Research Translation Podcast

David Newman
  • 5.0 (5)
  • Science
  • Updated Weekly

Translating Medical and Health Research For All researchtranslation.substack.com

  1. 9h ago

    The Early Detection Cancer Test That Doesn’t Detect Early

    In college our physics teacher, Professor McGinnis, held an annual competition. Using a container made only of glue and toothpicks, we had to keep an egg from cracking when dropped out of the classroom’s first-story window. My team built a geometric lattice, buttressed with layers of internal protection. It was beautiful. For our (mandatory) test run, we dropped it empty, to be sure the container would survive and the egg compartment would hold. The container fell apart, and the compartment was gone. So we redesigned. Again the empty drop failed. But the lattice was so pretty. Perhaps the weight of a settled egg would strengthen the wood, we hoped. And with no time for a loaded test, we prepared for Egg Drop Friday. Suzanne, our dropper, had magic hands. With a quick lift and a soft release, our soccer-ball-like structure seemed to fall slower than gravity should allow. All morning, cheers had erupted at each impact. But when our dazzler struck the ground, the crowd fell silent. Toothpicks everywhere. The egg was not cracked, it was obliterated. No piece of shell was larger than a coin. Professor McGinnis wondered aloud whether our structure had somehow funneled the energy straight into the egg. Then, shaking his head, he asked the question we all knew was coming. “What happened when you did the empty drop?” Last week the New England Journal of Medicine published NHS-Galleri, a massive trial that randomized more than 140,000 people in England to have, or not have, their blood run through a multi-cancer early detection (MCED) test once a year for three years. The trial was funded by the test’s manufacturer, Grail, and Grail employees are among the authors. Let’s start with what matters: The point of cancer screening is to keep people from dying. This paper doesn’t even report how many people died in each group, from cancer or from anything else. This is despite the fact that the trial enrolled 140,000 people, easily enough to have a count of cancer and overall mortality, Instead, they offer us a test run: did the early detection test detect cancer earlier? It is the empty egg drop. It failed. The primary outcome was whether annual testing reduced the number of people diagnosed with late-stage (stage III or IV) cancer. It didn’t, and it wasn’t a near miss. The screened people were found to have 706 late-stage cancer diagnoses and the unscreened people were diagnosed with 688. In other words there were MORE, LATER STAGE, cancers in the tested group. Grail’s press release, however, leads with something different. They tout “20% drop in stage IV cancers”! This was a secondary endpoint, and the result’s confidence interval brushes up against no effect at all. Moreover this reduction, they confess was “offset by a higher rate of stage III cancer diagnosis.” The test didn’t lift anyone out of late-stage disease. Why? The companion paper in Nature Medicine shows us. In years two and three, which is what a real annual screening program looks like, the test missed about 73% of cancers, and roughly half of the 12 deadly types it was designed to catch. Worse yet, about half its positive results were false alarms: 864 healthy, cancer-free people were told they might have cancer. The test’s specificity, around 99.5%, sounds superb. But it isn’t. When you screen millions of healthy people for rare diseases, even 99.5% means half of all positive results are wrong. Scale it to 100 million Americans and roughly 450,000 people a year would be told they might have cancer when they don’t, each launched into scans, scopes, and biopsies. And Grail tuned the test for that specificity. The price was sensitivity, which is why it missed MOST cancers. This isn’t a bug to fix in version 2.0. Small, early tumors shed very little DNA into the blood. Crank up the sensitivity to find them and you drown in false positives and overdiagnosis, the treatment of ‘cancers’ that would never have hurt anyone. Cancer is not one disease marching in a straight line from small and curable to big and deadly. Some kill no matter how early you find them. Some never would have mattered. The mythical sweet spot, where finding it early changes everything, has never been proven to exist. And if it does, we know this: It’s way smaller than the brochures suggest. But fortunately for the test manufacturers, and unfortunately for us, early detection is more than a strategy. It’s an ideology, and for many it’s even a faith. And faith can ‘succeed’ even when the empty drop fails. Which brings us to September 23, exactly one day after the NEJM paper went online. An FDA advisory committee met to consider Grail’s application and voted to recommend approving the test. Someone on that panel should have asked Professor McGinnis’s question: What happened when you did the empty drop? Get full access to Research Translation at researchtranslation.substack.com/subscribe

    The Early Detection Cancer Test That Doesn’t Detect Early

  2. 5d ago

    Deconstructing the Alcohol-Cancer Obsession

    Below is a summary of the audio transcript, written by Gemini AI with some heavy editing by me. Let me know if you like or hate these, I’m still experimenting! “STROLLERS THAT KILL! At eleven.” RT is reader-supported. To help, PLEASE become a paid subscriber. For the price of a cappucino once a month, you’ll keep us afloat. This is one of my all-time favorite teasers for the evening news. These days, it’s pretty typical for headlines and news teasers to threaten you or your children’s lives. Which is why it wasn’t so surprising to see recent headlines about how alcohol-linked cancer deaths have doubled—AND ONE DRINK A DAY COULD KILL YOU. Behind this sensationalism, however, lies a not-so-master class in statistical deception. To understand why these claims are absurd, we can look at how public health authorities manipulate observational data—and why a glass of wine with dinner is not exactly a death sentence. 1. Severe Alcoholism vs. Moderate Drinking First, let’s stipulate what everyone agrees on: chronic, severe alcoholism is bad. Heavy, years-long drinking leads to liver cirrhosis and hepatocellular carcinoma (often fatal liver cancer). It also contributes to violence and fatal accidents. Public health efforts should target severe addiction, drunk driving, and alcohol-related violence. But authorities and mass media are conflating severe alcoholism with moderate intake, and they’re doing it on purpose. Why? Because vilifying the glass of wine with dinner is a surefire path to more clicks and more attention. 2. The Observational Evidence Trap Every study claiming moderate drinking causes cancer is observational, in which researchers observe people over time and collect data including self-reported surveys. Here are a few classic problems with using observational data to make cause-and-effect claims. * Self-Selection & Confounders: People who drink and those who abstain carry vastly different cultural, socioeconomic, genetic, and lifestyle traits. In observational data, these background differences—called confounders—distort any comparisons that are, in theory, supposed to be about alcohol consumption. * The Trial Difference: In randomized trials, flipping a coin balances background traits equally between groups. Historically, observational ‘discoveries’ (like hormone replacement for post-menopausal women preventing heart disease, or high fiber diets preventing colon cancer) evaporated when tested in trials. * Underreporting: Patients famously underreport alcohol intake. When a study links “a drink a day” to a health outcome, it may really mean two to four drinks. Weak Relative Risks Are Not Causation To evaluate a health claim, you have to look at the relative risk. When scientists look at cigarette smoking and lung cancer, smokers face a staggering 2,000% to 4,000% increase in risk. Multiple severe sunburns double or triple the risk of melanoma (200% to 300%). Those are huge, indisputable signals. Compare that to moderate alcohol intake and breast cancer, where public health authorities point to a paper-thin 4% relative increase. A relative risk that low is not a cause-and-effect relationship—it is the classic residue of lifestyle confounding. Furthermore, public health agencies selectively report these tiny upward bumps while ignoring the other side of the ledger. The same studies show that moderate drinking is associated with a 15% lower rate of kidney cancer, thyroid cancer, and lymphomas. Neither direction represents true causation, but health agencies cherry-pick the upward noise, tipping their hand: This isn’t about truth, it’s about pushing an agenda. Absolute Risk Puts Fear in Perspective When agencies cite relative multipliers like ‘1.4-fold increases’, converting them to absolute numbers exposes how negligible the real-world risk is. For esophageal cancer, that relative bump translates to 1 additional case per 7,000 moderate drinkers over a decade. For oral cancer, it works out to 1 additional case per 700 moderate drinkers over ten years. To put those numbers in perspective, the average American has a 1 in 93 lifetime chance of dying in a car crash and a 1 in 2,482 chance of choking on food—making daily life far riskier than a decade of moderate wine consumption. In private admissions, public health researchers acknowledge that these numbers do not represent true cause-and-effect, but justify scaring moderate drinkers under the assumption that it might indirectly curb drunk driving or severe abuse. Deconstructing the “Doubling” Headline The recent headline claiming alcohol-attributable cancer deaths doubled between 1990 and 2023 comes from a Lancet secondary analysis of Global Burden of Disease data, citing a jump from roughly 11,000 deaths to 23,000. When you strip away the spin, the “doubling” vanishes. Ninety-two percent of that increase was driven entirely by population growth and the aging of the Baby Boomer generation into peak cancer years. In fact, per-capita alcohol consumption in the U.S. was exactly the same in 2023 as it as in 1990. And the small remaining statistical increase? It was driven almost entirely by liver cancer deaths—the majority of which are caused by Hepatitis C infections, not alcohol. Meanwhile, actual U.S. age-adjusted cancer mortality has dropped massively in the last three decades, thanks to improved treatments. The Bottom Line When public health modeling turns aging into scary headlines, it isn’t science—it’s spin. True causal relationships are strong, clear, and undisputed. When studies contradict each other and rely on 4% relative increases, there is no meaningful real-world risk. Public health agencies should address severe alcohol abuse and drunk driving directly, rather than using statistical ghosts to fearmonger moderate drinkers. Get full access to Research Translation at researchtranslation.substack.com/subscribe

    Deconstructing the Alcohol-Cancer Obsession

  3. Oct 1

    The Cardiologist Contorting To Compliment the STAREE Trial

    Below is a summary of the transcribed audio, brought to you by Gemini. I’m trying different AI options for these summaries, please feel free to let me know which ones you like. For an additional note answering one subscriber’s excellent question about statins, check out the audio. Enjoy! RT is 100% subscriber-supported. To help us endure and expand, become a paid subscriber! The New England Journal of Medicine recently published its print edition featuring the long-awaited STAREE trial, alongside an editorial authored by Dr. Martin Mortensen. STAREE set out to answer a fundamental, patient-centered question for millions of aging adults: Can statins help healthy people 70 and older live longer, preserve cognitive function, or prevent disability? To appreciate why STAREE matters, you have to understand that over 90% of the statin trial literature driving clinical guidelines originates from pharmaceutical companies. Drug manufacturers design studies around tightly controlled, easily measurable markers carefully engineered to secure regulatory approvals and support marketing claims. But a commercially viable, marketable endpoint does not automatically translate into a meaningful clinical benefit for a human being taking a daily pill. STAREE flipped that approach on its head. Publicly funded and conducted across Australian primary care practices with roughly 10,000 enrolled participants aged 70 and older, the trial asked a far more rigorous question. While the investigators measured traditional cardiovascular events, those outcomes were explicitly treated as an intermediary step. The true goal was disability-free survival—a hard composite of all-cause mortality, incident dementia, and physical disability. The trial tested whether reducing cardiovascular events translated into a tangible improvement in the duration and quality of older adults’ lives. The answer was a resounding, unambiguous “no.” STAREE demonstrated no reduction in death, no reduction in dementia, no reduction in disability, and no reduction in the composite combining all three. On the outcome it was designed to test, the trial was a complete failure. Enter the accompanying editorial. Despite these definitive findings, Dr. Mortensen summarizes STAREE as providing “compelling evidence for the benefit and safety of initiating statin therapy in relatively healthy older adults.” Arriving at an enthusiastic, green-light recommendation from a trial that completely missed its primary patient-centered endpoint requires a masterclass in rhetorical and analytical contortion. First, the editorial obscures the fact that the trial investigators moved the goalposts mid-study. Mortensen opens his analysis by focusing heavily on the trial’s cardiovascular composite endpoint, presenting it as a combination of cardiovascular death, nonfatal heart attack, stroke, or coronary revascularization. What readers are not told is that revascularization was not part of the original outcome. According to the original protocol and registry, the outcome was strictly limited to cardiovascular death, nonfatal heart attack, and stroke. Enrollment began in 2015. Revascularization was tacked on in 2024—nine years into the study—after investigators realized event rates were falling short of projections. So they moved the goal posts. Expanding an endpoint mid-trial to capture procedural numbers is critical context, but the editorial presents it as if it were the plan all along. Second, the editorial equates procedural interventions with true clinical preservation. The statistically significant reductions touted in the study were driven almost entirely by nonfatal heart attacks and coronary revascularizations. Cardiovascular mortality itself was not reduced in the slightest. Placing a stent in a stable coronary artery is a procedure that is subjectively chosen, and proven not to reduce heart attacks, strokes, or death. So how, pray tell, can it act as a surrogate for those outcomes? Third, when Mortensen directly confronts the trial’s central primary outcome—the complete lack of impact on disability-free survival—he labels it a “neutral finding that warrants a nuanced interpretation.” Calling the failure of a 10,000-person, multi-year trial’s main outcome ‘neutral’ is pure gas-lighting. Thousands of seniors participated precisely so we could learn whether statins preserve independent living. that failure isn’t an incidental or ‘neutral’ footnote—it is the core finding of the entire enterprise. Fourth, the editorial performs a complete logical inversion by reframing a lack of harm as a positive justification for intervention. Mortensen points out that statins did not statistically increase death or dementia, and declares victory. The drugs were supposed to reduce both, and failed. But Mortensen uses the absence of benefit to reassure clinicians about prescribing statins. Gas-lighting. Mortensen’s mandatory disclosures reveal active financial relationships with major pharmaceutical manufacturers Novartis and Eli Lilly. Once again, the Death Star selected an editorialist with commercial ties to companies that sell cholesterol-lowering drugs, to provide commentary on cholesterol-lowering trial results. While Mortensen may sincerely believe his conclusions serve public health, good intentions do not compensate for an unpersuasive, contorted interpretation. Mortensen concludes by advocating for informed, shared decision-making in practice—a principle on which we agree. However, genuine shared decision-making cannot happen when trial results are repackaged to hide their primary failures. Telling an older patient “your cholesterol is high, take this pill” is not shared decision-making. Honest consent requires informing patients that while statins may lower the risk of certain nonfatal and procedural events, large-scale public trials prove they do not help healthy older adults live longer, prevent dementia, or avoid disability—all while carrying established risks of diabetes and muscle problems. STAREE asked the right question, and came to the right answer. Perhaps the commentary should do the same. Get full access to Research Translation at researchtranslation.substack.com/subscribe

    The Cardiologist Contorting To Compliment the STAREE Trial

  4. Sep 25

    The Single Most Important Trick in Medical Research

    Today’s episode explains the broader theme that ran through our episode from earlier this week. I wasn’t sure if I was clear enough, so here it is. Since it was just a quick audio I had ChatGPT turn the transcript into a written piece for the many written word addicts in our group (respect!). Naturally, I couldn’t help but heavily edit. RT is 100% reader-supported, and we want to keep doing it. You can help ensure that, by becoming a paid subscriber. Thanks. Here’s a trick that, once you see it, is hard to unsee. It’s everywhere in research, and understanding it can change how you read a study in about ten seconds. The trick is called a surrogate, something researchers measure that stands in for the thing we actually care about. Let’s start with an easy one: Cholesterol. Nobody wants a lower cholesterol number for its own sake. You don’t frame your lab report and put it next to the grandkids. You want to live longer and live better. Which illustrates just one of the problems with surrogates: Sometimes they’re not good surrogates. For most people, lower cholesterol does not predict longevity or health. But that’s a story for another day. The important part here is that it’s not the true prize. When researchers do a study to prove they can lower cholesterol—which is common—the cholesterol is acting as a surrogate for what matters. Imagine I offered you a pill that raised your cholesterol into the thousands, but research showed it helped you live longer and better. You’d be interested, of course. Because lower cholesterol isn’t the prize. Your health is. Now, for a less obvious example: breast cancer mortality. This sounds like the real thing. Death! What could be more important? When people hear “fewer breast cancer deaths,” their brains instantly translate it to “fewer deaths.” But those aren’t the same. This is the central tension in debates about whether cancer screening saves lives. No randomized trial has ever found that mammography reduces deaths—but a few have found it reduces breast cancer deaths. How can that be? In my book I talk about an imaginary device called the HeartSaver 3000. Touch it to someone’s chest during a heart attack and—presto!—it eliminates their chance of dying from that heart attack. Unfortunately, it replaces that risk with an equal chance of dying from a stroke. In such a case heart attack deaths are reduced, but the total number of deaths is unchanged. Suppose mammography screening prevents some deaths due to breast cancer but also leads to extra, unnecessary biopsies and surgeries and chemotherapy, some of which cause deaths—perhaps enough to offset any lives saved. Or, suppose there’s ambiguity about which deaths are due to breast cancer. A woman treated for breast cancer dies from a blood clot. Is that a blood clot death or a cancer death? This kind of ambiguity can sway the numbers in one direction. Counting total deaths, rather than breast cancer deaths, avoids the labeling problem. It also keeps deaths caused by treatment in the calculation (where they should be). This is why I call disease-specific mortality (like breast cancer-, or heart attack mortality) a surrogate. It’s standing in for the true prize, longer life. Which brings us to the adult vaccines. The big trial of the pneumococcal vaccine found fewer pneumonias due to certain bacterial strains. But not fewer pneumonias (or deaths). And ten years later, CDC data confirm that pneumococcal disease rates have not changed with the introduction of the adult vaccine. In other words, the surrogate suggested a benefit that failed to appear. Yet, when we hear a study found “less pneumococcal pneumonia” our brains often drop the middle word. We hear ‘less pneumonia’—which the study never found. This week, along comes the new RSV trial, of more than half a million people. RSV-associated respiratory hospitalizations fell from 64 to 19. But neither hospitalizations nor deaths fell. At this point someone invariably says: perhaps the study just wasn’t large enough to detect the broader benefit. But it would probably be there if the study was bigger. Perhaps. But another possibility is that the surrogate failed. Perhaps harms offset gains, or blurry definitions swayed the result. Whatever the reason, history tells us we cannot assume a benefit in the true prize is hiding below the statistical radar when we see only a surrogate succeed. Meanwhile, the burdens need no such leap of faith. The appointment takes time. The shot costs money. Pain, swelling, fatigue, and serious effects like Guillan-Barré Syndrome all count. Whether it’s a vaccine, a surgery, or a pill, there’s no free lunch in medicine. So here’s your thirty-second research-reading habit: Find out what they actually measured. Then ask whether it’s the true prize. Is it a blood test? Or a type of pneumonia? Or death from one disease? Those are surrogates. Is it living longer and living better? These are prizes. Don’t let anyone sell you a better number, and call it better health. Get full access to Research Translation at researchtranslation.substack.com/subscribe

    The Single Most Important Trick in Medical Research

  5. Sep 22

    The New Adult RSV Vaccine, the Death Star, and Promises Broken

    Apologies to my paid subscribers, who are getting this for the second time. I sent it out as a paid-only post, which was a mistake. I will be having paid subscriber-only content fairly soon, but I do try to make most of my content available to everyone, so I’m re-sending to all. Sorry for the inbox clutter! In third grade Marc Diaz explained to me, very earnestly, that if one’s hand was larger than their face, that person was technically deformed. Concerned, I held my palm to my face, carefully comparing. At which point he cracked the back of my hand with his own palm, sending my hand into my face and me toppling backwards out of my chair. Good one, I thought, ready to line it up on the next sucker. But he wasn’t finished. “Sorry sorry,” he said “I had to do it, so funny, everyone checks. But seriously,” he went on, “I swear it’s totally true. My dad told me, it means you’re deformed.” “No way” I said, and checked again. So he cracked me again. Fool Us Once: The Adult Pneumococcal Vaccine Below is a graph of CDC data tracking the rates of serious pneumococcal illness since 1997 in the United States. Uptake of the adult pneumococcal vaccine, a shot intended to reduce such illnesses, began in earnest in the 1990s. The only two inflection points in the timeline are at 2000 and 2010. Each point appears to initiate a 5-year downward trend in pneumococcal disease, followed by five years of no change. Both points coincide almost perfectly with newly implemented childhood pneumococcal vaccines—a classic manifestation of herd protection, where immunizing toddlers stops them from passing bacteria to their grandparents. The adult pneumococcal vaccine, conversely, first gained widespread use in the late 1990s—where the line lies stubbornly flat for its first three years on record. Then, in 2014, the CDC’s vaccine advisory committee recommended routine PCV13 (pneumococcal) vaccination for those over 65. With expanded Medicare coverage, this fueled a surge in uptake from 3% to 49% of Medicare beneficiaries by 2019. And yet, through the surge years, pneumococcal disease remained unchanged. These findings fit with results from the vaccine’s one relevant large trial—but only if you know how to read it. The CAPITA trial randomized 85,000 older adults in the Netherlands to a pneumococcal vaccine or a placebo, and found the vaccine did not measurably reduce either pneumonia or mortality. It did, however, achieve its stated goal: reducing illnesses labeled as due to vaccine strains of the bacteria. But that’s a shifty substitution. Patients want fewer pneumonias, fewer hospital stays, and fewer deaths. Instead, the trial claimed a tiny reduction in pneumonias with selected bacterial names. It was a classic bait and switch, since doctors and patients would see the ‘reduction in pneumonias’ and equate it with a reduced chance of pneumonia overall. But that’s not what the trial showed. Pneumonia was as common in the vaccine group as in the placebo group. How can one subtype be reduced, while pneumonia rates overall are not? One way would be if vaccination increased pneumonia due to other bacteria or viruses, offsetting those it prevented. Another would be if pneumonia was just as common but vaccination made it harder for labs to detect and isolate covered strains, making those pneumonias appear less common in the vaccine group. Or, perhaps the vaccine prevents so few cases that the difference disappears among all the other pneumonias. Whatever the mechanism, people who received the vaccine weren’t less likely to experience pneumonia. Instead, a tiny reduction in selected bacterial strains stood in for what people actually care about. And that surrogate became a brilliant sales pitch in the vaunted New England Journal of Medicine. But the broader benefit it claimed to represent never materialized—either in trials or in population trends. RT is fully reader supported. Please become a paid subscriber, so we can continue to TR! Fool Us Twice: The Adult RSV Vaccine Last week, trial findings from Pfizer’s RSV vaccine appeared in NEJM Evidence, accompanied by an editorial titled “A Hat Trick for RSV Vaccines — Prevention of Disease, Hospitalization, and Beyond.” And beyond! Now we’re vaccinating Buzz Lightyear. But before boarding the tiny spaceship, here’s a question that hearkens back to our first rodeo: Were vaccinated people less likely to end up in the hospital, or to die? The Pfizer-funded DAN-RSV trial randomized an incredible half million adults to vaccine or no-vaccine. During the study more than 27,000 of them were hospitalized and 1,200 died—enough to show even tiny benefits in the two outcomes. But the vaccine reduced neither. So where does the victory dance come from? Their primary outcome: RSV-attributed respiratory hospitalizations, at 19 vs 64 in the two groups. Sound familiar? This isn’t our first rodeo. In the DAN-RSV study the adjective ‘RSV-attributed’, much like ‘vaccine-type strains’ in the CAPITA study, is doing serious work. If we drop it, and instead simply ask whether people were less likely to be hospitalized or die, the triumph disappears. People deserve to understand that distinction before rolling up their sleeves, because people don’t lie in hospital beds hoping for a more flattering diagnostic code. They hope not to be in a hospital at all. And of course, we all deserve to know if the vaccine truly works, because vaccination comes with ginormous individual and societal costs, both human and monetary. Setting aside appointments, out-of-pocket costs, then pain, swelling, and fatigue from the shot, there were also 21 people in the trial judged to have had serious vaccine side effects. These included cardiac arrhythmia, pericarditis, facial paralysis (Bell’s palsy), and five cases of Guillain–Barré Syndrome. Financially, Abrysvo (the shot’s brand name) generated roughly $1.7 billion globally in 2023 and 2024—numbers poised to explode if professional societies and governments are convinced by NEJM’s ‘hat trick’. Similarly, the pneumococcal vaccine, recommended by the CDC and dozens of societies, has added a billion per year in Medicare spending alone—i.e. tax dollars. Which raises a question: At this point, given the data, why does anyone get that shot? The answer is that by the time the pneumonia shot enters the room, the sales job is largely complete. The New England Journal of Medicine called it a success. The FDA approved it. Public-health authorities recommended it. Your doctor brought it up. Who could imagine the best science shows it failed to reduce pneumonias or deaths, or that public health trends show it had no impact? That is the awesome power of the Death Star, turning a narrow surrogate success into a sweeping promise for less illness and lives saved. Meanwhile, the price we pay is billions of dollars and millions of side effects—all while the promise is quietly broken. And up goes our hand again. Footnotes: Get full access to Research Translation at researchtranslation.substack.com/subscribe

  6. Sep 22

    The New Adult RSV Vaccine, the Death Star, and Promises Broken

    In third grade Marc Diaz explained to me, very earnestly, that if one’s hand was larger than their face, that person was technically deformed. Concerned, I held my palm to my face, carefully comparing. At which point he cracked the back of my hand with his own palm, sending my hand into my face and me toppling backwards out of my chair. Good one, I thought, ready to line it up on the next sucker. But he wasn’t finished. “Sorry sorry,” he said “I had to do it, so funny, everyone checks. But seriously,” he went on, “I swear it’s totally true. My dad told me, it means you’re deformed.” “No way” I said, and checked again. So he cracked me again. Fool Us Once: The Adult Pneumococcal Vaccine Below is a graph of CDC data tracking the rates of serious pneumococcal illness since 1997 in the United States. Uptake of the adult pneumococcal vaccine, a shot intended to reduce such illnesses, began in earnest in the 1990s. The only two inflection points in the timeline are at 2000 and 2010. Each point appears to initiate a 5-year downward trend in pneumococcal disease, followed by five years of no change. Both points coincide almost perfectly with newly implemented childhood pneumococcal vaccines—a classic manifestation of herd protection, where immunizing toddlers stops them from passing bacteria to their grandparents. The adult pneumococcal vaccine, conversely, first gained widespread use in the late 1990s—where the line lies stubbornly flat for its first three years on record. Then, in 2014, the CDC’s vaccine advisory committee recommended routine PCV13 (pneumococcal) vaccination for those over 65. With expanded Medicare coverage, this fueled a surge in uptake from 3% to 49% of Medicare beneficiaries by 2019. And yet, through the surge years, pneumococcal disease remained unchanged. These findings fit with results from the vaccine’s one relevant large trial—but only if you know how to read it. The CAPITA trial randomized 85,000 older adults in the Netherlands to a pneumococcal vaccine or a placebo, and found the vaccine did not measurably reduce either pneumonia or mortality. It did, however, achieve its stated goal: reducing illnesses labeled as due to vaccine strains of the bacteria. But that’s a shifty substitution. Patients want fewer pneumonias, fewer hospital stays, and fewer deaths. Instead, the trial claimed a tiny reduction in pneumonias with selected bacterial names. It was a classic bait and switch, since doctors and patients would see the ‘reduction in pneumonias’ and equate it with a reduced chance of pneumonia overall. But that’s not what the trial showed. Pneumonia was as common in the vaccine group as in the placebo group. How can one subtype be reduced, while pneumonia rates overall are not? One way would be if vaccination increased pneumonia due to other bacteria or viruses, offsetting those it prevented. Another would be if pneumonia was just as common but vaccination made it harder for labs to detect and isolate covered strains, making those pneumonias appear less common in the vaccine group. Or, perhaps the vaccine prevents so few cases that the difference disappears among all the other pneumonias. Whatever the mechanism, people who received the vaccine weren’t less likely to experience pneumonia. Instead, a tiny reduction in selected bacterial strains stood in for what people actually care about. And that surrogate became a brilliant sales pitch in the vaunted New England Journal of Medicine. But the broader benefit it claimed to represent never materialized—either in trials or in population trends. Fool Us Twice: The Adult RSV Vaccine Last week, trial findings from Pfizer’s RSV vaccine appeared in NEJM Evidence, accompanied by an editorial titled “A Hat Trick for RSV Vaccines — Prevention of Disease, Hospitalization, and Beyond.” And beyond! Now we’re vaccinating Buzz Lightyear. But before boarding the tiny spaceship, here’s a question that hearkens back to our first rodeo: Were vaccinated people less likely to end up in the hospital, or to die? The Pfizer-funded DAN-RSV trial randomized an incredible half million adults to vaccine or no-vaccine. During the study more than 27,000 of them were hospitalized and 1,200 died—enough to show even tiny benefits in the two outcomes. But the vaccine reduced neither. So where does the victory dance come from? Their primary outcome: RSV-attributed respiratory hospitalizations, at 19 vs 64 in the two groups. Sound familiar? This isn’t our first rodeo. In the DAN-RSV study the adjective ‘RSV-attributed’, much like ‘vaccine-type strains’ in the CAPITA study, is doing serious work. If we drop it, and instead simply ask whether people were less likely to be hospitalized or die, the triumph disappears. People deserve to understand that distinction before rolling up their sleeves, because people don’t lie in hospital beds hoping for a more flattering diagnostic code. They hope not to be in a hospital at all. And of course, we all deserve to know if the vaccine truly works, because vaccination comes with ginormous individual and societal costs, both human and monetary. Setting aside appointments, out-of-pocket costs, then pain, swelling, and fatigue from the shot, there were also 21 people in the trial judged to have had serious vaccine side effects. These included cardiac arrhythmia, pericarditis, facial paralysis (Bell’s palsy), and five cases of Guillain–Barré Syndrome. Financially, Abrysvo (the shot’s brand name) generated roughly $1.7 billion globally in 2023 and 2024—numbers poised to explode if professional societies and governments are convinced by NEJM’s ‘hat trick’. Similarly, the pneumococcal vaccine, recommended by the CDC and dozens of societies, has added a billion per year in Medicare spending alone—i.e. tax dollars. Which raises a question: At this point, given the data, why does anyone get that shot? The answer is that by the time the pneumonia shot enters the room, the sales job is largely complete. The New England Journal of Medicine called it a success. The FDA approved it. Public-health authorities recommended it. Your doctor brought it up. Who could imagine the best science shows it failed to reduce pneumonias or deaths, or that public health trends show it had no impact? That is the awesome power of the Death Star, turning a narrow surrogate success into a sweeping promise for less illness and lives saved. Meanwhile, the price we pay is billions of dollars and millions of side effects—all while the promise is quietly broken. And up goes our hand again. RT is totally reader-supported. Please become part of the support—become a paid subscriber. Get full access to Research Translation at researchtranslation.substack.com/subscribe

  7. Sep 16

    What If Medicine’s Most Trusted Journal Can’t Be Trusted?

    Jeffrey Flier, former dean of Harvard Medical School, recently accused the New England Journal of Medicine of editorial failure. Flier and others were responding to an essay from the Journal’s new Voices section about Jason Arday, a Cambridge professor who committed suicide after being outed for academic work and a life story littered with fabrications. The essay, which began “They are trying to kill us,” invoked white supremacy and called Arday’s lies irrelevant. I respect Dr. Flier’s opinion. And I do find much of the NEJM’s advocacy undignified, screed-ish, and misplaced. But I don’t typically look to the politics and culture section of a medical journal. To me, the editors have a right to share their views and readers can tolerate, appreciate, or ignore them. I choose the latter. Welcome to RT, a reader-supported joint. If you want it to keep going become a paid subscriber. And yet. The Journal does have an urgent editorial problem—just not in the opinion pages. It’s in the science. In the two years since I’ve been writing a Substack, nearly every biomedical study I’ve reviewed from NEJM has failed to meet the basic standards and obligations of peer review and scientific editing. This matters. The 214 year-old Journal is not a printing press, or a pay-to-play predator that publishes every submission. NEJM’s vaunted editorial staff chooses the studies, reviews their methods, sculpts the conclusions, commissions and approves the editorials, and puts its seal of approval on every piece. When that process fails to be scientifically grounded—or even, in some cases, intellectually honest—the imprimatur of ‘peer review’ no longer denotes rigor or truth. Instead, it becomes a marketing slogan. Below are seven examples from the past two years, each describing one paper’s most brazen editorial lapses. Recall that I only review papers when they are relevant to headlines or topics I’m writing about. Which means I’ve sampled about 1% of the Journal’s output in that time. And yet, what I’ve seen could fill a book on editorial failures and violations of public trust. #1 The Alzheimer’s drug with an invisible ‘benefit’ The NEJM published the lone randomized trial testing lecanemab, a $30,000 per year drug for Alzheimer’s Disease. The 1,800-person trial reported a difference of 0.45 points between drug and placebo on an 18-point dementia scale. But numerous prior studies have carefully shown that the smallest detectable change is 1–2 points. In other words, the drug failed to have an effect on dementia that was visible to patients, families, or doctors. Obviously, that’s the headline. But it wasn’t. There could be nothing more relevant to patients and doctors than the proven absence of a perceivable benefit, particularly for a pricey drug that comes with fatal risks and common infusion reactions. Yet the Journal ‘s ‘experts’ either didn’t know this fact, or they did—and chose not to tell readers or the public. American patients and insurers have spent roughly $600 million, so far, on lecanemab. #2 The COVID vaccine that prevented car crashes An Israeli observational study reported that people who got vaccine boosters were 95% less likely to die of COVID than those who didn’t. The Journal offered this as ‘real world’ evidence of the booster’s life saving effects. In a letter to the editor, Dr. Tracy Høeg and others used the same data to show that people who received the shots were also 95% less like to die from everything else. Obviously, the vaccine can’t prevent car crashes, cancers, or heart attack deaths. Which means the shots weren’t preventing any deaths, from anything. The group differences were due to healthy-user bias: people who sought out boosters were much healthier than those who didn’t, and therefore died less from everything. Peer reviewers and editors with any education in research, particularly vaccine studies, know this. But either the Journal staff didn’t—or they did, and purposefully ignored it. #3 A statin trial that changed its outcome when it was nearly done. The STAREE trial randomized nearly 10,000 older adults to statins or placebo. In 2024, nine years after it began, the researchers added coronary revascularization to the primary outcome. The paper makes no mention of the change, which is in the trial registry, and the protocol linked from the Journal’s site. Trial registries were created decades ago precisely to prevent outcome switching. But it only works if journals check the registry, a basic and essential function of peer review. Again, the NEJM either did not check or (worse) they did, and decided not to inform the rest of us. #4 A flu drug that altered swab results—not illnesses The NEJM published a trial of the flu drug baloxavir, finding it reduced positive nasal swabs in household contacts by 4%. But flu illnesses were unchanged. The NEJM not only accepted, and thus tacitly endorsed, the conclusion that baloxavir “reduced transmission,” they published an editorial supporting the drug. U.S. sales are now in the hundreds of millions. #5 A drug for people ‘without a previous heart attack’ A paper from 2025 was titled “Evolocumab in Patients without a Previous Myocardial Infarction or Stroke.” This sounds like a study of ordinary people who’ve never had cardiovascular disease. But no. Participants all had established vascular disease, or high-risk diabetes, plus additional mandatory high-risk features. That’s not primary, or even typical secondary, prevention. It is a study of ultra-high risk people. But you’d never know it from the title, which is all most people will see. The result is virtually guaranteed to be use of the drug in people nothing like the study cohort, and therefore less likely to benefit—but just as likely to experience harms and side effects. The study’s title is a perfect advertising bait-and-switch for this $8,000 per year drug, and a marketing coup for the drug company—but a dark day for truth in publishing. #6 The subgroup, masquerading as a full study. Pfizer tested its new mRNA flu vaccine in 46,000 adults, most of whom were 65 and older—the critical group for whom an effective vaccine could potentially save lives. The paper claims success, but presents a study of 18,476 people. Why did 60% of the data disappear? Because people 65 and older saw no benefit, dooming the vaccine to failure in the overall study. So Pfizer removed them, and presented the younger subgroup as though it was the whole study. The study’s registry page fearlessly includes this information. It seems the NEJM editors and reviewers (again) either didn’t check it, or else knowingly endorsed the publication of a partial dataset, pretending to be a study. #7 When knee surgery made knees worse, NEJM said: Write a letter. The FIDELITY trial compared arthroscopic meniscectomy with a sham procedure, a rare gold standard trial with a surgical sham arm. At one, two, and five years the surgery was no better than the fake. But by ten years, the real surgery group had more degeneration, more pain, and three times as many knee operations. Arthroscopic meniscal surgery, performed hundreds of thousands of times per year in the U.S., currently costs Americans roughly $3 billion annually. But the NEJM editors refused to publish the study’s long term results as an Original Article, asking the researchers to compress the findings into a letter for the barely-noticed ‘Letters to the Editor’ section. So they did. Meanwhile, the week the Journal published the letter the lead Original Research was a non-randomized study of a $300,000-a-year drug for a rare molecular cancer subtype. Taken together, this is not a collection of random or unrelated mistakes. These failures operate in one direction: Away from truth and transparency, and toward money. They make potentially profitable drugs, devices, and procedures look better. They allow lab measurements, and effects too small to perceive, to be called ‘benefits’. Outcomes are switched, populations showing no benefit are removed, and results of urgent importance to hundreds of thousands of people are diverted into letters. And in each case they are stamped: PEER REVIEWED, by the NEJM. Peer review is not science. But for better or worse it is the border between a researcher’s claims and spin, and what the world comes to see as accepted science. It is the process through which data acquires the institutional authority of science. And no journal has more of that authority than the New England Journal of Medicine. Most doctors, patients, and reporters will never scrutinize an original protocol, reconstruct a statistical analysis, or compare a publication with its registry. Neither will legislators, or insurers, or the professional societies that convert findings into guidelines. Why? Because they assume peer reviewers did it. Peer review is why scientific journals exist, a domain of expertise that is supposed to curate, improve, and ensure the validity and trustworthiness of what we call ‘science’. A study’s appearance in the storied and respected NEJM is therefore treated as de facto evidence that its methods were scrutinized, its work has been checked, and its registry data align with its paper. Which means that top journals like NEJM do not just report science. They help create it, infusing studies with the faith and implied scientific validity of findings that have been vetted by a bastion of science and truth. This is why NEJM’s failures carry enormous weight. When the Journal permits an invisible effect to be sold as a benefit, or a goal post to be moved, or a subgroup to masquerade as a trial, it doesn’t just platform bad science. It launders falsehoods, turning them into accepted ‘facts’. Recent history has seen a collapse in public trust that is often blamed on the CDC and other government agencie

    What If Medicine’s Most Trusted Journal Can’t Be Trusted?

  8. Sep 8

    Two New Statin Trials the AHA Won’t Want to Talk About

    Once again the audio this week is for people who enjoy a bit more wonkiness. Below is what ChatGPT wrote when I pasted in the transcript and asked for a piece emulating my voice. And as always, I still couldn’t stop myself from editing pretty heavily. Enjoy! RT is fully reader-supported. To help me keep doing it, please become a paid subscriber. Just when I thought I was out, they pulled me back in. I’d finished my cholesterol series, a history of the Lipid Hypothesis, and a review of data on statins for primary prevention. Now, along come two major trials published within days of each other—one in the New England Journal of Medicine and one in The Lancet Healthy Longevity. Both were publicly funded, a rarity for statin trials. Both studied older people without cardiovascular disease. And both tell us—again—that lowering cholesterol in primary prevention neither prolongs nor improves life. The larger trial was STAREE, a huge Australian study of nearly 10,000 adults age 70 and older. Half received 40 milligrams of atorvastatin daily while the rest received placebo, for a median of six years. This was an ultra-idealized group. Participants lived independently, were cognitively and functionally intact, and had no known cardiovascular disease, diabetes, dementia, or major illness. Their doctors had to approve them, and they even completed a four-week run-in period proving they would reliably take the pill. In other words these were healthy people, optimized to see benefits and less likely to suffer harms. The investigators originally planned two major composite outcomes. The first asked whether statins reduced cardiovascular deaths, heart attacks, or strokes. Then, if the drugs accomplished this task, the researchers would calculate the true prize: Overall deaths, dementia, or disability. Drum roll, please. The cardiovascular outcome occurred in 4.7% of the atorvastatin group and 6.3% of the placebo group—a statistically significant difference of 1.6%. That means 98.4% of participants saw no benefit. Put another way, roughly 63 people had to take atorvastatin for six years for one of them to avoid an event. But which event? Remember, there were three. There was no significant reduction in cardiovascular death, or stroke. The difference in the overall outcome was driven overwhelmingly by a 1.1% reduction in nonfatal heart attacks. IMHO, that endpoint deserves serious skepticism. Nonfatal heart attacks are included in trials because they’re supposed to predict what people care about: disability and death. But enormous meta-analyses of coronary trials show that these events do not reliably track with mortality. Which means ‘nonfatal heart attack’ probably does not mean what most people think it means. Most of us imagine crushing chest pain, a damaged heart, an emergency stent, and a life permanently shortened. But many events called heart attacks in modern trials may instead be small blood test changes seen during medical stressors like pneumonia, surgery, sepsis, or other hospitalizations—events with little or no lasting effect on lifespan or quality of life. STAREE’s protocol says the investigators planned to distinguish and track these ‘type 2’ heart events, but the paper did not report it. They did show us, however, that despite a 1.1% reduction in nonfatal cardiovascular events, there was no reduction in death. Or dementia. Or disability. Or hospitalizations. It is worth pausing for a moment to contemplate and summarize: The researchers performed a 10,000 person, publicly funded, decade-long randomized trial to determine whether statins delayed or prevented death, dementia, or disability. They did not. It’s a stinging rebuke to any notion that statins have a meaningful medical impact in primary prevention. Which is incredible—because we haven’t yet looked at the drugs’ harms. Perhaps we should. Diabetes and related events were 1.3% more common with atorvastatin. Musculoskeletal problems were 2.6% more common. Overall adverse events were 4.3% more common. The clean comparison for benefits vs harms therefore, is a 1.1% reduction in nonfatal heart attacks versus a combined 3.9% increase in diabetes-related and musculoskeletal problems. The harms were more than three times as common as the only benefit. Case closed. Although there is, (to me), an interesting footnote. When the trial began in 2015, coronary revascularization was not part of the primary outcome. Nine years into the study, the researchers were seeing fewer cardiac events than expected, so they added it. To repeat: Nine years into the trial they changed their primary outcome. You will not find that disclosure anywhere in the published manuscript. To discover it, you have to dig through the trial’s 182-page protocol. The New England Journal of Medicine either failed to notice this tectonic change or allowed the authors to bury it. Either possibility is worth pondering. (And in a future piece I will, because NEJM’s failure to even pretend to uphold the central purpose of trial registration—preventing selective outcome reporting—is breathtaking). Adding revascularization increased the reported benefit from 1.6% to 2.3%. But outside an active heart attack, placing a coronary stent does not prevent future heart attacks, or strokes, or deaths. COURAGE, ISCHEMIA, and other trials settled that question. A procedure that does not prevent meaningful outcomes cannot legitimately stand in for them, a fact that many cardiology researchers have openly discussed. Moreover, ‘revascularization’ carries a built-in bias in any statin trial. Cardiologists are more likely to suspect coronary disease, refer for angiography, and place a stent when cholesterol is higher. But statins lower cholesterol. Participants in the placebo group are therefore more likely to undergo stent procedures. Even accepting the altered endpoint, the difference in groups was driven entirely by nonfatal heart attacks and revascularizations—1.1% and 0.7%, respectively. Meanwhile, diabetes and musculoskeletal harms still totaled 3.9%. Harms therefore outnumbered cardiovascular benefits by more double, even while statins had no impact on the true prize, death, dementia, or disability. On to the second trial. SAGA/SITE was a smaller and simpler trial conducted across France. It enrolled roughly 1,200 primary-prevention adults aged 75 or older who were taking statins. They were randomized to either continue or stop. After three years, stopping produced no increase in mortality, heart attacks, or any other major outcome. Nothing measurable was lost by discontinuing the drug. In summary, STAREE and SAGA answer two sides of the same question: What happens when healthy older people start statins? They do not live longer, avoid dementia, remain independent, or stay out of the hospital—and statin harms far outnumber their plausible benefits. What happens when they stop statins? Nothing bad. What strikes me is not that these results differ from any other primary-prevention literature. They don’t. What differs is clarity. These publicly funded trials present the individual outcomes plainly enough for everyone to see what industry-run statin trials spent decades hiding. In fact, older adults are precisely the people in whom statins should have had their best chance to produce a meaningful advantage. Yet the results were exactly the same as dozens of other trials. Statins lowered cholesterol. But they didn’t help people. Get full access to Research Translation at researchtranslation.substack.com/subscribe

    Two New Statin Trials the AHA Won’t Want to Talk About
See All (108)

Ratings & Reviews

5
out of 5
5 Ratings

About

Translating Medical and Health Research For All researchtranslation.substack.com

Information

You Might Also Like