Snowball in a Blizzard

by Steven Hatch

  TABLE 4.1

  Yet the real dispute came in patients between the ages of sixty and eighty who did not have kidney disease or diabetes. The JNC7 goal was a pressure below 140/90, whereas in JNC8, the committee basically said that they should be thought of as the same way we think of patients over eighty, with a revised target blood pressure below 150/90. At its core, the dissenters were saying, we believe that we should get our patients below 140! while the majority opinion offered the following reply: we can’t confidently say that dropping below 140 confers unambiguous benefits.

  Currently, an estimated population of nearly 42 million people are above the target pressure by JNC7 criteria; with the revised JNC8 guidelines, that number drops to about 14 million.† The minority essentially rejected this reclassification scheme, but, as I noted, they didn’t reject all of it. The sticking point revolved around whether it was legitimate to raise the goal pressure of patients between ages sixty and eighty if they did not have the complications of diabetes and kidney disease. Should the goal be a systolic blood pressure less than 140 or 150? That was the crux of the matter: one group insisted the lower target was the ideal, while the other group (the majority) was skeptical of the benefits of this lower target.

  With the reclassification, a little under 10 percent of patients would no longer be regarded as in need of therapy—about 6 million out of the 70 million people whom the JNC7 guidelines considered treatment eligible. Therefore the bulk of the controversy lay in how aggressively to approach the approximately 30 million people who are already on treatment but are above the previous (JNC7) guideline’s goals.

  How could such experts be so divided? In large part, it’s because of a legitimate dispute about what we can reasonably infer about three points of systolic pressure.

  One particular contested trial highlights this conundrum: the Systolic Hypertension in Elderly Patients, or SHEP, trial. SHEP included nearly 5,000 patients older than sixty, most of whom had not been treated for hypertension before enrolling in the trial. Researchers gave them a combination of antihypertensive medications or placebos, and followed them for more than four years on average.* The trial was a resounding success: the total number of strokes was nearly cut in half, the amount of heart failure cases decreased by more than half, and there were more modest reductions in heart attacks, heart bypass surgeries, and transient ischemic attacks (so-called ministrokes, which can be harder to diagnose than full-blown stroke events). SHEP showed, without any doubt, that lowering blood pressure results in big gains.

  A placebo is a “do nothing” pill, which I’ll discuss in Chapter 6.

  But how to translate SHEP to a set of guidelines by a committee that has been tasked to weigh in on the bright line of 140/90 versus a higher target? Herein lies the problem. The average starting systolic blood pressure of the SHEP participants, that is, both the placebo and treatment groups, was 170. The placebo group dropped its average pressure to 155 over the course of the trial, while the treatment group dropped its average pressure to . . . one hundred forty three. Therefore, the majority of the JNC8 panel concluded this study provided strong evidence that lowering systolic blood pressure below one hundred fifty carried benefits. However, because the final blood pressure of the treatment group was just above that 140 cutoff, SHEP was not deemed to support the lower target that the previous guidelines had advocated.

  Before I continue to analyze the dispute, it’s worth considering how the placebo group dropped its pressure if the researchers were not giving them pharmacologically active medications. This is a beautiful illustration of what is known as the “placebo effect,” where symptomatic or other changes take place in patients who are not given pharmacologically active drugs but believe they are. In the eyes of many believers in “alternative medicine,” the placebo effect is evidence of the magical and secretive powers of the human mind to heal the body, and much claptrap is devoted to the topic.

  More on alternative medicine and pseudoscience in the next chapter, but in this case I think there’s a pretty good explanation for why the placebo group in SHEP dropped its average systolic pressure by fifteen points. Simply put, by becoming aware that they had sky-high blood pressures—which they had been told could lead to very scary outcomes as part of the consent process for enrolling in the trial—they suddenly became focused on their mortality and made, consciously or otherwise, healthy lifestyle changes, probably by increasing their physical activity and eating either less or better, or less and better. In this respect, SHEP (and all trials of this ilk) probably underestimates the magnitude of benefit of lowering blood pressure, because a comparison of properly treated patients with a group of people who maintained a systolic blood pressure of 170 would almost certainly show an even greater reduction in cardiovascular events and overall mortality when compared to the difference between the treated and the placebo group (with its lowered average pressure of 155, which was still higher than the treatment group). Such a trial is no longer considered ethical, however.

  Keep in mind that SHEP was but one among many tiles in the mosaic of studies that the JNC8 experts were forced to sift through on their way to issuing blood pressure guidelines. Thus, one can pretty easily comprehend why a minority group could have split away and cried foul at the decision to change the recommendations—keeping in mind that the previous guidelines had taken nearly a decade to inculcate into the brains of primary care physicians, in the estimates of one group that wrote a mildly skeptical editorial in the same issue in which the JNC8 report was published.

  The question the dissenters were asking was this: Was the JNC8 willing to scrap the “140/90 is the target” message, which had become the equivalent of a public service announcement, especially to the community of primary care docs who may not have the time to delve into arguments about the nuances of study design, because the average blood pressure of this universally acknowledged well-designed trial was three points higher than the previous guideline? No matter where one falls on the specific issue of what the optimal blood pressure of this patient group should be, it isn’t especially difficult to see why they were so troubled by the change.

  Moreover, SHEP wasn’t the only prickly piece of evidence that caused headaches and led to the minority group’s decision to break with JNC8. One can get the feeling of entering Wonderland when reading about the controversy, but here is a glimpse into the debate. In expressing its skepticism of the benefits of lowering pressures below 140, the JNC8 cited two other studies that went by the acronyms JATOS and VALISH. These studies did get the average systolic blood pressures below the magic 140 mark, but neither study showed a benefit. So the majority of the committee believed that represented evidence that 140 should no longer be the goal, but 150 instead.

  Unfortunately, both JATOS and VALISH were considered to be of inferior trial design largely because they didn’t enroll enough patients to demonstrate positive effects if such effects were present (I’ll talk more in the later chapters, and in the Appendix, about the size of trials). Nobody, as far as I can tell, disputed the value of the SHEP study. The majority of the JNC8 felt the JATOS and VALISH studies provided indirect evidence against the lower target, but the minority said these studies shouldn’t be considered at all and were biasing the group. The dissenters further noted that JATOS and VALISH were Japanese studies and thought that it was potentially inappropriate to generalize studies from Japan anyway, given the significantly different diet and lifestyle of those patients.

  The dissenters were also upset about the exclusion of a trial that went by the acronym FEVER, which did show benefits for the lower target blood pressures, even though FEVER was a Chinese study, enrolling patients with what one would assume are equally different dietary and lifestyle differences compared to Americans (even if these differences aren’t the same differences as those of the Japanese study subjects). All I can conclude with confidence after poring over these papers is that the lead scientists need to work on developing better acronyms for their trials.


  As we saw in the data on screening mammography, another issue that at least indirectly drove the debate about the optimal target blood pressure involved the difference between the relative and absolute benefits of aggressive blood pressure control. Recall that when we looked at mammography, we found data showing that even big reductions in relative risk might translate to only modest absolute benefits. A look at the SHEP trial highlights a similar phenomenon with respect to blood pressure control.

  At the end of the trial, 149 out of the 2,371 patients in the placebo group of the trial had a stroke (in this trial, stroke was considered to be the primary outcome); in the treatment group, there were 96 strokes out of 2,365 patients. So although the relative risk reduction in strokes is 0.63—that is, treatment cuts the rate of strokes almost by half—the absolute risk reduction is about 2 percent, or about fifty strokes saved per nearly 2,400 patients over five years’ time.* What that probably means is that there’s a big benefit to dropping your blood pressure if you start out at 170, but those benefits get smaller once you’re down in the low 140s and you and your physician are contemplating adding a drug to get you to the low 130s.

  Because of dropout rates over the course of the trial, the study authors used some more sophisticated statistical tests and concluded that the medications reduced thirty strokes per one thousand patients, a slightly higher benefit than my raw calculation implies.

  Stop for a moment, and reread that last sentence, for it is absolutely critical to contextualizing the JNC8 dispute. If one can grasp the essential principle, one can apply the lesson to much more than just blood pressure management. The biggest benefits of blood pressure management come for those already at the higher end and diminish as we move closer to the “normal” value of 120/80. That’s true for many other diseases as well: for instance, diabetes may be diabetes, but very poorly controlled diabetes is much, much more dangerous than even moderately controlled diabetes. So patients need not feel like failures if they aren’t “at goal” with some disease—any kind of improvement tends to reap huge benefits. Even a little exercise is likely to provide real protection against heart attacks, heart failure, and strokes. Do patients who walk for half an hour twice a week derive the same level of protection from heart disease as those who transform themselves from couch potatoes to marathoners? No, but instead of becoming dispirited for not hitting that goal and losing their motivation to keep walking, such patients should feel accomplished that they’ve made a big difference in their lives. And their doctors should be encouraging this way of looking at the world instead of obsessing about hitting a given target.*

  The increasing emphasis on adherence to national guidelines, coupled with the computerization of patient health information, has had the unintended effect of forcing doctors, in particular primary care physicians, on achieving bright-line goals for their patients. That is, if some patient lowers their blood pressure from a systolic of 180 to 152—which confers a huge protective benefit—the physician may face lower “performance scores” because he or she is not at the goal of less than 150, and such performance scores can affect reimbursement.

  This is how understanding uncertainty can aid you as a patient when you talk with your physician, for with uncertainty comes a framework for knowing when it’s a really, really good idea to get treatment for some condition, or it’s perfectly fine to hold off. We’ll see the same kind of effect when we look at how high cholesterol is treated later in the book.

  Harm

  It might cross your mind to ask why we would want to stop lowering blood pressure, even if the benefits diminish as the numbers decline. Isn’t any benefit, even a little one, worth it, if it prevents something as devastating as a stroke? Again, the SHEP data can help explain in part why we shy away from being overly aggressive. While the reduction in cardiovascular events from SHEP is quite striking, hovering around 40 percent, the reduction in “all-cause mortality” (that is, just adding up all the people who died in each group regardless of cause) is a much more modest 10 percent, sufficiently small that the lower number of deaths in the treatment group might simply have been a matter of chance.

  Why is there such a discrepancy between these two measurements? Why doesn’t the reduction in mortality match that of strokes and cardiac events seen in the trial? The authors of SHEP don’t speculate, but all internists routinely encounter medical problems in the elderly that are often direct consequences of a blood pressure being too low. Keep in mind that blood pressures, like body temperatures or heart rates, fluctuate throughout the course of a day, and in older patients the swings can be more dramatic as the elasticity of blood vessels decrease. Thus, even simple maneuvers like rising from a seated position can cause major drops in pressure, and if the starting point is lowered through medications, such precipitous shifts can cause people to faint.

  Among the most common problems caused by fainting is a broken hip. Far worse than a broken bone almost anywhere else in the body, broken hips cause patients to be bedbound for long stretches, which increases their risk for blood clots, pneumonias, and other complications associated with surgery. Estimates vary, but the one-year mortality following a broken hip is probably somewhere between 10 and 40 percent—at least as bad as or worse than many cancers. At first glance, hip fractures may not seem like the kind of event that a blood pressure study should evaluate (and few have done so), but that hidden variable probably accounts for a chunk of the reduction in lifesaving benefit seen in SHEP. That is why being overly aggressive in lowering blood pressure in patients over the age of sixty might be dangerous.

  The question, then, is where this happy medium can be found, and the current evidence suggests that the benefits clearly outweigh the risks for these patients when their pressures are lowered to somewhere in the range of the low to mid-140s on average. Whether there is additional benefit to lowering one’s pressure even further than that is very much a matter of debate.

  The key to understanding how one can use uncertainty as a helpful tool lies in realizing that nobody truly knows the optimal low point. Indeed, it’s very clear that there is no single optimal low point for all patients, because factors like genetics, other disease, and smoking will all contribute to one’s risk of stroke, and some will require lower sustained pressures over time than others. But appreciating the value of uncertainty involves knowing oneself.

  Nobody expects patients to solve the problem of knowing their optimal blood pressure. But what goals do they wish to achieve, and are they willing to tolerate side effects, and even more serious problems, in pursuit of a particular goal? It may be that a given patient, having watched a loved one suffer a major stroke, wants to do as much as possible to minimize that risk, even if he or she understands that means complications like a broken hip—which has the potential to be lethal—may arise. Other patients may have very wide pulse pressures—high systolic blood pressures coupled with normal or even low diastolic pressures—for which aggressive approaches may carry significantly more danger. Others still may hate taking medications and find themselves after one drug in the mid-140s, but are looking at having to start a new class of medications to drop them further. These are the kinds of factors that patients can think about and should discuss with their doctors. Similarly, they should expect honest and informed replies about relative risks and relative benefits (to say nothing of absolute risks and benefits). I’ll say more about how to have that conversation in the Conclusion.

  The JNC8 brouhaha highlights as much about our emotional responses toward data as it does about the underlying disease process and the science we use to analyze it. Glenn Kershaw, a nephrologist (kidney and blood pressure doctor) with whom I work, summed up the controversy this way: “The old school view is that guidelines express certainty. The new school view is that guidelines emphasize the limits of evidence, and the unique role of the practitioner and patient.” To my mind, it is not that one approach is right and the other wrong; they’re doing different things, and, when we are at the li
mits of our understanding, struggling in the world of known unknowns, finding ourselves in the middle of the spectrum of certainty, conversations about medical guidelines must take this into account.

  In the previous chapter, we encountered a body of experts who presented a set of guidelines that incorporated uncertainty to its core. Although they were unified, the guidelines led to public controversy. In reviewing the JNC8 report, we observed how uncertainty and our attitudes about uncertainty can lead to controversy even among the experts. The next chapter will look at what happens when the experts are unified but a small and highly motivated activist group creates the impression that there is very little consensus, even though there is a very high level of it, not just among the experts, but among tens of thousands of physicians.

  5

  LYME’S FALSE PROPHETS: CHRONIC FATIGUE, TICK-BORNE ILLNESS, AND THE OVERSELLING OF CERTAINTY

  We are all capable of believing things which we know to be untrue, and then, when we are finally proved wrong, impudently twisting the facts so as to show that we were right. Intellectually, it is possible to carry on this process for an indefinite time: the only check on it is that sooner or later a false belief bumps up against solid reality, usually on a battlefield.

  —GEORGE ORWELL

  In the two previous chapters, we have moved along the spectrum of certainty into ever murkier territory. In the case of mammograms, I tried to show how the divide between the experts and the public occurred because most experts were always less certain of mammography’s benefits than was popularly understood. Moreover, there was a general consensus among experts that the magnitude of the benefit wasn’t as profound as laypeople thought. The latest blood pressure guidelines in the JNC8 investigated a question that involved so much uncertainty that even the experts were split—indeed, they were so split that one group simply broke off and issued a competing set of guidelines. The JNC8 controversy also highlights the issue of what the ultimate purpose guidelines should serve: Are they simply bright-line goals, about which the less asked the better, or should they reflect the limits of our knowledge and go no further? These are, by no means, easily answered questions.