I did a TON of things in 2020, but one thing I didn't manage to keep up with is writing #tweetorials for all of the academic papers I published.

So here's a thread of very short summaries of the papers I've been a part of. Let me know which ones you want to hear more about!
My first paper of 2020 was the final paper from my dissertation. I really love this paper because it packs so many ideas into such a short amount of text, but one thing I regret is not giving it a more general title.

Despite the title's focus on individual-level simulation models, the paper has important implications for all mediation analyses and for defining causal questions for non-manipulable exposures.

I also published pair of papers applying g-methods for causal inference to improving the analysis of randomized clinical trials. These papers are from my postdoc work, and include my first senior author paper.
The first of this set demonstrates how to estimate the per-protocol effect with a binary adherence measure, and outlines sensitivity analyses for checking whether key assumptions are violated.
The second of this set analyses a trial with a continuous measurement of adherence. This is more complicated than dealing with binary adherence, and this paper, led by @KerollosWanis, nicely explains why & what analytic options exist.
A third paper, still in pre-print, estimates the effect of daily zinc supplementation versus placebo, taken as prescribed (ie per-protocol) on HIV disease symptoms among heavy alcohol users.

On the theme of randomized trials, I also published my first solo authored paper explaining why the concept of a "placebo effect" is generally misapplied, and how we would have to formally define a placebo effect in order to meaningfully estimate it. academic.oup.com/aje/advance-ar…
I also spent a fair amount of time critiquing studies which I felt violated causal inference assumptions. The most high profile of these was a critique of the research on hydroxychloroquine with @LucyStats, @ProfMattFox, and others: academic.oup.com/aje/advance-ar…
With @PWGTennant, I also published a critique of the OpenSAFELY study which made several important errors that have led to bad policy decisions. We have more on this topic coming soon, but also check out the critiques on OpenSAFELY from @EpidByDesign

Another critique I published continued on the theme of identifying appropriate causal questions that can be asked in our data. This one focused on the challenges of interpreting life-course exposures based only on cohorts recruited in middle age. jamanetwork.com/journals/jama/…
Because it's also important to call attention to work that's done well, I also published a paper with @EpiSconroy on a study that CORRECTLY DIDN'T control for confounding. Descriptive studies are important so it's great to see them done well. nature.com/articles/s4141…
Descriptive studies often still require causal thinking. I had the great privilege of working with @LaurenAnneWise and her team to think through the role of selection bias for a longitudinal descriptive study of behavior change in pregnancy: ingentaconnect.com/content/wk/ede…
Back to the theme of asking good causal questions: @BreskinEpi & I wrote a commentary to accompany a lovely paper by @kellyn_arnold et al on compositional exposures. Our commentary focused on causal consistency & the weighted average causal effect (WACE) academic.oup.com/ije/article-ab…
I also collaborated with a number of research groups thinking about how to handle the causal consistency assumption in complex settings.

Option 1: acknowledge you are estimating a weighted average causal effect, as in this paper led by @palolili23
Option 2: Identify a more appropriate causal estimand by thinking about the target trial.

This approach was applied in a newly accepted paper designed to estimate the causal effects of pre-exposure prophylaxis for HIV prevention. academic.oup.com/aje/advance-ar…
A companion piece explaining the rationale and methodology behind this approach is also forthcoming in @amjepi, so stay tuned for that in the new year!
A similar idea but in the setting of social epidemiology (where the causal questions and estimands are equally complex though in somewhat different ways) was outlined in this piece led by my fabulous students @h_kouser and @rubymaei👇
I also wrote some tutorial-style papers . The first, with @elliecaniglia, unpacked difference-and-difference methods for an epidemiology audience, describing the assumptions in our familiar language of DAGS and counterfactual contrasts.
This diff-and-diff paper is a fave of mine because it incorporates historic epi (John Snow's cholera data). I also wrote another historic epi paper which was technically accepted last year but only published this year, about epi pioneer John Graunt academic.oup.com/aje/article-ab…
The second tutorial, with @elliecaniglia & @lucia_petito, is anoverview to all things causal survival analysis, and includes code in SAS, R, and stata, and a workbook, dataset, and solutions manual.

If you're teaching survival analysis, check this out! journals.sagepub.com/doi/full/10.11…
I also collaborated with a group of biologists to write an explainer on causal inference in the setting of ecology, evolution, and behavior research. This paper is still a pre-print but you can read it here: arxiv.org/abs/2010.07506
And I tried my hand at talking to economists about the ways they can help with COVID-19 and other pandemic research. This was a challenging piece to write & I'm so thankful to the editors for their support and advice. aeaweb.org/articles?id=10…
Another pre-print I love, but which is having a bit of trouble fining a home, is this paper on knowledge, attitudes, and practices related to Directed Acyclic Graphis in medical research, led by @rubymaei medrxiv.org/content/10.110…
This is one of a pair of DAG papers I've worked on over the past year. The other was led by @PWGTennant and looks at how DAGs have been applied in medical research: academic.oup.com/ije/advance-ar…
That's not all, but it's all that could fit in the first thread!
A paper which is 100% a twitter collaboration & was a lot of fun but also a challenge to write, is this viewpoint on machine learning and precision medicine. Tl;dr: these methods don't do what you think they do!

Thanks @jd_wilko for inviting me to join!
Finally, some pre-prints and in progress papers which you can look forward to over the next few months:

First, a review of all the false dichotomies that have filled our conversations about COVID: osf.io/k2d84/?trk=org…
Next, a paper that was really fun to write with @sciCarly and others on Science Communication in the Age of Misinformation will be out soon in Annals of Behavioral Medicine. Can't wait to share that one with you all!
Led by @michelle_caunca and Haadiya Cheema (a fantastic undergrad in my lab!), the @EpiCOVIDCorps wrote a rebuttal to the NASEM's framework for COVID vaccine allocation calling for more focus on reducing racial and ethnic disparities. This work is ongoing. osf.io/x87ua
Haadiya Cheema also led another paper which I just love! This paper required her to read hundreds of papers about randomized trials, so that we could track how opinions of scientists and trialists have evolved over time.
Pre-print here: osf.io/69e2h
With @LucyStats and Kyra Grantz, we wrote about quantifying uncertainty in infectious disease mechanistic models. This paper is accepted at @amjepi and will hopefully be available soon!
And with @nabuelezam, I wrote about how to understand pandemic prediction models in @elemental elemental.medium.com/a-guide-to-und…
Building a collaboration with @elemental has been one of the most rewarding parts of my 2020 #scicomm, and with them we have written a series of @EpiCOVIDCorps COVID Diary Audits: elemental.medium.com/tagged/the-cov…
And led by @caredwen, we evaluated the good & bad about Farhad Manjoo's Thanksgiving plans. elemental.medium.com/an-epidemiolog…
I also wrote a guide to living in a pandemic in March, featuring a cartoon infosheet I created with @BenjaminLinas for @The_BMC
And in January, a Regular Person's Guide to Pandemic Preparedness to try to help people understand what might be coming. I failed to emphasize masks or ventilation but the bulk of this article is still relevant today. medium.com/@EpiEllie/a-re…
All that, and I have 7 other articles under review -- some have been looking for homes for several years & have @arxiv pre-prints I've shared before; others are new this year but aren't pre-printed for a range of reasons. I hope to share them with you soon!

Phew! What a year!

• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with Dr Ellie Murray

Dr Ellie Murray Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @EpiEllie

29 Dec 20
Despite the pandemic, 2020 was a wildly productive year for me academically. A big reason is I took the advice of my mentor, @_MiguelHernan, and as a postdoc & junior faculty said "yes" to every opportunity I could, in order to learn what did and didn't work for me.
Some things I've learned: A🧵
* In person talks & workshops are so energizing to me and can lead to great collaboration opportunities.
* Online talks are draining & make it much harder to network.
* Guest lectures can be fun, but recorded guest lectures are more bang for my buck.
* Doing peer review was a good use of time when I was learning to write & publish papers, but less now I am so busy.
* Pre-prints are a valuable way to get rapid feedback + reduce the pain of publication timelines.
* @UpcycledScience is a great tool for desk drawer papers.
Read 6 tweets
29 Dec 20
This past semester I taught two courses: an advanced (causal inference) methods course, and an outbreak investigation course.

I’d like to share some thoughts on teaching during a pandemic.

1. The pandemic is the major thing happening in everyone’s lives, in one way or another. Acknowledge that it exists and state clearly how your course expectations are changing because of it.
1b. Change your course expectations because of the pandemic! Relax late work policies, provide alternative grading options for people that need more flexibility, use projects or homeworks rather than exams, find ways to allow for asynchronous course participation.
Read 10 tweets
28 Dec 20
On the first day of Christmas, my quarantine activity:
A puzzle of a pear tree
On the second day of Christmas, my quarantine activity:
Two knitted gloves
And a puzzle of a pear tree
On the third day of Christmas, my quarantine activity:
Three French crime shows
Two knitted gloves
And a puzzle of a pear tree
Read 5 tweets
26 Dec 20
There’s a lot of talk about how anti-expertise sentiment is problematic, but I’d like to offer a slightly different take.

The issue isn’t that people don’t want to listen to experts, it’s that experts aren’t being honest with ourselves about the limits of our own knowledge.

Although many people take the concept for granted, KNOWLEDGE is a hard thing to define.

Philosophically, to *know* something you generally need three things: (1) for it actually to be true; (2) for you to believe it is true; and (3) for you to have learnt it in a reliable way.
The first requirement—that the think you think you know is actually true—is something we can’t ever really be sure of because The Truth is largely an unknowable philosophical mystery, in the same way that a perfect circle is an idea that doesn’t exist in the world.
Read 25 tweets
22 Dec 20
The past 12 months, I’ve heard a disturbingly large number of people, even respected senior scientists, ask the question: “how do I design a study to prove this fact I already believe is true?”

Um, you shouldn’t?!

Ask instead: “how do I design a study to learn *if* it’s true?”
Read 4 tweets
6 Dec 20
People often ask me if they should do some optional activity.

I work from home & only leave to exercise outside or run errands (eg groceries, take-out). This is all I’ve been doing since March.

So, no I don’t think you should do that thing. But if you must, please do it safer.
I miss my family, I miss my friends, I miss casual water cooler chats and awkward small talk, I miss getting lost in a crowd, I miss browsing craft & art markets, I miss IT NOT BEING A PANDEMIC.
I’m not wading in to the comments to argue about my choices, but let me just say, risk to *myself* is not my primary driver here.

I stay home because of risk to us *all*. I stay home because I can, and, by staying home, I make myself a firebreak for my contact network.
Read 5 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!