Iodine status of consumers of milk-alternative drinks versus cows’ milk: data from the UK National Diet and Nutrition Survey

Milk is the main source of iodine in the United Kingdom (UK), however, the consumption and popularity of plant-based milk-alternative drinks is increasing. Consumers may be at risk of iodine deficiency as, unless fortified, milk alternatives have a low iodine concentration. We therefore aimed to compare the iodine intake and status of milk-alternative consumers to that of cows’-milk consumers. We used data from the UK National Diet and Nutrition Survey from Years 7-9 (2014-2017; before a few manufacturers fortified their milk-alternative drinks with iodine). Data from four-day food diaries were used to identify consumers of milk-alternative drinks and cows’ milk, along with the estimation of their iodine intake (µg/day) (available for n=3976 adults and children ≥ 1.5 years). Iodine status was based on urinary iodine concentration (UIC, µg/L) from spot-urine samples (available for n=2845 adults and children ≥ 4 years). Milk-alternative drinks were consumed by 4.6% (n=185; n=88 consumed these drinks exclusively). Iodine intake was significantly lower in exclusive consumers of milk-alternatives than cows’-milk consumers (94 vs 129 µg/day; P <0.001). Exclusive consumers of milk-alternatives also had a lower median-UIC than cows’-milk consumers (79 vs 132 µg/L; P <0.001) and were classified as iodine-deficient by the WHO criterion (median-UIC < 100 µg/L) whereas cows’-milk consumers were iodine-sufficient. These data show that consumers of unfortified milk-alternative drinks are at risk of iodine deficiency. As a greater number of people consume milk-alternative drinks, it is important that these products are fortified appropriately to provide a similar iodine content to that of cows’ milk.


Introduction
Iodine is an essential component of the thyroid hormones which are required for brain and neurological development in fetal and early life (1,2) . There is emerging evidence that even mild-to-moderate iodine deficiency in pregnancy is associated with suboptimal child neurodevelopmental outcomes (3) . It is also increasingly clear that the role of pre-pregnancy iodine stores is important (4) and that women of childbearing age need to ensure adequate iodine intake to maximise their thyroid iodine stores prior to pregnancy.
As there is no formal universal salt-iodisation policy in the United Kingdom (UK), dietary sources are more important determinants of iodine intake. In the UK, as is the case in many industrialised countries, milk and dairy products are an important dietary source of iodine, contributing 25-70% of total daily iodine intake (5) . Results from the National Diet and Nutrition Survey (NDNS) between 2014 and 2016 indicate that milk and dairy products accounted for 34% of the average daily iodine intake of UK adults (19-64 years) and 51% and 40% of the iodine intake of children (4-10 years) and adolescents (11-18 years), respectively (6) . Although fish is a rich natural source of iodine (7) , it is not consumed widely and its contribution to iodine intake in the UK is relatively small (6) .
Cows' milk has a low natural iodine content but various dairy-farming practices, including the use of iodine-supplemented cattle-feed and iodophor disinfectants for teatcleaning and sterilising milking equipment, increase its iodine content as a result of carryover into the milk (8)(9)(10) . A recent study that collected milk samples from UK supermarkets in the summer and the winter reported an overall milk-iodine concentration of 427 µg/L (11) , providing approximately 85 µg iodine per glass (200 ml), which would contribute 57% of the World Health Organisation (WHO) recommended iodine intake for adolescents and adults (150 µg/day) and 34% of that for pregnant and lactating women (250 µg/day) (12) . UK milkiodine concentration, however, can vary with season (e.g., milk has a higher iodine concentration in winter than in summer as a result of varying farming practice) (11,13) .
With increasing awareness of the environmental impact of food production (14) and rising concerns over the health effects and sustainability of current eating habits (15) , plantbased diets have become an area of growing interest. This shift to plant-based diets is accompanied by an increase in the consumption of plant-based drinks (e.g., soya, almond, rice drinks) as alternatives to cows' milk (16) . According to Mintel market research, 23% of UK adults used plant-based milk-alternative drinks in 2019 (17) compared to 19% in 2018 (18) and 14% in 2017 (19) . The most recent Mintel data also show that plant-based milk-alternative Accepted manuscript drinks were used mainly by women (26% vs 19% of men) and younger age groups (32% in those under 35 years vs 18% in those of 35 years or above) (17) . People tend to substitute cows' milk with milk-alternative drinks for various reasons, including environmental, health and ethical concerns (16) .
Milk-alternative drinks, however, unless fortified, have a lower iodine content than cows' milk (median 7 vs 438 µg/kg, respectively) (20) . Data from 2015 indicated that the vast majority of milk-alternative drinks available in the UK were not fortified with iodine (20) , which is of concern because the consumers of unfortified milk-alternative drinks might be at risk of iodine deficiency unless another source of iodine is consumed. This is particularly worrying for women of childbearing age, since it is becoming increasingly clear that the preconceptional period may be an important window of opportunity to optimise iodine stores prior to pregnancy and young women are more likely to consume these milk-alternative drinks.
Given the fact that cows' milk is the principal source of iodine in the UK diet, the aim of this study was to investigate the iodine intake and status of consumers of unfortified milkalternative drinks in the UK. We hypothesised that those who consumed milk-alternative drinks exclusively would have lower iodine intake and status than those who consumed only cows' milk.

Study population
We used data from the UK National Diet and Nutrition Survey (NDNS) Rolling Programme (RP) from Years 7 to 9 (i.e., in the period between 2014/15 and 2016/17). The NDNS methodology has been described in detail elsewhere (6,21) . Briefly, the NDNS is a continuous, cross-sectional survey of the UK population that is designed to recruit annually a nationally representative sample of individuals aged 1.5 years and above and it collects information on UK dietary habits, nutrient intakes and nutritional status (6,21)

Dietary assessment in NDNS
Diet of the participants in the NDNS RP was assessed using four-day food diaries.
Participants were asked to record all foods and drinks consumed at home and outside the home for four consecutive days with a randomly assigned start day. The portion sizes were reported using standard household measures or weights from food labels. The diaries were self-completed by adults and children above 12 years while the diaries of children aged 12 years and under were completed by a parent or carer, with assistance from the child. To ensure the completeness and accuracy of the reported dietary data, the diaries were reviewed in the presence of the participants by a trained interviewer. Participants also completed a computer-assisted personal interview about their eating habits and food avoidance (e.g., whether following vegetarian or vegan diets). The food diaries were coded and analysed by researchers at the Medical Research Council Elsie Widdowson Laboratory (MRC-EWL).

Iodine intake
Nutrient intakes, including iodine, were available and had been calculated from the food diaries using the UK Nutrient Databank which is based on the food tables from "McCance and Widdowson's Composition of Foods", the Food Standards Agency's food-portion sizes and data from manufacturers. In the current study, we used data on the average daily iodine intake from food only (µg/day) (i.e., excluding iodine-containing supplements) as our primary outcome but we also examined the total iodine intake (µg/day) (i.e., from food and iodine-containing supplements).

Iodine status
Iodine status was assessed using the urinary iodine concentration (UIC, µg/L) measured in spot-urine samples. From 2012, as part of the NDNS RP, spot-urine samples were collected from participants aged four years and above for the purpose of assessing population iodine status. Detailed description of the methods for urinary iodine analysis is available in the NDNS documentation (NDNS Appendix N) (6,21) . Briefly, UIC was measured at MRC-EWL using an inductively coupled plasma mass spectrometer. The coefficient of variation against quality control samples was ≤ 5% for each of the NDNS years included in the current analyses (i.e., Years 7-9), indicating acceptable accuracy and precision.

Estimation of milk-alternative and cows'-milk consumption
The intakes of milk-alternative drinks and cows' milk were calculated from the food diaries of the fully productive individuals in NDNS (i.e., those who completed three or four fooddiary days) as the average weight consumed daily (g/day). For this analysis, the cows'-milk food-group included whole milk, semi-skimmed milk, skimmed milk and 1%-fat milk food groups. In the NDNS data, the consumption of milk-alternative drinks was not recorded as a separate food group but under the food subgroup "other milk" (as part of the main food group "other milk and cream"). As the food subgroup "other milk" also included other types of milk that were not plant-based milk-alternative drinks (e.g., goats milk, lactose-free milk, evaporated milk, condensed milk, dried milk powder, coffee creamer, milk shakes and others), we further disaggregated this food subgroup to create a separate variable for the consumption of milk-alternative drinks only. This new food group included all types of plantbased milk-alternative drinks that were consumed (e.g., almond, soya, oat, rice, coconut and hemp drinks). We then calculated the total amount of milk-alternatives consumed by each individual over the three or four days of dietary assessment (g) and also the mean daily amount consumed (g/day). These calculations were also done for each type of milkalternative drink (e.g., soya, almond, rice, coconut drink etc.). Individuals were categorised as consumers of milk-alternative drinks [i.e., those who consumed a milk-alternative drink in at least one instance during the three/four days of dietary assessment (average intake > 0 g/day)] and non-consumers [i.e., those who did not consume any milk-alternative drinks during any of the three/four days of dietary assessment (average intake 0 g/day)]. The same categorisation was used to define consumers and non-consumers of the different types of milk-alternative drinks and cows' milk. We further grouped individuals based on how they used milk-alternative drinks and cows' milk in their diet in three categories: i) those who consumed cows' milk exclusively (reference group); ii) those who consumed milk-alternative drinks exclusively (exclusive consumers); and iii) those who consumed milk-alternatives, as well as cows' milk (mixed consumers).

Statistical analyses
We investigated the proportion of consumers of milk-alternative drinks and cows' milk by NDNS year [i.e., Years 7-9 of RP (2014/15, 2015/16 and 2016/17)], sex, age group and diet type (vegetarian, vegan or neither) using Chi-square tests. In the milk-alternative consumer group, we examined the total average daily intake of milk-alternative drinks (g/day) and the differences by sex, age group, diet type and their use in the diet (e.g., used exclusively or used alongside cows' milk) using Mann-Whitney U tests (for two groups) or Kruskal-Wallis tests (for more than two groups).
The distributions of both the estimated daily iodine intake from food (µg/day) and spot-UIC (µg/L) were skewed positively; we therefore reported the medians (25-75 th percentiles) and used non-parametric tests throughout. We examined the differences in daily iodine intake and UIC by survey year, sex, age group and diet type using Mann-Whitney U tests or Kruskal-Wallis tests. We also investigated the differences in daily iodine intake and UIC between those who consumed milk-alternative drinks exclusively (exclusive consumers) and those who consumed only cows' milk using Mann-Whitney U tests. In addition, we repeated these analyses in the women of childbearing age (16-49 years) only. We performed sensitivity analyses that excluded children of 1.5-10 years, as they are high milk consumers.
Outliers for iodine intake and UIC were identified using box-plots and their influence on the analyses was investigated by excluding these values in sensitivity analyses.
We performed multiple linear regression models to examine the association of the type of milk consumed exclusively [milk-alternative drinks vs cows' milk (reference group)] with iodine intake and UIC, using (natural) log-transformed iodine intake and UIC (as the dependent variables) and adjusting for total energy intake (kcal/day), age and sex.
In addition to iodine intake from food only, we also examined the total iodine intake (i.e., from food and supplements); this variable was also positively skewed hence we reported the medians (25-75 th percentiles) and used non-parametric tests. We investigated whether milk-alternative consumers were more likely to use iodine-containing supplements than nonconsumers using a Chi-square test. As with the iodine intake from food, we examined the difference in the total iodine intake (i.e., from food and supplements) between exclusive milk-alternative consumers and cows'-milk consumers using a Mann-Whitney U test.
The median (25-75 th percentile) daily intake of milk-alternative drinks was 92 (38-191) g/day in all consumers of milk-alternatives (n=185) but was significantly higher in the group of exclusive milk-alternative consumers than in the group of mixed consumers (i.e., those who consumed milk-alternatives as well as cows' milk) (P=0.79) and the mean daily fish intake between these groups was similar (28 vs 21 g/day, respectively; P=0.95).
The median (25-75th percentile) estimated daily iodine intake from food (excluding iodine-containing supplements) was 124 (86-176) µg/day in the total sample (n=3976) and 106 (73-154) µg/day in the women of reproductive age (n=772). Iodine intake differed by sex, age group and diet type ( Table 2). Those who exclusively consumed milk-alternative drinks had a lower dietary iodine intake than those who consumed only cows' milk [median (25- respectively; P<0.001; Table 3]. Women of reproductive age who were exclusive consumers of milk-alternative drinks also had a lower dietary iodine intake than reproductive-age women who consumed only cows' milk, though this difference was not statistically significant (Table 3). In sensitivity analyses that excluded the 1.5-10-year age group, the results remained largely unchanged (online Supplementary Table S1). Excluding the outliers did not change the results substantially. Adjustment for total energy intake, age and sex in multiple linear regression analyses did not change the conclusions.
Only 3.7% (n=147) of the included NDNS sample used iodine-containing supplements, of whom 16 were milk-alternative consumers (n=8 were exclusive consumers).
Those who consumed milk-alternative drinks exclusively (n=62) had a significantly lower UIC than cows'-milk consumers (n=2426) [median (25- Table 3]. When comparing to the median-UIC cut-off defined by the WHO for iodine sufficiency in populations or groups (i.e., median-UIC ≥ 100 µg/L) (12) , only the group of cows'-milk consumers was iodine-sufficient (Fig. 1). The results for UIC were similar in the subsample that included only the women of childbearing age (Table 3; Fig. 1), in the sensitivity analyses that excluded the children in the 1.5-10-year age group, and after excluding the outliers.
The findings and overall conclusions remained unchanged after controlling for total energy intake, age and sex in multiple linear regression analyses.

Discussion
In the current study, we found that those who consumed milk-alternative drinks exclusively had a lower iodine intake than those who consumed only cows' milk. Furthermore, the median UIC of the exclusive consumers of milk-alternative drinks indicated that this group was classified as iodine-deficient by WHO criteria (12) whereas cows'-milk consumers were iodine-sufficient.
To our knowledge this is the first study to report on the iodine intake and status of the consumers of milk-alternative drinks in the UK. Previous studies have investigated the iodine intake and/or status of vegetarians and/or vegans in the UK (23)(24)(25) and in other countries (26)(27)(28)(29)(30)(31)(32) and have mostly shown that vegans are at particular risk of iodine deficiency. Although studying these groups provides an insight into how plant-based diets can influence iodine intake and status, it does not necessarily reflect the impact of substituting cows' milk with milk-alternative drinks (e.g., vegetarian diets may include some milk-alternative drinks but usually also include milk and dairy products; vegan diets exclude milk and dairy products and are more likely to include plant-based substitutions but additionally, vegans do not consume other good sources of iodine, such as fish).
Our results show that milk-alternative drinks were typically consumed by women, particularly women of childbearing age. Although the median UIC in the group of women of childbearing age was above the WHO criterion for iodine sufficiency in the general population (i.e., median UIC ≥ 100 µg/L) (12) , this value is lower that the median UIC in the total NDNS sample (107 vs 124 µg/L) and it is well below the WHO criterion for sufficient iodine intake in pregnancy (i.e., median UIC ≥ 150 µg/L) (12) . Moreover, as was the case for the total sample, we found that women of childbearing age who consumed milk-alternatives exclusively had a significantly lower iodine status (measured by spot-UIC) than those who consumed only cows' milk. These findings are of public-health significance as only those women who consumed cow's milk were iodine-sufficient by the WHO criteria. Furthermore, women of childbearing age who were exclusive consumers of milk-alternative drinks had an estimated iodine intake from food (86 µg/day) that was below the Estimated Average Requirement set by the United States Institute of Medicine (95 µg/day) (33) and considerably below the UK Reference Nutrient Intake (140 µg/day) (34) ; this is of concern because it is important to have sufficient iodine intake prior to pregnancy to ensure optimal thyroidal iodine stores that can be drawn upon to support the higher demand for iodine during gestation (35) . Although our results indicate that consumers of milk-alternative drinks might be at risk of iodine deficiency, it is important to note that the influence of consumption of milkalternative drinks on iodine intake and status will depend on how these drinks are used in the diet (e.g., used exclusively or used alongside cows' milk) and the composition of the overall diet (i.e., intake of other iodine-rich food sources, such as fish). Though in the current analyses, we found that some 70% of the consumers of milk-alternative drinks consumed fish, that did not make up for the missing iodine from cows' milk, hence their estimated iodine intake and median-UIC were lower than those of cows'-milk consumers. We also found that although milk-alternative consumers were more likely to use iodine-containing supplements than non-consumers, only a small proportion (< 10%) took such a supplement and exclusive milk-alternative consumers still had a lower total iodine intake (from food and supplements) than cows'-milk consumers.
The differences between the exclusive consumers of milk-alternative drinks and cows' milk were more apparent with spot-UIC than with the estimated iodine intake from the food diary. This could be because iodine intake from fish, which is usually consumed less frequently, might not be reflected in a casual spot-urine sample which reflects iodine intake only in the last 24-48 hours (36) . By contrast, milk is usually consumed daily and therefore, the observed difference in spot-UIC between the two study groups (i.e., consumers of milkalternative drinks vs consumers of cows' milk) is more likely to reflect the difference in the type of milk consumed.
The overall proportion of milk-alternative consumers in the NDNS was low (4.6%) but we observed a small increase over the years of the survey (2014-2017). As indicated by more recent Mintel market-research data (2017-2019) (17)(18)(19) , milk-alternative drinks are indeed becoming more popular, with the proportion of UK adults who consume these drinks increasing from 14% in 2017 to 23% in 2019. Consumption of plant-based milk-alternative drinks will probably continue to increase as people become more concerned about the environmental sustainability of their diets (e.g., greenhouse-gas emissions, land use and water use from dairy production vs plant-based alternatives) (37) . The revised UK Eatwell Guide also aims to address sustainability and now includes milk-alternatives in the milk-and dairy-food group (38) . Furthermore, the British Dietetic Association has recently launched the One Blue Dot toolkit which is a guide to an environmentally-sustainable diet; one of the sustainable diet recommendations is moderate dairy consumption and use of fortified plant-based alternatives where necessary (39) . Although these guides recommend that plant-based milk- alternatives should be fortified, they focus mostly on the comparability to cows' milk in terms of calcium content and do not focus on iodine.
As previously noted (20) , most milk-alternative drinks on the market are fortified with calcium, vitamin D and vitamin B 12 to match the content in cows' milk but it is rare to match for iodine. Considering our findings, it is important for manufacturers to fortify their plantbased milk-alternative drinks with iodine, so that the iodine content is comparable to that of cows' milk. It is also essential that the products are fortified appropriately, for example by the use of potassium iodide, instead of seaweed species (e.g., the brown seaweed, kelp/kombu) because the iodine concentration of the latter is highly variable and can provide excessive amounts (40,41) . Soya and almond drinks were the most frequently reported milk-alternative types in the NDNS suggesting that more fortified options of these milk-alternatives might be needed. The NDNS data we report, however, only go up to 2017 and the situation regarding the consumption of these drinks overall and their popularity by type is constantly changing.
Since there is no iodine-fortification policy in the UK, iodine intake is mainly dependent on individual food-choice. Individuals who substitute cows' milk exclusively by plant-based milk-alternative drinks need to increase their iodine intake from other food sources of iodine (e.g., fish, eggs). This may not be possible for some individuals (e.g., vegans), in which case, iodine-fortified milk-alternatives (if available) or an iodinecontaining supplement (not kelp or seaweed) should be considered. Since the publication of our results on the low iodine content of milk-alternative drinks in the UK (20) , some brands have either started to fortify their plant-based milk-alternative drinks with iodine or have increased the iodine content (e.g., Alpro Soya and Oatly drinks are now fortified at 22.5 µg/100 ml; Marks & Spencer plant drinks are now fortified at 30 µg/100 ml), but most milkalternative drinks available on the market are currently not fortified. Consumers should therefore check the labels of such products as the brands fortifying with iodine may well change over time.
Our study has several limitations that should be considered when interpreting the findings. We categorised individuals into consumers and non-consumers of milk-alternative drinks and cows' milk based on their reported intakes of these foods in the food diary; misclassification might have occurred as a result of misreporting (42) . Additionally, the food diaries capture only short-term food and nutrient intakes (43) (in this case, over three or four days) and this may not fully reflect the day-to-day variability in dietary iodine intake (e.g., an individual day-to-day variation in 24-hour urinary iodine excretion of 33% is suggested as representative for Western populations with varied diets (44) ). We used UIC measured in a single spot-urine sample to assess iodine status. Although this measure is misleading when used to estimate iodine status for individuals as a result of variation in hydration status and day-to-day differences in iodine intake (44)(45)(46) , we used this measure at a group level where these differences usually even out (47) . In the NDNS, the urine samples for the measurement of UIC were not necessarily collected during the four-day dietary assessment. UIC reflects iodine intake in the last 24-48 hours (47) whereas the diary data relate to milk-alternative and cows'-milk consumption during the dietary assessment. This possible mismatch in the timing of the exposure (i.e., milk-alternative/cows'-milk consumption) and one of the outcomes (i.e., UIC) might have influenced our results; however, this would probably have reduced the observed differences in UIC between the study groups (i.e., milk-alternative consumers vs cows'-milk consumers). Additionally, we cannot completely exclude the possibility that some of the milk-alternatives consumed by individuals in the current study were fortified with iodine; however, previous data from 2015 [i.e., during the period when the NDNS data used in this study were collected (2014-2017)] showed that across 20 retail brands, only one brand, not the market leader, was fortifying its products with iodine (20) . Furthermore, had this been the case, the observed differences in iodine status between milk-alternative consumers and cows'-milk consumers would have been less apparent. Finally, because of the cross-sectional design of the NDNS, the observed association between the type of milk consumed (milkalternative drinks or cows' milk) and iodine intake and status cannot infer causality.
In conclusion, the NDNS data, collected prior to the fortification by manufacturers of their milk-alternative drinks with iodine, show that consumers of unfortified milk-alternatives are at risk of iodine deficiency. With the continuing rise in consumption of milk-alternative drinks, it is important for manufacturers to fortify their products appropriately with iodine (e.g., potassium iodide or iodate) to provide a similar iodine content to that of cows' milk and thus reduce the risk of iodine deficiency in milk-alternative consumers. Furthermore, individuals who completely substitute cows' milk by milk-alternative drinks should ensure that they are buying iodine-fortified versions or should increase their iodine intake from other food sources or iodine supplements. NDNS, National Diet and Nutrition Survey; n/a, not applicable. * P-values are from Chi-square tests (after Continuity Correction for comparisons in 2x2 tables). † The percentages and numbers for each type of milk-alternative drink do not add up to the total percentage and number of individuals who consumed any milk-alternative drink (n=185) as some individuals consumed more than one type of milk-alternative drink. ‡ P-value from a Fisher's Exact Test was reported due to cells with an expected count < 5. Table 3. Comparison of iodine intake (µg/day) and urinary iodine concentration (µg/L) between consumers of cows' milk and consumers of milkalternative drinks (exclusive consumers only). Results shown for the total sample, and for women of childbearing age separately.  Data are presented as medians and 25-75 th percentiles for all individuals and separately for those who consumed cows' milk exclusively and those who consumed milk-alternative drinks exclusively. * P-values are from Mann-Whitney U tests comparing UIC of the exclusive consumers of milk-alternative drinks with that of cows'-milk consumers performed in the total included NDNS (Years 7-9) sample (white bars) and separately in the NDNS (Years 7-9) women of childbearing age (16-49 years) (grey bars). † Based on the World Health Organisation median-UIC cut-off for iodine sufficiency in populations or groups of school-age children and/or adults (median-UIC ≥ 100 µg/L).