Explanation

The study objective reflects the scientific question to be answered by the trial and defines its purpose and scope, with profound implications for many other aspects of the trial, e.g., the design (Item 12) and analysis (Item 27).

The "PICO" format to formulate the trial objective is often used. This entails defining the participant population (P); intervention (I); comparator (C); and outcomes (O) of main interest.(165) PICO is sometimes styled as PICOTS to include the timeframe and setting.

Protocol authors should further report whether the intention of the trial is to evaluate potential benefits and harms,(166) and whether the aim is to assess superiority of the intervention, or non-inferiority/equivalence,(167) versus the comparator. For example, the sample size calculation and statistical analyses for superiority trials will differ from those investigating non-inferiority.

Authors should also indicate which treatment effect they are planning to investigate (e.g., the effect of assignment to the intervention irrespective of adherence, or the effect of adhering to the intervention (Box 1)). They should also report whether the trial is intended to provide preliminary data (a pilot or feasibility trial) or confirmatory results.

If authors are planning a trial that involves re-adjusting the objective during the trial, for example in some platform trials or basket trials(168, 169), this should be reported. Trials can be designed to study the effect of the intervention under different conditions, often described on a spectrum from ideal conditions (explanatory trial) to routine clinical care conditions (pragmatic trial).(170)

The objectives should generally be phrased using neutral wording (e.g., “to compare the effect of treatment A versus treatment B on outcome X for persons with condition Y …”) rather than in terms of a particular direction of effect.(171) For multi-arm trials, the objectives should clarify which treatment group comparisons are of interest (e.g., A versus B; A versus C).

Reviews of two samples of 108 and 292 trial protocols from 2016 found that 91% and 94% described the specific objectives of the trial, respectively.(9, 10)

Recently, some trials are being designed using the estimands framework to define the research question and trial objectives.(172) Key concepts of this framework are presented in Box 1.

Box 1 Estimands

Concerns have been raised that the precise     research questions that randomised trials are intended to answer are often     unclear.(173) In particular, there     is often ambiguity around how events occurring after randomisation (termed     intercurrent events) are handled. Research objectives can be specified using an estimands     framework to improve clarity. We provide a     brief overview of estimands and introduce terminology, so this framework     can be applied and reported if used. A more detailed primer on the estimand     framework, which provides practical guidance on estimands in studies of     healthcare interventions, can be found elsewhere.(174)

The European Medicines Agency(175) defines an estimand     as “a precise description of the treatment effect reflecting the clinical     question posed by a given clinical trial objective”. The estimands     framework provides a structured description of the treatment effect in an     attempt to bring clarity in specifying the research question, which can be     used to guide the study design, data collection and statistical analysis     methods. In brief, an estimand comprises five key attributes: 1)     population, 2) treatment conditions, 3) endpoint, 4) summary measure and 5)     handling of intercurrent events (Table i). A separate estimand should be     defined for each study outcome and for some outcomes, more than one     estimand may be defined.

Box 1,     Table i: Five key attributes of the estimands     frameworka

Attribute

Definition

Population

Patients for whom researchers want       to estimate the treatment effect

Treatment conditions

Different intervention strategies       being compared in the treatment effect definition

Endpoint

Outcome for each participant that is       used in the treatment effect definition

Summary measure

Method used to summarise and compare       the endpoint between treatment conditions (e.g., risk ratio, odds ratio)

Handling of intercurrent events

Strategies used to handle each       intercurrent eventb

in the treatment effect definition;       different strategies could be used for different types of intercurrent       events

aTaken from(174); bIntercurrent     events are post-baseline events (or post-randomisation events in randomised     trials) that affect the interpretation or existence of outcome data. These     events frequently affect receipt of treatment (e.g., treatment switching or     treatment discontinuation) or preclude existence of the outcome (e.g.,     death, if it is not defined as part of the outcome).

The European Medicines Agency outlines five strategies for handling     intercurrent events, which are at the core of the estimands framework(174, 175), (see Table ii).

Box 1,     Table ii: Strategies for handling intercurrent events

Strategy

Description

Treatment policy

The occurrence of the intercurrent       event is considered irrelevant in defining the treatment effect of       interest: the value for the outcome of interest is used regardless of       whether or not the intercurrent event occurs

Hypothetical

The treatment effect in a scenario       where the intercurrent event did not occur is of interest

Composite

The intercurrent event is       incorporated into the outcome definition

While on treatment

The outcome prior to the occurrence of the       intercurrent event is of interest

Principal stratum

The outcome in a subpopulation of       patients who would not (or would) experience the intercurrent event is of       interest.

While the terminology surrounding estimands     may be new to some investigators, the concepts defined by the attributes and strategies of the framework are not new. A number of     existing reporting guidelines have recently included estimands in     their recommendations.(176-178) If the estimands     framework has been used to design the trial and data collection or inform     the statistical analysis (by choosing appropriate methods), then this     should be clearly reported within the manuscript.

Summary of key elements to address

● Trial objectives related to the benefits and harms, including the:

o Participants

o Intervention

o Comparator

o Primary outcome(s)

o Time point of main interest

● Description of trial estimand(s), as appropriate