Reporting animal research: Explanation and elaboration for the ARRIVE guidelines 2.0

In the paper

“However, evidence shows that the majority of publications fail to include key information and there is significant scope to improve the reporting of studies involving animal research [1–4].”

What the sources say

• Supported

  The Results (Quality of Reporting; page not available) show that only 43% of the 271 surveyed publications reported animal age and 46% reported weight—details the Discussion (Animal Characteristics) calls scientifically important. The Abstract says the survey identified issues requiring improvement in reporting. The findings concern surveyed rodent and primate studies from UK and US laboratories, rather than all animal research publications. (LLM, conf=0.88)

  cites [1] C Kilkenny, 2009, Survey of the quality of experimental design, statistical analysis and reporting of research using animals, PLoS ONE source
• Partly supported

  The Results (‘Reporting of sex and age’; page not available) say that only about 50% of mouse-model papers published since 2012 reported both variables. The Discussion identifies this as a reporting problem, but the study does not establish that a majority of animal-research publications generally omit key information. (LLM, conf=0.86)

  cites [2] O Florez-Vargas, 2016, Bias in the reporting of sex and age in biomedical research on mouse models, Elife sourceExperimental — accuracy not yet measured
• Supported

  The Abstract reports that papers in the review routinely omitted essential statistical information, including 95% of papers using ANOVA lacking enough detail to identify the test. The section “Papers rarely contain information needed to verify the test result” also reports that more than two thirds of papers using animal models of stroke or cancer lacked sufficient information to determine whether animals were excluded. These findings support the broad claim that reporting in animal research has substantial room for improvement. (LLM, conf=0.88)

  cites [3] TL Weissgerber, 2018, Why we need to report more than ‘Data were Analyzed by t-tests or ANOVA’, Elife source
• Supported

  The paper reports that over two-thirds of more than 1,000 publications from leading UK institutions did not report even one of four measures considered critical to reducing bias. It also concludes that there is substantial room to improve reporting of in vivo research (passages [1] and [4]; section labelled Abstract, no page supplied). (LLM, conf=0.94)

  cites [4] MR Macleod, 2015, Risk of bias in reports of in vivo research: a focus for improvement, PLoS Biol source

In the paper

“Reporting n = 50 in this case is an example of pseudoreplication [26].”

What the sources say

• Supported

  The Background defines n as the number of experimental units and says that multiple observations from one animal cannot be added together to obtain the sample size. Its Figure 1 example calls treating 20 measurements from two rats as independent an example of pseudoreplication. Thus reporting n = 50 is pseudoreplication if the 50 are observations from fewer than 50 mice, rather than 50 independent mice. (LLM, conf=0.88)

  cites [26] SE Lazic, 2010, The problem of pseudoreplication in neuroscientific studies: is it affecting your analysis?, BMC Neurosci source

In the paper

“The use of a one-sided test requires justification of why a treatment effect is only of interest when it is in a defined direction and why they would treat a large effect in the unexpected direction no differently from a nonsignificant difference [43].”

What the sources say

• Supported

  Split into 2 atoms — 2 confirmed, 0 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

  cites [43] GD Ruxton, 2010, When should we use one-tailed hypothesis testing?, Methods Ecol Evol source

In the paper

“The Experimental Design Assistant (EDA) is a platform to support researchers in the design of in vivo experiments; it can be used to generate diagrams to represent any type of experimental design [19].”

What the sources say

• Partly supported

  The Introduction describes the EDA as a web-based tool that guides researchers in designing animal experiments and produces diagrams of their plans. The section on standard notation describes diagrams for different designs, but our copy does not establish the broader claim that it can represent any type of experimental design. (LLM, conf=0.82)

  cites [19] N Percie du Sert, 2017, The Experimental Design Assistant, PLoS Biol sourceExperimental — accuracy not yet measured

In the paper

“For hypothesis-testing experiments, in which inferential statistics are used to estimate the size of the effect and to determine the weight of evidence against the null hypothesis, the sample size needs to be justified to ensure experiments are of an optimal size to test the research question [33,34] (see Item 13. Objectives).”

What the sources say

• ?Not checked — no full text of the sourcecites [33] F Vahidy, 2016, Reporting standards for preclinical studies of stroke therapy, Stroke source
• Supported

  The “Sample Size Issues” section says that calculating sample size before an RCT ensures sufficient statistical power and recommends that animal researchers report how they determined sample size a priori. The recommendations table likewise calls for justification of sample size selection based on a prespecified primary outcome. (LLM, conf=0.90)

  cites [34] BS Muhlhausler, 2013, Whole animal experiments should be more like human randomized controlled trials, PLoS Biol source

In the paper

“and there is evidence that in vivo scientists are not sufficiently aware of the importance of reporting the information covered in the guidelines and fail to appreciate the relevance to their work or their research field [12].”

What the sources say

• Supported

  In “Implementation of ARRIVE,” the authors report that many authors paid no attention to the guidelines and that some did not see the point of particular requirements for their work. In “Have our efforts made a difference?” and the subsequent discussion, they describe colleagues who considered the effort a waste of time and authors who claimed compliance without fully following the guidelines. These passages support the cited statement about insufficient appreciation of the reporting requirements and their relevance. (LLM, conf=0.88)

  cites [12] JC McGrath, 2015, Implementing guidelines on reporting research using animals (ARRIVE etc.): new requirements for publication in BJP, Br J Pharmacol source

In the paper

“Consequently, low-powered studies contribute to the poor internal validity of research and risk wasting animals used in inconclusive research [37].”

What the sources say

• Supported

  In the Hanno Würbel section (page unavailable), the paper identifies samples that are too small as a threat to scientific validity and includes sample-size calculation in its assessment of internal validity. In “Bumping up the benefits” (page unavailable), it says animals may be wasted when a study produces inconclusive results. (LLM, conf=0.90)

  cites [37] H Würbel, 2017, More than 3Rs: The importance of scientific validity for harm-benefit analysis of animal research, Lab animal source

In the paper

“To ensure scientific rigour, the criteria should be defined before the experiment starts and data are collected [8,33,48,49].”

What the sources say

• Partly supported

  The Results section identifies inclusion and exclusion criteria as core reporting items, and the Discussion links rigorous implementation and reporting of such measures to reduced risk of bias. Our copy does not say that these criteria must be defined before the experiment starts or before data collection; that timing is not supported by this source. (LLM, conf=0.82)

  cites [8] MT Avey, 2016, The devil is in the details: incomplete reporting in preclinical animal research, PLoS ONE sourceExperimental — accuracy not yet measured
• ?Not checked — no full text of the sourcecites [33] F Vahidy, 2016, Reporting standards for preclinical studies of stroke therapy, Stroke source
• ?Not checked — no full text of the sourcecites [48] ASC Rice, 2013, Transparency in the reporting of in vivo pre-clinical pain research: The relevance and implications of the ARRIVE (Animal Research: Reporting In Vivo Experiments) guidelines, Scandinavian Journal of Pain source
• ?Not checked — no full text of the sourcecites [49] Salkind NJ. Encyclopedia of research design: Sage; 2010. source

In the paper

“To ensure that the recommendations are as clear and useful as possible to the target audience, this document was road tested alongside the revised guidelines with researchers preparing manuscripts describing in vivo research [13].”

What the sources say

• Supported

  The ‘Introducing ARRIVE 2.0’ section states that the revised guidelines and accompanying Explanation and Elaboration document were road tested with researchers preparing manuscripts describing in vivo studies. It also gives the stated purpose: checking that the documents were understood and useful to their intended users. (LLM, conf=0.95)

  cites [13] N Percie du Sert, 2020, The ARRIVE guidelines 2.0: Updated guidelines for reporting animal research, PLoS Biol source

In the paper

“Statistical programmes to help perform a priori sample size calculations exist for a variety of experimental designs and statistical analyses, both freeware (web-based applets and functions in R) and commercial software [38–40].”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites [38] R Core Team, 2017
• ?Not checked — no full text of the sourcecites [39] C-YJ Peng, 2012, Power analysis software for educational researchers, The Journal of Experimental Education source
• Supported

  The passage on software for sample-size calculation (section labeled Abstract) recommends freely downloadable G*Power, points to other free calculators, and names nQuery Advisor and MINITAB for more complex calculations. It also discusses calculations for different statistical tests, including t-tests, chi-square tests, ANOVA and non-parametric tests. This supports the attributed claim that free and commercial programs are available for a variety of analyses. (LLM, conf=0.90)

  cites [40] J Charan, 2013, How to calculate sample size in animal studies?, J Pharmacol Pharmacother source

In the paper

“It underestimates the true variability in a study, which can lead to false positives and invalidate the analysis and resulting conclusions [26,27].”

What the sources say

• Supported

  In the Background section on false precision, the paper explains that pseudoreplication produces an incorrectly narrow error bar and confidence interval, understating uncertainty about the population. The same section reports increased false-positive rates and says analyses that ignore this data structure can be meaningless and lead to false conclusions. (LLM, conf=0.93)

  cites [26] SE Lazic, 2010, The problem of pseudoreplication in neuroscientific studies: is it affecting your analysis?, BMC Neurosci source
• ?Not checked — no full text of the sourcecites [27] SH Hurlbert, 1984, Pseudoreplication and the design of ecological field experiments, Ecol Monogr source

In the paper

“Missing data, common in all areas of research, can impact the sensitivity of the study and also lead to biased estimates, distorted power, and loss of information if the missing values are not random [54].”

What the sources say

• Partly supported

  The Abstract (page not provided) says missing data reduce statistical power and can bias parameter estimates; the Missing completely at random section adds that power can be lost even when estimates remain unbiased. The paper also describes lost data, but it does not say missing data affect the study’s sensitivity. Its Sensitivity analysis section discusses a method for testing the robustness of results, not a loss of study sensitivity. (LLM, conf=0.84)

  cites [54] H Kang, 2013, The prevention and handling of the missing data, Korean J Anesthesiol sourceExperimental — accuracy not yet measured

In the paper

“In a hypothesis-testing study powered to detect an effect on the primary outcome measure, data on secondary outcomes are used to evaluate additional effects of the intervention, but subsequent statistical analysis of secondary outcome measures may be underpowered, making results and interpretation less reliable [88,93].”

What the sources say

• ?Not checked — no full text of the sourcecites [88] LK John, 2012, Measuring the prevalence of questionable research practices with incentives for truth telling, Psychol Sci source
• Partly supported

  Split into 2 atoms — 0 confirmed, 2 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

  cites [93] SC Landis, 2012, A call for transparent reporting to optimize the predictive value of preclinical research, Nature sourceExperimental — accuracy not yet measured

In the paper

“Registering a protocol in advance protects the researcher against concerns about selective outcome reporting (also known as data dredging or p-hacking) and provides evidence that the primary outcome reported in the manuscript accurately reflects what was planned [95] (see Item 19. Protocol registration).”

What the sources say

• Supported

  In “Reporting and dissemination,” the paper says that pre-registering a study with its primary outcome specified in advance addresses outcome switching and P-hacking. The advance specification provides a record against which the manuscript’s reported primary outcome can be compared; the paper does not suggest that registration guarantees adherence. (LLM, conf=0.89)

  cites [95] MR Munafò, 2017, A manifesto for reproducible science, Nature Human Behaviour source

In the paper

“Analysing the data using different methods and selectively reporting those with statistically significant results constitutes p-hacking and introduces bias in the literature [90,94].”

What the sources say

• Supported

  The Introduction describes using different statistical approaches but reporting only the analysis with the best results, calling this selective analysis and outcome reporting bias. It says these biases inflate the proportion of published statistically significant findings. Although the paper does not use the term “p-hacking” in the quoted passage, it describes the practice the claim names. (LLM, conf=0.92)

  cites [90] KK Tsilidis, 2013, Evaluation of excess significance bias in animal studies of neurological diseases, PLoS Biol source
• Supported

  The paper lists post-analysis choices, including changing which covariates to include and stopping data exploration when an analysis yields a significant p-value, as p-hacking practices (introductory discussion; page not provided). Box 3 calls for reporting the entire analysis process rather than only the final tests, and the Summary and Conclusions says p-hacking has probably inflated estimated mean effect sizes. Together, these support the claim that selectively reporting significant results from different analyses is p-hacking and biases the literature. (LLM, conf=0.90)

  cites [94] ML Head, 2015, The extent and consequences of p-hacking in science, PLoS Biol source

In the paper

“The immune or microbiological status of the animals can also influence welfare, experimental variability, and scientific outcomes [112–114].”

What the sources say

• Supported

  The opening passage (Abstract) states that animals’ microbiological quality directly influences welfare, experimental variability, and scientific research projects. The discussion of infections (Abstract) also says they can confound scientific results and increase experimental variability. (LLM, conf=0.93)

  cites [112] M Mahler Convenor, 2014, FELASA recommendations for the health monitoring of mouse, rat, hamster, guinea pig and rabbit colonies in breeding and experimental units, Lab Anim source
• ?Not checked — no full text of the sourcecites [113] DG Baker, 1998, Natural pathogens of laboratory mice, rats, and rabbits and their effects on research, Clin Microbiol Rev source
• Supported

  The Abstract identifies gut microbiota as a source of variation in animal experiments and links understanding that variation to reproducibility. The Results report that microbiota transplants reproduced differences in Salmonella susceptibility and that one donor microbiota led to greater lethal morbidity—a welfare-relevant outcome. Together, these passages support the claim that animals’ microbiological status can affect welfare, experimental variability, and scientific outcomes. (LLM, conf=0.88)

  cites [114] EM Velazquez, 2019, Endogenous Enterobacteriaceae underlie variation in susceptibility to Salmonella infection, Nat Microbiol source

In the paper

“The animals’ provenance, their health or immune status, and their history of previous testing or procedures can influence their physiology and behaviour, as well as their response to treatments, and thus impact on study outcomes.”

What the sources say

• ?Not checked — no full text of the sourcecites [16] MF Festing, 2002, Guidelines for the design and statistical analysis of experiments using laboratory animals, ILAR J source
• Partly supported

  The paper’s introductory passage (indexed as Abstract; page not given) says infections can affect animals’ physiology, including behaviour and immune response, and can confound experimental results. Its health-monitoring discussion also considers immune status and the health history of the facility of origin. It does not establish that previous testing or procedures alter animals’ physiology, behaviour, or responses to treatments. (LLM, conf=0.85)

  cites [112] M Mahler Convenor, 2014, FELASA recommendations for the health monitoring of mouse, rat, hamster, guinea pig and rabbit colonies in breeding and experimental units, Lab Anim sourceExperimental — accuracy not yet measured
• ?Not checked — no full text of the sourcecites [113] DG Baker, 1998, Natural pathogens of laboratory mice, rats, and rabbits and their effects on research, Clin Microbiol Rev source
• Partly supported

  The Results show that mice from different vendors had different susceptibility to Salmonella infection, and the Abstract links that variation to their gut microbiota and to reproducibility of animal experiments. The cited paper does not establish the claim about prior testing or procedures, behaviour, or responses to treatments generally. The copy provides section labels but no page numbers. (LLM, conf=0.88)

  cites [114] EM Velazquez, 2019, Endogenous Enterobacteriaceae underlie variation in susceptibility to Salmonella infection, Nat Microbiol sourceExperimental — accuracy not yet measured

In the paper

“Animal handling can be a source of stress, and the specific method used (e.g., mice picked up by tail or in cupped hands) can affect research outcomes [61,121,122].”

What the sources say

• Supported

  The Abstract identifies handling as a source of stress and says that handling method affects mouse anxiety and test performance. The General Discussion reports that tail-handled mice avoided the handler, while mice handled non-aversively explored more and performed better in behavioural tests. (LLM, conf=0.93)

  cites [61] K Gouveia, 2017, Optimising reliability of mouse performance in behavioural testing: the major role of non-aversive handling, Sci Rep source
• Supported

  The Introduction describes tail handling as more anxiety-provoking than tunnel or cupped-hand handling and says handling technique can influence behavioural-study data. The Results report greater anxiety-like behaviour in tail-handled mice, and the Discussion warns that handling may affect responses to rewards and the interpretation of findings. The paper itself tested tail versus tunnel handling; cupped hands appear as an illustrative alternative, not a tested group. (LLM, conf=0.93)

  cites [121] JM Clarkson, 2018, Handling method alters the hedonic value of reward in laboratory mice, Sci Rep source
• ?Not checked — no full text of the sourcecites [122] JL Hurst, 2010, Taming anxiety in laboratory mice, Nat Methods source

In the paper

“RRIDs are unique and stable, allowing unambiguous identification of reagents or tools used in a study, aiding other researchers to replicate the methods.”

What the sources say

• Supported

  The “Why Not Just Use a Catalog Number or the Company Website?” section explicitly says RRID syntax is unique and stable. “One Practical Solution: Research Resource Identifiers” explains that identifiers point to individual resources, while “Simple Common-Sense Steps Help Authors Avoid Reproducibility Pitfalls” says RRIDs make it easier to identify the exact resource used in a study. No page numbers are provided in our copy. (LLM, conf=0.94)

  cites [126] AE Bandrowski, 2016, RRIDs: a simple step toward improving reproducibility through rigor and transparency of experimental methods, Neuron source
• Supported

  The Abstract says RRIDs identify research resources, including reagents and tools, to improve identifiability and scientific reproducibility. The Discussion (“Why unique identifiers?”) says identifiers must be unique and persistent, and the Results say an RRID combined with its resource accession number uniquely retrieves articles using that resource. (LLM, conf=0.92)

  cites [127] A Bandrowski, 2016, The Resource Identification Initiative: a cultural shift in publishing, J Comp Neurol source

In the paper

“Reporting the correct nomenclature is crucial to understanding the data and ensuring that the research is discoverable and replicable [116–118].”

What the sources say

• Supported

  The opening of “In the name of reproducibility” says that adhering to mouse nomenclature guidelines makes research discoverable and replicable. “Who provides it now?” explains that inconsistent gene names make it difficult to tell which gene or mutation a paper describes, supporting the point about understanding the data. (LLM, conf=0.91)

  cites [116] S Mallapaty, 2018, In the name of reproducibility, Lab Animal source
• Partly supported

  The Abstract says standard nomenclature is vital for interpreting and recovering mouse experimental data, and the Conclusions link it to searching and retrieving archived data. The Mutant Gene Symbols section says correct annotation allows studies to be compared accurately. These passages support understanding and discoverability, but do not establish that correct nomenclature ensures research is replicable. (LLM, conf=0.83)

  cites [117] JP Sundberg, 2010, Commentary: mouse genetic nomenclature: standardization of strain, gene, and protein symbols, Vet Pathol sourceExperimental — accuracy not yet measured
• ?Not checked — no full text of the sourcecites [118] L Montoliu, 2011, Using standard nomenclature to adequately name transgenes, knockout gene alleles and any mutation associated to a genetically modified mouse strain, Transgenic Res source

In the paper

“Maternal diet affects offspring body weight [183].”

What the sources say

• Supported

  The Results and discussion (p. 5) report significantly lower birth weights and weaning body masses in offspring of mothers fed a junk-food diet than in the chow-fed comparison group. This directly supports the broad claim that maternal diet affects offspring body weight, as demonstrated in rats. (LLM, conf=0.94)

  cites [183] SA Bayol, 2007, A maternal ‘junk food’ diet in pregnancy and lactation promotes an exacerbated taste for ‘junk food’ and a greater propensity for obesity in rat offspring, Br J Nutr source

In the paper

“Testing a specific hypothesis has implications for both the study design and the data analysis [16,167].”

What the sources say

• ?Not checked — no full text of the sourcecites [16] MF Festing, 2002, Guidelines for the design and statistical analysis of experiments using laboratory animals, ILAR J source
• Supported

  In “Improving Design and Interpretation of Preclinical Research” (page not provided), the authors say that confirmatory studies should match experimental design to the intended application and use pre-specified statistical analyses. This supports the claim that the purpose of hypothesis testing has implications for both study design and data analysis. (LLM, conf=0.90)

  cites [167] J Kimmelman, 2014, Distinguishing between exploratory and confirmatory preclinical research will improve translation, PLoS Biol source

In the paper

“Temperature and humidity | Modifies tumour progression [186].”

What the sources say

• Partly supported

  The Results report slower tumor growth and fewer lung metastases in mice housed at thermoneutral rather than standard temperatures. Materials and Methods says humidity was controlled, but the paper does not report testing whether humidity modifies tumour progression. (LLM, conf=0.89)

  cites [186] KM Kokolus, 2013, Baseline tumor growth and immune control in laboratory mice are significantly influenced by subthermoneutral housing temperature, Proc Natl Acad Sci U S A sourceExperimental — accuracy not yet measured

In the paper

“The ethical statement provides editors, reviewers, and readers with assurance that studies have received this ethical oversight [12].”

What the sources say

• Supported

  The Ethical statement section requires authors to disclose ethical review permissions. The Practicalities section says this information should be available to reviewers and readers for assessment, while Table 1 says it enables editors and reviewers to judge the work’s ethical quality. Together, these support the claimed purpose of the statement. (LLM, conf=0.88)

  cites [12] JC McGrath, 2015, Implementing guidelines on reporting research using animals (ARRIVE etc.): new requirements for publication in BJP, Br J Pharmacol source

In the paper

“They can also inform the choice of animal model by providing a comprehensive overview of the models used along with their benefits and limitations [160–162].”

What the sources say

• ?Not checked — no full text of the sourcecites [160] MT Avey, 2015, The use of systematic reviews and reporting guidelines to advance the implementation of the 3Rs, Journal of the American Association for Laboratory Animal Science: JAALAS
• Supported

  The Scientific Rationale section says systematic reviews make differences in animal-study designs transparent and help select the most suitable animal models. That supports the claim that reviewing models and their relative strengths and limitations can inform model choice; the paper does not use the phrase “comprehensive overview.” (LLM, conf=0.88)

  cites [161] CR Hooijmans, 2013, Progress in using systematic reviews of animal studies to improve translational research, PLoS Med source
• Partly supported

  The Introduction says systematic reviews provide a less biased summary of preclinical evidence, and the ‘External and Construct Validity’ section gives an example of a review identifying which animals predominate in stroke studies. That section also discusses limitations in how animal models represent human disease. Our copy does not say that reviews provide a comprehensive account of the benefits and limitations of each model or explicitly recommend them for choosing a model. (LLM, conf=0.81)

  cites [162] ES Sena, 2014, Systematic reviews and meta-analysis of preclinical studies: why perform them and how to appraise them critically, J Cereb Blood Flow Metab sourceExperimental — accuracy not yet measured

In the paper

“However, abstracts often either do not contain enough information for this purpose [11] or contain information that is inconsistent with the results in the rest of the manuscript [154,155].”

What the sources say

• ?Could not verify

  This source is cited for the claim that abstracts contain information inconsistent with the rest of the manuscript; the insufficient-information claim has a separate citation. The Discussion and Results address compliance with ARRIVE reporting guidelines, but no passage found by searching for abstracts, inconsistencies, and manuscript results assesses whether abstracts contradict manuscript results. (LLM, conf=0.73)

  cites [11] K Hair, 2019, A randomised controlled trial of an intervention to improve compliance with the ARRIVE guidelines (IICARus), Research Integrity and Peer Review sourceExperimental — accuracy not yet measured
• ?Not checked — no full text of the sourcecites [154] RM Pitkin, 1999, Accuracy of data in abstracts of published research articles, JAMA source
• ?Not checked — no full text of the sourcecites [155] I Boutron, 2014, Impact of spin in the abstracts of articles reporting results of randomized controlled trials in the field of cancer: the SPIIN randomized controlled trial, J Clin Oncol source

In the paper

“In studies in which statistical power is low and/or hypothesis-testing is inappropriate, providing the effect size and confidence interval indicates how small or large an effect might really be, so a reader can judge the biological significance of the data [151,152].”

What the sources say

• Partly supported

  Split into 2 atoms — 0 confirmed, 2 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

  cites [151] D Moher, 2010, CONSORT 2010 Explanation and Elaboration: updated guidelines for reporting parallel group randomised trials, BMJ sourceExperimental — accuracy not yet measured
• ?Not checked — no full text of the sourcecites [152] S Nakagawa, 2007, Effect size, confidence interval and statistical significance: a practical guide for biologists, Biol Rev Camb Philos Soc source

In the paper

“Different studies may be sensitive to different environmental factors, and particular aspects of the environment necessary to report may depend on the type of study [176].”

What the sources say

• Supported

  Section 4, “The Research Animal Environment (Study Conditions)” (page not provided), says environmental factors can affect study outcomes and recommends describing factors known to affect the particular study type or endpoints. Its examples show that the environmental details warranting fuller reporting differ by study type. (LLM, conf=0.94)

  cites [176] 2014, Guidance for the description of animal research in scientific publications, ILAR J source

In the paper

“For example, in a systematic review of the effect of the MVA85A vaccine on tuberculosis challenge in animals, the largest preclinical trial did not include the vaccine name in the abstract or keywords of the publication; the paper was only included in the systematic review following discussions with experts in the field [157].”

What the sources say

• Supported

  Split into 2 atoms — 2 confirmed, 0 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

  cites [157] R Kashangura, 2015, Effects of MVA85A vaccine on tuberculosis challenge in animals: systematic review, Int J Epidemiol source

In the paper

“Environmental enrichment | Reduces anxiety [196,197], stress [196,197], and abnormal repetitive behaviour [198–200]. Reduces susceptibility to epilepsy [201] and osteoarthritis [202] and modifies the pathology of neurological disorders [203]. Increases foraging behaviour in fish [204].”

What the sources say

• Partly supported

  The Severity of OA section reports lower histological OA scores with one enrichment device, the Shepherd Shack, than with the other devices, but the difference was not statistically significant (p = 0.07). All mice developed OA after surgery. The paper therefore suggests a device-specific effect on disease severity in a surgically induced mouse model, not a general reduction in susceptibility to developing OA. (LLM, conf=0.89)

  cites [202] A Salvarrey-Strati, 2008, The influence of enrichment devices on development of osteoarthritis in a surgically induced murine model, ILAR J sourceExperimental — accuracy not yet measured

In the paper

“Similarly, human clinical trials initiated based on insufficient or misleading animal research evidence increase research waste and negatively influence the risk-benefit balance for research participants [229,232].”

What the sources say

• Partly supported

  The Introduction (page not specified) says risk–benefit assessment in early human trials depends heavily on animal evidence, and the paper states that poor preclinical study design and reporting thwart that assessment. It does not establish that trials launched on insufficient or misleading evidence increase research waste or demonstrate a worsened risk–benefit balance for participants. (LLM, conf=0.84)

  cites [229] S Wieschowski, 2018, Preclinical efficacy studies in investigator brochures: do they enable risk–benefit assessment?, PLoS Biol sourceExperimental — accuracy not yet measured
• ?Not checked — no full text of the sourcecites [232] I Chalmers, 2014, How to increase value and reduce waste when research priorities are set, The Lancet source

In the paper

“Limitations are unavoidable in scientific research, and describing them is essential to share experience, guide best practice, and aid the design of future experiments [228].”

What the sources say

• Partly supported

  The Introduction (page not provided) calls candid discussion of limitations crucial to scientific discourse and says it informs higher-quality future studies. The Discussion section (page not provided) recommends using limitations to guide future research and study design. The copy does not establish that limitations are unavoidable in all scientific research or specifically say that describing them guides best practice. (LLM, conf=0.83)

  cites [228] MA Puhan, 2012, Discussing study limitations in reports of biomedical studies- the need for more transparency, Health Qual Life Outcomes sourceExperimental — accuracy not yet measured

In the paper

“If the experiment is testing the efficacy of a treatment, the occurrence of adverse events may alter the balance between treatment benefit and risk [34].”

What the sources say

• Supported

  The Adverse Events section states that adverse events may tip the balance of benefit and risk for the intervention being tested. This directly supports the citing sentence’s qualified claim. (LLM, conf=0.94)

  cites [34] BS Muhlhausler, 2013, Whole animal experiments should be more like human randomized controlled trials, PLoS Biol source

In the paper

“Environmental enrichment | Reduces anxiety [196,197], stress [196,197], and abnormal repetitive behaviour [198–200]. Reduces susceptibility to epilepsy [201] and osteoarthritis [202] and modifies the pathology of neurological disorders [203]. Increases foraging behaviour in fish [204].”

What the sources say

• Supported

  The Discussion reports that cod reared with variable food cues were faster to approach and consume live prey than cod given food at an unchanging location. It concludes that varying food supply may improve the foraging capacity of hatchery-reared fish. (LLM, conf=0.89)

  cites [204] VA Braithwaite, 2005, Environmental variability in the early rearing environment generates behaviourally flexible cod: implications for rehabilitating wild populations, Proceedings of the Royal Society B-Biological Sciences source

In the paper

“Untreated pain can affect the animals’ biology and add variability to the experiment; however, specific pain management procedures can also introduce variability, affecting experimental data [212,213].”

What the sources say

• Supported

  The review states that untreated pain can affect organs, physiology and behavior and may increase data variability (Conclusion). It also says that both the pain-management protocol and untreated pain can affect results and increase variability (Focus on Reproducibility Review). (LLM, conf=0.94)

  cites [212] P Jirkof, 2017, Side effects of pain and analgesia in animal experimentation, Lab Anim (NY) source
• Supported

  The Introduction (page not supplied) states that untreated pain affects animals’ biology and that pain can skew experimental data by introducing unwanted variability. It also says pain-relieving drugs can have these effects, supporting the sentence’s second clause. (LLM, conf=0.95)

  cites [213] L Carbone, 2016, Pain and laboratory animals: publication practices for better data reproducibility and better animal welfare, PLoS ONE source

In the paper

“The EDA can also be used to generate a time-stamped PDF, which sets out key elements of the experimental design [19].”

What the sources say

• Partly supported

  The Introduction section on improving the reliability of experimental results and analysis (Fig. 1 workflow; page not available in our copy) says users can export a PDF report containing the EDA diagram and information about the experiment’s internal validity. It does not say the PDF is time-stamped. (LLM, conf=0.87)

  cites [19] N Percie du Sert, 2017, The Experimental Design Assistant, PLoS Biol sourceExperimental — accuracy not yet measured

In the paper

“Registering a protocol before the start of the experiment enables researchers to demonstrate that the hypothesis, approach, and analysis were planned in advance and not shaped by data as they emerged; it enhances scientific rigour and protects the researcher against concerns about selective reporting of results [239,240].”

What the sources say

• Partly supported

  The Results section says prospective declaration of primary outcomes prevents investigators from selecting a favorable outcome to report, and the Discussion says preregistration constrained post hoc changes to hypotheses. The paper thus supports the selective-reporting and transparency rationale for preregistration, but it does not say that registration protects researchers personally from concerns about selective reporting, and its evidence concerns clinical trials rather than experiments generally. (LLM, conf=0.83)

  cites [239] RM Kaplan, 2015, Likelihood of null effects of large NHLBI clinical trials has increased over time, PLoS ONE sourceExperimental — accuracy not yet measured
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites [240] C Allen, 2018, Open Science challenges, benefits and tips in early career and beyond, PsyArXiv

In the paper

“For example, in the Registered Report format offered by an increasing number of journals, protocols undergo peer review, and if accepted, the journal commits to publishing the completed research regardless of the results obtained [237].”

What the sources say

• Supported

  The Registered Reports section describes peer review of a Stage 1 protocol and says that a sound rationale and methods lead to an agreement to publish regardless of the outcome. The submission-pipeline passage adds that in-principle acceptance virtually guarantees publication if the authors follow the approved protocol. (LLM, conf=0.94)

  cites [237] CD Chambers, 2014, Instead of "playing the game" it is time to change the rules: Registered Reports at AIMS Neuroscience and beyond, AIMS Neuroscience source

In the paper

“An increasing number of publishers and funding bodies require authors or grant holders to make their data publicly available [247].”

What the sources say

• Partly supported

  The Discussion reports an increase from 2011 to 2012 in journals requiring data sharing, and the Introduction says NSF and NIH require dataset disclosure. But the Introduction describes sharing with other researchers, while some journal policies provide data only upon request; it does not establish that authors or grant holders must make data publicly available. Page numbers are not supplied in our copy. (LLM, conf=0.82)

  cites [247] V Stodden, 2013, Toward reproducible computational research: an empirical analysis of data and code policy adoption by journals, PLoS ONE sourceExperimental — accuracy not yet measured

In the paper

“In such cases, power calculations are not appropriate and sample sizes can be estimated based on operational capacity and constraints [44].”

What the sources say

• ?Not checked — no full text of the sourcecites [44] PS Reynolds, 2019, When power calculations won’t do: Fermi approximation of animal numbers, Lab Animal source

In the paper

“Two-sided tests with a nondirectional alternative hypothesis are much more common and allow researchers to detect the effect of a treatment regardless of its direction.”

In the paper

“Study design can influence the statistical power of an experiment, and the power calculation used needs to be appropriate for the design implemented.”

In the paper

“The choice of control or comparator group is dependent on the experimental objective.”

In the paper

“Reporting the total number of animals used in the study is also useful to identify whether any were reused between experiments.”

In the paper

“For complex study designs, a visual representation is more easily interpreted than a text description, so a timeline diagram or flowchart is recommended.”

In the paper

“For example, the number of animals needed for antibody or tissue production is determined by the amount required and the production ability of an individual animal.”

In the paper

“Controversy exists within the statistics community on when it is appropriate to use a one-sided test [42].”

What the sources say

• ?Not checked — no full text of the sourcecites [42] LS Freedman, 2008, An analysis of the controversy over classical one-sided tests, Clin Trials source

In the paper

“Note that analyses such as ANOVA and chi-squared are based on asymmetrical distributions (F-distribution and chi-squared distribution) with only one tail. Therefore, these tests do not have a directionality option.”

In the paper

“In early feasibility or pilot studies, the number of animals required depends on the purpose of the study.”

In the paper

“Diagrams facilitate the identification of which treatments and procedures were applied to specific animals or groups of animals and at what point in the study these were performed.”

In the paper

“Selectively excluding some experimental groups (for example, because the data are inconsistent or conflict with the narrative of the paper) is misleading and should be avoided [20].”

What the sources say

• Not checked yetcites [20] The BMJ. Scientific misconduct [cited 2020 Jan 10]. https://www.bmj.com/about-bmj/resources-authors/forms-policies-and-checklists/scientific-misconduct

In the paper

“The sample size in each group at the start of the study may be different from the n numbers in the analysis (see Item 3. Inclusion and exclusion criteria); this information helps readers identify attrition or if there have been exclusions and in which group they occurred.”

In the paper

“Sample sizes that are too small (i.e., underpowered studies) produce inconclusive results, whereas sample sizes that are too large (i.e., overpowered studies) raise ethical issues over unnecessary use of animals and may produce trivial findings that are statistically significant but not biologically relevant [35].”

What the sources say

• ?Not checked — no full text of the sourcecites [35] MD Jennions, 2003, A survey of the statistical power of research in behavioral ecology and animal behavior, Behav Ecol source

In the paper

“Positive controls can be used to support the interpretation of negative results or determine if an expected effect is detectable.”

In the paper

“When the objective of the preliminary study is primarily logistic or operational (e.g., to improve procedures and equipment), the number of animals needed is generally small.”

In the paper

“If, however, each cell taken from the mouse is then randomly allocated to different treatments and assessed individually, the cell might be regarded as the experimental unit.”

In the paper

“Choosing the appropriate calculator or algorithm to use depends on the type of outcome measures and independent variables, and the number of groups.”

In the paper

“For studies in which the outcome is the successful generation of a sample or a condition (e.g., the production of transgenic animals), the number of animals is determined by the probability of success of the experimental procedure.”

In the paper

“They also help to communicate complex design features such as whether factors are crossed or nested (hierarchical/multilevel designs), blocking (to reduce unwanted variation, see Item 4. Randomisation), or repeated measurements over time on the same experimental unit (repeated measures designs); see [16–18] for more information on different design types.”

What the sources say

• ?Not checked — no full text of the sourcecites [16] MF Festing, 2002, Guidelines for the design and statistical analysis of experiments using laboratory animals, ILAR J source
• ?Not checked — no full text of the sourcecites [17] ST Bate, 2014 source
• Not checked yetcites [18] G Ruxton, 2017

In the paper

“Exclusion criteria may result from technical or welfare issues such as complications anticipated during surgery or circumstances in which test procedures might be compromised (e.g., development of motor impairments that could affect behavioural measurements).”

In the paper

“It may also provide important safety information for human trials (e.g., exclusions related to adverse effects).”

In the paper

“Reporting losses will help the reader to understand the experimental design process, replicate methods, and provide adequate tracking of animal numbers in a study, especially when sample size numbers in the analyses do not match the original group numbers.”

In the paper

“There are many legitimate reasons for experimental attrition, some of which are anticipated and controlled for in advance (see subitem 3a on defining exclusion and inclusion criteria), but some data loss might not be anticipated.”

In the paper

“For example, data points may be excluded from analyses because of an animal receiving the wrong treatment, unexpected drug toxicity, infections or diseases unrelated to the experiment, sampling errors (e.g., a malfunctioning assay that produced a spurious result, inadequate calibration of equipment), or other human error (e.g., forgetting to switch on equipment for a recording).”

In the paper

“Blinding is a strategy used to minimise these subjective biases.”

In the paper

“Blinding is especially important during outcome assessment, particularly if there is a subjective element (e.g., when assessing behavioural changes or reading histological slides) [83].”

What the sources say

• ?Not checked — no full text of the sourcecites [83] R Rosenthal, 1963, The effect of experimenter bias on the performance of the albino rat, Behav Sci source

In the paper

“Using a validated method of randomisation helps minimise selection bias and reduce systematic differences in the characteristics of animals allocated to different groups [62–64].”

What the sources say

• ?Not checked — no full text of the sourcecites [62] KF Schulz, 1995, Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment effects in controlled trials, JAMA source
• ?Not checked — no full text of the sourcecites [63] KF Schulz, 2002, Allocation concealment in randomised trials: defending against deciphering, Lancet source
• ?Not checked — no full text of the sourcecites [64] TC Chalmers, 1983, Bias in treatment assignment in controlled clinical trials, N Engl J Med source

In the paper

“Using appropriate randomisation methods during the allocation to groups ensures that each experimental unit has an equal probability of receiving a particular treatment and provides balanced numbers in each treatment group.”

In the paper

“Exclusion criteria may also reflect the ethical principles of a study in line with its humane endpoints (see Item 16. Animal care and monitoring).”

In the paper

“Reporting exclusions and attritions provides valuable information to other investigators evaluating the results or who intend to repeat the experiment or test the intervention in other species.”

In the paper

“Whereas exclusion criteria and humane endpoints are typically included in the ethical review application, reporting the criteria used to exclude animals or data points in the manuscript helps readers with the interpretation of the data and provides crucial information to other researchers wanting to adopt the model.”

In the paper

“Criteria for excluding samples or data include failure to meet quality control standards, such as insufficient sample volumes, unacceptable levels of contaminants, poor histological quality, etc.”

In the paper

“The exact number of experimental units analysed in each group (i.e., the n number) is essential information for the reader to interpret the analysis; it should be reported unambiguously.”

In the paper

“If the person assessing the outcome cannot be blinded to the group allocation (e.g., obvious phenotypic or behavioural differences between groups), some, but not all, of the sources of bias could be mitigated by sending data for analysis to a third party who has no vested interest in the experiment and does not know whether a treatment is expected to improve or worsen the outcome.”

In the paper

“Randomising the order of examination can also reduce bias.”

In the paper

“Together with randomisation, allocation concealment helps minimise selection bias, which can introduce systematic differences between treatment groups.”

In the paper

“This type of blinding is often neglected but is important, as the analyst makes many semisubjective decisions such as applying data transformation to outcome measures, choosing methods for handling missing data, and handling outliers.”

In the paper

“Summary/descriptive statistics provide a quick and simple description of the data; they communicate quantitative results easily and facilitate visual presentation.”

In the paper

“The outcome measure is potentially affected by the treatments or interventions being tested but also by other factors, such as the properties of the biological samples (sex, litter, age, weight, etc.) and technical considerations (cage, time of day, batch, experimenter, etc.).”

In the paper

“For example, categorical data are summarised by counts and percentages or proportions and are analysed by tests of proportions; these analysis methods assume that data are binary, ordinal or nominal, and independent [18].”

What the sources say

• Not checked yetcites [18] G Ruxton, 2017

In the paper

“Standard acclimatisation periods may vary between research laboratories, and this information cannot be inferred by readers.”

In the paper

“This relationship has to be managed in the statistical analysis by including cage/litter/animal effects in the model or by aggregating the outcome measure to the cage/litter/animal level.”

In the paper

“If the assumptions are incorrect, the conclusions may not be valid.”

In the paper

“However, these tests have to be used cautiously. If the sample size is small, they will struggle to detect non-normality; if the sample size is large, the tests will detect unimportant deviations.”

In the paper

“It enables other researchers to repeat the experiment and generalise the findings.”

In the paper

“For example, the assumptions for data used in parametric tests (such as a t test, z test, ANOVA, etc.) are that the data are continuous, the residuals from the analysis are normally distributed, the responses are independent, and different groups have similar variances.”

In the paper

“Effect sizes are often not reported in animal research, but they are relevant to both exploratory and hypothesis-testing studies.”

In the paper

“If the residuals of the analysis are not normally distributed, the assumption may be satisfied using a data transformation in which the same mathematical function is applied to all data points to produce normally distributed data (e.g., loge, log10, square root).”

In the paper

“Categorical data can be expressed as counts, frequencies, or proportions.”

In the paper

“Other types of outcome measures (binary, categorical, or ordinal) will require different methods of analysis, and each will have different sets of assumptions.”

What the sources say

• Not checked yetcites [18] G Ruxton, 2017

In the paper

“The hierarchy can induce a clustering effect; for example, cage, litter, or animal effects can occur when the outcomes measured for animals from the same cage/litter, or for cells from the same animal, are more similar to each other.”

In the paper

“For continuous data, these descriptors include a measure of central tendency (e.g., mean, median) and a measure of variability (e.g., quartiles, range, standard deviation) to help readers assess the precision of the data collected.”

In the paper

“For example, animals of the same strain but from different sources, or animals obtained from the same source but at different times, may be genetically different [16].”

What the sources say

• ?Not checked — no full text of the sourcecites [16] MF Festing, 2002, Guidelines for the design and statistical analysis of experiments using laboratory animals, ILAR J source

In the paper

“It also enables readers to assess whether the animal characteristics chosen for the experiment are relevant to the research objectives.”

In the paper

“In some cases, however, reporting the developmental stage is more informative than chronological age—for example, in juvenile Xenopus, in which rate of development can be manipulated by incubation temperature [109].”

What the sources say

• ?Not checked — no full text of the sourcecites [109] MK Khokha, 2002, Techniques and probes for the study of Xenopus tropicalis development, Dev Dyn source

In the paper

“Statistical significance is usually quantified and evaluated against a preassigned threshold, with p < 0.05 often used as a convention.”

In the paper

“Hypothesis tests are based on assumptions about the underlying data.”

In the paper

“To reduce the risk of bias, some of these factors can be taken into account in the design of the experiment, for example, by using blocking factors in the randomisation (see Item 4. Randomisation).”

In the paper

“For example, an experiment designed to detect a hypothesised effect will likely need to be analysed with inferential statistics, and a statistical estimation of the sample size will need to be performed a priori (see Item 2. Sample size).”

In the paper

“Explaining the purpose of the study by describing the question(s) that the research addresses allows readers to determine whether the study is relevant to them.”

In the paper

“This also promotes transparency and understanding about the use of animals in research and fosters public trust.”

In the paper

“Readers can also assess the relevance of the model organism, procedures, outcomes measured, and analysis used.”

In the paper

“In contrast, exploratory research investigates many possible effects and is likely to yield more false positive results because some will be positive by chance.”

In the paper

“Knowing whether a study is exploratory or hypothesis-testing is critical to its interpretation.”

In the paper

“The abstract is often used as a screening tool by readers to decide whether to read the full article or whether to select an article for inclusion in a systematic review.”

In the paper

“Hypothesis-testing studies also have a predefined primary outcome measure, which is used to assess the evidence in support of the specific research question (see Item 6. Outcome measures).”

In the paper

“Tank colour affects stress in aquatic species [179,180].”

What the sources say

• ?Not checked — no full text of the sourcecites [179] AM Holmes, 2016, Impact of tank background on the welfare of the African clawed frog, Xenopus laevis (Daudin), Appl Anim Behav Sci source
• ?Not checked — no full text of the sourcecites [180] M Pavlidis, 2013, Husbandry of zebrafish, Danio rerio, and the cortisol stress response, Zebrafish source

In the paper

“The environment determines the health and wellbeing of the animals, and every aspect of it can potentially affect their behavioural and physiological responses, thereby affecting research outcomes [175].”

What the sources say

• ?Not checked — no full text of the sourcecites [175] T Nevalainen, 2014, Animal husbandry and experimental design, ILAR J source

In the paper

“A transparent and accurate abstract increases the utility and impact of the manuscript and allows readers to assess the reliability of the study [153].”

What the sources say

• ?Not checked — no full text of the sourcecites [153] RB Haynes, 1990, More informative abstracts revisited, Ann Intern Med source

In the paper

“Exploratory research helps researchers to design hypothesis-testing experiments by choosing what variables or outcome measures to focus on in subsequent studies.”

In the paper

“Regulates sexual differentiation in zebrafish [187].”

What the sources say

• ?Not checked — no full text of the sourcecites [187] C Lawrence, 2007, The husbandry of zebrafish (Danio rerio): A review, Aquaculture source

In the paper

“Hypothesis-testing studies may collect additional information for exploratory purposes; it is important to distinguish which hypotheses were prespecified and which originated after data inspection, especially when reporting unanticipated effects or outcomes that were not part of the original study design.”

In the paper

“A typical exploratory study may measure multiple outcomes and look for patterns in the data or relationships that can be used to generate hypotheses.”

In the paper

“Independent replication and meta-analysis can further increase the confidence in conclusions.”

In the paper

“The confidence interval for the effect indicates how precisely the effect has been estimated and tells the reader about the strength of the effect [150].”

What the sources say

• ?Not checked — no full text of the sourcecites [150] DG Altman, 2005, Why we need confidence intervals, World J Surg source

In the paper

“Systematic reviews of the animal literature provide the most convincing evidence that a research question has not been conclusively addressed, by showing the extent of current evidence within a field of research.”

In the paper

“Environmental enrichment | Reduces anxiety [196,197], stress [196,197], and abnormal repetitive behaviour [198–200]. Reduces susceptibility to epilepsy [201] and osteoarthritis [202] and modifies the pathology of neurological disorders [203]. Increases foraging behaviour in fish [204].”

What the sources say

• ?Not checked — no full text of the sourcecites [201] R Auvergne, 2002, Delayed kindling epileptogenesis and increased neurogenesis in adult rats housed in an enriched environment, Brain Research source

In the paper

“The use of humane endpoints can help minimise harm while allowing the scientific objectives to be achieved [219].”

What the sources say

• Not checked yetcites [219] C Hendriksen, 2011

In the paper

“Reporting adverse events allows other researchers to plan appropriate welfare assessments and minimise the risk of these events occurring in their own studies.”

In the paper

“A thorough description of the procedures used to alleviate pain, suffering, and distress provides practical information for researchers to replicate the method.”

In the paper

“Discussing the limitations of the work is important to place the findings in context, interpret the validity of the results, and ascribe a credibility level to its conclusions [227].”

What the sources say

• ?Not checked — no full text of the sourcecites [227] JP Ioannidis, 2007, Limitations are not properly acknowledged in the scientific literature, J Clin Epidemiol source

In the paper

“Exploratory results derived from additional outcomes should not be described as conclusive, as they may be underpowered and less reliable.”

In the paper

“Publishing score sheets of the clinical signs that were monitored [221] can help guide other researchers to develop clinically relevant welfare assessments, particularly for studies reporting novel procedures.”

What the sources say

• ?Not checked — no full text of the sourcecites [221] DB Morton, 2000, A systematic approach for establishing humane endpoints, ILAR J source

In the paper

“Datasets can also be independently cited in their own right, which provides additional credit for authors.”

In the paper

“Sharing adequately annotated data allows others to replicate data analyses so that results can be independently tested and verified.”

In the paper

“The identification of funding sources allows the reader to assess any competing interests and any potential sources of bias.”

In the paper

“This allows others to explore new topics and increases the impact of the study, potentially preventing unnecessary use of animals and providing more value for money.”

In the paper

“Alternatively, enough evidence may have accumulated in the literature to justify further research in another species or in humans.”

In the paper

“Findings from a novel, exploratory study may be used to inform future research in a broadly similar context.”

In the paper

“This practice is gaining increasing recognition and acceptance [250].”

What the sources say

• Not checked yetcites [250] Data Citation Synthesis Group, 2014

In the paper

“Data sharing allows the data to be repurposed and new datasets to be created by combining data from multiple studies (e.g., to be used in secondary analyses).”

In the paper

“This can be used to demonstrate that the study conduct, analysis, and reporting were not unduly driven by emerging data.”