“Doc, I can’t take the pain anymore,” Mrs. Wiedner said as she hobbled with a cane back to the examining room in the office. “I’ve had enough!” At eighty-six years of age she had suffered from severe arthritis in her knees for many years. “What about this Celebrex I keep hearing about? My neighbor says it works like a charm, but I hear it causes heart attacks.”
Celebrex is a pain reliever in the family of medications called COX-2 inhibitors and part of the broader group of pain relievers called non-steroidal anti-inflammatory drugs (NSAIDs) that include over the counter medications such as ibuprofen and naproxen. These medications have received widespread attention for their potential to increase the risk of heart attacks and strokes, and Vioxx, a COX-2 inhibitor, has been removed from the market.
Unfortunately all medicines (even over the counter medicines) have risks, but to understand the risks associated with a medication you first have to understand something about the numbers behind risk - a topic often neglected by patients and their doctors.
What if I told you that I have secretly developed a new drug, a potion that after only one dose insures perfect physical health. Would you be interested? Of course. Would I be rich? Probably. What if I told you that taking the drug, unfortunately, also doubles your risk of developing Alzheimer’s disease at an early age. Would you take the drug?
I assume you’re not too keen on getting early Alzheimer’s, let alone doubling your risk, but what if before taking the drug your risk of developing early Alzheimer’s is only one in a billion. Would you then be willing to double your risk of early Alzheimer’s to insure perfect health? I probably would. Sure, your risk relative to someone who didn’t take the drug would be double, but your absolute risk would only be two in a billion – a trivial number or risk to insure perfect physical health.
In selling the drug I would for full disclosure mention the increased risk of early Alzheimer’s. Though unless I wanted to scare you, rather than stating that the drug doubles a patient’s risk of early Alzheimer’s, I would state simply that it increases your risk by one in a billion. Either way I would be accurate, but why scare you?
What if, however, your risk of developing early Alzheimer’s before taking the drug is instead one in five or 20%. If the drug that guarantees perfect physical health doubles your risk of early Alzheimer’s (that is, the relative risk is again doubled), your absolute risk of developing early Alzheimer’s after taking the drug would now be two in five, 40%. With that risk for early Alzheimer’s, would you still want to take the drug? Maybe, but those odds are clearly less favorable.
Therefore understanding relative risk, knowing whether something doubles or halves your risk, is often insufficient. It is important to ask what the initial risk is and think about what the change in absolute risk really is. Never going outside may decrease your risk of being hit by a meteor or being robbed, but what is your risk to begin with? Most people would probably still decide to walk outside.
Another way to look at the numbers is to calculate the number needed to treat, the number of patients needing to take the drug to affect one person. The number needed to treat (NNT) is equal to 100 divided by the difference in absolute risk expressed as a percent. In the last example, the NNT for one additional person to develop early Alzheimer’s is 5 (100 divided by 20). In contrast, in the first example the NNT for one additional person to develop early Alzheimer’s is 10 million (100 divided by 0.00001).
Drug companies, the media, and others have their reasons for either magnifying or diminishing the perception of medications’ risks. In reading numbers it is always helpful to question what the vested interests are of those who quote you the numbers. Then you can look objectively and evaluate the numbers.
Celebrex is, therefore, an excellent example. The media has certainly stressed the potential increased risk of heart attacks from COX-2 inhibitors, but for many people the absolute risk may be low. What is Mrs. Wiedner’s risk of a heart attack to begin with? Are there other more common risks from Celebrex she should worry about? How much is the pain really bothering her? Are there alternative medications that she might consider taking? Are the risks of Celebrex to her worth the benefits? These are questions that only she can answer with the help of her doctor and understanding what the numbers really mean. Once a person understands risk, however, they can make their own decision.
Hopefully this look at the numbers behind risk assists you in having informed discussions with your doctor about medications and medical treatments.
Posts
3 comments:
Minor additional point about risk: what is the severity of the risk in question? If you have a doubled risk of a headache, it changes your calculation relative to a doubled risk of a lethal heart attack. Pedantic point, but something I think about when assessing the relative risk of vaccinating my child.
How honest do you think drug companies are when disclosing the relative frequency of various side effects associated with the use of their product in specific patient populations? I find it surprising that the heart attack issue didn't come up in the pre-debut controlled trials of Vioxx.
Great points!
As for the first, worth discussing further in another entry. Other examples: in the context of elective surgeries such as knee replacements, lasik, etc. Huge upside but also potential for huge downside..
As for the second, there are several reasons why major adverse effects don't necessarily get discovered before FDA approval. One is merely numbers, that the number of people tested in clinical trials (a few hundred to a few thousand in phase III trials) pales next to the population of patients that take a drug once it is approved (general population). A good reason for more post-marketing surveillance and developing a better system to do so in the U.S.. Also, a great reason not to be the first to try new drugs unless there's a compelling reason.. Hmm, worth its own entry as well!
Thanks so much for leaving a comment!
"Post-marketing surveillance" in the U.S. is not up to snuff, but is it better elsewhere?
Post a Comment