FDA publishes two guidances on collecting clinical outcomes data

The US Food and Drug Administration (FDA) has finalized two guidance documents in its series of guidances related to Patient-Focused Drug Development (PFDD). The documents outline the agency’s thinking on clinical outcome assessment (COA) data that can be used in premarket applications.

FDA published the COA guidances on 3 November 2023. The first guidance details technical specifications that sponsors should consider including in their premarket applications when submitting COA information that uses Item Response Theory (IRT). The COA data could come from a number of sources, including the clinician in the form of a clinician-reported outcome (ClinRO), the patient in the form of a patient-reported outcome (PRO), a non-clinician observer in the form of an observer-reported outcome (ObsRO), and a performance-based assessment in the form of performance outcome (PerfO) measurement.

The guidance lists certain terms and their definitions, which may help sponsors better understand FDA’s thinking. For instance, the agency notes that IRTs are mathematical models that describe the relationship between the performance and characteristics of an item and the patient’s status on the measurement in question. In that vein, the agency lists COAs that use IRTs, including those that are static COAs and COAs that are administered using IRT-based Computerized Adaptive Testing (CAT).

“CAT is a sequential form of individual testing administered by a computer in which successive items in the COA measure are selected for administration based primarily on the item’s psychometric properties and content in relation to the patient’s responses to previous items,” FDA wrote. “COAs that use IRT can leverage an item bank for item selection.”

The agency added that an item bank “represents the total set of items from which a subset is selected during COA measure development or selected for the patient during adaptive testing.”

The guidance also details specifications on how to submit Clinical Data Interchange Standards Consortium (CDISC) Study Data Tabulation Model (SDTM) and Analysis Data Model (ADaM) datasets. They are meant to be general guidelines that sponsors can use to submit COA data in their premarket application.

The guidance delves into the specifics of what documentation sponsors need to provide when submitting COA data and its timing, the dataset standards sponsors should use and how they should address issues such as missing data.

“Missing data should be represented within the [questionnaire] dataset with the reason for missingness captured under ‘Reason Not Performed’ (QSREASND),” FDA wrote. “The handling of missing data may differ depending on whether the covered COA is fixed/static or is administered using CAT.”

FDA also published a final guidance on how sponsors can submit patient-reported outcome (PRO) data from clinical trials when developing cancer drugs. The agency defines a PRO as information that comes directly from the patient based on self-reports and interviews that are not interpreted by a clinician or other parties. This latest guidance supplements the agency’s draft guidance -- Core Patient-Reported Outcomes in Cancer Clinical Trials -- which was published in June 2021.

Similar to the broader COA guidance, the PRO guidance details what FDA is looking for in premarket applications that contain Clinical Data Interchange Standards Consortium (CDISC) Study Data Tabulation Model (SDTM) and Analysis Data Model (ADaM) datasets. More specifically, the agency outlines its thinking on how to submit standardized dataset content and present tables and figures.

FDA, however, notes that sponsors may choose to present their PRO data in other ways, and there may be additional datasets and tables that are necessary to satisfy the agency’s regulatory requirements. Regulators will still consider such datasets and tables in the premarket application as long as there is a good justification.

“The dataset specifications and specifications for tables and figures are pursuant to discussions with FDA and may vary by clinical drug development program and clinical trial therein,” FDA wrote. “These specifications are intended to be applicable to any PRO data used to inform the evaluation of (1) safety and tolerability or (2) clinical benefit in randomized studies (i.e., improvement in disease symptoms) within a cancer clinical trial.”

The guidance does not delve into information needed to support the PRO measures used by sponsors or their interpretation. Instead, the agency recommends that sponsors should talk with regulators early on about the right PRO measurements needed to support their drug development. The guidance also states that some of the recommendations in the document may be useful for collecting other COA datasets in cancer trials.

After detailing terms and outlining FDA's thinking on how to collect COA information in clinical trials, both guidances also provide examples of SDTM and ADaM questionnaire datasets that sponsors can use as comparisons when putting together premarket applications.

Final guidance on COAs that use IRT
Final guidance on PRO data from cancer trials

FDA publishes two guidances on collecting clinical outcomes data

Related topics

Recommended content

Welcome to the new RAPS Digital Experience