cases (5.8%).The intrapericardial great vessels were penetrated in 61 cases (33%).The present series confirms the frequency of this distribution. The overall survival in this series was 67 of 109 patients (61%), a better survival being accomplished with stab wounds, 47 survivors of 60 (78%), when compared with
Ann Thorac Surg 1991;51:7114
20 survivors of 49 gunshot wounds (40%).Moreno and associates [5] reviewed 100 consecutive unselected patients with acute cardiac injuries. Overall salvage was 31%:27 (47%)of 57 stab wounds and 4 (9%)of 43 gunshot wounds. Tavares and co-workers [6] reported 64 consecutive patients with penetrating cardiac injuries with 45 survivors (70%). Analysis of the factors affecting survival is summarized in Tables 3 and 4. The most significant factors were the clinical condition of the victim on arrival to the emergency room center represented by the PI, the method of injury (knife or gunshot), and the need for emergency room thoracotomy. The presence of tamponade, the involvement of multiple cardiac chambers, and injury to extrathoracic organs were not statistically significant. This is different from some reported series in the literature. Moreno and associates [5] reported that the presence of tamponade, vital signs, and wound site were the most critical factors for survival. Cardiac tamponade was considered to be a critical independent factor in patient survival, even more influential than presenting vital signs in determining outcome. Ivatury and associates [l] reported 112 patients with penetrating cardiac injuries. Analysis revealed that the indices, penetrating cardiac trauma index and PI, showed an excellent correlation with survival, as did the total extent of trauma (PTI). A composite prognostic score of the sum of PI and PTI demonstrated a significant separation of survivors from nonsurvivors. Our data did not confirm these findings in full. Although penetrating cardiac trauma index and PI correlated well with survival, we could not separate the survivors from nonsurvivors based on the penetrating trauma index. Serious postoperative complications after repair of cardiac wounds in this series included coagulopathy, sepsis, wound dehiscence, and encephalopathy. Coagulopathy was secondary to cardiogenic and hemorrhagic shock as well as to multiple blood transfusions. Sepsis was frequently due to Staphylococcus aureus. Coagulopathy and encephalopathy were generally associated with a fatal outcome. As to the residual and delayed sequelae of penetrating cardiac injuries in this series, ventricular septal defects developed in 4 patients after stab wounds of the heart; 2 underwent repair of the defect, and the other
ATCARETAL PENETRATING CARDIAC INJURIES
715
2 patients had small ventricular septal defects and were asymptomatic. They continue to be treated conservatively. Mitral regurgtation requiring mitral valve replacement developed in 1 patient, and pericarditis requiring pericardiectomy developed in 2 patients. All patients with residual complications who underwent surgical correction survived. Penetrating cardiac injuries will continue to increase commensurate with increased crime and violence in our cities. Moribund and lifeless patients will be brought to the emergency area. Our experience has demonstrated that aggressive resuscitation, including endotracheal intubation, volume replacement, and chest tube placement followed by resuscitative emergency thoracotomy has resulted in improved survival. Emergency room thoracotomy is indicated in penetrating cardiac injuries in patients who are "lifeless" or in extremis who do not respond to resuscitative measures. Patients with stable vital signs, and hypotensive patients who respond to resuscitative measures, should be transported t o the operating room for definitive repair. This selective approach to penetrating cardiac injuries has salvaged a significant percentage of the patients seen "lifeless" or in extremis and has resulted in excellent survival of the patients in stable condition.
Ref erenees 1. Ivatury RR, Nallathambi MN, Rohman M. Penetrating cardiac trauma: quantifying the severity of anatomic and physiologic injury. Ann Surg 1987;205:61-6. 2. Moore EE, Dunn EL, Moore JB, Thompson JS. Penetrating abdominal trauma index. J Trauma 1981;21:439-44. 3. Ivatury RR, Shah PM, Katsuki I, Ramirez-Schon G, Suarey F, Rohman M. Emergency room thoracotomy for the resuscitation of patients with fatal penetrating injuries of the heart. Ann Thorac Surg 1981;32:377-84. 4. Karrel R, Shaeffer MA, Franaszek JB. Emergency diagnosis, resuscitation, and treatment of acute penetrating cardiac trauma. Ann Emerg Med 1982;11:504-17. 5. Moreno C, Moore EE, Majure JA, et al. Pericardial tamponade: a critical determinant for survival following penetrating cardiac wounds. J Trauma 1986;26:821-5. 6. Tavares S, Hankins JR, Moulton AL, et al. Management of penetrating cardiac injuries: the role of emergency room thoracotomy. Ann Thorac Surg 1983;38:183-7.
DISCUSSION DR MICHAEL D. HOROWITZ (Miami, FL): Approximately a year ago we had a gentleman brought to the emergency room with an anterior thoracic stab wound. He had had a cardiac arrest and required an emergency room thoracotomy. He was found to have cardiac tamponade due to an injury to the right ventricle. His initial pH was 6.8. After repair and closure, the patient had persistent, profound low cardiac output. An emergency echocardiogram showed a ventricular ejection fraction of approximately 0.15. We considered that we might have missed a coronary artery lesion. Cardiac catheterization showed normal coronary arteries. We placed an intraaortic balloon pump, and within the next 12 to 24 hours his clinical picture changed very impressively. In fact,
this man turned around and survived. Late echocardiography showed normal ventricular function. I attribute much of the cardiac dysfunction to his profound shock and severe acidosis. I would like to ask if you have any experience with use of intraaortic balloon counterpulsation in the group of patients with penetrating cardiac injuries.
DR ATTAR: In our group, we had 2 patients who required intraaortic balloon pulsation after the repair of the cardiac injury. We had only 1 survivor of the 2, which is about 50%. I think it is worthwhile using the intraaortic balloon to support the patients; however, the need for it is really rare, because in most of these
716
ATTARETAL PENETRATING CARDIAC INJURIES
patients the injury is localized, especially if it is a stab wound. Stab wounds produce a very small, neat cut in the ventricle in contradistinction to gunshot wounds, which produce substantial tissue injury. We had 1 patient who required reconstruction of the left anterior ventricular wall and the right ventricle as well as a bypass graft to the left anterior descending coronary artery, and he survived. We have been following him up now for 10 years, and he has an ejection fraction of about 0.25.
DR WILLIAM A. COOK (North Andover, MA): Two years ago, the nursing supervisor in a 100-bed hospital where I work asked, “What d o you think would be desirable in a thoracotomy kit?” I asked, “What for?” She replied, “The emergency room physicians want to be prepared to d o thoracotomies.” Here is a group of physicians proposing to do thoracotomies who do not know enough about it to select the instruments. I would just like to emphasize that there is a great deal of difference between the average community hospital emergency ward and a class I trauma center that probably has an emergency room operating suite with lights, instruments, and all of the other things needed to provide standard thoracotomy in the emergency room setting and probably a full-time thoracic surgical resident. We must not be confused regarding emergency room physicians, now being trained to think about thoracotomy in their training programs, as to what they are capable of doing and equipped to d o in an emergency room thoracotomy, especially in the smaller hospital with the occasional patient and no thoracic resident. In the group of patients we looked at in Dallas, the single most important factor in survival was routine transfer of patients suspected of having a cardiac injury to the operating room, bypassing emergency room stabilization. Even though the patient was simply observed in the operating room, and perhaps never even operated on, if he or she did need to be operated on everything was available. I think this is very important. I would like to be sure that thoracic surgeons are pointing out this aspect of cardiac injury care so the patient does not get lost in the shuffle with emergency room personnel. DR ATTAR: Dr Cook has raised two issues. The first one is, who should d o emergency thoracotomy-the thoracic surgeon, the emergency room surgeon, or the general surgeon? In general, I
Ann Thorac Surg 1991;51:7114
agree that not every community hospital should have a thoracotomy kit in the emergency room. At the same time, we should not deny the fact that patients with cardiac wounds are brought to the emergency room. The staff of the ambulance service does not know whether the patient has a heart wound or a chest wound that requires operating or if it may be treated conservatively, so they bring the patient to the emergency area. The first person who sees the patient is either the emergency room surgeon or the general surgeon. In our hospital, it is the general surgical residents who attend to the initial care of the patient. But as soon as the patient hits the emergency room with a presumptive diagnosis of cardiac injury, the thoracic resident, who is on call 24 hours at our university hospital, is called on to supervise the treatment. However, the initial emergency thoracotomy is performed by the general surgical resident in certain conditions. That is, if the patient comes in with cardiac arrest or is under cardiopulmonary resuscitation, or if the patient goes into cardiac arrest, or if there is massive bleeding and deterioration in the emergency room, then the general surgical resident will undertake the emergency thoracotomy. If the patient is in relatively stable condition, having a blood pressure of 80 mm Hg or greater, then he or she can afford to have the luxury of going to the operating room. There has been a study reported in the literature in which there was a comparison of cardiac wound patients who were operated on in the emergency room versus those who were taken right away to the operating room with a delay of 7 minutes. This 7 minutes was very significant as far as survival or death in that series. Therefore, I think we should have an agreed-upon plan between the emergency room staff, the general surgical staff, and the thoracic surgical staff so that when a cardiac emergency is seen in the emergency room, it can be handled properly. In our hospital emergency room, a specific area is designated for emergency thoracotomies and is fully equipped for this procedure. The second question that Dr Cook raised was whether or not the patients should be taken directly to the operating room. Not every hospital has the luxury of having an operating room ready and available. Sometimes the operating rooms are all full, and you cannot wait until a room opens up. So, you have to take care of the emergency and at least open the chest, control the bleeding, and then take the patient to the operating room when he or she is stabilized for definitive repair.
One hundred nine penetrating cardiac injuries were reviewed: 49 gunshot wounds and 60 stab wounds. They were classified into four groups: group 1 (lifeless), 38; group 2 (agonal), 16; group 3 (shock), 33; and group 4 (stable), 22. Thirty-six patients in group 1 (94%)and 8 of 16 patients in group 2 (50%)underwent emergency room thoracotomy; 24 of 33 in group 3 (73%)and 20 of 22 (90%) underwent thoracotomy in the operating room. Twentyone (38%)of 55 patients undergoing emergency room thoracotomy survived, whereas 47 (87%)of 54 patients undergoing operating room thoracotomy survived. Sur-
viva1 was 12 of 38 (31%)in group 1, 11 of 16 (69%)in group 2,26 of 33 (79%)in group 3, and 18 of 22 (82%)in group 4 with an overall survival of 67 of 109 (61%). Gunshot wounds of the heart portend a worse prognosis than stab wounds. Survival of gunshot wounds was 20 of 49 (40%)compared with 47 survivors of 60 stab wounds (78%).Aggressive treatment, including emergency room thoracotomy, is justified for lifeless and deteriorating cardiac injury victims.
P
area of the emergency room. The cardiac tamponade was relieved by pericardiotomy, and bleeding from the cardiac wound was controlled by finger pressure and pledgeted sutures. If the condition of the patient stabilized, then he or she would be moved to the operating room for further repair of the heart wound, if necessary, and control of other thoracic and abdominal injuries. Patients who arrived in stable condition were taken to the operating room for repair of the cardiac injury. The patients ranged in age from 12 to 84 years (mean age, 32.6 years). There were 97 male and 12 female patients. They were classified into four groups according to their clinical status on admission, as suggested by Ivatury and associates [l ] (Table 1). Group 1 comprised 38 patients who arrived unconscious and without vital signs. Group 2 comprised 16 patients who were semiconscious with gasping respiration, a thready pulse, and no measurable blood pressure. Group 3 comprised 33 patients who were hypotensive with a blood pressure of 80 mm Hg or less. Group 4 comprised 22 patients who were in stable condition with a systolic blood pressure of 90 mm Hg or greater. There were 49 (45%)gunshot wounds and 60 (55%)stab wounds; the distribution in each group is shown in Figure 1. There were 55 emergency room thoracotomies, compared with 54 operating room thoracotomies. Actually, the operating room approach was through a left anterolateral thoracotomy in 31 cases and through a median sternotomy in 23 cases. Pericardiocentesis was performed before thoracotomy in 4 patients early in the series, and pericardial window was performed in 10 patients; two of the latter procedures proved to give false-negative results. The right ventricle was the most commonly involved cardiac chamber in 51 patients, followed by the left ventricle in 30 patients, the right atrium in 11 patients, and the left atrium in 4 patients (Fig 2). Multiple cardiac
enetrating cardiac injuries continue to increase in proportion to the steady rise in violence in our society. The method of assault has changed from knives and icepicks to more lethal low-velocity hand guns. With the widespread use of rapid medical transportation systems, many of the victims who would have been declared dead at the scene are reaching the trauma centers in urban areas “lifeless” or in extremis. With aggressive resuscitative therapy and emergency room thoracotomy, the salvage rate of these patients can reach up to 35%. This study reviews our experience with penetrating cardiac injuries over the past 23 years.
For editorial comment, see page 701.
Material and Methods One hundred nine patients with penetrating wounds of the heart were operated on at the University of Maryland Hospital between 1967 and 1990. Most of the patients were brought to the emergency room area within 30 minutes of the time of injury. The directional and anatomical location of the site of injury suggested a presumptive diagnosis of cardiac injury. Resuscitative measures including endotracheal intubation, volume replacement, and placement of chest tubes were carried out soon after admission to the emergency room. If the patient was “lifeless” or in extremis and did not respond to resuscitative measures, an emergency anterolateral thoracotomy was performed in a designated Presented at the Thirty-seventh Annual Meeting of the Southern Thoracic Surgical Association, Dorado, Puerto Rico, Nov 6-10, 1990. Address reprint requests to Dr Attar, Division of Thoracic and Cardiovascular Surgery, University of Maryland Medical Systems, 22 South Greene St, Baltimore, MD 21201.
0 1991 by The Society of Thoracic Surgeons
(Ann Thoruc Surg 1991;51:711-6)
0003-4975/91/$3.50
712
ATTAR ET AL PENETRATING CARDIAC INJURIES
Ann Thorac Surg 1991;51:711-6
Table 1. Classification According to Clinical Status on Admission Group
GS S Total No. Wounds Wounds of Patients
1
27
11
38
2
4
12
16
3
10
23
33
4
8
14
22
Total 49 (45%) 60 (55%) BP
=
blood pressure;
Clinical Status Unconscious, no vital signs, some sign of life in transit to hospital Semiconscious, thready pulse, gasping respiration, no measurable BP Hypotensive (BP, 330 mm Hg), conscious Stable
109
GS = gunshot;
S = stab.
chamber involvement was noted in 13 patients: the right and left ventricles in 7, the right ventricle and right atrium in 3, and the right ventricle and left atrium in 1. Involvement of the ventricular septum was noted in 4 patients, transection of the left anterior descending coronary artery proximally in 6 patients, and circumflex coronary artery transection in 2. Mitral valve disruption resulting in mitral regurgitation occurred in a stab wound of the left ventricle. The superior and inferior venae cavae were both disrupted in 2 patients, the inferior vena cava in 4 patients, the superior vena cava in l patient, the internal jugular vein in 1 patient, and aorta in 2 patients. A left anterior descending-right coronary artery-ventricular fistula developed in 1 patient. Most of the cardiac wounds were treated by simple suture repair over Teflon pledgets. Emergency coronary artery bypass grafting was necessitated in 3 patients, with reconstruction of the anterior walls of the right and left ventricles using a Dacron patch in 1 of them. Associated injuries were present in 77 patients. The
Fig I . Distribution of patients according to clinical group and mode of penetrating cardiac inju y.IGSW = gunshot wound; SW = stab wound.)
lungs were involved in 24 patients, liver in 15, stomach and intestines in 13, colon in 2, diaphragm in 4, spleen in 5, gallbladder in 3, pancreas in 2, kidney in 3, extremity in 6, and spine in 1. Quantification of the anatomical extent of injury to the heart was done according to the method suggested by Ivatury and associates [l].The method consists of assigning a risk factor for each thoracic organ based om the reported mortality and morbidity rates from injury to that organ. For example, the heart and major vessels of the thorax receive a score of 5, whereas the lung is given a risk factor of 4. The extent of injury to each organ is graded on a scale of 1 to 5, 1 being minor and 5 being the most severe. The penetrating cardiac trauma index is organ risk factor (5) x injury severity estimate. The product of the risk factor and the estimate of the severity of injury equals the organ injury score. The sum of the thoracic organ injury scores in a patient constitutes the penetrating thoracic trauma index (PTTI). In patients with combined thoracoabdominal injury, the total extent of trauma represented by the penetrating trauma index (PTI) can be expressed by the sum of thoracic and abdominal indices (PTTI + penetrating abdominal trauma index [PATI]). Penetrating abdominal trauma index is measured b y the method of Moore and co-workers (21. The clinical status of the patient is represented by a graded physiologic index (PI)of increasing severity from 5 to 20. Patients who are in stable condition (group 4) have a PI of 5, patients in shock (group 3) have a PI of 10, group 2 patients have a PI of 15, and group 1 patients have a PI of 20. All data were analyzed using Statistical Analysis System, version 5 programs (SAS Institute, Cary, NC). Nominal data were analyzed using a x2 test or Fisher exact probability test for two-by-two tables. Numeric data were analyzed using the Wilcoxon rank-sum test.
Results The survival rate of penetrating cardiac injuries was related to the time elapsed between injury and initiation of resuscitation, the clinical status on arrival to the emergency room area, the mechanism of injury (whether by
Group I GSE
Stab Wounds 60
Group Ill
sw
GSW
Gun Shot Wounds 49
ATTARETAL PENETRATING CARDIAC INJURIES
Ann Thorac Surg 1991;51:711-6
Single Wound &Patients
Multiple Wounds 13 Patients
\
RV+LV
RV 51 LV 30 RA 11 L A 4
7
713
Fig 2. Cardiac chamber involvement in penetrating cardiac injuries. (A0 = aorta; CIRC = circumflex artery; LA = left atrium; LAD = left anterior descending coronary artery; LV = left ventricle; RA = right atrium; RV = right ventricle; svc = superior vena cava.)
Include 6 with transection LAD
1
stab wound or gunshot wound), the extent of injury, the need for emergency room thoracotomy for resuscitation, and the presence of cardiac tamponade. The time interval was not a factor in this series of patients, because they were all transported to the emergency room center within 30 minutes of injury. No resuscitative measures were undertaken in the field other than the application of MAST trousers or cardiopulmonary resuscitation when indicated. There were 12 survivors (31%)among 38 patients who were lifeless on initial presentation (group 1). There were 16 patients in group 2, who were semiconscious but without measurable blood pressure, with 11 survivors (69%). There were 33 patients in group 3, who were in shock, with 26 survivors (79%).Twenty-two patients were stable clinically (group 4) with 18 survivors (82%). As to the mechanism of injury, there were 60 stab wounds and 49 gunshot wounds with 20 surviving the gunshot wounds (a survival rate of 40%)and 47 surviving the stab wounds (a survival rate of 78%).The need for emergency room thoracotomy was very significant, because it reflected the critical clinical condition of the patient. There were 55 patients who required emergency room thoracotomy with 21 survivors (a survival rate of 38%) as compared with 54 patients undergoing operating room thoracotomy with 47 survivors (a survival rate of 87%).When
the clinical condition of the patient was correlated with the need for emergency room thoracotomy, there were 36 patients in group 1 with 10 survivors (a survival rate of 27%)and 8 patients in group 2 with 4 survivors (a survival rate of 50%), as compared with the 24 relatively stable patients in group 3 and 20 patients in group 4 who underwent operating room thoracotomy with 19 survivors in group 3 (a survival rate of 79%)and 18 survivors in group 4 (a survival rate of 90%) (Table 2). The PI representing the clinical status of the patient on admission to the emergency room was very significantly correlated with survival or death ( p < 0.001). The effect of the thoracic injury superimposed on the cardiac injury was not significant in affecting survival. As to the extrathoracic injuries represented by PATI, they did not affect survival. Quantification of the injuries by the Ivatury method did separate the 49 cardiac injury patients with a mean PATI of 14 from the 60 patients with multiple injuries, who had a mean PATI of 32. The difference was significant ( p < 0.0001) (Tables 3, 4). The prognostic score, which represents the sum of the PI and the PTI, was also significantly different for the cardiac injury group alone versus the cardiac injury group with extrathoracic injuries. The mean prognostic score was 27 for the cardiac group, compared with 45 in the group with multiple injuries ( p < 0.0001). Although the
Table 2. Penetrating Cardiac Injuries: Efect of Clinical Presentations and Treatment on Survival
Group 1 2 3 4
38 16 33 22
Total
109
Survival a
Gunshot Wounds
No. of
Patients
ERT 27 2 3 2
ORT
(7)a (0) (2) (0)
0 2 (2) 7 (4) 6 (5)
15 73%
34 26 %
Stab Wounds
Subtotal 27 4 10 8 49 40%
Numbers in parentheses indicate survivors.
ERT
=
emergency room thoracotomy;
ORT
=
operating room thoracotomy
Subtotal
Overall
Survival
ERT
ORT
9 (3) 6 (4) 6 (5)
2 (2) 6 (5) 17 (15) 14 (13)
11 12 23 14
31% 69% 79% 82%
39 89%
60 78%
61%
0
21 61 %
714
AlTARETAL PENETRATING CARDIAC INJURIES
Ann Thorac Surg 1991;51:7114
Table 3. Effect of Age and Indices on Survival
Table 5. Complications in Resuscitated Patients ~
Variable
No.
Mean
Standard Error
p Value NS
Age (Y) D S
42 67
30.8 34.0
21.976 21.941
42 67
84.0 81.0
20.664 20.727
PI
0.001
D S
PTTI D S PATI D S PTI D S
22.0 19.0
21.776 k1.421
NS 42 67
3.66 3.38
k1.234
-co.841 NS
42 67
26.0 23.0
Penetrating cardiac injuries have shown a progressive increase over the past two decades. Although the literature indicates the increase is mostly due to gunshot
Table 4. Factors Afecting Suwival
GSW
sw
ERT
ORT TAM+ TAM-
SCH MCH
Survivors
Survivors
49 60 55 54 60 49 96 13
20 47 21 46 40 27 60 7
40 78 38 85 67 55 62 54
9 (3)" 5 (3)
Cardiopulmonary Myocardial infarction Postcardiogenic shock Pulmonary embolus Respiratory insufficiency Cerebral Anoxic encephalopathy
Paraplegia
Comment
No. of
Coagulopathy Reexploration for bleeding Sepsis Wound infection Wound dehiscence
Total
prognostic score was different for the two groups, the survival rate was 63% (31/49) for the cardiac group and 60% (36/60) for the multiple injury group, which was not significant (see Table 3). The most frequently associated injuries were injuries to the lungs and liver. Complications encountered in resuscitated patients are summarized in Table 5. Forty patients had complications, most of which were nonfatal. The most serious complications were cerebral damage in 5 patients, all of whom died, and coagulopathy in 9 patients, 3 of whom survived.
No. of Patients
~~~
Hemorrhagic
k2.422 ?1.648
NS = not significant; D = died; PATI = penetrating abdominal PI = physiologic index; PTI = penetrating trauma trauma index; index; PTTI = penetrating thoracic trauma index.
Factor
~
No. of Patients
Empyema NS
42 67
Complication
%
p Value 0.001 0.001
NS NS
GSW = gunshot wound; ERT = emergency room thoracotomy; ORT = operating NS = not significant; MCH = multiple chambers; room thoracotomy; SCH = single chamber; SW = stab wound; TAM- = no tamponade. TAM+ = tamponade;
a
Numbers in parentheses indicate survivors.
wounds [3, 41 the current series showed a concomitant increase in both gunshot and stab wounds, the implication being that gunshot wounds carry a worse prognosis [4]. They are associated with larger defects in the pericardium and more destruction of myocardial tissue than stab wounds. Subsequently, they are more likely to produce cardiac hemorrhage that results in exsanguination, in contradiction to stab wounds, which produce a small1 rent in the pericardium that seals off and produces cardiac tamponade. Eighty to 90% of stab wounds demonstrate pericardial tamponade, but only 20% of gunshot wounds of the heart have cardiac tamponade when first seen [4]. Moreno and associates [5] observed that the presence of tamponade improved survival for patients with stab and gunshot wounds of the right as well as the left ventricle. They reported a survival rate of 73% (2433) in patients with tamponade, compared with 11% (5/44) without its protective effect. Our series does not confirm this observation, because the survival rate was not affected by tamponade. The relative frequency of involvement of the cardiac chambers depends on their anatomical location. The right ventricle is most frequently involved, considering its anterior position, followed by the left ventricle. Karrel and co-workers [4] reported the distribution of involvement of cardiac chambers in a review of 1,802 penetrating cardiac wounds. The right ventricle was involved in 765 cases (42.5%), the left ventricle in 594 cases (33%), the right atrium in 277 cases (15.4%), and the left atrium in 105 cases (5.8%).The intrapericardial great vessels were penetrated in 61 cases (33%).The present series confirms the frequency of this distribution. The overall survival in this series was 67 of 109 patients (61%), a better survival being accomplished with stab wounds, 47 survivors of 60 (78%), when compared with
Ann Thorac Surg 1991;51:7114
20 survivors of 49 gunshot wounds (40%).Moreno and associates [5] reviewed 100 consecutive unselected patients with acute cardiac injuries. Overall salvage was 31%:27 (47%)of 57 stab wounds and 4 (9%)of 43 gunshot wounds. Tavares and co-workers [6] reported 64 consecutive patients with penetrating cardiac injuries with 45 survivors (70%). Analysis of the factors affecting survival is summarized in Tables 3 and 4. The most significant factors were the clinical condition of the victim on arrival to the emergency room center represented by the PI, the method of injury (knife or gunshot), and the need for emergency room thoracotomy. The presence of tamponade, the involvement of multiple cardiac chambers, and injury to extrathoracic organs were not statistically significant. This is different from some reported series in the literature. Moreno and associates [5] reported that the presence of tamponade, vital signs, and wound site were the most critical factors for survival. Cardiac tamponade was considered to be a critical independent factor in patient survival, even more influential than presenting vital signs in determining outcome. Ivatury and associates [l] reported 112 patients with penetrating cardiac injuries. Analysis revealed that the indices, penetrating cardiac trauma index and PI, showed an excellent correlation with survival, as did the total extent of trauma (PTI). A composite prognostic score of the sum of PI and PTI demonstrated a significant separation of survivors from nonsurvivors. Our data did not confirm these findings in full. Although penetrating cardiac trauma index and PI correlated well with survival, we could not separate the survivors from nonsurvivors based on the penetrating trauma index. Serious postoperative complications after repair of cardiac wounds in this series included coagulopathy, sepsis, wound dehiscence, and encephalopathy. Coagulopathy was secondary to cardiogenic and hemorrhagic shock as well as to multiple blood transfusions. Sepsis was frequently due to Staphylococcus aureus. Coagulopathy and encephalopathy were generally associated with a fatal outcome. As to the residual and delayed sequelae of penetrating cardiac injuries in this series, ventricular septal defects developed in 4 patients after stab wounds of the heart; 2 underwent repair of the defect, and the other
ATCARETAL PENETRATING CARDIAC INJURIES
715
2 patients had small ventricular septal defects and were asymptomatic. They continue to be treated conservatively. Mitral regurgtation requiring mitral valve replacement developed in 1 patient, and pericarditis requiring pericardiectomy developed in 2 patients. All patients with residual complications who underwent surgical correction survived. Penetrating cardiac injuries will continue to increase commensurate with increased crime and violence in our cities. Moribund and lifeless patients will be brought to the emergency area. Our experience has demonstrated that aggressive resuscitation, including endotracheal intubation, volume replacement, and chest tube placement followed by resuscitative emergency thoracotomy has resulted in improved survival. Emergency room thoracotomy is indicated in penetrating cardiac injuries in patients who are "lifeless" or in extremis who do not respond to resuscitative measures. Patients with stable vital signs, and hypotensive patients who respond to resuscitative measures, should be transported t o the operating room for definitive repair. This selective approach to penetrating cardiac injuries has salvaged a significant percentage of the patients seen "lifeless" or in extremis and has resulted in excellent survival of the patients in stable condition.
Ref erenees 1. Ivatury RR, Nallathambi MN, Rohman M. Penetrating cardiac trauma: quantifying the severity of anatomic and physiologic injury. Ann Surg 1987;205:61-6. 2. Moore EE, Dunn EL, Moore JB, Thompson JS. Penetrating abdominal trauma index. J Trauma 1981;21:439-44. 3. Ivatury RR, Shah PM, Katsuki I, Ramirez-Schon G, Suarey F, Rohman M. Emergency room thoracotomy for the resuscitation of patients with fatal penetrating injuries of the heart. Ann Thorac Surg 1981;32:377-84. 4. Karrel R, Shaeffer MA, Franaszek JB. Emergency diagnosis, resuscitation, and treatment of acute penetrating cardiac trauma. Ann Emerg Med 1982;11:504-17. 5. Moreno C, Moore EE, Majure JA, et al. Pericardial tamponade: a critical determinant for survival following penetrating cardiac wounds. J Trauma 1986;26:821-5. 6. Tavares S, Hankins JR, Moulton AL, et al. Management of penetrating cardiac injuries: the role of emergency room thoracotomy. Ann Thorac Surg 1983;38:183-7.
DISCUSSION DR MICHAEL D. HOROWITZ (Miami, FL): Approximately a year ago we had a gentleman brought to the emergency room with an anterior thoracic stab wound. He had had a cardiac arrest and required an emergency room thoracotomy. He was found to have cardiac tamponade due to an injury to the right ventricle. His initial pH was 6.8. After repair and closure, the patient had persistent, profound low cardiac output. An emergency echocardiogram showed a ventricular ejection fraction of approximately 0.15. We considered that we might have missed a coronary artery lesion. Cardiac catheterization showed normal coronary arteries. We placed an intraaortic balloon pump, and within the next 12 to 24 hours his clinical picture changed very impressively. In fact,
this man turned around and survived. Late echocardiography showed normal ventricular function. I attribute much of the cardiac dysfunction to his profound shock and severe acidosis. I would like to ask if you have any experience with use of intraaortic balloon counterpulsation in the group of patients with penetrating cardiac injuries.
DR ATTAR: In our group, we had 2 patients who required intraaortic balloon pulsation after the repair of the cardiac injury. We had only 1 survivor of the 2, which is about 50%. I think it is worthwhile using the intraaortic balloon to support the patients; however, the need for it is really rare, because in most of these
716
ATTARETAL PENETRATING CARDIAC INJURIES
patients the injury is localized, especially if it is a stab wound. Stab wounds produce a very small, neat cut in the ventricle in contradistinction to gunshot wounds, which produce substantial tissue injury. We had 1 patient who required reconstruction of the left anterior ventricular wall and the right ventricle as well as a bypass graft to the left anterior descending coronary artery, and he survived. We have been following him up now for 10 years, and he has an ejection fraction of about 0.25.
DR WILLIAM A. COOK (North Andover, MA): Two years ago, the nursing supervisor in a 100-bed hospital where I work asked, “What d o you think would be desirable in a thoracotomy kit?” I asked, “What for?” She replied, “The emergency room physicians want to be prepared to d o thoracotomies.” Here is a group of physicians proposing to do thoracotomies who do not know enough about it to select the instruments. I would just like to emphasize that there is a great deal of difference between the average community hospital emergency ward and a class I trauma center that probably has an emergency room operating suite with lights, instruments, and all of the other things needed to provide standard thoracotomy in the emergency room setting and probably a full-time thoracic surgical resident. We must not be confused regarding emergency room physicians, now being trained to think about thoracotomy in their training programs, as to what they are capable of doing and equipped to d o in an emergency room thoracotomy, especially in the smaller hospital with the occasional patient and no thoracic resident. In the group of patients we looked at in Dallas, the single most important factor in survival was routine transfer of patients suspected of having a cardiac injury to the operating room, bypassing emergency room stabilization. Even though the patient was simply observed in the operating room, and perhaps never even operated on, if he or she did need to be operated on everything was available. I think this is very important. I would like to be sure that thoracic surgeons are pointing out this aspect of cardiac injury care so the patient does not get lost in the shuffle with emergency room personnel. DR ATTAR: Dr Cook has raised two issues. The first one is, who should d o emergency thoracotomy-the thoracic surgeon, the emergency room surgeon, or the general surgeon? In general, I
Ann Thorac Surg 1991;51:7114
agree that not every community hospital should have a thoracotomy kit in the emergency room. At the same time, we should not deny the fact that patients with cardiac wounds are brought to the emergency room. The staff of the ambulance service does not know whether the patient has a heart wound or a chest wound that requires operating or if it may be treated conservatively, so they bring the patient to the emergency area. The first person who sees the patient is either the emergency room surgeon or the general surgeon. In our hospital, it is the general surgical residents who attend to the initial care of the patient. But as soon as the patient hits the emergency room with a presumptive diagnosis of cardiac injury, the thoracic resident, who is on call 24 hours at our university hospital, is called on to supervise the treatment. However, the initial emergency thoracotomy is performed by the general surgical resident in certain conditions. That is, if the patient comes in with cardiac arrest or is under cardiopulmonary resuscitation, or if the patient goes into cardiac arrest, or if there is massive bleeding and deterioration in the emergency room, then the general surgical resident will undertake the emergency thoracotomy. If the patient is in relatively stable condition, having a blood pressure of 80 mm Hg or greater, then he or she can afford to have the luxury of going to the operating room. There has been a study reported in the literature in which there was a comparison of cardiac wound patients who were operated on in the emergency room versus those who were taken right away to the operating room with a delay of 7 minutes. This 7 minutes was very significant as far as survival or death in that series. Therefore, I think we should have an agreed-upon plan between the emergency room staff, the general surgical staff, and the thoracic surgical staff so that when a cardiac emergency is seen in the emergency room, it can be handled properly. In our hospital emergency room, a specific area is designated for emergency thoracotomies and is fully equipped for this procedure. The second question that Dr Cook raised was whether or not the patients should be taken directly to the operating room. Not every hospital has the luxury of having an operating room ready and available. Sometimes the operating rooms are all full, and you cannot wait until a room opens up. So, you have to take care of the emergency and at least open the chest, control the bleeding, and then take the patient to the operating room when he or she is stabilized for definitive repair.
