Evidence based medicine has much to offer, but one has to remember Einstein’s famous remark: “Not everything that counts can be counted, and not everything that can be counted, counts.”
Mrs. R comes in to discuss some recent lipid testing. She is a 44 year old married woman with a BP of 136/72, has a minimally elevated BMI of 27, and has never smoked. She has no family history of coronary artery disease or stroke. She exercises 3-4 times weekly for 30-45 minutes, either at the gym or jogging with her husband. She is careful about salt and fat. She takes several health supplements and practices yoga for relief of stress from her job as administrative assistant for an aggressive trial attorney. She had a health assessment through her gym and her trainer suggested she see her doctor to talk about treating her lipids with a statin, telling her it will reduce her risk of heart attacks by 30%. Her total cholesterol is 260, with an HDL of 36 and LDL of 152. Her fasting sugar is normal. She has no symptoms.
I explain the following*: Her risk of a cardiovascular event (fatal or non-fatal heart attack or stroke) in the next 10 years is about 7/100. Current studies suggest that she can reduce this to just under 5/100 over a 10 year time period by taking a statin. If I treat 100 patients just like her for 10 years I will probably prevent three events (number needed to treat or NNT). Roughly 1 person in 100 treated with a statin will have a serious but treatable complication from statin treatment during that 10 years and roughly 10 will stop taking it because of non-threatening but bothersome side effects.
She decides to think about. Six weeks later she comes in to discuss her plan: no medication at this time, but come in annually to check her lifestyle, BP and lipids, reassess her risk. She notes: “I’m pretty healthy, and I work really hard to take care of myself. My risk is low and I don’t like the idea of taking a medication to prevent something that probably isn’t going to happen anyway. My aunt takes medication for her BP and is always sick, and has lots of troubles with her medications.”
That same week, I also see Mrs. B, 43 yo married teacher with a BP of 133/78, a BMI of 26. She has never smoked and has no family history of either coronary artery disease or stroke. She exercises 4-5 times weekly for 30 minutes jogging with friends. She enjoys regular day hikes and snowshoeing with her husband. She had several successful surgeries as a high school athlete for knee injuries and considers herself well. She comes in several months after her annual exam because a well-liked supervisor at work, an overweight male smoker, died unexpectedly of an MI. Her lab shows a cholesterol of 262, an HDL of 35 and an LDL of 150. Her fasting sugar is normal. She has checked on line and has read that a statin will reduce her risk of a heart attack by about a third.
I explain the following*: Her risk of a cardiovascular event (fatal or non-fatal heart attack or stroke) in the next 10 years is about 7/100. Current studies suggest that she can reduce this to just under 5/100 over a 10 year time period by taking a statin. If I treat 100 patients just like her for 10 years I will probably prevent three events (number needed to treat or NNT). Roughly 1 person in 100 will have a serious but treatable complication from statin treatment during that 10 years and roughly 10 will stop taking it because of non-threatening but bothersome side effects.
She asks several questions about lab monitoring, generics and side effects, and then says she wants to take a statin and treat to an LDL of 70 if possible. “I work hard to be healthy and I feel good, but I’d hate to let all my effort to stay well go to waste because I skimped on something simple like taking a pill every evening. I know my risk is pretty low, but I certainly don’t want to blow it off and then be sorry. Bad things can happen, you know.”
These two patients (who were, in fact, next door neighbors) with nearly identical health status and risk, came to very different decision based on the same evidence. What can we learn?
Numbers do not tell us what is right or wrong. At best, they help clarify the relative size of the potential risks and benefits. After that, decisions are made using personal preferences and personal contexts.
In this example, both patients see themselves as healthy and both are aggressive about maintaining health. Both understand their risk of cardiovascular disease is low. Both recognize the potential role of lipids for their enduring health. Despite this, they come to very different decisions because of things not so easy to measure. What explains the difference? One way to think of this is the availability heuristic, one of many common causes of cognitive bias in decision making:
- Mrs. R doesn’t picture herself as ill. Her life experience means that the most potent and available image of a medication-taker is her aunt, and she is more concerned about the potential downside of the medication.
- Mrs. B has had a personal positive experience with medical care, and her most potent and available image is of someone dying unexpectedly of heart disease.
Many clinicians (and quality reviewers, payors and auditors) use evidence based medicine as if it were The Rule Book, a collection of recipes capable of providing The Answer to clinical questions about diagnosis and treatment.
If it were only that simple.
Evidence based medicine is only one tool in the decision making process. It allows us to use validated information to anchor and inform our decisions, a big improvement over tradition based or expert based care where unilateral decisions are made by the clinician based on tradition, theory, and the clinician’s personal values and preferences (prejudices).
It is neither quick nor easy to marshall and explain the often complicated evidence available. It requires research and preparation of patient information sheets for the commonest problems, and it may be necessary to admit ignorance and set aside time to collect and review the relevant information before a return visit for discussion and decision making. Helping the patient use the data to make a personally suitable decision takes considerable time and requires both a knowledge of the patient and considerable listening skill. Not all patients are willing or able to participate in this to the same degree. The hardest part is often letting patients we care deeply about make decisions that are right for them, but not the decision we would make in the same circumstances.
Footnote: (*Note: I take pains to present statistics to patients using whole numbers, and not percents or ratios. I try to use a consistent denominator, and round to avoid decimals. I try not to use numbers larger than 100, and if I do, I try to use common multiples like 200, 500, or 1000.)