Publishing in the Journal of Clinical Sleep Medicine: tips from an associate editor

Citation:

Jindal RD. Publishing in the Journal of Clinical Sleep Medicine: tips from an associate editor. J Clin Sleep Med. 2020;16(7):997–998.

I have served the Journal of Clinical Sleep Medicine as an associate editor for approximately 3 years. It is a privilege to have a front-row seat as the journal goes about fulfilling its charter of meeting the needs of readership. At the Journal of Clinical Sleep Medicine, we focus on publishing evidence that either affirms current clinical practice or supports change. We do, however, make an exception for the publication of exciting exploratory work, especially on topics of high public health importance. Although not all exciting early work will eventually translate into new standards of patient care, it is valuable for us to inform our readers about what may be on the horizon.

It is a cliché but true that we want everyone’s manuscript to be published. It is disappointing to issue a rejection, especially when it is clear that the authors have put in much time, skill, and effort. We encourage contributions from authors across the spectrum, including early-career, midcareer, and independent researchers. We also want our readers to learn from the wisdom of clinicians on the front lines who may not have advanced research training. For potential contributors with limited experience in publishing, I further discuss how we prioritize manuscripts.

We appreciate that not all our readers are trained clinician-scientists. Because an untrained eye may not fully appreciate the nuances of a hypothesis testing vs post hoc analysis, it is helpful when our contributors use language that allows readers to accurately appraise the evidence presented. In this regard, our motivation is not different from that of institutional review boards, which mandate that the language used in consent forms is understandable to a person with an eighth-grade education. The language that we like to see in our published manuscripts is such that is understandable to physicians, nurse practitioners, physician assistants, psychologists (with a PsyD degree), and others who may or may not have formal research training.

We prioritize studies that prove or disprove beyond a reasonable doubt a thought-out hypothesis. For clinical trials, we require that the trial was registered on a trial registry. The list of acceptable registries is as follows: the Australian New Zealand Clinical Trials Registry (http://anzctr.org.au/), the Chinese Clinical Trial Registry (http://www.ChiCTR.org.cn), ClinicalTrials.gov (http://www.clinicaltrials.gov), the Clinical Trials Registry—India (http://ctri.nic.in/Clinicaltrials/login.php), the German Clinical Trials Register (http://www.germanctr.de), the ISRCTN Registry (http://isrctn.org), the Netherlands Trial Register (http://www.trialregister.nl), and the UMIN Clinical Trials Registry (http://www.umin.ac.jp/ctr).

Occasionally, our potential contributors inform us that their research office did not recommend registration for a given study. A ruling from their local authority is not sufficient. If a trial was not registered, then the authors need to send a presubmission query to the editor. The editor may waive the need for preregistration in exceptional circumstances, but the authors will still need to make a well-reasoned argument for exception in the body of the manuscript.

Apart from providing a prospective time stamp, a clinical trial that is preregistered helps reviewers make a clear distinction between primary and secondary endpoints. When multiple primary endpoints are listed on the website for a particular clinical trial, one can quickly check whether corrections for multiple comparisons are reflected in the interpretation of the primary findings.

While designing a clinical trial, it is tempting to have more than 1 primary endpoint, but the downside is the loss of statistical power to test the hypothesis. Even when the robustness of preliminary data allows sufficient statistical power for multiple primary endpoints for the subsequent definitive trial, there are caveats. Having a change in one endpoint and not the other/others makes it difficult to choose the variable that would drive clinical care. A possible exception would be when more than 1 clinical outcome is independently desirable. For example, in a clinical trial investigating the utility of modifying sleep duration, both glycemic control and blood pressure could be independently meaningful variables of interest.

A common drawback with the manuscripts that receive lower ranking is that causal inferences are made imprecisely. It is important to limit causal inferences to instances when the evidence for association is coupled with evidence that establishes temporal precedence and rules out alternative explanations.¹ Therefore, longitudinal studies are needed to establish an intermediary variable or mediator in a purported mechanism. Cross-sectional datasets can provide hypothesis-generating information on possible mediators, but it is important to clarify that the hypothesis needs to be tested in longitudinal studies for it to be clinically meaningful. Cross-sectional datasets can provide clinically meaningful information when there is evidence of association between variables in 1 subpopulation but not the other, or when the association is stronger in 1 subpopulation.² Two published articles from the Multi-Ethnic Study of Atherosclerosis seem to clarify this point well. In 1 article, an association of body mass index and waist circumference with apnea-hypopnea index (AHI) was described to be stronger among Chinese Americans than among other racial groups.³ In the second, AHI was associated with left ventricular mass in those younger than age 65 years but not in older individuals.⁴ The association between AHI and left ventricular mass was stronger when 3 covariates—diabetes, systolic blood pressure, and antihypertension medications—were not included. Because race and age distinguished subpopulations at differential risk, the first 2 findings have potential clinical implications. However, the third of the aforementioned findings only generated a hypothesis that diabetes and hypertension are potential mediating variables. Testing this hypothesis in a longitudinal cohort can answer clinically consequential “how” and “when” questions. It is important to clarify whether one is testing for a moderator, which potentially distinguishes a subpopulation, or a mediator, which is meant to unravel mechanisms, because a moderator does not need longitudinal studies to yield “actionable intelligence” for a clinician but a mediator does. These are, of course, general guidelines and it is possible to conjure up situations in which the risk-benefit calculus could prompt a clinical decision based solely on preliminary evidence. Nevertheless, it is important that the authors provide nuance and not try to over-sell their findings.

Once an association is found, excluding alternative explanations is as important as establishing temporal precedence, if not more. We have seen manuscripts in which potential contributors argued that the study’s uncontrolled switch from one intervention to the other has clinical implications. An acknowledgment that “absent a control group, the findings could simply be a regression to mean” could have won support from the reviewers. If an author is somehow able to win acceptance without fully acknowledging the limitations, then the editor will have a stronger need for an accompanying commentary that puts the results in context. I think it is better form for authors to describe the limitations of their work—to stay ahead of their story—and not invite a critical commentary.

For presentations of preliminary work, we appreciate when the exploratory nature of the study is reflected in the title, abstract, discussion, and conclusions. For secondary analysis, it is important to explicitly acknowledge the likelihood of false positivity for a given observation and spell out the need for confirmatory studies before adoption to routine clinical practice.

In sum, although our goal is to publish studies that establish standards of clinical care, we appreciate that evidence needs to be stacked “brick by brick” for it to become compelling. Providing the context for observations in a neutral tone with appropriate acknowledgment of the limitations of a given study and putting the results in perspective to existing standards of care are attributes we hold in high regard for manuscripts submitted to our journal.

DISCLOSURE STATEMENT

The author reports no conflicts of interest.