New paper and software alert! https://t.co/cZNy82J1kQ Interested in modern mediation analysis methods with machine learning and multivariate mediators? Take a look at this joint work with Richard Liu, @nickWillyamz , and @kara_rudolph Short 🧵...

So excited to see this work by truly incredible PhD student @cagimbrone out in the world! Adding quantitative evidence for the potential benefits of red flag laws.

Getting to hang out in person is the best

Have you every wondered how generalizability or transportability work with difference-in-differences? Check it out in a new preprint with @EMatthay and @kara_rudolph.

We learned optimal treatment rules to match patients w the medication for #OUD that would give them the lowest risk of relapse. These rules improved outcomes relative to had everyone received methadone or naltrexone but similar to buprenorphine. Also adding amazing @nickWillyamz

Academic moms - trying to quantify how Covid impacted your research productivity? Our newest paper can help. Check out this 🧵 (@leefay_low @AndrewJBaillie @syd_health @Momademia)

This was the best ❤️

And a huge thanks to @nickWillyamz @ildiazm @AliciaThilani and not-on-Twitter Matis, Sean, Marc, John and Ned for working on this with me

Thank you!! tl;dr: we quantified how well 4 different dosing strategies for #buprenorphine and #methadone work for treating OUD. For both medications, dose should be increased to minimum thresholds, and, for buprenorphine, increased in response to continued use after.

This is the true story of 7 epidemiologists... Picked to give presentation... To find out what happens... When people stop making assumptions... And start getting REAL. This is the real world - Epidemiology! #SER2022 #epitwitter 1/3

Tarlie was incredibly thoughtful, incredibly insightful, and so kind. I miss and will continue to miss her, but am grateful to have known and worked with her.

I am so deeply sad today to have lost a wonderful colleague and human. @TarlieTownsend was one of the strongest and brilliant people I’ve had the fortune to work with. We will miss her tremendously. May her words below and her memory be a blessing to all of us who knew her.

Specifically, we consider the scenario with one IV for the the exposure, and another with 2 IVs—one for the exposure and another for the mediator—that may be related. We propose non parametric, robust estimators for each. Thanks to @nickWillyamz we also have an R package!

As a bright point (/distraction from nyc’s Covid numbers with kids too young to be vaccinated), I’m excited to post this paper that’s been a pandemic project with @ildiazm and @nickWillyamz. We use instrumental variables to identify mediation effects

8. Message 1: next time, I’ll do the math 9. Message 2: binax is a useful screener, but important to confirm with PCR

7. *Cannot believe I failed to calculate the positive predictive value of those binax tests* WHY. PPV: P(D+ | T+) = P(T+ | D+) P(D+) / P(T+) Plugging in the sensitivity and specificity of the Binax test and the 7 day Covid prevalence in our zip code, the PPV is 6.3%. 🙄

6. Day after Thanksgiving: no results yet. I book her a 3 hour PCR test. Get results: negative. Get results of second PCR later that evening: negative.

3. I retested her, and she tested positive again! 4. We put a mask on her. I book a PCR appointment. Some family chooses not to come. 5. I get her a PCR. Few options Thanksgiving day, but I find one that guarantees results in 48 hours. Rest of family masks and leaves later.

1. We Binax tested everyone getting together inside for the holiday 2. My 6 year-old daughter tested positive. She’s had a cold with a cough for more than 2 weeks. She’s had her first vaccine dose but only recently.

How me not calculating the positive predictive value of a positive binax test messed up Thanksgiving: a thread