User talk:Psarka

Hi! Psarka (talk) 14:33, 9 May 2018 (UTC)[reply]

What a mess et al: proposed changes

Hi! Thanks for leaving me a message about your proposed changes. I was only involved in editing these pages during my studies, and I´m no longer actively involved in the topic, though I still watch them. And my background is STEM studies, although I´m now in medicine - whereas it seems your background is Epidemiology.

As for the question about notation/signs: the problem with your cited books (as do most epidemiology books and published papers) is that they obviously are relying heavily on context and based on this context change the formulas around to avoid negative numbers. While you can technically do this, it´s not very nice from a mathematical point of view as you always need the text along with the numbers. For example, when studying the effect of NSAIDs you could in a study look at cardiac risk (risk reduced, therefore NNT) and gastric ulcer incidence (risk increased, therefore NNH), and therefore in the same paper switch effect and control group depending on context just to avoid negative numbers. This kind of paper obviously would never be published (as you could produce 2 separate papers from the same data) - and who´s interested in pure mathematics, anyway ;) So what they do is they calculate the numbers, find that they´re negative, then go back and switch to a different formula (if they ever cite one) and then publish this with a text explaining that it´s for example "a risk reduction". So for your cited books in regard to the proposed changes to NNT, EAI p.59 is the mathematically most correct book, though it´s the minority of your references.

So, asides of the differences in view of the math (where I believe a context independent approach is useful), I believe your other proposed/planned changes are a good addition.

BR, Tom T.pienn (talk) 10:33, 18 August 2018 (UTC)[reply]

@T.pienn: Thanks for your feedback! Reading what you wrote and rereading the EAI, I realized that there is a pretty nice solution to the sign flipping conundrum: it is to have *three* versions of the metric: generic signed one, and two directional unsigned ones. Say "absolute risk change", "absolute risk reduction", and "absolute risk increase". The "reduction" and "increase" ones are frequently used by researchers which deal with one direction only (pills mainly reduce risks, pollution & unsocial behavior mainly increase risks), and they will continue doing so with reciprocals NNT and NNH fitting them well. More generic, signed one, can be used when direction is not assumed. The "risk difference" as defined in EAI p.59 is exactly the generic signed one, as they have it both positive and negative in the text.

This separation into "difference", "reduction", and "increase" also works well semantically, so I'm starting to have a good feeling about this :) By the way, maybe you could share some of the main references of your field (authoritative dictionaries or "bible" textbooks), I'd like to cover as much ground as I can. Psarka (talk) 11:45, 18 August 2018 (UTC)[reply]

Your submission at Articles for creation: Attributable fraction for the population has been accepted

Attributable fraction for the population, which you submitted to Articles for creation, has been created.

You are more than welcome to continue making quality contributions to Wikipedia. If your account is more than four days old and you have made at least 10 edits you can create articles yourself without posting a request. However, you may continue submitting work to Articles for Creation if you prefer.

If you have any questions, you are welcome to ask at the help desk.
If you would like to help us improve this process, please consider leaving us some feedback.

Thank you for helping improve Wikipedia!

Brad v 13:53, 3 June 2018 (UTC)[reply]

Ways to improve Preventable fraction for the population

Hi, I'm Dan Koehl. Psarka, thanks for creating Preventable fraction for the population!

I've just tagged the page, using our page curation tools, as having some issues to fix. Please categorize

The tags can be removed by you or another editor once the issues they mention are addressed. If you have questions, you can leave a comment on my talk page. Or, for more editing help, talk to the volunteers at the Teahouse.

Dan Koehl (talk) 17:47, 23 September 2018 (UTC)[reply]