SurvBin

SurvBin implements two-sample nonparametric statistical tests to compare binary outcomes (proportions) and time-to-event outcomes (survival functions) without relying on parametric assumptions such as proportional hazards.

Key Features:

• Outcome types: Tests equality of proportions for binary outcomes and equality of survival functions for time-to-event data.
• Nonparametric approach: Operates without parametric model assumptions, accommodating departures from proportional hazards.
• Weighted combination statistics: Integrates a weighted combination of a score test for proportions and a Kaplan–Meier statistic–based test for survival functions.
• Customizable weights: Allows selection of weighting schemes to control the relative contribution of the binary and survival components.
• Asymptotic distribution and variance estimation: Provides asymptotic distribution results and a variance estimator to support inference under fixed and local alternatives.
• Performance evaluation: Validated via simulation studies with emphasis on small sample sizes.
• Implementation: Provided as an R package.

Scientific Applications:

• Oncology clinical trials: Comparison of treatment arms in trials such as randomized phase III cancer vaccine studies using both binary and time-to-event endpoints.
• Treatment efficacy assessment: Joint evaluation of proportions and survival functions to inform patient outcome analyses.

Methodology:

Constructs a weighted combination of a score test for proportions and a Kaplan–Meier statistic–based test for survival functions, supports customizable weights, derives asymptotic distributions with a variance estimator for inference under fixed and local alternatives, and evaluates performance via simulation studies focused on small sample sizes.