1. Hypothesis and purpose
This study is based on the hypothesis that normal preoperative cardiac biomarkers, lack of intraoperative hypotension and BIS values between 45-60 are associated with better outcomes compared with elevated biomarkers, intraoperative hypotension and/or BIS values below 45. Additionally, the investigators hypothesize that tight blood pressure management reduces the incidence of adverse outcomes compared to liberal blood pressure management (see below for definitions). Lastly, the investigators will test the hypothesis that individualized, relative lower limits of blood pressure are a better predictor of adverse outcomes than absolute, non-individualized limits.
2. Study design
The study will be comprised of a cohort study (RISK) and a nested randomized controlled trial (RCT) \[HEMODYNAMICS\]. In the period between September 2015 and August 2018 all patients meeting the inclusion criteria will be included in the cohort study to assess the association of perioperative biomarkers, intraoperative blood pressure and BIS values with the occurrence of MACE. A smaller subset of patients will additionally be included in the RCT to assess the effect of tight versus liberal blood pressure management on MACE.
Cardiac risk assessment and optimization will be performed according to clinical practice in the investigators' institution, following internal guidelines (based on the 2007 ACC/AHA guidelines. In addition, preoperative cardiac biomarkers (BNP and hsTnI) will be measured. Intraoperative blood pressure management will be based on an internal hemodynamic management algorithm, however in the cohort study the intervention threshold will be left at the discretion of the anesthesiologist. The occurrence of MACE will be assessed 30 days and one year after surgery. The association of preoperative biomarker levels and the occurrence of intraoperative hypotension will be investigated.
In the RCT, patients will be assessed similarly to the protocol of the cohort study, however these patients will be randomly assigned to either tight (mean arterial pressure \[MAP\] \> 75 mmHg) or liberal (MAP at the discretion of the anesthesiologist, but not \< 60mmHg) blood pressure management using the same hemodynamic management algorithm as in the whole cohort. In a subset of the RCT patients, a nighttime blood pressure profile will be obtained additionally to define a presumed individual lower blood pressure limit (i.e. the mean MAP during natural sleep). The association of intraoperative time spent below the individual lower blood pressure limit with the occurrence of adverse outcomes will be assessed.
Additionally, the association of postoperative parameters (e.g. the time spent below a given blood pressure threshold) with the occurrence of MACE will be investigated.
Further, the association of perioperative levels of the biomarkers midregional proadrenomedullin (MR-proADM) and high-sensitivity cardiac troponin T (hs-TnT) with the occurrence of MACE will be investigated.
Lastly, fully anonymised patient data from the BBB-Study may be included in multicenter cohort studies investigating perioperative myocardial ischemia and hemodynamic stability.
3. Sample Size
The sample size calculation of the Clinical Trials Unit St. Gallen indicates that in the RCT, assuming a 10% drop-out rate, 458 patients will be necessary to detect an a difference of 33% between the liberal and tight BP groups in our primary composite endpoint of MACE (α=0.05, β=0.2).
4. Statistical Analysis
In the RCT, potential differences between the incidence of the primary endpoint in the tight and liberal BP groups will be assessed using χ2 statistics. In the cohort study, the investigators will generate logistic regression models using the primary (or secondary) outcome as a dependent variable. The investigators will model the preoperative level of cardiac biomarkers (in quartiles and dichotomized), time spent below a prespecified mean arterial pressure (\< 75 mmHg, \< 60 and below the individual average mean arterial pressure during natural sleep) and time spent below a BIS \< 45. The model will be adjusted for the following prespecified independent variables: Charlson Comorbidity Index, Lee Revised Cardiac Risk Index, age and sex. The number of independent variables will be limited to avoid overfitting of the model. For dichotomization of BNP, the investigators will define the optimal cut-off value using receiver operating characteristics curves for the primary outcome under assumption of equal weight sensitivity and specificity. The null hypothesis will be considered refuted if the 2-sided p-value is \< 0.05.
Based on previous experience in similar studies, the investigators expect a very small number of loss of follow up (\< 2%). If the number exceeds 5% a Cox regression model using time to the composite endpoint (or time to last contact, if loss of follow up has occurred) as a dependent variable will be calculated. Missing perioperative BNP values will be handled as normal (\< 50 pg/ml). Any other missing perioperative laboratory values shall also be assumed to be normal.
