Landis, S. C. et al. A call for transparent reporting to optimize the predictive value of preclinical research. Nature 490, 187-191

National Institute of Neurological Disorders and Stroke, NIH, Bethesda, Maryland 20892, USA.
Nature (Impact Factor: 42.35). 10/2012; 490(7419):187-91. DOI: 10.1038/nature11556
Source: PubMed

ABSTRACT The US National Institute of Neurological Disorders and Stroke convened major stakeholders in June 2012 to discuss how to improve the methodological reporting of animal studies in grant applications and publications. The main workshop recommendation is that at a minimum studies should report on sample-size estimation, whether and how animals were randomized, whether investigators were blind to the treatment, and the handling of data. We recognize that achieving a meaningful improvement in the quality of reporting will require a concerted effort by investigators, reviewers, funding agencies and journal editors. Requiring better reporting of animal studies will raise awareness of the importance of rigorous study design to accelerate scientific progress.

Download full-text


Available from: Oswald Steward, Aug 30, 2015
  • Source
    • "Similar suggestions have been made for biological studies (Kozlov and Hurlbert, 2006; Verbitsky, 2013) and research in sports medicine and sports science (Hayen, 2006). It was recently noted that when a biological variation in response to some intervention was the variable of interest in the analysis of samples, considering the natural grouping of objects was essential (Landis et al., 2012). Therefore, the current study has three main goals: (i) to determine gender differences in start reaction times for four different 50 m stroke final events at six Swimming World Championships, (ii) to assess the effects of ignoring natural groupings at Swimming World Championships, and (iii) to evaluate the effects of the large sample size on statistical inferences when gender differences are examined with respect to start reaction times among elite swimmers at Swimming World Championships. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Four 50 meter male/female finals ‐ the freestyle, butterfly, breaststroke, and backstroke ‐ swum during individual events at the Swimming World Championships (SWCs) can be defined in four clusters. The aim of the present study was to use a single‐unit design structure, in which the swimmer was defined at only one scale, to evaluate gender differences in start reaction times among elite swimmers in 50 m events. The top six male and female swimmers in the finals of four swimming stroke final events in six SWCs were analyzed. An unpaired t‐test was used. The p‐values were evaluated using Neo‐Fisherian significance assessments (Hurlbert and Lombardi, 2012). For the freestyle, gender differences in the start reaction times were positively identified for five of the six SWCs. For the backstroke, gender differences in the start reaction times could be dismissed for five of the six SWCs. For both the butterfly and breaststroke, gender differences in the start reaction times yielded inconsistent statistical differences. Pooling all swimmers together (df = 286) showed that an overall gender difference in the start reaction times could be positively identified: p = 0.00004. The contrast between the gender differences in start reaction times between the freestyle and backstroke may be associated with different types of gender adaptations to swimming performances. When the natural groupings of swimming stroke final events were ignored, sacrificial pseudoreplication occurred, which may lead to erroneous statistical differences.
    Journal of Human Kinetics 09/2014; 42(1):215‐222. DOI:10.2478/hukin‐2014‐0075 · 0.70 Impact Factor
  • Source
    • "In addition, commonly used methods for statistical interpretation of treatment effects are at high risk of introducing bias [46], [47]. Second, several groups have called for improved reporting standards to improve reproducibility, rigorous assessment of results and efficient use of animals [41], [42], [48], [49]. A broadly applicable tool would need to incorporate both of these aspects. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Sensory processing in the spinal cord during disease states can reveal mechanisms for novel treatments, yet very little is known about pain processing at this level in the most commonly used animal models of articular pain. Here we report a test of the prediction that two clinically effective compounds, naproxen (an NSAID) and oxycodone (an opiate), are efficacious in reducing the response of spinal dorsal horn neurons to noxious knee joint rotation in the monosodium iodoacetate (MIA) sensitized rat. The overall objective for these experiments was to develop a high quality in vivo electrophysiology assay to confidently test novel compounds for efficacy against pain. Given the recent calls for improved preclinical experimental quality we also developed and implemented an Assay Capability Tool to determine the quality of our assay and ensure the quality of our results. Spinal dorsal horn neurons receiving input from the hind limb knee joint were recorded in anesthetized rats 14 days after they were sensitized with 1 mg of MIA. Intravenous administered oxycodone and naproxen were each tested separately for their effects on phasic, tonic, ongoing and afterdischarge action potential counts in response to innocuous and noxious knee joint rotation. Oxycodone reduced tonic spike counts more than the other measures, doing so by up to 85%. Tonic counts were therefore designated the primary endpoint when testing naproxen which reduced counts by up to 81%. Both reductions occurred at doses consistent with clinically effective doses for osteoarthritis. These results demonstrate that clinically effective doses of standard treatments for osteoarthritis reduce pain processing measured at the level of the spinal cord for two different mechanisms. The Assay Capability Tool helped to guide experimental design leading to a high quality and robust preclinical assay to use in discovering novel treatments for pain.
    PLoS ONE 08/2014; 9(8):e106108. DOI:10.1371/journal.pone.0106108 · 3.23 Impact Factor
  • Source
    • "This has also been found in other highly researched medical fields such as stroke [7], asthma [16], cancer [17], and pharmaceutical drug development [18]. Second, poor methodological quality of AR has been reported in many publications over the past four decades [5]–[9],[19]–[26]. The lack of randomization, allocation concealment, blinding, primary outcome and sample size calculation, as well as multiple statistical testing, and publication bias have been assumed to account for the poor translation of AR to human medicine [3],[6],[8],[27]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background Animal research (AR) findings often do not translate to humans; one potential reason is the poor methodological quality of AR. We aimed to determine this quality of AR reported in critical care journals. Methods All AR published from January to June 2012 in three high-impact critical care journals were reviewed. A case report form and instruction manual with clear definitions were created, based on published recommendations, including the ARRIVE guidelines. Data were analyzed with descriptive statistics. Results Seventy-seven AR publications were reviewed. Our primary outcome (animal strain, sex, and weight or age described) was reported in 52 (68%; 95% confidence interval, 56% to 77%). Of the 77 publications, 47 (61%) reported randomization; of these, 3 (6%) reported allocation concealment, and 1 (2%) the randomization procedure. Of the 77 publications, 31 (40%) reported some type of blinding; of these, disease induction (2, 7%), intervention (7, 23%), and/or subjective outcomes (17, 55%) were blinded. A sample size calculation was reported in 4/77 (5%). Animal numbers were missing in the Methods section in 16 (21%) publications; when stated, the median was 32 (range 6 to 320; interquartile range, 21 to 70). Extra animals used were mentioned in the Results section in 31 (40%) publications; this number was unclear in 23 (74%), and >100 for 12 (16%). When reporting most outcomes, numbers with denominators were given in 35 (45%), with no unaccounted numbers in 24 (31%), and no animals excluded from analysis in 20 (26%). Most (49, 64%) studies reported >40, and another 19 (25%) reported 21 to 40 statistical comparisons. Internal validity limitations were discussed in 7 (9%), and external validity (to humans) discussed in 71 (92%), most with no (30, 42%) or only a vague (9, 13%) limitation to this external validity mentioned. Conclusions The reported methodological quality of AR was poor. Unless the quality of AR significantly improves, the practice may be in serious jeopardy of losing public support.
    07/2014; 4(1):26. DOI:10.1186/s13613-014-0026-8
Show more