Michael Baum
@MichaelBaum11
Professor emeritus of surgery and visiting professor of medical humanities University College London. Artist, novelist, suffers from scepticaemia.
28-05-2014 14:54:12
9,6K Tweets
2,2K Followers
471 Following
Doug McKechnie Andrew Vickers Paul Pharoah (R soul) πͺπΊπ΅π¬πΏπ¦π¬π§ Ash Paul Susan Bewley Steven Laitner Naser Turabi It seems to make sense but the problem is the deaths from over diagnosis and over-treatment. Yes small numbers but so might be the the numbers claimed for screen detected PC.
Michael Baum Doug McKechnie Paul Pharoah (R soul) πͺπΊπ΅π¬πΏπ¦π¬π§ Ash Paul Susan Bewley Steven Laitner Naser Turabi Overall survival (OS) terrible endpoint for screening RCT. ~zero health interventions have been shown to improve OS. This for statistical reasons. Hard to show reduction in cancer mortality from eg 3% to 2.5%, but doable. ~impossible to show improvement in OS from 69.5% to 70%
Michael Baum Doug McKechnie Paul Pharoah (R soul) πͺπΊπ΅π¬πΏπ¦π¬π§ Ash Paul Susan Bewley Steven Laitner Naser Turabi Sample sizes. 3% to 2.5%: about 50,000. 69.5% to 70%: ~350,000
Andrew Vickers Michael Baum Doug McKechnie Paul Pharoah (R soul) πͺπΊπ΅π¬πΏπ¦π¬π§ Ash Paul Susan Bewley Steven Laitner Naser Turabi If 350,000 people are needed to show a benefit, then we have to question the validity of screening...
What will be the number need to screen?
Vinay Prasad MD MPH
Boussageon RΓ©my Michael Baum Doug McKechnie Paul Pharoah (R soul) πͺπΊπ΅π¬πΏπ¦π¬π§ Ash Paul Susan Bewley Steven Laitner Naser Turabi Vinay Prasad MD MPH The number needed to screen to save a life is the same in both settings. To restate, no screening test, zero, has ever (or likely ever will) show an effect on overall mortality. This is simply a math problem.
Andrew Vickers Michael Baum Doug McKechnie Paul Pharoah (R soul) πͺπΊπ΅π¬πΏπ¦π¬π§ Ash Paul Susan Bewley Steven Laitner Naser Turabi Vinay Prasad MD MPH OK. How is it that some drugs reduce specific and total mortality at the same time?
See simvastatin in HPS and ramipril in HOPE...?
Boussageon RΓ©my Andrew Vickers Michael Baum Doug McKechnie Paul Pharoah (R soul) πͺπΊπ΅π¬πΏπ¦π¬π§ Ash Paul Susan Bewley Steven Laitner Naser Turabi My question to Andrew is:
1. Is it possible that a screening test comes out that has off-target deaths that are difficult to quantify (bc they span domains) but in total negate deaths averted from screening?
2. If so, how would you discover this?
Vinay Prasad MD MPH Andrew Vickers Michael Baum Doug McKechnie Paul Pharoah (R soul) πͺπΊπ΅π¬πΏπ¦π¬π§ Ash Paul Susan Bewley Steven Laitner Naser Turabi Andrew, thank you for your answer...
twitter.com/VPrasadMDMPH/sβ¦
Vinay Prasad MD MPH Andrew Vickers Michael Baum Doug McKechnie Paul Pharoah (R soul) πͺπΊπ΅π¬πΏπ¦π¬π§ Ash Paul Susan Bewley Steven Laitner Naser Turabi How do you interpret this result in the NELSON trial?
Boussageon RΓ©my Vinay Prasad MD MPH Andrew Vickers Michael Baum Doug McKechnie Paul Pharoah (R soul) πͺπΊπ΅π¬πΏπ¦π¬π§ Susan Bewley Steven Laitner Naser Turabi Nelson was underpowered to detect all-cause mortality
ncbi.nlm.nih.gov/pmc/articles/P⦠via David Baldwin et al
Ash Paul Boussageon RΓ©my Vinay Prasad MD MPH Andrew Vickers Michael Baum Doug McKechnie Susan Bewley Steven Laitner Naser Turabi David Baldwin All screening trials are underpowered for ACM. The point is if you only include cause specific mortality the off target harms are effectively ignored. E.g. over treatment of breast cancer with radiotherapy causes death from heart disease.