Título Vascular Events In Non Cardiac Surgery Patients Cohort Evaluation Study

TÍTULO Vascular events In noncardiac Surgery patIents Cohort evaluatioN study. The Vision study. INVESTIGADORES PRINCIPALES Dr. PJ Deveraoux INVESTIGADORES PRINCIPALES EN EL HOSPITAL GREGORIO MARAÑÓN: Dr. Ignacio Garutti Martinez Dra Patricia Cruz Pardos Dra. Carmen Fernández Riveira

INTRODUCCIÓN There are 2 major types of surgery. One is surgeries that are on the heart (cardiac surgeries) and the other is surgeries that are not on the heart (noncardiac surgeries, of which examples include bowel, lung, and bone surgeries). The noncardiac surgeries account for 95% of all surgeries performed and these surgeries are the focus of our study. Over the last 20 years, surgeries have made big advances in treating diseases like cancer and improving patient quality of life (for example, hip replacement). Concurrent with this development patients are living longer and they are now surviving to develop conditions that require a wide variety of noncardiac surgeries. As a result, the number of patients having surgery is growing. 100 million adults worldwide – have major noncardiac surgery every year. Despite the improvements in surgical procedures and anesthetic care some patients having surgery suffer major vascular events (that is: death, heart attack, cardiac arrest, and stroke). However, we do not know how many major vascular events are currently happening to patients having surgery. Further, we do not have an accurate tool to estimate a patients’ risk of such events Most patients having a heart attack after surgery will have their heart attack within 3 days of surgery. Most patients take pain drugs for the first 3 days after surgery. Some patients who have a heart attack during this time will not have chest pain because they are taking pain drugs. As a result of this, doctors may not detect all heart attacks after surgery. We do not know how many heart attacks around the time of surgery are missed because patients do not have chest pain. When a patient has a heart attack after surgery it is unknown what impact this will have on the patient’s health over the next year. A multicentre, international, prospective cohort study is the best design to address our study objectives as it will allow us to prospectively collect complete and accurate risk factor data, ensure patients have troponins measured, closely monitor patients for perioperative events, ensure unbiased and comprehensive ascertainment of events, and include a heterogenous group of patients who are > 45 years of age undergoing a broad spectrum of noncardiac surgeries in both university and nonuniversity hospitals. Our study has 4 objectives. The first objective is to learn how many patients have a major vascular event within 30 days of their surgery. Our second objective is to establish a tool that will allow us to accurately predict before surgery a patient’s risk of a major vascular event around the time of surgery. Our third objective is to learn

how many patients have a heart attack after surgery without symptoms and are at risk of being missed. The final objective of our study is to learn when a patient has a heart attack after surgery does this impact their health over the next year. Throughout the last few decades, noncardiac surgery has made substantial advances in treating diseases (e.g., cancer) and improving patient quality of life (e.g., arthroplasty). As a result, the number of patients undergoing noncardiac surgery is growing and worldwide an estimated 100 million adults undergo noncardiac surgery requiring hospital admission annually. Despite its benefits, noncardiac surgery is associated with adverse major vascular events (i.e., vascular death, nonfatal myocardial infarction, nonfatal cardiac arrest, and nonfatal stroke). For patients undergoing noncardiac surgery uncertainty exists regarding: the current incidence of major vascular events; the optimal clinical risk estimation model for predicting major vascular events; whether screening troponin measurements after surgery can allow physicians to avoid missing myocardial infarctions; and whether post surgical troponin measurements predict vascular mortality and major vascular events in the first year following surgery. We will undertake a large multicentre international cohort study to address these issues. - What is the incidence of major vascular events in patients undergoing noncardiac surgery? Because outpatient noncardiac day surgery is associated with a low risk of major complications (e.g., approximately 1 in 11,000 patients die within 30 days of ambulatory surgery) the focus of this study is on noncardiac surgery requiring hospital admission. The studies evaluating noncardiac surgery patients with, or at risk of, coronary artery disease or referred for a preoperative medical consult demonstrated event rates that ranged from 0.7 to 2.4% for vascular death, 0.7 to 5.6% for myocardial infarction, 0 to 1.5% for cardiac arrest, 0.1 to 1.6% for stroke, and 2.2 to 7.4% for major vascular event. Because these studies focused on select groups, they are limited in their capacity to inform the incidence of major vascular events in unselected adult patients. The only study that included relatively unselected patients suggests that major perioperative vascular events occur in 1.4% (95% CI 1.0-1.8%) of adults > 50 years of age undergoing elective noncardiac surgery with an expected length of stay > 2 days. Although this study provides the best estimate of the incidence of major vascular events in unselected adult patients undergoing noncardiac surgery requiring hospital admission, it suffers from important limitations. To inform this study we recently conducted the Vascular events In noncardiac Surgery patIents cOhort evaluatioN (VISION) Pilot Study. In this prospective cohort study we included patients undergoing noncardiac surgery who were > 45 years of age and receiving a general or regional anesthetic. The most common surgery was intraabdominal (50 patients) and 17 patients underwent surgery within 72 hours of an acute event. During the first 30 days after surgery 8% (95% CI 3-15) of the patients suffered a major vascular event (2 vascular deaths [1 due to stroke and 1 sudden death thought due to arrhythmia] and 6 nonfatal myocardial infarctions). Despite its potential limitations (e.g. small sample size), our pilot study suggests that the incidence of major perioperative vascular events may be increasing. Given that there are probably at least 50 million adults worldwide who are > 45 years of age undergoing noncardiac surgery requiring hospital admission annually a 2%, 2.5%, or 3% incidence of major vascular events would indicate that there are approximately 1, 1.25, or 1.5 million major perioperative vascular events annually. - What is the optimal clinical model to predict major vascular events in patients undergoing noncardiac surgery? Researchers have developed 2 types of clinical models – generic and Bayesian – to estimate perioperative cardiac risk in patients undergoing noncardiac surgery. The

generic risk models estimate a patient’s risk of a perioperative cardiac event through determination of how many predictors of risk (e.g., history of angina, diabetes, emergent surgery) an individual patient fulfills. The Bayesian risk models modify the average cardiac event rate for a specific surgery or group of comparable surgeries (pre-test probability) through use of a patient’s individual index score (likelihood ratio), which is based upon how many predictors of risk (e.g., history of angina, diabetes) an individual patient fulfills. The result provides an estimate of the patient’s risk of a perioperative cardiac event (post-test probability). The Revised Cardiac Risk Index is the simplest of the models to use and is more predictive than the Original Cardiac Risk Index. Considering the 2 Bayesian models the Veterans Affairs Model is simpler to use and the 1 study that compared these 2 models suggested the Veterans Affairs Model was as predictive as the Modified Cardiac Risk Index. Although several studies have compared the predictive accuracy of the generic and Bayesian risk models only 2 have used pretest probabilities based upon contemporary data from the hospitals studied at that time in the Bayesian models. The current predictive accuracy of the Modified Cardiac Risk Index and the Veterans Affairs Model is uncertain because there is no high quality study that has established contemporary complication rates for individual surgeries or groups of comparable surgeries, and it is unknown if contemporary complication rates at one institution are applicable to another institution. Until an accurate risk estimation model is established, physicians will remain limited in their capacity to undertake informed patient and physician decision-making regarding the appropriateness of noncardiac surgery. - Does the monitoring of troponins after noncardiac surgery help physicians to avoid missing myocardial infarctions? Several factors suggest the possibility that a substantial proportion of perioperative myocardial infarctions may go undetected if there is no monitoring of cardiac enzymes or troponins after surgery. First, the majority of perioperative myocardial infarctions occur during the first 3 days after surgery, a period during which most patients receive analgesic medication (e.g., narcotics) that can blunt cardiac pain perception. Second, a small but high-risk group of surgical patients will remain intubated and sedated during the highest risk period, limiting their ability to communicate symptoms. Third, surgical patients experiencing possible symptoms (e.g., shortness of breath, nausea) or signs (e.g., hypotension, tachycardia) of myocardial infarction have a host of more common potential explanations (e.g., atelectasis, pneumonia, hypovolemia, bleeding, incisional pain, medication side effect), and physicians may therefore not consider myocardial infarction. The available studies in prospective cohort studies suggest that only 16% (95% CI 631) of patients who suffered a perioperative myocardial infarction experienced chest discomfort. Approximately half (45%, 95% CI 29-62%) had neither chest pain nor any other symptoms or signs to suggest myocardial infarction. Although these studies offer encouraging evidence that the monitoring of cardiac enzymes or biomarkers after surgery may allow physicians to avoid missing a substantial proportion of myocardial infarctions, the studies are limited in that there were only a total of 38 myocardial infarctions, used CK-MB and they were restricted to patients with, or at high-risk for, coronary artery disease. If monitoring troponins after noncardiac surgery allows physicians to avoid missing perioperative myocardial infarctions, troponin screening will facilitate appropriate timely interventions. From a human and economic perspective, it is a tragedy if patients are undergoing noncardiac surgery to obtain a cure of their cancer or to become mobile after a new prosthetic hip but they fail to enjoy these benefits because they suffer an undetected myocardial infarction that ultimately takes their life. This point and previous research outlined above provide the impetus for a large

multicentre study of a heterogeneous group of patients undergoing noncardiac surgery to determine if monitoring troponins after surgery will avoid missing myocardial infarctions. - What is the relationship between postoperative troponin measurements and the 1 year risk of total mortality and major vascular events? Most available studies restricted patients to those undergoing vascular surgery or to those with, or at relatively high risk of, coronary artery disease. All studies were underpowered for the number of variables assessed in their analyses, and they did not include in multivariable analyses all the known intermediate to long-term predictors of death and major vascular events. Results demonstrated markedly varied associations with adverse outcomes, with wide confidence intervals. All but one study focused on total mortality. Although there is a strong rationale to support an association between an elevated perioperative troponin and vascular death, there is no plausible rationale to support an association between an elevated perioperative troponin and nonvascular mortality. If monitoring troponins after noncardiac surgery allows physicians to stratify patients according to their risk of vascular death or major vascular events over the subsequent year this will facilitate appropriate timely interventions and help to identify patients for future interventional randomized controlled trials. References: - Devereaux PJ, Goldman L, Yusuf S, Gilbert K, Leslie K, Guyatt GH. Surveillance and prevention of major perioperative ischemic cardiac events in patients undergoing noncardiac surgery: a review. CMAJ 2005;173(7):779-88. - Lee TH, Marcantonio ER, Mangione CM, et al. Derivation and prospective validation of a simple index for prediction of cardiac risk of major noncardiac surgery. Circulation 1999;100(10):1043-9. - Chan A, Livingstone D, Tu J. The Goldman and Detsky cardiac-risk indices: Do they work in patients undergoing hip-fracture surgery? Annals RCPSC 1999;32(6):337-341. - Fleisher LA, Beckman JA, Brown KA, et al. ACC/AHA 2007 guidelines on perioperative cardiovascular evaluation and care for noncardiac surgery: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2007;50(17):e159-241. F- leischmann KE, Goldman L, Young B, Lee TH. Association between cardiac and noncardiac complications in patients undergoing noncardiac surgery: outcomes and effects on length of stay. Am J Med 2003;115(7):515-20. - Filipovic M, Jeger R, Probst C, et al. Heart rate variability and cardiac troponin I are incremental and independent predictors of one-year all-cause mortality after major noncardiac surgery in patients at risk of coronary artery disease. J Am Coll Cardiol 2003;42(10):1767-76. - Devereaux PJ, Yusuf S, Yang H, Choi PT, Guyatt GH. Are the recommendations to use perioperative beta-blocker therapy in patients undergoing noncardiac surgery based on reliable evidence? CMAJ 2004;171(3):245-7. - Devereaux PJ, Goldman L, Cook DJ, Gilbert K, Leslie K, Guyatt GH. Perioperative cardiac events in patients undergoing noncardiac surgery: a review of the magnitude of the problem, the pathophysiology of the events and methods to estimate and communicate risk. CMAJ 2005;173(6):627-34.

This is a multicentre, international, prospective cohort study of 40,000 adults undergoing noncardiac surgery requiring hospital admission. Outcome adjudicators will adjudicate all major vascular events without knowledge of a patient’s vascular risk factors. JUSTIFICATIO FOR STUDY DESIGN We are currently conducting a 8,000 – 10,000 patient beta-blocker trial (the POISE Trial) in patients undergoing noncardiac surgery.34 Data from the POISE Trial or another randomized control trial (RCT) cannot address our objective regarding the incidence of major vascular events in adults undergoing noncardiac surgery because of the selective nature of patients who participate in trials and their restrictive eligibility criteria that target high or moderate to high-risk patients. Further, studies have shown limited performance of prediction models developed from RCT data when applied to general populations.35,36 An administrative database study is a suboptimal design to address our objectives for the following reasons: the lack of details on important prognostic factors;37 the inaccuracy of comorbidity data;38,39 the data distinguish only between events that happened pre and post admission but not pre and post surgery; the data do not capture whether patients experienced symptoms or signs of a myocardial infarction; the data do not capture which patients had troponins after surgery; and if monitoring troponins after surgery facilitates diagnosing myocardial infarctions, and given that troponin screening is unusual,40 then the administrative data are very likely to underestimate the incidence of perioperative myocardial infarctions. A multicentre, international, prospective cohort study is the best design to address our study objectives as it will allow us to prospectively collect complete and accurate risk factor data, ensure patients have troponins measured, closely monitor patients for perioperative events, ensure unbiased and comprehensive ascertainment of events, and include a heterogenous group of patients who are > 45 years of age undergoing a broad spectrum of noncardiac surgeries in both university and nonuniversity hospitals. OUTCOMES For our first objective (i.e., the incidence of major perioperative vascular events) our primary outcome is major vascular events (i.e., a composite of vascular death, nonfatal myocardial infarction, nonfatal cardiac arrest, and nonfatal stroke) at 30 days post surgery. Individual secondary outcomes for our first objective at 30 days after surgery include congestive heart failure, new clinically important atrial fibrillation, and deep venous thrombosis, and at 30 days and 1 year after surgery will include total mortality, vascular mortality, myocardial infarction, nonfatal cardiac arrest, stroke, cardiac revascularization procedures, pulmonary embolus, and rehospitalization for vascular reasons. For our second objective (i.e., the optimal clinical model to predict major perioperative vascular events) our primary and only outcome is major vascular events at 30 days after surgery. For our third objective (i.e., the proportion of patients with perioperative myocardial infarctions that may go undetected without perioperative troponin monitoring) our primary outcome at 30 days post surgery is myocardial infarction that probably would have gone undetected without perioperative troponin monitoring (those without chest discomfort or other symptoms or signs suggesting myocardial infarction), and our secondary outcome is myocardial infarction that possibly would have gone undetected without perioperative troponin monitoring (no chest discomfort). For our fourth objective (i.e., the relationship between postoperative troponin measurements and the 1 year risk of vascular death) our primary outcome is vascular death at 1 year after surgery, and our secondary outcome is major vascular events (i.e., vascular death, nonfatal

myocardial infarction, nonfatal cardiac arrest, and nonfatal stroke) at 1 year after surgery. Appendix I provides definitions for all outcomes. STUDY POPULATION A total of 40,000 patients undergoing noncardiac surgery will participate.Inclusion criteria: all patients who undergo noncardiac surgery are eligible if they are > 45 years of age and receive a general or regional anesthetic (i.e., plexus block, spinal, or epidural). We will include patients for whom we can obtain consent within the first 24 hours after their surgery. Exclusion criteria: we will exclude patients undergoing noncardiac surgery who do not require at least an overnight hospital admission after surgery or who only receive infiltrative (i.e., local) or topical anesthesia. Screening and Enrolment of Patients: through simple eligibility criteria, screening procedures, and enrolment procedures, we have ensured that investigators will require minimal effort to recruit patients. Once a patient has consented and undergone surgery they are considered enrolled in the VISION study. VARIABLES COLLECTED AT THE TIME OF ENROLMENT after obtaining written informed consent from eligible patients or their family members, research personnel will interview and examine patients and review their charts to obtain information on patient characteristics that are potential predictors of major perioperative vascular events. The VISION study will evaluate patient characteristics that were associated with major perioperative vascular events in prior studies and patient characteristics that are clinically sensible. The preoperative patient characteristics that study personnel will record include and their definitios are detailed in Appendix I. MONITORING, FOLLOW-UP AND DATA COLLECTION We have kept the follow-up procedures simple and to a minimum. All patients will have a troponin T drawn sometime between 6 to 12 hours postoperatively and on the 1st, 2nd, and 3rd day after surgery. Standard orders will ensure these tests are undertaken. Orders will also ensure that an ECG is undertaken immediately after an elevated troponin measurement is detected. If a troponin measurement is elevated but the patient has no ECG changes, ischemic symptoms, or pulmonary edema to fulfill the diagnostic criteria for myocardial infarction, then the patient will undergo an echocardiographic study. Details to ensure patient enrolment and protocol adherence are provided in Appendix II. SAMPLE SIZE Our sample size calculation is based upon our second primary objective (i.e., to determine the optimal clinical model to predict major perioperative vascular events among patients undergoing noncardiac surgery). Of the 4 primary hypotheses, this is the 1 that requires the largest number of patients to ensure the stability of the prediction model. To determine the optimal clinical model we will undertake a multivariable analysis. Simulation studies demonstrate that logistic models require 12 to 15 events per predictor to produce stable estimates.22,47 We will evaluate 55 predictors in our multivariable analysis: 16 preoperative patient characteristics (Appendix II); 24 groups of comparable surgeries with 1 reference low-risk surgical group (Appendix III); 2 types of surgical categories (emergent and urgent surgery) with 1 reference category (elective surgery) (Appendix IV); and 13 centre variables (i.e., to assess centre effect, 1 variable for each of 13 centres with 1 reference centre). Table 9 presents the number of variables possible to test in a multivariable analysis based upon various sample sizes, event rates, and the required number of events per variable. The shaded cells identify cells that meet or exceed the number of variables we will evaluate in our multivariable analysis. We propose a sample size of 40,000 patients that will ensure a stable model as we will have over 12 events per variable in our multivariable analysis based upon our anticipated 2.0% event rate.

STATISTICAL ANALYSES PLAN For our first primary objective we will determine the proportion of patients suffering a major perioperative vascular event and separately the individual secondary outcomes and the associated 95% confidence intervals. For our third primary objective we will report the proportion of patients suffering a perioperative myocardial infarction that probably and separately possibly would have gone undetected without perioperative troponin monitoring and the associated 95% confidence intervals. We will also construct time to event curves for these outcomes. To address our second objective we will undertake a multivariable logistic regression analysis to develop a generic model in which the dependent variable is a major perioperative vascular event at 30 days after surgery and the independent variables are the 16 preoperative patient characteristics (Appendix II), the 25 groups of comparable surgeries (Appendix III), and the 3 types of surgical categories (Appendix IV). For all logistic regression analyses we will use forced simultaneous entry (all candidate variables will remain in the model) as opposed to automated stepwise selection, because simulation studies have demonstrated a higher risk of overfitting with the latter approach.48,49 We will repeat the first multivariable logistic regression analysis but add an independent variable for each centre to assess if there is a centre effect. If we find a centre effect, we will then generate a Bayesian model based upon the average centre pretest probabilities for each of the 25 groups of comparable surgeries and a risk index based upon the first multivariable logistic regression analysis without the 25 groups of comparable surgeries. To assess the reliability of our models we will undertake bootstrapping,50 because this technique is superior to cross-validation and jack-knife techniques.51 We will compare our models to the Revised Cardiac Risk Index11 and the Veterans Affairs Model8 using the area under the receiver operating characteristic curve.52 To address our fourth objective we will undertake a Cox proportional hazard regression analysis in which the dependent variable is time to vascular death during the first year after surgery and the independent variables are the 16 preoperative patient characteristics (Appendix II) and the peak post surgical troponin value. We will repeat this analysis changing the dependent variable to major vascular events. For regression models, we will report the odds ratios [OR] (for logistic regression) or hazard ratio [HR] (for Cox proportional hazard regression), corresponding standard error, 95% confidence intervals and associated p-values. We will report P-values to 3 decimal places with p-values less than 0.001 reported as p<0.001. For all tests, we will use alpha = 0.05 level of significance.