Tuesday, April 13, 2021

Premature adoption of 3D mammography threatens study that aims to prove its value

Although digital breast tomosynthesis (DBT; 3D mammography) was approved by the U.S. Food and Drug Administration a decade ago and has since been rapidly adopted by breast imaging centers, no studies have shown that it is more beneficial or less harmful for breast cancer screening than traditional digital mammography. In a Diagnostic Tests review in the April 1 issue of American Family Physician, Drs. Kathleen Barry and Chelsea Evans noted that DBT offers a "modestly increased cancer detection rate" and lower recall rate, but also costs about 40% more per test and exposes patients to a higher dose of radiation. It is unclear if additional cancers detected by DBT would have eventually become symptomatic, and "no studies have evaluated mortality as an outcome in women screened with DBT compared with digital mammography."

To address these important questions, in 2017 the U.S. National Cancer Institute (NCI) and the Canadian Cancer Trials Group launched a $100 million randomized trial, TMIST (Tomosynthesis Mammographic Imaging Screening Trial). With a planned enrollment of 165,000 women at 100 North American clinical sites, TMIST was designed to compare the incidence of advanced breast cancer after 4.5 years of follow up in women receiving either digital mammography or DBT. However, by early 2020, investigators had managed to enroll fewer than 23,000 women and were forced to expand to include overseas sites in Asia and Europe. The reason, according to an article in Medscape: already convinced that DBT was a superior technology, large numbers of U.S. and Canadian radiologists were declining to participate. Then the COVID-19 pandemic hit, further hindering recruitment efforts.

Meanwhile, a national study of Breast Cancer Surveillance Consortium sites found substantial racial and ethnic disparities in DBT access, with Black, Asian, and Hispanic women significantly less likely than White women to be screened at a facility with DBT or to receive DBT if the facility offered both DBT and digital mammography. If DBT is truly superior to digital mammography, this apparent disadvantage may worsen existing disparities in breast cancer outcomes. If it isn't, then widespread DBT use is unnecessary and wasteful.

After the NCI director suggested in the fall of 2020 that the "feasibility and relevance" of TMIST was in jeopardy, a working group was formed to reevaluate the trial. In its report released last month, the group recommended that the trial continue, but with protocol revisions that included reducing the sample size to 102,000 and developing specific targets for enrolling women from racial and ethnic minority groups. Unlike European trials that are also evaluating DBT, the group noted, TMIST is the only study that is representative of the U.S. population, includes women younger than age 50, and includes multiple rounds of screening.

In a 2019 commentary, Drs. Joy Melnikow and Joshua Fenton observed of DBT:

Diffusion of medical technology ahead of definitive evidence is common in the United States. ... Societal attitudes that place high value on innovation and technology create a fertile environment for the rapid adoption of novel but unproven interventions. ... When the evidence from randomized clinical trials catches up, interventions shown to add little value to previous approaches are often already embedded in practice, widely covered by health insurance (sometimes by mandate), and difficult to withdraw.

Initial findings from TMIST will be available in 2027 at the earliest if investigators can reach its lower enrollment goal. As radiology facilities across the country continue to "upgrade" from digital mammography to DBT and more states mandate insurance coverage of the newer but unproven technology, I'm not optimistic that this important study's findings will have beneficial effects on clinical practice.


This post first appeared on the AFP Community Blog.

Thursday, April 8, 2021


I have started asking every patient I see, for whatever reason, if they have received or intend to receive one of the three available U.S. vaccines against COVID-19. In less than two weeks, every adult in every state will be eligible to get the shots, and in the next few months the Pfizer vaccine should become available to children as young as age 12. There is no question in my mind that getting vaccinated as soon as possible is the right choice to protect my patients and their loved ones and communities from the ravages of the virus. But like most primary care physicians, I don't have access to a supply of vaccines for use in my office - instead, I must direct patients to a website to sign up for an appointment to get it at another time, somewhere else. And for vaccine-hesitant patients or those who for whatever reason are unable to find the time to do this, this obstacle may leave many unvaccinated who (at least when they saw me) were perfectly willing to get the shot. In a recent Medscape commentary, I argued why "it's time to hand the [vaccine supply] baton to primary care for the final leg of the race to end this pandemic."

I ran track relays in high school: the 4 X 400 meters and 4 X 800 meters. The latter distance is long enough that baton-passing technique makes little difference in the final result, but in the shorter relay (just short of a mile in total) it matters. We runners on deck were taught two techniques: accepting the baton from a standing start or a running start. A standing start usually ensures a clean exchange, but the runner then loses time by being unable to accelerate until the baton is in hand. A running start ensures that momentum transfers smoothly from one runner to the next, but if not executed precisely, it can result in a botched handoff (either a dropped baton or disqualifying exchange outside of the legal zone), dooming the team's hopes. The running start also requires a lot more practice to get right.

What I and family physician colleagues are seeing across the U.S. right now, as primary care practices are gradually becoming involved in vaccination efforts in Maryland and other states, is handoffs from a standing start. Even though many practices went to great lengths to identify and reach out to their patients at the highest risk of complications in anticipation of being able to administer vaccines (running start), the belated recognition by the federal government and state health departments of their critical role in reaching more reluctant patients has forced them to halt these efforts while they wait for adequate supplies to trickle in. This is a huge lost opportunity, and I fear that this lack of coordination between public health and primary care will result in unnecessarily prolonging the pandemic.

Monday, March 29, 2021

Increasing clarity about benefits and harms of screening for diabetes

Affecting about 6 percent of pregnancies in the United States, gestational diabetes increases risks of preeclampsia, shoulder dystocia, and macrosomia, and is associated with a 10-fold greater risk of developing type 2 diabetes mellitus in later life. In a recent draft statement, the U.S. Preventive Services Task Force (USPSTF) affirmed its previous recommendation to screen pregnant persons for gestational diabetes at or after 24 weeks of gestation. Historically, there have been two screening options: a non-fasting 50 gram oral glucose challenge test followed by a fasting 100 gram glucose tolerance test if the first test result exceeds a threshold value (typically 130-140 mg/dL), or a single fasting 75 gram glucose tolerance test. Although two-step screening is more commonly used in the U.S., until recently the comparative outcomes of these approaches were uncertain. This evidence gap is important because the diagnosis is associated with increased psychological and emotional burden; labeling more persons as having gestational diabetes with the one-step screening approach would only be justified if doing so resulted in better pregnancy outcomes than the two-step approach.

A pragmatic, randomized trial recently compared the one-step and two-step approaches in more than 23,000 women who received prenatal care at Kaiser Permanente Northwest and Kaiser Permanente Hawaii. Researchers evaluated five primary outcomes: "diagnosis of gestational diabetes, large-for-gestational-age infants, a perinatal composite outcome (stillbirth, neonatal death, shoulder dystocia, bone fracture, or any arm or hand nerve palsy related to birth injury), gestational hypertension or preeclampsia, and primary cesarean section." As expected, a diagnosis of gestational diabetes was more common in participants who underwent one-step screening (16.5%) compared to the two-step approach (8.5%). However, intention-to-treat analyses found no statistically significant differences in any perinatal or maternal complications. Although the trial was not designed to measure potential long-term benefits of post-pregnancy risk-reduction strategies to prevent type 2 diabetes, the results suggest that the two-step approach produces equivalent benefits, and fewer harms, than the one-step approach.

The USPSTF is also updating its recommendation on screening for prediabetes and type 2 diabetes in nonpregnant adults. Compared to its 2015 statement, which recommends screening overweight or obese adults between the ages of 40 and 70, the updated draft statement lowers the age range to include persons aged 35 to 39 years. Although the focus of the Task Force's old and new diabetes screening guidelines is identifying persons with prediabetes in order to prevent them from developing diabetes and its complications, the utility of the term "prediabetes" is controversial, as Dr. Jennifer Middleton discussed in a previous AFP Community Blog post. In older adults, prediabetes is extremely common. In a prospective cohort study of community-dwelling adults aged 71 to 90 years, 73 percent met one or both of the diagnostic criteria for prediabetes (hemoglobin A1c level of 5.7% to 6.4%, impaired fasting glucose of 100-125 mg/dL). After 6.5 years of follow-up, persons with prediabetes at baseline were substantially more likely to revert to normoglycemia or to die than to progress to diabetes. Based on these findings, stopping diabetes screening after age 70 will avoid overdiagnosis and unnecessary treatment.


This post first appeared on the AFP Community Blog.

Sunday, March 14, 2021

Can Choosing (more) Wisely prevent chronic low back pain?

More than a dozen Choosing Wisely campaign recommendations concern what not to do for patients with acute low back pain. Based on a clinical practice guideline from the American College of Physicians, the American Society of Anesthesiologists and several other groups recommend avoiding imaging studies within the first six weeks in patients without red flags or specific clinical indications. The American Academy of Orthopedic Surgeons and the American Academy of Physical Medicine and Rehabilitation (AAPMR) advise avoiding opioids for these patients unless other alternatives have not provided pain relief. The AAPMR and the North American Spine Society discourage bed rest as treatment for acute low back pain. These recommendations are intended to reduce downstream harms and costs: for example, a spurious finding on MRI could lead to unnecessary surgery; use of opioids could lead to physical dependence and opioid use disorder; bed rest and avoidance of physical activity could increase the risk of long-term disability.

Can guideline-discordant care for patients with acute low back pain increase the risk of progression to chronic low back pain? In an inception cohort study published in JAMA Network Open, researchers enrolled 5233 adults with acute low back pain from 77 U.S. primary care practices, assessed their baseline risk of transition to chronic pain using the Subgroups for Targeted Treatment (STarT) Back prognostic tool, and followed them for 6 months. 32% of participants met clinical criteria for chronic low back pain at the study's end. Characteristics associated with transition to chronic pain included obesity (adjusted odds ratio, 1.52), tobacco use (aOR, 1.56), severe baseline disability (aOR, 1.82), and a depression and/or anxiety diagnosis (aOR, 1.66). Researchers also examined associations between chronic low back pain and inappropriate care processes within 21 days of the initial visit: 1) any opioid prescriptions, or benzodiazepines or systemic steroids prescribed without an NSAID or skeletal muscle relaxant; 2) diagnostic imaging; 3) medical subspecialty referral for back pain. Compared to the 52% of participants who received none of these, patients with 1, 2, or 3 inappropriate care processes were 1.39, 1.88, and 2.16 times more likely to develop chronic low back pain after controlling for clinical characteristics.

Although adherence to Choosing Wisely recommendations was associated with a lower risk of patients developing chronic low back pain in this study, it is disappointing that almost half of them received at least some inappropriate care. Is care for back pain an outlier, or does it reflect national trends? A recent cross-sectional study examined the use of 32 low-value health services in Medicare fee-for-service beneficiaries. The study found modest progress from 2014 to 2018 in the percentage of persons receiving any low-value service (declined from 36.3% to 33.6%), number of low-value services per 1000 persons (declined from 678 to 633), and spending per 1000 persons on low-value care (declined from $52,766 to $46,922). Three services comprised about two-thirds of low-value care: preoperative laboratory testing, opioids for back pain, and antibiotics for upper respiratory infections. While preoperative testing decreased during the study period, opioid and antibiotic prescribing both increased.

Since the campaign's inception in 2012, Choosing Wisely recommendations have been widely disseminated in American Family Physician (AFP), FPM (formerly Family Practice Management), and other family medicine journals. The American Academy of Family Physicians periodically updates and adds new "don't do" recommendations, most recently in 2018. Other studies have recognized that clinicians in underserved or "safety net" practices are as likely to provide low-value care as clinicians in better resourced settings. A 2018 AFP editorial by Dr. Jennifer Middleton recognized that increasing awareness of best practices is necessary but not sufficient to drive implementation: "For meaningful change to occur, the workflows and systems we operate within must change so that new habits become routine." The Medicare study suggests that workflows and systems have not changed enough in the past decade to undo entrenched low-value practices.


This post first appeared on the AFP Community Blog.

Thursday, March 4, 2021

Misinformation can be as contagious and harmful as COVID-19

At the end of August 2020, when around 1,000 Americans were dying from COVID-19 each day, I posted on my personal Facebook account a straightforward translation of the deaths statistic: averaged over 24 hours (86,400 seconds), one U.S. resident was dying every 86 seconds. In another post, I expressed my concern for the mental health and safety of hundreds of public health officials all over the country who were being threatened and harassed for recommending or enforcing measures aimed at stemming these horrifying losses, such as restricting restaurant capacity, banning large gatherings, and mandating mask-wearing when social distancing wasn't possible.

It didn't occur to me that either of these posts would be the least bit controversial among my friends and family. But later that night, a friend from college whom I've known for more than a quarter of a century commented that I was greatly exaggerating the death toll and "fear mongering." Do the math, I responded. He countered by claiming that doctors were inflating COVID-19 death counts for financial benefit and that just because thousands were dying with COVID, they weren't necessarily dying of COVID. The flu kills 100,000 people every year, he wrote, and we don't shut down the country for that. I gently pointed out that according to official CDC statistics, no more than 60,000 people have died in a single influenza season for the past decade, and those numbers are estimates, not actual case counts. We went back and forth like this for a while. Finally, exasperated, I observed that my friend is a lawyer, while I have medical and public health degrees and was probably a little bit more informed than he was. "Elitist," he taunted, your fancy education doesn't mean anything. The great thing about social media is that it's the ultimate equalizer - no one needs to "bow down" to your supposed expertise. By the way, he wrote, right before I wished him well and blocked him from posting more misinformation on my page, those public health officials you support all march in lockstep with the Democratic Party, so as for the harassment, well, they had it coming.

Six months later, more than half a million Americans have died from the pandemic, and though new infections have fallen from their post-holiday peak, we're still losing about 2,000 people every day (or one every 43 seconds). Despite the greatly increased pace of vaccination since President Biden took office, as of today less than 10 percent of the U.S. population is fully vaccinated, and less than 20 percent has received at least one dose of the Pfizer or Moderna two-dose vaccines (the Johnson & Johnson vaccine is a single dose). Nonetheless, the misinformed governors of Texas and Mississippi have declared victory, discarding their mask mandates and restrictions on businesses and sending the message that if COVID-19 was ever a serious public health problem, it isn't any longer.

Of course, online misinformation abounded well before COVID, infamously leading many parents to refuse the measles, mumps, and rubella vaccine for their children because they feared, based on a single fraudulent, discredited and retracted scientific article, that it might cause autism. And I'm deeply concerned that the crazy myths now proliferating online about COVID vaccines - that they make changes to your DNA, they cause infertility, they contain tiny microchips to allow Bill Gates can track your movements, that their side effects are worse than the disease - pose a huge obstacle to achieving herd immunity through vaccinating 70 percent or more of the population. (And don't get me started on the barbaric "let's achieve herd immunity through natural infection." One might as well go back to the days of bloodletting and Hippocrates' four humors, for the immense harms that strategy would cause.) Let me be clear: I don't think that vaccination should be mandatory for every adult (or child, when it is shown to be safe and effective in children), but those who decline it for themselves or on someone else's behalf should do so because they have concerns that medicine can't yet answer (e.g., are there any side effects that don't show up for a year or more?), not because they fall for misinformation or conspiracy theories.

Monday, February 22, 2021

Are physicians overdiagnosing melanoma?

In an editorial in the February 1 issue of American Family Physician, Dr. Jenny Doust and colleagues wrote about the problem of widening disease definitions, a common phenomenon in which the definition of a disease is "broadened over time to include milder and earlier cases," leading to harm "by exposing more patients to the adverse effects of treatments, triggering investigation and prescribing cascades, increasing anxiety, and placing a financial burden on patients and the wider society." Expanding the number of patients diagnosed with disease increases the burden on primary care physicians called on to manage these additional cases, even when it is uncertain if earlier interventions prevent morbidity or mortality. Illustrative examples of wider disease definitions include hypertension, polycystic ovary syndrome, breast cancer, and autism. What can family physicians do about it? The authors responded:

Recognizing the problem is the first step in tackling it. In particular, family physicians should not blindly accept new definitions and testing guidelines without an adequate understanding of the harms and benefits of the changes and the implications for our patients and wider practice.

Along similar lines, a recent analysis in the New England Journal of Medicine by Dr. H. Gilbert Welch and colleagues examined the drivers of the dramatically increased incidence of cutaneous melanoma in the U.S., which today is 6 times as high as in 1975 despite essentially no change in melanoma mortality. They pointed out that exposure to ultraviolent (UV) radiation (including tanning bed use) cannot account for more than a small portion of this increase. Instead, they argued that increased diagnostic scrutiny - "the combined effect of more screening skin examinations, falling clinical thresholds to biopsy pigmented lesions, and falling pathological thresholds to label the morphologic changes as cancer" - is most likely to be responsible for the epidemic of new diagnoses. Not only has the annual percentage of fee-for-service Medicare beneficiaries undergoing skin biopsies nearly doubled since 2004, but pathologists frequently upgraded skin biopsy specimens obtained in the late 1980s from benign to malignant when evaluating the same specimen two decades later. Primary care physicians contribute to widening the definition of cutaneous melanoma by performing or referring for biopsy small (<6 mm), incidentally detected skin lesions and screening patients with dermoscopy, which identifies more melanomas than visual inspection alone but is not well studied in primary care settings.

The U.S. Preventive Services Task Force (USPSTF) has concluded that current evidence is insufficient to assess the balance of benefits and harms of skin cancer screening in asymptomatic adults. Nonetheless, more than half of family physicians and general internists in a 2011 survey reported performing full-body skin examinations for skin cancer screening. In a 2020 AFP editorial, Drs. Michael Pignone and Adewole Adamson (Dr. Adamson also co-authored the NEJM analysis) observed that "compared with usual care, potential effects of screening on morbidity and mortality from keratinocyte carcinoma are at most small, and screening cannot be justified based on the impact on keratinocyte carcinoma alone." Dr. Welch and colleagues went one step further, arguing that the established harms of skin cancer screening already outweigh any potential benefits:

The increase in melanoma diagnoses by a factor of 6, with at least an order of magnitude more persons undergoing a biopsy and no apparent effect on mortality, is more than enough to recommend against population-wide screening. ... It [screening] has been effectively promoted under the guise of public health, with the combination of frightening messages about skin cancer and the premise that screening can only help. However, medical care should be driven by patient needs, not system needs. Now is not the time to add more anxiety and expense to an already anxious and expensive world.

Not surprisingly, dermatologists have a more positive view of skin cancer screening, as reported in a news story about the analysis by Dr. Welch and colleagues that quoted the president of the American Academy of Dermatology as stating that "an aggressive approach to prevention and treatment is entirely appropriate for a disease that kills 20 Americans each day." Of course, no one is urging clinicians to stop counseling patients on minimizing their exposure to UV radiation; indeed, the USPSTF recommends behavioral counseling to prevent skin cancer, particularly for children, their parents, and young adults. But screening for skin cancer, which has effectively widened the definition of cutaneous melanoma and driven widespread overdiagnosis - is a different story. To give Dr. Doust and colleagues the last word: "We [primary care physicians] are not here to passively enact specialist recommendations. Instead, we need to more assertively act as advocates for our patients and our communities."


This post first appeared on the AFP Community Blog.