My interview with @AshishKJha46, leader of the @WHCOVIDResponse, about the pandemic
We covered many important issues🧵 medscape.com/viewarticle/98… 1. Underuse of Paxlovid in older patients: not backed up by low incidence of rebound and ease of sidestepping potential drug interactions
2. Pharmacists unwilling to prescribe Paxlovid, even though empowered (these are results from new @Medscape survey) 3. Dr. Jha is a champion for developing nasal and universal coronavirus vaccines and explains why it hasn't happened (yet)
4. Explanation for the reasons we don't have any monoclonal antibodies that work now (no incentives)
A couple of weeks ago I reviewed the BA.5 bivalent booster data and concluded it has worked better than expected erictopol.substack.com/p/the-bivalent…
Today there are 3 new reports that strongly reinforce that 1. Comparing bivalent vs original for protection vs hospitalizations & deaths
2. Superior neutralizing antibodies vs the new variants with marked immune evasion: XBB.1, BQ.1.1, BA.2.75 compared with the original booster (adds to 12 lab studies)
3. Higher protection vs symptomatic infections (against XBB.1.5, which has an extreme ° of immune escape) than has been seen for the original booster, for at least 3 months
[Original booster vs Omicron lineages ≤ 30% protection, and short-lived]
Just published @Nature
Impressive. A wearable patch that acquires continuous, real-time ultrasound images of the heart at rest, on the go, and during exercise (echocardiograms) with automated processing nature.com/articles/s4158…
The benefit of bivalent boosters for enhanced, ~80% protection from hospitalizations and deaths (vs vaccine plus prior prior booster or unvaccinated) has now been shown in 6 countries (Israel, US, Denmark, Norway, Finland, Sweden)
The #AI takes the medicine qualifying exam.
Three to take note: 1. China, 2017, Xiaoyi passed the licensing exam with a score of 456 (passing 360-600) based on memorization and information recall, with poor performance on patient cases
2. UK, 2022, the machine took the radiology qualifying exam, compared with 26 radiologists who recently passed. Overall accuracy 79.5% compared with 84.8% for the radiologists.
3. US, 2022, the large language model (LLM) Med-PaLM takes the US medical licensing exam. A huge (33% relative) jump in accuracy of answering medical questions: 92.6% doctors say the AI was correct vs 92.9% other doctors were correct. Now replicated by ChatGPT.
A new retrospective nationwide EHR assessment of #LongCovid in ~230,000 people w/ mild Covid and matched controls for age, sex, vaccination status, variant
Some downplaying of persistent symptoms at 1 year: 1. Median age of cohort = 25 years. People age 41-60 had worse outcomes
2. Major symptoms of difficulty breathing, cognitive impairment and weakness were present at 1 year. Also loss of smell, taste, dizziness and cough
3. The vaccinated individuals did better: less dyspnea and more likeness to controls. 4. Children had more Strep throat & conjunctivitis 5. The retrospective design w/reliance only on electronic records confounds the conclusions dx.doi.org/10.1136/bmj-20… medicalxpress.com/news/2023-01-c…