ORIGINAL ARTICLE – ADULT CARDIAC
Interactive CardioVascular and Thoracic Surgery 19 (2014) 419–425 doi:10.1093/icvts/ivu139 Advance Access publication 17 June 2014
Mesenteric ischaemia following cardiac surgery: the influence of intraoperative perfusion parameters Priya Sastrya, Gillian Hardmana, Aravinda Pagea, Richard Parkerb, Martin Goddarda, Stephen Largea and David P Jenkinsa,* a b
Department of Cardiothoracic Surgery, Papworth Hospital, Cambridge, UK Department of Public Health and Primary Care, Centre for Applied Medical Statistics, Robinson Way, UK
* Corresponding author. Department of Cardiothoracic Surgery, Papworth Hospital, Cambridge CB23 3RE, UK. Tel: +44-1-480830541; fax: +44-1-480831514; e-mail: [email protected] (D.P. Jenkins).
Abstract OBJECTIVES: Mesenteric ischaemia (MesI) remains a rare but lethal complication following cardiac surgery. Previously identified risk factors for MesI mortality (age, poor left ventricular (LV) function, cardiopulmonary bypass time and blood loss) are non-specific and cannot necessarily be modified. This study aims to identify potentially modifiable risk factors for MesI mortality through analysis of periand intraoperative perfusion data. METHODS: Patients who underwent cardiac surgery between 2006 and 2011 at Papworth Hospital were retrospectively divided into 3 outcome categories: death caused by MesI; death due to other causes and survival to discharge. A published MesI risk calculator was used to estimate risk of MesI for each patient and then to create 3 cohorts of matched patients from each outcome group. Pre-, intra- and postoperative variables were collected and conditional logistic regression methods were used to identify parameters associated specifically with MesI deaths after cardiac surgery. RESULTS: A total of 10 409 patients underwent cardiac surgery between 2006 and 2011. The incidence of MesI was 0.3% (30 patients). Two hundred and sixty-one patients died of non-MesI causes and 10 118 survived. It was possible to identify 25 patients in each group at equivalent risk of MesI. The following parameters were found to be associated with MesI mortality: recent myocardial infarction [odds ratio (OR) 4.98, 95% confidence interval (CI) 1.58–15.71, P = 0.01], standard EuroSCORE (OR 1.12, 95% CI 1.03–1.21, P = 0.01), vasopressor dose on bypass (OR 1.28, 95% CI 1.04–1.57, P = 0.02), metaraminol dose on bypass (OR 1.52, 95% CI 1.12–2.06, P = 0.01) and lowest documented mean arterial pressure (OR 0.90, 95% CI 0.83–0.97, P = 0.01). No other intraoperative perfusion-related parameters (e.g. flow, average activated clotting time or pressure) were associated with MesI mortality. CONCLUSIONS: Our study not only confirms previously known predictive factors, but also demonstrates a new association between intraoperative vasopressor use and MesI mortality. Keywords: Intestinal ischaemia • Mesenteric ischaemia • Cardiac surgery
INTRODUCTION Mesenteric ischaemia (MesI) is a rare complication following cardiac surgery, with an estimated incidence of 0.49–2% [1, 2]. Although it is rare, it carries a high risk of mortality (60–100%), prolonged postoperative recovery and impaired long-term survival [3–5]. Diagnosis of MesI following cardiac surgery is difficult. The initial symptoms and signs of abdominal pain, distension and guarding are not apparent in sedated, ventilated patients. Even in the awake patient, the signs may be masked following cardiac surgery due to confounding pain from the sternotomy incision or drain sites. Other clinical signs may be equally explained by ileus, postoperative nausea and vomiting or opiate analgesia, all of which are common after cardiac surgery. If MesI is present, patient survival relies on prompt identification of the condition and laparotomy. Attention has therefore been focused on predicting which
patients are particularly vulnerable to MesI. Thus, it is important to understand the causative process with a view to identifying modifiable risk factors. It has been speculated that occlusive mesenteric arterial or venous thromboembolism cause MesI. However, more recent evidence suggests a mechanism of low mesenteric blood flow following a low cardiac output state as an important factor in the development of MesI following cardiac surgery [6–8]. In a previous study from our own institution, we found that, in patients with MesI following cardiac surgery, the incidence of a non occlusive cause was as high as 96% [2]. In keeping with a hypothesis of low-flow aetiology, previous researchers have independently identified that the following parameters significantly increase the risk of MesI mortality following cardiac surgery: age >65 years, poor left ventricular function, recent myocardial infarction (MI, within 90 days of surgery),
© The Author 2014. Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.
ORIGINAL ARTICLE
Received 10 December 2013; received in revised form 2 April 2014; accepted 7 April 2014
420
P. Sastry et al. / Interactive CardioVascular and Thoracic Surgery
METHODS
for age, postoperative inotrope, postoperative renal replacement therapy, postoperative ventilation >48 h and postoperative atrial fibrillation. Post-mortem reports for patients in Groups A and B (if available) were collected and reviewed to identify evidence of perioperative MI, left ventricular hypertrophy, mesenteric arteriopathy and the pattern of bowel ischaemia (localized, likely to be large vessel aetiology versus generalized, and likely to be microvascular malperfusion). For all patients in Groups A, B and C, preoperative, intraoperative and postoperative data were reviewed. The full list of parameters collected is in Table 2.
Patient selection, data collection and definition of parameters
Intraoperative management
peripheral vascular disease (PVD), prolonged bypass time, haemodynamic instability, excessive blood loss and postoperative atrial fibrillation [5, 8–10]. One group has even proposed a risk prediction tool for MesI based on these parameters [11]. Unfortunately, most risk factors identified thus far are not modifiable. This study focuses on identifying any modifiable risk factors in patients who are known to be at high risk of MesI mortality.
Our institutional database was interrogated retrospectively to identify patients who underwent cardiac surgery between 2006 and 2011. One patient undergoing thoraco abdominal aneurysm repair was excluded, because this operation adds further technical causes of MesI. Patients undergoing off-pump procedures were also excluded from our analysis, partly because only a small proportion of patients undergo off-pump surgery, but also because we specifically aimed to study parameters related to extracorporeal perfusion during the operation. The remaining patients were stratified according to clinical outcomes in the following manner: (i) Group A patients died following cardiac surgery with MesI stated as the primary cause of death on the death certificate; (ii) Group B patients died following cardiac surgery, but from another cause (as declared on the death certificate); (iii) Group C patients had cardiac surgery and survived to discharge despite any complications. This study was facilitated by the fact that 70% of our in-hospital deaths undergo a post-mortem examination. However, a postmortem is not mandatory if the cause of death can be confidently declared by the treating physicians. Therefore, causes of death were taken from the death certificates. After dividing the surgical population into the 3 groups, the MesI risk prediction tool published by Chauduri et al. [11] was used to estimate risk of MesI for each patient and then create 3 cohorts of patients from Groups A, B and C matched for risk of MesI, rounded to whole integers (Table 1). The tool matches
At our institution, cardiopulmonary bypass is run in accordance with Cambridge Perfusion Services perfusion protocol. The targets are average activated clotting time over 400 s and cardiac index above 2.4 throughout the case. During the operation, perfusion parameters are recorded every 15 min and the charts are stored on site. Flows, pressures and vasopressor usage during cardiopulmonary bypass are adjusted at the discretion of the surgeon, anaesthetist and perfusionist.
Postoperative management After the operation, patients are transferred to intensive care with invasive arterial and central venous pressure monitoring. In cases where a low cardiac output state is suspected, invasive cardiac monitoring is achieved with a Swann-Ganz pulmonary artery catheter. First-line inotropic agents in our hospital are dopamine, adrenaline, noradrenaline and enoximone. Mechanical support (i.e. venoarterial extracorporeal membrane oxygenation or ventricular assist devices) is available for patients in unremitting low cardiac output state, but was not indicated in any of the patients in the study populations. Intra-aortic balloon pumps are also available and their pre- and postoperative usage were also collected. Patients who die of any cause within 30 days of cardiac surgery are referred to HM Coroner’s Officers and most proceed to postmortem examination. Cause of death (stratified into parts 1a, 1b, 1c and 2) is recorded within our institutional database.
Table 1: Risk prediction tool used to determine risk of ischaemic bowel (multivariate logistic regression model from Chauduri et al.) [11]
Postop inotrope and dialysis support Postop ventilation >48 h Age at operationa Postoperative atrial fibrillation Blood loss in ICU >700 ml Intercept
Coefficient
SE
OR
95% CI
P-value
1.9013 1.6287 0.0619 0.8276 0.6976 −10.9553
0.3784 0.3858 0.0186 0.3376 0.3344
6.7 5.1 1.06 2.3 2.0
3.2–14.1 2.4–10.9 1.03–1.1 1.2–4.4 1.04–3.9
48 ha Renal replacement therapya Inotrope usagea
a
These variables were collected for the purposes of matching the groups.
Statistical analysis Descriptive summary statistics relating to preoperative, interoperative and postoperative characteristics were calculated within each patient group. Statistically significant between-group differences were identified using Cochan’s Q test for categorical variables and Friedman’s test for the ordinal or continuous variables. Conditional logistic regression was used to investigate predictors of death from MesI, based on 1:2 matching comparing Group A with Groups B and C together. We used conditional logistic regression rather than ordinary logistic regression because patients were matched by MesI risk score. Separate univariate conditional logistic regression models were fitted to the outcome variable (MesI mortality) for each explanatory variable of interest, such that only one explanatory variable was included in the regression models at any one time. It was not possible to perform multiple regression analysis or investigate interaction effects due to the small sample size. A 5% significance level was used throughout. R version 2.15.1 [R Development Core Team (2012) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project. org/] was used for the conditional logistic regression analysis, and SPSS version 21 (IBM Corp. Released 2012. IBM SPSS Statistics for Windows, Version 21.0.IBM Corp., Armonk, NY, USA) was used to perform the Cochran’s Q tests. All other analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC, USA).
RESULTS During our study period, 10 409 patients underwent cardiac surgery. Thirty patients were confirmed to have died of MesI. Two hundred and sixty-one patients died of other causes. Ten thousand
one hundred and eighteen patients had cardiac surgery and survived to discharge (Fig. 1). The mean logistic EuroSCORE for the overall population was 20.2 (SD 19.9), mean age was 75.5 years (SD 7.12) and overall in-hospital mortality was 2.9%. Five cases of MesI deaths had to be discarded because case notes were unavailable or electronic records were incomplete, leaving 25 cases to be riskmatched. Thus, 3 matched cohorts were created from Groups A, B and C, as described above and each containing 25 patients. Fourteen of the 25 patients (64%) in Group A and 21 (84%) of the 25 patients in Group B had post-mortems. After review, it was found that 1 patient in Group A and none in Group B had mesenteric arteriopathy. In that 1 patient, the superior mesenteric artery was narrowed, but the pattern of bowel ischaemia was generalized and not restricted to superior mesenteric artery territory. Twenty-one percent of Group A patients and 33% of Group B patients had post-mortem evidence of perioperative myocardial infarct that may have contributed to a low cardiac output state. Ninety-two percent of patients in Group A and 76% of patients in Group B had evidence of left ventricular hypertrophy (defined as heart weight >500 g). Table 3 demonstrates the preoperative, intraoperative and postoperative characteristics of the 3 groups. Significant differences between groups were found for the following variables: redo rate, recent MI, standard EuroSCORE, metaraminol dose, percentage filtered while on bypass, percentage intra-aortic balloon pump (IABP), lowest documented mean arterial pressure, total volume of blood products transfused in hospital, volume of fresh-frozen plasma transfused, volume of platelets transfused and volume of red blood cells transfused. Most variables showed a similar pattern of results whereby the largest differences were seen between Groups A and C (MesI deaths vs survivors), with a few exceptions: 48% of patients had a recent MI in the MesI deaths group compared with only 8 and 20% in the non-MesI death and survivor
ORIGINAL ARTICLE
Operation status Recent myocardial infarction (MI within 90 days of cardiac surgery) Diabetes Peripheral vascular disease Cerebrovascular accident/transient ischaemic attack Smoking history (past or present) Hypertension Serum creatinine Left ventricular function Standard EuroSCORE Logistic EuroSCORE Preoperative heart rhythm
422
P. Sastry et al. / Interactive CardioVascular and Thoracic Surgery
Figure 1: Flowchart of patients. MESI: mesenteric ischaemia.
groups, respectively; the survivors group (Group C) was observed to have a substantially lower percentage of patients with IABP, lower percentage filtered while on bypass and lower volume of platelets transfused compared with the other 2 groups. Univariate conditional regression analysis was performed to analyse which factors were predictive of MesI deaths. Recent MI (OR 4.98, 95% CI 1.58–15.71), standard EuroSCORE (OR 1.12, 95% CI 1.03–1.21), metaraminol dose (OR 1.52, 95% CI 1.12–2.06) and total vasopressor use (OR 1.28, 95% CI 1.04–1.57) were significant predictors of death from MesI at the 5% level. Patients with lower minimum values of mean arterial pressure were also significantly more likely to die from Mesl (OR 0.90, 95% CI 0.83–0.97). PVD, red blood cell transfusion and cryoprecipitate transfusion were also significant predictors at the 10% level (Table 4). The variable ‘Redo’ had sparse data and, therefore, the conditional logistic regression model did not fit when this variable was included as an explanatory variable in the model. An McNemar’s test was performed instead (comparing Group A with the other two groups), which gave a P-value of 0.0005. The regression results corresponding to ‘age’ and ‘blood loss’ were not presented because they were used as matching variables in the study design.
DISCUSSION Our study design has yielded new information about risk factors for MesI mortality and the role of intraoperative factors. Our data demonstrate that redo surgery, recent MI, standard EuroSCORE, metaraminol dose, total vasopressor dose and lowest documented blood pressure (BP) are associated with death from MesI following cardiac surgery. Patients undergoing cardiac surgery inherently have risk factors and associations with PVD. The risk of embolic phenomena following cardiopulmonary bypass is also well documented. It has been speculated that occlusive arterial or venous thromboembolism may be responsible for MesI following cardiac surgery. Recent
evidence, however, disputes this, with only 4% of MesI after cardiac surgery being related to thromboembolism [2]. Current opinion favours the mechanism of low mesenteric blood flow following a low cardiac output state as the most important association with the development of MesI following cardiac surgery [6–8]. The hypothesis is that inadequate perfusion leads to intestinal intramucosal acidosis, which leads to a ‘vicious cycle’ of increased microvascular permeability and cytokine production that worsens sub-mucosal oxygen delivery and creates the substrate for MesI. It is possible that MesI after cardiac surgery follows after a multihit phenomenon where there are sequential low-flow insults to the periphery of the intestinal villi. Our data support this hypothesis by demonstrating that recent MI and intraoperative vasopressor dose are specifically predictive of MesI. The first ‘hit’ to intestinal perfusion may occur during the preoperative MI. The secondary hit(s) may occur intraoperatively, either due to the low pressure state requiring administration of vasopressor or due to the effects of vasopressor itself. Subsequent hits may occur due to microemboli while coming off bypass, low cardiac output or atrial fibrillation postoperatively. The preoperative risk factors cannot be modified. Postoperatively, the low cardiac output state is already recognized as a forerunner to high morbidity and mortality. It is therefore already treated aggressively with invasive cardiac output monitoring and inotropic support. However, there may be scope for modification of intraoperative variables, specifically as regards the management of extracorporeal perfusion. In an emergency case, there may be no alternative to high vasopressor dose if the patient is profoundly vasodilatated for any reason. However, in an elective scenario, measures such as mandatory cessation of vasodilatory drugs may be helpful for patients who are anticipated to be at high risk of MesI. Intraoperatively, it may be possible for the perfusionist to ‘overflow’ on cardiopulmonary bypass rather than use vasopressors to generate systemic pressure. Equally, it may even be appropriate to accept lower pressure as long as adequate flow (as evaluated by cardiac index) is maintained. It is also possible that normothermia with warm blood cardioplegia could be protective against MesI by
P. Sastry et al. / Interactive CardioVascular and Thoracic Surgery
423
Table 3: Preoperative, intraoperative and postoperative characteristics of the 3 cohorts propensity matched for risk of MesI
n Preoperative variables Mean agea Male sex, n (%) Redo rate, n (%) Mean body surface area Nonelective operation status, n (%) Recent myocardial infarction, n (%) Diabetes, n (%) Peripheral vascular disease, n (%) Cerebrovascular accident/transient ischaemic attack, n (%) Smoking history, n (%) Hypertension, n (%) Creatinine >200, n (%) Poor left ventricular function, n (%) Standard EuroSCORE, mean ± SD Logistic EuroSCORE, mean ± SD Preoperative atrial fibrillation, n (%)a Intraoperative parameters Cardiopulmonary bypass time (min) mean, SD Cross-clamp time (min) mean, SD Lowest temperature intraoperatively (°C) Average flow (l/min) Average cardiac index (l/min/m2) Average perfusion pressure on bypass (mmHg) mean ± SD Phenylephrine dose (mg) mean ± SD Metaraminol dose (mg) mean ± SD Total vasopressor dose (mg) mean ± SD Average activated clotting time, mean ± SD Fluid balance at the end of bypass (l) mean ± SD % Filtered while on bypass % Intra-aortic balloon pump Postoperative variables Temperature on arrival to intensive therapy unit (°C) Lowest documented blood pressure (mmHg), mean ± SD Blood loss (ml) mean ± SDa Total volume of blood products transfused in hospital (ml), mean ± SD Volume of cryoprecipitate transfused in hospital (ml), mean ± SD Volume of FFP transfused in hospital (ml), mean ± SD Volume of platelets transfused in hospital (ml), mean ± SD Volume of red blood cells transfused in hospital (ml), mean ± SD
All 75 patients matched for risk of MesI P-valueb
Group A MesI deaths
Group B non-MesI deaths
Group C survivors
25
25
25
75.7 ± 5.93 17 (68%) 0 (0%) 1.88 ± 0.18 13 (52%) 12 (48%) 6 (24%) 2 (8%) 1 (4%) 13 (52%) 22 (88%) 4 (16%) 7 (28%) 28.3 ± 25.0 18.1 ± 20.6 8 (32%)
74.1 ± 8.26 17 (68%) 5 (20%) 1.83 ± 0.24 14 (56%) 2 (8%) 3 (12%) 7 (28%) 2 (8%) 15 (60%) 17 (68%) 5 (20%) 5 (20%) 11.0 ± 4.01 28.4 ± 23.6 5 (20%)
76.7 ± 7.04 18 (72%) 7 (28%) 1.84 ± 0.20 11 (44%) 5 (20%) 7 (28%) 7 (28%) 0 (0%) 19 (76%) 23 (92%) 0 (0%) 4 (16%) 8.28 ± 2.79 14.0 ± 11.2 6 (24%)
0.538 0.943 0.039 0.820 0.738 0.007 0.395 0.168 0.223 0.247 0.103 0.066 0.529 0.009 0.147 0.459
135 ± 83.2 84.6 ± 45.8 30.3 ± 4.98 4.57 ± 0.51 2.43 ± 0.17 55.8 ± 9.84 1.17 ± 3.48 2.60 ± 2.31 3.58 ± 4.49 516 ± 144 1.10 ± 1.99 10 (40%) 8 (32%)
129 ± 96.9 71.7 ± 73.4 31.2 ± 4.87 4.56 ± 0.73 2.50 ± 0.32 58.1 ± 6.37 0.31 ± 0.97 1.73 ± 1.89 2.0 ± 2.32 517 ± 150 2.22 ± 1.86 10 (40%) 12 (48%)
97.3 ± 39.3 65.2 ± 31.9 33.0 ± 2.72 4.55 ± 0.69 2.49 ± 0.34 59.3 ± 7.28 0.09 ± 0.24 0.55 ± 0.23 0.65 ± 0.37 463 ± 60.4 1.82 ± 0.79 3 (12%) 2 (8%)
0.317 0.582 0.261 0.870 0.331 0.291 0.764 0.025 0.051 0.585 0.069 0.018 0.015
35.1 ± 0.98 47.4 ± 11.3 3885 ± 9231 3467 ± 4087 98.0 ± 199 817 ± 1576 459 ± 887 2092 ± 1805
35.0 ± 1.84 53.1 ± 9.16 2394 ± 1923 3441 ± 3091 33.6 ± 93.7 676 ± 1017 669 ± 950 2062 ± 2127
35.4 ± 0.86 61.7 ± 9.83 1308 ± 1431 951 ± 1075 12.5 ± 62.6 121 ± 337 114 ± 204 703 ± 748
0.438 0.0003 0.141 0.006 0.215 0.046 0.003 0.001
a
These parameters were among those used to propensity match patients in the three cohorts. Cochran’s Q-test for categorical variables and Friedman’s test for ordinal or continuous variables. FFP: fresh-frozen plasma.
b
reducing the severity of perioperative MI and/or low cardiac output state. Our study has shown that low postoperative blood pressure was associated with MesI, but we cannot determine from our recorded data whether the hypotension was due to low cardiac output, sepsis or vasoplegia. On the other hand, intra-aortic balloon pumps, which are used often and early in our institution for patients with low cardiac output syndrome were not associated with MesI. Therefore, we suspect that normothermia and warm blood plegia may help prevent MesI in high-risk patients, but there is inadequate information from this study to support this proposal. Indeed, the benefit of any preventative strategies for MesI is difficult to evaluate scientifically because the incidence of MesI after cardiac surgery is so low. A prospective randomized study would require several thousand patients to be adequately powered. Therefore, we believe that, despite the limitations of a retrospective analysis, matched cohort studies may be the best approach to
studying clinical methods of preventing MesI. Preventing MesI is a particularly worthwhile goal because diagnosing and treating MesI in patients after cardiac surgery is so difficult. The most useful diagnostic test available for a sedated patient with abdominal signs is CT with contrast medium. However, this has poor sensitivity and specificity for MesI. Furthermore, it is relatively contraindicated in our patient group, who are likely to be unstable by the very nature of the problem and at heightened risk of contrast-induced nephropathy. CT mesenteric angiography is the most sensitive and specific test for acute MesI and has the additional benefit of enabling interventions, such as injections of papaverine to cause local vasodilatation and embolization of thrombus. However, this is a sophisticated investigation that can be performed only in specialist centres. With regard to treatment options, laparotomy and bowel resection is the most common treatment for MesI, but only if a limited length of bowel is affected. The morbidity and mortality of such an operation is high.
ORIGINAL ARTICLE
Parameters
P. Sastry et al. / Interactive CardioVascular and Thoracic Surgery
424
Table 4: Results of univariate conditional logistic regression
Female Recent myocardial infarction Nonelective operation Diabetes Smoking history Intra-aortic balloon pump Peripheral vascular disease CVA/TIA Hypertension Preop creatinine Standard EuroSCORE Logistic EuroSCORE Cardiopulmonary bypass time XC time Lowest temp intraop Average flow Cardiac index Average PP on bypass Any phenylephrine dose Aramine dose Total vasopressors Average activated clotting time Fluid balance at end of CPB Temperature on arrival to ITU Lowest documented BP Total transfusion volume Volume of RBCS transfused Left ventricular function Good (Reference) Moderate Poor AF preop rhythm Filtered on bypass Any cryoprecipitate transfused Any FFP transfused Any PLTS transfused
Odds ratio
95% CI
P-value
1.09 4.98 1.07 1.23 0.56 1.18 0.25 1.00 2.25 1.00 1.12 0.99 1.00 1.01 0.92 1.13 0.56 0.95 1.59 1.52 1.28 1.00 1.00 0.89 0.90 1.00 1.00
0.40–2.98 1.58–15.71 0.45–2.55 0.42–3.66 0.22–1.39 0.45–3.10 0.05–1.16 0.05–18.91 0.49–10.41 1.00–1.00 1.03–1.21 0.97–1.02 1.00–1.01 1.00–1.02 0.83–1.03 0.52–2.46 0.11–2.83 0.89–1.01 0.46–5.49 1.12–2.06 1.04–1.57 1.00–1.01 1.00–1.00 0.61–1.30 0.83–0.97 1.00–1.00 1.00–1.00
0.864 0.006 0.883 0.706 0.211 0.739 0.077 1.000 0.300 0.561 0.008 0.539 0.285 0.243 0.142 0.754 0.481 0.116 0.461 0.008 0.022 0.394 0.920 0.531 0.007 0.113 0.095
1.00 1.93 2.35 2.12 3.56 0.76 1.08
0.35–2.91 0.53–7.03 0.55–10.05 0.74–6.11 0.88–14.49 0.27–2.15 0.41–2.83
0.996 0.321 0.249 0.164 0.076 0.608 0.871
CI: confidence interval; ITU: intensive therapy unit; FFP: fresh-frozen plasma; PLTS: platelets; BP: blood pressure; RBCS: red blood cells; AF: atrial fibrillation; preop: preoperative; XC:cross clamp; PP: perfusion pressure. CVA/TIA: cerebrovascular accident/transient ischaemic attack.
Intra-arterial injection of papaverine has been shown to be effective, but as explained above, requires specialist input from interventional radiologists. Until diagnostic and therapeutic techniques improve, we believe that prevention is the best strategy to improve outcomes. This study has corroborated the findings of other groups that poor left ventricular function, recent MI, age and blood loss are predictive of MesI. In addition, we have shown that patients who died of MesI had significantly higher doses of vasopressor while on cardiopulmonary bypass compared with patients who died of other causes or patients who survived the operation.
CONCLUSION Our study suggests that standard EuroSCORE, redo surgery, recent MI and vasopressor dose are significant associations of MesI. The
conduct of cardiopulmonary bypass, and the use of vasopressors, should be carried out with an awareness of mesenteric perfusion, particularly in high-risk patients. We advocate the use of the risk prediction tool published by Chauduri et al. to identify vulnerable patients who are due to undergo cardiac surgery with the use of cardiopulmonary bypass, a low threshold of suspicion for MesI and prompt investigation in these patients.
LIMITATIONS There are several difficulties inherent in studying MesI after cardiac surgery. The first is that the incidence of the complication is very low. Consequently, in this study, only 25 MesI patients could be identified who could be matched. Despite the small sample sizes, we were able to identify statistically significant differences in the perfusion parameters. However, the sample size was not high enough to attempt multivariate analysis, and it is possible that some of the relations observed may be due to the presence of confounding variables. Another major limitation in studying the topic is that there is no practical method for analysing mesenteric perfusion. This causes a problem both for diagnosing MesI and for evaluating the relative effects of drugs, perfusion methods or treatments on mesenteric perfusion. Mesenteric flow probes might alleviate this problem but are impractical to use in postoperative patients. Therefore, their use is likely to remain experimental. Conflict of interest: none declared.
REFERENCES [1] Allen KB, Salam AA, Lumcden AB. Acute mesenteric ischaemia after cardiopulmonary bypass. J Vasc Surg 1992;16:391–6. [2] Venkateswaran RV, Charman SC, Goddard M, Large SR. Lethal mesenteric ischaemia after cardiopulmonary bypass: a common complication? Eur J Cardiothoracic Surg 2002;22:534–8. [3] Schutz A, Eichinger W, Breuer M, Gansera B, Kemkes BM. Acute mesenteric ischaemia after open heart surgery. Angiology 1998;49:267–73. [4] Andersson B, Nilsson J, Brandt J, Higlund P, Andersson R. Gastrointestinal complications after cardiac surgery. Br J Surg 2005;92:326–33. [5] Filsoufi F, Rahmanian PB, Castillo JG, Scurlock C, Legnani PE, Adams DH. Predictors and outcomes of gastrointestinal complications in patients undergoing cardiac surgery. Ann Surg 2007;246:323–9. [6] Moneta GL, Misbach GA, Ivey TD. Hypoperfusion as a possible factor in the development of gastrointestinal complications after cardiac surgery. Am J Surg 1985;149:648–50. [7] Williams LF. Mesenteric ischaemia. Surg Clin North Am 1988;68:331–53. [8] Christenson JT, Schmuziger M, Maurice J, Simmonet F, Velebit V. Post operative visceral hypoperfusion the common cause for gastrointestinal complications after cardiac surgery. Thorac Cardiovasc Surg 1994;42: 152–7. [9] Mangi AA, Christison-Lagay ER, Torchiana DF, Warshaw AL, Berger DL. Gastrointestinal complications in patients undergoing heart operation: an analysis of 8709 consecutive cardiac surgical patients. Ann Surg 2005;241: 895–901. [10] Spotnitz WD, Sanders RP, Hanks JB, Nolan SP, Tribble CG, Bergin JD et al. General surgical complications can be predicted after cardiopulmony bypass. Ann Surg 1995;221:489–96. [11] Chaudhuri N, James J, Sheikh A, Grayson AD, Fabri BM. Intestinal ischaemia following cardiac surgery: a multivariate risk model. Eur J Cardiothorac Surg 2006;29:971–7.
Interactive CardioVascular and Thoracic Surgery 19 (2014) 419–425 doi:10.1093/icvts/ivu139 Advance Access publication 17 June 2014
Mesenteric ischaemia following cardiac surgery: the influence of intraoperative perfusion parameters Priya Sastrya, Gillian Hardmana, Aravinda Pagea, Richard Parkerb, Martin Goddarda, Stephen Largea and David P Jenkinsa,* a b
Department of Cardiothoracic Surgery, Papworth Hospital, Cambridge, UK Department of Public Health and Primary Care, Centre for Applied Medical Statistics, Robinson Way, UK
* Corresponding author. Department of Cardiothoracic Surgery, Papworth Hospital, Cambridge CB23 3RE, UK. Tel: +44-1-480830541; fax: +44-1-480831514; e-mail: [email protected] (D.P. Jenkins).
Abstract OBJECTIVES: Mesenteric ischaemia (MesI) remains a rare but lethal complication following cardiac surgery. Previously identified risk factors for MesI mortality (age, poor left ventricular (LV) function, cardiopulmonary bypass time and blood loss) are non-specific and cannot necessarily be modified. This study aims to identify potentially modifiable risk factors for MesI mortality through analysis of periand intraoperative perfusion data. METHODS: Patients who underwent cardiac surgery between 2006 and 2011 at Papworth Hospital were retrospectively divided into 3 outcome categories: death caused by MesI; death due to other causes and survival to discharge. A published MesI risk calculator was used to estimate risk of MesI for each patient and then to create 3 cohorts of matched patients from each outcome group. Pre-, intra- and postoperative variables were collected and conditional logistic regression methods were used to identify parameters associated specifically with MesI deaths after cardiac surgery. RESULTS: A total of 10 409 patients underwent cardiac surgery between 2006 and 2011. The incidence of MesI was 0.3% (30 patients). Two hundred and sixty-one patients died of non-MesI causes and 10 118 survived. It was possible to identify 25 patients in each group at equivalent risk of MesI. The following parameters were found to be associated with MesI mortality: recent myocardial infarction [odds ratio (OR) 4.98, 95% confidence interval (CI) 1.58–15.71, P = 0.01], standard EuroSCORE (OR 1.12, 95% CI 1.03–1.21, P = 0.01), vasopressor dose on bypass (OR 1.28, 95% CI 1.04–1.57, P = 0.02), metaraminol dose on bypass (OR 1.52, 95% CI 1.12–2.06, P = 0.01) and lowest documented mean arterial pressure (OR 0.90, 95% CI 0.83–0.97, P = 0.01). No other intraoperative perfusion-related parameters (e.g. flow, average activated clotting time or pressure) were associated with MesI mortality. CONCLUSIONS: Our study not only confirms previously known predictive factors, but also demonstrates a new association between intraoperative vasopressor use and MesI mortality. Keywords: Intestinal ischaemia • Mesenteric ischaemia • Cardiac surgery
INTRODUCTION Mesenteric ischaemia (MesI) is a rare complication following cardiac surgery, with an estimated incidence of 0.49–2% [1, 2]. Although it is rare, it carries a high risk of mortality (60–100%), prolonged postoperative recovery and impaired long-term survival [3–5]. Diagnosis of MesI following cardiac surgery is difficult. The initial symptoms and signs of abdominal pain, distension and guarding are not apparent in sedated, ventilated patients. Even in the awake patient, the signs may be masked following cardiac surgery due to confounding pain from the sternotomy incision or drain sites. Other clinical signs may be equally explained by ileus, postoperative nausea and vomiting or opiate analgesia, all of which are common after cardiac surgery. If MesI is present, patient survival relies on prompt identification of the condition and laparotomy. Attention has therefore been focused on predicting which
patients are particularly vulnerable to MesI. Thus, it is important to understand the causative process with a view to identifying modifiable risk factors. It has been speculated that occlusive mesenteric arterial or venous thromboembolism cause MesI. However, more recent evidence suggests a mechanism of low mesenteric blood flow following a low cardiac output state as an important factor in the development of MesI following cardiac surgery [6–8]. In a previous study from our own institution, we found that, in patients with MesI following cardiac surgery, the incidence of a non occlusive cause was as high as 96% [2]. In keeping with a hypothesis of low-flow aetiology, previous researchers have independently identified that the following parameters significantly increase the risk of MesI mortality following cardiac surgery: age >65 years, poor left ventricular function, recent myocardial infarction (MI, within 90 days of surgery),
© The Author 2014. Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.
ORIGINAL ARTICLE
Received 10 December 2013; received in revised form 2 April 2014; accepted 7 April 2014
420
P. Sastry et al. / Interactive CardioVascular and Thoracic Surgery
METHODS
for age, postoperative inotrope, postoperative renal replacement therapy, postoperative ventilation >48 h and postoperative atrial fibrillation. Post-mortem reports for patients in Groups A and B (if available) were collected and reviewed to identify evidence of perioperative MI, left ventricular hypertrophy, mesenteric arteriopathy and the pattern of bowel ischaemia (localized, likely to be large vessel aetiology versus generalized, and likely to be microvascular malperfusion). For all patients in Groups A, B and C, preoperative, intraoperative and postoperative data were reviewed. The full list of parameters collected is in Table 2.
Patient selection, data collection and definition of parameters
Intraoperative management
peripheral vascular disease (PVD), prolonged bypass time, haemodynamic instability, excessive blood loss and postoperative atrial fibrillation [5, 8–10]. One group has even proposed a risk prediction tool for MesI based on these parameters [11]. Unfortunately, most risk factors identified thus far are not modifiable. This study focuses on identifying any modifiable risk factors in patients who are known to be at high risk of MesI mortality.
Our institutional database was interrogated retrospectively to identify patients who underwent cardiac surgery between 2006 and 2011. One patient undergoing thoraco abdominal aneurysm repair was excluded, because this operation adds further technical causes of MesI. Patients undergoing off-pump procedures were also excluded from our analysis, partly because only a small proportion of patients undergo off-pump surgery, but also because we specifically aimed to study parameters related to extracorporeal perfusion during the operation. The remaining patients were stratified according to clinical outcomes in the following manner: (i) Group A patients died following cardiac surgery with MesI stated as the primary cause of death on the death certificate; (ii) Group B patients died following cardiac surgery, but from another cause (as declared on the death certificate); (iii) Group C patients had cardiac surgery and survived to discharge despite any complications. This study was facilitated by the fact that 70% of our in-hospital deaths undergo a post-mortem examination. However, a postmortem is not mandatory if the cause of death can be confidently declared by the treating physicians. Therefore, causes of death were taken from the death certificates. After dividing the surgical population into the 3 groups, the MesI risk prediction tool published by Chauduri et al. [11] was used to estimate risk of MesI for each patient and then create 3 cohorts of patients from Groups A, B and C matched for risk of MesI, rounded to whole integers (Table 1). The tool matches
At our institution, cardiopulmonary bypass is run in accordance with Cambridge Perfusion Services perfusion protocol. The targets are average activated clotting time over 400 s and cardiac index above 2.4 throughout the case. During the operation, perfusion parameters are recorded every 15 min and the charts are stored on site. Flows, pressures and vasopressor usage during cardiopulmonary bypass are adjusted at the discretion of the surgeon, anaesthetist and perfusionist.
Postoperative management After the operation, patients are transferred to intensive care with invasive arterial and central venous pressure monitoring. In cases where a low cardiac output state is suspected, invasive cardiac monitoring is achieved with a Swann-Ganz pulmonary artery catheter. First-line inotropic agents in our hospital are dopamine, adrenaline, noradrenaline and enoximone. Mechanical support (i.e. venoarterial extracorporeal membrane oxygenation or ventricular assist devices) is available for patients in unremitting low cardiac output state, but was not indicated in any of the patients in the study populations. Intra-aortic balloon pumps are also available and their pre- and postoperative usage were also collected. Patients who die of any cause within 30 days of cardiac surgery are referred to HM Coroner’s Officers and most proceed to postmortem examination. Cause of death (stratified into parts 1a, 1b, 1c and 2) is recorded within our institutional database.
Table 1: Risk prediction tool used to determine risk of ischaemic bowel (multivariate logistic regression model from Chauduri et al.) [11]
Postop inotrope and dialysis support Postop ventilation >48 h Age at operationa Postoperative atrial fibrillation Blood loss in ICU >700 ml Intercept
Coefficient
SE
OR
95% CI
P-value
1.9013 1.6287 0.0619 0.8276 0.6976 −10.9553
0.3784 0.3858 0.0186 0.3376 0.3344
6.7 5.1 1.06 2.3 2.0
3.2–14.1 2.4–10.9 1.03–1.1 1.2–4.4 1.04–3.9
48 ha Renal replacement therapya Inotrope usagea
a
These variables were collected for the purposes of matching the groups.
Statistical analysis Descriptive summary statistics relating to preoperative, interoperative and postoperative characteristics were calculated within each patient group. Statistically significant between-group differences were identified using Cochan’s Q test for categorical variables and Friedman’s test for the ordinal or continuous variables. Conditional logistic regression was used to investigate predictors of death from MesI, based on 1:2 matching comparing Group A with Groups B and C together. We used conditional logistic regression rather than ordinary logistic regression because patients were matched by MesI risk score. Separate univariate conditional logistic regression models were fitted to the outcome variable (MesI mortality) for each explanatory variable of interest, such that only one explanatory variable was included in the regression models at any one time. It was not possible to perform multiple regression analysis or investigate interaction effects due to the small sample size. A 5% significance level was used throughout. R version 2.15.1 [R Development Core Team (2012) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project. org/] was used for the conditional logistic regression analysis, and SPSS version 21 (IBM Corp. Released 2012. IBM SPSS Statistics for Windows, Version 21.0.IBM Corp., Armonk, NY, USA) was used to perform the Cochran’s Q tests. All other analyses were conducted using SAS version 9.3 (SAS Institute, Cary, NC, USA).
RESULTS During our study period, 10 409 patients underwent cardiac surgery. Thirty patients were confirmed to have died of MesI. Two hundred and sixty-one patients died of other causes. Ten thousand
one hundred and eighteen patients had cardiac surgery and survived to discharge (Fig. 1). The mean logistic EuroSCORE for the overall population was 20.2 (SD 19.9), mean age was 75.5 years (SD 7.12) and overall in-hospital mortality was 2.9%. Five cases of MesI deaths had to be discarded because case notes were unavailable or electronic records were incomplete, leaving 25 cases to be riskmatched. Thus, 3 matched cohorts were created from Groups A, B and C, as described above and each containing 25 patients. Fourteen of the 25 patients (64%) in Group A and 21 (84%) of the 25 patients in Group B had post-mortems. After review, it was found that 1 patient in Group A and none in Group B had mesenteric arteriopathy. In that 1 patient, the superior mesenteric artery was narrowed, but the pattern of bowel ischaemia was generalized and not restricted to superior mesenteric artery territory. Twenty-one percent of Group A patients and 33% of Group B patients had post-mortem evidence of perioperative myocardial infarct that may have contributed to a low cardiac output state. Ninety-two percent of patients in Group A and 76% of patients in Group B had evidence of left ventricular hypertrophy (defined as heart weight >500 g). Table 3 demonstrates the preoperative, intraoperative and postoperative characteristics of the 3 groups. Significant differences between groups were found for the following variables: redo rate, recent MI, standard EuroSCORE, metaraminol dose, percentage filtered while on bypass, percentage intra-aortic balloon pump (IABP), lowest documented mean arterial pressure, total volume of blood products transfused in hospital, volume of fresh-frozen plasma transfused, volume of platelets transfused and volume of red blood cells transfused. Most variables showed a similar pattern of results whereby the largest differences were seen between Groups A and C (MesI deaths vs survivors), with a few exceptions: 48% of patients had a recent MI in the MesI deaths group compared with only 8 and 20% in the non-MesI death and survivor
ORIGINAL ARTICLE
Operation status Recent myocardial infarction (MI within 90 days of cardiac surgery) Diabetes Peripheral vascular disease Cerebrovascular accident/transient ischaemic attack Smoking history (past or present) Hypertension Serum creatinine Left ventricular function Standard EuroSCORE Logistic EuroSCORE Preoperative heart rhythm
422
P. Sastry et al. / Interactive CardioVascular and Thoracic Surgery
Figure 1: Flowchart of patients. MESI: mesenteric ischaemia.
groups, respectively; the survivors group (Group C) was observed to have a substantially lower percentage of patients with IABP, lower percentage filtered while on bypass and lower volume of platelets transfused compared with the other 2 groups. Univariate conditional regression analysis was performed to analyse which factors were predictive of MesI deaths. Recent MI (OR 4.98, 95% CI 1.58–15.71), standard EuroSCORE (OR 1.12, 95% CI 1.03–1.21), metaraminol dose (OR 1.52, 95% CI 1.12–2.06) and total vasopressor use (OR 1.28, 95% CI 1.04–1.57) were significant predictors of death from MesI at the 5% level. Patients with lower minimum values of mean arterial pressure were also significantly more likely to die from Mesl (OR 0.90, 95% CI 0.83–0.97). PVD, red blood cell transfusion and cryoprecipitate transfusion were also significant predictors at the 10% level (Table 4). The variable ‘Redo’ had sparse data and, therefore, the conditional logistic regression model did not fit when this variable was included as an explanatory variable in the model. An McNemar’s test was performed instead (comparing Group A with the other two groups), which gave a P-value of 0.0005. The regression results corresponding to ‘age’ and ‘blood loss’ were not presented because they were used as matching variables in the study design.
DISCUSSION Our study design has yielded new information about risk factors for MesI mortality and the role of intraoperative factors. Our data demonstrate that redo surgery, recent MI, standard EuroSCORE, metaraminol dose, total vasopressor dose and lowest documented blood pressure (BP) are associated with death from MesI following cardiac surgery. Patients undergoing cardiac surgery inherently have risk factors and associations with PVD. The risk of embolic phenomena following cardiopulmonary bypass is also well documented. It has been speculated that occlusive arterial or venous thromboembolism may be responsible for MesI following cardiac surgery. Recent
evidence, however, disputes this, with only 4% of MesI after cardiac surgery being related to thromboembolism [2]. Current opinion favours the mechanism of low mesenteric blood flow following a low cardiac output state as the most important association with the development of MesI following cardiac surgery [6–8]. The hypothesis is that inadequate perfusion leads to intestinal intramucosal acidosis, which leads to a ‘vicious cycle’ of increased microvascular permeability and cytokine production that worsens sub-mucosal oxygen delivery and creates the substrate for MesI. It is possible that MesI after cardiac surgery follows after a multihit phenomenon where there are sequential low-flow insults to the periphery of the intestinal villi. Our data support this hypothesis by demonstrating that recent MI and intraoperative vasopressor dose are specifically predictive of MesI. The first ‘hit’ to intestinal perfusion may occur during the preoperative MI. The secondary hit(s) may occur intraoperatively, either due to the low pressure state requiring administration of vasopressor or due to the effects of vasopressor itself. Subsequent hits may occur due to microemboli while coming off bypass, low cardiac output or atrial fibrillation postoperatively. The preoperative risk factors cannot be modified. Postoperatively, the low cardiac output state is already recognized as a forerunner to high morbidity and mortality. It is therefore already treated aggressively with invasive cardiac output monitoring and inotropic support. However, there may be scope for modification of intraoperative variables, specifically as regards the management of extracorporeal perfusion. In an emergency case, there may be no alternative to high vasopressor dose if the patient is profoundly vasodilatated for any reason. However, in an elective scenario, measures such as mandatory cessation of vasodilatory drugs may be helpful for patients who are anticipated to be at high risk of MesI. Intraoperatively, it may be possible for the perfusionist to ‘overflow’ on cardiopulmonary bypass rather than use vasopressors to generate systemic pressure. Equally, it may even be appropriate to accept lower pressure as long as adequate flow (as evaluated by cardiac index) is maintained. It is also possible that normothermia with warm blood cardioplegia could be protective against MesI by
P. Sastry et al. / Interactive CardioVascular and Thoracic Surgery
423
Table 3: Preoperative, intraoperative and postoperative characteristics of the 3 cohorts propensity matched for risk of MesI
n Preoperative variables Mean agea Male sex, n (%) Redo rate, n (%) Mean body surface area Nonelective operation status, n (%) Recent myocardial infarction, n (%) Diabetes, n (%) Peripheral vascular disease, n (%) Cerebrovascular accident/transient ischaemic attack, n (%) Smoking history, n (%) Hypertension, n (%) Creatinine >200, n (%) Poor left ventricular function, n (%) Standard EuroSCORE, mean ± SD Logistic EuroSCORE, mean ± SD Preoperative atrial fibrillation, n (%)a Intraoperative parameters Cardiopulmonary bypass time (min) mean, SD Cross-clamp time (min) mean, SD Lowest temperature intraoperatively (°C) Average flow (l/min) Average cardiac index (l/min/m2) Average perfusion pressure on bypass (mmHg) mean ± SD Phenylephrine dose (mg) mean ± SD Metaraminol dose (mg) mean ± SD Total vasopressor dose (mg) mean ± SD Average activated clotting time, mean ± SD Fluid balance at the end of bypass (l) mean ± SD % Filtered while on bypass % Intra-aortic balloon pump Postoperative variables Temperature on arrival to intensive therapy unit (°C) Lowest documented blood pressure (mmHg), mean ± SD Blood loss (ml) mean ± SDa Total volume of blood products transfused in hospital (ml), mean ± SD Volume of cryoprecipitate transfused in hospital (ml), mean ± SD Volume of FFP transfused in hospital (ml), mean ± SD Volume of platelets transfused in hospital (ml), mean ± SD Volume of red blood cells transfused in hospital (ml), mean ± SD
All 75 patients matched for risk of MesI P-valueb
Group A MesI deaths
Group B non-MesI deaths
Group C survivors
25
25
25
75.7 ± 5.93 17 (68%) 0 (0%) 1.88 ± 0.18 13 (52%) 12 (48%) 6 (24%) 2 (8%) 1 (4%) 13 (52%) 22 (88%) 4 (16%) 7 (28%) 28.3 ± 25.0 18.1 ± 20.6 8 (32%)
74.1 ± 8.26 17 (68%) 5 (20%) 1.83 ± 0.24 14 (56%) 2 (8%) 3 (12%) 7 (28%) 2 (8%) 15 (60%) 17 (68%) 5 (20%) 5 (20%) 11.0 ± 4.01 28.4 ± 23.6 5 (20%)
76.7 ± 7.04 18 (72%) 7 (28%) 1.84 ± 0.20 11 (44%) 5 (20%) 7 (28%) 7 (28%) 0 (0%) 19 (76%) 23 (92%) 0 (0%) 4 (16%) 8.28 ± 2.79 14.0 ± 11.2 6 (24%)
0.538 0.943 0.039 0.820 0.738 0.007 0.395 0.168 0.223 0.247 0.103 0.066 0.529 0.009 0.147 0.459
135 ± 83.2 84.6 ± 45.8 30.3 ± 4.98 4.57 ± 0.51 2.43 ± 0.17 55.8 ± 9.84 1.17 ± 3.48 2.60 ± 2.31 3.58 ± 4.49 516 ± 144 1.10 ± 1.99 10 (40%) 8 (32%)
129 ± 96.9 71.7 ± 73.4 31.2 ± 4.87 4.56 ± 0.73 2.50 ± 0.32 58.1 ± 6.37 0.31 ± 0.97 1.73 ± 1.89 2.0 ± 2.32 517 ± 150 2.22 ± 1.86 10 (40%) 12 (48%)
97.3 ± 39.3 65.2 ± 31.9 33.0 ± 2.72 4.55 ± 0.69 2.49 ± 0.34 59.3 ± 7.28 0.09 ± 0.24 0.55 ± 0.23 0.65 ± 0.37 463 ± 60.4 1.82 ± 0.79 3 (12%) 2 (8%)
0.317 0.582 0.261 0.870 0.331 0.291 0.764 0.025 0.051 0.585 0.069 0.018 0.015
35.1 ± 0.98 47.4 ± 11.3 3885 ± 9231 3467 ± 4087 98.0 ± 199 817 ± 1576 459 ± 887 2092 ± 1805
35.0 ± 1.84 53.1 ± 9.16 2394 ± 1923 3441 ± 3091 33.6 ± 93.7 676 ± 1017 669 ± 950 2062 ± 2127
35.4 ± 0.86 61.7 ± 9.83 1308 ± 1431 951 ± 1075 12.5 ± 62.6 121 ± 337 114 ± 204 703 ± 748
0.438 0.0003 0.141 0.006 0.215 0.046 0.003 0.001
a
These parameters were among those used to propensity match patients in the three cohorts. Cochran’s Q-test for categorical variables and Friedman’s test for ordinal or continuous variables. FFP: fresh-frozen plasma.
b
reducing the severity of perioperative MI and/or low cardiac output state. Our study has shown that low postoperative blood pressure was associated with MesI, but we cannot determine from our recorded data whether the hypotension was due to low cardiac output, sepsis or vasoplegia. On the other hand, intra-aortic balloon pumps, which are used often and early in our institution for patients with low cardiac output syndrome were not associated with MesI. Therefore, we suspect that normothermia and warm blood plegia may help prevent MesI in high-risk patients, but there is inadequate information from this study to support this proposal. Indeed, the benefit of any preventative strategies for MesI is difficult to evaluate scientifically because the incidence of MesI after cardiac surgery is so low. A prospective randomized study would require several thousand patients to be adequately powered. Therefore, we believe that, despite the limitations of a retrospective analysis, matched cohort studies may be the best approach to
studying clinical methods of preventing MesI. Preventing MesI is a particularly worthwhile goal because diagnosing and treating MesI in patients after cardiac surgery is so difficult. The most useful diagnostic test available for a sedated patient with abdominal signs is CT with contrast medium. However, this has poor sensitivity and specificity for MesI. Furthermore, it is relatively contraindicated in our patient group, who are likely to be unstable by the very nature of the problem and at heightened risk of contrast-induced nephropathy. CT mesenteric angiography is the most sensitive and specific test for acute MesI and has the additional benefit of enabling interventions, such as injections of papaverine to cause local vasodilatation and embolization of thrombus. However, this is a sophisticated investigation that can be performed only in specialist centres. With regard to treatment options, laparotomy and bowel resection is the most common treatment for MesI, but only if a limited length of bowel is affected. The morbidity and mortality of such an operation is high.
ORIGINAL ARTICLE
Parameters
P. Sastry et al. / Interactive CardioVascular and Thoracic Surgery
424
Table 4: Results of univariate conditional logistic regression
Female Recent myocardial infarction Nonelective operation Diabetes Smoking history Intra-aortic balloon pump Peripheral vascular disease CVA/TIA Hypertension Preop creatinine Standard EuroSCORE Logistic EuroSCORE Cardiopulmonary bypass time XC time Lowest temp intraop Average flow Cardiac index Average PP on bypass Any phenylephrine dose Aramine dose Total vasopressors Average activated clotting time Fluid balance at end of CPB Temperature on arrival to ITU Lowest documented BP Total transfusion volume Volume of RBCS transfused Left ventricular function Good (Reference) Moderate Poor AF preop rhythm Filtered on bypass Any cryoprecipitate transfused Any FFP transfused Any PLTS transfused
Odds ratio
95% CI
P-value
1.09 4.98 1.07 1.23 0.56 1.18 0.25 1.00 2.25 1.00 1.12 0.99 1.00 1.01 0.92 1.13 0.56 0.95 1.59 1.52 1.28 1.00 1.00 0.89 0.90 1.00 1.00
0.40–2.98 1.58–15.71 0.45–2.55 0.42–3.66 0.22–1.39 0.45–3.10 0.05–1.16 0.05–18.91 0.49–10.41 1.00–1.00 1.03–1.21 0.97–1.02 1.00–1.01 1.00–1.02 0.83–1.03 0.52–2.46 0.11–2.83 0.89–1.01 0.46–5.49 1.12–2.06 1.04–1.57 1.00–1.01 1.00–1.00 0.61–1.30 0.83–0.97 1.00–1.00 1.00–1.00
0.864 0.006 0.883 0.706 0.211 0.739 0.077 1.000 0.300 0.561 0.008 0.539 0.285 0.243 0.142 0.754 0.481 0.116 0.461 0.008 0.022 0.394 0.920 0.531 0.007 0.113 0.095
1.00 1.93 2.35 2.12 3.56 0.76 1.08
0.35–2.91 0.53–7.03 0.55–10.05 0.74–6.11 0.88–14.49 0.27–2.15 0.41–2.83
0.996 0.321 0.249 0.164 0.076 0.608 0.871
CI: confidence interval; ITU: intensive therapy unit; FFP: fresh-frozen plasma; PLTS: platelets; BP: blood pressure; RBCS: red blood cells; AF: atrial fibrillation; preop: preoperative; XC:cross clamp; PP: perfusion pressure. CVA/TIA: cerebrovascular accident/transient ischaemic attack.
Intra-arterial injection of papaverine has been shown to be effective, but as explained above, requires specialist input from interventional radiologists. Until diagnostic and therapeutic techniques improve, we believe that prevention is the best strategy to improve outcomes. This study has corroborated the findings of other groups that poor left ventricular function, recent MI, age and blood loss are predictive of MesI. In addition, we have shown that patients who died of MesI had significantly higher doses of vasopressor while on cardiopulmonary bypass compared with patients who died of other causes or patients who survived the operation.
CONCLUSION Our study suggests that standard EuroSCORE, redo surgery, recent MI and vasopressor dose are significant associations of MesI. The
conduct of cardiopulmonary bypass, and the use of vasopressors, should be carried out with an awareness of mesenteric perfusion, particularly in high-risk patients. We advocate the use of the risk prediction tool published by Chauduri et al. to identify vulnerable patients who are due to undergo cardiac surgery with the use of cardiopulmonary bypass, a low threshold of suspicion for MesI and prompt investigation in these patients.
LIMITATIONS There are several difficulties inherent in studying MesI after cardiac surgery. The first is that the incidence of the complication is very low. Consequently, in this study, only 25 MesI patients could be identified who could be matched. Despite the small sample sizes, we were able to identify statistically significant differences in the perfusion parameters. However, the sample size was not high enough to attempt multivariate analysis, and it is possible that some of the relations observed may be due to the presence of confounding variables. Another major limitation in studying the topic is that there is no practical method for analysing mesenteric perfusion. This causes a problem both for diagnosing MesI and for evaluating the relative effects of drugs, perfusion methods or treatments on mesenteric perfusion. Mesenteric flow probes might alleviate this problem but are impractical to use in postoperative patients. Therefore, their use is likely to remain experimental. Conflict of interest: none declared.
REFERENCES [1] Allen KB, Salam AA, Lumcden AB. Acute mesenteric ischaemia after cardiopulmonary bypass. J Vasc Surg 1992;16:391–6. [2] Venkateswaran RV, Charman SC, Goddard M, Large SR. Lethal mesenteric ischaemia after cardiopulmonary bypass: a common complication? Eur J Cardiothoracic Surg 2002;22:534–8. [3] Schutz A, Eichinger W, Breuer M, Gansera B, Kemkes BM. Acute mesenteric ischaemia after open heart surgery. Angiology 1998;49:267–73. [4] Andersson B, Nilsson J, Brandt J, Higlund P, Andersson R. Gastrointestinal complications after cardiac surgery. Br J Surg 2005;92:326–33. [5] Filsoufi F, Rahmanian PB, Castillo JG, Scurlock C, Legnani PE, Adams DH. Predictors and outcomes of gastrointestinal complications in patients undergoing cardiac surgery. Ann Surg 2007;246:323–9. [6] Moneta GL, Misbach GA, Ivey TD. Hypoperfusion as a possible factor in the development of gastrointestinal complications after cardiac surgery. Am J Surg 1985;149:648–50. [7] Williams LF. Mesenteric ischaemia. Surg Clin North Am 1988;68:331–53. [8] Christenson JT, Schmuziger M, Maurice J, Simmonet F, Velebit V. Post operative visceral hypoperfusion the common cause for gastrointestinal complications after cardiac surgery. Thorac Cardiovasc Surg 1994;42: 152–7. [9] Mangi AA, Christison-Lagay ER, Torchiana DF, Warshaw AL, Berger DL. Gastrointestinal complications in patients undergoing heart operation: an analysis of 8709 consecutive cardiac surgical patients. Ann Surg 2005;241: 895–901. [10] Spotnitz WD, Sanders RP, Hanks JB, Nolan SP, Tribble CG, Bergin JD et al. General surgical complications can be predicted after cardiopulmony bypass. Ann Surg 1995;221:489–96. [11] Chaudhuri N, James J, Sheikh A, Grayson AD, Fabri BM. Intestinal ischaemia following cardiac surgery: a multivariate risk model. Eur J Cardiothorac Surg 2006;29:971–7.
