Abdominal complications following cardiothoracic surgery are infrequent (0.3–5.5%) but carry a significant mortality rate of around 30% (14–63%). This figure is far greater than is seen with the same pathologies in the general population, suggesting the physiological changes imparted by cardiothoracic surgery have a major and deleterious result in the body’s ability to respond to abdominal pathology. Many of the abdominal complications discussed, such as mesenteric ischaemia, can have subtle initial presentations where early intervention can have the most benefit to prevent abdominal catastrophe. These findings suggest there should be a high index of suspicion for abdominal pathology in the postoperative cardiothoracic patient who is failing to progress in the intensive care environment.
Large retrospective and prospective studies have identified several risk factors for the development of abdominal pathology after cardiothoracic surgery:
Old age (>70 years)
Low cardiac output (secondary to AF or congestive cardiac failure)
Peripheral vascular disease
Reoperation secondary to haemorrhage
Renal failure (chronic and acute)
Prolonged cardiopulmonary bypass (CPB) time (>150 minutes)
Use of an intra-aortic balloon pump
Preoperative inotropic support
Chronic obstructive pulmonary disease
Deep sternal infection
Although extended CPB times are associated with elevated risk, the issue of whether off-pump CABG (OPCABG) is protective has been extensively debated in the literature. A study reported in 2003 by Raja et al. suggested that OPCABG decreased the incidence of gastrointestinal complications from 7.3% to 0.6%. These results, however, have failed to be replicated by other groups; indeed it has further been shown that during OPCABG when accessing difficult to reach coronaries, the superior mesenteric artery blood supply is disturbed. General consensus currently is that OPCABG is not protective but the question remains to be definitively answered.
Although not designed specifically for this purpose, EuroSCORE has been used to predict the development of gastrointestinal complications. An alternative scoring model termed the gastrointestinal complication score (GICS) has been developed with the aim to more accurately predict the development of these pathologies. Using prospective collection of data from over 5500 patients, the authors identify that those with a GICS of ≥15 have a probability of developing GI complications of >20% whereas a score of ≤5 has a probability of <0.4%. The authors identify several independent predictors of these complications, which are weighted accordingly (Table 34.1).
|Postoperative vascular complication||9.5|
|Preoperative inotropic support||4.0|
|Postoperative heart failure||3.5|
|Reoperation for haemorrhage||3.5|
|Age >80 years||2.5|
|Cardiopulmonary bypass >150 minutes||2.5|
|Postoperative atrial fibrillation||2.5|
|NYHA class III–IV||2.0|
The GICS has been tested on a validation set of over 1000 patients and is reported to have a sensitivity of 56% and a specificity of 88% for a GICS of 5; and a sensitivity of 13% with a specificity of 99.8% when the score is ≥15. GICS had a larger area under the curve than EuroSCORE when receiver operating characteristic (ROC) curve analysis was performed, although of note this only reached statistical significance in the developmental data set. Overall, GICS appears to provide a useful adjunctive tool to help identify patients at risk of postoperative gastrointestinal complications. Although GICS is useful in predicting which patients may develop abdominal complications, the nature of the pathologies is such that prophylactic measures introduced in those with high scores are unlikely to be of benefit.
Incidence of Abdominal Complications
Abdominal complications after cardiac surgery are not common, although the associated risk of mortality is great (Table 34.2). In addition to the postoperative pathologies described below, it is important not to forget that common causes of the acute abdomen in the general population can also, coincidentally, occur in the patient after cardiac surgery. These can include adhesive small bowel obstruction, incarcerated/strangulated herniae, appendicitis or pyelonephritis to name but a few.
|Gastrointestinal bleed (upper GI (90%) lower GI (10%))||0.39||19|
|Perforated peptic ulcer||0.07||36|
|Pseudo-obstruction||0.06||50 (with perforation)|
Diagnosis and Management of the Common Intra-abdominal Complications
The vast majority of postoperative gastrointestinal bleeds are from the upper gut. Patients in the cardiothoracic ICU are more susceptible due to the common use of antiplatelet agents and also instrumentation of the upper GI tract with transoesphageal echo (TOE). The patient with an upper GI bleed may present with frank haematemesis, cardiogenic shock or melaena. It is important not to forget that a significant upper GI bleed can present solely with fresh blood per rectum, thereby mimicking a lower GI bleed. Upper GI bleeds can cause elevations in serum urea levels as well as the more characteristic drop in haemoglobin. Abdominal signs are often lacking and tenderness is normally absent. After resuscitation and correction of any clotting abnormalities, the first line investigation in any significant GI bleed is an upper GI endoscopy to exclude a bleeding peptic ulcer or alternative upper GI pathologies such as oesophageal varices. Should a peptic ulcer be identified, dual-modality therapy is recommended with the options including adrenalin injection, clips or diathermy. High dose proton pump inhibitor therapy should be commenced and Helicobacter pylori eradication therapy started if biopsy (CLO test) results confirm infection. Should the patient suffer a rebleed after primary endoscopy, the options are either repeat endoscopy, interventional radiology (IR) with possible embolisation of the bleeding vessel (usually the gastroduodenal artery) or if these fail, surgery. Surgery involves an upper midline laparotomy and under-running of the bleeding vessel following enterotomy. Alternative diagnoses such as bleeding oesophageal varices can be managed endoscopically with the application of ligation rubber bands in conjunction with medical control of portal venous hypertension.
Significant lower GI bleeds which fail to stop early with resuscitation and correction of clotting abnormalities can be extremely difficult to manage.The use of adjunctive medical therapies can be helpful in these situations, in particular intravenous tranexamic acid can, on occasion, slow down or even stop significant colonic bleeds. It is imperative in situations of massive rectal bleeds to first exclude an upper GI bleed by endoscopy as well as bleeding haemorrhoids by using bedside proctoscopy. Should these tests fail to identify a source for the bleed and if the patient is generally haemodynamically stable, then a CT angiogram should be obtained. In order for this investigation to successfully identify a bleeding point, the patient needs to be bleeding at a rate of >0.5 ml/minute. However, if a location is identified this facilitates the possibility of an IR approach using conventional angiography in conjunction with embolisation. In the absence of IR, CT angiogram can still be of help as it can direct targeted therapeutic colonoscopy or surgical resection. If the bleeding patient is more unstable, or if CT angiogram fails to identify a bleeding location, the next option is therapeutic colonoscopy which can be performed in the operating theatre. Careful wash-out and mucosal inspection can permit diagnosis and therapeutic intervention, with the use of clips and/or adrenaline injection to achieve haemostasis. In the case of a patient continuing to bleed profusely and no active bleeding site identified, the endoscopy can be augmented by a laparotomy to aid wash-out of the colon through the appendix, and should the colon be cleared as a source of bleeding, sequential enterotomy and small bowel endoscopy to identify the site can then be undertaken. However, undertaking a laparotomy when the site of bleeding is not known in advance should be avoided whenever possible.
Postoperative mesenteric ischaemia carries a very high mortality rate. Two forms of ischaemia are described: occlusive and non-occlusive. Occlusive disease occurs as a result of migration of emboli or the development of de novo thrombus. Non-occlusive disease is more common in this patient group and develops secondary to hypoperfusion from low cardiac output or the use of vasoconstrictors. Occlusive ischaemia tends to follow a very rapid course with the development of acidosis and ultimately intestinal infarction. Non-occlusive disease generally follows a slower clinical trajectory and in the intubated and sedated patient can be problematic to diagnose. In the awake patient, pain, nausea/vomiting and bloody diarrhoea can be present with the pain classically being ‘out of proportion’ to the abdominal signs, which commonly only occur late. However, in the sedated patient the suggestion of non-occlusive ischaemia can be subtle, varying from increasing nasogastric outputs, diarrhoea ± blood or a slowly rising unexplained metabolic acidosis in combination with raised serum lactate. In the stable patient, the investigation of choice if mesenteric ischaemia is suspected is a contrast enhanced CT scan. Endoscopy (both upper and lower) can also be performed if CT is equivocal. Ultimately, however, laparotomy is not uncommonly used as a diagnostic tool in sick patients, and should be considered in a patient where there is a high risk of clinical suspicion for mesenteric ischaemia even if the CT scan is non-diagnostic.
In non-occlusive mesenteric ischaemia, priority should be given to resuscitation and improvement of cardiac output once the diagnosis is made or suspected. If the clinical picture progresses and the patient deteriorates then laparotomy may be the only option for survival.
In both occlusive and non-occlusive mesenteric ischaemia it is not uncommon to find the entire intestine may be non-viable. Depending on the amount of intestine and other associated viscera that have infarcted, the surgical options are either resection with stoma formation accepting the high likelihood of lifelong parenteral nutrition dependency or ‘open-and-shut’ where the degree of ischaemia is extensive and not compatible with ongoing life.
As with all the other pathologies here described, the mortality rate from pancreatitis in this patient cohort (20%) exceeds that of pancreatitis in the general population (5%). However, the management of the condition follows the same principles. As with mesenteric ischaemia, the diagnosis of pancreatitis can be difficult as the symptoms and signs can be impossible to elicit in the sedated patient. The aetiology of pancreatitis here is more commonly secondary to reduced perfusion although the possibility of ‘traditional’ gallstone pancreatitis should not be dismissed.
An elevated serum amylase level is reportedly present in 30–40% of all patients following cardiac surgery, although only 1–3% of these will have pancreatitis. Elevated serum lipase levels can be used to aid the diagnosis but do not appear to confer any additional sensitivity over and above amylase per se. It has been suggested, however, that the combination of an elevated amylase and lipase level can imply subclinical pancreatitis. Contrast enhanced CT scanning can be used to confirm the diagnosis, and the management for the first couple of weeks is invariably organ supportive. Parenteral or nasojejunal nutrition should be commenced if nasogastric (NG) aspirates are high, although NG feeding itself is not contraindicated in this group. There is no evidence for the use of prophylactic antibiotics and these should be reserved for cases of infected necrosis where sensitivities can be obtained by radiologically guided percutaneous aspiration. Rarely, in cases of advanced necrotising pancreatitis, surgical necrosectomy is required which should ideally be performed via a minimally invasive approach, or via an open transabdominal or retroperitoneal approach if not. Attention should be drawn during the course of pancreatitis to development of acute haemodynamic instability as this can suggest a ruptured splenic artery aneurysm secondary to the local inflammatory process. A specialist HPB opinion should always be sought in cases of acute pancreatitis in this patient group.
In the general population acalculous cholecystitis is relatively rare, presenting in the sick and diabetics. In patients following cardiac surgery it occurs at least as frequently as calculous cholecystitis. The aetiology of acalculous cholecystitis in this group is not entirely clear but associations have been drawn with reduced gallbladder contraction and ischaemia. The disease is suggested by intra-abdominal sepsis in the presence of abnormal liver function tests as well as right upper quadrant pain in the awake patient. Diagnosis is confirmed by ultrasound or CT scanning. Aggressive and early treatment is essential to prevent deterioration and should consist of resuscitation and broad spectrum antibiotics. The risk of gangrene and subsequent perforation is far higher in patients with acalculous compared to classical calculous cholecystitis and for those that fail to improve rapidly early surgical intervention is recommended, with cholecystectomy for those fit enough and percutaneous cholecystostomy for those more acutely unwell.
Pseudo-obstruction or Ogilvie’s syndrome is not common after cardiac surgery but is an important condition to diagnose as failure to treat this appropriately can lead to haemodynamic compromise due to an abdominal compartment syndrome and colonic perforation, which carries a 50% mortality rate. It is defined as massive acute colonic dilatation as a result of an imbalance in the autonomic supply to the colon rather than a true mechanical blockage. The diagnosis is suggested upon the presence of increasing abdominal girth with a tense, tympanic abdomen together with the inability to pass faeces or flatus. Diagnostic confirmation can be made with an abdominal radiograph; however, a CT scan provides greater information and rules out the small possibility of a true mechanical blockage being present (e.g. sigmoid or caecal volvulus). Pseudo-obstruction is more likely to occur in patients with electrolyte imbalances, in particular potassium and magnesium. Indeed sometimes the condition can be adequately treated with restoration of homeostatic levels of these cations alone. Should colonic diameter exceed 10 cm or if the patient (if awake) complains of right iliac fossa tenderness, then urgent intervention is required as the possibility of imminent perforation exists. Generally, endoscopic decompression with a colonoscope can relieve the impending perforation and on occasion a flatus tube is left in situ to aid the ongoing decompression. In the general population the parasympathomimetic neostigmine has been used in refractory cases of pseudo-obstruction; however, the cardiac risk profile of this drug suggests its use may be limited in patients following cardiac surgery. Surgical intervention is only required in cases of perforation where colonic resection, abdominal wash-out and stoma formation are generally required. A drainage caecostomy tube, which can be placed radiologically, is only used in very rare cases of repeated caecal dilatation in a patient where laparotomy is not possible or safe.