The most appropriate database structure for data collected repeatedly over time for each subject is a normalized, "long" format. In this scenario, vital signs are a set of measurements collected at multiple, distinct events (visits). Structuring the table with one record per patient per visit creates a unique entry for each data collection instance. This relational model ensures data integrity, prevents redundancy, and provides the flexibility to handle scheduled, unscheduled, or missed visits without altering the table structure. The primary key for such a table would typically be a composite of the patient identifier and the visit identifier.

A. One record per visit: This is ambiguous and fails to associate the data with a specific patient, which is a critical requirement for clinical trial data.
B. One record per patient per study: This "wide" format is inefficient, creating a single large record with columns for each visit (e.g., Visit1BP, Visit3BP), leading to many null values and making the database difficult to query and scale.
D. One record per patient: This structure is only suitable for static, non-repeating data (e.g., demographics) and cannot accommodate longitudinal data like vital signs collected at multiple time points.

1. Society for Clinical Data Management (SCDM). (2023). Good Clinical Data Management Practices (GCDMP).

Section 4.3.2, Database Design: "The database should be designed to reflect the data collection requirements as specified in the protocol... The design should be based on data standards where applicable." A protocol defines data collection on a per-patient, per-visit basis. A structure of "one record per patient per visit" directly models this requirement, which is fundamental to relational database design for clinical trials.

2. Prokscha, S. (2012). Practical Guide to Clinical Data Management (3rd ed.). CRC Press.

Chapter 6, Database Design and Build, Page 85: The text explains that for repeating data, such as vital signs collected at multiple visits, "the data are stored in a table with one record for each time the measurements were taken." This directly corresponds to the "one record per patient per visit" structure.

3. Krishnankutty, B., Bellary, S., & Kumar, N. B. R. (2012). Data management in clinical research: An overview. Indian Journal of Pharmacology, 44(2), 168–172. https://doi.org/10.4103/0253-7613.93842

Page 170, Section: Database designing: The article states, "The data from a clinical trial are collected on Case Report Forms (CRFs)... The structure of the database is based on the CRFs." Since CRFs are completed for each patient at each visit, the underlying database must have a corresponding structure, which is one record per patient per visit, to accurately capture the collected data.