An organization deleted 1,417 studies that had never been conducted, in what was described as a record case of fraud involving engineers.