INCEPT: The Intensive Care Platform Trial-Design and protocol

Abstract

Background: Adult intensive care unit (ICU) patients receive many interventions, but few are supported by high-certainty evidence. Randomised clinical trials (RCTs) are essential for trustworthy comparisons of intervention effects, but conventional RCTs are costly, cumbersome, inflexible, and often turn out inconclusive. Adaptive platform trials may mitigate these issues and have higher probabilities of obtaining conclusive results faster and at lower costs per participant.

Methods: The Intensive Care Platform Trial (INCEPT) is an investigator-initiated, pragmatic, randomised, embedded, multifactorial, international, adaptive platform trial including adults acutely admitted to ICUs. INCEPT will assess comparable groups of interventions (primarily commonly used interventions with clinical uncertainty and practice variation) nested in domains. Interventions may be either open-label or masked. New domains will continuously be added to the platform. INCEPT assesses multiple core outcomes selected following substantial stakeholder involvement: mortality, days alive without life support/out of hospital/free of delirium, health-related quality of life, cognitive function, and safety outcomes. Each domain will use one of these core outcomes as the primary outcome. INCEPT primarily uses Bayesian statistical methods with neutral, minimally informative or sceptical priors, adjustment for important prognostic baseline variables, and calculation of absolute and relative differences in the intention-to-treat populations. Domains and intervention arms may be stopped for superiority/inferiority, practical equivalence, or futility according to pre-specified adaptation rules evaluated using statistical simulation or at pre-specified maximum sample sizes. Domains may use response-adaptive randomisation, meaning that more participants will be allocated to interventions with higher probabilities of being superior.

Conclusions: INCEPT provides an efficient, pragmatic, and flexible platform for comparing the effects of many interventions used in adult ICU patients. The adaptive design enables the trial to use accumulating data to improve the treatment of future participants. INCEPT will provide high-certainty, conclusive evidence for many interventions, directly inform clinical practice, and thus improve patient-important outcomes.

Keywords: adaptive platform trial; intensive care; randomised clinical trial; trial protocol.

FIGURE 1

Multiple domains in an adaptive platform trial, with the domains and interventions reflecting the tentative first domains and interventions on INCEPT. Each domain consists of comparable interventions (similar to what could be compared in a single stand‐alone trial), and all domains use the same overall platform trial infrastructure. All domains use the same overall platform eligibility criteria, and the overall eligible population thus consists of adult, acutely admitted ICU patients. Additional domain‐specific inclusion criteria further narrow down this population, for example, the albumin domain will be restricted to patients with shock, the thromboprophylaxis domain to patients with an indication for thromboprophylaxis with LMWH, and the glucose domain to patients with an insulin requirement for blood glucose management. Randomisation to the interventions in each domain is separate. Collection of data, including outcome data, is joint across domains. Based on adaptive (interim) analyses of outcome data, domains will adapt by changing the available interventions and domains and by adapting allocation probabilities, as described in the text and full core protocol available at the trial website. Domains will be stopped once there is sufficiently strong evidence for a conclusion according to pre‐defined criteria or at a pre‐specified maximum sample size. Additional details on all INCEPT domains will be specified in domain‐specific appendices to the core protocol. At the time of writing, the albumin domain has been approved, while the other domains illustrated are under preparation and have not yet been submitted for approval. CGM: continuous glucose monitoring; ICU: intensive care unit; INCEPT: the Intensive Care Platform Trial; LMWH: low‐molecular weight heparin.