There’s an interesting study making the rounds on Twitter:
- Dulal, S., Liégeois, F., Osrin, D., Kuczynski, A., Manandhar, D. S., Shrestha, B. P., … & Prost, A. (2018). Does antenatal micronutrient supplementation improve children’s cognitive function? Evidence from the follow-up of a double-blind randomised controlled trial in Nepal. BMJ Global Health, 3(1), e000527.
Introduction Multiple Micronutrient (MMN) supplementation during pregnancy can decrease the proportion of infants born low birth weight and small for gestational age. Supplementation could also enhance children’s cognitive function by improving access to key nutrients during fetal brain development and increasing birth weight, especially in areas where undernutrition is common. We tested the hypothesis that children whose mothers received MMN supplementation during pregnancy would have higher intelligence in early adolescence compared with those receiving Iron and Folic Acid (IFA) only.
Methods We followed up children in Nepal, whose mothers took part in a double-blind Randomised Controlled Trial (RCT) that compared the effects on birth weight and gestational duration of antenatal MMN versus IFA supplementation. We assessed children’s Full Scale Intelligence Quotient (FSIQ) using the Universal Non-verbal Intelligence Test (UNIT), and their executive function using the counting Stroop test. The parent trial was registered as ISRCTN88625934.
Results We identified 813 (76%) of the 1069 children whose mothers took part in the parent trial. We found no differences in FSIQ at 12 years between MMN and IFA groups (absolute difference in means (diff): 1.25, 95% CI −0.57 to 3.06). Similarly, there were no differences in mean UNIT memory (diff: 1.41, 95% CI −0.48 to 3.30), reasoning (diff: 1.17, 95% CI −0.72 to 3.06), symbolic (diff: 0.97, 95% CI −0.67 to 2.60) or non-symbolic quotients (diff: 1.39, 95% CI −0.60 to 3.38).
Conclusion Our follow-up of a double-blind RCT in Nepal found no evidence of benefit from antenatal MMN compared with IFA for children’s overall intelligence and executive function at 12 years.
The hypothesis tested is plausible given the results seen earlier for pregnant women with iodine deficiency as well as some very limited previous evidence (see Protzko 2017). Of the 1200 mothers, only 800ish children could be found, leading to questions about selective drop-out. The main results can be seen here:
It can be seen that the two groups had roughly the same number of participants, thus making selective drop-out an unlikely explanation for the pattern seen, which is, basically no effect of the intervention, with a best estimate of 1.25 IQ (simple comparison) or 1.12 IQ (with controls, yes, you want to use controls in randomized trials). It’s a bit of a let down, but an important finding that should be replicated before we are too sure about it, especially in even poorer countries.
Furthermore, it can be seen that the estimated mean IQ is 77 in the follow-up sample. Lynn 2012 reported an IQ of 78 based on a single 1968 study:
- Sundberg, N., & Ballinger, T. (1968). Nepalese children’s cognitive development as revealed by drawings of man, woman, and self. Child development, 969-985.
Curiously, Becker’s dataset v 1.2 (built upon Lynn’s) contains 6 samples from 3 studies, none of which overlap with Lynn 2012’s study (from what I could see):
- Buckley, G. J., Murray-Kolb, L. E., Khatry, S. K., LeClerq, S. C., Wu, L., West Jr, K. P., & Christian, P. (2013). Cognitive and motor skills in school-aged children following maternal vitamin A supplementation during pregnancy in rural Nepal: a follow-up of a placebo-controlled, randomised cohort. BMJ Open, 3, e002000.
- Christian, P., Murray-Kolb, L. E., Khatry, S. K., Katz, J., Schaefer, B. A., Cole, P. M., LeClerq, S. C., & Tielsch, J. M. (2010). Prenatal micronutrient supplementation and intellectual and motor function in early school-aged children in Nepal. Journal of the American Medical Association, 304, 2716-2713.
- Jamison, D. T., & Lockheed, M. E. (1985). Participation in Schooling: Deferminants and Learning Outcomes in Nepal. World Bank.
The values of some samples are very suspiciously low, a point on which Becker agrees (private comm.).
What’s a plausible IQ for Nepal? Well, one idea is to look at it’s residual in terms of relationship to S factor. Looks like this (see this post):
Nepal is right on the regression line (at IQ 78), so we don’t really have any reason to doubt the new IQ or the study Lynn cites. Rather, we should doubt the studies that produce IQs in the 50s. Lynn seems to have set a lower range of plausibility at 65 and winsorized values to that level. His notes (2002 book, I guess) about the studies are:
(no note for Buckley)
Christian et al. (2010) report a cohort study from a micronutrient supplementation trial. For children age 7-9 they report a score of 51 on the UNIT (Universal Intelligence Test) but we do not know what meaning can be attached to this. The mothers (n = 676) obtained a CPM score of 16 and is eqivalent to 15 on the SPM. This is at approximately the 1st percentile on the British 1992 standardisation and a British IQ of 65.
Jamison & Lockheed (1985) report CPM data for 369 household heads, average age 42 years, in rural Nepal, carried out in 1977-8. The sample obtained a mean score of 13.5. The 1998 Raven manual has a CPM-SPM conversion table showing that this is equivalent to an SPM score of 12.5. Tis is at about the 1st percentile of the British 1992 norms and an IQ of 65. Add 3 for a 15 year Flynn effect gives the sample an IQ of 68.
Other nutritional studies?
I note that one of the prior studies is also a nutritional RCT, and it cites Christian et al 2010 in the discussion:
Our follow-up of a double-blind RCT in Nepal found no evidence of overall benefit of antenatal MMN over IFA for children’s intelligence or executive function at 12 years of age. Eight other studies of antenatal MMN supplementation examined cognitive outcomes, and only two found benefits for children’s cognitive function in intention-to-treat analyses. One trial from China by Li et al. found a small increase in mean mental development raw scores using the Bayley Scales of Infant Development at 12 months (1.22, 95% CI 0.32 to 2.12; P=0.02) with MMN supplementation compared with IFA only. 44 This difference had disappeared by 7–10 years of age. 45 The second positive study found a 0.11 SD gain (95% CI 0.01 to 0.20, P=0.03) in procedural memory among children whose mothers received ante- natal MMN supplementation compared with IFA only. 46 Three studies found small benefits in subgroup analyses of children born to undernourished or anaemic mothers. 46–48 The remaining three trials found no benefits of antenatal MMN supplementation compared with placebo or control. 35 49 50 Online supplementary table 1 describes all eight previous trial follow-up studies and their results. Online supplementary table 2 describes the micronutrients given in antenatal supplementation trials. Except for two trials, all tested the UNIMMAP. 35 49
With only two trials showing small positive effects and seven, including ours, showing no benefits in mid-child- hood or early adolescence, the evidence suggests that antenatal MMN supplementation is unlikely to lead to large, population-level improvements in children’s long-term overall intelligence and executive function. These findings also support evidence from 10 single micronutrient supplementation trials, none of which showed effects on cognitive outcomes. It is possible that any gains in cognitive development seen in infancy, such as those seen in the early study by Li et al , are over – whelmed by the subsequent influences of the home and school environments. 6 It is also possible that follow-up studies of micronutrient supplementation have not measured important domains of cognitive function and missed more subtle effects. For example, only one study measured procedural memory, an ability with distinct neural substrates that appeared to be affected by supplementation. 46 Future research could explore such effects further.
I looked in the online supplements, but there were no other studies of Nepal (except Christian et al 2010). Most of these studies were about early moter development, not so much standard measures of intelligence, though a few of them were. So it seems that we are in a position where different samples give quite different results and it doesn’t appear to be due to sampling error, which reminds me of Malloy’s review of the Vietnamese data.