Optimality in Multivariate Tie-breaker Designs

\begin{abstract} Tie-breaker designs (TBDs), in which subjects with extreme values are assigned treatment deterministically and those in the middle are randomized, are intermediate between regression discontinuity designs (RDDs) and randomized controlled trials (RCTs). TBDs thus provide a convenient mechanism by which to trade off between the treatment benefit of an RDD and the statistical efficiency gains of an RCT. We study a model where the expected response is one multivariate regression for treated subjects and another one for control subjects. For a given set of subject data we show how to use convex optimization to choose treatment probabilities that optimize a prospective $D$-optimality condition (expected information gain) adapted from Bayesian optimal design. We can incorporate economically motivated linear constraints on those treatment probabilities as well as monotonicity constraints that have a strong ethical motivation. Our condition can be used in two scenarios: known covariates with random treatments, and random covariates with random treatments. We find that optimality for the treatment effect coincides with optimality for the whole regression, and that the RCT satisfies moment conditions for optimality. For Gaussian data we can find optimal linear scorings of subjects, one for statistical efficiency and another for short term treatment benefit. We apply the convex optimization solution to some real emergency triage data from MIMIC.