Medical research reports and publication practice in biomedicine have been under increased scrutiny over the last decades. Selective reporting of study results and related publication bias has been confirmed in several empirical studies in different disciplines and settings[1]. Further, there is continued concern regarding conflicts of interest that are not disclosed by article authors [2] and cases of scientific misconduct[3]. Non-reporting of study results has been identified also in pediatric research[4]. For instance, significant differences in the risk-benefit profile of selective serotonin re-uptake inhibitors (SSRIs) in children were found in a meta-analysis of studies published in peer-reviewed journals and unpublished data[5]. The data suggesting that SSRIs are linked to an increased risk of suicide or suicidal thoughts had not been published[6].

In the past, several recommendations have been proposed to improve the reporting and publication practice in biomedicine: First, the International Committee of Medical Journal Editors (ICMJE) published the "Uniform Requirements for Manuscripts submitted to Biomedical Journals"[7]. This widely used guideline is currently endorsed by over 700 journals and covers issues such as ethical conduct and reporting of biomedical research, preparation and publishing of manuscripts and editorial policies. Second, the problem of publication bias and selective outcome reporting has been widely analyzed over the last 15 years[1]. Registration of clinical trials and studies of other types prior to patient enrollment has been advocated as an important first step to tackle this problem. Third, authors but also journal editors and reviewers might have financial ties or personal interests in conflict with an article being submitted for publication[8, 9]. The debate about this problem has led journals to require disclosure of potential conflicts of interest. Recently, a uniform conflict of interest disclosure form was proposed jointly by major medical journals[10]. Finally, published reporting guidelines such as the CONSORT Statement provide guidance to authors and aim at improving the completeness and accuracy of publications[11, 12]. Further, they facilitate the critical appraisal by readers. Endorsement and implementation of these reporting guidelines has been studied for general medicine journals [13, 14] but less so for journals in specialties such as pediatrics.

Journals and their editors play a key role in promoting and ensuring transparency in biomedical publishing. Previously, we focused on pediatric journals indexed in the Journal Citation Report and found that the advice given to authors regarding the above mentioned four domains was moderate to low[15]. Reflecting these findings, we wondered whether the low uptake might be due to hesitation or even reluctance of editors of these journals to experiment with editorial procedures.

Open-access electronic publishing has been associated with a number of editorial innovations aiming at improved access to and transparency of research results[16, 17]. The new model was a response to the dilemma between increasing prices for journal subscriptions on one side and decreasing resources of academic institutions to finance access to the scientific literature. Open Access publications are generally made available online to anyone anywhere with no charges for access while recovering costs by charging publication fees from authors. Open Access journals usually provide peer review like journals following the conventional publishing model. It has been argued that electronic Open Access publishing does not change significantly content and quality of research articles but improves access to research findings[18].

We wondered whether journals adopting this new publication model take up recommendations which aim to ensure publication of research results in an unbiased and transparent manner. We therefore set out to elucidate the coverage of the four domains Uniform Requirements, trial registration, conflicts of interest and reporting guidelines in Open Access pediatric journals. We then compared our results with findings from "conventional" JCR-indexed pediatric journals analyzed earlier[15].

We accessed the Directory of Open Access Journals http://www.doaj.org on 4^th of September 2009, identified 43 journals listed in the "Health Sciences - Medicine (General) - Pediatrics" category and extracted information on their start year and publication language. We excluded one journal that does not publish original research articles (Foro Pediátrico) and another journal (Pediatric Cardiology Today) because it was continued as one of the included journals (Congenital Cardiology Today) in 2005. From the websites of the 41 included journals we downloaded the author instructions in September 2009. Two authors (JJM and RFW) read each document and classified information about the geographical location of the main editorial office using the following groups: Africa, Australasia, Europe (without UK), United Kingdom (UK), North America, and South America. Further, we defined three categories of publishers: open access publishing houses, professional societies/academic institutions and other publishers. Using relevant keywords in electronic full text searches we then identified any information on the following four domains of good publication practice:

For each of these items, we analyzed the wording of the author instructions and determined whether it was "required" (i.e. a submitted manuscript would not be accepted if the item was not considered) or "recommended" (i.e. its use or fulfillment ought to be considered). Two investigators (JJM, RFW) extracted and categorized information independently. Discrepancies occurred in less than 1% of items and were all resolved by discussion among the investigators.

For comparison, data from a previous study on 69 pediatric journals indexed in the Journal Citation Report (JCR) were used[15].

Only one journal, "Revista Chilena de Pediatria", started publishing in the 1980s. Six journals started in the 1990s, while most of them (n = 34) started publishing in 2000 or later. The majority are published in English (n = 23), while 6 are published in Spanish, 4 in Turkish and 2 in Portuguese. Six journals publish their articles in two or more different languages with all but one publishing in English.

The interpretation of these data and the comparison with our previous survey of author instructions [15] need to consider several aspects. First, the number of journals examined was not very large with 41 Open Access journals and 69 journals indexed in the Journal Citation Report. However, both selections did not represent samples drawn from a larger group but an analysis of all journals meeting the entry criteria. Second, clustering of journals that are run by the same publisher might have influenced our results. We aimed to address this by analysing the data by type of publisher. Third, our analysis was based on the policies as documented in the online instructions that potential authors would consult as first reference. We did not determine whether the editorial staff of the journals applies additional procedures that are not reflected in the author instructions e.g. to encourage or even enforce adherence of authors to the promoted policies. Such additional procedures could only be identified by surveying editorial staff to elucidate their motivations to implement new policies or any barriers to do so. However, journal procedures were not the interest of our current study. Finally, our study is of exploratory nature. We are not aware of any other studies on author instructions of Open Access journals or comparisons with journals with another publishing model. One might speculate that the situation is similar in journals in general medicine or other specialties. Additional studies including other than pediatric journals should be undertaken to determine whether our results can be generalized to other Open Access journals.

Several of the included Open Access journals are more recent than the journals that are indexed in the Journal Citation Report. When they set up their author instructions they might have sought guidance on the most recent developments in good publication practice and consequently included recommendations on the four domains investigated in this survey from the start.

Although the empirical evidence on improvement of publication practice due to endorsement of these recommendations is still limited, providing such recommendations in the author instructions might be considered a means to enhance reporting quality by newly founded or less well-established journals[16, 29].

Several of the included journals were run by publishers dedicated to Open Access publishing. This group of journals offered the most extensive guidance to their authors. It is conceivable that these publishers provided a master copy of author instructions to be adapted by the journal's editorial team. This circumstance might have contributed to the higher proportion of guidance that is offered by these OA journals.

Finally, our analysis of author guidelines could be complemented by an assessment of the content published by the Open Access journals. Previous empirical studies have looked at the quality of published articles, in particular before and after introduction of reporting guidelines, but did not focus on different publishing models[11, 30].

Pediatric OA journals are a heterogeneous group of journals ranging from journals run by national societies to new publications launched by dedicated OA publishers. Overall, they do give at least as much guidance to authors as do conventional pediatric journals. Interestingly, differences exist between types of publishers. However, the uptake of good publication practices could still be improved in all groups. Whether these results can be generalized to Open Access journals in other specialties or in general medicine needs to be investigated. Reasons for differences should be elucidated e.g. by surveys of journals editors.