Abstract
We are coming to this discussion a bit late, but after attending some recent meetings where regulatory agency personnel and Society of Toxicologic Pathology (STP) members discussed
As most of you probably know from reading other articles (Kerlin 2012; Lepore 1996; Tuomari et al. 2004) and hearing various STP members and other thought leaders speak on this topic, the definition of pathology raw data most commonly presented is that it consists of the final, signed, pathology report (referred to as the pathology report from here on), and sometimes the associated findings incidence and severity tables. However, there is considerable consternation on this definition (Kerlin 2012).
Our rebuttal to this definition, specifically the inclusion of the pathology report as raw data, is that a search for “definition of raw data” does not identify one result that resembles the content of the typical pathology report. Also, to us anyway, basic reasoning would indicate that a report is not “raw;” a great deal of thought and deliberative writing goes into the creation of a pathology report.
“Raw Data” definitions and explanations that we have found include the following (italics added for emphasis): Raw data are data that have not been processed for use. A distinction is sometimes made between Raw data have not been processed to a presentable form for anyone other than someone with special training to understand them. Raw data are usually processed to make them more refined. The important point is the distinction between Raw data have not been processed in any manner. The term may also refer to recently captured data that has been placed into a database structure (such as a pathology database software program—our addition), but not yet processed. (encyclopedia2.thefreedictionary.com)
Our interpretation, paraphrased from wisegeek.org:
Additionally, the initial data set from computerized evaluations of digital images/digital pathology (including stereology and morphometry) should be considered pathology raw data.
In the current practice of toxicologic pathology, more specifically, histopathology, there are often changes to the original diagnostic entries made by the study pathologist, such as after formal or informal peer review, consultation with colleagues, reading book chapters, and so on. At this stage of the process, the individual animal findings may be termed working raw data. When the individual animal findings are used to create a pathology report (the final, signed version), they should be termed final raw data. Such final data sets are usually “locked” (Tuomari et al. 2004). Working changes to the raw data are common, and necessary, to create the final, locked, raw database; it is simply the process by which microscopic toxicologic pathology raw data are created.
Changes are certainly needed to the Good Laboratory Practice (GLP) and Federal Drug Agency/Environmental Protection Agency definitions of pathology raw data. What exists now is not clear, and in light of modern definitions of the term raw data, do not hold up. The GLP definition of raw data is “the first recording of an original experimental observation” (Lepore 1996).
Since histopathology findings are often changed after the first recording, our definition of microscopic pathology raw data does not fit this definition of raw data, dating back 30 years or more. How the regulatory agencies reached the conclusion that the final, signed pathology report would become the raw data is lost in the mists of history; there is discussion about the “pathologist’s interim notes” not being raw data because they do not contribute to reconstruction of the final report (Lepore 1996). Pathologist’s interim notes are important only if one is unyieldingly tied to the definition of raw data as the “first recording of an observation.” We agree that this definition (the “first recording of an observation”) does not apply to the practice of histopathology, but it should not force toxicologic pathologists to accept another definition of raw data that is unsuited to our modern world of computerized creation, modification, and storage of raw data, and is inconsistent with current definitions of the term.
There is a scenario, however, wherein the pathology report may contain the histopathology raw data, such as small studies with one, or very few, animals. Immunohistochemistry studies and studies with implants, and so on, are commonly done without the use of a computerized data retrieval program. Some of these studies use a computer software program, such as Microsoft Excel or Word, to record and store the raw data. For these types of studies, the signed, dated, pathology report may have the raw data attached as an addendum or it may be included in the body of the report. As in a traditional toxicology study, however, the raw data are the individual animal findings rather than the interpretive narrative portion of the report.
Although it is tempting to include the summary incidence and severity tables as raw data, these are data derived, or processed, from the individual animal findings and should be considered the first level of processed data.
If we continue to consider the pathology report as raw data, then the obvious question is as follows: “Where is the explanation of these raw data and who is processing these data into information?”
Our main point, obviously, is that the pathology report is not raw data; it typically contains selected
The goal of the pathology report is to explain the raw data—to put the explanations in a narrative that is understandable. The report needs to be created by the study pathologist; no one can interpret the raw data better than the study pathologist.
Our opinion is not intended to diminish the importance of the signed pathology report nor to alter the established procedures for creation, review, and submission of those narrative reports. It does, however, define microscopic observations (including manual or computer-generated observations from digital images) as raw data and thus makes them subject to GLP and other regulatory constraints. By strict definitions and terminology, individual histopathology observations currently are not defined as raw data, and there may be questions as to the applicability of pertinent regulations, as indicated above in the outdated, ambiguous, regulatory agency definition of pathology raw data. Given that the overall goal of GLP regulations is to ensure the integrity of nonclinical safety assessment studies, it seems patently absurd to exclude histopathology observations from the provisions of those regulations because of a narrow and somewhat antiquated interpretation of definitions and historical precedents. Therefore, we are proposing the following definitions to differentiate “raw data” from its current limitation to the “final, signed pathology report.” Microscopic pathology raw data = individual animal findings; these may be divided into “working” raw data (prior to locking, or prior to creation of the final, signed, report) and “final” raw data (the locked data used to create the final, signed, pathology report). Microscopic pathology processed data (first iteration) = summary incidence/severity tables. Pathology information (additional iterations of processed data and narrative explanation of raw data) = pathology report.
Footnotes
*This is an opinion article submitted to the Regulatory Forum and does not constitute an official position of the Society of Toxicologic Pathology or the journal
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Abbreviations
Acknowledgment
We greatly appreciate the input from members of the Regulatory Forum, for helping us present our position in a clearer format.
Authors’ Note
Several years prior to the preparation of this Opinion Piece, one of us (JRL) had a discussion with Dr. Steven Stefanski, which helped immensely to understand the concept of pathology raw data.