Predictors of Outcome After Perioperative Stroke in the Non-Cardiac Surgery Population

Summary

Participation Requirements

Description

BACKGROUND Perioperative stroke is a devastating complication of surgery that is currently poorly characterized with limited clinical tools available to detect and prevent its occurrence. Perioperative stroke is a cerebrovascular event that occurs after surgery, and affects between 0.1-1.9% of patie...

BACKGROUND Perioperative stroke is a devastating complication of surgery that is currently poorly characterized with limited clinical tools available to detect and prevent its occurrence. Perioperative stroke is a cerebrovascular event that occurs after surgery, and affects between 0.1-1.9% of patients having non-cardiac, non-neurologic surgery. Perioperative stroke is relatively understudied compared to postoperative complications of similar incidence and severity, such as cardiac complications. Over the past few decades, significant efforts have been undertaken to reduce the risk of perioperative myocardial infarction and have led to a decrease in incidence over time due to an advancements in risk stratification and perioperative management. Although the incidence of perioperative stroke is similar or higher than that of perioperative myocardial infarction, the risk of this complication has been increasing over time and perioperative stroke remains relatively neglected. The current literature has identified that patients who experience a stroke after surgery have a higher rate of mortality, length of stay and discharge to a facility, but given the rare nature of this complication relatively little is known about which factors predict these outcomes amongst those who experience a perioperative stroke. OBJECTIVES The primary study objective is to identify predictors of mortality after perioperative stroke in non-cardiac, non-neurological surgery. The secondary objectives are to 1) create a predictive model for length of stay after perioperative stroke; 2) create a predictive model for discharge to an alternate facility after perioperative stroke; 3) 3) Describe the temporal trend in mortality after perioperative stroke over time (2004 to 2015); and 4) explore mediators of mortality risk after perioperative stroke in non-cardiac, non-neurological surgery. METHODS This study is a retrospective analysis of the prospectively-collected American College of Surgeons National Surgical Quality Improvement Program database between 2004 and 2017. The primary outcome is mortality. The secondary outcomes are hospital length of stay and discharge to an alternate facility. Outcome after perioperative stroke is potentially related to patient, surgical and anesthetic factors, as well as characteristics of the stroke. The candidate predictors variables will include patient characteristics (demographics, BMI, comorbidities including history of stroke), surgical characteristics (specialty, complexity, type, emergency status, transfusion), characteristics of the stroke (timing relative to the operation, unplanned reoperation after stroke, readmission for stroke vs inpatient stroke), and anesthetic technique used. History of TIA or CVA is available only until 2015, therefore our multivariable analysis will exclude years 2016 and 2017 of the dataset given the clinical importance of this variable. SAMPLE SIZE ESTIMATION The sample size will be determined by the number of perioperative stroke events in NSQIP database. It is estimated that there will be a total number of 6.6 million patients in the NSQIP database between 2004 and 2017, of which 80% will be non-cardiac surgery. Assuming a 0.1% overall perioperative stroke risk, we estimate a total of approximately 5000 cases of perioperative stroke will be included in the dataset. As only years 2004 through 2015 will be included in the prediction modeling, it is estimated that approximately (0.8*0.001*4,608,309) 3700 cases will be available for this portion of the analysis. The 30-day mortality rate in a prior study using the NSQIP database was estimated to be as high as 25% in patients who experience a stroke,(2) therefore it is anticipated that a sufficient number of outcome events (n=900) to fit and validate a stable multivariable regression model (i.e., we anticipate >10 events per covariate). STATISTICAL ANALYSIS Data will be described using frequency, mean (standard deviation [SD]) or median (interquartile range [IQR]). For predictive modelling, all continuous variables will be standardized (i.e., centered with mean=0 and SD=1). To avoid over-fitting our model, a data reduction strategy will be undertaken and variables with greater than 10% missing data or less than 20 observations will be excluded. Where >1% but <10% data are missing we will consider multiple or mean imputation; variables with <1% missing data will be handled through complete case analysis. Pre-specified predictor variables will be used to construct a logistic regression model where coefficients will be shrunken using elastic net penalization.(11) This method for variable selection and penalization was chosen based on our modelling scenario. A large number of potential predictors relative to the number of outcomes may occur. In addition, several important predictors will be collinear. Elastic net regularization can be used to employ a balance between least absolute shrinkage and selector operator (LASSO) penalization and RIDGE regression. RIDGE regression will allow us to keep important collinear predictors, but shrink their coefficients (which could be inflated due to collinearity). LASSO methods will allow the coefficients of some non-contributory predictors to zero, therefore, eliminating them from the model. Included variables will be assessed for interactions and addition terms considered to improve model performance as needed. The a priori interactions to be examined will include the following: age*gender; surgical complexity (WRVU)*age; WRVU*specialty; specialty*anesthetic technique. Model discrimination will be evaluated using the area under the receiver operating characteristic curve (C-statistic). Model calibration will be assessed with a loess-smoothed plot of observed vs predicted risks over the risk spectrum. A similar analysis will be used to create a prediction model for our secondary outcomes, length of stay (which will be a log-transformed linear model) and discharge to an alternate facility (logistic regression). Following derivation, 5000 bootstrap samples will be used to perform internal validation and generate an estimated optimism. To assess the impact of year of surgery, a sensitivity analysis of model performance over different time epochs will be performed (ie, 2-3 year epochs, depending on the number of observations available in each year). Following derivation of the final model to predict mortality, a simplified predictive index will be derived by converting the regression coefficients to points that reflect their relative weights. The predictive accuracy of the resulting tool will be assessed in a similar fashion to above. To address secondary objective 3, the annual mortality rate after perioperative stroke will be described using a scatter plot. We will also create an exploratory univariate ordinary least squares regression model with annual mortality rate as the dependent variable and year as the predictor to estimate the yearly change in mortality rate over time. A further multivariable adjusted linear regression model will be specified adjusting for important predictors. To explore possible effect modification of pre-stroke mortality risk by post-stroke events, a series of logistic regression models with death as the dependent variable and the probability of death from our primary model as a linear predictor will be specified. We will then add pre-specified effect modifiers (as above), each as an additional predictor variable, to explore whether the probability of death increases or decreases based on addition of the postulate effect modifying variable. A p value <0.05 will be considered significant for all analyses and all data analysis will be performed using STATA 15 (StataCorp, Texas, USA).

Tracking Information