Randomized, Embedded, Multifactorial Adaptive Platform Trial for Community- Acquired Pneumonia

Background: Respiratory tract infections that are of sufficient severity to require admission to hospital are associated with substantial mortality. All hospitalized patients with respiratory tract infections will receive therapy that consists of a combination of multiple different treatments. For many of these treatments, different options are available currently. For example, several antibiotics exist that are active against the microorganisms that commonly cause pneumonia but it is not known if one antibiotic strategy is best, or whether all suitable antibiotic strategies have similar levels of effectiveness. Of all the treatments that clinicians use for patients with respiratory tract infections, only a small minority have been tested in randomized controlled trials to determine their comparative effectiveness. As a consequence, the standard treatments that are administered vary between and within countries. Current conventional clinical trials methods to assess the efficacy of treatments for respiratory tract infections generally compare two treatment options (either two options for the same treatment modality, where both are in common use; or a new treatment against no treatment or placebo, where the effectiveness of the new treatment is not known). Using this approach, in a series of separate and sequential trials, it will take an inordinate length of time to study all the treatment options. Additionally, with conventional trial designs it is not possible to evaluate interactions between treatment options. Aim: The primary objective of this platform trial is, for patients hospitalized with respiratory tract infection, to identify the effect of a range of interventions to improve patient outcomes. Methods: The study will enroll hospitalized patients with respiratory tract infection using a trial design known as a REMAP, which is a type of platform trial. Within this platform trial, eligible participants will be randomized to receive one intervention in each of one or more domains. The primary outcome is a composite of 90-day all-cause mortality and, among survivors, a daily ordinal scale analyzed as a trajectory of illness and recovery to 28-days. There will also be both Platform and domain-specific secondary outcome measures. In a conventional trial enrollment continues until a pre-specified sample size is obtained, at which time enrollment ceases and the trial data are analyzed to obtain a result. The possible results are that a difference is detected, or no that no difference is detected. However, when the conclusion of the statistical test is "no difference", this could reflect that there truly is no meaningful difference or that the result is indeterminate (i.e., it is possible that if more patients had been enrolled a clinically relevant difference may have been detected). In comparison to a conventional trial this Platform uses an adaptive design that: * avoids indeterminate results by concluding an answer to a question when sufficient data have accrued (not when a pre-specified sample is reached); * utilizes already accrued data to increase the likelihood that patients within the trial are more likely to be randomized to treatments that are more likely to be beneficial; * is intended to be perpetual (or at least open-ended), substituting new questions in series as initial questions are answered; and * can evaluate the interaction between interventions in different domains, added at different timepoints into the Platform. Bayesian statistical methods will be used to define Statistical Triggers that reflect thresholds for decision making within a domain. While a limited number of initial treatments and treatment domains were specified at initiation, other interventions and domains have been evaluated over time, and it is planned that this platform trial will continue to evaluate other treatments in the future. Furthermore, in the event of a future epidemic, this Platform is able to adapt to evaluate the most relevant treatment options.