Country of Origin of Medical Products and Risk of Labour Rights Abuse: A Cross-Sectional Analysis Using Four Procurement Datasets

Background Case studies have highlighted labour rights abuse in the manufacture of several healthcare products, but little is known about the scale of the problem or the specific products involved. We aimed to quantify and compare the overall and product-specific risks of labour rights abuse in the manufacture of healthcare products supplied to high-income settings using multiple datasets on the product country of origin (COO). Methods Public procurement data from South-Eastern Norway (n=23,972 products) were compared to datasets from three other high-income settings: procurement data from Cambridge University Hospitals, trade data from UN Comtrade, and registry data from the US Food and Drug Administration (FDA). In each dataset, the product COO was matched to the International Trade Union Confederation risk rating for labour abuse and deemed high-risk when rated 4, 5, or 5+. Results In the Norway data, 55.4% of products by value had a COO declared, 49.1% of which mapped as high-risk of labour rights abuses. COO was identified for 70/100 products in the Cambridge data, with COO matching high-risk at 59.9% by value. The level of risk for specific medical product categories varied between the Norway, US FDA, and UN Comtrade datasets, but those with higher proportional risk included medical/surgical gloves and electrosurgical products. Conclusion Evidence of high-risk of labour rights abuse in the manufacture of healthcare products present in these data indicates a likely high level of risk across the sector. There is an urgent need for global legislative and political reform, with a particular focus on supply chain transparency as a key mechanism for tackling this issue.


Introduction
Reports in the last 15 years have documented multiple cases of labour rights abuses in the manufacture of medical products.This includes sweatshop and child labour in the manufacture of steel surgical instruments in Pakistan [1] and endemic forced migrant labour in medical glove manufacturing in Malaysia [2].It also includes state-sponsored forced labour in China: in the manufacture of masks in the Xinjiang province using forced Uyghur labour and of medical gowns in Dandong using alleged forced North Korean labour [3].Together, these reports suggest labour abuse could be widespread in the medical product sector.However, risk analysis of labour rights abuses in this sector to date has been limited [4], despite medical goods being amongst the top 20 traded globally, with annual trade in 2021 estimated at US$150 billion [5].
Medical products have unique specifications, and supply chains have a mix of public and private stakeholders, yet the impact of these factors on the risk of labour rights abuse is poorly understood.Assumptions of risk made in other sectors [6] should not be presumed.Determining, quantifying, and qualifying this risk has implications for researchers, supply chain stakeholders, governments, and international policymakers in how they evaluate and target such issues through their activities.
Challenges for risk analysis include the sector's scale, poor public data availability on medical product procurement, and associated risks or actual incidents of labour abuse [7].However, existing guidance in ethical procurement suggests the risks of labour rights abuse are highest when manufacturing occurs in a country known to have weak legislation, policies, or a track record of protecting workers [8].Thus, the country of origin (COO) of a product can serve as a suitable proxy for the risk of labour abuse and is a method already used by government and inter-governmental agencies to assess supply chain risk in other sectors.An analysis of 100,000 audits on the Sedex platform (a risk assessment tool built by a UK-based company using supplier environmental, social, and governance data) found that the COO was predictive of the risk of labour rights abuse [9].
Here, we quantify the risk of labour rights abuses in the manufacture of medical products supplied to a highincome country (Norway) based on product COO and compare this with COO data on medical products purchased in other high-income settings to assess the generalisability of our findings.We compare risk at the level of the whole dataset and for specific product categories.Specifically, we analysed four datasets on COO for medical products purchased for use in high-income countries: procurement data for South-Eastern Norway and Cambridge University Hospitals (England), global trade data from the UN Comtrade database, and registry data from the US Food and Drug Administration (FDA) database.We focused on high-income contexts due to the high level of consumption of medical goods in these countries and chose to focus on Norway given the quality of data available for this region.However, given the nature of globalised supply chains, our study is relevant to those in countries of all income levels who procure or supply medical products.We argue that if there is evidence of a prevalent risk of labour rights abuses across these multiple datasets, this indicates likely risk in the sector as a whole.
Parts of the work underlying this article were previously presented as a poster at the 2019 Public Health@Cambridge Network Showcase meeting on November 15, 2019, and as a conference abstract with an accompanying oral presentation at the 2021 Sustainable Healthcare Academic Research and Enterprise Conference on May 1, 2021.This article was also presented as an abstract with an accompanying oral presentation at the Brighton and Sussex Health Research Partnership conference on October 19, 2023.

Primary dataset: medical products purchased by the South-Eastern Norway Regional Health Authority
In the 2015-16 financial year (and only this year), suppliers to publicly funded hospitals in the South-Eastern region of Norway (around half of Norway's population or >2.5 million people) were contractually required by the regional health authority (Sykehusinnkjop) to self-declare the production location of the main component of the products they supplied.This included products classified as 'medical consumables' or 'surgical products', excluding medicines, medical aids, laboratory products, and supplies through local contracts with individual hospitals.We will refer to this data as the 'Norway dataset'.These data were supplied to the senior author (MFB) on request (in accordance with Norwegian law that obligates the sharing of data related to public expenditure).This dataset was analysed using a stepwise approach.First, we assigned products to categories based on intended usage.Categories were not prespecified, as there is no validated approach for assigning medical products to such categories, and the contents of the data were unknown prior to exploration.Aberrant or out-of-remit products, non-medical products (e.g., kitchenware, cleaning products, packaging), capital goods (e.g., sterilisation equipment), and medicines were excluded.We assigned the remaining product categories to 'super-categories' based upon similarities in material composition and physical design, on the assumption that such products would likely be manufactured in similar factories.For example, metal surgical instruments and metal laryngoscope blades are made in the same factories in Pakistan, and plastic kidney dishes and plastic specimen containers are made in the same factories in Mexico [10].Multi-component products (e.g., procedure packs) were assigned a separate category.Efforts to reduce categorisation bias included cross-checking of excluded product categories and assignation to product super-categories by two authors (JA and MFB), with disagreements resolved by discussion.
For all products, we equated the self-declared location of production to the product COO.We determined a product to be at high-risk of labour rights abuse where the COO was one with known systematic violations of worker rights (rating 4) or no guarantee of rights (rating 5 or 5+) as defined by the International Trade Union Confederation (ITUC) Global Rights Index for 2016 [11].This rating scale enables comparison of risk between datasets and utilises a variety of sources on real rights abuse instances alongside assessment of each country's legal processes to inform contemporaneous risk ratings, minimising bias compared to supplier reporting [12,13].We calculated the proportion of products at high-risk of labour abuse for all medical products and the proportion within each super-category.
The Norway data are, to our knowledge, the most comprehensive and granular data on the origin of medical products used within healthcare systems in high-income countries.To ascertain congruity and relevance to other contexts, we obtained data from three other high-income settings and cross-analysed the findings to compare the overall risk and the risk for specific product lines.We call these 'comparative datasets'.

Comparative dataset on overall risk: hospital procurement data from Cambridge, England
We obtained procurement data for the 2018-19 financial year for Cambridge University Hospitals NHS Trust (a tertiary-level healthcare provider in England, providing core care to a local population of half a million and regional specialist services to six million) through finance (Qlikview, Qlik, Pennsylvania) and inventory (Powergate, GHX, Louisville) databases.We call this the 'Cambridge dataset'.We extracted the top 100 items by spend under the categories 'medical or surgical equipment' and 'dressings', excluding capital purchases (which could skew data).For products in this dataset, we followed a standardised method of data collection.One author (JA) physically examined product packaging for statements in English on the country of manufacture (=COO), including statements using the terms 'made in', 'manufactured in', or 'assembled in' and any location identified through a manufacturer symbol and listed address.JA also searched supplier and manufacturer websites and performed Google searches using a systematic method to corroborate data on packaging.Where there was a discrepancy between the packaging and online sources, the COO on the packaging was used.
Mirroring our analysis for the primary Norwegian dataset, for the Cambridge dataset we matched COO to classification on the ITUC Global Rights Index but for 2019 [14] (the financial year of this analysis) to ascertain the overall proportion of goods classified as high-risk (ITUC rating 4, 5, or 5+).For multicomponent products, we assigned the product to the highest ITUC rating for the COO of any of its components.Where we could obtain data on only some components, we recorded COO as 'partially available'.

Comparative datasets on risk for specific product lines: global trade data and US registry for medical products
We know of two public databases documenting the COO of products that contain data on some medical products: (1) the United Nations (UN) Statistics Division Commodity Trade database (UN Comtrade), built on import and export data supplied by member countries of the UN [15], and (2) the US FDA Medical Devices Product Registration database, which registers all producers or distributors of medical devices intended for sale in the USA (where registration is a legal requirement) [16].From each database, we extracted information on medical products also present in the Norway dataset.
We searched the UN Comtrade database using the World Integrated Trade System ( WITS, The World Bank, Washington, DC) online platform for exports in 2019.We identified medical products using World Customs Organisation HS 2017 nomenclature [17] and subtracted 're-exports' from 'gross exports' to calculate net exports.For countries with a net export, we assumed the product was manufactured in that country (=COO), and export volumes equated to the quantity manufactured and supplied to the global market.We matched the COO to the 2019 ITUC ratings [14] and calculated the proportion of products at high-risk of labour abuse (rating 4, 5, or 5+).
During October and November 2021, we conducted a search of records on the FDA database for all product categories that comprised >2.5% of contract value (spend) in the Norway dataset, aiming to identify all product codes related to that category.We extracted all 'contract manufacturers' or 'manufacturers' listed on the database under these product codes and equated the COO of the product to the manufacturer's registered address.The FDA database does not include trade volumes.To estimate the proportion of a product coming from each country, we assumed every registered manufacturer of a particular product supplied equal volumes of that product.We matched the COO to 2021 ITUC ratings [18] and calculated the proportion of products at high-risk of labour abuse (rating 4, 5, or 5+).
For both comparator datasets, product codes were identified by one author (JA) with a review by a second (MFB).We compared the proportion of each product category or super-category in the Norway dataset to the equivalent product categories in other datasets.

Primary dataset
There were 23,972 products in the Norway dataset, 22,739 after excluding out-of-remit products (see the Appendices section for a list of out-of-remit products).COO was declared by the supplier for 55.4% of these by value (Table 1).The 2016 ITUC rating is also shown [11], and cells representing ITUC ratings of 4, 5, or 5+ are equated to a high-risk of labour abuse.A dash in the ITUC column represents there is no ITUC rating for that country or that the COO is unknown.Countries supplying <2.5% of the value of contracts for that 'super-category' are not shown.COO is mapped to the 2016 ITUC country rating for labour rights risk [11], with an ITUC rating of 4, 5, or 5+ equated to a high-risk of labour abuse COO: country of origin, ITUC: International Trade Union Confederation

Comparative datasets
In the Cambridge dataset, four items (of 100) were excluded as capital goods.Packaging was available for 83/96 items and supplier and manufacturer websites for 95/96 items (one website was under maintenance).Seventy items (70/96; 73.0%) had information available on COO, including six multi-component products with partially available information (see the Appendices section for the item categories and associated annual spend for products included in the Cambridge dataset).For 10 products, online data were incomplete or did not match product packaging.Of the 70 products for which the COO was identified (  In this dataset, for products of 'multiple' COOs, the highest ITUC-rated country is used to represent the product.COO is mapped to the 2019 ITUC country rating for labour rights risk [14], with ITUC rating 4, 5, or 5+ equated with a high-risk of labour rights abuse.Note: Malta and the European Community are not included in ITUC 2019 ratings, so the ITUC rating is listed as 'not recorded'

COO: country of origin, ITUC: International Trade Union Confederation
Eight medical product codes were identified in the UN Comtrade database with equivalent products in the Norway dataset (see the Appendices section for the UN Comtrade product categories and codes).The COO for these products compared to the nearest equivalent product in the Norway dataset is summarised in Table 4.    [15].The first six columns relate to the Norway dataset: the name of the product category or super-category is shown in column 1 (with category/super-category in brackets indicating which of these is true for this group), columns 2-5 list the COO by contract value (in Euros, €) and the proportion of products from that COO alongside its ITUC rating, column 6 describes the overall proportion by value of that group from high-risk COO.The last six columns (columns [7][8][9][10][11][12] relate to data drawn from the UN Comtrade database, with column 7 listing the HS 2017 product code [17] used to search the database (shortened names are used in this table, see the Appendices section for the full category code name used in HS 2017 nomenclature).The value of net exports in column 9 is reported in US Dollars (USD), where a value of 1 in the table equates to 1000 USD.For both datasets, countries supplying <2.5% of the value of contract/net exports are not shown, percentage by value with a high-risk of labour abuse includes only products with a recorded COO (i.e., missing is excluded), and high-risk COO is defined as matching ITUC rating 4, 5, or 5+.COO is mapped to the 2016 (Norway data) and 2019 (UN Comtrade data) ITUC country rating for labour rights risk [11,14].Note: the European Union is not included in the 2019 ratings and the value of contracts/exports values are listed in their respective reported currencies rather than being converted to a common denominator COO: country of origin, ITUC: International Trade Union Confederation

Norway
Codes for five medical products were identified in the FDA database (see the Appendices section for the FDA product category codes and descriptors).Table 5 summarises the COO for these products compared to the nearest equivalent in the Norway dataset.[16].The first six columns relate to the Norway dataset: the name of the product category is shown in column 1 (no super-categories are included in this table); columns 2-5 list the COO and ITUC rating by contract value and the proportion of product by value (in Euros, €) from that COO; and column 6 describes the overall proportion by value of that product category from high-risk COO.The last six columns (columns 7-12) relate to data drawn from the October/November 2021 FDA Medical Devices Database: column 7 lists the category codes used to search the database for items equated to the Norway product category; columns 8-11 list the COO and ITUC rating by number of manufacturers listed and the proportion of manufacturers from that COO; and column 12 describes the overall proportion of manufacturers listed from high-risk COO.For both datasets, countries supplying <2.5% of the value of contract or number of total manufacturers listed are not shown, percentage by value with a high-risk of labour abuse includes only products with a recorded COO (i.e., missing is excluded) and high-risk COO is defined as matching ITUC rating 4, 5, or 5+.COO is mapped to the 2016 (Norway data) and 2021 (FDA data) ITUC country rating for labour rights risk [11,18].See the Appendices section for the descriptors of FDA database category codes COO: country of origin, ITUC: International Trade Union Confederation, FDA: Food and Drug Administration

Risk of labour rights abuse in healthcare supply chains
This is the first study to comprehensively evaluate the risk of labour abuse in healthcare supply chains.The use of the Norway dataset alongside three comparative datasets on COO from different global contexts is a key strength of this study.The Norway data is a large sample containing granular information on product types, including volumes, value, and COO.The size and granularity of the Norway data increase its suitability in assessing risk for the region.The product types in the Norway data are likely similar for many high-income contexts, although this has not been formally assessed.
Data from the Norway and Cambridge datasets (Table 1, Table 3) indicate that for medical products with known COO, approximately half are manufactured in countries where there is a high-risk of labour rights abuse (COO rating of ITUC 4 or 5 at 49.1% of products by value in Norway and 59.9% in Cambridge).Across both datasets, the main countries contributing to this risk are the USA, Mexico, Malaysia, and China.
When this risk is broken down by product type (Table 2, Tables 4-5), the proportion of products manufactured in any country (and the proportion at high-risk of labour abuse) varied across datasets.For example, the only listed COO for cardiac pacemakers and implantable defibrillators in the Norway data is the USA (contract value €1,309,438), with the majority of COOs for this category missing.While the USA does make up the biggest number of manufacturers in the FDA data (Table 5), there are several other COOs listed here, and the UN Comtrade data (Table 4) does not include the USA as a COO.Variation in the proportion of each product type manufactured in any country likely relates to the supplier used (since suppliers use networks of manufacturers across different locations) and is also impacted by missing data.Moreover, individual supplier practices, including policies on preventing modern slavery and relations with manufacturers, will vary.For most products, therefore, it would be erroneous to generalise the level of risk from each category in the Norway dataset and make blanket statements on the level of risk: assessment should be related back to the supplier chosen for a product and their responsible business practices.
In contrast, our data show most medical gloves are manufactured in South and South-East Asia (Malaysia, Thailand, and China; Table 4), where issues of forced labour have long been documented.Therefore, this product category should be considered high-risk for the majority of suppliers.These issues were made evident during the COVID-19 pandemic when supply chains for personal protective equipment gained media and political interest, leading to the USA banning imports of gloves from several manufacturers in Malaysia, which in turn spurred improvements to working conditions [19].While caution is needed in making comparisons between datasets to ascertain the specific levels of risk (as described above and in the following section), comparative datasets also suggest certain product categories, in addition to gloves, may be at higher (electrosurgical products, clips, and staples) or lower-risk (non-adhesive dressings and examination gel) of labour rights abuse, although even 'lower-risk' items contain significant contributions from countries with high ITUC ratings.We note that China is the top site of production of medical textiles in the Norway data (with contracts totaling €2.4 million), despite having been noted as a location for forced labour by Uyghur populations in picking cotton (although we recognise many medical textiles are made of plastic polymers) [20].
The only preceding study with an assessment of risk in healthcare supply chains is a recent policy paper from the NHS in England on the risk of modern slavery (including slavery, servitude, forced labour, and human trafficking) in its supply chain [21].Predominantly informed by a supplier questionnaire, this assessment investigated whether companies reported supplying any of the five medical products previously highlighted (including by our group [10]) as at risk of modern slavery: face masks, gloves, gowns, surgical instruments, and uniforms.That report found 21% of suppliers to be at risk of modern slavery.In comparison, our study has looked at the proportion of products (not suppliers) at risk of labour abuse and includes the full range of medical products and all forms of labour abuse (beyond those that define modern slavery).Our approach enables the identification of risk in medical supply chains where products might not previously have been highlighted as high-risk of labour rights abuses and allows for variation in risk between suppliers supplying the same products.This is important given the dynamic nature of supply chains, where the supplier chosen is subject to constant change.

Limitations to assessment of risk
There are little or no data on actual working conditions at the site of manufacture for the products we have included, and so our findings are an estimation of labour abuse risk.While there are strong pieces of evidence for instances of labour rights abuses in medical product manufacture from case studies [1-3], we are not aware of previous studies that have analysed the COO of medical products and their associated risk of labour rights abuses.Countries rated by the ITUC as 4, 5, or 5+ show persistent or repeated failures to protect workers, and prior evidence suggests using COO is predictive of real-world labour rights issues [9].However, there are caveats: where manufacturing is largely mechanised and automated, the risk of labour rights infringement is normally lower than where products are largely manufactured through unskilled labour [22].Conversely, our methodology highlights the proportion of products from high-risk COOs (i.e., ITUC rating of 4 or more), but this does not exclude risk where the COO is one with a lower ITUC rating (e.g., ITUC rating of 3, which indicates known 'regular violations of rights').
Our study is a cross-sectional analysis where each dataset represents only a sample of activity within complex medical supply chains and has its own strengths and limitations.Differences in the construction and recording of these data between the four locations could limit comparison and generalisability.
Although we compare the overall risk between the Norway and Cambridge datasets, the Cambridge dataset is a relatively small sample and derives from a specialist centre, which could have different procurement activities than other hospitals in England.There are also some unbalanced data contributions.For example, electrical devices, peripherals, and attachments constituted 50.5% of spend in the Cambridge dataset (see the Appendices section for the item categories and associated annual spend for products included with the Cambridge dataset), but only around 17% in the Norway dataset.We correlated the proportion of products at risk to expenditure, which could be unduly influenced by very high-cost items, such as cardiac pacemakers.
The FDA database does not include data on quantity or value of trade; our assumption that manufacturers supply equal volumes is unlikely to reflect reality, and we recognise the limitations in comparing these proportions to the Norway data.Additionally, our assumption that export volumes in the UN Comtrade dataset equate to COO and quantity supplied to the global market may not always be true due to the complexities of global supply chains.
Our analysis relies on COO reporting by supply chain stakeholders, and our datasets contain incomplete or potentially inaccurate data.Importantly, we note the scale of the missing data.Suppliers to Norway did not declare the COO of their products for nearly half of the products by value, despite this being mandated.It is likely that in many instances, the supplier was not aware of where their product was manufactured or did not wish to declare it, circumstances often associated with poor labour practices [23].It is possible, therefore, that the proportion of medical products at high-risk of labour abuse is even higher than estimated here.
Even where the COO was declared, this may represent a superficial or misleading statement.For example, the USA is a major site for the manufacture of medical products, supported by federal legislation that promotes the purchase of products wholly manufactured or substantially transformed in the USA.This policy does not, however, guarantee that products are actually manufactured in the USA; for instance, the medical suppliers Smith & Nephew and Medtronic have both received substantial fines for falsely stating medical products to be made in the USA when they were in fact manufactured in Malaysia or China [24,25].We should also be concerned about multi-component products, where those components may come from different countries, but the country of final assembly is stated as the COO, masking the risk.In the Norway data, the Czech Republic was stated as the COO of nearly a quarter of procedure packs and multi-component products, yet other data in this paper suggest that the Czech Republic is not a manufacturer of the typical constituents of such packs (including gloves, gauze, cannulae, or catheters).We should also show caution towards electrical devices, peripherals, and attachments, whose manufacture is stated to be predominantly in the USA, Europe, and Central America.It is highly likely that some (or even the majority) of components for these products are manufactured elsewhere; for example, although USA companies are responsible for 49% of global sales of semiconductors, they only manufacture 12%, with most made in China and East Asia [26].
Finally, there is a risk of selection and misclassification biases in the methods to categorise products in the Norway dataset, to identify product COOs in the Cambridge dataset, and to identify equivalent codes in the comparator datasets.However, these should not significantly affect the assessment of overall risk, and the use of multiple datasets as comparators acts towards mitigating this.

Implications for research, policy, and practice
Our analysis provides strong evidence that many medical goods purchased in high-income countries are at high-risk of labour rights abuse in their manufacture.Around half of the products by value in the Norway data with known COO derived from countries at high-risk of labour rights abuse, constituting contracts of over €34 million in 12 months, and we found evidence of such risk in data from different global contexts.These contexts included a smaller procurement sample from an acute hospital in a high-income country where the overall level of risk was comparable to that in the Norway data and two large global trade datasets that demonstrated significant risk for specific product categories.Despite the limitations of our study, this suggests that the overall high-risk of labour rights abuses in the manufacture of medical products identified in the Norway data based on COO is likely representative of the sector.The EU recently published proposals for a new directive on mandatory human rights and environmental due diligence for companies established in the EU, recognising as high-risk sectors the textile, clothing, footwear, food, agriculture, forestry, fisheries, and extractive industries [27].The data available to public buyers here establishes that medical goods are also a high-risk industry, implicating the need for coordinated, multi-disciplinary approaches involving researchers, policymakers, and supply chain stakeholders focusing on this sector.
Our study also highlights the potential utility of identifying high-risk products through COO as an initial approach to identifying risk, which could then be complemented by a rigorous investigation of potential high-risk areas [12].However, in spite of our analysis being to date the most comprehensive assessment of the risk of labour rights abuses in medical supply chains, our attempts to identify high-risk products were limited by poor availability and transparency of COO data.The volumes of production by each COO and the exact level of risk for most products remain largely unknown.Further approaches to identifying high-risk products, including by ethical procurement practitioners, would require improvements to the availability and reliability of data on product COO.Thus, we believe that the proposed EU (and UK) diligence could be furthered by including the mandatory public declaration of product COO, which is already in place in countries such as Australia, Canada, South Korea, and the USA, but not the UK or EU [28].Given the complexity of some products, in particular those with multiple components, the depth and granularity of transparency on COO will need to be sensitive to the particular product type and associated risk and also not simply focus on the first tier of a supply chain [12].Transparency on COO will also help in quantifying and addressing environmental risks, given that typically 88% of the carbon footprint of single-use medical products is due to production [29].
Transparency in the COO should be seen as a first and necessary step in evaluating and tackling labour abuses in medical supply chains.What is also important is the governance in place to tackle such risks once identified, as well as the incentives and barriers to change through strategies of compliance or cooperation [30].This would be complemented by further research into how supplier practices and supply chain stakeholders influence the risk of rights abuse, paying attention to the nuances specific to medical product supply chains.Our study does not include procurement data from middle-or low-income countries; future studies could assess if there are differences in the types and proportions of products purchased in these contexts and whether this impacts the level of risk taken on by the purchaser.Our analysis also did not include pharmaceuticals, but this is an important area for future research: major global sites of pharmaceutical production include China, India, and the USA, which are countries rated 4 or 5 on the latest (2023) ITUC Global Rights Index.Together, these approaches could enable a more comprehensive, validated assessment of risk at the level of specific medical products and specific supply chains, which would provide valuable information to supply chain practitioners and policymakers in all countries towards ending the practice of labour rights abuses in medical supply chains.

Conclusions
There has been previous case-based evidence of labour rights issues in the manufacture of specific medical products, but our analysis (using multiple data sources) suggests high-risk across this entire sector.We add insight into the way that public procurers could undertake risk assessments of their medical supply chains and highlight limitations.Our findings demonstrate a critical and urgent need for greater transparency in global supply chains for medical products, which could be addressed through legislative or regulatory reform.

Appendices
The following supplementary materials provide further supporting information and analysis to aid in the interpretation of the associated manuscript.As per the data availability statement, further original data and analysis are available upon reasonable request to the corresponding author.The appendices use the same nomenclature to refer to each dataset and reference as described in the main text.The methods used to obtain these data are described in the main manuscript.Medical product category names and code from HS 2017 [17] nomenclature that was explored and data obtained from UN Comtrade [15] via World Integrated Trade System (WITS, The World Bank, Washington, DC) online platform (as per method in associated manuscript).The table also describes the shorthand names used throughout the manuscript and supplementary information (which was generated by the authors) to refer to these categories FDA product code FDA product description  Medical product category codes from the FDA product code database [16] that were identified and data was collected as in Table 5 of the main manuscript.All these codes were those identified in searching the FDA product code database for equivalent product categories that comprised >2.5% of contract value for spend in the Norway dataset following the exclusion of out-of-remit products

TABLE 4 : Comparison of COO for equivalent product categories/super-categories in the 2015-16 Norway and 2019 UN Comtrade datasets
COO is compared for equivalent product categories/super-categories in the 2015-16 Norway dataset and the 2019 UN Comtrade database

TABLE 7 : Value of contracts for product super-categories and the product categories they contain from the analysis of the Norway dataset
Medical products supplied to the South-Eastern Norway Regional Health Authority in 2015-16 and the categories included in the final analysis, breakdown of assigned product categories contained within each super-category (super-category name and total contract value in bold), and the value of contracts per product category in Euros (€) 2024 Abbott et al.Cureus 16(2): e54258.DOI 10.7759/cureus.542582024 Abbott et al.Cureus 16(2): e54258.DOI 10.7759/cureus.54258

TABLE 8 : Item categories and associated annual spend for products included in the Cambridge dataset
Breakdown by item category of the top 100 medical consumables (excluding four out of remit products as described in the main manuscript) by spend in the year April 2018-April 2019 supplied to Cambridge University Hospitals NHS Trust.The numbers of items in each category and the related annual spend in British Pounds (£) are provided.The item categories and super-categories are chosen based on the same categories used in the Norway dataset analysis, except for new categories of 'other direct imaging equipment', 'organ transplant media', and 'wound care pack' which have been assigned to appropriate super-categories.Super-categories and the total in that category are listed in bold, with item categories contributing to that category listed directly underneath Pharmaceutical goods; gel preparations designed to be used in human or veterinary medicine as a lubricant for parts of the body for surgical operations or physical examinations or as a coupling agent between the body and medical instruments