Penetrating injuries to the heart are caused by stabs, gunshot wounds and rarely, accidental impalements [1, 2]. The mechanism of penetrating cardiac injuries varies worldwide [1, 3, 4] . Data from the United States reported 70% gunshot wounds and 30% stab wounds distribution [5, 6]. Intentional gunshot wounds to the heart also contribute to the predominance of gunshot wound distribution [5,6,7,8]. However, in South Africa, the opposite is reported, with 82–94% stab wounds and 6–18% gunshot wounds [3, 9, 10].

Chest injuries are on the rise in South Africa and are often associated with high mortality rates [3, 9,10,11]. The management of penetrating chest injuries has evolved to mostly being non-operative in nature [11]. A subset of penetrating chest injuries presents following gunshot wounds [11, 12]. Gunshot wounds to the chest and abdomen are common in our institution. The management of these patients can be a challenge. Gunshot wounds to the heart are regarded as one of the most lethal penetrating injuries [2]. Survivors of gunshot wounds put an economic burden on the healthcare system [13]. The morbidity affects not only the victims but also the victim’s families [13]. Gunshots that result in homicide or mass shootings get more attention than the survivors that present to the hospital [13, 14].

These gunshot injuries can be perforating or non-perforating [15,16,17]. They can also be full thickness, tangential or result from shockwaves [15, 16]. Bullets do not always follow a straight trajectory to the target [15, 16]. Injuries resulting from shockwave injuries may be challenging to diagnose [16]. Full-thickness injuries are likely to result in multi-chamber injuries [16]. Data on multi-chambered injuries are contradictory and inconsistent [3, 4, 9,10,11,12]. A study by Degiannis showed this was not associated with poor patient outcomes [10].

The literature suggests that patients who have sustained gunshots present in a worse physiological state than those who sustained stab wounds [5, 18]. More than 70% of patients who sustain gunshot wounds present with no signs of circulation or respiration [5, 18]. Furthermore, cardiac tamponade was reported to be less and exsanguination more common in gunshot injuries than in stabs [12, 15, 19]. Gunshot wounds can cause disruptive injuries to the heart, which promote exsanguination and decompression into the chest cavity [5, 18]. When cardiac tamponade is present, it has favourable outcomes [18].

It is unsurprising that gunshot heart injuries are rare and are said to have poor outcomes [3, 4, 9,10,11,12]. However, despite the high mortality rates, with improvements in prehospital care and surgical skills, we see more case reports of patients with penetrating cardiac injuries being salvaged [5, 18, 20, 21]. It was worth looking at our experience in a low-to-middle income country with a high volume of penetrating trauma. Our unit works as a level 1 trauma that caters to all types of trauma, and the procedures are done by trauma surgeons or specialist surgeons in the department. The hospital is a 1100 bedded hospital, seeing 220 priority one cases per month and a total of 1200 total trauma cases per month. There are 18 critical beds and 26 other dedicated beds for visceral trauma (excluding orthopaedics and neurosurgery. Prehospital services are variable, with some patients arriving directly to the facility on private transport and many more arriving with prehospital services that vary from basic to advanced care. This prehospital setup is better than most Sub-Saharan countries. We hope our experience will shed light on managing these highly lethal cases.

Our study aims to describe the outcomes and injuries of patients who sustained penetrating cardiac injuries at a single institution in Johannesburg, South Africa.

Materials and methods

We retrospectively reviewed all patients who presented to Charlotte Maxeke Johannesburg Academic Hospital Trauma Unit with penetrating cardiac injuries from gunshots over a period of 8 years (01 January 2011–31 December 2018). The patients were identified from the hospital trauma database. All patients above the age of 18 with gunshot wounds to the heart who presented with a carotid pulse and a cardiac rhythm were included in the study. Patients who presented pulseless were not included, regardless of presence of other signs of life. Four patients who had significant missing data were excluded from the study.

Blood pressure of less than 90 mmHg was considered haemodynamic instability.

The patient’s demographics, mechanism of injury, injury severity score (ISS), physiological parameters at the scene and on presentation, investigation findings, type of operation, length of hospital stay, morbidities and in-hospital mortalities were recorded. Echocardiography was requested in all survivors post-surgical repair. All surviving candidates were followed up at least twice on an outpatient basis.

Statistical analysis

Means (± SD) are presented for continuous variables, and frequencies (%) are presented for categorical variables. All analyses were done using STATA version 15. Continuous variables were first tested for normality using the Shapiro–Wilk test. Fisher’s exact test was used to test the significance of the relationship between categorical variables. Univariate and multivariate analyses were conducted on the data. A p-value of < 0.05 was considered statistically significant.

Ethics approval was obtained from the University of the Witwatersrand Human Ethics committee and the hospital CEO. Ethics number M180463 was allocated to the study. The study was also registered on Clinical Trial.gov NCT05016986.

Results

After excluding four patients for missing data, 37 patients that sustained penetrating gunshot wounds (GSW) to the heart over eight years were enroled in the study. Thirty five (94.59%) were males and 2 (5.41%) females. The median age was 30 years (IQR 24–36). Sixty-four per cent of patients presented haemodynamically unstable (n = 24), and 13 (35%) patients were haemodynamically normal on presentation to the emergency department (see Table 1). Beck’s triad of hypotension, distended neck veins/increased CVP, and muffled heart sounds were present in about 51% of patients.

Table 1 Demographics and physiological parameters
Full size table

All our patients presented from the scene and were all victims of interpersonal violence. The median systolic blood pressure prehospital was 92 mmHg, with a diastolic median of 58 mmHg. The median pulse rate prehospital was 89 beats per minute. On presentation to the emergency department, the median vital signs were a systolic BP of 69 mmHg, diastolic BP of 55 mmHg and a pulse rate of 92 beats per minute. The physiological parameters in the hospital showed a median acidaemia of 7, 16 ± 0.16 and a hyperlactatemia of 7 mmol/l (IQR 4.7–10.1) with a base excess of − 11, 4 mmol/l ± 8.33. The median RTS was 5, 6 (IQR 3.6–7.6). The median ISS was 32 (IQR 26–41) (see Table 1).

The most common investigations in the emergency department (ED) were chest X-rays (73%) and chest ultrasound (55%). Echocardiograms in ED (9%) and cardiac enzymes (6%) were only done in selected cases. Intravenous lines were established in all cases, with 41% receiving central venous catheters in the ED.

Most patients sustained other associated chest injuries (59.4%) besides cardiac injuries, abdominal injuries (64.9%), head injuries (16.2%) and extremities (16.2%) (see Table 2). All patients had sustained gunshot wounds to the chest, nine cases were purely chest injuries, fourteen had thoraco-abdominal injuries, 12 had separate chest and abdominal gunshot injuries and, two had chest and extremity gunshot wounds. The most common injuries from the chest were parenchymal lung injuries (n = 13). Combined hollow and solid organs (n = 10) were the most common injuries in the abdomen (see Table 2).

Table 2 Associated injuries
Full size table

Ten patients had an emergency room thoracotomy performed as they arrived in extremis. Twenty patients had an anterolateral thoracotomy in theatre. Only five had sternotomy incisions to access the thorax (see Table 3). Two other patients had clamshells for right atrial injuries. The most injured chamber was the right ventricle (n = 28), followed by the left ventricle (n = 4), right atrium (n = 2) and left atrium (n = 1) (see Table 3). There were 2 combined chamber injuries and no injuries to coronary vessels. All cardiac repairs were done with a non-absorbable polypropylene suture depending on the chamber involved. (Prolene 2/0, Prolene 3/0 and Prolene 4/0). Pledgets were utilised in atrial or auricular repairs only. All patients who survived were admitted to ICU. The complications among survivors were local wound sepsis (n = 1) and empyema (n = 1).

Table 3 Type of incision used to access the chest cavity and cardiac site injured
Full size table

The mortality rate was 81.1%, with six patients (16.21%) dying in the emergency department, twelve patients (32.43%) died on the table in the theatre, eleven (29.72%) died shortly after admission to ICU, and one (2.7%) later in the ward. Nine of the ten patients who had emergency room thoracotomy due to extremis died. Six of the ten died in the emergency department and the remaining 3 in theatre. Most of the deaths occurred in theatre. Among the survivors (7/37), 6 were discharged home, and one stepped down to another hospital. Associated chest injury apart from the heart was a predictor of death (p < 0.05) (see Table 4).

Table 4 Gun shot wounds (Survivor–Non-Survivor)
Full size table

Discussion

Gunshot wounds to the heart are often fatal due to the damage they inflict on the heart [1]. These can be challenging cardiac injuries to manage. There has been an increase in gunshot wounds in several countries including South Africa due to increased use of firearms, organised crime and interpersonal violence [1, 15, 22]. Most patients with gunshot wounds to the heart will die on the scene [12]. The ones that present to hospitals may be salvaged even in the absence of cardiothoracic surgery services in specialised centres [10]. Our study was focused on those who presented directly to our institution with signs of life. Our general approach to these injuries is depicted in the flow chart. Those declared dead at the scene are taken directly to the mortuary for forensic investigations.

Trauma patients are usually young males, as shown in our cohort of patients [3, 4, 9,10,11,12] [1,2,3,4,5,6]. The clinical status on presentation of these patients influences the outcomes [18]. The median systolic blood pressure was 69 mmHg; physiologically, these patients had acidaemia, hyperlactatemia and a high base deficit. These parameters reveal very sick patients when they arrive in the emergency department. The median ISS was 32, and RTS 5, 6 (see Table 4) in keeping with the energy mechanism associated with gunshot compared to stabs [23]. Up to 65% of patients with penetrating cardiac injuries present to the emergency department with no recordable blood pressure [5]. Ten patients presented agonal and had an emergency room thoracotomy in keeping with our previously reported experience [24]. Only two had a clamshell thoracotomy as we tend to reserve this approach to more complex injuries compared to other centres [25]. Other studies also showed that clinical status on presentation had a significant bearing on mortality [10, 18].

The most injured chamber of the heart was the right ventricle in 75.7% of our patients. This is in keeping with other previous studies regarding penetrating chest injuries [4,5,6, 8, 9, 12]; the explanation offered was the fact that the right ventricle is the most presenting chamber to the anterior mediastinum [1]. There were 2 patients with combined chamber injuries. This was not associated with increased mortality, as in a previous study [11]. No coronary injuries were reported in these patients who managed to present to the hospital. This may be due to selection bias, as those with complex coronary vessel injuries may have died at the scene [5, 18, 20]. Though sternotomy was utilised in only 13, 5% of the cases, it was associated with survivability (p = 0.003) (see Table 4), in keeping with our surgical practice to do sternotomy in the more stable cases.

Alive patients were admitted to ICU at the end of surgery for further monitoring and care. We conducted a postoperative echocardiogram on all survivors, and no repeat echocardiogram was ever deemed necessary at outpatient follow-up. The overall recorded morbidity rate was 5.4%, with one patient developing local wound sepsis and another developing an empyema which required a relook thoracotomy. Tang et al., in their cohort of penetrating cardiac injuries, had a complication rate of 17% in their survivors [26]. Carr et al. had up to 43% confirmed immediate complications, which settled in time [27]. Our morbidity was, therefore, lower than reported in other studies, probably in keeping with the young age noted in our cohort.

Most of our patients had combined extra-cardiac injuries, with only two having sustained isolated gunshots to the heart. Other studies showed no association between associated injuries and mortality [9, 18]. In our study, other associated chest injuries were a predictor of mortality (see Table 4). This may reflect a high injury burden to the chest cavity, which has close proximity of pivotal structures. Associated abdominal injuries were not statistically associated with mortality (p > 0.05) compared to what we had experienced when looking at outcomes in those requiring Emergency Room thoracotomy, where combined chest and abdominal injuries had poorer outcomes [24].

The overall survival of patients in our series was 18.9% (7/37), in keeping with another reported mortality rate after gunshot wounds ranging from 12 to 81% [1]. This survival rate must consider that 64.86% presented with haemodynamic instability. However, some of these studies included patients with other missiles causing penetrating cardiac injuries [1]. The 90% mortality seen in those who had emergency room thoracotomy is in keeping with our unit’s policy to perform this procedure in those presenting in extremis only [24]. A study on penetrating cardiac injuries in South Africa (stab and gunshot combined) had a mortality rate of 81% in gunshot injuries patients [10], which is similar to our findings. Our series reports on civilian gunshots only. Cardiopulmonary resuscitation before arrival, high Injury Severity Score (ISS), high Revised Trauma Score (RTS), hypotension and greater vessel injuries are known to be associated with poor outcomes in penetrating cardiac injuries [18].

Conclusion

Gunshot wounds to the heart have a high mortality rate. A survival rate of 18.9% is acceptable considering the energy transfers associated with gunshot injuries and the fact that the majority presented with haemodynamic instability. The right ventricular chamber was the most common injury site. The morbidity rate was lower than in other series. We propose a prospective multicentre study be considered in low-to-middle income countries to clarify outcomes and patterns of gunshot injuries to the heart under these circumstances.

Limitations to the study

This study was a retrospective study looking only at gunshot injuries to the heart. Bias might have been introduced by this being a single-centre study. The total study numbers were few due to only a few survivors being able to make it to an institution. This sample size could explain why we could not statistically demonstrate the direct link of BP < 90 to the mortality outcomes. The morbidity reporting was self-reported, limited to short-term follow-ups, and may have missed some minor complications. No long-term follow-up was done in our cohort.