What is Danshen?

Danshen, the dried root of Salvia miltiorrhiza, is the herb in three Hong Kong case reports published between 1995 and 1998 in which patients on long-term warfarin reached INR values from 5.5 to above 8.4. Two of those three patients bled (one was also using a methyl salicylate rub), which is why this issue spends more of its length on safety than on benefit.

Salvia miltiorrhiza Bunge is a perennial in the mint family (Lamiaceae). The medicinal part is the dried root and rhizome. Red sage is not culinary sage: Salvia officinalis is a different species whose leaves are used.

The Chinese Pharmacopoeia (2020 edition) lists the drug as Salviae Miltiorrhizae Radix et Rhizoma, gives a practitioner dose of 10 to 15 g, and carries a single caution: do not combine it with Veratrum (li lu). American retail sells it under dietary supplement law, mainly as root powder, 5:1 water extract granules and multi-herb "teapills."

Common Names

Danshen, dan shen, tan shen, tanjin (Japanese), red sage, Chinese red sage, salvia root, Radix Salviae Miltiorrhizae. On a Chinese medicine label "Salvia" with no species almost always means this root, not the kitchen herb.

Primary Active Compounds

Two chemical classes with almost opposite properties. The water-soluble phenolic acids: salvianolic acid B (also called lithospermic acid B, the most abundant), salvianolic acid A, rosmarinic acid, lithospermic acid, caffeic acid, and danshensu (salvianic acid A, 3,4-dihydroxyphenyllactic acid). The fat-soluble diterpene quinones called tanshinones: tanshinone IIA, cryptotanshinone, tanshinone I and dihydrotanshinone I, which give the root its red color. Sodium tanshinone IIA sulfonate is not a plant constituent; it is a semisynthetic, water-soluble salt made from tanshinone IIA for injection.

Key Note

Which class you swallow depends on how the root was processed. A water decoction or water extract granule delivers mostly phenolic acids and very little tanshinone. Finely milled whole root or an ethanol extract delivers far more tanshinone. The tanshinones are the compounds that switch drug-metabolizing enzymes on, so the form on the label changes the interaction profile, not just the potency.

Primary Functions & Benefits

Danshen is used in China for "blood stasis" conditions: angina and coronary disease above all, then ischemic stroke, menstrual pain and irregular periods.

Angina: the Compound Danshen Dripping Pill record

Compound Danshen Dripping Pill (CDDP) is a Chinese prescription drug, made from extracts of danshen and Panax notoginseng (sanqi) with borneol, formed into 27 mg pills taken 10 at a time, three times a day (810 mg a day). A 2023 meta-analysis in Frontiers in Cardiovascular Medicine included 29 Chinese trials (3,185 patients, published 1997 to 2013) comparing CDDP with nitrates (isosorbide dinitrate in most, isosorbide mononitrate in 7) for stable angina; in the 9 trials (1,116 patients) that measured symptom response at about 4 weeks, CDDP did better (odds ratio 1.95, 95% CI 1.25 to 3.05). The authors rated the certainty of that evidence very low to low by GRADE, noted that most trials had fewer than 100 participants and that selection and performance bias were unclear in every one, and four of the five authors worked for Tasly, the manufacturer.

The test meant to settle the question was the US program. After a US phase II trial (NCT00797953) of what the registry calls a "two-herb botanical product," the phase III trial, CAESA (NCT01659580), began in August 2012, mostly at US sites, randomized about 960 patients (the figure reported by BioCentury in August 2017) with chronic stable angina to placebo or T89, and used treadmill exercise duration at week 4 as its primary endpoint. By 2017, according to company and trade-press reports rather than a published paper, week 4 had been missed (p = 0.06) and week 6, a secondary endpoint, met (p = 0.02), and FDA had asked for a confirmatory trial before any application. That study (NCT03789552, ORESA in its European registration) tests 225 or 300 mg twice daily against placebo with exercise duration at day 57 as the primary endpoint; it began in August 2019, opened European sites in May 2025, and halted European recruitment in September 2025 to prepare a protocol amendment. I could find no T89 phase III result published in a peer-reviewed journal, and no FDA approval exists.

Myocardial infarction and stroke: mostly injections, poorly tested

The Cochrane review of danshen preparations for acute myocardial infarction (Wu, Ni and Wu, 2008) found 6 studies with 2,368 participants, of which only one was judged a genuine randomized trial; that trial showed no significant mortality difference (Peto odds ratio 0.55, 95% CI 0.23 to 1.32). A quasi-randomized trial reported lower mortality, and pooling the two gave a Peto odds ratio of 0.46 (95% CI 0.28 to 0.75), but the reviewers concluded that the evidence was too weak to make any judgment. The Cochrane review of danshen agents for acute ischemic stroke (Wu, Liu and Zhang, 2007) included 6 trials with 494 patients; none reported death or dependency at 3 months, only two reported adverse events, and although more treated patients showed improved neurological deficit at the end of treatment (Peto odds ratio 3.02), the reviewers judged all six of poor methodological quality. Both reviews covered Chinese hospital "danshen preparations," where injectables dominate.

Cholesterol, blood pressure and vascular markers: the single-herb test

The cleanest test of the root alone is the Nijmegen trial (van Poppel and colleagues, PLoS One, 2015): 20 analyzed adults aged 40 to 70 with high cholesterol and high blood pressure, randomized double-blind crossover, 28 days of a Taiwanese water extract granule (4 capsules three times daily, 3 g of extract a day, from 15 g of root) versus placebo. LDL cholesterol rose by 0.29 mmol/L (95% CI 0.02 to 0.55) relative to placebo. Twenty-four hour blood pressure, forearm endothelial function, C-reactive protein, oxidized LDL, glucose, insulin and thrombin generation did not change.

The positive vascular trials are on a two-herb formula. Hong Kong groups gave a danshen plus kudzu root (Pueraria lobata, gegen) capsule to 100 coronary patients for 24 weeks at 3 g a day (Tam and colleagues, 2009), to 90 high-risk hypertensive patients for 12 months at 1 or 2 g a day (Woo and colleagues, 2013), and to 165 postmenopausal women with borderline high cholesterol for 12 months (Kwok and colleagues, 2014). Each reported small improvements in carotid intima-media thickness or flow-mediated dilation. These are surrogate markers, and none of the three trials can tell you what danshen did on its own.

Menstrual pain

The one European regulatory foothold is a Dutch traditional herbal registration from 2016: a 90% ethanol dry extract "for alleviation of mild menstrual pain." A traditional registration rests on a history of use and safety, not on trial evidence of efficacy.

Forms & Standardization

Whole root, sliced or powdered

The pharmacopoeial crude drug. The 2020 Chinese Pharmacopoeia requires, on a dried basis, not less than 3.0% salvianolic acid B and not less than 0.25% for tanshinone IIA, cryptotanshinone and tanshinone I combined, plus water-soluble extractives of not less than 35.0% and ethanol-soluble extractives of not less than 15.0%. Hong Kong's HKCMMS monograph is stricter on the phenolic side: not less than 4.4% salvianolic acid B, 0.17% rosmarinic acid and 0.12% tanshinone IIA, with water-soluble extractives of at least 57.0%. These are minimums for raw material.

Water decoction and water extract granules

The traditional preparation. Hot water extracts salvianolic acid B, danshensu and rosmarinic acid well and tanshinones poorly. In the best human comparison (Xing and colleagues, Scientific Reports, 2017), 24 healthy volunteers took 20 g of the same root batch either as a classic two-stage decoction or as a micronized whole root powder stirred into water. After dosing, peak plasma tanshinone IIA was 1.75 ng/mL after the decoction and 25.76 ng/mL after the powder, and the relative bioavailability of tanshinone IIA, cryptotanshinone and tanshinone I rose 45.9, 43.6 and 123.7 fold with the powder. Tanshinone I was undetectable in 11 of the 24 decoction drinkers.

The Nijmegen trial's granule (Kaiser Pharmaceutical, Taiwan) was 50% water extract and 50% starch at a stated 5:1 ratio; TCMzone's TD025B is labeled 5:1; Mayway's 5730C lists maltodextrin and magnesium stearate and no ratio. A 5:1 granule is a water extract: expect phenolic acids, not tanshinones.

Ratio claims do not guarantee marker content. Zhou and colleagues (Journal of Traditional Chinese Medical Sciences, 2022) compared 12 anonymized granule products with 6 raw root samples and reported significantly lower salvianolic acid B, sodium danshensu and cryptotanshinone in the granules; the authors stated that none of the granules met the pharmacopoeial salvianolic acid B or tanshinone IIA minimum, though none of the raw samples met the tanshinone IIA minimum either and only some met the salvianolic acid B one. That applies a raw-root standard to an extract with excipients, and the paper's table conflicts with its text in places, so treat it as a signal.

Ethanol extracts

Alcohol pulls tanshinones. In two of the human enzyme studies by Qiu and colleagues the product was a danshen ethanol extract (1 g three times daily in the fexofenadine study), and it measurably induced drug-metabolizing enzymes and transporters (see Interactions). If a capsule says "ethanol extract" or "standardized to tanshinone IIA," it belongs in the higher-interaction category.

Multi-herb teapills

Mayway's Plum Flower "Salvia Teapills" (Dan Shen Yin Wan, MW Code 3368) are not single-herb danshen. The ingredient list is Salvia miltiorrhiza root, Santalum album wood (sandalwood), Amomum longiligulare fruit and Glycyrrhiza uralensis root (licorice), with no amounts per herb shown. The product page states "Contraindicated during pregnancy."

Compound Danshen Dripping Pill and T89

A Chinese prescription drug, not a supplement: danshen and notoginseng extracts with borneol, 27 mg per pill. Borneol is pharmacologically active in its own right and is reported to raise absorption of the pill's other components. Any result from CDDP is a result for that combination.

Injectables

Danshen injection, Compound Danshen injection (danshen with Dalbergia odorifera), Danhong injection (danshen with safflower), salvianolate (about 85% salvianolic acid B) and sodium tanshinone IIA sulfonate are Chinese hospital drugs, not supplements.

Food Sources

Danshen is not a food and is not eaten in any cuisine. The exception is rosmarinic acid, which also occurs in culinary mint-family herbs such as rosemary, but nobody takes danshen for its rosmarinic acid.

Exposure comes through six routes: practitioner decoctions; water extract granules; capsules of root powder or extract; multi-herb teapills such as Dan Shen Yin Wan; Chinese patent drugs such as Compound Danshen Dripping Pill; and, in Chinese hospitals only, injections. The one food-category precedent is Dutch: the European Medicines Agency's 2022 assessment records that danshen root tea has been sold in the Netherlands as a food supplement since 1991, labeled for 9 to 15 g of root a day.

Who Should Take Danshen

No group has been shown in a published trial to benefit from a danshen supplement of the kind sold at retail in the US.

A person with angina considering danshen in place of nitrates, a beta blocker or an antiplatelet drug has no evidence to stand on, and is very likely already taking aspirin, clopidogrel, ticagrelor or an anticoagulant, the population in which danshen is most dangerous. A person with high cholesterol has one single-herb placebo trial, in which LDL went up by 0.29 mmol/L (about 11 mg/dL). A woman with menstrual pain has a Dutch traditional registration and no trial.

The realistic candidate is an adult under a licensed Chinese medicine practitioner's care who takes no anticoagulant, antiplatelet, digoxin or narrow-margin drug, is not pregnant or trying to conceive, and has no surgery scheduled. Anyone else should treat danshen as a drug with an interaction record, not a heart tonic.

Who Should AVOID or Use Caution

Contraindications

• Anyone taking warfarin or another vitamin K antagonist (acenocoumarol, phenprocoumon). Published cases of INR rising to above 5.5, 5.6 and above 8.4, two with bleeding, and the European Medicines Agency's 2022 assessment states that danshen should be avoided in patients taking warfarin.

• Anyone taking a direct oral anticoagulant (apixaban, rivaroxaban, edoxaban, dabigatran) or heparin. No interaction studies exist, which means no reassurance, and danshen's antiplatelet action (see How Danshen Works) makes an additive bleeding effect expected rather than speculative.

• Anyone taking clopidogrel, prasugrel, ticagrelor or dual antiplatelet therapy after a stent. Danshen pushes clopidogrel both ways: its phenolic acids add antiplatelet effect, while 7 days of danshen capsules in 20 healthy men roughly halved clopidogrel exposure and weakened its antiplatelet activity (see Interactions). After a stent, an unpredictable change in platelet inhibition in either direction can be fatal.

• Pregnancy. There are no reproductive toxicity studies on the root. The European Medicines Agency concluded in 2022 that use in pregnancy and lactation "cannot be recommended." In pregnant mice, sodium tanshinone IIA sulfonate raised uterine artery resistance and lowered offspring body weight (Morton and colleagues, PLoS One, 2015), and in zebrafish embryos tanshinone IIA caused malformations at about 1 micromolar (Wang and colleagues, Molecules, 2017). Mayway's product page for Plum Flower Salvia Teapills states "Contraindicated during pregnancy."

• Breastfeeding, for the same absence of data.

• Known bleeding disorders (hemophilia, von Willebrand disease, severe thrombocytopenia) and active bleeding, including peptic ulcer.

• Scheduled surgery, dental extraction or other invasive procedure within the next two weeks (see Special Considerations).

Use Caution

• Digoxin users. Danshen interferes with some digoxin blood tests, making the measured level falsely high on certain assays and falsely low on others (see Interactions). A wrong digoxin level can lead to a wrong dose.

• Anyone taking a drug with a narrow margin that is cleared by CYP3A4 or carried by P-glycoprotein: tacrolimus, cyclosporine, sirolimus, some anticancer kinase inhibitors, oral midazolam, and statins that depend on CYP3A4. Ethanol extracts of danshen changed midazolam clearance roughly sixfold in human volunteers after 10 days.

• Low blood pressure or a tendency to faint. The European Medicines Agency's adverse event summary lists hypotension and dizziness.

• Children. No pediatric data exist for any oral danshen product.

• Older adults on several cardiovascular drugs, who are the population most likely to be taking an anticoagulant, an antiplatelet drug and digoxin at the same time.

Critical Safety Point

The controlled human studies of danshen with warfarin point in different directions. A 40-patient controlled Chinese trial, summarized in a 2015 Singapore Medical Journal review (Chua and colleagues; the primary report could not be traced for this issue), reported that two weeks of danshen with warfarin significantly prolonged prothrombin time and INR compared with controls, and in 14 Hong Kong volunteers a danshen and kudzu product lowered exposure to warfarin's 7-hydroxy metabolite (Zhang and colleagues, 2020). Studies of the standardized prescription pill also disagree: in a 12-volunteer Chinese crossover, Compound Danshen Dripping Pill significantly changed prothrombin time (Yi and colleagues, Chinese Journal of Pharmacovigilance, 2013), while two others did not reproduce the case reports. In the 23 of 24 healthy volunteers who reached a stable INR on warfarin, 7 days of added T89 did not raise INR (Wang and colleagues, Journal of Clinical Pharmacology, 2013), and in 59 Tianjin patients with coronary disease and atrial fibrillation, adding Compound Danshen Dripping Pill for at least 4 weeks did not change mean INR (2.42 before, 2.40 after) or warfarin dose (Lv and colleagues, Frontiers in Pharmacology, 2017). Those results apply to one standardized prescription pill, given under INR monitoring, to small groups. The Hong Kong cases involved danshen of unreported dose, taken as a home decoction or tea where the form was described; in the 1998 case a 62-year-old man with a replaced mitral valve bled into his chest after two weeks of daily danshen. A supplement buyer cannot know which of these two worlds their bottle belongs to. If you take warfarin, do not take danshen. If you already have, tell the clinic that manages your INR and get an INR checked within days, not at your next routine visit.

Recommended Dosages

No trial has shown a danshen supplement working at any particular amount, so no supplement dose has been established. What follows describes what labels print and what those amounts contain. Two figures a reader may go looking for, the Chinese Pharmacopoeia practitioner quantity and the prescription dose of Compound Danshen Dripping Pill, sit in Special Considerations with their context attached, because neither is a supplement dose.

Water extract granules

Labeled 5:1 and directed at 1 to 2 g a day in two or three servings (TCMzone, TD025B), the extract of 5 to 10 g of root at a true 5:1 ratio. The one placebo-controlled single-herb granule trial used 3 g of extract a day for 28 days and found no cardiovascular benefit.

Root powder

Mayway's unsulfured danshen powder lists a "standard dosage" of 6 to 15 g. That is the raw root amount used in decoction, printed on a product that is not decocted. Micronized root powder delivered 15 to 46 times more tanshinone IIA, by peak level and overall exposure, than a decoction of the same weight in the Xing volunteer study, so a gram of swallowed powder is not equivalent to a gram boiled in water.

Multi-herb teapills

Plum Flower Salvia Teapills are directed at 8 pills three times a day, with no per-herb amount disclosed. The danshen dose cannot be calculated from the label.

Ethanol extract capsules

The human enzyme studies of ethanol extracts found a marked change in drug clearance within 10 days (see Interactions). No efficacy trial supports any ethanol extract dose.

Duration

The Nijmegen single-herb placebo trial ran 28 days, and a Chinese "hydrophilic extract" trial ran 60 days. The two-herb danshen and kudzu formula has been given for up to 12 months in Hong Kong trials without serious adverse events attributed to it. Nothing in the record supports open-ended daily use of the single root.

Timing & Administration

Timing matters here for one reason that is specific to danshen: the direction of its effect on drug metabolism flips over the first one to two weeks. In 12 healthy men, a single dose of danshen ethanol extract raised peak midazolam levels by 87%, consistent with inhibition of CYP3A in the gut, while 10 days of the same extract cut midazolam exposure (AUC) by 79.9% and raised its clearance by 501.6%, consistent with induction (Qiu and colleagues, Evidence-Based Complementary and Alternative Medicine, 2013). Someone on a CYP3A4 drug who starts danshen may see that drug's levels rise for a few days and then fall below their usual level, and stopping danshen after weeks of use may push levels back up as the induction wears off. Spacing danshen and a medicine several hours apart does not prevent either effect, because both act on enzyme activity rather than on contact in the stomach.

For anyone on warfarin the onset in the published cases ranged from 3 days to 4 weeks after starting danshen, so there is no safe interval of "just trying it."

There are no human data on morning versus evening dosing or on taking it with food.

Timeline of Effects

Every time point here describes what a named study measured; none is a schedule for taking a supplement.

Absorption

After a 20 g decoction, plasma tanshinone IIA peaked at about 0.5 hours with a half-life of about 4.5 hours; after the same weight of micronized root, it peaked at about 0.9 hours and the half-life stretched to about 14.8 hours (Xing and colleagues, 2017). The phenolic acids are poorly absorbed: in rats given purified salvianolic acid B (lithospermic acid B) by mouth, absolute bioavailability was 5% at 50 mg/kg and the compound was undetectable in plasma at 10 mg/kg (Kim and colleagues, Journal of Toxicology and Environmental Health, 2005).

Effects on drug metabolism

Inhibition of CYP3A after a single dose; induction of CYP3A and intestinal P-glycoprotein measurable after 10 to 14 days (Qiu and colleagues, 2010, 2013 and 2014). None of these studies measured how long the induction lasted after danshen was stopped.

Effects on warfarin in the case reports

About 3 days (Tam and colleagues, 1995, in a patient also using a methyl salicylate oil for 9 days), 2 weeks (Izzat and colleagues, 1998) and about 4 weeks (Yu and colleagues, 1997) before a markedly raised INR was found.

When to conclude it is not working

No single-herb trial shows benefit at any time point for angina, cholesterol or circulation, so no duration makes a lack of effect informative. A cholesterol panel after a month is the one objective self-check, and in the only trial that ran it the result went the wrong way.

Benefits of Taking Danshen

The benefit case for the oral root rests on three things. First, its phenolic acids have measurable antiplatelet and antioxidant activity in human cells and, in salvianolate's case, in patients receiving an injected purified fraction. Second, a large body of Chinese trials, all of low or very low certainty, reports symptom and ECG improvement in angina with Compound Danshen Dripping Pill. Third, three Hong Kong trials of a danshen and kudzu capsule, totaling 355 patients, report small improvements in carotid wall thickness and artery dilation over 6 to 12 months. None of those three supports a single-herb supplement directly, and the one single-herb placebo trial in Western patients found no benefit on any of 10 or more cardiovascular measures.

If danshen does anything useful by mouth, the evidence points to it doing so in a formula, prescribed by a practitioner, for a patient not taking blood thinners.

Potential Negatives & Side Effects

Bleeding and over-anticoagulation

The serious adverse effect of oral danshen. In the three Hong Kong warfarin cases the INR rose to above 5.5 (a 66-year-old man admitted with bleeding from a gastric cancer, who had also been applying a 15% methyl salicylate medicated oil, itself a warfarin potentiator, for 9 days), 5.6 (a 48-year-old woman with rheumatic heart disease; Yu and colleagues, 1997) and above 8.4 (the 62-year-old man with bleeding into the chest). A 1998 Hong Kong audit of 35 warfarin patients admitted with INR above 3.0 (Chan, International Journal of Clinical Pharmacology and Therapeutics) attributed 7 to interactions, one of them danshen plus a 15% methyl salicylate medicated oil, apparently the same 1994 to 1995 case. A 2014 review by Ge, Zhang and Zuo at the Chinese University of Hong Kong graded danshen's interaction with warfarin "highly probable," the strongest of four levels, alongside cranberry, soy and St John's wort.

Common, milder effects

In the Nijmegen trial, people taking 3 g a day of water extract reported headache (5), dizziness (3), change in stool frequency (3) and flatulence (2). One participant stopped because of peripheral facial nerve paralysis (Bell's palsy) during the danshen period, classified as a serious adverse event; a single case in a 23-person trial cannot establish cause. The European Medicines Agency's 2022 summary of reported events adds hypotension, abdominal discomfort, decreased appetite and allergic reactions.

A rise in LDL cholesterol

The Nijmegen trial's primary result: LDL 0.29 mmol/L higher than placebo after 4 weeks of water extract. The authors compared it with the cholesterol-raising effect of unfiltered boiled coffee and suggested that something in the water preparation might be responsible. It has not been replicated or refuted.

Injectable preparations, which are a different risk

Guo and colleagues (BioMed Research International, 2015) analyzed 316 Chinese reports from 1980 to 2013 describing 350 episodes of anaphylactic shock and 10 deaths across 17 herbal injections for heart and brain disease, including danshen injection, Compound Danshen injection, Danhong injection and sodium tanshinone IIA sulfonate; 346 of the 350 followed intravenous use. They cite China's national adverse drug reaction center as attributing 53 anaphylactic shock cases to Compound Danshen injection, and record that on March 24, 2009, China's Ministry of Health and drug regulator ordered one manufacturer's Compound Danshen injection (Taizhou Tianrui Pharmaceutical) withdrawn from use, sale and production. In a meta-analysis of 17 randomized trials of sodium tanshinone IIA sulfonate added to standard care for unstable angina, adverse reactions were more frequent with the injection than without it (risk ratio 3.57, 95% CI 1.28 to 9.94; Tan and colleagues, Chinese Journal of Integrative Medicine, 2018). The European Medicines Agency also records one case of liver damage and two of shock after injected preparations. None of this is oral-root data, and equally, reassuring safety language from injectable trials says nothing about a capsule taken with warfarin.

Deficiency Symptoms

What danshen addresses

Nothing that can run low. The body has no requirement for tanshinones or salvianolic acids, no tissue store of them, and no physiological process that fails in their absence. Danshen is taken for symptoms and risk factors (chest pain, cholesterol, menstrual pain), and in each case the target is a disease process, not a shortfall.

Bottom line

No deficiency state exists and no blood test measures danshen status. With danshen the only meaningful measurement is the INR of someone on warfarin, and the right value of danshen intake for that person is zero.

Toxicity Symptoms

At high intake

Acute toxicity of the water-soluble fraction is low in animals: the oral LD50 of a water-soluble extract in mice is reported at 25.8 g per kilogram, and 2.5 g per kilogram a day for 90 days was reported non-toxic in rats (both cited in Xing and colleagues, 2017, from earlier Chinese work). Those numbers describe one extract type in rodents, not the tanshinone-rich forms, and they are not a margin of safety for a human on anticoagulants. The documented human harm at ordinary intake is excessive anticoagulation and bleeding in warfarin users; I found no published case of poisoning from oral danshen taken on its own.

Signs to reduce or stop

Stop immediately and contact a clinician for: nosebleeds or bleeding gums that are new or slow to stop, unexplained bruises, red or brown urine, black or tarry stools, vomiting blood or material like coffee grounds, unusually heavy periods, a sudden severe headache, or new shortness of breath with chest pain. For anyone on warfarin, an INR above the target range after starting danshen is itself the stop signal. Lightheadedness on standing, palpitations or fainting fit the reported hypotension and are also a reason to stop.

General note

The European Medicines Agency's 2022 assessment found no carcinogenicity data, one negative Ames mutagenicity test on an unspecified extract, and no reproductive toxicity studies, alongside a zebrafish study in which tanshinone IIA caused spinal, tail and heart-sac malformations at about 1 micromolar. That gap is why pregnancy appears under Contraindications rather than Use Caution.

How Danshen Works

Phenolic acids and platelets

Salvianolic acid B, the dominant water-soluble compound, inhibits human platelet activation in the laboratory by two routes: it antagonizes the P2Y12 receptor, the same receptor clopidogrel, prasugrel and ticagrelor block, and it inhibits phosphodiesterase, which raises the platelet's internal cAMP and damps activation (Liu and colleagues, Thrombosis Research, 2014). Salvianolate, an injected fraction that is about 85% salvianolic acid B, measurably added to aspirin plus clopidogrel in acute coronary syndrome patients. Danshensu adds a cyclooxygenase component: in a 14-person crossover in healthy volunteers, a danshen and kudzu product taken with aspirin lowered systemic thromboxane B2, which the authors linked partly to higher aspirin exposure and partly to danshensu's own COX-2 inhibition (Zhang and colleagues, Journal of Ethnopharmacology, 2020). Antiplatelet action plus an anticoagulant is the pharmacodynamic half of the warfarin problem.

How danshen raises warfarin's effect: the pharmacokinetic half

Rat work in the 1990s showed danshen changing warfarin's handling directly. Chan, Lo, Yeung and Woo (Journal of Pharmacy and Pharmacology, 1995) gave rats oral danshen extract at 5 g per kilogram twice daily for 3 days and found higher plasma levels of both R- and S-warfarin over 24 hours, higher AUC and peak concentration, longer half-life, lower clearance and a longer prothrombin time, both after a single warfarin dose and at warfarin steady state. A second mechanism is displacement from albumin. Warfarin is about 99% protein-bound, and only the unbound fraction acts; binding studies summarized in a 2017 review of danshen drug interactions found several danshen compounds binding the same albumin site as warfarin (site I) and lowering warfarin's binding constant 1.4- to 8.7-fold, which would raise free warfarin even if total levels did not change. Not every rat study agrees: at 240 and 480 mg per kilogram, Ge and colleagues (Chinese Medicine, 2016) saw no change in liver CYP or vitamin K epoxide reductase with danshen, while kudzu in the same study accelerated warfarin metabolism.

CYP450: induction, inhibition and why the form matters

An inhibitor slows clearance and raises drug levels within hours; an inducer makes cells produce more enzyme, lowering drug levels, but only after days of exposure. Danshen does both, through different tanshinones. In cell work by Qiu and colleagues (2013), dihydrotanshinone I inhibited CYP3A while tanshinone IIA and cryptotanshinone induced it, and the same two inducers raised expression of the P-glycoprotein transporter gene (2014). That is the mechanism behind the single-dose versus 10-day flip described under Timing.

The phenolic acids do not appear to drive this. Kuo and colleagues (Journal of Pharmacy and Pharmacology, 2006) fed mice either an ethyl acetate extract (tanshinone-rich) or a water extract: the ethyl acetate extract raised CYP1A activity 8-fold, CYP2C (tolbutamide hydroxylation) 2-fold, CYP3A (nifedipine oxidation) 3-fold and warfarin 7-hydroxylation 3-fold, and the water extract changed none of them. Human data line up by enzyme: CYP3A is clearly affected (midazolam, and clopidogrel, which needs CYP activation); CYP1A2 apparently is not, since 14 days of danshen tablets left theophylline pharmacokinetics unchanged in 12 volunteers (Qiu and colleagues, British Journal of Clinical Pharmacology, 2007). A water decoction or 5:1 granule is the low-tanshinone, low-induction form; micronized root powder, ethanol extracts and "tanshinone-standardized" capsules are the high-induction forms.

Coronary flow and microcirculation

The microcirculation case is built mainly on Compound Danshen Dripping Pill. A 2018 review by Tasly scientists (Liao and colleagues, Biomedicine and Pharmacotherapy) attributes its anti-anginal action to improved coronary microcirculation, a shift in myocardial energy metabolism and platelet inhibition, drawing on animal and cell studies. For the root alone, the Nijmegen trial's direct measure of vascular function, forearm blood flow in response to acetylcholine, did not change.

Synergistic Supplements

Kudzu root (Pueraria lobata, gegen)

The pairing with the most human data: the three Hong Kong vascular trials, at 1 to 3 g a day for 6 to 12 months. It also interacts with antithrombotic drugs in two directions (raised aspirin exposure in 14 volunteers; faster warfarin metabolism from kudzu in rats), so it is not a combination to assemble from two bottles while on any blood thinner.

Panax notoginseng (sanqi) and borneol

The partners in Compound Danshen Dripping Pill. No US supplement replicates the drug's composition or quality control, and buying the three separately does not recreate it.

What not to stack

Anything else with antiplatelet or anticoagulant reports: ginkgo, garlic supplements, dong quai, high-dose fish oil, high-dose vitamin E, nattokinase, and turmeric or curcumin extracts. Fugh-Berman's 2000 review in The Lancet named ginkgo, garlic, dong quai and danshen together as plausible causes of bleeding in warfarin users.

Interactions & What NOT to Take

Warfarin and other vitamin K antagonists

Warfarin (Coumadin, Jantoven), acenocoumarol, phenprocoumon. Do not combine. A 2001 review in Annals of Pharmacotherapy by Chan concluded that danshen "should be avoided in patients taking warfarin," and the European Medicines Agency reached the same conclusion in 2022. A 2024 analysis of 737 hospitalized Chinese patients with type 2 diabetes found danshen with warfarin to be the most frequent herb-drug pair carrying an "avoid combination" rating (Ren and colleagues, Frontiers in Endocrinology), so this is not a rare combination in practice.

Direct oral anticoagulants and heparins

Apixaban (Eliquis), rivaroxaban (Xarelto), edoxaban, dabigatran (Pradaxa), enoxaparin. No interaction studies exist. Danshen's antiplatelet action adds bleeding risk regardless of anticoagulant type, and its CYP3A and P-glycoprotein induction could in principle lower apixaban and rivaroxaban levels, which would raise stroke risk in atrial fibrillation. Avoid.

Antiplatelet drugs

Clopidogrel (Plavix): 7 days of danshen capsules in 20 men halved clopidogrel exposure and weakened its antiplatelet effect. Aspirin: a danshen and kudzu product raised aspirin exposure and lowered thromboxane B2 in 14 volunteers. Prasugrel, ticagrelor, cilostazol, dipyridamole and NSAIDs (ibuprofen, naproxen, diclofenac): additive bleeding risk. Avoid.

CYP3A4 and P-glycoprotein substrates

Ethanol extract and tablet forms of danshen reduced midazolam exposure (90% CI for AUC ratio 0.592 to 0.921 after 14 days of tablets, Qiu and colleagues, 2010) and fexofenadine exposure (AUC down 37.2%, clearance up 104.9% after 10 days, Qiu and colleagues, 2014). Drugs where a change of that size matters: tacrolimus, cyclosporine, sirolimus, everolimus, ticagrelor, apixaban, rivaroxaban, simvastatin, atorvastatin, lovastatin, oral midazolam and triazolam, HIV protease inhibitors, and oral cancer kinase inhibitors. Water extracts appear to carry this risk far less, but labels rarely name the solvent.

Statins

Beyond the CYP3A4 statins above, a 6-person Japanese crossover found a tanjin product lowered rosuvastatin peak level by 26.85% and AUC by 19.43% (Yang and colleagues, Yonago Acta Medica, 2019). Small, but the direction is reduced statin exposure.

Digoxin: a laboratory interaction

Danshen can change the number on the report without changing digoxin's effect. Some danshen components cross-react with certain digoxin antibodies. With the Abbott fluorescence polarization immunoassay, danshen extract falsely raised measured digoxin; with Abbott's microparticle enzyme immunoassay it falsely lowered it (Wahed and Dasgupta, American Journal of Clinical Pathology, 2001). Because the interfering compounds are strongly protein-bound while only about 25% of digoxin is, measuring free digoxin removed the interference. Chemiluminescent and several newer immunoassays were largely free of interference, and different danshen brands produced widely different degrees of interference (Datta and Dasgupta, Therapeutic Drug Monitoring, 2002). Anyone on digoxin who takes danshen should tell the lab and the prescriber.

Antihypertensives

The European Medicines Agency lists antihypertensives among possible interactions, consistent with reported hypotension. Watch for dizziness on standing.

What does not seem to interact

Theophylline: 14 days of danshen tablets did not change its pharmacokinetics in 12 volunteers, which suggests CYP1A2 drugs (theophylline, clozapine, tizanidine) are lower priority.

Quality, Testing & Adulteration

The best market survey

Kum and colleagues at University College London (Frontiers in Pharmacology, 2021) analyzed 71 samples: authenticated roots from China, Taiwan, Germany and the US, 15 from a Hanoi herb market, 17 from Chinese online stores, and concentrated extracts. Six of 60 samples tested for metals exceeded the Chinese or US Pharmacopoeia cadmium limit (0.3 mg/kg in the 2015 Chinese edition), four raw roots and two concentrated extracts. Lead stayed under 1.5 mg/kg in every sample against a 5 mg/kg limit, the highest arsenic was 1.33 mg/kg in a concentrated extract, and one online sample carried 19 mg/kg copper against a 20 mg/kg limit. Chemically, the online-store samples were the most variable, and one (C11) had high caffeic acid and low salvianolic acid B and rosmarinic acid, matching the authors' authenticated reference samples of a substitute Salvia species rather than S. miltiorrhiza.

Substitution

At least 19 other Salvia species have been used locally as danshen, and the common substitutes contain salvianolic acids and tanshinones in different proportions, so a crude color or marker test may not catch them. In China's 2015 national drug sampling, 39.7% of danshen crude drug samples failed on appearance, extractives, water or marker content, one of the highest failure rates of any Chinese medicinal material, as cited by Kum and colleagues.

Pesticides and sulfur

Zhang and colleagues (Journal of Chromatography B, 2017) screened 32 batches for 12 organophosphorus pesticides and found dimethoate in 4 (0.016 to 0.02 mg/kg), dichlorvos and omethoate together in one Sichuan sample (0.004 and 0.027 mg/kg), and malathion and monocrotophos in two others (0.014 and 0.028 mg/kg). Those are low concentrations, but they show the crop is sprayed. Mayway sells a danshen powder labeled "unsulfured," which tells you sulfur-fumigated root exists in the supply chain; the label claim is the only way a buyer can tell.

Third-party testing

I could not find a USP Verified, NSF or ConsumerLab certification for any danshen product, and the practitioner product pages reviewed for this issue (TCMzone TD025B, Mayway 5730C) published no contaminant results.

What to ask the seller, and what the label alone can settle

Ask for the certificate of analysis for your lot before buying. A useful one names Salvia miltiorrhiza Bunge and the root and rhizome as the part, gives an identity method (HPTLC or DNA), states salvianolic acid B and the three main tanshinones as measured percentages with the method and lab, and reports cadmium, lead, arsenic and mercury in mg/kg against a named limit, plus organophosphorus pesticide and sulfur dioxide results. If a granule claims 5:1, ask for its salvianolic acid B figure and compare it with the raw root minimum of 3%; in the 2022 granule comparison, granules ran below raw root. On your own, check that the Supplement Facts panel names the species rather than just "Salvia" or "red sage"; that a formula product discloses how much of it is danshen; and whether the extract solvent is stated, since water versus ethanol decides the tanshinone and enzyme-induction profile. Real root powder is reddish brown and tastes slightly bitter and astringent, per the Hong Kong monograph.

Special Considerations

Practitioner and prescription quantities, in context

Chinese Pharmacopoeia (2020): 10 to 15 g of crude root per day in decoction, for practitioner prescription. Dutch food-supplement tea: 9 to 15 g a day. Dutch registered traditional product: 975 mg of 90% ethanol dry extract twice daily for mild menstrual pain. Compound Danshen Dripping Pill: 10 pills of 27 mg three times daily as a Chinese prescription drug. T89 in US trials: 250 to 375 mg a day (phase II) and 225 or 300 mg twice daily (confirmatory phase III). None of these is a supplement recommendation, and the prescription figures apply to products that are not supplements.

Surgery and dental work

Stop danshen at least two weeks before scheduled surgery, the interval the American Society of Anesthesiologists' patient guidance mentions for supplements, and tell the surgical team you took it, especially if anticoagulation or antiplatelet therapy is planned afterward.

Regulatory status

US: sold under DSHEA as a dietary ingredient without FDA review; T89 has investigational status only. European Union: on March 30, 2022 the Committee on Herbal Medicinal Products adopted a public statement (EMA/HMPC/509931/2019) that an EU monograph could not be established because neither well-established use nor 15 years of traditional use in the EU was demonstrated; 17 member states reported no danshen preparations on the market, and the Netherlands reported one registered traditional product. China: crude drug, patent drugs and injections are all regulated medicines.

Research Status & Evidence Quality

Tier 1, the interaction evidence (strongest)

Three Hong Kong warfarin cases with dechallenge, rat pharmacokinetic and albumin-binding data, and controlled human pharmacokinetic studies (12 to 24 volunteers each) showing CYP3A inhibition then induction, P-glycoprotein induction and halved clopidogrel exposure. Two small controlled studies of T89 and CDDP with warfarin found no INR change, a 12-volunteer CDDP study reported changed prothrombin time, and a 40-patient Chinese controlled trial reported prolonged prothrombin time and INR with danshen, so the concern stands.

Tier 2, the Compound Danshen Dripping Pill record

Dozens of Chinese randomized trials rated very low to low certainty, and one US phase III trial with a missed primary endpoint. Evidence about a three-ingredient prescription drug.

Tier 3, injectables

Cochrane reviews for myocardial infarction (6 studies, 2,368 participants, one genuine RCT) and ischemic stroke (6 trials, 494 patients, all poor quality) that could not reach conclusions, plus a substantial anaphylaxis and adverse reaction record.

Tier 4, the oral single root

One Western placebo-controlled crossover (20 people, 4 weeks) with no benefit and an LDL increase; a Chinese trial of a "hydrophilic extract" in 62 diabetic patients with coronary disease (Qian and colleagues, Phytotherapy Research, 2011 and 2012; PubMed indexes the product with notoginseng and borneol, so it may not be single-herb) reporting lower oxidative and endothelial markers, with only 24 treated and 30 placebo patients followed to completion; and three trials of a danshen and kudzu formula showing small surrogate improvements.

Research Limitations

The interaction case reports are three, all from Hong Kong in the 1990s, with unreported doses, and in the 1995 case the patient was also using a methyl salicylate oil, a second agent that potentiates warfarin, and bled from a gastric cancer. The negative warfarin studies used a single prescription product in small, monitored groups. No study has measured interactions with direct oral anticoagulants. Almost no trial reports the tanshinone and phenolic acid content of what was given. There are no reproductive, pediatric or carcinogenicity data for the oral root, and no published US retail product survey.

Summary & Key Takeaways

Bottom Line

Danshen is a red root whose reputation as a heart herb was built mainly on a three-ingredient Chinese prescription pill and on hospital injections, while its best-documented effect by mouth is making blood thinners stronger and less predictable. How the root is extracted decides whether it delivers mostly phenolic acids or tanshinones, and the tanshinone-rich forms are the ones that alter drug metabolism.

Key Safety Points

• Do not take danshen with warfarin; published cases reached INRs of 5.5 to above 8.4, two with bleeding.

• Do not take it with apixaban, rivaroxaban, dabigatran, heparins, clopidogrel, prasugrel, ticagrelor or after a coronary stent; there is either interaction data showing harm or no data at all.

• Avoid it in pregnancy and while breastfeeding, and stop it at least two weeks before surgery.

• If you take digoxin, tacrolimus, cyclosporine or another narrow-margin drug, do not start or stop danshen without your prescriber and appropriate blood tests.

• Stop at once for unusual bleeding, black stools, blood in the urine, heavy bruising or fainting.

Special Note

Stopping danshen is also an event for anyone on warfarin. In the published cases, withdrawing danshen and resuming warfarin brought the INR back into range, which means a person who has been taking both and quietly stops the herb can see their INR fall. And because the tanshinone-rich forms induce CYP3A over 10 to 14 days, stopping them can let levels of drugs such as tacrolimus or cyclosporine drift upward afterward. Tell the prescriber before you start, and again before you stop.

Researched and drafted with AI assistance. Every claim verified against primary sources and human-reviewed before publication. Last reviewed: October 19, 2026. Corrections: reply to any issue.

This issue is educational and is not medical advice. If you take a blood thinner, an antiplatelet drug, digoxin or any other prescription medicine, have a heart or bleeding condition, are pregnant or breastfeeding, or have surgery coming up, talk to a clinician before taking danshen or changing anything.