Developing Optimal Design Theory for Adaptive Randomised Clinical Trials in Resource-Limited Nigerian Hospital Settings

Adaptive randomised clinical trials that modify design features including allocation ratios and sample sizes based on accumulated data offer substantial statistical and ethical advantages in Nigerian hospital settings where patient enrollment is slow, resources are limited, and ethical imperatives to minimise patient exposure to inferior treatments are strong. This dissertation develops optimal design theory for adaptive randomised clinical trials in resource-limited Nigerian hospital settings, making original contributions to adaptive design theory. A unified Bayesian-frequentist framework for adaptive allocation and early stopping design is developed, with theoretical guarantees on type I error control, power, and expected sample size under Nigerian hospital enrollment rate distributions. New optimality conditions for response-adaptive randomisation under ethical and regulatory constraints are derived, extending existing Urn model optimality theory. Sample size re-estimation rules with type I error control are derived for two-stage adaptive designs under non-normal primary endpoints common in Nigerian clinical data. Optimal designs are derived for three Nigerian clinical contexts: malaria treatment non-inferiority trials, tuberculosis treatment time-to-negativity superiority trials, and malnutrition management equivalence trials. Theoretical analysis shows proposed adaptive designs require 25 to 38 percent fewer patients than fixed designs for equivalent power in the malaria and TB contexts. Simulation experiments verified type I error control at the 0.05 nominal level across all scenarios. The dissertation provides Nigerian Clinical Trial Registry-compatible adaptive trial design guidelines for Nigerian academic hospital research units. Keywords: adaptive clinical trial design, response-adaptive randomisation, optimal design theory, Nigerian hospitals, statistical methodology