Trial results records for public registries

A trial within the scope of US or EU disclosure law owes the public registries a structured results record, separate from its clinical study report. ClinicalTrials.gov and the European database, EudraCT, each define what the record contains. Its data has to be planned alongside the clinical study report and entered by people each database grants access to, and what it discloses has to stay consistent with the rest of the trial's public record.

What each record contains

ClinicalTrials.gov defines the results data that must be submitted in its Results Data Element Definitions for Interventional and Observational Studies, and divides the record into seven modules. ^{1, 2}

• Participant flow: a tabular summary of the progress of participants through each stage of the trial, by arm or comparison group, giving the numbers who started, completed and dropped out of each period of the trial based on the sequence in which the interventions were assigned.
• Baseline characteristics: a tabular summary of the data collected at the beginning of the trial for all participants, by arm or comparison group, covering demographics such as age and gender and trial-specific measures as appropriate.
• Outcome measures: a tabular summary of outcome measure values, by arm or comparison group, with tables for each pre-specified primary and secondary outcome. It may also include other pre-specified outcomes, post-hoc outcomes and any appropriate statistical analyses.
• Adverse event information: a tabular summary of all serious adverse events and another of the non-serious adverse events exceeding a specified frequency threshold of more than 0, 1, 2, 3, 4 or 5 per cent. Each event is given with its term, the affected organ system, the number of participants at risk and the number affected, by arm or comparison group.
• Limitations and caveats: significant limitations of the trial, such as not reaching the target number of participants needed for target power and statistically reliable results, or technical problems with measurements leading to unreliable or uninterpretable data.
• Certain agreements: whether an agreement between the sponsor or its agent and the principal investigators, unless the sponsor employs them, restricts in any manner their ability after the trial to discuss its results at a scientific meeting or any other public or private forum, or to publish information about its results in a scientific or academic journal.
• Results point of contact: the point of contact for scientific information about the results.

¹

EudraCT defines the results data that must be submitted in its Result Related Data Dictionary, and divides the record into six modules. ³

• Trial information: the trial identification details, paediatric regulatory details, sponsor details, the results analysis stage, general information about the trial, the number of participants enrolled per country, and a breakdown of the trial population by age group.
• Subject disposition: recruitment, screening, blinding implementation, the trial products, and a tabular summary of the progress of participants through each stage of the trial, by arm or comparison group.
• Baseline characteristics: a tabular summary of the data collected at the beginning of the trial for all participants, by arm or comparison group, covering demographics such as age and gender and trial-specific measures as appropriate.
• End points: a tabular summary of endpoint values, by arm or comparison group, with tables for the primary and secondary endpoints. It may also include other pre-specified endpoints, post-hoc outcomes and any appropriate statistical analyses.
• Adverse events: a tabular summary of all serious adverse events and another of the non-serious adverse events exceeding a specified frequency threshold of more than 0, 1, 2, 3, 4 or 5 per cent. Each serious event is given, by arm or comparison group, with its term, the affected organ system, the number of participants at risk, the number affected, the number of occurrences, the number of occurrences causally related to treatment, the number of fatalities and the number of fatalities causally related to treatment. Each non-serious event is given, by arm or comparison group, with its term, organ system, number at risk, number affected and number of occurrences.
• More information: substantial global protocol amendments, global interruptions to the trial, limitations and caveats, and online references.

⁴

Which endpoints have to be disclosed

The US Final Rule requires all primary and secondary outcome measures to be disclosed on ClinicalTrials.gov, whether or not target accrual was met, the trial was terminated, or the planned analyses were expected to yield statistical significance. ² So the choice of primary and secondary endpoints should be considered carefully when the protocol is drafted, and each endpoint needs a specific definition to avoid ambiguity over what must be disclosed later. ² For example, naming the pharmacokinetics of a drug as a secondary endpoint means all derived pharmacokinetic parameters should be disclosed. ² If only certain parameters are of interest, they should be specified, such as maximum plasma concentration, area under the plasma concentration-time curve from time zero to infinity, and half-life. ² Where data are collected and analysed at multiple time points, whether to restrict the primary or secondary endpoints to the time points of interest should also be considered. ² Under US law all endpoints in the statistical analysis plan must be disclosed, while endpoints designated "other" are not expected to be, though voluntary disclosure of them, or of analyses such as meta-analyses, is encouraged. ⁵

Planning alongside the study report

The disclosure of trial results must be planned alongside the preparation of the clinical study report. ⁶ Some data the records require, such as non-serious adverse events and the number of participants enrolled per country, are not commonly summarised for the clinical study report. ^{7, 8} The data requirements should be checked carefully, ideally while the statistical analysis plan is produced, so that all required data will be tabulated and summarised. ⁷ They should be considered early enough for the regulations to be met. ⁸

The information ClinicalTrials.gov requires can be mapped to where a full clinical study report typically provides similar information. ⁹ A synoptic or abbreviated report may not address all of it. ⁹ Best practice in planning a posting takes account of timelines, collaboration with cross-functional team members, and strategies for when only an abbreviated report is available. ⁹

For the European record, a list of randomised participants should be prepared by country and by pre-specified age category, for all participants and for each treatment group. ¹⁰ Statistical analysis of results is expected at least for the primary endpoints. ¹¹ Where a primary endpoint has no statistical analysis, the database's validation system requires a justification, so a brief justification statement should be ready. ¹¹

Non-serious and serious adverse events should be presented in separate tables. ¹² For non-serious events, the number of participants affected in each treatment group should be shown, with outputs prepared according to the safety dictionary used, for example by system organ class and preferred term. ¹² The number of occurrences of a particular adverse event should be shown overall and for each treatment group, and a threshold of up to 5% of participants affected by the event in any treatment group can be applied. ¹² All serious adverse events should be shown, with outputs prepared according to the safety dictionary used, and with the number of participants affected, the number of occurrences of each, and relatedness to treatment. ¹² All fatalities should be shown, with those related to treatment specified. ¹² Some sponsors upload the adverse event information as a customised file, for example in extensible markup language (XML), and instructions for preparing such a file are available on the European Medicines Agency's website. ¹²

Preparation of the required records should be part of the overall study timeline to ensure compliance with the regulations. ^{13, 14} The time it takes is driven mainly by the number of endpoints and any associated statistical analyses, and by the number of serious and non-serious adverse events that occurred during the trial. ¹³ If both US and EU results records are required, efficiencies can be made by completing one record before starting the other. ¹³

Access to the databases

Records for ClinicalTrials.gov can be prepared directly in its web-based data entry system, the Protocol Registration and Results System, and records for the EU directly in the EudraCT database. ² ClinicalTrials.gov establishes one account for an organisation, managed by the organisation's administrator, who can then grant individuals access to specific trials as required. ² In the EU a primary user is assigned for a trial through a Clinical Trial Assignment Request Letter, which must be completed by the sponsor, by the addressee of the decision on a paediatric investigation plan, or by the marketing authorisation holder. ² A template for the letter and accompanying instructions are available on the European Medicines Agency's website. ¹⁵ The agency then grants access for the trials listed in the letter to one primary user, who can assign one backup user and multiple delegated results preparers and posters for each listed trial. ² Individual users apply for a single personal account, and the primary user then assigns them specific trials to edit. ²

Where the sponsor does not have a primary results user, the agency can take up to 25 calendar days to process a request for access to EudraCT, which impending registry deadlines sometimes do not allow. ¹⁶

Entering the data

Data can be entered manually or uploaded as XML files, for which schemas are available online for both ClinicalTrials.gov and EudraCT. ¹³ Specialist vendors also offer tailored authoring software for managing the authoring, approval and release of records to both systems. ¹⁵ One contract research organisation's disclosure group invested in its own software for creating, reviewing and uploading results postings to both registries, which avoids entering the data twice where results must be posted to both. ¹⁶ The software also lets the group start a posting and download drafts for the sponsor to review, then upload the entry to EudraCT as an XML file once access is granted. ¹⁶

Three properties of the database fields should be shared with anyone performing quality control on a record, to avoid redundant findings. ⁷

• Fields are annotated with symbols marking information as mandatory, conditionally required or optional.
• Some fields can be completed only by selecting from a drop-down menu.
• Some fields have character limits on free text.

⁷

In the Protocol Registration and Results System the official title of a trial is limited to 600 characters, and in both EudraCT and ClinicalTrials.gov an outcome measure title is limited to 255 characters. ¹⁷ Outcome measures should include the measure, units and time points. ¹⁷

Consistency of the public record

The range of trial results information that must be made publicly accessible keeps growing in both the US and the EU. ⁶ This brings challenges, not least keeping the publicly available data for a trial consistent. ⁶ The specific disclosure requirements also differ between ClinicalTrials.gov and EudraCT. ⁶ Where a trial is conducted in both the EU and the US, consideration must be given to keeping its results on the two registries consistent, given the different regional reporting requirements. ¹ Consistency must also take account of the lay summaries and redacted clinical study reports that will be released for public viewing in the EU. ⁸

The US quality control review

Once results have been released through the Protocol Registration and Results System, the record undergoes a quality control review by the National Institutes of Health for the clarity and completeness of the information submitted. ¹⁴ The record is not released publicly until it passes this review, which may take several iterations. ¹⁴ In 2018 no such process occurred for records submitted to EudraCT, so addressing the review comments may add to the inconsistencies between the results disclosed in the US and the EU. ¹⁴