Disease-area comparison
Oncology has the most stopped trials, but not the highest biological-signal share
Oncology contains 7,941 stopped records, but its 7.3% biological-signal share does not lead the disease-area comparison. Ophthalmology ranks highest among areas with at least 200 records.
- The current database contains 23,617 stopped clinical trial records across its disease-area taxonomy.
- Oncology is the largest disease-area slice with 7,941 stopped records, including 582 biological signals, a 7.3% share.
- Ophthalmology has the highest biological-signal share among disease areas with at least 200 stopped records: 13.4% (35 of 261).
- The comparison includes only disease areas with at least 200 stopped records to reduce small-sample distortion.
- These percentages describe efficacy/futility and safety signals within stopped records; they are not overall clinical trial failure rates.
The short version
Oncology is the largest disease area in the stopped-trial database with 7,941 records. That makes it the easiest area to notice, search, and quote. It does not make oncology the disease area with the highest concentration of likely biological failure signals: 582 oncology records are classified as efficacy/futility or safety, a 7.3% share.
Ophthalmology ranks highest among disease areas with at least 200 stopped records, with 35 biological signals among 261 records (13.4%). This is a useful reminder that volume and concentration answer different questions.
Volume and share answer different questions
Raw volume tells us where the database contains the most stopped trials. Share asks a narrower question: among stopped records in one disease area, what proportion carries source language classified as efficacy/futility or safety? Both views are useful, but they should not be substituted for each other.
A large area can produce many scientific failure signals while still having a lower signal share because it also contains a very large number of operational, strategic, enrollment, regulatory, or unclear stops. A smaller area can have fewer signals in absolute terms but a higher concentration within its stopped-trial slice.
Why oncology volume can be misleading
Oncology has more stopped records than any other disease area in this dataset. It also has substantial efficacy and safety counts. If I looked only at totals, I might conclude that oncology is the clearest failure area. The denominator changes that interpretation.
Cancer development includes a wide variety of mechanisms, combinations, investigator-led studies, biomarker populations, and operationally complex protocols. The large denominator includes many stops that do not establish failed biology. That is why the share of classified biological signals is more informative than the headline count alone.
What a higher share does and does not mean
A higher share means that efficacy/futility or safety language appears more often within the stopped records assigned to that disease area. It does not mean that drugs in that disease area have a higher overall clinical failure rate. We do not have the full denominator of all successful, ongoing, and completed trials in this analysis.
The ranking is therefore a stopped-trial signal comparison, not a probability of technical success and not a league table of therapeutic quality. It is best used to decide where source-level review may be especially valuable.
How I would use this result
I would use the disease-area comparison as a triage layer. First identify areas with a meaningful record count and a comparatively high signal share. Then separate efficacy/futility from safety, because those categories can imply very different development problems.
After that, I would move to phase, intervention, sponsor, and individual NCT records. The useful question is not simply which area ranks first. It is whether the pattern persists inside a comparable phase, modality, mechanism, or patient population.
The limits of the comparison
Disease areas are assigned through a keyword-based taxonomy derived from conditions and MeSH terms. Some trials span more than one clinical area, and the primary assignment can simplify that complexity. Classification is also based on registry language, which can be brief or incomplete.
To reduce unstable small-sample rankings, this comparison includes only disease areas with at least 200 stopped records. Even with that threshold, every percentage should be read as an analytical screening signal and verified against the underlying trial records.
Biological-signal share by disease area
| Disease area | Signals / stopped records |
|---|---|
| Ophthalmology | 13.4% (35 / 261) |
| Neurology | 11.9% (119 / 1,003) |
| Dermatology | 10.1% (32 / 317) |
| Immunology & Autoimmune | 10.0% (67 / 669) |
| Respiratory | 9.7% (61 / 629) |
| Gastroenterology & Hepatology | 9.1% (139 / 1,529) |
| Hematology (non-onc) | 8.9% (30 / 338) |
| Infectious Disease | 8.5% (145 / 1,709) |
| Endocrine & Metabolic | 7.4% (59 / 799) |
| Oncology | 7.3% (582 / 7,941) |
Efficacy and safety composition
| Disease area | Efficacy / safety |
|---|---|
| Ophthalmology | 22 / 13 |
| Neurology | 92 / 27 |
| Dermatology | 28 / 4 |
| Immunology & Autoimmune | 41 / 26 |
| Respiratory | 37 / 24 |
| Gastroenterology & Hepatology | 87 / 52 |
| Hematology (non-onc) | 11 / 19 |
| Infectious Disease | 87 / 58 |
| Endocrine & Metabolic | 28 / 31 |
| Oncology | 305 / 277 |
Continue from here
FAQ
Which disease area has the most stopped clinical trials?
Oncology has the largest stopped-trial volume in the current dataset. That does not mean it has the highest share of efficacy or safety signals within its disease-area slice.
Is this a clinical trial failure-rate ranking?
No. The denominator contains stopped trials only. It does not include every successful, completed, or ongoing trial, so the percentages must not be interpreted as overall failure rates.
Why exclude disease areas with fewer than 200 records?
The minimum-record threshold reduces rankings driven by very small samples. It does not remove all uncertainty, but it makes comparisons more stable and useful.
Source note: counts are generated from the current ClinicalTrials.gov-derived stopped-trial dataset used by ClinicalTrialFailures.com. These labels are analytical screening signals, not medical advice.