https://rwednesdays.com/blog.html Study design notes and opinions on real-world evidence methods, held with evidence. en-us Sun, 05 Jul 2026 20:51:06 GMT https://rwednesdays.com/posts/specialized-model-theater.html https://rwednesdays.com/posts/specialized-model-theater.html Wed, 24 Jun 2026 12:00:00 GMT A June 2026 Nature Medicine study found frontier general-purpose models beat purpose-built clinical AI tools across knowledge, alignment, and real physician queries — and clinicians preferred them. The lesson is not that specialization is dead. It moved up the stack, from the model to the evidence system around it. https://rwednesdays.com/posts/ispor-2026-recap.html https://rwednesdays.com/posts/ispor-2026-recap.html Thu, 11 Jun 2026 12:00:00 GMT Four days in Philadelphia made one thing clear — the field has moved from debating whether AI belongs in HEOR to arguing about how to govern the AI already doing the work. https://rwednesdays.com/posts/eu-jca-year-one.html https://rwednesdays.com/posts/eu-jca-year-one.html Thu, 11 Jun 2026 12:00:00 GMT The EU's Joint Clinical Assessment finished its first year. The machinery works. The PICOs multiply. And "one Europe, one assessment" remains a slogan, not a description. https://rwednesdays.com/posts/ira-round-three-part-b.html https://rwednesdays.com/posts/ira-round-three-part-b.html Thu, 11 Jun 2026 12:00:00 GMT CMS's third negotiation cycle crosses a line the first two did not: the selected list now includes drugs payable under Part B and/or covered under Part D. The evidence questions just got more interesting, and the stakes moved into the infusion suite. https://rwednesdays.com/posts/mfn-least-favored-evidence.html https://rwednesdays.com/posts/mfn-least-favored-evidence.html Thu, 11 Jun 2026 12:00:00 GMT Seventeen manufacturers, a White House savings report, and a transparency problem — the MFN pricing experiment is generating policy claims faster than anyone can verify them. https://rwednesdays.com/posts/fda-elsa-halo.html https://rwednesdays.com/posts/fda-elsa-halo.html Thu, 11 Jun 2026 12:00:00 GMT Elsa 4.0, a consolidated data platform called HALO, and one-day AI-assisted inspections — the FDA is becoming an AI-native organization faster than most sponsors. Your evidence package will be machine-read before it is human-read. Write accordingly. https://rwednesdays.com/posts/automated-systematic-reviews.html https://rwednesdays.com/posts/automated-systematic-reviews.html Thu, 11 Jun 2026 12:00:00 GMT LLM pipelines now screen abstracts more sensitively than human dual review and extract data more accurately. The systematic review industry's response so far has been to update its reporting checklists. That will not be enough. https://rwednesdays.com/posts/tte-ehr-operational-framework.html https://rwednesdays.com/posts/tte-ehr-operational-framework.html Thu, 11 Jun 2026 12:00:00 GMT A new framework in npj Digital Medicine draws the line that target trial emulation has needed for years — between the trial you specify and the data that can (or cannot) realize it. https://rwednesdays.com/posts/global-rwe-trust-gap.html https://rwednesdays.com/posts/global-rwe-trust-gap.html Thu, 11 Jun 2026 12:00:00 GMT China's real-world evidence machine was one of the loudest stories at ISPOR 2026 — and the polite skepticism in Western HTA hallways is the most important methods problem nobody has formalized. https://rwednesdays.com/posts/heor-careers-ai-premium.html https://rwednesdays.com/posts/heor-careers-ai-premium.html Thu, 11 Jun 2026 12:00:00 GMT AI skills carry a documented wage premium and appear in a growing share of job postings — but the career advice circulating in HEOR ("learn to prompt") is aimed at the wrong target. Learn to verify. https://rwednesdays.com/posts/teaching-methods-age-of-ai.html https://rwednesdays.com/posts/teaching-methods-age-of-ai.html Thu, 11 Jun 2026 12:00:00 GMT If an LLM pipeline can outperform human dual review on screening and extraction, it can certainly complete your pharmacoepi problem set. The assessment crisis in methods education is here — and it is also the best teaching opportunity in a generation. https://rwednesdays.com/posts/icpe-milan-field-guide.html https://rwednesdays.com/posts/icpe-milan-field-guide.html Thu, 11 Jun 2026 12:00:00 GMT Two thousand pharmacoepidemiologists, four days, one Italian city in late August. Here is how to extract career-changing value from ICPE — and the mistakes that turn it into an expensive poster-printing trip. https://rwednesdays.com/posts/devices-meet-jca.html https://rwednesdays.com/posts/devices-meet-jca.html Thu, 11 Jun 2026 12:00:00 GMT 2026 is the year medical devices meaningfully enter Europe's joint assessment process — a trial-shaped evaluation system colliding with an evidence ecosystem that has never run on trials. Real-world evidence will either bridge that gap or watch it become a crisis. https://rwednesdays.com/posts/tte-oncology.html https://rwednesdays.com/posts/tte-oncology.html Sun, 12 Apr 2026 12:00:00 GMT Immortal time, treatment switching, and line-of-therapy definitions break naive new-user designs in solid tumor RWE. https://rwednesdays.com/posts/harper-start.html https://rwednesdays.com/posts/harper-start.html Sat, 28 Mar 2026 12:00:00 GMT Both are excellent. The choice depends on whether you need a full protocol template or a reporting checklist for regulatory submissions. https://rwednesdays.com/posts/phenotype-1ip2op.html https://rwednesdays.com/posts/phenotype-1ip2op.html Thu, 19 Feb 2026 12:00:00 GMT The difference between 30-day and 365-day windows for acute vs chronic events is massive in claims data, and it changes your incidence rates and cohort size.