In the largest study of its type to date, and in a pregnant population characterised as having insufficient iodine intake according to WHO-outlined thresholds (UIC < 150 μg/L), evidence was found to support a small association between lower I:Cr measured at 26–28 weeks’ gestation and lower birthweight centile, birthweight in grams and higher probability of SGA. In absolute terms, between the 25th and 75th I:Cr percentile (59 and 121 μg/g), birthweight centile was estimated to be 2.7% higher, birthweight was 41 g higher and the probability of being SGA was 1.9% lower, for a typical participant in this cohort. The estimated birthweight difference of 41 g is small but of comparable size to the birthweight differences observed with environmental tobacco smoke exposure in pregnancy [28]. There was no evidence to support associations between I:Cr and all other outcomes examined, including measures of intrauterine growth from ultrasound scans at 34 weeks’ gestation, head circumference at birth, APGAR score, low birthweight, stillbirth, preterm birth and congenital anomalies. Additionally, UIC was not found to be associated with any outcomes.
Amongst comparable studies, this is the largest to date, with the next largest study including 3140 pregnancies [11]. Our findings for birthweight are concordant with three prior studies reporting associations of lower birthweight with lower iodine status [4, 5, 7]. However, in each case, the association was only examined with UIC, but UIC was not associated with any outcomes in this study. Six other studies also report no evidence of associations between birthweight and UIC [6, 8,9,10,11,12] or I:Cr [11, 12]. The higher probability of SGA with lower I:Cr observed here also supports observations from one prior study [4], but not five others reporting no evidence of associations [6, 7, 10,11,12]. Inconsistency between studies and within specific outcomes may be attributed to differences in the time period of assessment, with iodine status during key developmental stages potentially having differential effects on foetal growth. Furthermore, only three prior studies corrected for urinary dilution using creatinine concentration in urine. The absence of associations across all birth or pregnancy outcomes in this study may reflect measurement error in some outcome assessments such as for intrauterine growth. Alternatively, iodine status at 26–28 weeks’ gestation may be outside of critically important time windows for some aspects of development, or iodine status may not be responsible for the effects on these birth or pregnancy outcomes.
In this study, a sampling point of 26–28 weeks was selected to ensure iodine exposure assessment was conducted before the ultrasound outcomes measured at around 34 weeks. Foetal iodine demands are known to increase throughout pregnancy [29], with demands peaking during the second half of pregnancy [30]. However, despite this increase in demand, iodine excretion has been demonstrated to remain somewhat constant throughout the pregnancy in other populations presenting mild iodine deficiency [31, 32]. Our time point therefore likely represents a snapshot of a pregnant population during a period of peak thyroid stress and of stable iodine excretion.
According to the WHO-outlined threshold [3], pregnant populations with sufficient iodine have a median UIC > 150 μg/L, but the median UIC in this study was just 76 μg/L. Whilst observing a wide range of iodine concentrations in the BiB cohort, median UIC was the lowest of all previous studies [4,5,6,7,8,9,10,11,12]. The difference potentially positions this work to better quantify associations in lower sufficiency settings. Higher concentrations in some settings may explain why several previous studies did not report associations for key growth outcomes, as any effect of iodine is likely to be dose-dependent. The effect sizes reported here are dependent on the iodine distribution in the BiB study population, and replication is therefore warranted in other settings.
In the UK, where salt iodisation or supplementation for iodine is not routine, iodine is primarily sourced from only a few foods (largely from milk but also from some fish, meat and cereals) [33]. Women who avoid dairy or are vegetarian and who do not use supplements may therefore be at greater risk of insufficiency, particularly during pregnancy when iodine demands rise, and consequently may be at greater risk of having lower-weight babies. Lower weight at birth is a well-established risk factor for chronic disease in later life [13, 34]. Furthermore, some women may be at greater risk of having low iodine status: the UK Low Income National Diet and Nutrition Survey reported that Black or Asian women were more likely to be below the UK’s lower reference nutrient intake than White women [35]. The predicted estimates in this study may also be amplified amongst those with several risk factors for smaller-birthweight babies. The majority of women of reproductive age are likely unaware of their iodine status or good dietary sources, as only 23% of women who were pregnant or of reproductive age in the UK had heard of iodine, compared to 100% for folic acid [36]. Given the presence of these at-risk groups, and evidence of a continuing decrease in iodine consumption amongst UK women of childbearing age [35], our finding of an association between lower iodine status and increased risk of lower birthweight is highly relevant for the UK population.
There was evidence of different associations between I:Cr and birthweight centile in women according to deprivation and education category, with more apparent associations in those who were ‘more deprived and less educated’. These differences may result from residual confounding or could indicate better resilience or some compensatory mechanisms amongst more affluent and educated women, possibly driven by generally better general health or dietary status, and thus, women who are ‘less educated and more deprived’ may be at greater risk of potential negative effects of lower iodine status in pregnancy.
Severe iodine deficiency results in a range of disorders in offspring, including mental deficiency and short stature [3]. Iodine plays an essential role in thyroid-mediated foetal growth and development [2], but there is little evidence to indicate an optimum iodine concentration or a threshold associated with better pregnancy or birth outcomes. Within the range of comparatively low iodine concentrations observed in this study, evidence for a threshold or plateau in associations was weak.
Study strengths include the large study size and power to detect potential differences in associations. Data are from a well-characterised and multi-ethnic cohort with a comprehensive range of objective outcomes, allowing for evaluation of potential associations with different aspects of growth and development in utero and at birth. A more accurate measure of iodine status in pregnant women was achieved by accounting for urine dilution variability using I:Cr [37]. Spot urine samples are considered a reliable marker of population iodine status by the WHO [3], and all samples used here were collected after overnight fasting and at a similar time of day. The robust and validated urine sample analysis, including participation in the international EQUIP programme [16], is a strength in this work. The use of standardised birthweights, calculated using nationally representative datasets, is also a strength, along with data-driven exploration of potential associations and threshold effects, rather than the arbitrary categorisation of iodine concentration. Participants were drawn from one geographic region but covered a wide range of social backgrounds and different ethnicities. However, in sensitivity analyses, our results are consistent across various subgroups suggesting generalizability.
Despite carefully controlling for potential confounders, there remains the chance that observed associations result from residual confounding in unmeasured or inaccurately characterised variables. The single assessment of urinary iodine also limits the exploration of changes through pregnancy and any possible effects on stages of foetal development. Random measurement error in estimating iodine status from single spot urine samples may also limit analyses; however, this would bias estimates towards the null [38] and may only explain why associations were not seen for other outcomes.
It remains challenging to establish whether the absence of associations in some other studies resulted from measurement error, assessment during less critically important time periods in pregnancy or the fact that other populations had higher median UIC and may therefore have been less likely to observe associations. Further studies in populations without routine fortification programmes or with relatively low status or that evaluate birth outcomes in relation to iodine status throughout pregnancy (or prior to pregnancy) may provide clarity on these issues. Research into alternative biomarkers will also help to improve characterisation of usual or changes in maternal iodine status.