Patients with schizophrenia who don't meet eligibility criteria for clinical trials (~80% of patients) respond much worse to antipsychotics than those who do. How much worse is "much worse"? . Let's find out🧵

RT @DavidPuder: Great thread: . Shows my intuition regarding .Antipsychotics to be accurate…. 1) clozapine wins.2) long acting injectables….

💭The Bottom Line. ► Differences between RCTs and real-world → efficacy-effectiveness gap. ► Antipsychotics work in the real-world, but they work 40% less on average for patients who don't represent RCT study populations - and that's 80% of patients. ► The

💉A silver lining?. Effectiveness can also reveal treatments that do better in the real-world than in clinical trials. In real-world effectiveness, LAI formulations did better than their oral counterparts (HR: 0.88). However, in RCTs, LAI efficacy was equivocal, with a HR of

📊 The Efficacy-Effectiveness Gap. Comparing RCT efficacy vs real-world effectiveness across antipsychotics shows a jarring picture. There is a range of differences in the efficacy-effectiveness gap (hazard ratios relative to haloperidol):. Smallest gap:.► Risperidone in RCTs:

🚨 Are RCTs generalizable?. How can we tell whether RCTs generalize? One way would be to:. 1. Measure effectiveness in patients who would have met criteria for an RCT (20% of patients). 2. Measure effectiveness in patients who do not meet criteria for an RCT (80% of patients).

💊 Efficacy in clinical trials. What about in clinical trials, where things are controlled, and strict inclusion/exclusion criteria are used to filter patients?. Winners (compared to oral haloperidol):.► None - all antipsychotics did about the same. Losers (compared to oral

📈Effectiveness in the real-world. In real-world settings (which looks at effectiveness), antipsychotics were compared to a reference point (oral haloperidol). Who were the winners?.► Clozapine (oral, 36% lower relapse).► Olanzapine (LAI, 27% lower relapse). Who were the

🗺️Combining RCTs with real-world data. The authors use network meta-analysis (NMA) to compare and synthesize data from multiple sources. ► RCTs: 30 trials (n=10,091) of antipsychotic efficacy. ► Real-world data: Individual patient data from Swedish (n=29,637) and Finnish

🥇Are RCTs the gold standard?. RCTs provide evidence on efficacy against placebo. But real-world effectiveness versus no antipsychotic may vary due to:.► Patient selection biases .► Exclusions for comorbidities .► Differences in med compliance. This leads to the

As a trainee, I try and follow clinical trials closely because they are the best level of evidence we have for helping patients make treatment decisions. But if the trials don’t represent our patients, how do we make informed decisions?. A tough conundrum.

Source:

80% of patients with schizophrenia in the real world do not meet eligibility criteria for the very trials their antipsychotics were approved for.

Woah. This is actually really wild. It's easy to be fooled by how good a LLM is based on how sycophantic it is. But the bulletproof benchmark is in predicting the future, which has ground truth. @elonmusk has bet on a maximally truth-seeking LLM based on math, physics, and.

One of the most common mistakes I hear when therapists coach about meditation is this. Meditation isn’t about not thinking. It’s about noticing thoughts as they arise. Without judgement. Without avoidance. Just observing and being curious.

Awesome summary by @DrDominicNg of how machine learning identified a diet that reduces dementia risk by a third. It's called the MODERN diet - and it's fairly easy to follow.

RT @paulg: Believe it or not, it's usually wise to walk investors through the risks involved in your startup. Investors know there's risk.….

RT @davidycli: sorry to report but on the current trajectory there will only be two types of biotechs in the US in < 5 years: . 1. tech bi….

A standard practice for antidepressant deprescribing is to cross-taper someone to fluoxetine because it has a long half-life and a liquid formulation. The theory is that by replacing one med for another, there, then slowly tapering off the other, antidepressant discontinuation.

The bear case: . "The prognosis was good, a surgeon assured him, with a 65 percent chance of a cure. But the man said he didn’t want treatment; he wanted MAID. What I’ve learned since is: Eligible doesn’t mean you should provide MAID,” Li told me. “You can be eligible because.