Reviews are papers that compile, summarize, critique, and synthesize the available information on a topic. Among Environmental Toxicology and Chemistry's 100 most cited papers (Supplemental Data, Table S1) they are disproportionately represented (Supplemental Data, Table S2) for good and obvious reasons. If you wish to provide citation to support your use of a method or assumption, a good review paper will provide that support by showing that the method or assumption is consistent with current knowledge or best practices. The popularity of reviews is illustrated by the criticism of Impact Factors as being biased in favor of reviews relative to original research . In this editorial, I will not review the individual highly cited reviews. Rather, I will encourage you to write more and better reviews as a service to the scientific community and to enhance your career.
WHY WRITE A REVIEW?
Writing a good review is a service to the scientific community. The progress of science depends on the development of a consensus version of current scientific truth . That consensus is created and modified by various activities, including the creation, peer review, and publication of review papers. A good review paper can clarify the state of knowledge, explain apparent contradictions, identify needed research, and even create a consensus where none existed before.
Writing a good review can also help to advance your career. Because reviews tend to be highly cited, they help with recognition and promotion. They show that you have mastered a topic that may be important to your organization or clients. They may even demonstrate that you are a good synthetic thinker. Finally, if you are in a situation of not producing original research or not being able to publish original work because of legal or policy constraints, writing a review allows you to continue adding to your publication record.
HOW TO WRITE A REVIEW
The most important decision in developing a review is selecting a topic. In general, the topic should be something in which you and others are interested and a topic that has not been reviewed recently. Once upon a time, the next step of compiling the literature to be reviewed was arduous and time consuming, involving traveling to libraries, leafing through bound journal volumes, mailing reprint requests, and so on. Now, thanks to the Internet, tools such as Google Scholar and JSTOR, and the availability of electronic articles, literature compilation is relatively easy. Much of the work now involves extracting the critical information from the publications and organizing it in a way that illuminates the issue.
Finally, you must ask what sort of conclusion is appropriate. At one extreme, the conclusions are simply a condensed summary of the contents of the literature, without interpretation. Other reviews provide an overall interpretation of the literature, such as Ratte's conclusion that the high silver toxicity seen in the laboratory is not expected in ambient waters because of the many processes that render silver unreactive . Erickson's review of the biotic ligand model highlighted complexities in metal toxicity that are not well treated by the model and are overlooked by most users . Some reviews verge on polemic. For example, my contribution to Environmental Toxicology and Chemistry's 100 most cited papers reviewed the uses of the concept of ecosystem health and concluded that the concept should be abandoned . At the highest level, a review identifies new patterns in the data that have not been previously recognized.
WHO SHOULD WRITE REVIEWS?
Many reviews are written by researchers who study the topic being reviewed. They are in a good position to review the topic; but unfortunately, they also are biased in favor of their own results and against those who have contradicted them. It takes a real effort of will to be objective in such circumstances. Workshop participants may be good sources of unbiased reviews because their diverse points of view can reduce bias, increase the likelihood of identifying all relevant publications, and promote creative interpretations . In my opinion, the ideal review writers are users of data rather than generators of data. They are likely to have a practical view of the issues and to be less biased. The best standard for science is the question, “Does it work in practice?” —and users know the answer to that question better than generators.
HOW SHOULD REVIEWS BE USED?
Reviews provide, at minimum, an entrée to an unfamiliar topic and a quick way to gain some familiarity with the issues and major contributors. They may also provide useful syntheses and original insights. However, their conclusions must be used with caution. Some of the highly cited reviews, for example, were funded by the industries that released the chemicals being reviewed. That does not negate the utility of the review, but it should increase the reader's circumspection. Reviews should not be used as a substitute for reading the original research papers that are important to your own work. It is particularly bad practice to cite a paper based on the description of it in a review. By all means, take advantage of reviews; but do not consider them to be the final authority.
Unlike original research, most review papers lack defined methods. I believe that they would benefit from clearer definitions and descriptions of the methods used to search the literature, identify relevant publications, extract information, and generate conclusions. This is done for reviews in support of regulatory activities such as the US Environmental Protection Agency's Integrated Science Assessments (www.epa.gov/ncea/isa) and Integrated Risk Information System (IRIS) documents (www.epa.gov/iris). Such definition of methods has been recommended in the medical literature  and in environmental management . In medicine, it has been standardized to include common procedures, formats, and scoring systems [9-13]. It is easy to imagine conventional methods for reviews of the environmental chemistry or toxicity of chemicals. In any case, greater attention by review authors to their methods seems likely to improve the quality of results, and the reporting of methods could provide greater assurance of quality and lack of bias. Online supplementary materials provide a venue for that material.
I hope to have convinced you to be a generator as well as a user of review papers. In particular, I encourage risk or impact assessors to consider publishing reviews. You review models, data, and assumptions every time you begin an assessment of a new chemical, effluent, site, species, or other issue. Why not share the results with your colleagues and enhance your reputation at the same time?
Tables S1. (49 KB PDF).
Table S2. (53 KB DOC).
This editorial benefitted from comments by J. Lipscomb and M. Kravitz. The document has been reviewed in accordance with US Environmental Protection Agency policy and approved for publication. The views expressed are those of the author and do not necessarily reflect the views of policies of the US Environmental Protection Agency.