The paradox of health and screening for prediabetes

In a classic article from the 1980s, Harvard psychiatrist Arthur Barsky described what he termed the "paradox of health": despite great improvements in many objective measures of health in America, people were actually becoming less satisfied with their personal health. Dr. Barsky suggested several explanations for this paradox, including "greater self-scrutiny and an amplified awareness of bodily symptoms and feelings of illness"; "a climate of apprehension, insecurity, and alarm about disease"; and "the progressive medicalization of daily life." He was especially critical of what he saw as the increasing commercialization of health services:

Each producer tries to convince the public that something is dangerously wrong, or about to go wrong, and that immediate steps must be taken to remedy the situation. An indifferent public is first galvanized into becoming a market of alarmed consumers by constant reminders of myriad threats to health, and then convinced of the need for more and more products and services to protect them. In the process, deep-seated fears about disease are mobilized, and insecurities about health and longevity are intensified. As a result, many come to feel less secure about their health, more worried about the possibility of disease, more absorbed in trivial symptoms.

Fast forward to this year, when a blog post from the U.S. Department of Health and Human Services advised, "It's never too early to prevent type 2 diabetes." Noting that women with high blood sugar levels in pregnancy (gestational diabetes) are more likely to develop type 2 diabetes later in life, this post suggested that mothers be tested for diabetes 6 to 12 weeks after the birth, then tested again periodically, with more frequent testing if their results showed prediabetes.

When medical and public health professionals talk about diabetes prevention, they are generally referring to helping people with prediabetes (high blood sugar levels that aren't high enough to meet diagnostic criteria for diabetes) improve their diets, become more physically active, and (if overweight) lose extra weight. In March, the Centers for Disease Control and Prevention and the American Medical Association announced the Prevent Diabetes: STAT campaign, with STAT standing for Screen, Test, Act-Today. Since few persons with prediabetes have recognizable symptoms, the reasoning goes, physicians need to find them by screening. But what is the evidence that screening for prediabetes makes a difference in long-term health outcomes?

Let's start with screening for diabetes. In 2008, the U.S. Preventive Services Task Force recommended that adults with blood pressures over 135/80 be screened for diabetes, not from direct evidence that screening was beneficial, but because at that time blood pressure goals were lower for diabetic patients. However, the 2014 JNC-8 hypertension guideline set the same goal blood pressure of 140/90 for adults younger than 60 years regardless of diabetes status, removing this rationale for screening.

So when the USPSTF updated their 2008 recommendations, they would need to find proof that earlier treatment of screen-detected diabetes (as opposed to clinically detected) prevented or delayed patients from going blind, losing limbs, requiring dialysis, having heart attacks, or dying as a result of their disease. The Task Force completed its review last November. The review found no evidence that screening for diabetes, even in high-risk persons, prevented or delayed any of the those outcomes. In fact, two randomized trials of screening for diabetes found no differences in mortality rates after 10 years of follow-up. These findings would typically have resulted in a "I" (insufficient evidence) statement or "D" (don't do) recommendation. Instead, the USPSTF provisionally recommended screening a much larger group of adults than before, reasoning that in the process of screening for diabetes, physicians would find lots of people with prediabetes and hopefully be able to prevent them from developing diabetes.

Huh?

Both the USPSTF review and a more recent review commissioned by the Community Preventive Services Task Force clearly showed that combined diet and physical activity promotion programs can slow (or sometimes stop) the progression from prediabetes to diabetes. No argument there. What we don't know is if finding more adults with prediabetes (and, since so many are thought to be undiagnosed, creating an epidemic of prediabetes) will help them live longer or better. Most persons with prediabetes are overweight, sedentary, and/or consume an unhealthy diet. Family physicians already provide lifestyle counseling to these patients. And if (like a few of my own patients with prediabetes) they happen to be normal-weight long-distance runners who haven't touched junk food in years, I just tell them to keep doing what they're doing.

Two editorials that accompanied publication of the USPSTF and Community Task Force reviews in Annals of Internal Medicine essentially argued that we don't need any more evidence to go out and screen everybody for prediabetes. Drs. K.M. Venkat Narayan and Mary Beth Weber wrote, "Much of the debate around screening for prediabetes and diabetes focuses on the lack of direct evidence from randomized, controlled trials comparing screened with unscreened persons on a hard outcome, such as CVD or mortality. However, such a definitive trial of hyperglycemia screening is infeasible, unrealistic, and arguably unethical ..."

Tell that to the authors of exactly this type of randomized trial which has been going on for more than a decade! Similarly, Dr. Ronald Ackermann argued, "As a society, we should no longer hold for ransom the potential to reduce the burden of diabetes by continuing to await further evidence that interventions will be cost-saving or prevent myocardial infarctions or deaths."

Well, while we're talking about hostage-taking, consider that being told that he or she has diabetes or prediabetes makes a patient immediately feel worse (or as Dr. Barsky would have put it, feel less secure about their health, more worried about the possibility of disease, [and] more absorbed in trivial symptoms). In my view, widespread screening for prediabetes will only become ethically justifiable when we have proof that it will lead to people living longer or better.

Financial conflicts of interest have no place on guideline panels

In a series of editorials in the New England Journal of Medicine earlier this year, Dr. Lisa Rosenbaum stirred controversy by suggesting that policing of pharmaceutical conflicts of interest in medicine had gone too far. These editorials and the passionate responses they elicited (especially those authored by three former NEJM editors and leaders of the Lown Institute) are worth reading in their entirety. I recently weighed in on this debate in a Medscape video commentary (the fifth in a continuing series) by discussing the narrower question of whether it is a problem if practice guidelines are developed by members with financial conflicts of interest. Below are two excerpts:

I strongly disagree with Dr Rosenbaum that preventing physicians with financial conflicts from developing guidelines or from creating continuing medical education materials will reduce their quality or usefulness. In fact, I believe the opposite to be true. Experts in medical subspecialties are not only more likely to be financially biased but intellectually biased as well. Is it any surprise that radiologists are up in arms about the US Preventive Services Task Force's conclusion that screening mammography in women in their forties should be optional or that urologists are opposed to the Task Force's recommendation against prostate-specific antigen screening in men? ... As the Institute of Medicine has affirmed, the primary qualification for membership on a trustworthy guideline panel should be expertise in evidence-based medicine. ...

The VA/DoD [cholesterol treatment] guideline suggested higher 10-year cardiovascular risk thresholds [than the ACC/AHA guideline] for recommending a statin for primary prevention (12% vs 7.5%) and for considering a statin for primary prevention (6% vs 5%). They also preferred moderate-dose to high-dose statins in most clinical situations. Why is the VA/DoD guideline more conservative about statin use? Same evidence, different panel, different conclusions. It's worth noting that none of the VA/DoD panelists had any financial conflicts of interest.

How to make sure patients understand health information

In the current issue of American Family Physician, Dr. Barry Weiss shares an anecdote about a hospitalized patient's confusion about a conversation between himself and a consultant who said that it was okay to move her from the intensive care unit to the floor. The perplexed patient interpreted "floor" literally to mean that the hospital was so overcrowded that she would not be able to sleep in a bed! Although this mildly comical story might elicit chuckles at health professional gatherings, it also highlights the serious potential for limited health literacy to lead to misunderstandings between doctors and patients.

Health literacy encompasses essential skills that patients need to access health services, understand and apply health information, and make good decisions: reading, writing, numeracy, and communication. A large body of evidence demonstrates strong associations between low health literacy and poorer health outcomes; compared to patients with high health literacy, patients with low literacy have more hospitalizations, more emergency department visits, and are less likely to receive appropriate preventive and chronic care services. According to a clinical review by Dr. Lauren Hersh and colleagues, "More than one-third of U.S. adults, an estimated 80 million persons, have limited health literacy, making it more difficult for them to read, understand, and apply health information. ... Although U.S. adults on average read at an eighth-grade level, more than 75% of patient education materials are written at a high school or college reading level." Limited health literacy is more common in patients age 65 years and older and in minority populations.

How can physicians make sure their patients understand health information? Rather than routinely screening for low health literacy, which has not been shown to improve outcomes, Dr. Hersh and colleagues recommend taking "universal health literacy precautions," a group of strategies for enhancing verbal and written communication, and visual aids (including online videos). For example, clinicians can use "teach back" to ask patients to restate the medical concept or plan in their own words to confirm understanding. Online tools are available to quickly assess the reading level of written patient materials. Patients also benefit from being encouraged to participate actively in formulating their care plans and being connected to community literacy resources, if needed.

For more information on how to bridge the health literacy gap, see this recent Family Practice Management article and the Agency for Healthcare Research and Quality's Health Literacy Universal Precautions Toolkit.

**

This post first appeared on the AFP Community Blog.

Countering a too-neat narrative on screening mammography

The Men Against Breast Cancer Fund and several cancer advocacy and physician organizations have posted a petition on Change.org asking Congress to place a two-year moratorium on finalizing the U.S. Preventive Services Task Force's draft recommendations on screening for breast cancer to forestall the possibility that health insurers will stop covering screening mammograms for women younger than age 50. Since 2010, the Affordable Care Act has mandated that private insurers provide "first-dollar coverage" (no copayments or deductibles) for all preventive services that the USPSTF assigns an "A" or "B" grade - services for which, in Task Force lingo, there is moderate or high certainty of at least moderate net benefit on the population level.

Screening mammography for women age 50 to 74 received a "B" grade the last time the Task Force reviewed the evidence, but it was their "C" grade for women age 40 to 49 that became a lightning rod for health reform opponents. Even though that grade aligned with the small health benefits in this age group and was intended to allow women to make their own decisions about what is, scientifically speaking, a close call, a "C" grade did not guarantee coverage for mammography. So Congress rewrote the ACA to disregard the 2009 recommendations and instead use the most recent version otherwise, dating from 2002 when a previous Task Force had given a "B" grade to mammography starting at age 40. But should the 2015 draft recommendations be finalized, the "C" grade will finally be enacted into law.

That's the backstory. The petition itself is riddled with misrepresentations and errors. For example, it states: "The draft recommendations give mammograms for women ages 40-49 a “C” grade, meaning most women in this age group, according to the Task Force, do not need an annual exam." Not so. Actually, the Task Force is telling physicians and patients to weigh the benefits (a 1 in 2000 chance over 10 years of avoiding death from breast cancer) against the harms (a 6 in 10 chance of a false positive test result, unnecessary biopsy, and/or unnecessary treatment during the same time period).

The petition says: "Of the approximately 40,000 women who die from breast cancer each year, up to 10,000 had cancer that potentially could have been diagnosed with screening mammography prior to the age of 50." I don't know the source of this statistic, but let's say for the sake of argument that it's true. It implies, falsely, that by not undergoing mammography in their 40s, 10,000 American women will die preventable deaths. Unfortunately, a cancer that can be diagnosed (no matter how "early") with screening mammography is not a cancer that can be successfully treated. Even under ideal scientific study conditions with perfect followup, screening mammography reduces mortality in younger women by no more than 20 percent. In real-life settings? A study published recently in JAMA Internal Medicine found that more intensive mammography is associated with the detection of more small breast cancers, but is not associated with fewer large breast cancers or breast cancer deaths.

Finally: "The draft recommendations also deem the evidence for 3D mammography insufficient, dismissing hundreds of peer-reviewed publications that clearly and consistently show 3D mammography is a more accurate exam." Accurate in what way? That it finds more cancers than conventional mammography does? That it leads to more biopsies, more radiation, more surgery? The Task Force contends (and I agree) that the only reliable measure of a screening test's superiority is whether or not it leads to fewer deaths. For 3D mammography, there's absolutely no proof that it does.

I recognize that for women or loved ones of women who believe their lives to have been saved by mammography, no amount of scientific evidence that I or anyone else can marshal will change their minds. As Dr. Louise Aronson wrote this week in JAMA:

As journalists, marketers, fundraisers, and politicians know all too well, the best strategy for education and persuasion in the public arena is usually a story that shows rather than tells how the salient facts are relevant in a particular life. If a person’s sister (wife, daughter, lover, friend) has breast cancer, and that cancer was found by a mammogram, they are not interested in populations. He or she is only interested in the sister. In that context, science seems cold, clinical, and uncaring, perhaps even irrelevant.

So how can we counter the prevailing narrative of the Task Force as a group of cold-hearted scientists who are more concerned about population-level data than the individual lives of the women we love? We can tell the human story of the guideline developers - half of whom are women over the age of 40 who have personally faced the mammography decision at some point themselves - but we can do much more than that. We can tell a representative story of the hundreds of thousands (or millions, perhaps) of women who experienced serious emotional or physical harm as a result of screening mammography, as Dr. Aronson did:

I would begin by saying that Elizabeth ... had a good life and loving family. And then, at her primary care physician’s recommendation, she had her first mammogram, which was, as we like to say in medicine, “grossly abnormal.” She was called back initially for a diagnostic mammogram and then for an ultrasound with biopsy. ... Over the next three weeks, she had several more mammograms and ultrasounds and 2 MRIs suggestive of a large, aggressive tumor; findings in the other breast that required investigation (“Probably nothing,” she was told, “but once we’ve seen it … ”); and 22 biopsies, several of which resulted in painful collections of clotted blood in her breasts.

Normal life and work were impossible. Elizabeth and her family endured a Thanksgiving punctuated by tears and carefully chosen words so as not to increase the already-suspicious children’s concern. Meanwhile, her physicians were at war: based on the x-ray films, the radiologists argued she had metastatic cancer with a less than 50% chance of 5-year survival, while her surgeons, based on the biopsy pathology, contended she had a rare, mostly benign condition. Fortunately, the surgeons were right. Still, sorting that out took weeks, and because the condition was associated with increased cancer risk, they insisted on bilateral surgery to remove all of the suspicious areas. So Elizabeth’s mammogram didn’t find cancer, but it did lead to the permanent mutilation of her breasts, huge medical bill copays, significant lost time from work, months of extreme stress, and ongoing anxiety about her disfigurement and risk of cancer.

To those who say, as if to brook no further argument: "Screening mammography in young women saves lives, that's all there is to it," I respond, no, that is not all. That too-neat narrative is only a small part of the story. And it's past time that the full story is told.

Decontaminating the medical literature

To ensure that its clinical review articles reflect current medical literature, the American Family Physician journal (of which I am an associate deputy editor) requires prospective authors to consult evidence-based resources that synthesize the best available evidence from clinical trials and other high-quality studies. The goal of this process is to produce unbiased recommendations for primary care physicians. But what if the authors of clinical reviews are actually professional scientific writers paid by pharmaceutical companies, rather than the physicians who are named as authors?

In fact, drug-company funded "ghostwriters" have been publishing articles in the medical literature for years. A 2011 study found that from 2 to 11 percent of articles published in 2008 in six major journals were actually written by people who were not named as authors. Although the study could not establish that these ghostwriters had been directly financed by industry, the practice of writing up a scientific study and then recruiting a lead author (usually an academic physician under pressure to "publish or perish") has been well-documented in the case of previous "blockbuster" drugs that were taken by millions of patients for common conditions but later turned out to have dangerous or fatal side effects, including Prempro and Vioxx.

Ghostwriting is not the only way that the pharmaceutical industry is able to influence the interpretation of evidence in its favor. As a Letter to the Editor pointed out, a 2005 Cochrane Review on medications for diabetic neuropathic pain unintentionally exaggerated the effectiveness of gabapentin in treating this condition due to the manufacturer's selective publication of favorable trials and suppression of unfavorable ones. (Antivirals for influenza present a similar problem.) In an accompanying editorial, Drs. Adriane Fugh-Berman and Jay Siwek reviewed these and other "stealth marketing" tactics that contaminate the medical literature, along with ways that readers can help identify and correct these biases:

Distorted information, once ensconced in the medical literature, is propagated by industry and by well-intentioned authors who unwittingly cite these studies. The medical literature is a permanent record that scientists and physicians rely on for decisions that ultimately affect patient care. Although the scientific process is never linear, the self-correcting process by which evidence is continually refined can be corrupted by the infiltration of medical journals with research studies and review articles distorted by a hidden marketing agenda.

Although there is no foolproof way for readers to detect undue industry influence, readers should be alert for marketing messages that disparage older, generically available drugs or that position newer branded (or upcoming) drugs as more effective, more convenient, safer, or filling an unmet need. The last sentence of the abstract is typically where the marketing spin is inserted. Readers should alert medical journals to suspicious articles by writing letters to the editor.

**

This post originally appeared on Common Sense Family Doctor under the title "Compromising the medical literature" on September 8, 2011.

CT scan after unprovoked blood clot is too much medicine

The diagnosis and and treatment of patients with deep venous thrombosis and pulmonary embolism is a common task in family medicine. In many instances, venous thromboembolism (VTE) is "provoked" by one or more predisposing risk factors, such as prolonged immobility, major surgery, pregnancy, or thrombophilia. In patients with a first episode of VTE without any known risk factors, one question that arises is how aggressively to test for a possible occult cancer, which may be present in up to 10 percent of cases. A Cochrane review of two small studies found inconclusive evidence that extensive testing for cancer improved cancer-related mortality compared to limited testing, although one study suggested that extensive testing advanced the time of cancer diagnosis.

A multicenter, randomized, controlled trial of 854 adults (mean age 54 years) with unprovoked VTE published in the New England Journal of Medicine provided more evidence to guide clinical decision-making. The control group underwent basic blood testing, chest radiography, and screening for breast, cervical, and prostate cancer; the intervention group received these tests and a computed tomography (CT) scan of the abdomen and pelvis. (25 percent of both groups had already received CT pulmonary angiography to diagnose a pulmonary embolism.) The primary outcome was confirmed cancers detected after one year of follow-up that were not diagnosed by the initial screening strategy. The trial found no statistical differences between the two groups in the primary outcome (missed cancers), mean time to a cancer diagnosis, or cancer-related mortality.

Since a CT of the abdomen and pelvis exposes a patient to a considerable dose of radiation and greatly increases the risk of detecting an incidentaloma, the harms of routinely providing this test to search for occult cancers in patients with unexplained VTE clearly outweigh the benefits.

**

This post first appeared on the AFP Community Blog.

Book Review: "Epic Measures" and evidence-driven health workforce policy

Last month, I attended a talk by Jeremy N. Smith, whose book Epic Measures follows Dr. Christopher Murray's lifelong quest to improve measurement of global health problems, which culminated in the Global Burden of Disease study. Thanks to Tracy Kidder, whose 2003 bestseller Mountains Beyond Mountains made Murray's medical school and residency classmate Paul Farmer a minor celebrity, most people's idea of global health work is setting up clinics in impoverished countries such as Haiti and Rwanda. In contrast, after Epic Measures opens with a young Murray traveling with his family across the Sahara Desert, it mostly describes more prosaic activities: meetings with statisticians, navigating entrenched national and global health bureaucracies, and appealing to philanthropists for research support. What I loved about this book, I told Smith as he was signing my copy, is that it made sitting in an office cubicle performing complex statistical analyses seem almost as cool as practicing front-line medicine in remote areas of the world.

In his talk, Smith focused on three questions that drove Murray's research and the narrative arc of Epic Measures, beginning with: "What is the scientific evidence base for health policy?" Before Murray arrived on the scene, there was literally no one systematically measuring the overall global burden of death and disability. The World Health Organization was divided into mini-fiefdoms, each possessing its own set of disease-specific estimates and intent on defending its turf and budget share. Murray recognized the scope of this problem when he added up all of the disease-specific mortality estimates produced by various divisions of the WHO and found that the sum (30 million) was 10 million higher the United Nation's estimate of total child deaths in the same year! Non-governmental organizations were similarly haphazard in their approaches to global health problems, which received funding commensurate to the strength of political advocacy rather than an impartial assessment of their effects on specific populations. (Even today, as Murray's team recently illustrated in JAMA, the proportional match between relative disease burden and financial assistance to developing countries is imperfect, but at least the shortfalls are more transparent.)

Smith's second question: "Are we measuring the right thing?" Many chronic diseases cause enormous suffering, but may not result in death. Accounting for the amount of disability it causes over a lifetime to millions of adults, the burden of neck pain is twice as high as breast cancer in the U.S. This isn't to say that breast cancer is an unimportant health issue, but the countless walkathons, pink ribbons, and calls to raise money for breast cancer research and treatment far outpace the attention and dollars public and private sources devote to relieving neck pain.

Finally: "If we use new evidence and new measures, how far and how fast can we improve?" Quite dramatically, Epic Measures illustrated. Mexico prioritized coverage decisions for its national health insurance plan, Seguro Popular, on its national burden of disease data and saw its child mortality rate fall by almost half within a decade. Based on its burden of disease, Australia started paying for short-term depression therapy and dropped routine prostate cancer screening (a no-brainer in retrospect, given that screening has little to no effect on prostate cancer morbidity or mortality, but a hard call to make before 2011).

Smith's talk got me thinking along the same lines about the state of the health care workforce in the U.S., and how far we are from what is needed based on our national burden of disease. There isn't an easy fix when so much of our health care "system" is driven by disease-specific advocacy and perverse financial incentives to do more rather than less, and we spend more than $1 trillion each year on health professional salaries alone. But rather than simply projecting future workforce needs based on past experience or best guesses about evolving models of care in the post-ACA era, we should be asking some more basic questions. What is the scientific evidence base for U.S. health workforce policy? Are we measuring the right thing? And if we use new evidence and new measures, how far and how fast can we improve?