The HRT "window of opportunity" for brain health: what the trials actually showed. KEEPS: 727 women, ages 42-58, within 36 months of final period. The exact population proponents said would benefit.
Paul Abramson MD
@paulabramsonmd.mydoctorsf.com
Private physician in SF @mydoctorsf.com, MedEd at UCSF, concierge medicine, #addictionmed, infectious diseases, hospital med, gerontology, fmr EE, armchair ethicist. Bunny dad. #medsky
Linoleic acid consumption increased from 3% to 7% of calories over the last century. Soybean oil up 1,000-fold. This is a real dietary shift worth investigating.
The seed oil discourse gets one thing right: linoleic acid consumption increased 1,000-fold last century. It gets more wrong: RCTs show no increase in inflammation. Large cohorts show lower heart disease with higher intake. What it misses: the real issue is ultra-processed food. #medsky
EVOKE Phase 3 trial results: Semaglutide for Alzheimer's disease: 3,808 patients with MCI or mild dementia. 104 weeks. Primary endpoint: no difference from placebo
Why do red light therapy studies contradict each other? Biphasic dose-response. Too little does nothing. Too much backfires. The sweet spot is narrow. Most consumer devices are not independently verified to hit it. The honest answer: we do not know if yours works. #medsky
Red light therapy has better evidence than most optimization interventions. The mitochondrial mechanism is plausible. Systematic reviews show modest benefits for muscle recovery and knee pain. Still not strongly recommended by major societies. But not dismissible either. #medsky
Peptide evidence: essentially zero adequate human trials. Red light therapy evidence: multiple systematic reviews, plausible mechanism, moderate effects for some conditions. This is what engaging seriously with the evidence looks like. Not all optimization is equal. #medsky
"Can I reverse my plaque?" Less important than you think. Plaque regression is a surrogate endpoint. A smaller unstable plaque is worse than a larger stable one. Statins regress plaque AND reduce mortality. Focus on outcomes, not imaging.
"I want the PCSK9 inhibitor, it is newer." You would not fund a startup with one Phase 2 trial when the competitor has 22 RCTs and mortality data. Why would the bar be lower for your coronary arteries?
When your kid brings home flu, Xofluza is one dose for contacts vs 10 days of Tamiflu. Sounds better. But households aren't single-exposure events. Ongoing exposure may need ongoing coverage.
In this flu season, we are reviewing cases of rhabdomyolysis with super high CK levels in the context of influenza A infection Here's a case report, but surprised there's not more in the literature: pmc.ncbi.nlm.nih.gov/articles/PMC...
pmc.ncbi.nlm.nih.gov
Each day you delay flu treatment increases death risk by 14-40%. Both antivirals shorten illness by about one day if started within 24 hours. After 48 hours, Xofluza has no data at all.
If you care about healthspan, you have to care about blood pressure. Uncontrolled BP claims 10 million lives yearly. Intensive control cuts cardiovascular events by 30%. Taking a peptide feels sophisticated. Checking your BP feels ordinary. The evidence base isn't close.
The choice that matters isn't Tamiflu vs Xofluza. It's today vs tomorrow. Same-day flu treatment was associated with 64% lower mortality. Drug selection? Similar outcomes.
Beyond the headline: cystatin C discordance with creatinine eGFR is about more than just kidney function; it predicts outcomes, may suggest systemic problems, and the directionality is critical (GFRcr > eGFRcys is a good sign). buff.ly/CscgFAJ #medsky
Discordance in Creatinine- and Cystatin C-Based eGFR and Clinical Outcomes: A Meta-Analysis - PubMed
In the CKD-PC, 11% of outpatient participants and 35% of hospitalized patients had an eGFRcys that was at least 30% lower than their eGFRcr. In the outpatient setting, presence of eGFRcys at least 30%...
buff.ly
Software can be patched. Markets recover from failed experiments. The body's irreplaceable systems cannot. "Move fast and break things" doesn't work here.
Patients overestimate statin side effects by an order of magnitude. True attributable muscle symptoms: less than 1%. Severe myopathy: less than 0.1%. The perception gap drives people toward less-proven alternatives.
How smart people fool themselves about health interventions: Subtle adjustments to probability estimates. Shifts in confidence levels. Quiet downgrades of source reliability for inconvenient findings. Small biases that accumulate into confident but unfounded conclusions.
The same traits that drive innovation—systematic thinking, high openness, confidence in analysis—become vulnerabilities when applied to human biology. The mindset that builds companies requires different guardrails when applied to health.
Statins reduce all-cause mortality in high-risk primary prevention (RR 0.92). PCSK9 inhibitors do not demonstrate that benefit in this population. The boring generic pill has the outcome data the fancy injectable lacks.
Dunning himself clarified his research: "You can be incredibly intelligent in one area and completely not have expertise in another." Tech founders approaching human biology with the confidence they bring to software development is a textbook case.
"Unintelligent people are more easily misled by others. Intelligent people are more easily misled by themselves." They construct more sophisticated rationalizations. Find creative ways to dismiss contrary evidence.
Aging muscle isn’t just smaller. It's different. Soendenbroe et al. 2024 (J Cachexia Sarcopenia Muscle): heavy resistance training reverses irregular myofiber shape, a marker of aging. Net: strength work improves muscle quality, not just size. For healthy aging, resistance training is non‑negotiable
I’m quoted in today’s @nytimes.com on the "Chinese peptides" trend in SF. She asked about the "founder mindset" applied to biology. My take: "The entrepreneurial parallel isn't funding a scrappy start-up. It's wiring money to an unregistered offshore entity based on a pitch deck." buff.ly/n5nEob
Creatinine-based eGFR estimation is the default, cheap, fast, and often wrong. Cystatin C is more accurate, must be ordered separately, and in many settings takes days to return. Why? #medsky
The simple pleasure of chewing on a stick in the morning.
I know it's been an eventful year but let's not forget that AI served up this monstrosity of a figure in a paper that was published and then retracted by one of the frontiers journals this year. The rat with giant balls lives in AI infamy for all time scienceintegritydigest.com/2024/02/15/t...
T-SPOT.TB performs much better in many immunocompromised populations IMO, but historically was logistically more difficult from CA due to time-sensitive FedEx of samples to Memphis which failed a lot with winter storms. Less an issue now that Quest bought T-SPOT.
Joseph Sassine@josephsassinemd.bsky.social · 2y agoOut now in @thetxidjournal.bsky.social! T-SPOT.TB is more likely to produce actionable results than QFT-TB in patients with cancer being screened for latent TB. #TxID #IDSky onlinelibrary.wiley.com/doi/10.1111/...
