Ireland is missing about 73 clinical trials a year, and comes last of 11 countries for trials of medicines

The gap

From 2021 to 2025, trials with at least one site in Ireland started at a rate of 21.38, computed, select to see the proof and check it a year for every million people. Ten comparison countries had a median of 34.79, computed, select to see the proof and check it. That puts Ireland 10th of 1110, computed, select to see the proof and check it, ahead only of Portugal. Denmark, which has a similar population, starts 82.611, computed, select to see the proof and check it per million, 3.9 times12, computed, select to see the proof and check it Ireland’s rate.

Which trials are missing

The registry lets us split the gap two ways: by what is being tested, and by who sponsors the trial.

Trials of medicines, which carry a phase 1 to 4 label, start in Ireland at 11.813, computed, select to see the proof and check it a year per million people, against a median of 18.114, computed, select to see the proof and check it. No other country here starts fewer. The margin is small: Portugal is just ahead, and 315, computed, select to see the proof and check it more Irish trials of medicines over the five years would have lifted Ireland to 10th. Reaching the median would mean about 3416, computed, select to see the proof and check it more trials of medicines a year. Other studies, of devices, surgery, exercise, diet or how care is organised, start at 9.517, computed, select to see the proof and check it, against 16.818, computed, select to see the proof and check it.

By sponsor, industry-sponsored trials start at 12.519, computed, select to see the proof and check it per million a year, against a comparison median of 15.920, computed, select to see the proof and check it: about four-fifths of the usual level. Counting only company-sponsored trials of medicines, Ireland is 11th of 1121, computed, select to see the proof and check it. Trials not sponsored by industry, from hospitals, universities and research groups in Ireland or abroad, start at 8.822, computed, select to see the proof and check it, against 19.123, computed, select to see the proof and check it: less than half.

Of the 73 missing trials a year, industry accounts for 1824, computed, select to see the proof and check it. The other 5525, computed, select to see the proof and check it are not sponsored by industry; that remainder is a difference, not a separate count. A caution on that second number: three in four non-industry studies in Ireland, and three in four in Denmark, are not trials of medicines. Countries differ in how routinely they register those studies, so the non-industry gap is less certain than the gap in medicine trials. Our reading For patients, trials of medicines are the main way to get a treatment before it is generally available. That is where Ireland ranks lowest.

Not a bad year, a pattern

Ireland has been below the comparison median in every year since 2014. In 2025 it started 21.926, computed, select to see the proof and check it trials per million against a median of 29.327, computed, select to see the proof and check it. Our reading The gap narrowed in 2025 more because the comparison median fell than because Ireland rose. Counts for 2025 may still grow as late registrations arrive.

Why: what officials, industry and patients say

The registry can show the gap. It cannot show the cause. For that we read what the Department of Health, the Health Research Board, the industry and patient groups have published, and we label each point by where it comes from.

In August 2024 the then Minister for Health, Stephen Donnelly, set up a National Clinical Trials Oversight Group, saying: “I want to double the number of clinical trials taking place in Ireland28, said on the record, select to see where.” The group reported in November 202529, sourced, select to see the document. It named the obstacles: contract negotiation, costing processes and data-protection compliance, national shortages of key skills and a reliance on short-term contracts, and the lack of a national electronic health record30, sourced, select to see the document. It set out 20 recommendations31, sourced, select to see the document, including a clinical trials advisory council by early 2026 and a national clinical trials body by the third quarter of 2027. In July 2026 the health minister appointed Prof Donal Brennan32, sourced, select to see the document, who chaired the oversight group, to lead the new council. Most of the years this index counts, 2021 to 2025, came before those reforms. We have asked the Department of Health what has been done since.

The industry measures the delay. In IPHA’s figures, the time from EU approval to the first Irish site being ready was 213 days33, sourced, select to see the document, the slowest in the EU. Patient groups count the cost. The Irish Cancer Society found that only 2 to 3 per cent34, sourced, select to see the document of cancer patients take part in trials, against a national target of 6 per cent.

Money has gone in. By 2023 the Health Research Board had invested more than €150 million35, sourced, select to see the document in clinical trials infrastructure since 2002. Our reading The obstacles on that list, contracts, data-protection paperwork, short-term posts, slow every trial, and fall hardest on trials a hospital runs itself, with no company team to absorb them. It is a reading, not a proof: the registry records that a trial started, not why another one did not.

What the leader does differently

Denmark starts more trials per head than any other country here. Since autumn 2025 its medicines agency and ethics committees have reviewed eligible phase I and phase I/II applications within 14 days36, sourced, select to see the document, and phase I applications rose by almost 80 per cent in a year, according to the government’s investment agency. Denmark starts 38.237, computed, select to see the proof and check it trials of medicines per million a year against Ireland’s 11.8, and 55.138, computed, select to see the proof and check it trials not sponsored by industry against Ireland’s 8.8. Our reading Fast approval helps the first; a health service that treats research as part of care fills the second.

Two other counts point the same way

A single analysis can be wrong in ways its authors cannot see. So we compared our result with two others, made by different people over different years. They measure different things, so the ratios should not match.

Our reading All three put Denmark at between two and a half and four times Ireland. The cancer count uses the same registry as we do, so it checks our arithmetic more than our source. The industry count comes from a different source and agrees.

Where Ireland does well

Trials with Irish sites tend to be seen through. Of those that closed from 2016 to 2025, 86.9%42, computed, select to see the proof and check it completed rather than stopping early, 4th of 1143, computed, select to see the proof and check it. Results were posted for 80.1%44, computed, select to see the proof and check it of completed phase 2 to 4 trials, 3rd of 1145, computed, select to see the proof and check it.

As in most of the comparison countries, few post results quickly: 9.1%46, computed, select to see the proof and check it posted results within a year of finishing, the same as the comparison median of 9.1%47, computed, select to see the proof and check it.

Where it lags

82.6%48, computed, select to see the proof and check it of new trials were registered before the first patient joined, 9th of 1149, computed, select to see the proof and check it. And 10.7%50, computed, select to see the proof and check it of records since 2014 now carry the registry’s “unknown status” label because the sponsor has not confirmed them in time, 8th of 1151, computed, select to see the proof and check it (lower is better). Our reading Once a trial is open in Ireland it is usually finished and reported well. The problem is getting trials opened, which is what the oversight group’s recommendations target.

What the authorities said

On 5 November 2026 we sent these findings, with our questions, to the Department of Health, the Health Research Board and the HRB National Clinical Trials Office, and asked for a reply by Wednesday 11 November 2026, 17:00. Replies are printed in full.

How to check this story

There are three ways to check this story, from one minute to an afternoon.

1. Look it up about 1 minute · anyone

   Select any marked number or phrase to see where it comes from. Computed numbers show the formula and inputs, and link to the live registry query. Sourced facts show the document, its date, what it says and any caveat.

2. Recompute it about 10 seconds · anyone

   The check button downloads the 47,265 country records behind this edition (one row per trial per country, 32,816 distinct trials) and reruns our calculation in your browser. It compares its own result with the value the story was written from for every published number, re-checks every stress test, and confirms that the file has the SHA-256 fingerprint published below, the same fingerprint as the archived copy.

   Results of the check

   Run the check to see the full list.

3. Rebuild it from scratch about an hour · researchers

   The code that made the edition, the raw-data collector and our independent quality-control report are public. Run node fetch.mjs, then node compute.mjs, and compare with the edition file. The browser engine and the page script carry the EUPL-1.2 open licence. The QC report includes a test where we planted errors on purpose to make sure our checks catch them.

   Code on GitHub Archived edition and trial-level file (DOI 10.5281/zenodo.24229455) Quality-control report Trial-level file Airmid Éireann open data

What this cannot tell you

• It counts trials registered on ClinicalTrials.gov. Trials registered only in the EU’s own trial system are not included.
• A trial with sites in several countries counts once in each of them, so the 44,770 country records cover 30,943 distinct trials.
• “Not sponsored by industry” uses the registry’s sponsor class. It includes universities and research groups outside Ireland, and a company can fund a trial that a hospital sponsors; the registry does not show either.
• “Trials of medicines” are trials with a phase 1 to 4 label, which the sponsor assigns. Most other studies, of devices, surgery or care, carry no phase. Registering those studies on ClinicalTrials.gov is less routine in some countries than others.
• Results measures look only at ClinicalTrials.gov. Many European trials post results to the EU’s own systems (EudraCT or CTIS) instead.
• Each year in the trend is divided by the 2025 population (see the stress test).
• The gaps of 73 and 18 trials a year are worked out from rates rounded to one decimal place, as in the edition. From unrounded rates the gap is 72.7, of which industry is about 19 and the remainder about 54. The remainder is a difference, not a separate count.
• Registry entries are written by sponsors. Where a sponsor has not updated a record, we report what the registry says, not what happened.
• The measures describe how much research happens and how well it is reported. They do not judge the quality of any single trial, hospital or sponsor.