The existence of publication bias and risk factors for its occurrence

Article Abstract:

Scientific researchers may be more likely to submit their results to journals for publication if the findings are dramatic, or support the original hypothesis. For example, if a study is comparing two treatments for a disease and one is found to be significantly more effective than the other, the researchers may be more inclined to submit their results than if both therapies were equally valuable. This is called publication bias, meaning a prejudice in deciding which research findings to publish. Publication bias is also practiced by the editors of journals and the reviewers they consult with while evaluating manuscripts that have been submitted. Three previous studies have documented that publication bias occurs in the field of medical research. But it is important from a practical and a theoretical standpoint that publication bias be prevented. In the example mentioned of the two treatments, it would be important for physicians to know that they were equivalent in effectiveness, because one might have fewer side effects or cost less, and with the information on effectiveness they could make a better decision on which to use. If data are obtained by a research group but never published, chances are great that few or no members of the scientific community will learn of the findings. Someone who is interested in a specific topic will most likely never obtain the unpublished results, which if considered might alter the overall picture given by the published studies. The remedy for publication bias is prevention, by setting the following standards. Researchers should submit the results of all studies to the appropriate journals, and journal editors should accept and publish studies based on the quality and logic displayed rather than the strength and direction of the results. Since these changes may be difficult to implement, perhaps a more practical approach would be to require registration of all studies before they begin. Currently, registers are in place for certain fields, such as cancer and AIDS (acquired immunodeficiency syndrome) research. (Consumer Summary produced by Reliance Medical Information, Inc.)

Medical literature, Scientific literature, Science literature

---

The communities of scientists and journal peer review

Article Abstract:

Peer review is the process by which scientific papers submitted to journals for publication are reviewed by experts in the field. The experts, or peer reviewers, are familiar with the topic area from their own work and thus can be considered peers of the authors. Reviewers are asked to evaluate the importance and usefulness of the research and judge whether it was performed carefully and accurately. Journal editors consider the reviewers' opinions in deciding whether to publish the submissions. Recently, some members of the scientific community have criticized the peer review process for lacking explicit standards; some claim the journals should make their peer review criteria and procedures known and open to evaluation. Other critics go further, stating that the entire peer review system should be abandoned because it favors conservative, traditional research and thereby excludes innovative work by independent scientists from publication. Only minimal attention has been paid the theory and practice of peer review until a recent conference, The First International Congress on Peer Review in Biomedical Publication, was held to encourage research regarding peer review. Peer review may be inherently difficult to analyze because it has no measurable outcome, only goals, such as sorting out the valid from the invalid studies. It also varies from journal to journal and from case to case, depending on the individual editors, authors, and reviewers involved; peer review is an inherently social process. Peer review can best be characterized as a discussion among competent and honest individuals who attempt to evaluate the information presented to them; it is not a machine that can identify truth in a rigid, constant manner, and perhaps scientists would not want it to be. It could only be beneficial to study peer review in a systematic way. With better understanding, more realistic expectations of peer review may develop. (Consumer Summary produced by Reliance Medical Information, Inc.)

---

An exploratory study of statistical assessment of papers published in the British Medical Journal

Article Abstract:

Research papers that are submitted to medical and scientific journals for publication may be evaluated by the journals in various ways. Based on the review process, journal editors decide whether to accept the manuscript as is, require revisions, or reject the manuscript. One aspect of the editorial process that receives varying amounts of attention from the editorial boards at different journals is statistical assessment. The British Medical Journal has consistently recognized the need for careful evaluation of the statistical procedures reported in manuscripts. The editors screen articles and recommend which should be submitted for a thorough review by a statistician. A study was performed to monitor how well this system is working at the British Medical Journal. The percentage of papers judged to be statistically acceptable at the time of first submission was only 11 percent, but had improved substantially to 84 percent after publication. These results were encouraging, but also indicated that the system could be improved, since several articles had not been properly revised but were still published. The statistical areas that were most frequently found unacceptable were sample size calculations, confidence intervals, and the method of randomization. (Consumer Summary produced by Reliance Medical Information, Inc.)

Medical statistics

Similar abstracts:

• Abstracts: Review of evidence on the early detection and treatment of breast cancer. Breast cancer detection and community practice: executive summary report of a workshop cosponsored by the General Motors Cancer Research Foundation and the American Cancer Society
• Abstracts: Seroprevalence and risk factors for HTLV-I/II infection among female prostitutes in the United States. Seroprevalence of HTLV-I and HTLV-II among intravenous drug users and persons in clinics for sexually transmitted diseases
• Abstracts: Long-term cost-effectiveness of various initial monotherapies for mild to moderate hypertension. Cost effectiveness of thrombolytic therapy with streptokinase in elderly patients with suspected acute myocardial infarction
• Abstracts: Intrauterine growth retardation and preterm delivery: prenatal risk factors in an indigent population. Smoking, maternal age, fetal growth, and gestational age at delivery
• Abstracts: Bacterial vaginosis: current review with indications for asymptomatic therapy. Statistical evaluation of diagnostic criteria for bacterial vaginosis

This website is not affiliated with document authors or copyright owners. This page is provided for informational purposes only. Unintentional errors are possible.