December 5, 2023

Trade and changes to several key micronutrients between 1962r and 2016r

Micronutrient analysis—trade analysis

We undertook a macroanalysis of FAO databases to understand the United Kingdom’s domestic and imported supplies of food contributing to micronutrient security in 2015–2017 (2016r, rolling average). A rolling average is a calculation used to analyse data points by creating a series of averages of different subsets of the full dataset. Our rolling average is an average subset of the yearly data localized to three years around the stated year (for example: 1962r is a local average for the years 1961–1963). We use methods and results developed by Macdiarmid et al.16 to analyse micronutrient security and self-sufficiency. To investigate the micronutrient challenges for the EU exit time period, notably the year during which the EU exit referendum took place, we updated Macdiarmid et al.’s16 2010r figures. We also utilize a series of datasets dating back to 1961 presented in Macdiarmid et al.16.

Our analysis used production, export and import statistics from the FAO Food Balance Sheets (FBS)32 database15. Macdiarmid et al.16 converted these same FBS figures into micronutrient masses for a time period ending at the rolling average for 2009–2011 (2010r). We looked to create micronutrient masses for a more up-to-date year. We sourced the FAO FBS figures in 2010r and compared with the figures for 2016r. The comparison between the two time periods enabled us to yield percentage changes in supply of different food categories and origins (domestic, exported, imported) between the two periods.

The FAO FBS statistics are converted into micronutrient supplies via an index, created by Macdiarmid for their figures16, for our 2016r time period. This index is needed because the FAO FBS data15 use broad food categories for their production statistics that cannot be easily associated with a micronutrient density (for example, ‘wheat and products’ is one such category). Macdiarmid et al.16 create an index (an imagined basket of goods) to estimate what each food category contains in terms of actual produce; this is derived from data in the Living Costs and Food Survey41. This basket can convert the FAO FBS’s reported mass of a food category into a total mass of different micronutrients.

We apply the percentage changes we calculated above to the basket-converted micronutrient masses calculated by Macdiarmid for the 2010r period. This effectively extrapolates 2010r figures to find the nutrient supply in 2016r. This, in effect, replicates the food basket process, converting the 2016r FBS supply figures into micronutrient supply.

The micronutrient supply figures are measured per capita per day to situate micronutrient supply within the context of health demands. In our extrapolation, we adjust for population changes to factor in any changes between 2010r and 2016r. To make this adjustment for population growth (using Office for National Statistics (ONS) data this is estimated at 6% between 2010r and 2016r)42, we subtract the population growth percentage from all percentage change in FBS supply figures when extrapolating between the two time periods. Although this controls for population size change, we did not control for changes to the population pyramid, due to the short time period.

To compare the per capita per day supply with the UK population’s health demands, we use the Reference Nutrient Intake (RNI) to calculate the recommended population-level intake for the UK population, which was weighted to the demographics at the time. RNI is the average daily dietary intake level that suffices to meet the nutrient requirements of nearly all (97.5%) healthy persons of a specific sex, age, life stage or physiological condition (such as pregnancy or breastfeeding)43. This approach allows us to take a macro-approach to the supply required for the population’s health.

From this analysis, we can track the total mass per capita per day supply of micronutrients in 2016r that are (1) produced in the United Kingdom, (2) exported from the United Kingdom and (3) imported to the United Kingdom and distinguishing micronutrients that are (1) animal-based, (2) plant-based (within which we identify (3) from fruits and vegetables). The latter distinction is a new one we have made by distinguishing FBS food categories.

We can also understand this mass per capita per day in the context of percentage fulfilment of the average UK resident’s RNI. This will provide the baseline for current supply, which will aid our exploration and visualization of future scenarios. To add further context, we have looked back at Macdiarmid’s results stretching back to 1962r and a series of time points thereafter to 2010r and our extrapolated 2016r16. We have analysed the extended time period according to domestic/import/export in the Supplementary Information. However, in the Results and Figures section, we make our novel distinction between animal and plant (fruit and vegetable) for all these time points.

HMRC trade data for granularity on country and commodity

To gain a more detailed breakdown of UK imported plant-based micronutrient supply—both by country and by food source, we use trade data from HMRC UKTRADEINFO (Overseas Trade Statistics, or OTS)44 supported by the Food Standards Agency (FSA) Trade Data Visualization Application45, which allows data scraping of HMRC trade data. The FSA application is a query tool and allows downloads of a simplified and more flexible dataset than provided by HMRC UKTRADEINFO directly. HMRC data has the advantage of being more detailed than the FAO data used in the 2010r analysis by Macdiarmid et al.16. HMRC’s OTS lists imported food products, detailing each imported product’s ten-digit commodity code, total mass and the specific country of origin. Our study utilizes the OTS’s reported weight of fruits and vegetables imported into the United Kingdom from around the world and thus we can understand imported, plant-based (fruit and vegetable and other plant-based) sources of micronutrients. HMRC trade data are collected through different methods, depending on whether the trade is with non-EU or EU countries. UK businesses trading with non-EU countries are required to declare their imports to pay the legally obligated tariff. Trade between UK businesses and EU countries is collected from declarations in businesses’ VAT returns and, for larger businesses, from a survey called the ‘Intrastat survey’46. We decided to analyse 2017 only, unlike the three-year period used for the FAO FBS figures, because of the different nature of the data. HMRC OTS is released on a monthly basis and updated and refined for a few years afterwards. The FAO FBS is yearly and subject to fewer revisions. We are confident that any systematic errors in a particular OTS period (month) would be resolved by 12 months of data collection and Office for National Statistics quality assurance. We chose 2017 as it was the latest year used in the FAO FBS period we analysed.

These data require knowledge of potential trade country-of-origin misattribution, commonly referred to as ‘The Rotterdam Effect’, which we have previously shown to influence meat trade analysis28. Imports can be misattributed (that is, to the wrong country of origin) because the last port of dispatch can be mistakenly entered as the country of origin. We observe this in how the EU OTS records bananas imported ‘from’ the Netherlands and oranges ‘from’ Ireland—both places where growing such products is infeasible. We have adjusted for this by filtering out of the EU OTS all commodity codes that can be determined as tropical and ungrowable in the EU nations (for example, bananas). However, while this eliminates some cases, there are fruits growable in the European Union that can still be misattributed to neighbouring countries, and so the Rotterdam Effect will still overstate certain neighbouring countries’ contributions in those products.

Understanding micronutrient density of different food products

To transform the HMRC OTS data on imported produce into imported micronutrients, we need to understand the micronutrient density of imported produce. We use micronutrient data from Public Health England’s McCance & Widdowson Composition of Foods Integrated Dataset (CoFID)47, which holds estimations of the milligrams (mg) or micrograms (μg) of various micronutrients per 100 g of hundreds of foods, drinks and recipes. Where necessary and possible, a United States Department of Agriculture (USDA) database was used for missing values48.

For some foods, nutritional content could not be found in the CoFID dataset for a specific micronutrient, although it was believed that the micronutrient was present; this is coded by CoFID as ‘N’ where ‘a nutrient is present in substantial quantities, but there is no reliable information on the amount’. In the very few cases where this occurs and we can’t substitute with values from USDA, we have coded the micronutrient density as zero. This potentially underestimates the nutritional content of some foods that are imported. However, one should expect most foods with particularly high density to have a reliable figure recorded.

Synthesis of trade and micronutrient analysis

For this analysis, we combine the OTS and CoFID to estimate the total mass of micronutrients imported by the United Kingdom and understand the contributions of different countries of origin and commodity types.

To enable this synthesis, we matched the HMRC OTS ten-digit commodity codes to compositional data for foods that CoFID provides. CoFID does not categorize its compositional data according to commodity codes and it provides data for all manner of food from raw produce to fully prepared and cooked dishes. Although a major portion of imported food will be prepared in some way before it is consumed, we match HMRC OTS commodity codes to CoFID’s corresponding raw food ones unless suggested otherwise by the HMRC OTS commodity code description. Commodity code matching can be difficult for a few reasons. Some commodity codes lack a perfect analogue in CoFID because CoFID only has a broader category similar to the sub-types in the OTS (that is, CoFID has only density information for the broad category of ‘easy peeler oranges’, which must act as a proxy for the more specific commodity codes of ‘clementines’, ‘tangerines’ and so on) and vice versa. A commodity code can have no information available in CoFID where any proxy would be too inaccurate (for example, the dried bananas commodity code lacks a corresponding CoFID entry). The supplementary material shows which fruit and vegetable commodity codes were matched to which CoFID fruits and vegetables.

Once matching is complete, we multiplied the HMRC OTS kilograms of net mass for each commodity code by the corresponding CoFID nutritional density measures to establish the net mass of micronutrients imported from different countries of origin. See the below formula using vitamin A as an example:

$$\beginarrayl\mathrmOTS\;\mathrmrecorded\;\mathrmkg\;\mathrmof\;\mathrmcommodity\;\mathrmimported\;\mathrmfrom\;\mathrmcountry\\\times\mathrmCoFID\;\mathrmmeasure\;\mathrmof\;\mathrmdensity\;\mathrmof\;\mathrmvitamin\;\mathrmA\;\mathrmin\;\mathrmcommodity\left( \frac\mathrmmg100\mathrmg\right)\\\times 10 = \mathrmnet\;\mathrmmass\;\mathrmof\;\mathrmvitamin\;\mathrmA\;\mathrmvia\;\mathrmcommodity\;\mathrmimported\;\mathrmfrom\;\mathrmcountry\;\left( \mathrmmg \right)\endarray$$

The HMRC OTS’ measure of weight can include packaging, which means that a record of 100 kg of tomatoes may include tin cans and therefore not actually mean 100 kg of solely tomatoes were imported. This means that our use of CoFID density figures converts weight attributable to inedible packaging into imported nutrients reflective of the food itself. This risks overestimating the nutritional value of imports. If the packaging problem is more or less uniform across products and countries of origin (that is, there are no particular countries/food types that on average report higher kg of packaging to HMRC), the packaging issue is not a problem when analysing the relative contributions of different products and countries of origin. Conversely, for example, if countries that are more distant can ship produce to the United Kingdom only in bulkier packaging to preserve freshness, then these countries may, in turn, have higher weight values assigned to exports, leading to an overestimation of nutritional contributions greater than that of closer countries.

One final complication is that much of what the United Kingdom imports as raw fruits and vegetables will go through preparations—chopping, peeling, cooking, frying, boiling and drying—before consumption. During this preparation, certain nutrients may be broken down or lost. This means that a proportion of certain nutrients that we calculate as part of the national supply may never actually be consumed by individuals.

One may consider using CoFID’s micronutrient density for prepared food instead of raw food density to reflect the loss of food. However, this would also be inaccurate as many foods lose or gain mass from phenomena such as water loss/gain, impacting the density without actually impacting the total supply of nutrients. Using CoFID’s prepared food density in those cases would actually distort the nutrients consumed. For example, to use the CoFID density for boiled beans on OTS mass of imported dried beans would understate the nutritional content of what is imported, prepared and ultimately consumed because the process of boiling increases the total mass of beans, in such cases, reducing its density without necessarily destroying any nutrients.

The two preparation effects of lost nutrients and transformed mass means that it is difficult to attribute nutrient density loss between raw food and food preparations in CoFID to single changes in either the mass of foods or to actual nutrient breakdown. For example, reduction in density for boiled beans and increases in density for dried tomatoes are probably the former, while reductions in folate and vitamin C in boiled broccoli are probably the latter. This means there is no easy fix. As stated earlier, we keep to CoFID’s raw food density unless the OTS commodity code description suggests otherwise. This will mean we overestimate the nutrient supply versus what is actually consumed.

It is important to keep the food preparation problem into perspective. It is always the case that there will be differences between looking at macro-supply levels as opposed to the actual uptake and incorporation of the micronutrients in the human body. For example, there is much debate about bioavailability of some micronutrients in plants compared with animals49, and the macroanalysis focuses on the supply in these two and not the bioavailability from these sources. Serra-Majem et al.50 finds that FAO FBS (our method) estimates consumption of fruits, vegetables and roots 23%, 59% and 64% higher than Individual Dietary Surveys, respectively. It is important to highlight that focusing on supply always considers a ‘best case’ as it considers what is actually potentially available in terms of the supply of micronutrients rather than amounts actually in the human body, reflecting consumption patterns and bioavailability of nutrients post-consumption. It also still enables us to understand where the supply comes from. This means that our results are applicable in analysis of supply of nutritional security but less accurate with respect to the physiological nutritional security of individuals/populations.

OTS micro analysis enables us to identify the top ten fruit/vegetable commodities imported to the United Kingdom for each vitamin/mineral. We converted trade figures to what percentage of the UK’s micronutrient RNI is supplied by specific fruits or vegetables in terms of the RNI. This conversion was done by referring back to the RNI percentage supplied by fruit and vegetable-based imports, according to FAO FBS. In addition, we made a comparison with the five Food Foundation Farming for 5-A-Day categories14, which illustrates where commodities might be sourced, self-sufficiency possibilities and possible trading relationships required if we source less from the European Union. We code the top ten fruits and vegetables based on the categorization created by the Food Foundation14, as it allows visualization of the ability for the United Kingdom to replace key products through trade and/or domestic production.

Scenarios—development and analysis

In this paper, we describe the United Kingdomʼs position in 2016r, before the COVID-19 pandemic and exit from the European Union, both of which have the potential to substantially change the micronutrient security of the United Kingdom. Our analysis presents the situation for a range of vitamins and minerals, chosen according to importance and/or level of insecurity observed in 2010r and expert opinion from nutritional epidemiology researchers on National Diet and Nutrition Survey (NDNS) data and published data51. We focus on five of the seven micronutrients previously analysed16 as these were most insecure and/or influenced by changes in plant-based diets or changing trade patterns. This gives us the ‘business as usual’ current situation, and comparing with 2010r, we can also look at where future trends would lead if the future were on the same trajectory as the previous decade. We used the approach described by Khalil and Alexander23.

The United Kingdom’s decision to leave the European Union and the COVID-19 pandemic have raised a number of important uncertainties, which can result in quite different scenarios/futures. A scenario can be seen as a set of plausible assumptions about the way the world works in future, and we undertook a scenario exercise developing a narrative of the four scenarios represented by the combinations of the extremes of the two critical uncertainties. We consider that commissions such as EAT–Lancet9, UN food systems summit52 and recent drivers/shocks such as COVID-19 and EU exit pose two key uncertainties that align with our supply and demand approach. The increasing calls for a change of diet might be considered a demand uncertainty, and the ability to domestically produce or import products could be considered as a supply uncertainty: (1) the amount of meat versus plant foods in diet and (2) the level of self-sufficiency versus global trading (potentially beyond European Union).

For these scenarios, we have sub-divided animal source data into red meat, white meat, milk/dairy and fish and focused our plant-import origin work on fruits and vegetables to allow us to undertake scenario work. We have sub-divided imports to the European Union and non-European Union (sub-divided where required). These four future worlds could be described as: (1) self-sufficient plant source-rich diets, (2) global trading plant source-rich diets, (3) self-sufficient animal source-rich diets and (4) global trading animal source-rich diets.

These four futures represent extremes, and our analysis has enabled us to determine where we sit for each micronutrient in 2016r. This approach would allow exploration of future diets, perhaps shifting to more plant-based or self-sufficiency. For example, we could explore the UK population shifting towards a more flexitarian, pescatarian, vegetarian and vegan diet by looking at movements along the plant/animal axis (25%, 50%, 75% and 100% plant). In a similar way, we could consider a 25%, 50%, 75% and 100% shift towards self-sufficiency or imports in the trading uncertainty, which we will explore for future publications. Our analysis of datasets back to 1961 highlights how changing trading relationships (joining the European Union in the 1970s) and dietary preferences have occurred in previous years and are thus uncertainties that cannot easily be predicted as certainly happening in one direction.


Leave a Reply

Your email address will not be published. Required fields are marked *