test data—how much is not enough?

1
Environmental Mutagenesis 6:119 (1984) Letter to the Editor Test Data-How Much Is Not Enough? When investigators publish mutagenicity test results in the scientific literature, they have an obligation to include sufficient data to permit the reader of the publication to make an independent evaluation of the results and conclusions. There is probably no one working in the field of genetic toxicology who would disagree with this statement; yet manuscripts continue to appear in print that do not contain the minimal data needed to evaluate the results. The most frequent “sin of omission” is the failure to present positive and/or negative control data. We find it difficult to understand how some investigators can reconcile the need for controls in conducting experiments with the subsequent omis- sion of the control data from their publications. A brief list of offenses includes: 1. Tables and figures in which control values have been subtracted from the treatment values, and the relevant control values are not provided elsewhere. 2. Presentation of control values that were not obtained concurrently with the treatment values. 3. Presentation of data that are so highly transformed that the reader cannot calculate back to more primary values. 4. Presentation of qualitative (plus/minus) data that are not accompanied by quantitative data. At the National Toxicology Program (NTP), we are acutely aware of this problem because we rely heavily on published mutagenicity test data in order to evaluate the overall toxicity of selected chemicals, especially those that are nominated or are being tested for carcinogenicity by the NTP. The publication of papers that contain inadequately presented data diminishes the usefulness of such data to us and, certainly, to others in this field. This problem is not limited to any one journal; in fact, our survey of the recent literature shows that all of the major journals in this field have published papers that contain the problems outlined above. Consequently, our purpose is not to criticize this journal or its editorial staff, but rather, to emphasize the importance of including sufficient control (and test) data in publications in order to allow an independent evaluation of the results. Clearly, we are not the first to recognize or publicize these inadequacies of the data. We do, however, think that it is necessary to stress again that authors, reviewers, and editors need to be aware of these problems and should strive to correct them so that future papers will provide a more useful contribution to the mutagenesis literature. David M. DeMarini and Michael D. Shelby National Toxicology Program National Institute of Environmental Research Triangle Park, North Health Sciences Carolina 0 1984 Alan R. Liss, Inc.

Upload: david-m-demarini

Post on 11-Jun-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Test data—how much is not enough?

Environmental Mutagenesis 6:119 (1984)

Letter to the Editor

Test Data-How Much Is Not Enough?

When investigators publish mutagenicity test results in the scientific literature, they have an obligation to include sufficient data to permit the reader of the publication to make an independent evaluation of the results and conclusions. There is probably no one working in the field of genetic toxicology who would disagree with this statement; yet manuscripts continue to appear in print that do not contain the minimal data needed to evaluate the results.

The most frequent “sin of omission” is the failure to present positive and/or negative control data. We find it difficult to understand how some investigators can reconcile the need for controls in conducting experiments with the subsequent omis- sion of the control data from their publications. A brief list of offenses includes:

1 . Tables and figures in which control values have been subtracted from the treatment values, and the relevant control values are not provided elsewhere.

2. Presentation of control values that were not obtained concurrently with the treatment values.

3. Presentation of data that are so highly transformed that the reader cannot calculate back to more primary values.

4. Presentation of qualitative (plus/minus) data that are not accompanied by quantitative data.

At the National Toxicology Program (NTP), we are acutely aware of this problem because we rely heavily on published mutagenicity test data in order to evaluate the overall toxicity of selected chemicals, especially those that are nominated or are being tested for carcinogenicity by the NTP. The publication of papers that contain inadequately presented data diminishes the usefulness of such data to us and, certainly, to others in this field.

This problem is not limited to any one journal; in fact, our survey of the recent literature shows that all of the major journals in this field have published papers that contain the problems outlined above. Consequently, our purpose is not to criticize this journal or its editorial staff, but rather, to emphasize the importance of including sufficient control (and test) data in publications in order to allow an independent evaluation of the results. Clearly, we are not the first to recognize or publicize these inadequacies of the data. We do, however, think that it is necessary to stress again that authors, reviewers, and editors need to be aware of these problems and should strive to correct them so that future papers will provide a more useful contribution to the mutagenesis literature.

David M. DeMarini and Michael D. Shelby

National Toxicology Program National Institute of Environmental

Research Triangle Park, North Health Sciences

Carolina 0 1984 Alan R. Liss, Inc.