Most medical journals do not have a mechanism for producing well-written evidence-based medicine on a measured monthly basis. Some published articles are poorly written and some have few usable clinical tidbits. The temperament of different editors and editorial boards may vary from seeking sensational content that may be risky and shortly abandoned to publishing only methodical proven science. In fact, some view journals as a place to deposit only vetted and established work for posterity. The AJO, as do many journals, occasionally publishes controversial or novel topics and may provide an editorial to challenge the content, or offers readers the opportunity to challenge content through the correspondence section. In reality, most peer-reviewed journals seek more trust than debate and seek more science than sensationalism or entertainment, although we must sell subscriptions to continue our mission. We recognize that there are other current venues for less strict science reporting and more debating. Publication does not imply editorial support, as some articles in the same issue may have contradictory views. The medical literature, moreover, is a history of many ineffective and potentially dangerous treatments, which ultimately are supplanted by better science. At best, published articles represent provisional truth or a continuation of a hypothesis, and only a few articles survive as applicable information into the next decades.
Readers assume that articles published in peer-reviewed journals are scientifically valid, but there is sufficient evidence to the contrary. Peer review is used specifically to detect errors, but it is not foolproof, even among the best journals. Common errors in articles include poor research methods, misused or incorrect statistical analysis, avoidance of alternate explanations, overgeneralization of results, selective reporting, and fraud, but the most common problems are methodological errors (using the wrong techniques) or study design errors (sometimes deliberately, to create an effect). Altman calls these scandals and attributes them to poor quality control and the publish-or-perish climate. Readers may not understand the methodology and depend on the peer reviewers to sort out errors; readers trust that someone has looked at the data and judged it as scientifically valid. Even randomized controlled trials have considerable errors, despite the costs and important clinical implications. Institutional review boards (IRBs) have a role here in assuring valid studies are set up, but many IRBs do not even have statisticians on their boards and are dealing more intensely with ethical issues than with study design (yet it is unethical to do poor research). To extend this problem further, the published practice guidelines depend on the published literature.
Poor research is the fault of the authors or sponsors, but journals are responsible for the integrity of the peer-reviewed literature. Peer review attempts to expose these errors in manuscripts but is dependent on the few individual reviewers. Many manuscripts are not reviewed by the best in the field, since these experts are frequently overwhelmed and many conserve their time for their own research. Specifically, expert methodological review is difficult to obtain on a consistent basis and an experienced methodologist might even reject most manuscripts submitted. Most journals do not provide an independent statistical analysis of each accepted manuscript, so most are probably admittedly performing inadequate peer review. Authors are always encouraged to publish more complete data as supplemental online content, which many journals now provide, and thus the authors might anticipate and answer some criticisms about their research. Not many take advantage of this added venue to further clarify their research.
The plea here is for readers to assist in postpublication peer review by way of correspondence in an attempt to correct the literature and ensure the accountability of authors and editors. The International Committee of Journal Medical Editors agrees that editors should correct the literature by critical critique of the articles through correspondence and then by publishing corrections or retractions. Correspondence is the hallmark of peer-reviewed journals and also provides interesting and lively information for the journal and its readers. Correspondence can challenge ideas that have gone through the peer-review process, correct mistakes, and initiate a dialogue between researchers and clinicians. In fact, most letters are accepted if they meet the journal guidelines and editors wisely do not usually interject their own opinions or prejudices into the letters. Over the recent 3-year period for the AJO, however, there were only 167 letters received and 118 (71%) were accepted.
The intellectual challenge of preparing a critical evaluation of published research is a valuable exercise, especially for resident or fellowship journal clubs. Indeed, formal letter writing to journals is now used in some medical curricula to teach critical appraisal skills. Better indexing and linkage to the original paper permits easier retrieval and correction of the literature. Unfortunately, the clarifying correspondence might never receive the same due as the original (faulty) research.
Admittedly, journals differ to a great extent in what type of content they permit in letters; some permit original research or case reports in this format. The AJO agrees with Squires and has chosen to have strict criteria for this correspondence section since it is a peer-reviewed journal and therefore should not provide an easy path to publish incomplete data, uncertain or vague hypotheses, case reports with uncertain associations or insufficient new information, and pointless venting or, even worse, social statements unrelated to the specific content of the journal. The correspondence in this journal does not go through rigorous peer review, but the authors of the original article are usually required to reply to the correspondence and the pieces then appear in tandem. Admittedly, the last word usually defaults to the original author. If the authors do not reply to a specific editor’s request, then the readers are notified of this failure to participate or correct.
More specifically, correspondence to the AJO should follow the principles as outlined by Brown: the criticism or correction should relate to an article recently published in the Journal (older content may be challenged but requires more justification); should be written clearly and succinctly; should contain data, facts, or references that support the critique; and should avoid inflammatory comments. Unpublished data or study results and promotion of products should be avoided. The editor may limit duplicative or repetitive correspondence. The issue comes when the correspondence seeks to publish research that has not been peer reviewed, or that has been rejected somewhere else during peer review. If published, it can be cited with seemingly the same validity as a full peer-reviewed article. Weak or incomplete research is worse than no research as it confuses the literature. Correspondence does not usually undergo peer review as the format does not permit a full presentation of the data and discussion of the contrast with existing literature. Some suggest that correspondence should be placed online only for a traditional journal, but the online content is now assuming the role of the final version of a journal and is usually citable. Editors and authors must therefore work diligently to ensure that authors respond fully to important criticism about their work; they frequently do not respond completely to the critical correspondence.
Any content produced by a peer-reviewed journal should be of high quality whether it is an article or a correspondence and whether it is online or in print, as it is now usually citable. There is no shortage of venues for content or opinions in ophthalmology (which is not necessarily a good thing in medicine) and the peer-reviewed journal should be more restrictive/selective. Correspondence regarding articles in the peer-review literature is a neglected form of critical appraisal. Journals need this venue to express critiques, corrections, and alternate opinions, and to shape better knowledge through intellectual challenge. Publication should be the start of the reader peer-review process, since many readers possess the critical skills to provide enhanced knowledge regarding the content or interpretation of studies at hand and can detect faulty data as well as counteract inappropriate spin or propaganda on the part of the original author.