The finder
Pick what you want to change. See who actually measured it.
Every other tool like this gives you a compound and a match score. Those scores are not measured. Ours does not exist. What you get instead is what was actually measured and where: trials in people first with the participant count, then the animal and cell work kept separate, then what people using it report, and the trials that found nothing, counted rather than dropped.
82 of the 444 compounds here have never been through a trial in people. That is worth knowing and it is not a verdict: a molecule nobody can patent has no sponsor to pay for a trial, so a thin human record is usually a fact about money rather than about the compound. Where that is the case, this says so and shows you what does exist.
What are you trying to change?
Pick as many as you like. The number on each one is how many people it has been measured in across the whole archive, so you can see where the evidence is before you choose.
Or start from the compounds
Somebody handed you a stack. What is actually known about it?
Add what you are looking at and you get every outcome those compounds have been measured for in people, and then the longer list of the ones they have not.
Heart attacks and strokes
Does it prevent an actual heart attack?
The hardest endpoint in this entire catalogue and the rarest. A trial that counted events is worth more than a hundred that counted markers.
178,846 people · 54 human studies · 40 compounds
- 23,347 people4 studies1 found nothing
found nothingEMPA-REG OUTCOME. 7,020 people with type 2 diabetes at high cardiovascular risk, median observation 3.1 years. The primary composite of cardiovascular death, nonfatal myocardial infarction or nonfatal stroke occurred in 10.5 percent on empagliflozin against 12.1 percent on placebo (hazard ratio 0.86, P = 0.04). Death...
New England Journal of Medicine, 2015 · 7,020 people
EMPA-KIDNEY. 6,609 people with chronic kidney disease, median 2.0 years. Kidney disease progression or cardiovascular death occurred in 13.1 percent against 16.9 percent (hazard ratio 0.72). Notably for anyone reading the drug as a mortality intervention, death from any cause was 4.5 percent against 5.1 percent and...
New England Journal of Medicine, 2023 · 6,609 people
EMPEROR-Preserved. 5,988 people with class II to IV heart failure and an ejection fraction above 40 percent, median 26.2 months. Cardiovascular death or heart failure hospitalisation occurred in 13.8 percent against 17.1 percent (hazard ratio 0.79). The authors state the effect was mainly driven by fewer heart failure...
New England Journal of Medicine, 2021 · 5,988 people
- 17,604 people1 study
17,604 adults with cardiovascular disease and BMI 27 or higher but no diabetes: semaglutide 2.4 mg cut major adverse cardiovascular events 20% (6.5% vs 8.0%; HR 0.80) over a mean 39.8 months.
New England Journal of Medicine, 2023 · 17,604 people
- 14,752 people1 study1 found nothing
found nothing14,752 patients, median 3.2 years: MACE in 11.4% on weekly exenatide vs 12.2% on placebo (HR 0.91, 95% CI 0.83 to 1.00), noninferior for safety but not statistically superior.
New England Journal of Medicine, 2017 · 14,752 people
- 13,299 people1 study
13,299 patients with type 2 diabetes and atherosclerotic disease: primary MACE in 12.2% on tirzepatide vs 13.1% on dulaglutide (HR 0.92), meeting noninferiority against an active comparator with proven benefit; basis of the August 2026 cardiovascular indication.
New England Journal of Medicine, 2025 · 13,299 people
- 9,901 people1 study
9,901 participants (only about a third with prior cardiovascular disease), median 5.4 years: MACE in 12.0% on dulaglutide vs 13.4% on placebo (HR 0.88; p=0.026).
The Lancet, 2019 · 9,901 people
- 9,503 people2 studies1 found nothing
found nothingThe historical result that kept niacin in the guidelines for thirty years, and it is weaker than its reputation. In the Coronary Drug Project, 8,341 men aged 30 to 64 with a previous myocardial infarction were treated between 1966 and 1975. Niacin modestly reduced definite non-fatal recurrent myocardial infarction but...
Journal of the American College of Cardiology, 1986 · 8,341 people
Observational metabolomics rather than a trial, and it proposes a mechanism for why more niacin might be actively unhelpful. In a discovery cohort of 1,162 stable cardiac patients and validation cohorts of 2,331 US and 832 European patients, serum levels of the terminal metabolites of excess niacin, 2PY and 4PY, were...
Nature Medicine, 2024 · 1,162 people
- 9,463 people1 study
9,463 adults aged 40 and over with type 2 diabetes and cardiovascular disease, 610 sites, 28 countries, randomised 1:1 to albiglutide 30 to 50 mg weekly or placebo, median follow-up 1.6 years. Primary composite (cardiovascular death, myocardial infarction, stroke): 338 of 4,731 (7 percent) against 428 of 4,732 (9...
Lancet, 2018 · 9,463 people
- 9,340 people1 study
9,340 patients with type 2 diabetes at high cardiovascular risk, median 3.8 years: primary MACE in 13.0% on liraglutide vs 14.9% on placebo (HR 0.87, a 13% relative reduction; P=0.01 for superiority).
New England Journal of Medicine, 2016 · 9,340 people
- 7,706 people3 studies2 found nothing
found nothingRELAX-AHF-2, the confirmatory trial, and it failed. 6,545 patients in the intention to treat analysis, same drug, dose, 48 hour infusion and entry criteria as RELAX-AHF. Death from cardiovascular causes at 180 days occurred in 285 of 3,274 patients (8.7 percent) on serelaxin and 290 of 3,271 (8.9 percent) on placebo,...
New England Journal of Medicine, 2019 · 6,545 people
found nothing1,161 patients hospitalised with acute heart failure randomised 1:1 to a 48 hour infusion of serelaxin at 30 micrograms per kilogram per day or placebo within 16 hours of presentation. Serelaxin improved the visual analogue scale dyspnoea endpoint by 448 mm times hours (95 percent CI 120 to 775, p = 0.007) but had no...
The Lancet, 2013 · 1,161 people
An adjudicated analysis of every death in RELAX-AHF-2, run specifically to understand why the two trials disagreed. By 180 days 11.5 percent of patients had died, 38 percent of those deaths from heart failure. Unlike in RELAX-AHF, there was no apparent effect of serelaxin on any category of cause of death. Patients...
JACC: Heart Failure, 2020 · no headcount in the line
- 7,141 people1 study
ASCEND-HF: 7,141 people hospitalised with acute heart failure randomised to nesiritide or placebo for 24 to 168 hours on top of standard care. Dyspnoea improvement 44.5 versus 42.1 percent at 6 hours and 68.2 versus 66.1 percent at 24 hours, neither meeting the prespecified threshold. Rehospitalisation or death within...
New England Journal of Medicine, 2011 · 7,141 people
- 7,050 people3 studies1 found nothing
found nothing4,228 asymptomatic people with ejection fraction 0.35 or below, mean follow-up 37.4 months. Total mortality 313 versus 334 deaths, risk reduction 8 percent (95 percent CI -8 to 21, p = 0.30), not significant. Death or development of heart failure 630 versus 818 (29 percent reduction, p < 0.001). The null mortality...
New England Journal of Medicine, 1992 · 4,228 people
2,569 patients with symptomatic heart failure and ejection fraction 0.35 or below, enalapril 2.5 to 20 mg a day or placebo, mean follow-up 41.4 months. Deaths 452 (35.2 percent) versus 510 (39.7 percent), risk reduction 16 percent (95 percent CI 5 to 26, p = 0.0036); death or heart failure hospitalisation 613 versus...
New England Journal of Medicine, 1991 · 2,569 people
253 patients with NYHA class IV heart failure, double-blind, enalapril 2.5 to 40 mg a day or placebo on top of conventional therapy. Six-month mortality 26 versus 44 percent (40 percent reduction, p = 0.002); one-year 31 percent reduction; 50 versus 68 deaths overall (27 percent). Entire benefit in progressive heart...
New England Journal of Medicine, 1987 · 253 people
- 6,800 people1 study1 found nothing
found nothing6,800 patients with heart failure and ejection fraction 0.45 or less randomised to digoxin or placebo, average follow-up 37 months. Mortality unchanged: 1,181 deaths (34.8 percent) versus 1,194 (35.1 percent), risk ratio 0.99, 95 percent confidence interval 0.91 to 1.07. Hospitalisation for worsening heart failure...
New England Journal of Medicine, 1997 · 6,800 people
- 6,068 people1 study
6,068 patients: primary MACE hazard ratio 1.02, showing noninferiority to placebo but no cardiovascular benefit.
New England Journal of Medicine, 2015 · 6,068 people
- 5,650 people2 studies
UKPDS post-trial monitoring of 4,209 randomised participants. Glycaemic differences disappeared within a year of the trial ending, yet at 10 years the sulfonylurea-insulin group retained reductions in any diabetes-related endpoint (9 percent, p = 0.04), microvascular disease (24 percent, p = 0.001), myocardial...
New England Journal of Medicine, 2008 · 4,209 people
DCCT/EDIC: 1,441 people with type 1 diabetes randomised to intensive or conventional insulin therapy for a mean 6.5 years, then followed observationally. Over a mean 17 years, intensive treatment reduced any cardiovascular event by 42 percent (95 percent confidence interval 9 to 63, p = 0.02) and non-fatal myocardial...
New England Journal of Medicine, 2005 · 1,441 people
- 5,470 people2 studies
TREAT: 4,038 people with type 2 diabetes, chronic kidney disease and anaemia randomised to darbepoetin alfa targeting 13 g/dL or to placebo with rescue below 9 g/dL. Death or a cardiovascular event, hazard ratio 1.05 (0.94 to 1.17); death or end-stage renal disease, hazard ratio 1.06 (0.95 to 1.19). Fatal or non-fatal...
New England Journal of Medicine, 2009 · 4,038 people
CHOIR: 1,432 people with chronic kidney disease randomised open-label to a haemoglobin target of 13.5 or 11.3 g/dL, median 16 months. The composite of death, myocardial infarction, heart failure hospitalisation and stroke occurred 125 times in the high-target group and 97 times in the low-target group, hazard ratio...
New England Journal of Medicine, 2006 · 1,432 people
- 5,246 people1 study
TRAVERSE. 5,246 men aged 45 to 80 with preexisting or high risk of cardiovascular disease, symptoms of hypogonadism and two fasting testosterone levels below 300 ng/dL, randomised double-blind to transdermal testosterone gel titrated to 350 to 750 ng/dL or placebo. Mean treatment 21.7 months, mean follow-up 33.0...
New England Journal of Medicine, 2023 · 5,246 people
- 5,246 people1 study
TRAVERSE, 5,246 men, no excess of cardiovascular death, heart attack or stroke against placebo. Included because it is the safety context for any testosterone ester, with the qualifier that TRAVERSE used a transdermal gel titrated to a defined range rather than an injectable.
New England Journal of Medicine, 2023 · 5,246 people
- 4,444 people1 study
4,444 people with angina or previous myocardial infarction, randomised double-blind, median follow-up 5.4 years. 256 placebo patients (12 percent) died against 182 on simvastatin (8 percent), relative risk of death 0.70 (95 percent CI 0.58 to 0.85, p equals 0.0003). Coronary deaths fell from 189 to 111. Major coronary...
Lancet, 1994 · 4,444 people
- 4,076 people1 study
4,076 high-risk patients: MACE hazard ratio 0.73 and composite renal outcome hazard ratio 0.68 versus placebo.
New England Journal of Medicine, 2021 · 4,076 people
- 3,080 people1 study
3,080 patients randomised at 300 sites before the trial was stopped for futility. The composite of death, non-fatal stroke or severe left ventricular dysfunction through day 28 occurred in 5.1% on acadesine against 5.0% on placebo (odds ratio 1.01). An earlier 1997 meta-analysis of five smaller trials had been...
JAMA, 2012 · 3,080 people
- 2,619 people4 studies1 found nothing
Pooled analysis of two Japanese cohorts, 2,435 patients with acute decompensated heart failure split by carperitide dose. Cardiovascular and all-cause mortality within 1 year were lower in the low-dose group (at or above 0.02 micrograms per kilogram per minute) than in the no-carperitide and very-low-dose groups. This...
ESC Heart Failure, 2022 · 2,435 people
109 hospitalised patients randomised to tolvaptan or carperitide. Symptoms and plasma BNP improved similarly in both groups. Blood pressure was lower after carperitide, and fewer adverse events such as worsening heart failure and hypotension requiring discontinuation occurred with tolvaptan (p = 0.027).
Journal of Clinical Pharmacology, 2013 · 109 people
75 patients with acute heart failure given 0.0125 micrograms per kilogram per minute of carperitide for 6 hours. Lower baseline plasma ANP (r = -0.35, p = 0.002) and lower vasopressin (r = -0.54, p < 0.001) predicted a larger diuretic response; baseline blood pressure, renal function and prior loop diuretic use did...
ESC Heart Failure, 2022 · 75 people
- 420 people1 study
420 patients with moderate to severe chronic heart failure, 100 mg three times daily, two years. Major adverse cardiovascular events 15% against 26%, hazard ratio 0.50 (95% CI 0.32 to 0.80, p = 0.003). Cardiovascular mortality 9% against 16% (p = 0.026) and all-cause mortality 10% against 18% (p = 0.018). The...
JACC Heart Failure, 2014 · 420 people
- 123 people1 study
123 elderly hip-fracture patients: gait speed improved and IGF-1 rose, but most functional measures did not, and the trial was terminated early over a congestive heart failure safety signal.
Archives of Gerontology and Geriatrics, 2011 · 123 people
- 110 people1 study
Russian clinical study of 110 post-stroke patients: Semax courses raised plasma BDNF and were associated with faster Barthel index improvement; not a blinded placebo-controlled trial.
Zhurnal Nevrologii i Psikhiatrii imeni S.S. Korsakova, 2018 · 110 people
How this is counted
Human studies only. Animal and test-tube work never enters this ranking, because the question you asked was about a person.
A compound is listed for an outcome only when a citation on its own page states that outcome and reports a result for it. Naming it is not enough: a paper that measures testosterone and reports a result about something else does not count as testosterone evidence, which is a mistake this tool made on its first run and no longer makes.
People counts are read out of the study lines themselves and are deliberate undercounts. Where a line states no number, the study still counts as a study and adds nothing to the headcount. Counts are never added across compounds, because the same person can appear in more than one trial.
Ranking is by people measured. That is a fact about the evidence, not a judgement about the compound, and it is the only ordering this site will defend. A compound at the top of a list is the most studied, which is not the same as the best.
Nothing here is advice, and nothing here is a recommendation to take anything.