Ann Thorac Surg 2006;82:1796-1801
© 2006 The Society of Thoracic Surgeons
Original Articles: Cardiovascular
Abdominal Complications After Heart Surgery
Jamal H. Khan, MD*,
April M. Lambert, DO,
Joseph H. Habib, MD,
Mike Broce, BA,
Mary S. Emmett, PhD,
Elaine A. Davis, EdD
Charleston Area Medical Center and West Virginia University School of Medicine, Charleston Division, Division of Cardiothoracic Surgery and CAMC Health Education and Research Institute, Charleston, West Virginia
Accepted for publication May 26, 2006.
* Address correspondence to Dr Khan, 1 Portview Drive, Charleston, WV 25311 (Email: jamal.khan{at}camc.org).
 |
Abstract
|
|---|
BACKGROUND: Up to 3% of patients undergoing heart surgery suffer from an intraabdominal complication (IAC). These complications carry a high mortality besides adding to the morbidity and cost. This review was undertaken to see if a subset of patients with increased risk of IAC could be identified.
METHODS: Medical records of 7,731 consecutive patients undergoing heart surgery in a single center were screened for identification of postoperative IAC. One hundred and twenty (120) cases were found. One hundred and six (106) cases were compared with the same number of matched controls.
RESULTS: Significant predictors of the development of IAC were increased cardiopulmonary bypass times (> 99 minutes), peripheral vascular disease, chronic steroid use, and low left ventricular ejection fraction. Patients on postoperative antiplatelet therapy or warfarin had a lower risk of IAC. Significant predictors of mortality in IAC were increased cardiopulmonary bypass times (
120 minutes.), use of inotropes, cerebral vascular disease, and incremental age.
CONCLUSIONS: A subset of patients can be identified who are at higher risk for IAC and an associated adverse outcome. Patients who have prolonged cardiopulmonary bypass, have a low left ventricular ejection fraction, are on steroids, and suffer from other vascular disease should be observed carefully for development of IAC. Postoperative anticlotting strategies may be helpful. Early diagnosis and intervention are essential for improving outcomes in cases of IAC.
|
Introduction
|
|---|
Intraabdominal complications (IAC) affect 0.5% to 3.0% of all patients undergoing cardiopulmonary bypass surgery [1, 2]. They are a significant source of morbidity and mortality. Mortality rates of 14.5% to 100% [2–4] have been reported from various centers for these complications after open-heart surgery. Intraabdominal complications manifest themselves in various forms but visceral hypoperfusion during the perioperative period is the common factor in most instances [5–7]. Of particular concern is ischemic damage to the gut, which carries a very high mortality [8–10]. Because timing of the first manifestation of these complications is unpredictable, close monitoring of the patients is needed. Early recognition and intervention are imperative for improving outcomes. Indeed, many authors emphasize the importance of this strategy [1, 7, 11, 12]. Although invasive instrumentation has been tried for studying hypoperfusion of gastric mucosa [2, 6, 13], for practical purposes a high index of suspicion and special attention to patients with predisposing factors are needed to make an early diagnosis.
Gastrointestinal (GI) bleeding, followed by acute pancreatitis, acalculous cholecystitis, peptic disease without bleeding or perforation, perforated ulcer, bowel infarction, diverticulitis, and liver dysfunction are all manifestations of hypoperfusion and response to the stress of surgery. Bowel infarction is the most lethal of all intraabdominal complications. Relative visceral ischemia may result from prolonged cardiopulmonary bypass, vasoconstriction, low cardiac output, maldistribution of perfusion, obstructive splanchnic disease and embolism [7, 11, 14]. Risk factors for intraabdominal complications parallel risk factors for open-heart surgery in general. Advanced age, female gender, renal insufficiency, redo operations, and chronic obstructive pulmonary disease (COPD) have all been implicated as significant patient- related risk factors [9, 15, 16]. Other factors mentioned by various authors include use of inotropes, use of intraaortic balloon pump (IABP), atrial fibrillation, preexisting gastrointestinal disease, congestive heart failure (CHF), peripheral vascular disease (PVD), and cerebral vascular disease (CVD). Surgical factors identified in the literature that add to the risk include emergent operations, redo operations, combined procedures, prolonged cardiopulmonary bypass and prolonged aortic cross-clamp times [9, 11, 12, 17]. Postoperative renal failure, bleeding, atrial arrhythmias, low cardiac output, and low left ventricular ejection fractions (LVEF) have also been implicated.
In general, the risk factors emphasized in the literature for IAC are not specific to this particular complication. The current study was undertaken to examine if a predictive model, specific to IAC, could be defined and what lessons could be learned about the management of IAC.
|
Patients and Methods
|
|---|
The study was conducted with the approval of the Institutional Review Board. The requirement for consent was waived as the data were collected retrospectively from existing data and met the requirement for minimal risk. Medical records of 7,731 consecutive patients who underwent cardiac surgery at a single center between January 1, 1997 and May 25, 2001 were reviewed retrospectively. One hundred and twenty patients (1.5%) with intraabdominal complications were identified. These were patients who required medical or surgical intervention for intraabdominal pathology during the postoperative period. Complications encountered included acute GI bleeding, peptic disease without bleeding or perforation, peptic disease with perforation, mesenteric ischemia confirmed by exploration, pancreatitis, cholecystitis, and bowel obstruction. Treatment was classified into medical-observation, abdominal exploration without resection and exploration with bowel resection. During the same study period data were collected on a randomly selected sample of 484 patients who did not experience an IAC for the purpose of propensity matching.
Information collected on all patients included age, gender, date of admission and discharge, date and type of procedure, date of diagnosis of the complication, and outcomes at discharge. Original operations included coronary artery bypass graft (CABG), valve repair or replacement, CABG combined with valve surgery, or other concomitant procedures. Use of cardiopulmonary bypass (CPB) with bypass times, cross-clamp times, redo operation, and use of intraaortic balloon pump were documented. Status at surgery (elective, urgent, emergent, or salvage) according to the Society of Thoracic Surgeons definitions was noted [18].
Comorbid conditions collected included tobacco use, diabetes mellitus (DM), PVD, CVD, COPD, and LVEF by visual or echocardiographic evaluation (Table 1). Postoperative use of heparin, warfarin, inotropes, and antiplatelet therapy was recorded. Also noted were incidents of supraventricular arrhythmias or heparin-induced thrombocytopenia (HIT).
Data Analysis
All analyses were performed using SPSS version 10.1 (SPSS, Chicago, IL). Descriptive statistics are expressed in terms of frequencies, percentages, or means. Categoric variables were tested by the 
2 or Fisher exact test and continuous variables were tested by the Student t test. A p value of 0.05 or less was considered significant.
Multivariate logistic regression was used to identify independent predictors of complications and to calculate propensity scores [19]. The adjusted model included age, gender, comorbidities, and all other patient-related variables (see Table 1 for a complete list). The propensity score (or probability of developing a complication) [20] was used to match each complication case with a no-complication control (1-to-1 match). A match was made when a case and a control had the same five digit propensity score. If not, 4, 3, 2, or 1 digits were used for matching. To determine a successful match, the two new groups (complication cases versus matched no-complication controls) were tested for possible differences. Next, the mortality rates for the two matched groups were compared. And finally, two multivariate logistic regression models were used to determine the independent predictors of mortality: (1) for the original complication cases, and (2) for the matched set of complication cases and controls.
|
Results
|
|---|
Before Matching
Of the 120 patients with IAC, 112 complete records were selected for analysis. Eight charts were excluded because of incomplete documentation. One hundred and twelve (112) patients had a total of 119 IACs. Acute GI bleeding was the most common representing 39.3% (n = 44) of the GI complications. Those patients had a mortality rate of 29.5%. Peptic ulcer disease without bleeding or perforation was next with an incidence of 32.1% (n = 36) and a mortality rate of 8.3%. Eighteen percent (n = 20) of the complications consisted of ischemic bowel, confirmed by exploration. This carried the highest mortality of 55%. Six percent (6.3%, n = 7) were perforations secondary to peptic ulcer disease, carrying a mortality of 43%. Over eight percent (8.9%, n = 10) were diagnosed with pancreatitis based on serum amylase or lipase, and had a mortality rate of 40%. Only 2 patients were diagnosed with acute cholecystitis and 2 with bowel obstruction.
Independent predictors of intraabdominal complications, identified from the multiple logistic regression model of the unmatched sample, included the following: (1) CPB time 100 to 149 minutes (odds ratio [OR]: 1.98; confidence interval [CI]: 1.12 to 3.51; p= 0.018); (2) CPB time 150+ minutes (OR: 1.97; CI: 1.07 to 3.63; p= 0.029); (3) presence of PVD (OR: 1.81; CI: 1.13 to 2.89; p =0.013); (4) an EF of less than 0.50 (OR: 1.64; CI: 1.04 to 2.61; p =0.034); and (5) chronic use of steroids within 30 days prior to the procedure (OR: 5.86; CI: 2.81 to 12.24; p< 0.001). Interestingly, if the patients received antiplatelet therapy or warfarin postoperatively, they were less likely to experience an IAC (OR: 0.40; CI: 0.24 to 0.68; p =0.001 and OR: 0.36; CI: 0.18 to 0.70; p =0.003, respectively) (Table 2).
Logistic regression was also used to determine independent predictors of mortality for the complication group (Table 3). Factors affecting mortality were CVD (OR: 4.80; CI: 1.4 to 17; p =0.015), age with an incremental risk of death with each additional year of age (OR: 1.10; CI: 1.02 to 1.20; p =0.013), CPB above 119 minutes (OR: 7.30; CI: 1.8 to 28.52; p =0.004), and use of inotropic therapy (OR: 14.31; 3.31 to 61.95; p< 0.001). Mesenteric ischemia, a highly lethal complication, only approached significance (OR: 3.60; CI: 0.85 to 15.06; p = 0.082). In contrast, postoperative antiplatelet therapy was possibly protective (OR: 0.34; CI 0.11 to 1.06; p =0.062).
Matched Sample
Data were collected on a randomly selected sample of 484 patients with no IAC from the same time period to find matched controls. Propensity scores were calculated for each patient using logistic regression based on all patient variables (see Table I). One hundred and six of the 112 complication cases (94.6%) were matched (1-to-1) with no-IAC controls. The c-statistic for the model was 0.79 (Hosmer-Lemeshow goodness-of-fit p = 0.732). Before the match, the two groups significantly differed on several patient variables, but these differences were removed after a successful match (see Table 4).
After matching, the mortality rate (27.4%) for the complication group was similar to that of the matched no-IAC control group (22.6%) (p= 0.526). However, the total length of hospital stay was longer in the complications group 23.5 days (±16.8 SD) versus 12.2 days (±10.9 SD) for the controls (t = 5.82, df = 210; p< 0.001).
Logistic regression was also used to determine independent predictors of mortality for the matched group of patients and controls (Table 5). It was determined that patients whose procedures were performed off pump (OR: 9.90; CI: 1.47 to 65.95; p =0.018) or with CPB time 150 or greater minutes (OR: 4.44; CI: 1.48 to 13.31; p= 0.008) were much more likely to expire. One-year increments in age (OR: 1.06; CI: 1.02 to 1.10; p= 0.006) and the use of intraaortic balloon pumps were found to be associated with increased risk of death (OR: 3.73; 1.54 to 9.01; p =0.003). Antiplatelet therapy and the use of warfarin were found to be possibly protective (OR: 0.10; CI 0.4 to 0.24; p < 0.001 and OR: 0.25; CI 0.07 to 0.93; p= 0.038, respectively).
|
Comment
|
|---|
Although the use of inotropic therapy was a significant predictor of mortality in the IAC group this finding possibly reflects the use of inotropes as a terminal event. Appropriate inotropes may be helpful when used early in the perioperative period to enhance cardiac output and perfusion pressures. Because of the wide variability in timing, doses, and types of inotropes used a meaningful analysis of their impact was not possible. Patients with depressed cardiac output and the other described risk factors should be monitored closely for IAC and attention to the abdomen should be an essential part of the daily evaluation of these patients.
Contrary to other studies we did not find diabetes, cerebral vascular disease, emergency surgery, redo operations, postoperative atrial arrhythmias, preoperative renal insufficiency (creatinine > 2.0 mg/dL), use of intraaortic balloon pump, use of heparin postoperatively, and preoperative length of hospital stay to be significant predictors of IAC (see Table 2).
However, the factors we found to be significant predictors of IAC differed somewhat from other reports and pointed mainly to causes that might reduce splanchnic perfusion and(or) resistance to infection. Patients with a combination of prolonged CPB and depressed left ventricular function in the presence of PVD may be particularly susceptible to abdominal complications. Chronic steroid use had a strong association with occurrence of IAC. Advancing age increased mortality in both the complication and matched control groups.
The influence of the GI tract in the production of multiorgan failure, the usual terminal event in IAC, has been recognized [8]. We found that antiplatelet therapy as well as warfarin had a protective effect on IAC. We feel that these medications may help by preventing platelet aggregation and thrombosis in the presence of depressed cardiac output and(or) advanced vascular disease. Strategies to maintain cardiac output and perfusion pressures in the postoperative period together with antiplatelet therapy may be helpful in the prevention of IAC.
Clearly, prevention and early intervention should be the aims in the management of IAC [9, 18, 21]. A high index of suspicion would lead to early diagnostic and therapeutic measures. Ileus, abdominal distention, tenderness, diarrhea, unexplained metabolic acidosis, and azotemia should lead to prompt investigation for IAC. Endoscopy and abdominal exploration are tolerated well by these patients and should be used without hesitation. Early surgical intervention seems to improve outcomes [7, 9, 11]. Mesenteric angiography and mesenteric infusion with vasodilators [14, 22, 23] have been suggested and may be helpful but should not be allowed to delay prompt surgical intervention. High-risk patients should be advised of the possibility of IAC. With an aging patient population that presents with a higher risk profile it is important that we devote time and resources to understand these complications and develop a comprehensive approach to their prevention and treatment.
We found the incidence (1.5%) and mortality (28.6%) of IAC to be similar to other reported series [1, 5, 9, 16]. Interestingly, after the propensity matching we found that the mortality rate was similar between the complication and the control groups. At first this seems counterintuitive and somewhat difficult to explain. It is possible that this is a true finding because normally IACs develop among patients who are at higher risk for mortality in general. To support this argument, we found the predictors of IAC to be PVD, an EF less than 0.50, chronic steroid use, and longer pump times. This suggests that the IAC patients were at high risk. Thus, when compared with matched patients who have similar risk profiles, it is likely that the mortality rates would be the similar. Furthermore, treatment after a complication would lead to a longer stay but not necessarily higher mortality. That is exactly what we found.
Also, after matching we found mortality risk to be much higher when the procedures were performed off pump. Although not the focus of the current study, this is a curious and unexpected finding. Debate exists in the current literature as to which technique is superior. Perhaps this finding is true, but there are reasons not to put much emphasis on it. The small number of off-pump cases made meaningful analysis difficult as evidenced by the large confidence intervals surrounding the odds ratio for mortality (1.47 to 65.95). However, we do believe that our somewhat unique approach to pump time is worth further consideration. Instead of having two variables, a dichotomized on-off pump variable and a continuous pump time variable (which would have to be set to missing or zero for off pump procedures), one categoric variable can be used.
Limitations
The major limitation of the current study is the retrospective design, which by default contains many limitations and inherent selection bias of a nonrandomized study design. Multivariate analyses and propensity matching were used in an attempt to overcome some of the limitations. These methods can never be a substitute for a randomized study, but often they are more practical, economical, and versatile. However, there are limitations that these methods cannot overcome. The study could only look at a limited number of variables and may have missed some unknown factors.
|
References
|
|---|
- Fitzgerald T, Kim D, Karakozis S, Alam H, Provido H, Kirkpatrick J. Visceral ischemia after cardiopulmonary bypass Am Surg 2000;6:623-626.
- Simic O, Strathausen S, Geidel S, Hess W, Ostermeyer J. Abdominal complications following cardiac surgery Acta Med Croatica 1997;51:191-196.[Medline]
- Baily RW, Bulkley GB, Hamilton SR, Morris JB, Haglund UF. Protection of the small intestine from non-occlusive mesenteric ischemic injury due to cardiogenic shock Am J Surg 1987;153:108-116.[Medline]
- Gennoro M, Ascer E, Mataro R, Jacobowitz IJ, Cunningham JN, Uceda P. Acute mesenteric ischemia after cardiopulmonary bypass Am J Surg 1993;166:231-236.[Medline]
- Christenson JT, Schmuziger M, Maurice J, Simonet F, Velebit V. Gastrointestinal complications after coronary artery bypass grafting J Thorac Cardiovasc Surg 1994;108:899-906.[Abstract/Free Full Text]
- Ohri SK, Bjarnason I, Pathi V, et al. Cardiopulmonary bypass impairs small intestinal transport and increases gut permeability Ann Thorac Surg 1993;55:1080-1086.[Abstract]
- Sakorafas GH, Tsiotos GG. Intra-abdominal complications after cardiac surgery Eur J Surg 1999;165:820-827.[Medline]
- Baue AE. The role of the gut in the development of multiple organ dysfunction in cardiothoracic patients Ann Thorac Surg 1993;55:822-829.[Abstract]
- Byhahn C, Strouhal U, Martens S, Mierdl S, Kessler P, Westphal K. Incidence of gastrointestinal complications in cardiopulmonary bypass patients World J Surg 2001;25:1140-1144.[Medline]
- Tsiotos GG, Mullany CJ, Zeitlow S, van Heerden JA. Abdominal complications following cardiac surgery Am J Surg 1994;167:553-557.[Medline]
- Huddy SJ, Joyce WP, Pepper JR. Gastrointestinal complications in 4473 patients who underwent cardiopulmonary bypass surgery Br J Surg 1991;78:293-296.[Medline]
- Ott MJ, Buchman TG, Baumgartner WA. Postoperative abdominal complications in cardiopulmonary bypass patients: a case-controlled study Ann Thorac Surg 1995;59:1210-1213.[Abstract/Free Full Text]
- Gaer JAR, Shaw ADS, Wild R, et al. Effect of cardiopulmonary bypass on gastrointestinal perfusion and function Ann Thorac Surg 1994;57:371-375.[Abstract]
- Biraima AM, Genoni M, Winkler MB, Bruhlmann W, Turina MI. Mesenteric ischemia after cardiac operation: arteriosclerotic versus vasospastic etiology J Cardiovasc Surg 2002;43:87-89.[Medline]
- Albes JM, Schistek R, Baier R, Unger F. Intestinal ischemia associated with cardiopulmonary bypass surgery J Cardiovasc Surg 1991;32:527-533.[Medline]
- Perugini RA, Orr RK, Porter D, Dumas EM, Maini BS. Gastrointestinal complications following cardiac surgery Arch Surg 1997;132:352-357.[Abstract]
- Leitman IM, Paull DE, Barie PS, Isom OW, Shires GT. Intra-abdominal complications of cardiopulmonary bypass operations Surg Gynecol Obstet 1987;165:251-254.[Medline]
- The Society of Thoracic Surgeons Cardiac Database Core Data Elements-Summary Version 2:35. 1998.
- Rosenbaum PR, Rubin DB. The bias due to incomplete matching Biometrics 1985;41:103-106.[Medline]
- Blackstone EH. Comparing apples and oranges J Thorac Cardiovasc Surg 2002;123:8-15.[Free Full Text]
- Carpati CM, Astiz ME, Rackow EC. Optimizing gastric mucosal perfusion: all catecholamines may not be created equal Crit Care Med 1997;25:1624-1625.[Medline]
- Schutz A, Eichinger W, Breuer M, Gansera B, Kemkes BM. Acute mesenteric ischemia after open heart surgery Angiology 1998;49:267-273.[Medline]
- Boley SJ, Regan JA, Tunick PA, Everhard ME, Winslow PR, Veith FJ. Persistent vasoconstriction- a major factor in non-occlusive mesenteric ischemia Current Topics in Surgical Research 1971;3:425-433.
This article has been cited by other articles:
|
|
|
|
 
P. C. Austin
Propensity-score matching in the cardiovascular surgery literature from 2004 to 2006: a systematic review and suggestions for improvement.
J. Thorac. Cardiovasc. Surg.,
November 1, 2007;
134(5):
1128 - 1135.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
