Our research study was approved by the Stanford University institutional review board (#43750). The data sets that support the findings of this study are publically available at the Harvard Dataverse: https://doi.org/10.7910/DVN/B5HJQX.

We randomly selected twelve journals from the InCites Journal Citation Reports (JCR) database in each of eight broad scientific (biology, biochemistry and molecular biology, microbiology, immunology, and cell biology) and clinical fields (cardiology, gastroenterology, oncology). The OpenEpi random number generator was used to select each numerically assigned journal from the InCites JCR website [24]. The basic and clinical sciences were chosen to encompass a wide range of primarily biomedical researchers. To be included in the study, the journal must have contained original research articles (as defined by each journal) and be either published in the United States or, in rare cases, have multiple US based editors. For each journal, we reviewed the instructions for authors and extracted key formatting requirements for publication including manuscript word limits, abstract word limits and structure requirements, maximum number of tables/figures, and citation limits. We also collected additional information on impact factor, cover letter requirements, and the number (if any) reviewers needed to be suggested. Upon completion of multiple reviews by members of the research team, we then contacted the editorial team (from online contact listings on each journal’s respective website) of journals included in this manuscript to verify the data we extracted and offer them an opportunity to comment on our assessments.

From each of the 96 journals selected, we randomly selected three recently published manuscripts. We again used a random number generator to pick each article, which was pre-designated a number. We picked articles starting from the most recent issue at that time of review (September—October 2017). The following inclusion criteria were used: 1) considered an original research article as defined by the journal, 2) publicly available contact information displayed for the first and/or corresponding authors and 3) the first author/corresponding author must be affiliated with a US-based institution. The latter criterion was required for communication purposes as regular mail was used to obtain a portion of surveys as described below.

All 288 first authors from the 96 journals randomly selected were sent an email with a link to complete an eight-question survey on reformatting for journal resubmissions (supporting material, S1 Text). These questions largely solicited information on the time spent on reformatting resubmissions, the reformatting categories responsible for the time spent, and the authors’ opinions on the resubmission process. It was explicitly stated in the survey that any estimates of time spent on reformatting should be not include any effort spent improving the actual scientific content of the manuscript.

Two email reminders to complete the survey were sent at weekly intervals to authors who did not respond. If the first author failed to respond after three total emails, the same initial email and two weekly reminder emails, as needed, were then sent to the listed corresponding author. If no response was received after emails to both the first and corresponding author, we sent a mailed copy of the survey with pre-paid return postage to the corresponding author with a nominal monetary incentive of $5 to be kept if they sent the completed survey back. We requested the money be returned if the survey was not completed. We accepted mailed survey responses up to one month after mailing. Those who completed the survey, either online or by mail, were entered in a drawing to win a $100 Amazon.com gift card.

To provide a rough, “back-of-the-envelope” calculation of the economic burden from reformatting, we used NIH research stipend data for first year postdoctoral scholars. This salary, $48,432, is consistent with previously published postdoctoral salaries, when adjusted for inflation to 2018 [25–26]. To determine the number of publications per year, we examined multiple previously published sources, all with several inherent limitations. The National Science Foundation (NSF), for example, estimates about 2.3 million publications worldwide yearly, but this estimate does include some non-medical engineering publications [27]. Another approximation cited in the literature indicates roughly 2.5 million annual publications [20, 28]. Other calculations, based on SCOPUS (Elsevier’s abstract and citation database) and Web of Science (Clarivate Analytics citation indexing service), have concluded there are about 2.2 million yearly research publications [29]. For the sole purpose of illustrating that the cost of reformatting is not insignificant, we multiplied the number of publications (based on NSF data) by the percent of articles requiring resubmission based on our survey data results. We then multiplied this number by the hours spent reformatting from our survey data and the hourly rate of a postdoctoral scholar as described above in order to determine a rough estimate of cost due to reformatting. For the hours spent on reformatting, we used the lower limits of time ranges shown in Fig 1A (1 hr, 4 hrs, 1 day, 3 days, 7 days) and Fig 1B (1 hr, 1 day, 3 days, 7 days, 14 days).

There was significant disparity in manuscript requirements across all journal types. We compared 12 key variables for manuscript publication across all 96 journals selected for review. To illustrate the heterogeneity of these manuscript format requirements, the results for microbiology and gastroenterology journals are shown in Table 1 as examples.

Even among journals with the same scientific focus, the requirements for publication were highly variable. For example, within gastroenterology journals, 7/12 format dimensions showed heterogeneity. Only 4/12 journals explicitly required an author contribution section and 8/12 required a cover letter. Manuscript word limits were only mentioned for 5/12 journals (and varied from 3500 to 6000 words). A request for suggested reviewers was made in 6/12 journals, 2/12 journals had specific table/figure limitations, and 4/12 had restrictions on the number of citations accepted. Additionally, while all of the gastroenterology journals required an abstract, the word limits for the abstract varied from 150 to 300 words.

For microbiology journals, the heterogeneity was even greater. For example, an author contribution section was required in only 4/12 journals. Suggested reviewers were mentioned in 8/12 publications. Standard paper body structure was either not required or not mentioned in 2/12. Though a structured abstract was commonplace in gastroenterology journals, this was not the case for microbiology, as 7/12 journals did not require standard abstracts. At the time of our review, only 4/96 (4%) of journals offered fully format-free initial submission.

We received a total of 206/288 responses (71.5%) to our survey of United States based biomedical researchers (Table 2). Mail-in responses accounted for about 20% of our total (41/206). First authors responded most often (158/206, 76.7%). Authors who were listed as both first and corresponding were categorized as first authors. The majority of manuscripts in our study (N = 118, 57.3%) were not accepted by the first journal to which it was submitted. Subgroup analysis comparing basic science and clinical journals revealed no significant differences in these resubmission rates, so this data was not included in manuscript.

Of note, we received eighteen email bounce backs from the initial email survey request. For six authors, we were able to publicly search for a replacement email address that was up to date. For the other twelve, which we were not able to find an updated email address, we re-selected twelve new articles in the same respective journals and resent the survey request email to the twelve new first authors with no further bounce backs.

The distribution of time spent reformatting by survey responder and by the entire research team is shown in Fig 1A and 1B. When asked how much time was needed for reformatting to all journals to which the paper was resubmitted to, the majority of authors (77/118, 65%; Fig 1A) reported that they spent 1–3 days or more (one day of effort was defined to the respondent as meaning eight hours). This did not include time spent on improving the scientific content or waiting for reviewer comments. Time spent on reformatting alone delayed resubmissions by over two weeks in most instances (60/118, 51%; Fig 1C). Though delay to publication did vary based on how many submissions were required before acceptance, 20% of submissions were postponed over three months due to format revisions alone.

Fig 2 shows the distribution of time spent by authors on common reformatting tasks such as word count, figures, references, manuscript structure, and publisher online submission requirements. Given the difficulty in quantifying exact hours on these tasks, authors were given options on a 0–4 effort scale ranging from none (0) to a great deal (4). These authors expressed that most time was spent on various journal/publisher online requirements (e.g., entering author contact information, disclosures, etc.) with 90/116 (78%) expressing that this took them some to a great deal of time. The only task that the majority of investigators felt was associated with minimal time expenditure (i.e. little, very little, or no time spent) was readjusting references (72/117, 62%).

Only 11.8% of 203 authors expressed satisfaction with the current resubmission process, with 59.3% reporting dissatisfaction (Fig 3C). 91.1% of authors favored a more streamlined approach to reformatting (Fig 3A).