PGY-4 @UW Psychiatry. Prev MD-PhD @WUSTL. Biomarker and drug discovery for psychiatry, starting with Schizophrenia. Dog Daddy x2 to Taro and Azuki.
Seattle, WA
- Tweets 4,323
- Following 842
- Followers 4,058
- Likes 14,969
Pinned Tweet
🚨 The first randomized trial of Open Dialogue for mental health crises is finally here.
It failed its primary endpoint of delaying relapse. But it dramatically reduced psychiatric hospitalizations and crisis re-referrals.
New ODDESSI trial results in Lancet Psych 🧵
2
6
11
988
🚨 The first randomized trial of Open Dialogue for mental health crises is finally here.
It failed its primary endpoint of delaying relapse. But it dramatically reduced psychiatric hospitalizations and crisis re-referrals.
New ODDESSI trial results in Lancet Psych 🧵
2
6
11
988
🔎 Unblinded care and a mismatched relapse scale
Because participants and clinicians knew which treatment they received, subjective secondary outcomes (like patient satisfaction) are highly vulnerable to expectancy bias.
Furthermore, the primary relapse measure used here was originally designed for first-episode psychosis. In this trial, fewer than 30% of participants had a diagnosis of psychosis or bipolar disorder (a 38% plurality had depression or anxiety). The tool may simply have lacked the sensitivity to detect meaningful relapse in this broader transdiagnostic group.
Finally, families were absent for the majority of interactions in the Open Dialogue arm, meaning the intervention lacked one of its core intended active ingredients.
1
1
104
💡 Takeaways
In its first rigorous RCT, Open Dialogue did not change the clinical trajectory of mental health relapse compared to standard care.
However, it kept patients out of inpatient psychiatric units at a remarkably high rate and resulted in better self-reported recovery.
The continuity of a single dedicated team, rather than bouncing between siloed crisis and outpatient services, likely drives this benefit.
Whether health systems will fund the upfront training and staffing required remains the multi-million-dollar question.
Source: Pilling et al. Lancet Psychiatry 2026 — doi.org/10.1016/S2215-0366(2…
1
1
92
🚨UPENDING GENETICS🚨
We are taught that being a "carrier" of a recessive genetic disease is biologically silent.
One allele = no disease
Two alleles = disease
A new analysis of 378,000 genomes OVERTURNS what we know about recessive disease 🧵
1
2
3
809
🔎 Evolutionary signals in a healthy volunteer cohort
It is a generally known fact outside this study that UK Biobank participants are often wealthier, healthier, and more educated than the baseline population.
If carriers of ID variants are less likely to volunteer for genomic studies in the first place, the true penalties of carrier status might be far steeper than what this cohort captures.
However, the authors stress their study is correlational. They note that measuring fitness via childlessness is complex, as medical conditions, societal trends, and polygenic background may be much more important for reproductive success, and this dataset is restricted entirely to European ancestry.
1
92
💡 Takeaways
"Recessive" and "dominant" are extremes on a biological spectrum, not absolute rules.
Carrying a single recessive intellectual disability mutation has measurable cognitive and reproductive impacts.
Evolution is constantly pruning these variants from the population, despite their more subtle effects on phenotype.
I think this matters because when we think about modeling disease in a dish, we often look at "pathogenic" or "benign" variants. But if there were a way to measure more subtle effects experimentally, we could make predictions about how much they would impact human phenotypes, however subtle.
That modeling question is something I’m actively looking into in the lab.
Source: Fridman et al. Nature Human Behaviour 2025 — doi.org/10.1038/s41562-025-0…
76
Timing couldn’t be better.
Alkermes just announced top line results for their orexin receptor agonist for…
ADHD!!!
wtf!!!! It WORKED
investor.alkermes.com/news-r…
🚨 BREAKTHROUGH FOR NARCOLEPSY 🚨
In two phase 3 trials for narcolepsy, a NEW OREXIN AGONIST took patients from 4 minutes of sustained wakefulness to over 20.
The current gold standard Modafinil only adds ~3.6 minutes.
P.S. $LLY paid $6.3 billion to buy into the same target. 🧵
6
12
159
20,321
🚨 RELAPSE OR WITHDRAWAL 🚨
A new prospective cohort study by @markhoro finds that ON AVERAGE, antidepressant tapering led to NO new or worsening symptoms of withdrawal.
If they did in select individuals, withdrawal symptoms emerged mostly when patients drop below 75% of the minimum effective dose.🧵
4
7
18
2,079
🔎 Fated to succeed
Okay, here's the BIGGEST concern I have:
Secondary analyses happen all the time. Moving from the prespecified protocol to look at the 75% minimum effective dose could be reasonable.
But the Methods paint a different picture. From the Methods:
"In the preliminary analysis shown in the supplement, we tested whether we would find the assumed interaction effect between dose reduction relative to the minimum effective dose and intra-individual increases in DESS total scores.
These models confirmed that dosage relative to 75% of the minimum effective dose (above vs. below) and dose change (reduction vs. no reduction) were statistically significant time-variant predictors both as main effects and in interaction.
Further testing of different cut-offs confirmed that 75% of the minimum effective dose provided a better fit to the data than other thresholds (compared to 100% of the minimum effective dose and 125% of the minimum effective dose)."
Translation: we tested different thresholds and chose the threshold that best fit with our hypothesis and then did analysis with that threshold that confirmed our hypothesis.
Why???
1
1
4
191
💡 Takeaways
A NEW prospective study by @markhoro finds that ON AVERAGE, individuals discontinuing antidepressants experience no new or worsening withdrawal symptoms over 26 weeks.
Suggestive evidence indicates that antidepressant withdrawal happens mostly at the bottom of the tapering curve, when the dose drops below 75% of the minimum effective dose.
But my read of the paper is that the analyses were heavily cherry-picked and that some of the statistical choices (or lack thereof) were... confusing...
Conflicts of interest happen when a party has a vested interest in seeing an outcome. In pharma, that could be seeing a drug in clinical trials work. But for someone who runs a deprescribing clinic, that could be pushing the analyses to show that withdrawal happens more frequently, or more severely, than the data support.
I hope the authors of the paper who are here on X can help me understand the analyses if I am misinterpreting it.
Whether personalized tapering plans using very small reductions can practically mitigate withdrawal remains an open question requiring PROSPECTIVE randomized studies to directly test.
These are exactly the types of studies that the authors were trying to run, which I respect.
Source: doi.org/10.64898/2026.05.12.…
2
3
187