Doing statistics well is hard. Avoiding egregious errors in analysis, reporting and interpretation isn't. Just read, understand and follow the guidelines. pubmed.ncbi.nlm.nih.gov/32451178/ pubmed.ncbi.nlm.nih.gov/37286459/ pubmed.ncbi.nlm.nih.gov/40914655/ pubmed.ncbi.nlm.nih.gov/30580902/

Just submitted a paper to "Diabetes & Metabolic Syndrome: Clinical Research & Reviews". Was asked by editor *before* they would peer review: my h-index ; all publications in past 12 months, along with impact factor of each journal in which they were published". #RealScience

If only AI / ML had been around when I was training, I wouldn’t have had to learn about things like causal inference, how to evaluate prediction models or even, say, the importance of data quality. What a waste of time all that was!

Any suggestions as to great resources (lectures on YouTube, short didactic papers etc), to teach novice researchers about RCTs? Design, endpoints, consent, eligibility criteria, IRB etc etc any and all of it.

COULD EVERYONE PLEASE STOP USING RANDOM EFFECTS META-ANALYSIS WHEN COMBINING 3 OR 4 TRIALS? AND COULD REVIEWERS STOP DEMANDING IT?

www.auajournals.org/doi/10.1097/... Lymph node dissection (LND) for radical prostatectomy: controversy about RCT, complication rate. Decision analysis puts numerical estimates on benefit, harm, uncertainty. Expected utility of LND was higher vs. no PLND across broad range of scenarios.

Yet again, PRS do not differentially distinguish aggressive from indolent cancer. & BARCODE RCT showed poor results compared to MRI etc. Yet the authors give a thumbs up to genomics in prostate cancer screening. When is the PRS fever going to break? www.nature.com/articles/s43...

Too many meta-analyses have findings equivalent to: “If you average the cost of a loaf of bread, car insurance for a year and a movie ticket, you get $752.36”

1) A covariate that is predictive of outcome should be in the model even if unpredictive of assignment (eg matched pairs design). 2) A covariate that is not predictive of outcome should not be in the model, even if predictive of assignment. 3) The propensity score is stupid.

When discussing PSA screening policy, we often contrast two options as opportunistic vs. population-based PSA screening. Would suggest a name change to disorganized vs. organized PSA screening.

Number of papers on PubMed using the term "real world data" in 2000: 6. Number in 2025: ~5000. Number of papers for which "real world data" would be a meaningful scientific term: 0.

@amit_sud PRS-based prostate cancer screening has worse properties than contemporary approaches: "BARCODE1 biopsied more men, diagnosed more low-grade PCs & detected fewer high-grade PCs versus Göteborg-2 and ProScreen." authors.elsevier.com/sd/article/S...

ProQuant collaboration: >100 urologists, radiation oncologists, pathologists, radiologists, biostatisticians, & ML experts from 34 institutions worldwide evaluating whether & how tumor quantification offers superior risk stratification to Gleason score. www.sciencedirect.com/science/arti...

Lysenko is a bit of a bogeyman in science. But I have to say, rereading his story, hard not to draw parallels with Prasad, Makary, Bhattacharya and Hoeg. en.wikipedia.org/wiki/Trofim_...

Our guidance regarding performance measures for medical AI models is finally out! - Stop bashing AUROC, although it does not settle things - Calibration and clinical utility are key - Show risk distributions - Classification statistics (e.g. F1) are improper www.thelancet.com/journals/lan...

Working on a paper on informed decision making for PSA screening. Looking for any literature demonstrating that PSA screening is a preference sensitive decision and / or any data that these preferences can be accurately elicited in primary care. Please advise.

Machine learning has developed remarkable new ideas about how to develop prediction algorithms. But always baffled me why the field had to reinvent how to evaluate models. Here we show F score should not be used to evaluate medical prediction models link.springer.com/epdf/10.1186...

Five things you need to know about prostate cancer diagnostic tests www.sciencedirect.com/science/arti...

For those who complained about the recent RCT on PSA not looking at overall mortality, note that Gil Welch, the well-respected screening skeptic, concluded "It is not feasible to test all-cause mortality when screening for an individual cancer" pubmed.ncbi.nlm.nih.gov/37639251/

MRI is a great tool in prostate cancer. The problem is, quality in practice might not be quality we see in studies. Nice study from @dr_coops showing that in the VA, MRI not good enough to rule out biopsy. jamanetwork.com/journals/jam...

RCT n= ~1500. >90% of patients walking through the door with target indication were randomized. Zero funding. Clinical integration and rethinking informed consent transformational. journals.lww.com/anesthesiolo...

Gosh, who could have predicted that the men who complained that they were being "censored!" during the pandemic phase of Covid (Prasad, Bhattacharya) are now leading the nation's health agencies & CENSORING others? "staffers are terrified of pushing back on Prasad, lest they face retaliation"

Recently been debating "cancer screening skeptics" about endpoints in screening RCTs. They insist on overall survival (OS); when I point out feasibility issues, they say infeasibility of RCT on OS proves screening doesn't work. Prejudice masquerading as methodologic rigor.

Appreciated this editorial on the long term results from the prostate cancer screening trial. It's hard to believe we are doing the best we can be when it comes to this screening tool. Screening effectiveness needs to be re-evaluated over time as practices change. www.nejm.org/doi/full/10....

After my son’s team lost in semi-finals of frisbee nationals 2024, he wrote to team “I love you all. This is how we do better next year …”. 2025: national champions. Something for science to learn from the sporting world: respect and motivate your team and good things will happen.

Folks (Prasad etc) who say “cancer screening must improve overall survival”, let's take a real example, ovarian cancer. ~0.7% mortality; salpingectomy reduces risk by ≥50%. Trial could be powered on cancer mortality (n~20,000) vs. overall (n~750,000). Which would you recommend?

Whether / how to do PSA screening complex topic highly dependent on detailed knowledge. Plenty of room for reasonable disagreement. But I’m getting pretty tired of self-proclaimed skeptics who, without specialist knowledge, smugly condescend anyone supporting PSA as naive.

Very common to ask patients to rate pain from "no pain" (0) to "worst pain imaginable / possible" (10). Here is a randomized trial showing anchor for pain score of 10 should be "extreme pain" (TL;DR: we want to evaluate pain, not imagination) pmc.ncbi.nlm.nih.gov/articles/PMC...

Guidelines group argues against lymph node dissection in radical prostatectomy by cherry-picking outlying studies eg systematic review finds 5 studies on lymphedema only 1 with p<5%; authors cite that one study and ignore systematic review. www.europeanurology.com/article/S030...

All wrapped up in a neat little bow. Covid conspiracy theorist encourages her lymphoma afflicted daughter to refuse chemotherapy. The daughter dies. The mother blames paramedics for the death. www.theguardian.com/society/2025...

Large multicenter study: "When grade group discordant between systematic & targeted biopsy, risk is intermediate. Current approach assigning highest grade should be abandoned ... consider de-escalating treatment when grade discordant." academic.oup.com/jnci/advance...

My theory of medical evidence is simple: observational studies for me, RCTs for you.

Remarkable study from the Malmo Preventive Project: ~1000 men with blood at 60 in 1981; no PSA screening, *lifetime* follow-up. >50% life years lost to prostate cancer in PSA>4.0; 85% for PSA >2.0. Focus screening for men in 60s in men with higher PSAs! academic.oup.com/jnci/advance...

Now seen two papers claiming PLND not needed in a subgroup with relatively low incidence of positive nodes at radical prostatectomy. Neither cite the RCT showing benefit of PLND. When RCT says “do X”, don’t eyeball a case-series to conclude “don’t do X”.

This is big! Worked with great statisticians on guidelines for meta-analysis & systematic review. We discuss rationales for systematic review, evaluation & interpretation of heterogeneity, & common errors in network meta-analysis, funnel plots etc. www.europeanurology.com/article/S030...

Urologists do well with probability estimates vs. clinical scenarios. Goes against 25 years of me hearing "clinicians are too dumb to understand numbers, just tell them what to do". doi.org/10.1097/UPJ....

Thrilled to share our new @BJUIjournal study! 📊 Quantifying Gleason pattern 4 via MRI & targeted biopsy predicts prostate cancer metastasis better than standard Gleason grading & risk scores. Precision matters in risk stratification. 🔗 doi.org/10.1111/bju.... @vickersbiostats.bsky.social

No.

Folks discuss replication crisis as if something new. But remember Hans Eysenck (an OG scientific racist)? Both "Eysenck was the most frequently cited living psychologist" & "14 papers retracted, 60 statements of concern issued by journals" en.wikipedia.org/wiki/Hans_Ey...

No justification for arbitrary age cut-offs in clinical trials (e.g. excluding patients over 75) www.theguardian.com/society/2025...

A comment on some invalid criticisms of the radical prostatectomy lymph node dissection trial www.sciencedirect.com/science/arti...

The Gleason score - the fundamental determinant of how we chose on treatments in localized prostate cancers - basically doesn't make any sense. grandroundsinurology.com/away-from-th...

I keep hearing "AI models have to be explainable otherwise doctors and patients won't trust them". But if I Google "does explainability improve trust in AI?", I get an AI summary (short answer "yes") created by an AI process that probably <1% of the population understands.

Results of observational study almost perfectly match those of our RCT. Hazard ratios of 1.07 / 0.72 for BCR / mets; RCT reported 1.05 / 0.82. Conclusions: do extended PLND during radical prostatectomy; effects of PLND occur after BCR. www.auajournals.org/doi/pdf/10.1...

You can have all the omni-omics data in the world and the bestest algorithms, but eventually a predicted probability is produced & it should be evaluated using well-established methods, and correctly implemented in the context of medical decision making. statsepi.substack.com/i/140315566/...

There is no scientific concept of "real world" data or evidence - there is nothing about this terminology that can't be more accurately and honestly conveyed using other accepted methodological terms.

With my kids after Robin (bottom left) wins Gold at U24 Ultimate Frisbee World Championships

When I first started covering Kate Shemirani during the acute phase of the pandemic I said her medical disinformation would kill people. Tragic that her own daughter is one of her victims - arrogant, narcissistic conspiracy theorists kill innocent people www.bbc.com/news/article...