FDA still preannounces part of its foreign drug plant inspections while domestic ones are almost always unannounced — as of its 2026-08-03 report, unannounced rates were 87 percent in India and 61 percent in China for FY2026 to date, and the report gives no rate for any other country

What is happening?

Most of the plants that make medicines for the United States are not in the United States. As of October 2022, 58 percent of roughly 4,800 drug manufacturing facilities serving the US market were located abroad, according to the Government Accountability Office.

Those plants are inspected under a different practice from domestic ones. GAO reported on 2024-02-06 that domestic inspections have almost always been unannounced, while the agency has generally preannounced foreign inspections up to 12 weeks in advance.

FDA has said publicly that it intends to close that difference. On 2025-05-06 the agency announced it was extending unannounced inspections beyond the India and China pilots to foreign facilities making food, drugs, biologics and medical devices, and described the change as treating all manufacturing facilities equally.

Its own numbers, published on 2026-08-03, show the difference narrowing without disappearing. The India office went from 47 percent unannounced across 139 inspections in FY2025 to 87 percent across 76 inspections in FY2026 to date. The China office went from 19 percent across 192 inspections in FY2025 to 61 percent across 114 in FY2026 to date. The same statement reported more than 17,000 inspections in FY2025, domestic and foreign combined.

Two countries are the whole published picture. Four years after the 2022 GAO recommendations and more than a year after the 2025 announcement, no source opened here gives an unannounced rate for any other country, and the list of countries covered by the expansion has not been published.

Whose problem is this?

The body that would have to finish the change is the same body that announced it and the same body that publishes the only measure of whether it is happening. Nothing outside that body sets the date, and nothing outside it decides what gets reported.

Where does this problem end?

The boundary here is the inspection method rather than the quality of any particular medicine, because no source opened here compares what is found on announced visits with what is found on unannounced ones.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state: no source opened here gives an official number that would count as parity, an official date by which foreign and domestic practice are meant to match, or the list of countries the 2025 expansion covers. The 2026 statement reports movement and names no destination. A widely repeated account holds that the 2012 generic drug user fee legislation set an equity goal between domestic and foreign inspection frequency, but that text could not be confirmed in a primary document in this round and it is not used here.

How big is it?

The scale that can be counted here is measured in facilities and in inspections, and not in people.

About 2,800 facilities abroad. That is 58 percent of roughly 4,800 facilities serving the US market as of October 2022. In fiscal 2022 FDA inspected 6 percent of them, down from 37 percent in 2019, and in India alone 3 percent, down from 45 percent in 2019. In 2021 foreign inspections numbered 99, under 4 percent of the facilities in scope. Over the same period the share of foreign facilities that had never been inspected or had gone more than five years without one rose from 30 percent in 2020 to more than 80 percent in 2022.

Those two counts answer different questions. The first says how many plants sit under a weaker version of the inspection instrument, and the second says how rarely any version of it arrives at all.

The patient-side number is not derived here. Two of the three terms that would be needed exist in these sources and the third does not. The population was 340.1 million as of 2024-07-01 and 90 percent of prescriptions dispensed in 2024 were generics, but nothing opened here states how much of what is dispensed in the United States is made abroad, and nothing counts the people who received a medicine from a plant that had not been inspected. Multiplying a national population by a share of buildings would treat a count of plants as a share of medicine supply, and no source opened here supports that step.

Under what conditions does it arise?

1. The instrument changes at the border. An unannounced inspection and a preannounced one are not the same check, which is why FDA described the 2025 expansion as treating all manufacturing facilities equally. What none of these sources supplies is a measurement of the difference. One industry newsletter states that foreign inspections turn up serious deficiencies more than twice as often as domestic ones, but it names no base year and no sample, and it compares foreign with domestic rather than announced with unannounced.

2. Notice compounds with rarity. A plant inspected once in five years and told twelve weeks beforehand is being checked on a schedule it can see coming twice over. The coverage figures and the notice practice are usually reported separately, and they land on the same facilities.

3. The workforce shrank while the task widened. GAO reported in November 2024 that the drug inspector vacancy rate rose from 9 percent in November 2021 to 16 percent in June 2024, that departures of experienced investigators outpaced hiring, and that this bears directly on carrying out the unannounced inspection work.

4. Nothing in the process names a finish line. Of the three recommendations GAO made in January 2022, the first was closed as implemented in October 2023, the second stood as awaiting resources as of July 2025, and the third had no implementation date as of July 2024. The 2025 announcement carried no deadline and the 2026 statement carries no target rate.

What has been tried?

Three directions have run at once — fund more foreign inspections, change how they are announced, and staff the people who do them. The first and second have produced measurable movement in two countries. The third moved in the opposite direction over the same years.

What was found?

Why is it still unsolved?

Enforcement absent — the check that a plant cannot prepare for is standard at home and partial abroad, and nothing in the record names the point at which the two are supposed to meet.

The first part is that the two inspection methods were never the same thing. GAO reported the difference in February 2024 as a persistent challenge in overseeing foreign drug manufacturing, and FDA itself framed the 2025 expansion as equal treatment of all manufacturing facilities. Both statements describe the gap. Neither of them, and no other source opened here, measures what the gap costs, because nobody opened here has compared what announced visits find with what unannounced ones find.

The second part is that method sits on top of frequency, and frequency has been the weaker of the two. A plant inspected in 6 percent of a year, as the foreign estate was in fiscal 2022, is one where the choice between notice and no notice arrives rarely enough that either version is a light touch. More than 80 percent of foreign facilities had gone over five years without an inspection or had never had one, as of 2022. Raising the unannounced share of a small number of visits and raising the number of visits are separate achievements, and only the first is being published.

The third part is capacity moving the other way. The vacancy rate among drug inspectors rose from 9 percent in November 2021 to 16 percent in June 2024, inspections fell 36 percent between FY2019 and FY2023, and GAO tied that staffing trend directly to the ability to carry out unannounced inspections. A policy that requires more travel by fewer experienced people has a ceiling that no announcement can lift.

The fourth part is that there is no clock and no destination. One of the three 2022 recommendations closed, one waits on resources, one has no date. The 2025 expansion named no countries and no deadline. The 2026 report gives two percentages that both went up and does not say what number would be enough. Movement without a stated target is difficult to tell apart from movement that has stopped, because the same publication would describe either one.

What observation would mean it is solved?

Candidates — (a) FDA publishes an unannounced inspection rate for every country where it inspects, on a stated schedule (b) the unannounced share of foreign inspections reaches the domestic level and holds there across several fiscal years (c) the share of foreign facilities inspected within a fixed interval rises toward the domestic norm.

(a) alone counts publication, not practice. Reporting a rate for every country would end the situation where two countries stand for a global programme, and it would still leave the rate itself at whatever level it happens to be. It is a precondition for reading the other two rather than an answer.

(b) alone can be produced by shrinking. A rising unannounced share is easy to read as progress and hard to read as parity. The India office reported 87 percent across 76 inspections in a partial fiscal year against 47 percent across 139 in a full one, and a partial year cannot be compared with a full year on totals. A share can also rise because the announced visits were the ones dropped. The share has to be read next to the count, and the count next to the number of facilities.

(c) alone is the old problem. Coverage without method is what the programme already had before the pilots, when foreign plants were visited rarely and told in advance. The three have to be read together, and none of the three has a published number attached to it that FDA has said it is aiming for.

What is it connected to?

Fills with researchdrug shortage and supply concentration in generic medicines, the reliance of the US market on active ingredient production abroad, trade and industrial policy affecting where manufacturing sits, and how other regulators inspect their own foreign suppliers. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• Why foreign inspections are preannounced at all. No source opened here carries an on-the-record explanation from FDA of what the notice period is for. Interpreter services and inspector staffing appear in the GAO recommendations as conditions attached to the foreign programme, which is not the same as a stated reason.
• What the unannounced rate is anywhere except India and China. No source opened here gives a figure for any other country. Whether that is because the data does not exist or because it exists and has not been published could not be settled in this round.
• What number counts as done. The 2026 statement reports two rising percentages and names no target rate, and no source opened here gives an official date by which foreign and domestic practice are meant to match.
• What unannounced inspections find that announced ones do not. No 2025 or 2026 quantitative comparison was found. The one figure located, that foreign inspections find serious deficiencies more than twice as often, has no base year and no sample stated in the source itself, and it compares foreign with domestic rather than announced with unannounced.
• Whether the inspection gap has reached any patient. No source opened here quantifies a link between preannounced inspection and defective medicine reaching a person, and this document does not assert one.
• Whether the 2012 user fee legislation set an equity goal. Secondary industry and academic accounts repeat that it did. The primary texts sought in this round, a 2017 House committee witness statement and an HHS Office of Inspector General report, could not be read — the first because the PDF carried no extractable text layer and the second because automated retrieval was refused with HTTP 403.
• How the FY2026 partial-year figures will end. The India and China rates are year-to-date at 2026-08-03 and no full-year figure exists yet.

See the evidence

No primary rulemaking or statutory text was read in full. The two GAO product pages, the FDA press release of 2025-05-06 and the FDA statement of 2026-08-03 were opened directly, and the figures graded high above come from those four. The inspection coverage figures for fiscal 2022 come from a news organisation citing a 2022 GAO report that was not itself opened for those numbers, which is why they are graded medium even though the underlying source is a government audit. Where sources overlap they agree: the two-country limit of the published rates appears both in the 2026 FDA statement and in an industry regulatory note on the 2025 expansion, and the direction of the staffing trend appears in the 2024 GAO report and in the coverage decline reported elsewhere. Where a claim rests on one weak source it is marked low and no statement in this document is built on it, which covers the deficiency multiple and the asserted statutory equity goal. The position of the agency named throughout is present in its own words through two of its own publications rather than through a request for comment, which is the only form of response available to this process. This is a Path A output, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range Not derivable

The reason and what is missing are listed under “What is missing” below