Beta · Free · No login

Trialever

Two ways to get therapies to patients sooner, compared in one unit: life-years.

Does a patient population gain more from trials that finish sooner, or from trials more people can join? Pick a scenario, move the sliders, and see which inputs decide the answer.

life-years per month of earlier approval = chance it works × eligible patients per year × share treated × life-years gained per patient ÷ 12

Pick a scenario

Each scenario sets every slider. Move any slider and the result updates. Sliders tagged “assumption” have little published data behind them.

Scenarios are illustrative. They do not describe a specific drug. Each assumes the trial reaches the participants it needs from under-represented groups.

The therapy
The slider moves in multiples, so it covers 100 to about 30 million.
The slider moves in multiples, from 0.001 to 10.
Earlier approval only helps if the drug succeeds. Many Phase 3 programs fail.
Speed lever
Time saved outside enrollment and FDA review: protocol and contract finalization, site startup, database lock and report writing. Same sites, same enrollment rate.
Access lever
Community sites, remote visits, looser eligibility, more sites. 1.5× means enrollment finishes in two-thirds of the time. Existing sites reaching their planned pace would add about 1.2× on their own (see Enrollment pace below). That counts the same in life-years, but it is a performance gain, not wider access.
Older, sicker, rural or minority patients that trials often leave out.
Covers fewer wasted or harmful treatments and better dosing when trial results apply to the patient. Little direct data exists.
The evidence gain grows with participants reached and is complete at this number.
Zero in most scenarios. High-quality studies find no reliable survival benefit from trial participation alone.

What drives this result

This section updates with the sliders. It breaks the result into parts, shows which single changes would flip it, and ranks the inputs by how much they move it.

Where the numbers come from

What would change the answer

    Each line moves one slider and holds the others where they are.

    Inputs that mostly change the size

    Which inputs move the answer most

    Each input is moved a quarter down and a quarter up from its current value, within the slider range, one at a time. For enrollment speed, the quarter applies to the gain above 1×.

    InputA quarter lowerNowA quarter higher

    How to read it

    Why each month matters

    Each month of earlier approval adds a month of treatment for everyone who would use the therapy in that month. In a large population, even a modest drug gains thousands of life-years per year of acceleration.

    Where the access gain comes from

    Most of the access bar is time saved in enrollment, and a month saved that way counts the same as a month saved by existing sites enrolling faster, which is often the cheaper route. Access wins where enrollment is long and patients are hard to find, as in Duchenne-type rare disease and Alzheimer’s. Speed wins where enrollment is already short or crowded with sites, as in large heart trials, outbreak vaccines and ultra-rare trials that fill quickly. When better evidence for excluded patients is valuable, as for older heart failure patients, access can win anyway.

    Better evidence for more patients

    Trials that exclude older, sicker or rural patients can overstate benefit for the people who end up treated. That value lasts as long as the evidence guides care, but nobody has measured it well. This is the input most worth debating, and the scenarios set it conservatively, so the evidence part of the access bar is small in most of them. Life-years are summed across everyone, so the model does not show who gains them.

    Benefit of joining a trial

    High-quality studies find no reliable survival benefit from trial participation alone, so most scenarios set it to zero. The exception is a disease with no effective treatment, where a trial can be the only route to something new.

    Evidence behind the model

    Some inputs come from the published work below, and each scenario note lists its own sources. The rest are assumptions, tagged on the sliders. Change any value to see what follows.

    Delay cost
    Stewart and colleagues multiplied median survival gains by annual cancer deaths. Summed over 21 cancer drugs, they estimated about 29 life-years lost in North America for every hour of approval delay (J Thorac Oncol 2015 abstract; full analysis in Cancer Medicine 2018). Trialever uses the same arithmetic, scaled down for the chance the drug fails and for partial uptake. Stewart et al., Cancer Med 2018
    Timelines
    Of 2,542 surgical trials registered from 2010 to 2014, 20% finished on time and 46% met their enrollment target. Shadbolt et al., JAMA Netw Open 2023
    Reach
    Across 8,883 cancer patients, 55.6% had no trial available at their institution and 8.1% enrolled. Earlier surveys find that more than half of eligible patients who are offered a trial agree to join. Unger et al., JNCI 2019
    Participation
    Pooled studies show a survival edge for cancer trial participants (HR 0.76). It shrinks when patients are matched on eligibility (HR 0.85) and disappears in high-quality studies (HR 0.91, CI 0.80 to 1.05). Iskander et al., JAMA 2024
    Fit
    Sorafenib added 2.8 months of median survival in its pivotal liver cancer trial. Among 1,532 Medicare patients with advanced liver cancer, treated patients lived a median of 3 months and showed no significant survival difference from matched untreated patients (HR 0.95). Another Medicare analysis found a modest benefit. Sanoff et al., Oncologist 2016
    Access to speed
    A Tufts CSDD model, co-authored and partly funded by a decentralized-trial vendor, assumed decentralized elements cut phase durations by about 10% (about 3 months). Vendor materials claim up to 40% shorter enrollment, but that figure has not been published. DiMasi et al., Ther Innov Regul Sci 2023
    Success rates
    Each scenario’s chance of success uses the phase 3 to approval rate for its area: about 44% in oncology, 46% in cardiovascular disease, 57% in rare disease outside cancer and 58% for vaccines. BIO, Informa and QLS, 2021
    Operations
    Site startup takes about five to six months, and database lock averages about 36 days. Saving 6 months also means shortening protocol finalization, submission preparation and report writing, so the 6-month default used in most scenarios is generous to speed. Lamberti et al., 2018 · Tufts and Veeva survey
    Enrollment pace
    In AIMR’s analysis of 5,465 completed industry trials with a US site, completed trials ran at about 82% of planned speed, and 70% in oncology. Bringing existing sites up to plan is worth about 1.2× on the enrollment slider, or 1.4× in oncology. AIMR analysis of ClinicalTrials.gov and AACT data, 2026 (unpublished; methods available on request).

    Limits of the model

    • The model is linear. It ignores uptake ramps, discounting, competing therapies and quality of life.
    • Shorter enrollment moves approval month for month only when enrollment is the bottleneck. In event-driven trials, such as many cancer trials, the gain is smaller, so treat the enrollment saving as an upper bound.
    • Adding sites also adds startup work, and new sites often enroll slowly. Looser eligibility can also dilute the measured benefit. The model ignores all three.
    • Benefit per patient applies to everyone treated, including patients unlike the trial population, so both bars may run high.
    • The levers add together. Doing both is worth roughly the sum of the two bars.
    • Use it to compare rough sizes. The exact values are not reliable.