When I started my company in 2009, personal experience was a driving force in identifying the problem I wanted to help solve. As a neurosurgeon, I often treated patients who were neurologically devastated by acute bleeding in the brain after years’ worth of poorly controlled high blood pressure. In many cases, when I asked the family why the patient quit taking their hypertension medication, they described the patient as “not a pill person.” Too often, I filled out their death certificates. We address medication nonadherence, the inadequate treatment of chronic disease due to the failure to take, or adequately adhere to, important medications. It plays a major role in poor outcomes. In my new entrepreneur role, creating a novel digital intervention to address the problem and raising money from others to do so, I needed to draw upon more than just my anecdotal experience. I needed to gather hard data. So I spent months reading the scientific literature, speaking with medication adherence researchers, and making myself an expert. The problem was even bigger than I had realized. The potential opportunity to make a difference was enormous. To that end, I needed to select the most compelling facts about the problem for our pitch deck. I kept coming across one fact again and again — not only in the medical literature but also in the popular press and on the websites of other companies in the medication adherence space: medication nonadherence causes 125,000 deaths per year in the U.S. As a physician, this struck me as a critical fact. If we had room for only the top five pieces of data on a PowerPoint slide introducing the problem, this would be one of them. In learning to write scientific papers, I was well-trained to chase down the primary source for each fact, read it, and cite it accurately. That was how to do science right. I would stick to those same high standards in my start-up, including on the marketing side. I needed to cite primary sources on our slides. The problem I ran into in chasing down the primary source for this particular piece of data, however, was that I reached a dead end. Each paper seemed to cite a different reference, a cascade of different papers citing different secondary sources. I suspected that the peer review process of most medical journals stopped short of verifying references, and understandably so. It’s an onerous task. Determined to find the primary source myself, I spent countless hours (and plenty of money, buying scientific papers behind paywalls) carefully tracking citations back in time. The earliest source I could find was from the CDC: National Vital Statistics Reports, Vol. 50, No. 15. I obtained that source and read it. There was no mention of deaths due to medication non-adherence. I had suspected that there wouldn’t be. The annual tabulation of deaths and the breakdown of their causes derive from official death certificates. From my experience in filling out these forms, I knew that medication nonadherence is not an allowable “cause” of death. As an aside, should it be? If someone with hypertension who was “not a pill person” dies from hypertension-related bleeding in the brain, did they die because of hypertension or because of a failure to treat hypertension? One could logically argue the latter. On the other hand, disease and death are complex and multifactorial. Trying to disentangle the specific role of medication nonadherence among the litany of other potential complicating factors (lifestyle, socioeconomic, coexisting medical conditions) is nearly impossible. But back to the PowerPoint quandary. Back in 2009, did I include this popular “fact” about 125,000 deaths despite my inability to find the primary source? Yes. Why? Momentum, and maybe a bit of peer pressure. This “fact” was cited in the introduction sections of articles in peer-reviewed journals that I respected. That justification, for me, was strongest. Plus, other adherence companies — our competitors — included the fact in their pitch decks and white papers. Further, my colleagues were in favor of including it. There seemed to be a momentum fueled by repetition and the desire for simple, straightforward, and compelling data. In the back of my mind, another justification for including it was this: maybe I was wrong. Maybe there was a reliable primary source, the 125,000 number was accurate, and I had simply stopped short of actually finding it. Luckily, the ramifications of using this potentially questionable “fact” — which still circulates to this day — are relatively benign. Based on plenty of rigorous studies over decades, there is consensus that the problem of medication nonadherence is a problem of tremendous significance in need of better solutions. We don’t really “need” this one piece of data to help convince anyone. I, for one, don’t use it anymore. (In case you’re wondering, medication nonadherence is not limited to the U.S. and is not explained simply by high drug prices. It’s a global problem. Although out-of-pocket costs impact adherence to some degree, lowering out-of-pocket costs to $0 makes a surprisingly small dent in the problem.) What bothers me is not the potential inaccuracy of the 125,000 number. In fact, if pressed, I would have to estimate that 125,000 is too small a number, especially if you consider the number of deaths that nonadherence “contributes” to rather than “causes.” What bothers me, instead, is the sheer momentum that a “fact” can achieve in the absence of a clear primary source. And I’m bothered by the lack of rigor in hunting one down. I’m pointing out one small example of laxness in one specific corner of the medical world, but it certainly exists more broadly. It exists most famously in politics, of course, where “facts” seem looser than ever and the tolerance for laxness legendary. Not all aspects of our lives need to be — or should be — informed by reliable facts. We tend not to seek out facts in religion, entertainment, or the arts. But the closer you get to the science, particularly when decisions that affect people’s lives are informed by facts, the more we really should know exactly where our facts come from. Source
