Table 1

Core CONSORT and SPIRIT elements addressing statistical considerations for trials

Statistical considerations for reporting, with guidance regarding core detailsDefined SPIRIT item?Defined CONSORT item?
Description of trial design. Specify clearly the type (eg, parallel, crossover, factorial, cluster randomised, etc), the framework (eg, superiority, non-inferiority, exploratory/pilot) and the ratio of allocation to intervention groups for the study.
Estimation of sample size. Specify the estimated number of patients for enrolment based on calculations performed. Provide clear mention of the outcome used as the basis for the estimation, as well as the assumptions made and/or additional information used (eg, assumed event rates or means and SDs per group, along with source or rationale for these values, type 1 error rate, chosen power level, test to be used to compare groups, adjustments for anticipated losses to follow up and other information which may be required for sample size estimations for trial designs beyond the standard parallel groups design) to arrive at the planned sample size.
Assignment of patients to intervention groups. Describe the approach to generate the randomisation scheme for patients with additional details as necessary (eg, computer generation of random sequence, stratification factors with boundary criteria, size of blocks to be implemented, use of simple randomisation or adaptive randomisation or other modified approaches).
Description of elements of the study's primary analysis. Specify the primary and secondary end points for comparison between groups, and the main analytic approach of focus for primary interpretation of findings (eg, statistical test). If multiple groupings of patients are to be considered in separate analyses (eg, related to adherence), clearly describe these groups. Describe the measures of effect (eg, risk ratio, mean difference) that will be used to summarise comparisons between groups, and indicate chosen type 1 error rates and reporting of CIs. Describe adjustments to be made for multiple testing, and any additional design-specific information required for analysis (eg, for cross-over or cluster-randomised studies).
Performance of additional data analyses. Efforts to consider additional analyses of the study data, for example, subgroups (either clinically or methods oriented), should be described. Clear details regarding any boundaries to group patients should be indicated. Statistical techniques for additional comparisons should be clarified, and a priori hypotheses regarding subgroups should be reported. Analyses incorporating adjustments (eg, via regression) should also be outlined. Specific approaches for missing data (including imputation or other approaches) should be described.
Interim analyses of study data. Specify details regarding interactions with a data monitoring committee and statistical analyses performed at key points during the trial (eg, timing of analyses, analytic methods), as well as criteria that will be considered for early study stopping (related to benefits, harms or futility), sample size adjustments or modifications to allocation ratios.
Assessment of harms observed during the study. Specify anticipated harms and whether data were actively solicited from patients, duration of time for which end points were recorded and how frequently. Make clear any formally planned statistical comparisons of harms to be made between groups, and/or descriptive techniques to be used to summarise the data.
Describe protocol changes made during the study. All changes made after protocol finalisation and study start should be clearly described. This includes the addition, removal and reordering of clinical end points, modifications to planned study sample size, changes to the data analysis plan and other changes of relevance for readers which represent changes from the study protocol.
Describe patient flow observed during the study. The numbers of patients assigned to intervention groups, the numbers receiving interventions and the numbers contributing to data analysis should be provided. Patient attrition should be clearly described. Use of a flow diagram is common and recommended.
Document baseline demographics across groups. A clear description of the distribution of all important patient demographics and clinical features should be provided. This addresses the comparability of groups, establishes the success of randomisation in balancing confounders and instructs readers on features of the study population.
Document the numbers of patients included in analyses. Report the numbers of study participants that were included for formal analysis for each end point assessed. Make clear whether analyses reflect the groups as originally randomised, or whether certain differences are present.