As an editor I often see authors fall into a trap of stressing some fancy statistical technique for estimating causal effects. Often neglected are the necessary assumptions to map the data to the estimand. @pausalz.bsky.social and I wrote about it here www.bmj.com/content/391/... #EpiSky

New paper in @jclinepi.bsky.social: **Estimands: what they are and why we should use them** We provide simple explanations of: -estimands and intercurrent events -how these relate to familliar concepts like ITT -how to write estimands for protocols -and more www.jclinepi.com/article/S089...

Just a reminder about the new all-island Irish Reproducibility Network. There are already lots of great people involved. You can sign up and find local site leads with the link, and if you are based in Cork feel free to reach out to me directly. irishrn.org

NEW PAPER: "Reporting guidelines for studies involving generative artificial intelligence applications: what do I use, and when?" --> www.nature.com/articles/s41... #HealthTech #ClinicalAI #MachineLearning #MedAI #AIinMedicine #TransparentAI #HealthInnovation #GenAI

The term #estimand is now part of the technical language of trials - but it remains a concept that many find hard to understand. Recent work has aimed to make the term more accessible to researchers, patients & public 1/8 #MethodologyMonday #127

NEW paper "Clinical prediction models using machine learning in oncology: challenges and recommendations" --> tinyurl.com/2h2mcy5h #machinelearning #prediction #statssky #MLsky #methodsmatter

📢 New publication in #Trials led by @suziecro.bsky.social with the HEALTHY STATS public involvement group: Supporting public involvement in defining estimands: a practical tool accessibly explaining the five key attributes of an estimand trialsjournal.biomedcentral.com/articles/10....

We're excited to share our latest work on estimands in clinical trials! 📄 Estimating treatment effects in trials with outcome data truncated by death: A case study on aligning estimators with estimands. 🔗 Read the full paper: journals.sagepub.com/doi/10.1177/...

Want to come work on SNAP, BALANCE+, and STRAP? And change how we treat and manage bloodstream infections? See below. #IDSky

📢 Interested in a PhD in clinical trials methodology? The Trials Methodology Research Partnership are advertising projects for PhDs starting Oct 2026, in statistics, health economics, epidemiology, social science, and more, hosted at one of 12 UK partner institutions. mrctmrpdtp.com/current-oppo...

📢 As one of the next round of TRMP PhD students (below) you could join me, Clare Robinson, @averoniki.bsky.social and @marionkcampbell.bsky.social to work on developing a typology of research outputs in trials methodology, based at @wiphcem.bsky.social. Details here: www.findaphd.com/phds/project...

🎓 Thinking about a PhD in clinical trial #methodology? We have two funded PhD opportunities for students passionate about trial design and statistical methods: • One in estimands • One in non-inferiority trials 🔗 More info and how to apply here: buff.ly/lCpqcF6 @mrcnihrtmrp.bsky.social

New PhD position available at @mrcctu.bsky.social to develop guidance on balancing statistical and clinical considerations when choosing an estimand in RCTs. www.findaphd.com/phds/project...

The recording of this webinar is now available on the @nihr-rss.bsky.social YouTube channel: www.youtube.com/watch?v=4cNz...

Lots of feedback and engagement after this session. Some gentle push-back still: "But what if there's even just a *perception* from sites that a stepped wedge is preferable?" 🧵

A scoping review identified additional considerations for defining estimands in cluster randomised trials - Journal of Clinical Epidemiology www.jclinepi.com/article/S089...

Introductory online INSPECT-SR workshop. November 6th, 12-2pm UK-time. Free, places limited. BOOK: www.trybooking.com/uk/events/la...

The number needed to treat: It is time to bow out gracefully - Journal of Clinical Epidemiology www.jclinepi.com/article/S089...

There has been a lot of debate recently about the promise of real world data - the routine (observational) data collected on patients eg  treatments received, clinical outcomes etc – for estimating treatment effects. But can they deliver? 1/9 #MethodologyMonday #123

Improving the design of staggered implementation cluster randomised trials qualitysafety.bmj.com/content/earl...

We often now hear the term “Bayesian borrowing” in trials. But what is Bayesian borrowing and what are the pros and cons around its use? 1/6 #MethodologyMonday #122

NEW PAPER in @bmj.com "Dealing with continuous variables and modelling non-linear associations in healthcare data: practical guide" --> www.bmj.com/content/390/... #methodologymatters #StatsSky #EpiSky

What's that? The target trial emulation obtained the same answer as the RCT? Wow, I guess RCTs are generalisable after all! Perhaps we should perform them rather than using interventions on hundreds of thousands of people before we know whether or not they work, then doing a TTE?

Reckon I can submit this as a figure for a paper?

New post: “Marginal” and “conditional” are relative descriptions of an estimand

New paper posted on Arxiv: "When do composite estimands answer non-causal questions?" This can happen more often than you think, and can have a dramatic impact on trial results (e.g. a false-positive rate of almost 90%) arxiv.org/abs/2506.22610 @timpmorris.bsky.social

The CONSORT 2025 Statement is here. Essential reading for those involved in the reporting & reviewing of randomized trials. We're proud to see so many #UKCTUNetwork units & individual researchers contributing to this crucial update! 🔗 tinyurl.com/yc3cbh33 #CONSORT2025 #ResearchTransparency

Would like to add… there are guidelines for all types of studies… www.equator-network.org/reporting-gu...

Please people. It's not hard. In fact, it makes your life easier. And when you don't use CONSORT, you look like a novice.

Authors, editors, reviewers, and other potential users should use CONSORT 2025 when writing and evaluating manuscripts of randomized trials to ensure that trial reports are clear and transparent. Learn more in this Special Communication. https://ja.ma/3TlnKQe

Stepped wedge devotees step up, and cluster enthusiasts get ready to cluster: it's conference time again.

🆕New CONSORT & SPIRIT 2025 Guidelines Our colleagues helped shape these updates, making #ClinicalTrial reporting more transparent & useful for everyone. Key changes: ✔️More open science ✔️Better reporting of harms & missing data ✔️Greater harmonisation between guidelines More info: buff.ly/HC1rx31

NEW PAPER "Importance of sample size on the quality and utility of AI-based prediction models for healthcare" www.thelancet.com/journals/lan... We discuss seven reasons why inadequate sample size negatively affects model training, evaluation, and performance. #StatsSky #machinelearning #AI

One of the trial designs that has been cropping up in my discussions recently is the cluster randomised crossover, or CRXO, design. But when is it best used & what are the key issues to look out for? 1/7 #MethodologyMonday #119

New 3-post for a research/senior research fellow to work on adaptive longitudinal cluster randomised trials in Melbourne. Interesting work with a great research group. careers.pageuppeople.com/513/cw/en/jo...

Next Thursday, June 5th at 1pm (BST) we host our 4th webinar in our adverse event special series. Rayan Taher & Jenny Yiend will present their work on developing a process for assessing the safety of a digital mental health intervention & gaining regulatory approval. Register here: lnkd.in/eaW3MQrB

Hot off the press: bmcmedresmethodol.biomedcentral.com/articles/10.... In this we confirm empirically some theoretical results about using different summary measures in non-inferiority trials with survival outcomes. tldr: RMST beats HR.

Read the original paper here: trialsjournal.biomedcentral.com/articles/10....

Next week (May 22nd) we host our third webinar in our adverse event special series. Regina Stegherr will be discussing the SAVVY project which aims to improve the analyses of AEs in clinical trials using survival methods to handle varying follow-up & competing events. Register here: lnkd.in/eaW3MQrB

In 2023 Trials invited submissions that related to the collection, analysis & reporting of adverse events in clinical trials. In this editorial @vcornelius.bsky.social & I summarise the submissions & discuss the implications on applied clinical trials & future methodological research rdcu.be/elLFa

A Consensus Statement in Nature Medicine from CONSORT 2025 provides updated guidance to authors, reviewers and editors, when writing and evaluating manuscripts of randomized trials to ensure that trial reports are clear and transparent. go.nature.com/3Si0p1f #medsky 🧪

This week saw the publication of SPIRIT 2025, a major update to the international SPIRIT Statement which provides guidance on what should be included in a trial protocol. 1/6 #MethodologyMonday #117

The CONSORT and SPIRIT statements for randomized trials have been updated for 2025 and published in JAMA, with a new focus on open science principles and guidance on trial registration, data access, and conflict of interest reporting. ➡ CONSORT 2025: ja.ma/3Yyu0Hh ➡ SPIRIT 2025: ja.ma/3YjFfn8

Estimands are a way to clearly define the research question in #ClinicalTrials. This animated video explains how using estimands can help doctors and patients make informed treatment decisions.▶️👇 buff.ly/xiz3mU6 #StatsSky

I call (3) the ‘realisability criterion’. Ok, so people did/do not adhere. Why? Could you tell who would not adhere in a future population and only offer the intervention to them? Or could you make everyone adhere? Time for a shameless plug @brennankahan.bsky.social www.bmj.com/content/384/...

Why? For (1) & (2), it’s a sleight-of-hand to claim an estimation method is diluted/biased compared to the ‘true’ effect of treatment if you have not said what true effect you think people should be estimating. There is usually more than one reasonable estimand.

Oh good ‘ITT analyses are biased’ is back, here using the euphemism ‘diluted’ 🙃 The rules are that you have to: 1. say what your estimand is before saying an analysis is ‘diluted’/‘biased’ 2. admit that there are other reasonable estimands 3. describe how it could be realised in a future population

NEW PAPER: Updated SPIRIT-2025 for reporting protocols of randomised trials is now available in the @BMJ.com, @JAMA.com, @thelancet.bsky.social, @PLoS.org, and @natmedicine.bsky.social. —> www.bmj.com/content/389/... #openscience #transparency #medsky #statssky #episky

SPIRIT 2025 guidelines for protocols now published in @bmj.com journal and others www.bmj.com/content/389/...

HIRING! We offer 3 fully funded PhD positions to work on research methodology with experts in a great medical research environment. More info: www.linkedin.com/feed/update/...