Nutrition research has been questioned for “cherry picking” significant findings being pedalled by the industry or for failing to follow the rules of nutritional epidemiology for correcting common nutrition bias.
Under-reporting of dietary intake consists of a fundamental problem in nutrition epidemiology, inducing bias in the habitual dietary intake data. In parallel, as far as nutrients are concerned, given that their intake is greatly dependent on the total energy consumption of the participants, with greater-energy consumers exhibiting an increased intake in all nutrients, a correction of all intakes based on the level of energy intake (EI) is necessary to balance the findings.
Similarly, dietary indexes based on nutrients depend on the EI, and, according to Willet, should also be adjusted to the EI of the participants.
The aim of the present meta-epidemiological study was to evaluate the statistical significance status of research items published in three academic journals (American Journal of Clinical Nutrition, European Journal of Clinical Nutrition, and Clinical Nutrition ESPEN).
They found that most of the examined research items had positive findings and failed to account for the EI of the participants. Reports of studies not identifying any funding source were more likely to produce positive findings. Positive findings differed according to study design but no differences were noted in the proportion of positive findings between the industry and other funding sectors.
"The present study revealed that nutrition research can greatly benefit from the adoption of more rigorous research methodologies and improved reporting," the authors of the analysis conclude.
"As in all fields, researchers involved in nutrition are greatly responsible for the integrity, transparency and reproducibility of the science and ought to make steps towards the improvement of the design, analyses and reporting of their studies.
"We should not forget that with regard to population health, nutrition research is competing against big pharma. Although this might increase the scrutiny against nutrition, it can also serve as an additional motive to implement the best and most up-to-date design standards and reporting guidelines to support the science of nutrition in a seemingly uneven war. As this is an exciting time for nutrition research, the findings herein provide food for thought and indicate some areas of nutrition research that can be improved."
All research items included in the analysis were published between the years 2015 and 2019. Study design, primary and secondary findings, sample size and age group, funding sources, positivist findings, the existence of a published research protocol and the adjustment of nutrients/dietary indexes to the energy intake (EI) of participants, were extracted for each study.
Out of 2,127 studies in total, the majority had positive findings, in all three journals.
Most studies had a published research protocol, however, this was mainly due to the randomised controlled trials and not to the evidence-synthesis studies.
No differences were found in the distribution of positive findings according to the existence/inexistence of a published research protocol. In the pooled sample of studies, positive findings differed according to study design and more significant findings were reported by researchers failing to report any funding source. Analyses also showed that more significant findings were reported among studies lacking funding, compared to those funded by the industry.
The analysis suggests the majority of published research on clinical nutrition are being funded by scientific organisations rather than the industry, and thankfully, no differences were noted in the proportion of positive findings between the industry and other funding sectors.
"Whether this means that scientists are beginning to understand the role of the industry, if this is particular to the 'clinical nutrition' research domain, or if this is coincidental, it is not known, as this was a cross-sectional study," the report notes.
It concludes: "The publication of research protocols prior to the commencement of a study is considered as an important parameter of research transparency, limiting selective reporting bias, regardless of the scope of the study. Nonetheless, it appears that not all nutrition studies, whether primary or evidence synthesis, adhere to the protocol publication standard."
The authors say an important question arising from the present work, lays on whether academic journals are responsible for specific “inadequacies” present in nutrition research. They say most scientists and statisticians involved in nutrition research "appear unaware of all the prerequisites needed to produce more robust findings".
They add: "According to the International Committee of Medical Journal Editors (ICMJE), one of the responsibilities of journal editors is to ensure that published research is of the best possible quality standards and scientific integrity. In reality though, a great load lays on the peer review process. In this manner, the responsibility for attaining research integrity in nutrition research is segmented between scientific and statistical editors and scientists, including peer reviewers, authors and statisticians working in the field."
Gkiouras, K.; Choleva, M.-E.; Verrou, A.; Goulis, D.G.; Bogdanos, D.P.; Grammatikopoulou, M.G.
2022, 14(23), 5164; https://doi.org/10.3390/nu14235164
"A Meta-Epidemiological Study of Positive Results in Clinical Nutrition Research: The Good, the Bad and the Ugly of Statistically Significant Findings"