ISSN 1662-4009 (online)

ESPE Yearbook of Paediatric Endocrinology (2020) 17 15.7 | DOI: 10.1530/ey.17.15.7

ESPEYB17 15. Editors’ choice (1) (18 abstracts)

15.7. Childhood obesity intervention studies: A narrative review and guide for investigators, authors, editors, reviewers, journalists, and readers to guard against exaggerated effectiveness claims

Brown AW , Altman DG , Baranowski T , Bland JM , Dawson JA , Dhurandhar NV , Dowla S , Fontaine KR , Gelman A , Heymsfield SB , Jayawardene W , Keith SW , Kyle TK , Loken E , Oakes JM , Stevens J , Thomas DM & Allison DB



To read the full abstract: Obes Rev. 2019 Nov;20(11):1523–1541. doi: 10.1111/obr.12923.

This is a thoughtful position statement by a number of leading obesity researchers and statisticians. They draw attention to 10 errors that are common in the scientific literature: 1. Using self-reported outcomes and ‘teaching to the test’ (achieving better scores simply by repeat testing); 2. Lack of control groups and ignoring regression to the mean over time; 3. Changing the goal posts (i.e. the study primary outcome); 4. Ignoring clustering when randomizing groups of children; 5. Subsetting the data, p-hacking and data dredging (to produce false positives); 6. Relying on difference from baseline (rather than between groups); 7. Equating ‘no statistically significant difference’ with ‘equally effective’; 8. Prioritising observational analyses over intervention effects; 9. One-sided tests for statistical significance; and 10. Stating that effects are clinically significant when not statistically significant.

These highlighted errors are by no means confined (or more common) in childhood obesity studies than other fields. However, they illustrate these errors and how to avoid them by examples from the childhood obesity literature as an important step towards effective advances in reversing the obesity epidemic.

Notably, this was the last paper by Douglas Altman, who for many years was the chief statistical adviser to the British Medical Journal and, with his co-author Martin Bland, wrote the highly-regarded series of authoritative and succinct Statistical Notes in that journal. He worked hard to improve the use of statistics in medical research, and once wrote ‘The majority of statistical analyses are performed by people with an inadequate understanding of statistical methods. They are then peer reviewed by people who are generally no more knowledgeable’.

Article tools

My recent searches

No recent searches.