A new study, published in Journal of Endocrinology, examines vitamin D consumption during pregnancy and breast feeding and potential development of autism-like behaviour.
A roundup accompanied this analysis.
Title, Date of Publication & Journal
Vitamin D is crucial for maternal care and offspring social behaviour in rats
To be published: Wednesday 21 March
Journal of Endocrinology
Study’s main claims – and are they supported by the data
The paper does not support the claim that children born to mothers with low vitamin D levels may develop autism-like behaviours.
The paper covers an experiment that was done to test the effect of vitamin D deficiency in pregnant rats, not humans. Although this is mentioned in the body of the press release, the crucial words “in rats” are missing from the title. The correct headline is the second sentence of the press release i.e. “Rats with vitamin D deficiency during pregnancy and lactation produced offspring that displayed altered social behaviours in adulthood”.
The main statistical concerns with the paper include the lack of sample size being stated upfront (it should be) and that a large number of naive statistical tests were carried out without any adjustment for multiple comparisons. Key information is missing or hard to find in the paper and there is very little in this paper to indicate that the results used to support the claims made in the press release are anything other than the result of chance. The basic experimental design appears to be valid and a better statistical analysis might still result in the essence of the conclusions being valid.
The study included a control group for comparisons, randomisation was clearly used (although see the caveat below) and observers were blinded to the diet when taking measurements which is good to see.
The study does highlight some potential limitations such as maximum litter size which could be different from real life.
Although male rats were assigned randomly to the females for the purpose of mating, it is not clear if there were systematic differences between the female rats selected to receive a vitamin D deficient diet and their counterparts in the control group. This could mean there was a bias between the groups of females.
For tests where a subjective assessment was required, the observer was blind to which diet a rat had received, although blinding is not mentioned for other tests. The implication here is that the investigators themselves were not blinded (although this is not explicitly mentioned in the paper), which could affect reporting of outcomes.
Many measures used here were observations made by a human being and a statement should have been made on whether these measures are repeatable i.e. if the same observer did the same thing again would they get the same measurement.
The vitamin D deficient group also received calcium supplements. There does not appear to be a discussion as to whether the effects they see are the result of vitamin D deficiency or calcium supplements.
A major limitation is that the sample size is very low. It is also very hard to discern the exact sample size from reading the paper which references different numbers throughout – no explanation is given for this discrepancy.
In general there is a lack of data displayed. If the sample size is as small as I suspect, I don’t see why data for individual litters could not have been published.
The statistical analysis used may not be appropriate and could mean that, as many of the variables appear to be correlated, the statistical significance of each variable measured is flawed. There were also a large number of comparisons, increasing the likelihood that spurious “significant” results could occur by chance. Methods exist for accounting for this problem and adjusting for it transparently. It appears that these weren’t used.
The combination of this and a sample size that looks to be low means there is a risk that the trial will not be reproducible.
* ‘Vitamin D is crucial for maternal care and offspring social behavior in rats’ by Yates et al. published in Journal of Endocrinology on Wednesday 21 March.
Before The Headlines is a service provided to the SMC by volunteer statisticians: members of the Royal Statistical Society (RSS), Statisticians in the Pharmaceutical Industry (PSI) and experienced statisticians in academia and research. A list of contributors, including affiliations, is available here.