How to Find Your Own Personal Miracle Cure
Modern medicine is driven by huge, randomized double-blind trials that may not apply to your own particular genes or lifestyle. Here’s how to join the truly personal medical revolution.
Illustration Credit: Linda Yan
Just in case you’ve recently crawled out from under a rock, let me be the first to tell you that the adoption of self-tracking electronic gadgetry has reached epidemic proportions. A hefty step up from climbing on the bathroom scale once a week to see if the new diet works, digital wireless devices now measure everything from the number of steps you take to actual medical data such as blood sugar levels, heart rate, and heart rhythm. Data acquisition can require operator input: “Here’s everything I ate, now you tell me the grams or calories of carbs, protein, and fat.” Or it can be completely automated: “Good morning, Deborah! Not such a good night, eh? You were tossing and turning for one-third of the time you spent in bed. Not very good sleep efficiency.” OK, the sleep monitor in FitBit isn’t quite that personal yet, but a mere glance at the data acquired while you sleep can evoke the same feelings as a first-grade teacher’s praise or criticism.
Given that one in 10 Americans over the age of 18 now own some kind of activity tracker, it’s time to ask, “What’s going on here?” Are the new monitoring devices merely a reflection of an avid industry’s hunches about our willingness to buy new ways to look in the mirror? Or is something greater at work? What can be achieved by monitoring, tracking, and possibly comparing our lives to other parts of our own lives—and to the lives of others? What’s the benefit? Let’s talk about what we might learn from allowing the focus of a rigorous medical research experiment to be on a single individual: namely, you. But first, a little history.
As medicine struggled to emerge from our own perceived “dark ages,” one key goal was to make the practice of medicine subject to the highest degree of rational judgment. To that end arose the double-blind clinical trial, in which a therapeutic intervention is applied to a sizable group of people who are assigned to receive either the active therapy or a placebo treatment, and no one knows who gets what. In general, the bigger the n (the number of study subjects), the better the trial. So results from a study where n=1,000 is considered more significant than a study where n=100, which in turn is better than a study where n=10—and a study where n=1 is not considered a study at all: It’s an anecdote rather than scientific data. The first randomized, double-blinded clinical trial appeared on the medical scene in 1948 with antibiotic treatment of tuberculosis, and became the gold standard for rational medicine by the end of the 20th century.
Almost immediately, however, pushback began. In 1953, a renowned medical statistician named Lancelot Hogben cautioned against the growing reverence for the clinical trial. Granting the theoretical rationale for generalizing experience, he acknowledged what many doctors still do not: the average experience in a clinical trial may have no relevance to a prescription for an individual patient, who is never “average.” Indeed, the patient’s varying responses to different treatments is ultimately all that matters in selecting the appropriate treatment for that patient at that particular time.
Put another way: When you start any new diet or new prescription, you are essentially starting a new clinical trial in which n=1=you. No matter how robust the results of all previous clinical trials, your own genetics and lifestyle may cause your particular results to vary—so it is in your interest to find a way to determine for yourself whether any long-term prescription is actually better for you than a placebo.
The concept of n=1 was not really applied nor championed until 1986, when Gordon Guyatt, MD, of McMaster University wrote about it in the New England Journal of Medicine. One of the fathers of what’s called evidence-based medicine, Guyatt described the case history of one single patient working with one physician, in a standardized format to distinguish useful from useless medications. In this first formal n=1 study, the patient recorded her responses to medications that were packaged to conceal their identity from both the patient and physician. The active medication and the placebo were also rotated to ensure that neither the doctor nor the patient knew which pill was being taken. Rotating, blinded medications were a key aspect of Guyatt’s physician-based n=1 trial. But the concept didn’t catch on.
Then, in 2010, Eric B. Larson, MD, suggested in the Journal of General Internal Medicine that physicians should be able to “order” an n=1 trial, administered by some experienced entity, for a variety of treatments.
Dr. Larson suggested that if you had a new diagnosis of hypertension, you might be randomly exposed to two different drugs and a placebo, one at a time. At the conclusion of the study period a combination of your own observations (“I felt better with the placebo”) and data from the blood pressure monitors (“But your blood pressure was much better on the first real drug!”) would enable you and your physician to make an informed decision about your future treatment. Nowadays that idea is gaining real traction.
“Whoa!” You might be saying. “Just give me the drug that’s likely to work—I don’t want to mess with all of that.” I hear you, but before you give up on the idea of such close and personal medical testing, consider how little we know about who is likely to benefit from a particular medication—and the potentially high costs of our unknowing.
Keep in mind that the all-too-literal gold standard for pharmaceutical companies is drugs that have been proven statistically to help large numbers of people who suffer from common, ongoing, and typically lifestyle-related afflictions like high cholesterol, high blood pressure, and heartburn. For such blockbuster drugs, astronomical sums are spent on development, huge clinical trials, marketing the successful results of those trials to doctors, and marketing directly to patients to ask their doctors to inquire about a prescription. Of course all that money can create problems. An alarming 2009 study mentioned in this September’s Atlantic found that as many as one-third of scientists confessed to “a variety of questionable research practices including…changing the design, methodology or results of a study in response to a funding source.” The Atlantic article also noted that in 2012, when researchers at Amgen “tried to replicate 53 landmark cancer studies, they could replicate just six.” All that glittering data is not actually gold.
An even more alarming—and perhaps related—number is what’s known as the NNT, the “number needed to treat” to create the desired helpful effect. When you take a drug your assumption is that the pill is going to help you. In other words, you are expecting an NNT of 1. In fact, the likelihood of many common drugs actually providing a significant benefit to you personally may be quite small. For example, for every person who gets the desired effect from each of the 10 best-selling drugs, between three and 24 people need to take the drug. The vast majority may get no benefit at all. (See theNNT.com.)
Let’s take a closer look at the NNT of what I consider a particularly egregious example among the 10 top-selling drugs: Crestor (rosuvastatin), which is prescribed for high cholesterol. Never mind the observation that people over 65 tend to live longer if they have high cholesterol or that women of almost any age live longer with high cholesterol. Let’s just look at Crestor’s NNT: please note that it requires treating 20 people with Crestor to prevent an adverse cardiac event in one person (and these are among the more favorable statistics on statins in general). Meanwhile, many of the other 19 who will not benefit from the drug are likely to suffer from muscle soreness, diabetes, depression, and/or cognitive impairment. Such statistics become mind-boggling when you think about how much money is spent selling these pills both to doctors and to patients. How much extra suffering is created to prevent that single cardiac event? Crestor may make sense within the statistical gold standard of large double-blind trials, but perhaps no sense at all when n=1=me.
Nationalized Individual Medicine
One hopeful sign is that in January of this year, President Obama announced a $215 million “Precision Medicine Initiative” to spearhead a new era of individualized medicine with two main aspects. First, the initiative will consider how taking into account genetic information and individual variability can better personalize cancer treatments and make them more effective. Second, the initiative will oversee the creation of a vast, voluntary national research cohort in collaboration with the NIH and others. Participants as well as agencies will decide on which data will be measured and preserved and create a database that might inform qualified research inquiries about individual variability. The thrust of the initiative is primarily to improve cancer treatment, but the database could also contribute to understanding other questions. Personally, I would like to see a collation of data regarding lifestyle choices, genetic findings, and clinical outcomes, which might help us avoid many of the top 10 prescription drugs altogether.
More personalized medicine is beginning to happen from the top down, but that’s because it’s already happening from the bottom up. From an initial meeting of 30 curious souls in San Francisco in 2008, the “Self-Quantification” movement has grown to include an annual conference, spin-off groups in 18 other cities, and over 200 attendees at some of the now bimonthly meetings of the original chapter. Devices unimaginable a few years ago can capture body movement, heart rate, skin temperature, and more. Self-measurement of blood sugar and cholesterol are available from simple pinpricks, and of course for less than $100 you can find out a lot about your genetics and your ancestry from 23andMe.com.
Not only are folks enjoying the different gizmos that make it possible to quantify different measures of self, but they are able to research on the Internet either professional or crowdsourced solutions to self-identified problems. (Check out CureTogether.com.) No longer are medical interventions cloistered away in medical libraries requiring professional membership; research abstracts are readily available to any curious seeker on the NIH’s PubMed website. You can learn about a solution you hadn’t considered, and take it to your doctor: “What about this?” Or carry the information to a meeting of your peers and ask, “What are you all thinking about this? What shall we try?”
How High Can We Fly?
In a private conversation, investigative journalist Nina Teicholz, author of The Big Fat Surprise and challenger of the status quo in her own big way, expressed concern at the n of 1 concept, stymied by the obvious flaws in relying on information based on self-reporting. But as we pursued the possibilities further, she saw n of 1 data in a new light, “more like folk wisdom…the way that knowledge grew before the scientific method.” By respecting our own folk wisdom we might gain a greater respect for the evolutionary tradition of thousands of years of accumulated life wisdom—and add their insights to our new and enlarging “folk wisdom database.”
In his TED talk WIRED editor Gary Wolf raised the possibility that with our gene readings and our sleep and heart rate monitors we might develop a more intimate relationship with our own self. I would add the hope for a more tender and insightful relationship as well. Personally, I’ve found that nothing reaffirms my meditation habit like a variable heart rate monitor.
Can you take this further? What can you learn about the nature of being the human you are from learning about your origins, your heart rate, your sleep? N of 1 research will truly take form in the eyes of the beholder. The self-knowledge that translates into greater work efficiency for the entrepreneur may manifest as deeper calm for the reflective yogi or meditator.
Creating an n=1 Trial
- Start with one patient with a clearly identified problem or symptom. Ideally it would be numerically quantifiable (like blood pressure), but it could be subjectively assessed (quality of last night’s sleep on a 1 to 10 scale). Or a combination of a sleep monitor data and self assessment.
- Identify an intervention that can also be administered as a placebo. Total body massage, for instance, would not be a good intervention. Sublingual melatonin would work: a compounding pharmacist would have to come up with a similarly flavored placebo. Call the intervention A, call the placebo B.
- Identify the question you are asking! You can’t test melatonin’s effect on sleep and decide after the fact that it had such a good effect on their digestion that you’d rather test that instead.
- Decide how much of an effect you want to see to consider it significant, clinically. (If you can, ask a statistician for help.)
- Determine how long it would take to identify the effects of the intervention. One week, two?
- Plan to carry the trial through SIX of those time periods, and randomize the intervention: AABBAB or ABAABB or BABBAA—you get the idea.
- Does the effect of the intervention carry over for a day? A week? After you stop taking it? If so, you need to schedule that “washout” period between each of your interventions.
- Analyzing the data may seem straightforward. But it may not. Take into account that the washout period may be longer than you think. For a serious statistical analysis—well, that is beyond me and the scope of this article!
Here’s a common scenario: Your doctor looks at your fasting blood sugar and your hemoglobin A1C and detects early signs of type 2 diabetes, a potentially devastating condition that is completely reversible if detected before insulin is required. Hopefully, your doctor encourages you to eat more wisely and exercise more often. But you leave the office wondering how worried you should be. Perhaps you’ve read on the Internet that both lab tests may give false positive results: making you look sick when you’re not. That’s true, but you should not assume your tests are wrong. What you really want now is ongoing data—more data than you can reasonably expect to get from your doctor—and it turns out that you can now do your own complete diabetes blood testing at home. Now is the time for n of 1!
- Buy a blood glucose meter, available in most drugstores and online, and learn how to measure your blood sugar with a simple finger stick.
- Test your blood sugar at the times when a doctor would test your blood in a full diabetes test: morning (after fasting for at least 10 hours), 2 hours after a really big meal with all kinds of food, and again an hour later. If you’re worried that you tend to get hypoglycemic, carry the test out to 5 hours: if you feel faint, hot, sweaty, and shaky and your blood sugar has plummeted more than 30 points from the previous number, you have hypoglycemia. (Eat more protein and fat, eat less refined carbs.)
- Fasting blood sugar should be 95 mg/dL or below, but can be much higher in some people who don’t have diabetes. At 2 hours, normal is below 140, prediabetes is 140–199, and outright diabetes starts at 200. These numbers are true for everyone. Same numbers for three hours.
- If your 2- and 3-hour tests are normal, but your fasting level is high, you might start an individualized “bio-hacking” program. Try eating a number of different ways, a week or two at a time, and keep checking your morning blood sugar: Give up alcohol, give up sugar. Give up that food you think you might be allergic to. Eat more carbohydrates or less. Take some berberine supplements. See what happens. Keep a log of all your numbers and your food experiments to share with your doctor and your new research group.
Three Gizmos Worth Getting
- Blood sugar and ketone monitor: I use a Precision Xtra monitor that rapidly checks my blood sugar and blood ketone level. I have relied on it when I experimented with a high fat/low carb, ketogenic diet, and I continue to check fasting blood sugars during different periods of diet trials, alcohol abstention, or change in exercise pattern. (Your health prognosis is excellent if blood sugar levels are lowish: 70–95.)
- Heart rate variability: More useful than a pulse monitor, the Inner Balance Trainer heart rate variability monitor is an ear clip that connects by Bluetooth to your smartphone to assess the balance between your sympathetic and parasympathetic nervous systems. The sympathetic nervous system yells, “Faster, here comes the lion!” and the parasympathetic system coos, “Relax, enjoy your meal, maybe nap.” It’s best to have them both informing your pulse at all times, because they influence such diverse areas of health as our cardiovascular system, our mood, our degree of inflammation, and our microbiome. Learn more at the HeartMath Institute website.
- Sleep well or not? The sleep monitors offered by FitBit and Zeo have informed the mainstream about the quality of their sleep, and consumers know they want better sleep. “Hey, Doc, look at my Zeo—what should I do?” (Watch the tide beginning to turn.)
Deborah Gordon, MD, has worked in a variety of medical settings, including private practice, emergency medicine, outpatient clinics, and as the Medical Director for the first Migrant Farmworker Clinic in Southern Oregon.