Accessible Search Form           Advanced Search

News & Resources

E-mail updates icon Email Alerts

Submit your email address below to receive email alerts about fact sheets, stories of success, and other NHLBI research-related articles. Learn more.

Questions and Answers on NHLBI’s Clarification of Optimal Anticoagulation through Genetics (COAG) study


Results from the COAG study were published in the New England Journal of Medicine on November 19, 2013, to coincide with a presentation of the study findings at the American Heart Association (AHA) Scientific Sessions in Dallas.

The NIH-funded clinical trial found that combining genetic data with clinical information to determine the initial dosage of the blood thinner warfarin, used to prevent blood clots in the circulatory system, was no more effective in achieving stable anticoagulation than using only clinical information. In addition, the study found that in African-Americans, anticoagulation control was worse in the genetics-based approach compared to the clinically based method.


What is COAG?

COAG was a large-scale, multicenter, double-blind, randomized clinical study designed to determine the benefit of pharmacogenetics in determining the proper dose of the blood thinner warfarin. Pharmacogenetic dosing involves applying a person’s genetic information alongside other factors when determining how much medication a person needs.

COAG compared two strategies for warfarin dosing. The first was a clinically based dosing formula that looked at clinical measures such as age, race, body size, smoking status, and use of certain cardiovascular medications. The second was a pharmacogenetics-based approach that incorporated information on two specific genes along with the clinical measures outlined above. Patients enrolled in the study were randomly (e.g., by chance) assigned to one of the two dosing strategies outlined above.

What is warfarin?

Warfarin is a drug classified as an anticoagulant, or blood thinner. Warfarin is prescribed to help prevent the formation of potentially dangerous blood clots in the body. Warfarin works better in areas where blood flows slowly, such as veins, and is typically prescribed for cardiovascular problems in these areas. Common conditions calling for warfarin use include deep vein thrombosis, pulmonary embolism, and atrial fibrillation. Warfarin has been approved for anticoagulation for decades and can be taken as a pill, which has made it a popular medication; it is the most commonly prescribed anticoagulant in the United States.

While warfarin is an effective drug for anticoagulation, it does pose some serious health risks if the dose is not administered correctly. If the dose is too low, the patient will have a high risk of clotting events; however, if the dose is too high, then the blood may not clot at all and the patient is at risk for serious bleeding. Proper dosing of warfarin is complicated, though, because the drug interacts with many other common medications as well as some foods, so the active dose can vary greatly among individuals. When patients start warfarin, they typically are closely monitored via blood tests during the first few days to see how well the warfarin is working; the dose is then adjusted up or down as needed. Continuous blood monitoring is required throughout the duration of the treatment (usually every month).

Why did NHLBI decide to fund this study?

In order to reach the proper warfarin dose more quickly, physicians have developed some formulas to try and predict an initial dose as opposed to using a standard starting level like 5 or 10 mg. These formulas typically consist of clinical measures such as age, race, body size, smoking status, and use of certain medications known to interact with warfarin.

In addition to food and drugs, recent research has shown that people’s genetic makeup may influence how they process and metabolize warfarin. In light of this, some dosing formulas have incorporated genetic information to try and make them more predictive. The Food and Drug Administration recently updated the labels for warfarin to note that gene information may be taken into consideration when prescribing an initial warfarin dose. However, the available data does not provide definitive evidence as to whether a gene-guided, or pharmacogenetics, approach is superior to other types of initial dosing strategies.

The NHLBI believes in the potential value of using a person’s genetic information to help inform treatment strategies, but the Institute also believes that the application of pharmacogenetics in clinical practice should be put through the same type of rigorous testing as any other treatment strategy. Therefore, the NHLBI decided to initiate and sponsor a large, randomized clinical trial for warfarin dosing to answer an important clinical question that affects millions of people. 

How can genetics influence warfarin activity?

Variations in two genes are principally involved in the different responses observed among individuals treated with warfarin. The first is VKORC1, which encodes an enzyme responsible for activating Vitamin K, which is essential for blood clotting. Some variations of VKORC1 make the enzyme more susceptible to be suppressed by warfarin, possibly increasing the risk of bleeding. The second gene is CYP2C9, which encodes a key enzyme in the liver responsible for breaking down a wide range of compounds, including warfarin. Some variants in CYP2C9 reduce its metabolic activity, thus leading to higher levels of warfarin in the blood, again potentially increasing the risk of bleeding.

Why did the study choose to compare a pharmacogenetic formula to a clinically guided formula?

By comparing two formulas that differ in just one aspect (genetic variables), COAG was able to address a specific question, namely, how much benefit does genetic data offer in a dosing formula. Had the study compared pharmacogenetics to a standardized dose of warfarin, the study would not have known whether the genetic or clinical components of the formula would be responsible for any observed benefits or risks. Identifying the specific benefit level of genetics can help determine if the value outweighs the costs associated with genotyping patients. 

What was the basic design of the study?

The COAG study recruited and enrolled 1,015 participants who were prescribed warfarin therapy and were randomized 50/50 to receive a warfarin dose based either on clinical variables alone or clinical variables plus genotype data. The target goal was to have the blood take about 2 to 3 times as long to clot as normal; in clinical language this is known as achieving an international normalized ratio (INR) of 2-3. The participants received their initial calculated doses for the first three days of treatment, and the dose was adjusted on day 4 and/or 5 using a similar formula. After day 5, the patients continued their treatment for 23 more days and any further changes in dosage were made using standard adjustments. During the first 28 days, the study was “blinded”, meaning that neither the participants nor the physicians knew which formula was used (the clinical or the pharmacogenetic one) nor did they know which dose of warfarin they were receiving. At day 28, the study was unblinded (meaning patients and doctors were told which dosing formula had been used) and data was collected for the primary and secondary measurements. Patients were monitored and data collected for another five months.

What were the study’s primary findings?

The primary statistic measured in this study was known as percent time in therapeutic range (PTTR). PTTR measures how often during the 28-day study period the patient was within the target INR range of 2-3. Among all participants, after four weeks, the PTTR was 45.2% in the genotype guided group and 45.4% in the clinically guided group. Among African Americans, the PTTR was 35.2% for the genotype guided group and 43.5% in the clinically guided group, indicating genotyping provided less effective anticoagulation control.

Why did the study choose Percent Time in Therapeutic Range as the primary measurement, and not bleeding or clotting events? Wouldn’t health outcomes be more relevant to people taking warfarin?

The primary goal when administering warfarin is achieving an effective and stable dose as quickly as possible and the COAG study measured that goal. If an effective and stable dose is achieved, it is expected that the number of bleeding or clotting events would decrease. In addition, the expected number of bleeding or clotting events was anticipated to be low. Therefore, a study designed for bleeding or clotting events would have required a much higher number of participants.

Were there any adverse health events?

Overall, serious health problems were rare, as expected. Fourteen people developed a major bleeding issue, 33 develop clinically relevant non-major bleeding, and nine participants had a thromboembolism (blockage of blood flow due to a clot). There was no significant difference in the occurrence of these events after four weeks of warfarin therapy between the two study arms. Three deaths occurred during this period; two in the genotype-guided group and one in the clinical group.

Do COAG’s results close the book on using gene information for warfarin, or are there some people who might benefit from pharmacogenetic dosing?

COAG was not able to identify any group of people who might benefit from treatment. Further research may be needed to determine if other genes may influence how warfarin works. In addition two other smaller studies done in Europe also were published at the same time as COAG. One study used anticoagulant drugs that are slightly different from warfarin and did not show any difference in the percent time in therapeutic range (PTTR) at 12 weeks. Another study showed a difference in PTTR at 4 weeks. However, unlike COAG, the second European study did not use a dosing formula to determine the initial dose of warfarin in the control group but rather used a fixed dose, so it is impossible to say whether the difference was due to the clinical or the genetic factors used in the formula. Taken together, these studies indicate that the incremental benefit of adding genetic information to the known clinical factors influencing warfarin dose is probably minimal and does not justify the costs and time inconvenience of genotyping. One should not use such an approach in some groups (such as African Americans) where the pharmacogenetic formula seemed to be result in worse anticoagulation control than the clinical formula.

Who funded this study?

COAG was supported by an NHLBI contract (HHSN-268200800003C). In addition, Bristol-Myers Squibb donated the warfarin (Coumadin) for the study, while GenMark Diagnostics and AutoGenomics loaned the devices needed to genotype patients to the participating clinical centers.


For More Information

For the Public: NHLBI Health Information Center

The NHLBI Health Information Center provides information to health professionals, patients, and the public about the treatment, diagnosis, and prevention of heart, lung, and blood diseases and sleep disorders.

P.O. Box 30105
Bethesda, MD 20824–0105
Phone: 301–592–8573, or dial 7–1–1 for access to free Telecommunications Relay Services (TRS)
Fax: 301–592–8563

For the media: NHLBI Communications Office

The NHLBI press officers can gladly provide additional information and assistance related to this study, or arrange interviews with spokespeople to discuss the COAG findings or talk about additional NHLBI research into warfarin and/or pharmacogenetics.

Phone: 301-496-4236
Ask for press officer on duty