Determinants of Gastrointestinal Complications in Cardiac Surgery

Abstract

We designed this study to define determinants of gastrointestinal complications after cardiac surgery. From January 1992 through December 2000, 11,058 patients underwent cardiac surgery on cardiopulmonary bypass at our institution. Data were prospectively collected and univariate and multivariate analyses conducted.

A total of 147 gastrointestinal complications occurred in 129 patients (129/11,058; 1.2%) including gastroesophagitis (18, 12.2%), upper gastrointestinal hemorrhage (42, 28.6%), perforated peptic ulcer (7, 4.7%), cholecystitis (10, 6.8%), pancreatitis (13, 8.8%), intestinal ischemia (17, 11.5%), colitis (18, 12.2%), diverticulitis (5, 3.4%), intestinal occlusion (2, 1.1%), lower gastrointestinal hemorrhage (1, 0.7%), and mixed gastrointestinal complications (14, 9.5%). Patients with gastrointestinal complications were significantly older and had significantly higher comorbidity (unstable angina, chronic renal failure, and peripheral vascular disease), morbidity (prolonged mechanical ventilation, intra-aortic balloon pumping, bleeding, acute renal failure, stroke, and infection), and mortality rates (22.5% vs 4%, P <0.0001). They also had longer cardiopulmonary bypass times and higher valvular surgery rates. Multivariate analysis identified 6 independent predictors for gastrointestinal complications: prolonged mechanical ventilation (odds ratio [OR], 5.5), postoperative renal failure (OR, 4.2), sepsis (OR, 3.6), valve surgery (OR, 3.2), preoperative chronic renal failure (OR, 2.7), and sternal infection (OR, 2.4).

Factors such as mechanical ventilation, renal failure, and sepsis are the stronger predictors for GI complications, causing splanchnic hypoperfusion, hypomotility, and hypoxia. Furthermore, excessive anticoagulation after valve replacement may lead to GI hemorrhage. Valve surgery, often requiring anticoagulation, increases bleeding. Monitoring mechanical ventilation and hemodynamic parameters, adopting early extubation and mobilization measures, preventing infections, and strictly monitoring renal function and anticoagulation may prevent catastrophic abdominal complications. (Tex Heart Inst J 2003;30:280–5)

Gastrointestinal (GI) complications after cardiac surgery with cardiopulmonary bypass (CPB) are rare, but they entail significant morbidity and mortality rates. Despite improvements in patient care, such complications remain a risk that has been recognized since the early days of cardiac surgery. ¹

Previous analyses have identified a plethora of preoperative and perioperative variables; nonetheless, the determinants of GI complications are still controversial. ^2–11 The variability of findings reported in the existing medical literature may be the result of applying only univariate statistical analyses to a limited number of patients with GI complications. In contrast, we have focused our research on a large number of cardiac surgery patients in order to identify, by use of multivariate analysis, those factors that would significantly increase the risk of GI complications.

Patients and Methods

From January 1992 through December 2000, 11,058 adult patients underwent cardiac surgery with CPB at our institution.

Surgical Technique. All patients underwent either coronary artery bypass grafting (CABG), valve surgery, combined CABG and valve surgery, aortic surgery, or surgical correction of adult congenital heart defects. Cardiopulmonary bypass was initiated after ascending aorta-to-right atrial or bicaval cannulation. Body temperature was maintained in the 28 to 32 °C range, and myocardial protection was achieved with cold, intermittent, antegrade, and retrograde blood cardioplegia in most patients.

Data Collection and Statistical Analysis. Preoperative, intraoperative, and postoperative data were prospectively collected by trained personnel and stored in the institution's surgical database. Univariate analysis was conducted using either the χ² test for categorical variables or the Student's t-test for continuous variables. Two-tailed P values of <0.05 were considered statistically significant. Multivariate analysis was conducted using stepwise logistic regression (SAS Institute Inc.; Cary, NC).

Results

A total of 147 GI complications occurred in 129 patients (129/11,058, 1.2%). Demographic and comorbidity data were compared between the groups with and without GI complications (Table I).

TABLE I. Demographic Data, Comorbidities, and Presence or Absence of GI Complications

Average patient age; presence of chronic renal failure (CRF), unstable angina, and peripheral vascular disease; reoperation rates; and Parsonnet scores were all significantly higher in the group with GI complications (Table I).

Sex, diabetes mellitus, low left ventricular ejection fraction, preoperative acute myocardial infarction (AMI), and preoperative use of intra-aortic balloon pumping (IABP) were all equally distributed in the 2 groups (Table I).

Intra- and postoperative data were also collected and compared (Table II). Valve surgery, combined CABG and valve surgery, reoperation for bleeding, perioperative AMI, postoperative IABP, acute renal failure (ARF) (creatinine >150 μg/dL), prolonged mechanical ventilation (>24 hours), deep sternal wound infection, sepsis, and acute cerebrovascular accidents were all significantly more frequent in the GI complications group (Table II). Average CPB and cross-clamp times were also significantly higher in the group with GI complications (Table II). The mortality rates were 22.5% and 4%, respectively, in the groups with and without GI complications (P <0.0001).

TABLE II. Intra- and Postoperative Variables, Including Morbidity and Mortality

Among the 129 patients with GI complications, upper GI bleeding was the most common event (42/147, 28.6%), followed by gastroesophagitis (18, 12.2%), colitis (18, 12.2%), intestinal ischemia (17, 11.5%), mixed GI complications (14, 9.5%), pancreatitis (13, 8.8%), and cholecystitis (10, 6.8%). Other complications are also listed inTable III. Endoscopic cauterization was required in 15 patients with upper GI bleeding. Abdominal surgery was performed in 18 patients: 15 with intestinal ischemia and 3 with cholecystitis. Intestinal ischemia carried the highest mortality rate (11/17, 64.7%) (Fig. 1).

TABLE III. Distribution of 147 GI Complications in 129 Cardiac Surgery Patients

Fig. 1 Mortality rate by gastrointestinal (GI) complication.

Multivariate analysis was conducted on the preoperative, intraoperative, and postoperative variables that had yielded significantly different results by univariate analysis.

Six independent predictors of GI complications were identified, including prolonged mechanical ventilation (odds ratio [OR], 5.5), postoperative ARF (OR, 4.2), sepsis (OR, 3.6), valve surgery (OR, 3.2), preoperative CRF (OR, 2.7), and deep sternal wound infection (OR, 2.4) (Table IV).

TABLE IV. Determinants for GI Complications in 11,058 Patients Undergoing Cardiac Surgery

Discussion

Intra-abdominal complications after cardiac surgery with patients on CPB are rare, with an incidence ranging from 0.3% to 2%, but the mortality rate varies from 11% to 59%. ^2–11 Our present series had a 1.2% incidence of GI complications and a 22.5% mortality rate, both of which are consistent with existing reports.

Interestingly, in a separate unpublished analysis, we noted that the GI complications rate had not changed through the years despite improvements in perioperative care, monitoring, anesthesia, and operative technique. These improvements have been offset by an actual change in the cardiac surgery referral pattern, which now includes older and sicker surgical candidates. On the other hand, independent determinants for GI complications historically have not been adequately studied; for this reason, accurate means to identify patients at risk for these deleterious occurrences have yet to be established.

Careful analysis of the existing world medical literature on this topic revealed only 4 investigations that have been conducted using multivariate analysis to identify the independent determinants of GI complications in cardiac surgery. ^2–5

Although the pathogenesis of GI lesions is complex and multifactorial, the major factor implicated in cardiac surgery is reduced systemic blood flow, which leads to inappropriate oxygen delivery and energy deficit. The GI organs are at particular risk for ischemic events for several reasons, and oxygen shunting and consequent distal hypoxia can occur at the tip of the intestinal villi even under normal conditions. ¹² The GI tract does not have the ability to autoregulate to compensate for reductions in blood pressure. Furthermore, due to persistent vasoconstriction, splanchnic hypoperfusion may continue even after hemodynamic stability has been regained.

Preoperative, intraoperative, and postoperative variables may all influence abdominal perfusion. Comorbid conditions, such as low left ventricular ejection fraction ^2–4 and peripheral vascular disease, ² may all cause splanchnic hypoperfusion and have actually been identified as determinants of GI complications in patients undergoing cardiac surgery. In our study, only peripheral vascular disease was more common in the group with GI complications, although statistical significance was achieved solely by univariate analysis.

During the intraoperative phases, hypovolemia, ³ prolonged CPB, ^2,4,5 and administration of vasocon strictors can cause GI hypoperfusion. Cardiopulmonary bypass is associated with a broad range of systemic complications, including nonpulsatile flow, hemolysis, activation of the inflammatory cascade, anticoagulation, hypothermia, and, finally, reduced end-organ perfusion. Furthermore, CPB can increase GI permeability and, as a consequence, enhance the release of cytokines that will lead to mucosal damage and microcirculation problems. ¹³

Findings regarding the relationship between CPB duration and GI complications are not uniform. Zacharias's ² and Perugini's ⁴ groups have found a strong relationship between these 2 variables, with ORs ranging from 1.3 to 1.7. In contrast, Spotnitz and colleagues ⁵ and Christenson's ³ group did not find any significant relationship upon multivariate analysis.

In our analysis, although CPB time and cross-clamp time were significantly higher in the GI complications group, they did not reach statistical significance after stepwise logistic regression. This finding confirms that perioperative factors other than CPB should be kept under consideration as determinants of general surgery complications.

In this regard, prolonged mechanical ventilation plays an important role in modifying the splanchnic blood flow and thus causing temporary GI ischemia. Splanchnic hypoperfusion during prolonged mechanical ventilation may arise from decreased mean arterial pressure and increased GI vascular resistance. Prolonged mechanical ventilation with high positive end-expiratory pressure (PEEP) can result in de creased cardiac output and hypotension; splanchnic blood flow in these settings decreases in parallel with PEEP-induced reductions in cardiac output. ¹⁴ Furthermore, high PEEP is also associated with increased renin-angiotensin-aldosterone activity and elevated catecholamine levels. ¹⁵ All these neurohormonal abnormalities lead to splanchnic vasoconstriction and redistribution of blood volume. The resulting imbalance between oxygen supply and demand causes mucosal damage (ulceration) and altered GI motility (ileus). Moreover, after ischemia, the reperfusion damage that can ensue with normalization of cardiac output and peripheral vascular resistance may be responsible for acute nonocclusive mesenteric ischemia. ¹⁶ Spotnitz and coworkers ⁵ first reported the importance of prolonged mechanical ventilation as an independent determinant for GI complications after cardiac surgery, with an OR of 6.6 after nontruncated multivariate analysis. Other authors ^6,7 reached the same conclusions with univariate analysis.

In our experience, prolonged mechanical ventilation (>24 hours) was the strongest determinant for abdominal complications, with an OR of 5.6. Perioperative factors such as the use of IABP and the development of ARF are good indicators of a low output state and may, directly or indirectly, be related to GI complications after cardiac surgery. Intra-aortic balloon pumping is generally used in patients with ongoing cardiac ischemia or cardiac failure that is unresponsive to medical treatment. These patients are already predisposed to GI hypoperfusion secondary to decreased cardiac output; therefore, the critical need for IABP, rather than the IABP itself, predisposes to GI complications. On the other hand, IABP itself may encourage thrombus formation, embolization, and platelet destruction and thus potentially contribute to the GI insult. The use of intra-aortic balloon pump ing has been identified as a determining factor for abdominal complications in both multivariate ⁵ and univariate ^6,8,9 studies. In our analysis, although postoperative IABP use was significantly more common in the GI complications group (Table II), its impact after logistic regression analysis was not significant.

Acute renal failure after cardiac operations is associated with high morbidity and mortality rates and results from generalized organ hypoperfusion during bypass and in the postoperative phases. Animal studies have shown that moderate renal failure alters duodenojejunal motility, decreases colonic transit time, and may contribute to GI injury. ¹⁷ The relationship between CRF, ARF, and GI complications after cardiac surgery has been emphasized in a limited number of univariate analysis studies. ^6,18 In our experience, both ARF and CRF are independent determinants for GI complications, with ORs of 4.2 and 2.7, respectively.

To our knowledge, the relationship between sepsis and GI complications after cardiac surgery has not previously been investigated. A septic state leads to endotoxemia, release of cytokines, and activation of the inflammatory response. All of these conditions increase gut permeability, impair its microcirculation, and damage its mucosal layers. Furthermore, the progression of a septic state may lead to systemic hypo-perfusion and, finally, to multiple-organ failure. In our analysis, both sepsis and deep sternal wound infection carried a strong relationship with GI complications, with ORs of 3.6 and 2.4, respectively.

Among the different perioperative variables that may be related to abdominal complications, valve surgery has been reported frequently in previous univariate analyses. ^7–11 Patients after valve surgery may be at higher risk for GI bleeding because of anticoagulant therapy, and upper GI bleeding remains the most common abdominal complication after cardiac surgery with CPB. Heikkinen and Ala-Kulju ¹⁹ found that 24% of GI bleeding after cardiac surgery occurred after excessive anticoagulation. As reported in our results, valve surgery was significantly related to the occurrence of GI complications, with an OR of 3.2.

Although we have focused our attention on those variables that had significance after logistic regression, some other preoperative and postoperative elements may be important in modifying the GI complications rate. In this regard, advanced age, female sex, pre-operative IABP, emergent operation, and blood transfusions have all been shown to be more frequent in patients that developed GI complications. ^2–11 Our findings confirm that, among these variables, only a higher average age was noticed, after univariate analyses, in the group with GI complications.

Conclusions

We have reported herein our experience with GI complications after cardiac surgery with CPB. To our knowledge, this is the largest experience described in the medical literature. After logistic regression analysis, 6 major determinants for abdominal complications were identified: prolonged mechanical ventilation, ARF, sepsis, valve surgery, CRF, and sternal wound infection. Our findings are partially consistent with the existing literature and confirm that variables such as CPB time, ischemic time, and IABP are marginal factors in the development of GI complications. It is to be hoped that our findings will prompt a higher level of vigilance toward selected high-risk patients and lead to rapid diagnosis and treatment of GI complications. Correct identification of the independent determinants for GI complications should help us in designing a more targeted prophylactic, diagnostic, and therapeutic strategy. In this regard, optimization of the hemodynamic status seems to play a focal role in the various perioperative phases. Furthermore, attentive monitoring of mechanical ventilation and the adoption of early extubation and mobilization protocols might also prevent catastrophic abdominal complications. During the early postoperative stages, prevention of infections and stimulation of renal function should be emphasized to maintain physiologic GI functions. Moreover, strict monitoring of anticoagulation levels may avert GI bleeding in patients undergoing valve surgery.

Acknowledgments

We would like to thank Brigitte Dionne, RN, and Sege Simard, PhD, for their assistance in collecting and analyzing the data for this study.