@floriannaudet.bsky.social
@paulglasziou.bsky.social
@fzores.bsky.social

Meta-analyses of biased RCTs give biased results, even on individual data We strongly disagree with Guo and colleagues,1 who suggest an overall benefit from blood pressure targets of less than 120 or 130 mm Hg. It would be surprising for this conclusion to be true given tha...

www.thelancet.com/journals/lan...

By contrast, in HOPE-3,6 a powerful double-blind RCT of dual BPLD in patients with a majority of mild hypertension, no benefit was seen.

These biases explain a substantial decrease of serious adverse events in CRHCP;3 the other RCTs did not show any decrease (odds ratio 1·02, 95% CI 0·97–1·06).

Similar biases led to the exclusion of the Hypertension Detection and Follow-up Program4 from all recent meta-analyses of BPLD, after erroneous inclusion in early meta-analyses.

First, these results must not be extrapolated to different health systems. Second and most important, such a comparison involves a lot of differences other than BPLD, definitely confounding the results and making them impossible to attribute to BPLD.

The authors' analyses are based on six open-label RCTs, prone to lack-of-blinding bias. Notably, the largest included trial, CRHCP,3 compared two health systems, its intervention involving non-physician health-care providers, health coaching, and free drugs for hypertensions

It would be surprising for this conclusion to be true given that there is currently no randomised controlled trial (RCT) evidence demonstrating the benefit of blood pressure lowering drugs (BPLD) in mild hypertension (untreated systolic blood pressure 140–160 mm Hg)

Benefit–harm trade-offs of intensive blood pressure control versus standard blood pressure control on cardiovascular and renal outcomes: an individual participant data analysis of randomised controlle... Compared with standard blood pressure control, intensive blood pressure control provides a net benefit between the reduction in cardiovascular events and the increase in adverse events, including rena...

We strongly disagree with Guo and colleagues,
who suggest an overall benefit from blood pressure targets of less than 120 or 130 mm Hg.
www.thelancet.com/journals/lan...

Good science is not only about data: are you asking the right question? We have built an extraordinary scientific machine. We can capture data at scale, store data cheaply, and analyse data in ways that were unthinkable even a

Re-analyses of existing data can be super important, as the RIAT initiative shows

Guideline or Evidence based medicine...that is the question

Childhood Cholesterol Screening Is Not Cost-Effective Childhood cholesterol screening is controversial. For decades, clinician-scientists have disagreed on the questions of whether, when, and how to screen for hypercholesterolemia in youths. For example,...

@fzores.bsky.social
@docdu16.bsky.social
jamanetwork.com/journals/jam...

@fzores.bsky.social
www.bmj.com/content/388/...

Inherent difficulties with active control equivalence studies A simple model is used to investigate the relevance of ‘competence’ to active control equivalence studies )ACES(. It is shown that to the extent that such trials are successful the results of such tr...

See onlinelibrary.wiley.com/doi/abs/10.1...
and www.senns.uk/You_may_beli... p56

Thank you. I'll read this carefully.

Falsificationism and clinical trials - PubMed The relevance of the philosophy of Sir Karl Popper to the planning, conduct and analysis of clinical trials is examined. It is shown that blinding and randomization can only be regarded as valuable for the purpose of refuting universal hypotheses. The purpose of inclusion criteria is also examined. …

Dear Stephen
@stephensenn.bsky.social
How do you reconcile Bayesianism with falsificationism ?
pubmed.ncbi.nlm.nih.gov/1792462/

6. In practice, therefore, treating patients with a systolic blood pressure of 140 mmHg has not been shown to provide a clear (without the risk of biased results)... for patients

5. I conclude that the lack of blinding explains the difference... This may be due (though not certain, since strokes were not reduced) to a larger BP difference in the open-label group..., which is not the specific effect of the medications...

4. And HOPE 3 is the only trial in which patients did not receive a drug at baseline

3. In double-blind RCTs, no benefit is observed with an average of one additional medication in the intervention group and a BP difference of around 3 or 4 mmHg

2. For the same baseline blood pressure of 140 mmHg:

In some open-label RCTs, benefits are observed with an average of one additional medication in the intensive treatment group and a blood pressure difference of around 15 mmHg.

1. A comment about this important meta-analysis :
bmjopen.bmj.com/content/9/9/...

Nice to see that our review on fluoxetine and the correspondence it triggered was selected for the editor's choice section in @jclinepi.bsky.social
www.jclinepi.com/article/S089...
@markhoro.bsky.social @richlyus.bsky.social @joannamoncrieff.bsky.social @rboussageon.bsky.social

Instead of making decisions based on the best available evidence, shouldn’t we make decisions based on the necessary evidence (which implies not acting in the absence of evidence defined in this way)?

Levels of evidence supporting American, European and international guidelines in psychiatry, 2014–2024: a systematic review with quantitative synthesis Question To what extent are psychiatry guidelines supported by high-level evidence?Study selection and analysis Guidelines from the American Psychiatric Association, European Psychiatric Association, ...

mentalhealth.bmj.com/content/29/1...
The guidelines are therefore consistent with EBM if they have conducted a systematic review and make recommendations based on the" best available evidence", even if that evidence is of low quality..
And that is what I am criticizing:

Of course, there was James Lind and his comparative experiment, but the evaluation of animal magnetism in Paris in 1784 marked a turning point for double-blind testing

Kaptchuk says it (placebo)all started in the Middle Ages, when genuine and fake cases of possession were exposed by holy water
www.thelancet.com/journals/lan...

Unless it can be demonstrated that an action is beneficial, and until proven otherwise, the action is harmful...
This is what justifies 250 years of evaluating treatments (since Mesmerism). Otherwise, there’s no point in testing anything...

Thank you.
But primum non nocere...
And that’s not from Confucius.😉

Thank you very much for your response.
And how far are you willing to compromise on the quality or certainty of the information you provide to your patient?
Observational studies? In vitro studies? Testimonials?