This page is DRAFT

Introduction

As part of the submission process, study data submitted to the FDA must meet structure (format) and completeness (content) standards detailed in the FDA Standards Catalog (Add Reference). The FDA may reject an application based solely on technical deficiencies. The “Technical Rejection Criteria for Study Data (TRC)”(Add Reference) details requirements of varying severity. In their analysis of data from 2016 to 201, Chen and Hussong (2019) reported that approximately 32% of submissions contained critical errors.

The submission rules are not complex, but confusion exists around their implementation. The FDA recently created a Self-Check Worksheet for Study Data Preparation and accompanying Instructions to assist applicants in preparing their submissions. The rules could easily be extended to additional structure and quality checks to further improve the submission process for both applicants and the FDA.

Rationale

The submission process can be greatly streamlined by automating conformance to the technical submission requirements when the process and data use Linked Data principles. Data quality will be increased. Rejection of submissions due to technical issues can be avoided when both sponsors and regulatory agencies employ the same, automated, evaluation processes.

Background

The data standards listed in the FDA Data Standards Catalog (Add Ref) must be applied for nonclinical and clinical study NDAs, BLAs and ANDAs (with study start after 2016-12-17) and INDs (study start after 2017-12-17). Both CDER and CBER centers are automating the review of study data submissions based on these specifications. Both regulatory agencies and submitters would benefit from basing the validation process on Linked Data.

Criteria for Automated Validation

Study Start Data (SSD) Example

Study Start Date (SSD) is a key component in submissions, so its accuracy is paramount. Automated processes at the FDA rely on the SSD value provided in the Trial Summary domain (TS) in the dataset file TS.XPT. The TS dataset is referenced in the Study Tagging File (STF). Therefore, both the TS file(s) and STF must be present and cross referenced. Furthermore, the study start date is a derived data point (e.g. for human trials, it is the earliest date informed consent is obtained for all study subjects). Currently it is difficult to automate the validation of SSD derivation. Linked Data can faclitate this validation automation.