Safety of esophagogastroduodenoscopy within 30 days of myocardial infarction : A retrospective cohort study from a Canadian tertiary centre

St Joseph’s Healthcare, McMaster University, Hamilton, Ontario. *Co-first authors Correspondence: Dr Khurram Khan, St Joseph’s Healthcare, Room H424, 50 Charlton Avenue East, Hamilton, Ontario L8N 4A6. Telephone 905-521-2100 ext 76768, fax 905-521-4958, e-mail khankj@mcmaster.ca Received for publication May 18, 2011. Accepted July 9, 2011 Esophagogastroduodenoscopy (EGD) is a safe, well tolerated and useful diagnostic and therapeutic procedure in the general population. Gastrointestinal (GI) bleeding is a common side effect of many antiplatelet/anticoagulation/thrombolysis drugs used in the management of ischemic heart disease, especially in the modern era of cardiology, in which patients are on multiple drugs that can increase the risk of bleeding (1-3). In specific high-risk patients, such as those with a recent myocardial infarction (MI), the safety of routine upper endoscopy is not well known but has been described as acceptable in several observational cohort studies, with complication rates ranging up to 7.5% (4-7). In other studies, there are significant differences in mortality in acute coronary syndrome (ACS) patients who experience GI bleeding, suggesting they are a higher-risk population (8,9). EGD is an important diagnostic tool in up to 85% of patients who experience post-MI bleeding (6); is generally safe in high-risk patients and can shorten hospitalization (10). Early endoscopy (<24 h) is recommended in most patients, and risk stratification is achieved with measures such as the Blatchford or Rockall score (11-13). However, post-MI patients may also be more susceptible to cardiopulmonary complications including recurrent myocardial ischemia, hypoxia, originAl ArtiCle

anxiety or discomfort, hypotension and aspiration due to sedation or the procedure itself.Some earlier studies demonstrated a risk of cardiopulmonary complications that is highest immediately after a myocardial event, and more common in sicker patients based on acute physiology and chronic health evaluation criteria (5)(6)(7).
The present cohort study was designed to capture the cardiopulmonary safety concerns around post-MI endoscopy, and to evaluate specific predictors of complications such as age, hemoglobin (Hgb) level and time from MI to endoscopy.

MEthodS
The study was conducted at the Hamilton Health Sciences Hospitals (McMaster University Medical Centre, Hamilton General Hospital, and Henderson Hospital) in Hamilton, Ontario, and approved by the hospitals' research ethics boards.Patients admitted between 2003 and 2006 were identified retrospectively by computerized analysis of primary or secondary medical disease codes (International Classification of Diseases -9th revision) for myocardial infarction and upper GI bleeding, and by procedure codes for EGD.
Inclusion criteria for the study were as follows: adult patients (age >18 years) with ACS (including ST elevation myocardial infarction [STEMI] and non-STEMI [NSTEMI] with documented elevation in troponin level) who underwent EGD for suspected GI bleeding within 30 days of an incident acute coronary event.Reasons for exclusion included troponin elevation for renal failure, upper GI bleed before MI, EGD before MI and/or misdiagnosis.
Individual patient charts that met the inclusion criteria were carefully reviewed, and data relevant to the study were extracted using a standardized data extraction form.This included baseline demographics of age, sex, medical history (hypertension, diabetes, congestive heart failure, stroke, chronic lung disease, bleeding disorder, renal failure, MI, GI bleeding or peptic ulcer disease), medication use before endoscopy (eg, acetylsalicylic acid, clopidogrel, heparin, warfarin, beta-blockers, nonsteroidal anti-inflammatory drugs, proton pump inhibitors), details surrounding the endoscopy (indication, medications, complications, vital signs, diagnosis) and any adverse cardiopulmonary outcomes including significant cardiac arrhythmias, congestive heart failure, respiratory failure, hypoxemia (oxygen saturation <85%), recurrent ACS, significant hypotension (mean arterial pressure <70 mmHg, or in patients hypotensive at baseline -a drop in mean arterial pressure of 15%) or hypertension, aspiration pneumonia and death.
Statistical analysis was performed using SPSS version 19 (IBM Corporation, USA).Categorical variables were analyzed using Fisher's exact test or the c 2 test.Continuous variables (mean ± SD) were analyzed using the Student's t test or the Mann-Whitney test, as appropriate.ORs and 95% CIs were calculated from stepwise logistic regression analysis of the a priori predictors for cardiopulmonary complications.

rESultS
During the four-year study period, a total of 121 patient charts describing a diagnosis of an MI with an EGD within 30 days were reviewed.However, after careful analysis, only 44 patients met the inclusion criteria.
The mean age of study patients was 73.5 years, and 55% were female.Most patients had at least one comorbidity, and one-quarter experienced a previous GI bleed.The median Rockall score was 5. Almost 80% of the patients were taking an antiplatelet agent at the time of the bleed (Table 1).The average Hgb level was 86 g/L at the time of EGD, and 86% of patients required a transfusion (Table 2).The most common indication for EGD was melena in 55% of cases (Table 2).More than 50% of patients underwent EGD within the first three days post-MI.The median number of days to endoscopy after MI was three with a range of zero to 31 days.
EGD was diagnostic for the cause of bleeding; the most common diagnoses were duodenal or gastric ulcer, erosions, varices and esophageal ulcers.Complications occurred in 12 (27.5%)patients, with recurrent ACS being the most common in seven of the 12 (Table 3).Other complications included arrhythmia, respiratory failure, aspiration pneumonia and perforation.Logistic regression was used to determine whether age, Hgb level or days since MI (more than three days or less than three days) were important in predicting which patients would experience a complication.None of the factors were significant (age [P=0.41];Hgb level [P=0.65]; or days since MI [P=0.81]).

diSCuSSion
The present study showed that patients who required an EGD after a recent (<30 days) MI demonstrated higher rates of serious complications (27.5%) than previously reported in similar retrospective cohort studies (4-7).However, there were no deaths, whereas other studies reported mortalities in this population (8,9).The reason for the increased complication rate is unclear; most were recurrent ACS possibly due to antiplatelet medications being discontinued for active bleeding.Other explanations may relate directly to endoscopy because some experiments have shown that gastric insufflation and vagally mediated reflexes could trigger ischemia (14)(15)(16).
According to some studies, the highest risk of complication is in the first 24 h post-MI (7); however, in our cohort, we divided the patients into those who underwent EGD within three days and after three days of an MI.The logistic regression model adjusted for Hgb level and age did not show a significant impact of the timing of the EGD.Units of blood, mean ± SD 4.6±3.5

Data presented as n (%) unless otherwise indicated
Another explanation for the higher complication rate is that patients in the study had many comorbidities and up to 80% were treated with an antiplatelet agent.The baseline Hgb level was also quite low (86 g/L), and most patients (86%) required transfusion.The average Rockall score was 5, but was skewed toward the higher values, which suggest there was at least an intermediate risk of rebleeding following endoscopy.EGD proved to be an important diagnostic tool in risk stratification for upper GI bleeding because most patients had evidence of GI pathology.Patients not selected for endoscopy were likely even sicker, but as newer endoscopic hemostasis techniques are more widely used, the modern endoscopist is likely better equipped and, hence, possibly more willing to perform EGD in patients with severe comorbidities.International consensus recommendations on the management of patients with nonvariceal upper GI bleeding recently published by Barkun et al (13) suggest that EGD should not be deferred for more than 24 h except in certain high-risk patients, such as those with ACS or a perforation; however, no specific time frame is mentioned.The utility of deferring EGD is still controversial in patients with a concomitant ACS and GI bleed.A retrospective study by Lin et al (10) determined that in 43% of cases, an urgent EGD influenced the timing of cardiac catheterization.Yachimski et al (17) performed a decision analysis in which they proposed that an urgent EGD before cardiac catheterization was safer in patients with an ACS and overt GI bleeding.

Study limitations
The present study was retrospective in nature and, thus, had some bias in its design.There is selective reporting of only patients who underwent EGD for GI bleeding within 30 days of an MI, but no collection of data on patients with a GI bleed who did not undergo endoscopy.Our study focused on the safety of EGD and peri-MI complications, and investigated an a priori question of whether the timing of endoscopy is related to complications (7).Data were collected, but specific criteria assessing the severity of the bleed and timing of bleeding symptoms to endoscopy were not assessed.Outcomes were subjectively assessed based on documentation in the clinical record.Definitive conclusions could not be drawn using a logistic regression model because of the small sample size.

Clinical implications
The current study demonstrated that EGD was associated with a substantial risk of complication post-MI, which may have been underestimated in previous studies.Reasons for the higher rates of complications are not clear.The patients may be sicker, and/or the more aggressive use of anticoagulation/antiplatelet agents for an older population in the modern era of cardiology is a plausible explanation.The decision to perform EGD in a sick patient will always be difficult, but the goal of reducing bleeding and hypoperfusion by endoscopically identifying and treating a bleeding lesion must remain paramount.Several other studies similarly conclude that endoscopic evaluation using flexible sigmoidoscopy, colonoscopy, endoscopic retrograde cholangiopancreatography and even percutaneous endoscopic gastrostomy tube placement, is safe in patients post-MI (18-21).When necessary, alternative options, such as wireless capsule endoscopy, may replace many indications for diagnostic EGD because of higher sensitivity, less procedural invasiveness and greater patient safety (22).

TAblE 2 Endoscopy indications and transfusion requirements
18. Cappell MS.Endoscopic retrograde cholangiopancreatography with endoscopic sphincterotomy for symptomatic choledocholithiasis after myocardial infarction.Am J Gastroenterol 1996;91:1827-31.19.Cappell MS.Gastrointestinal bleeding associated with myocardial infarction.Gastroenterol Clin North Am 2000;29:423-44.20.Cappell MS.The safety and clinical utility of flexible sigmoidoscopy and colonoscopy for gastrointestinal bleeding after myocardial infarction: A six-year study of 18 consecutive lower endoscopies at two university teaching hospitals.Dig Dis Sci 1994;39:473-80.21.Cappell MS, Iacovone FM Jr.The safety and efficacy of percutaneous endoscopic gastrostomy after recent myocardial infarction: A study of 28 patients and 40 controls at four university teaching hospitals.Am J Gastroenterol 1996;91:1599-603.22. Saurin JC, Delvaux M, Gaudin JL, et al.Diagnostic value of endoscopic capsule in patients with obscure digestive bleeding: