Mendelian randomization for causal inference of the relationship between obesity and 28-day survival following septic shock.
Septic shock is one of the most common causes of death in intensive care units. Septic shock occurs when a body-wide infection leads to low blood pressure, and ultimately organ failure. Some recent studies have suggested that overweight and obese patients have a better chance of survival following septic shock than normal or underweight patients. In this project we apply Mendelian randomization to assess whether the observed obesity effect on 28-day survival following septic shock is causal or more likely due to unmeasured confounding variables. Mendelian randomization is an instrumental variables approach that uses genetic markers as instruments. Roughly, the idea is to decompose obesity into a component caused by genetic and other known variables, and a residual component interpreted as obesity caused by unmeasured confounding variables. Under modelling assumptions, unconfounded estimates of the obesity effect can be obtained by fitting a model for 28-day survival that includes residual obesity as a model term. Data for the project comes from the Vasopressin and Septic Shock Trial (VASST), a multi-center clinical trial to evaluate the efficacy of vasopressin versus norepinephrine on mortality in patients with septic shock. The VASST data includes patient characteristics such as age, gender and body mass index (BMI), and genotypes at more than a million markers. Among these markers is a subset previously shown to predict BMI. We combine the BMI-influencing markers to form a genetic risk score and use this score as an instrument in an instrumental variables analysis. Our analysis suggests that the observed obesity effect on survival following septic shock is not causal.