Pre-clinical research has its main purpose in enhancing our understanding of physiologic and pathologic processes. However, many pre-clinical studies, in vivo animal experiments in particular, may also influence clinical care of patients through the following mechanisms: i) information for the design of clinical studies, ii) clinical guidelines considering pre-clinical evidence when clinical evidence is lacking, or iii) direct guidance of clinical practice. For instance, in the widely accepted Surviving Sepsis Guidelines recommendations regarding target arterial blood pressure for vasopressor use and positive end-expiratory pressure in the prevention of lung injury, draw on results from animal experiments . In addition, a recent empirical study showed that a substantial number of overviews of animal experiments investigating therapies for sepsis explicitly extrapolated results from animal studies to human patients and that subsequent citations of these overviews in publications of clinical studies reflect the influence of pre-clinical on clinical research .
Whenever a summary of the current evidence is needed for decision-making in clinical research, health-care policy, or clinical practice, systematic reviews (SRs) and, where appropriate, meta-analyses are most useful tools. SRs offer a structured and transparent way to comprehensively identify and evaluate the available evidence on a specific topic. Meta-analyses increase the precision and generalizability of effect estimates by quantitatively summarizing the results of individual studies included in an SR in order to provide a single best estimate with maximal statistical power . SRs and meta-analyses of pre-clinical studies are still relatively rare in the medical literature. Mignini and Khan identified 30 SRs of laboratory animal experiments in 2006, and Peters et al. found 86 using a more sensitive search strategy and a broader definition of laboratory animal experiments [4, 5]. However, in an update study Korevaar et al. reported that SRs of pre-clinical studies are getting more common over time, doubling their number roughly every 3 years between 1997 and 2010 .
The validity and usefulness of the results of SRs and meta-analyses strongly depend on their methodological rigor. Assessing bias, such as publication bias, is essential, when conducting SRs. If the publication or non-publication of research findings depends on the nature and direction of the results (definition of publication bias ) then the published studies are no longer a random sample of all studies that have been conducted, but constitute a biased sample leading to spurious summary results. Different methods have been developed to graphically and statistically investigate publication bias [8–12]. Although all these methods have their limitations one should still try to address the likelihood of publication bias in each SR as thorough as possible [9, 13].
While principles of critically appraising SRs of clinical research are well established, their application to SRs of pre-clinical studies appears variable . Previous empirical studies on SRs of animal experiments found that only 16% (5/30) and 37% (17/46) of meta-analyses considered publication bias [4, 5]. Korevaar et al. reported that between 2005 and 2010 the proportion of meta-analyses of in vivo animal studies that assessed publication bias increased to 60% (21/35) . A recently published survey conducted in animal laboratories in the Netherlands reported that just about 50% of animal experiments are published; employees of for-profit organizations even estimated that only 10% of animal experiments appear in peer-reviewed journals . Lack of statistical significance was identified as one of several important reasons for non-publication.
Korevaar et al. conducted their search for SRs of animal experiments in 2009/10 . It remains unclear how the number and methodological quality, in particular assessment of publication bias, of such SRs evolved up to the present (January 2013), and to what extent they are cited by clinical studies. We will therefore conduct a comprehensive SR summarizing the results from Peters et al. and Korevaar et al. and updating the evidence base for SRs of animal experiments. This SR will be part of the project, To Overcome failure to Publish nEgative fiNdings (OPEN Project), which was developed with the goal of elucidating the scope of publication bias and non-publication of studies through a series of SRs and policy evaluations (http://www.open-project.eu).
The specific goals of the present SR of animal studies are: determining the number of published SRs of animal studies up to the present; investigating methodological features of SRs and meta-analyses of animal studies especially with respect to assessment of publication bias, and investigating the influence of SRs of animal studies on clinical research by examining citations of the SRs in clinical studies.