Search

We consider the conditional randomization test as a way to account for covariate imbalance in randomized experiments. The test accounts for covariate imbalance by comparing the observed test statistic to the null distribution of the test statistic conditional on the observed covariate imbalance. We prove that the conditional randomization test has the correct significance level and introduce original notation to describe covariate balance more formally. Through simulation, we verify that conditional randomization tests behave like more traditional forms of covariate adjustment but have the added benefit of having the correct conditional significance level. Finally, we apply the approach to a randomized product marketing experiment where covariate information was collected after randomization.

This pilot study adapted family-based treatment (FBT) for youth with potentially prodromal anorexia nervosa (AN). Fifty-nine youth with clinically significant AN symptom constellations, but who never met full Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.) (DSM-IV) criteria for AN, were enrolled in a partially randomized preference design study. Participants were offered randomization to FBT or supportive psychotherapy (SPT); those who declined to be randomized because of a strong treatment preference were entered into a parallel, non-randomized self-selected intervention study. Without accessing outcome data, an observational analysis with three diagnostic subclasses was designed based on AN symptom severity profiles, combining randomized and non-randomized participants, such that participants receiving FBT and SPT within each subclass were similar on key baseline characteristics. Outcomes of this pilot study were explored by calculating effect sizes for end-of-treatment values within each subclass, and also with a longitudinal mixed effect model that accounted for subclass. Weight trajectory was measured by percent expected body weight. Psychological outcomes were fear of weight gain, feeling fat, importance of weight, and importance of shape. Results show that the pattern of symptom observations over time was dependent on subclass of SAN (least symptomatic, moderately symptomatic, or most symptomatic) and on the target outcome variable category (weight or psychological). Results from this study, which should be considered in the context of the small sample sizes overall and within groups, can generate hypotheses for future, larger research trials on early treatment strategies. Feasibility findings illustrate how the innovative partially randomized preference design has potential broader application for AN intervention research.Clinical trial registrationClinicalTrials.gov, identifier NCT00418977.

The colonization of the human gut microbiome begins at birth, and over time, these microbial communities become increasingly complex. Most of what we currently know about the human microbiome, especially in early stages of development, was described using culture-independent sequencing methods that allow us to identify the taxonomic composition of microbial communities using genomic techniques, such as amplicon or shotgun metagenomic sequencing. Each method has distinct tradeoffs, but there has not been a direct comparison of the utility of these methods in stool samples from very young children, which have different features than those of adults. We compared the effects of profiling the human infant gut microbiome with 16S rRNA amplicon vs. shotgun metagenomic sequencing techniques in 338 fecal samples; younger than 15, 15–30, and older than 30 months of age. We demonstrate that observed changes in alpha-diversity and beta-diversity with age occur to similar extents using both profiling methods. We also show that 16S rRNA profiling identified a larger number of genera and we find several genera that are missed or underrepresented by each profiling method. We present the link between alpha diversity and shotgun metagenomic sequencing depth for children of different ages. These findings provide a guide for selecting an appropriate method and sequencing depth for the three studied age groups. [ABSTRACT FROM AUTHOR]