Podcasts are full of health claims. Some hold up. Some don't. We check them against published research: every verdict cites the study (PMID), the exact quote, and the moment it was said. Supported. Overstated. Contradicted. With receipts.

"radiation and chemotherapy being a huge risk factor for recurrence," says Emily Kaplan. The trials say no. Post-op radiation roughly halves 10-year breast cancer recurrence rather than causing it. 17 trials, 10,801 women. PMID 22019144

"Stable plaque... It's about as likely to cause a heart attack as not having any plaque as all at all." - Ford Brewer: Doesn't hold up. Even minimal calcification tripled heart risk compared to zero plaque. 6,814 people, no control group. PMID 29688297

Sasha Hamdani: "almost 100% of people with ADHD have some form of RSD". Overstated. Rejection sensitive dysphoria is a clinical concept popular on social media, but published studies find no near-100% rate in ADHD. 31 people, no control group. PMID 17242422

"nattokinase, which can naturally help reduce LDL cholesterol" - Andrew Huberman: Overstated. Pooled trials show typical doses of nattokinase do not reliably lower LDL cholesterol compared to controls. Pooled analysis of several trials. PMID 39076715

Antidepressants showed virtually no benefit over placebo in moderate depression, reaching clinical significance only in very severe cases. The gap was driven by a shrinking placebo response. Pooled analysis of trials. PLoS Med 2008, PMID 18303940

Lifting weights builds muscle, but it failed to shrink visceral belly fat in head-to-head research. A pooled analysis of 35 randomized trials in overweight and obese adults tracked changes in deep abdominal fat using CT and MRI scans. Aerobic exercise significantly reduced visceral fat compared to inactive controls (effect size -0.33). Progressive resistance training did not produce a significant reduction, and combining weights with cardio showed no clear additional benefit. The limit: only 9 of the included trials compared cardio against weight training head-to-head. Obesity Reviews 2012, PMID 21951360

Magnesium supplements lower blood pressure, but the effect is modest: systolic dropped 2.00 mm Hg and diastolic 1.78 mm Hg over 3 months. A pooled analysis of several trials (2028 participants). Hypertension 2016, PMID 27402922

Statin therapy comes with a 9% increased risk for incident diabetes, but the absolute risk remains low compared to the reduction in coronary events. A pooled analysis of several trials. Lancet 2010, PMID 20167359

Across 522 trials and 116,477 adults, every single one of 21 antidepressants outperformed placebo for acute depression. A pooled analysis evaluated published and unpublished double-blind trials comparing 21 psychiatric drugs head-to-head and against placebo. All 21 drugs beat placebo for response, with odds ratios between 1.37 for reboxetine and 2.13 for amitriptyline. For acceptability, only agomelatine and fluoxetine had fewer dropouts than placebo. The honest limit: 73% of trials had moderate risk of bias, 9% had high risk, and certainty was rated moderate to very low. Pooled analysis of 522 randomized trials. Lancet 2018, PMID 29477251

CoQ10 is widely recommended for statin muscle pain, but pooled trial data found no significant relief. A pooled analysis evaluated 6 randomized trials with 302 patients receiving statin therapy to test whether coenzyme Q10 improved muscle symptoms. Across 5 trials tracking muscle pain (253 participants), CoQ10 had no statistically significant effect compared to control (standardized mean difference -0.53, P=.20). It also caused no significant change in plasma creatine kinase activity across 226 participants (P=.38), with no dose-response relationship. The honest limit: the trials were small, totaling only 302 people across all 6 studies, leaving wide confidence intervals. Mayo Clinic proceedings 2015, PMID 25440725

Taking 600 mg of magnesium daily cut migraine attack frequency by 41.6%, compared to 15.8% for placebo. Adults with migraines (averaging 3.6 attacks per month) took 600 mg of oral trimagnesium dicitrate or a placebo daily for 12 weeks. By weeks 9-12, attack frequency dropped by 41.6% in the magnesium group versus 15.8% with placebo. The number of migraine days and reliance on acute medication also fell. Attack duration and intensity tended to decrease, but those changes did not reach statistical significance against placebo. Digestive side effects were common: diarrhea occurred in 18.6% and gastric irritation in 4.7%. The limit: moderate-certainty evidence from a single trial of 81 people. Cephalalgia 1996, PMID 8792038

Most body-language tells for lying do not hold up under scrutiny. A review examining 1,338 estimates across 158 cues found that many behaviors showed no discernible link, or only weak links, to deceit. Liars were slightly less forthcoming, appeared more tense, and told less compelling stories with fewer ordinary imperfections. But standard physical tells failed to reliably distinguish truth from lies. Differences were slightly more visible only when the lie concerned a transgression or when people had identity-relevant motivations to succeed. A pooled analysis of several studies. Psychological bulletin 2003, PMID 12555795

Sleeping too much tracks with a higher mortality risk than sleeping too little. A pooled analysis tracking 1,382,999 people across 16 studies (with 4 to 25 years of follow-up) evaluated sleep duration against 112,566 recorded deaths: - Short sleep was linked to higher mortality (RR: 1.12) - Long sleep was linked to an even larger increase (RR: 1.30) The limit: this evidence is observational and relies on self-reported questionnaires, meaning preexisting illness could explain both longer sleep and higher mortality. Pooled analysis of several studies. Sleep 2010, PMID 20469800

About 25% of the weight lost on GLP-1 drugs comes from lean mass, not fat. Across 22 randomized trials (2,258 participants), GLP-1 medications reduced total body weight by 3.55 kg: fat mass dropped by 2.95 kg and lean mass dropped by 0.86 kg. Tirzepatide (15 mg weekly) and semaglutide (2.4 mg weekly) produced the largest reductions in body weight and fat mass, but were also the least effective at preserving lean mass. Liraglutide was the only drug to achieve significant weight loss without significantly reducing lean mass. One key nuance: relative lean mass (percentage change from baseline) was unaffected across the trials. Pooled analysis of 22 trials. Metabolism 2025, PMID 39719170

The rule that your body can only use 20-25 g of protein after a workout did not hold up when tested after whole-body lifting. Thirty resistance-trained young men completed a whole-body resistance workout and took either 20 g or 40 g of whey protein in a crossover trial. Muscle protein synthesis was significantly higher following 40 g of whey protein compared with 20 g. Total lean body mass did not change the result: participants with lower lean mass (≤65 kg) and higher lean mass (≥70 kg) showed a similar response. The limit: this measured short-term muscle protein synthesis during recovery in 30 young men, not multi-month muscle growth across diverse populations. Randomized crossover trial, 30 people. Physiological reports 2016, PMID 27511985

Creatine can protect brain function during severe sleep loss, but only for complex tasks and only after 36 hours awake. Participants took 5 g of creatine monohydrate four times a day for 7 days before undergoing 18-, 24- and 36-h sleep deprivation combined with moderate intermittent exercise. At 36 h without sleep, the creatine group performed significantly better on central executive tasks than the placebo group. The limit: it showed zero benefit for short-term memory, choice reaction time, balance, mood, cortisol, or melatonin. The cognitive buffer was restricted strictly to complex executive tasks under extreme exhaustion. Randomized controlled trial. Physiology & behavior 2007, PMID 17046034

Meditation helps with anxiety and depression, but pooled trials show zero evidence it works better than exercise or medication. A pooled analysis evaluated 47 trials with 3515 participants across adult clinical populations, testing mindfulness programs against active control groups to account for placebo effects. Mindfulness meditation produced modest improvements in anxiety (effect size 0.38 at 8 weeks) and depression (0.30 at 8 weeks), alongside modest relief for pain (0.33). The limit: researchers found low or insufficient evidence that meditation improved sleep, eating habits, weight, or positive mood. Across all 47 trials, meditation showed no measurable advantage over standard active treatments like exercise or prescription drugs. Pooled analysis of 47 randomized trials. JAMA Internal Medicine 2014, PMID 24395196

People who report severe illness from electromagnetic fields could not detect them in blind laboratory tests. Researchers evaluated 46 blind or double-blind provocation studies involving 1175 volunteers who reported electromagnetic hypersensitivity. When exposed to active versus sham fields under controlled conditions, volunteers reported symptoms regardless of whether a real field was active. The experiments found no robust evidence that electromagnetic fields triggered the symptoms, pointing instead to a nocebo effect. One honest limit: the included provocation studies tested acute laboratory exposures, and not all trials explicitly detailed randomization protocols. Pooled analysis of several studies. Bioelectromagnetics 2010, PMID 19681059