Interest in the exposure of U.S. drug supply chains to China surged during the COVID-19 pandemic, as Chinese manufacturing plant closures underscored the risks of relying on China for critical medicines and ingredients. This concern led to significant policy actions, including President Trump’s executive order in August 2020 directing the Food and Drug Administration (FDA) to create a list of essential medicines and critical inputs to aid in prioritizing domestic production and reducing dependence on foreign suppliers. Additionally, the Trump administration launched a Section 232 investigation to assess whether imports of pharmaceuticals and their ingredients pose a national security risk.
Much of the concern focuses on the reliance of the U.S. drug supply chain on Chinese producers of active pharmaceutical ingredients (APIs), which are the ingredients in medications that produce the intended therapeutic effect. Reports on the U.S. drug supply chain exposure to Chinese-made products have varied widely. Estimates range from as low as 8% to as high as 47% due to indirect impacts stemming from India’s reliance on China for API. Some reports even claim that China controls 80-90% of the global supply of API. Our analysis suggests that the aggregate exposure is probably less than a quarter of all API by volume.
Understanding the scope and characteristics of supply chain exposure to China has important policy implications. It matters for efforts to improve supply chain resilience because switching API suppliers is much harder than switching suppliers of ingredients for API. It also matters for onshoring efforts because onshoring the API production stage alone has a limited impact on derisking from China if the precursors needed for API production come from China. Additionally, it matters for tariff policy because the tariff rate is generally tied to API location, but not the location of API precursors.
To discuss the exposure that U.S. drug supply chains have to China, we first describe key stages in manufacturing drugs, focusing on small molecule drugs, which represent the lion’s share of drugs taken by U.S. patients. We then discuss various approaches to estimating the exposure and where they can falter. As part of that discussion, we provide new original analysis to highlight areas of vulnerability. We conclude with a discussion of policy implications.
Typical drug supply chain
The prescription drugs dispensed in hospital and retail pharmacies, known as finished dosage form (FDF) products, combine one or more APIs with excipients that have no therapeutic effect but either serve as fillers, diluents, or otherwise affect functionality of the product, such as drug release, stability, or dissolution rate. APIs differ across drugs in their complexity. Active ingredients for highly complex biologic drugs cannot be synthesized chemically but are instead grown in living cells. However, the drug products that U.S. patients overwhelmingly use are small molecule drugs that are chemically synthesized, either fully or partially.
API production for small molecule drugs involves a multistep process that begins with preparing key starting materials and carrying out controlled chemical synthesis through a series of reactions that use auxiliary chemicals. In some cases, the raw materials are first turned into intermediate compounds called intermediates. The process then requires purifying and isolating the API from impurities, filtering and drying the API, and conducting quality control and analytical testing to ensure purity and compliance with regulatory standards. To ensure that the resulting product is made to specification, the API and FDF steps must be tightly controlled.
FDA requires that API and FDF facilities register with FDA and undergo pre-approval inspections that verify that manufacturers can make the product as described in the drug application. Changes in the FDF or API process must be reported to the FDA, with some changes, such as FDF and API site changes, requiring additional FDA review. While FDF and API steps are closely regulated by FDA, the regulation of direct and indirect inputs into API production varies. There is no requirement to register, but manufacturers of certain intermediates will be reported in the initial drug application. In some circumstances, companies that have the FDF approval may have to report changes to intermediate manufacturers if the supplier change in some way affects product characteristics. A change in source only, without a change in process, will not require any amendments to the drug application.
How much API really comes from China?
In researching reports and articles characterizing U.S. reliance on Chinese API, we found a surprising number of statistics that had no reference, misinterpreted or mischaracterized the underlying data, or referenced a source that lacked information supporting its claims. In some cases, statistics seemed to result from a game of telephone, with the 80% China API exposure statistic a good example, which one article tracked to FDA’s reports that 80% of API facilities are not domestic. Overall, we identified several explanations for the variability in API numbers: differences in the unit of measurement used, mischaracterization of India’s reliance on Chinese API, improperly extrapolating from examples, and blending upstream and API exposure.
Determining the correct unit of measurement
One reason for differences in API exposure numbers is the unit of measurement. Some analyses measure the share of facilities in China or the share of APIs that are made in China, weighing each equally. Others apply weights to drugs: an analysis from Avalere Health weighed API exposure by dollar sales of the relevant FDF, while a recent U.S. Pharmacopeia analysis weighted APIs by unit volume. How the different APIs are weighted matters. When APIs are weighted by dollar sales, the resulting figure will heavily skew towards brands because of the extensive gap between branded and generic drug prices. However, American patients overwhelmingly use chemically-synthesized generic drugs, making volume weights more relevant for assessing public health impact from a geopolitical shock.
Identifying exposure through the share of APIs made in China is a better alternative, especially when the analysis is stratified by specific drug subcategories (e.g., essential medicines, most used). Exposure through the share of facilities would likely be an underestimate because of the reported economies of scale differential between Chinese and Indian manufacturers.
Constructing volume-based exposure measures is the best approach because of the public health focus of the policy discussion. But weighing by volume has been a major challenge because of data availability. While FDF sales are available through sources like IQVIA, such data then must be mapped to specific API sources-not an easy task when FDF manufacturers do not disclose publicly who makes their API.
Recently, the U.S. Pharmacopeia Medicine Supply Map made significant progress towards such mapping using various public and proprietary sources. Their analysis shows that 3% of branded volume has Chinese API and 8% of generic volume for drugs that do not include IV fluids, with an additional 6% and 16% respectively, where the country is unclear.
Differentiating between actual and potential exposure
Some API exposure statistics rely on certain publicly available FDA data, without differentiating between actual and potential reliance on Chinese API. One such data source is the list of Drug Master Files (DMFs). DMFs, specifically Type II DMFs, include information about chemistry, manufacturing, and controls (CMC) for the API or its intermediate, including all the synthesis steps from key starting materials.
By filing a DMF with FDA, a third-party API or intermediates manufacturer can submit the necessary CMC information to FDA without having to share the intellectual property information with the company applying for FDA approval of the FDF version. If the applicant plans to make the API in-house, the full CMC information may be directly included in the application, without a DMF.
FDA will not review a DMF’s content until it reviews the drug application that references it. With no review, listing DMFs is easy and can invite irrelevant DMF filings.
A recent U.S. Pharmacopeia analysis appropriately characterizes the DMF submission data as showing capability, not actual exposure, when discussing China’s share of DMF filings that rose from 5% in 2000 to 32% in 2024. However, it is easy to misinterpret DMF filings as actual reliance because FDA uses the term “active” when the DMF listing simply means that the DMF is available for reference in a new drug application.
We further characterize the extent to which DMFs are not used in drug applications, focusing on API. This is easier done for generics because an API DMF must pay a fee under the Generic Drug User Fee Act (GDUFA) before a generic drug application referencing it is submitted.
As such, the absence of a GDUFA fee is an indication that the API DMF is not referenced in a generic drug application. No such DMF fee exists for branded products, but a focus on generics is appropriate because of their prevalence and the small number of Chinese facilities involved in branded API production.
To identify which DMFs are for API in generics or soon-to-be generics, we eliminate biologics and DMFs for single source small molecule drugs filed at least two years before generic applications can be submitted. To separate out the API, we then cross-reference the DMF subject with the Orange Book.
We find that close to 20% of DMFs filed between 2013 and 2024 do not have a reasonable match with approved drug substances. These DMFs include intermediates, as expected, as well as drugs that are not approved in the U.S. market. This appears to be a result of companies wanting to use a DMF filing for promotional purposes.
Just as DMF filings have been mischaracterized as actual API sources, FDA’s facility registration data contained in the electronic Drug Registration and Listing System (eDRLS) has been mischaracterized as a set of facilities making prescription drug products for the U.S. market.
However, eDRLS is much broader, with registration requirements for facilities making over-the-counter products, unapproved prescription drug products, and facilities making API for research purposes. As with the DMF file, there are no restrictions to registering, so facilities listed in eDRLS may not be making product for the U.S. market.
To narrow down the list of API facilities making API for prescription drugs intended for the U.S. market, we limit the 2024 eDRLS list to those that either pay GDUFA API facility fees or have FDA compliance records indicating branded drug production. The impact is substantial: out of the 1,867 facilities listing API manufacture in their operations description, 46% have neither FDA compliance records suggesting branded drug production nor have they paid GDUFA fees.
Among the 417 registered Chinese API facilities, 53% have neither paid GDUFA fees nor have FDA compliance records suggesting branded drug production.
Properly interpreting Chinese API exports to India
En lugar de la asignación de nivel de API-FDF, una aproximación razonable para estimar la exposición de fármacos estadounidenses a API chinos es ajustando las compartes de mercado de determinados países por su exposición a API chinos. En particular, la exposición de India es importante debido a su alto porcentaje de genericos de fármacos sólidos vendidos en los Estados Unidos, el 65% de los genericos de fármacos sólidos. Si sabemos cuál es el porcentaje de las necesidades de API de India que son abastecidas por China, podemos ajustar el 65% de la participación de India de genericos de fármacos sólidos por ese número, aproximando la exposición que los Estados Unidos tiene a China a través de India.
Cómo mucho India importa de China es bien documentado a través de las estadísticas de importación de India, que muestran que la participación en dólares de las importaciones chinas (relativas a todas las API importadas) ha estado fluctuando alrededor del 68%, más o menos 2%. Pero entre 2018 y 2023, la participación en unidades de las importaciones chinas (relativas a todas las API importadas) ha aumentado de 58% a 75%. Esto significa que la media de los precios de las importaciones chinas está disminuyendo, either porque el mix de drogas importadas está cambiando o porque los puntos de precio de API chinos están bajando más rápido que los de API importados de otros países.
Estas estadísticas de importación son ampliamente pero incorrectamente interpretadas como indicativas de la dependencia general de India de China para API, porque India también tiene una industria de API doméstica robusta.
Evaluar el alcance de la producción de API doméstica de India es mucho más difícil que sus importaciones porque las estadísticas nacionales de cuentas de India no se desglosan en productos terminados y API.
Sin embargo, la producción de drogas domésticas de India parece ser considerable según la asociación de fabricantes de genericos de India, que en 2019 evaluó las importaciones a India de todos los países en un 32% de las necesidades de API de India, un número más tarde citado por una revisión de la dependencia de China de India escrita por funcionarios del gobierno indio.
La omisión de la producción de API doméstica de India puede hacer una diferencia sustancial en los cálculos de exposición China-a través-de-India. Si la producción agregada de API doméstica de India es realmente dos veces la tamaña de las importaciones de API de India, la exposición India-a través-de-China es un tercio del a menudo citado 70%.
Ejemplos adecuados para la extrapolación
Una zona conocida por su alta exposición a API chinos es los antibióticos. Sin embargo, en al menos un caso, las estadísticas de antibióticos se han convertido en “China controla el 80 al 90% de la suministra global de API activas”. Podemos aprender mucho sobre las áreas de dominación de China en API a partir de las evaluaciones de exposición a China del gobierno indio. Tal vez la evaluación más exhaustiva se publicó en 2020 por TIFAC, la versión del gobierno indio de MITRE. Usando los datos del gobierno indio y extensas entrevistas con fabricantes, el informe identificó varios APIs que India ya no produce (algunos antibióticos, vitamina C) y muchos donde su producción doméstica es limitada y las importaciones provienen principalmente de China (B vitaminas, heparina, y erythromicina).
Una imagen más centrada en los Estados Unidos
Para obtener una imagen más centrada en los Estados Unidos, analizamos los datos de DMF de la FDA para patrones. Un análisis de DMF solo no nos dará una imagen completa porque algunos fabricantes pueden ser verticalmente integrados y no referenciar un DMF, por lo que asumimos que si la FDF es hecha por una empresa china, también se hace la API (por ejemplo, asumimos que Qilu hace API internamente para su versión de FDF de cisplatino). Primero exploramos cuáles son las drogas más probablemente solo abastecidas por China. Las identificamos como aquellas moléculas donde todas las DMF de API activa son de China, o todas las versiones de FDF son de China. Identificamos 36 tales drogas, casi todas a través del camino de la DMF.
Quizás no sorprende que los antibióticos aparezcan prominentemente, con 12 drogas en la lista, incluyendo streptomicina, tobramicina, y gentamicina. Dos drogas en la lista son fármacos anticancerígenos (mitotano y abemaciclib), y tres están en la lista de medicamentos esenciales de la FDA (isoflurano, ipratropium bromuro, y baloxavir marboxil). Otras drogas en la lista incluyen dobutamina, iodixanol, y iopamidol. Esta lista es significativamente más larga que la lista de drogas identificadas en 2019 como solo abastecidas por China: capreomicina, streptomicina, y sulfadiazina. Más análisis es necesario para determinar cuánto de esta diferencia es debido a la expansión de China en el espacio API en lugar de nuestra incapacidad para observar si las DMF inactivas no no-chinas todavía están siendo utilizadas o si las empresas no-chinas están verticalmente integradas con sus fuentes de API.
La FDA puede hacer esta distinción mucho más fácil que nosotros. En nuestro análisis, nos detenemos a corto de caracterizar la exposición a API chinos para moléculas donde hay una mezcla de DMF chinas y no-chinas. Se podría pesar igual las DMF, justo como varias análisis han hecho con las instalaciones de API. Sin embargo, es poco probable que la producción esté uniformemente distribuida entre las fuentes de API de producción, limitando la información que podría ser útil. La análisis correcto aquí sería pesar las fuentes por el volumen de API. Este tipo de análisis será posible una vez que más fabricantes comiencen a informar el volumen de API a la FDA en sus informes anuales, como se requiere bajo la Ley de Ayuda de Emergencia de 2020.
No diferenciando entre la etapa de API y los componentes de entrada
Few would dispute that China’s grip on upstream drug supply chains is extensive. It is important, however, to differentiate between API exposure and other upstream component exposure to ensure that policy solutions effectively target the real problem. Even if API suppliers for a given drug are or can be moved outside of China, the precursors for it might still be coming from China. Visibility into upstream supply chains is extremely limited, making any existing numbers suspect. In fact, some of the circulating upstream exposure numbers are not actual statistics, but guesstimates. Any numbers would also need to be carefully interpreted – the share of all key starting materials (KSMs) made in China would not be as useful a measure as knowing how many drugs have at least one KSM sourced solely from China. Identifying single points of failure and common nodes in chemical synthesis is a better way to analyze exposure.
Common nodes is an approach explored in the TIFAC report mentioned earlier. The report clearly highlighted areas of vulnerability for specific KSMs, but also with common nodes created by shared technological processes to make intermediates, and shared auxiliary materials necessary for the synthesis of KSMs and intermediates. Table 1 characterizes some of these areas.
In speaking with industry experts, there is a common theme with what has been outsourced to China: chemical synthesis steps that are intrinsically more dangerous and require handling of highly toxic materials, and performing reactions which have the potential to run away. Over time, manufacturers in Europe and even in India have ceded such syntheses to Chinese producers. This is true for fluorination of intermediates and many chemicals needed for synthesis, including KSMs. It is worth pointing out that much of Table 1 – reagents, solvents, and fluorination – does not appear in other reports characterizing U.S. drug supply chain exposure to China.
It could be that Europe is less exposed to China in the areas listed in this table. But it is more likely that these elements are simply not on the radar screen for those analyzing the U.S. drug supply chain exposure to geopolitical risks because of their focus on specific drugs and not common nodes in drug production.
Historically, U.S. drug shortages have primarily been triggered by manufacturing quality problems at the FDF production stage for generic sterile injectables. In the last 20 years, no shortages appear to have been caused by export restrictions. However, shortages due to geopolitical conflicts are a potential future shock. Not only does the risk exist, but it could affect a much wider set of drugs than drugs that have historically been at greatest risk of shortage.
With the potential exposure so wide and a limited government budget, it is critical to fully understand where the greatest vulnerabilities exist.
Our analysis has four key takeaways:
– China’s share in API production of U.S.-bound drugs must be understood in the context of specific drug categories: it is high for some and low for others. The current aggregate exposure across all categories is much less than many reports suggest, likely a quarter of the U.S. drug volume.
– Exposure to China at the intermediates stage is significant for some drugs and is related to specific processes.
– U.S. reports analyzing exposure to China usually highlight their grip on drugs that require fermentation (antibiotics). However, reports do not appear to highlight the importance of China’s control of fluoridation processes used in formulating intermediates for some of the largest therapeutic classes: statins, sartans, and SSRIs.
– China plays a critical role in auxiliary chemicals, such as the production of reagents and solvents. These chemicals are necessary for the chemical synthesis of API and intermediates, yet are omitted from virtually all analyses.
These findings have several implications for policy:
– Policy interest should be broadened from KSMs to auxiliary chemicals that are needed in the chemical synthesis of API or as excipients. Auxiliary chemicals should be included in vulnerability assessments for essential medicines and considered in any stockpiling programs of precursors.
– The significant upstream supply chain exposure means that policies to onshore API will have very limited effect on derisking from China unless those efforts are also coupled with derisking the upstream supply chain for drugs supported by onshoring incentives.
– Policymakers should consider ways to limit China’s expanding API production stage ability. As one of us described elsewhere, imposing and reinforcing through Congress tariffs on Chinese drugs would limit Indian FDF manufacturers’ uptake of Chinese API sources.
The broad upstream exposure also reinforces the need for prioritizing how the U.S. government engages and what role foreign alliances play in solving the seemingly ever-increasing scope of China’s exposure to U.S. drug supply.
