Combining risks and benefits - summing it up.

Insights and discussion from the cutting edge with reference to journal articles and other research papers.
Post Reply
Fiver
Senior Contributor
Senior Contributor
Posts: 629
Joined: Wed Feb 01, 2017 12:51 pm

Combining risks and benefits - summing it up.

Post by Fiver »

I've been thinking a lot lately about how risks and benefits combine. But I'll admit to having no idea how to start answering this question.

For example, let's say there is a combination of risk factors - e.g., e4 genotype, standard American diet, diabetes, and some past head trauma. Are those risks additive? Seems that they would be. But the diet and diabetes are obviously related. And the genotype might make the head trauma worse. And at some point the risks, if there are enough of them, would add up to more than 100%, which can't be.

On the other hand, let's say someone adopts several lifestyle improvements - e.g., they follow a good anti-inflammatory diet, exercise, and reduce stress. We's assume that doing all three is better than doing any one alone. Again, they are clearly inter-related. And, in theory, maybe one could by math alone lower their risk way below "normal" or even below 0%....which can't be right either.

Maybe the first big change has a larger effect than the next, and the next....this makes sense. Diminishing returns.

....or do they combine in a synergistic way? This also makes sense.

.....or is it a matter of tipping points (which is a popular view)? Once the balance has tipped the rest of the risks/benefits don't change much.

It seems logical that the math would depend on the risks / benefits of each individual. But statisticians must have figured this out for groups. Anyone know of articles / books I could read about how this is applied in medicine or for AD specifically?

Practically speaking - is there enough information for us to even tease apart the benefits of one factor vs another?

In general, how does one properly combine % risks, HRs, ORs, etc?

Thanks.
User avatar
Stavia
Contributor
Contributor
Posts: 5255
Joined: Tue Apr 29, 2014 6:47 pm
Location: Middle Earth

Re: Combining risks and benefits - summing it up.

Post by Stavia »

Tom, members have asked this question many times over the years. I don't know of any robust work in this area. Plus we don't know our individual baseline risk. Also, remember all ORs have a 95% confidence interval, so its actually a range, not a definite number.
I've given up trying to figure it out years ago, I'm just doing my best. I personally believe its not a tipping point yes/no, but I am hoping that strategies will push out the AD until I die of something else first.
Fiver
Senior Contributor
Senior Contributor
Posts: 629
Joined: Wed Feb 01, 2017 12:51 pm

Re: Combining risks and benefits - summing it up.

Post by Fiver »

Thanks Stavia. I agree in terms of just doing our best. What an adventure!

I think I will do some research on this - just out of curiosity - and report back.

It doesn't seem like we have enough information yet to split hairs about how things do or do not combine. You're right about the error bars - they are often pretty large. And combining data sets, each with error terms, propagates the errors so much the data would quickly lose all meaning. It gets fuzzy fast. You just taught me some things about HRs and I was so interested to add them all up in my head that I (conveniently!) forgot we'd need to add up the error terms. Ugh.

I'm starting to see how, even with all the necessary data, this is more about some larger personal view about how one lives life.

Having said that.....I'm still curious....for other medical issues, or risks more generally, there must be a whole tool box of techniques available. Let me see what I can find out. :)
Searcher
Contributor
Contributor
Posts: 194
Joined: Thu Oct 26, 2017 5:42 am

Re: Combining risks and benefits - summing it up.

Post by Searcher »

Tom,

The Global Burden of Disease project (funded by the Gates Foundation) attributes the total number of premature deaths or disabilities to a variety of causes and risk factors.

http://www.healthdata.org/data-visualiz ... bd-compare

This avoids double-counting. I think they have copious descriptions of their methods.
Fiver
Senior Contributor
Senior Contributor
Posts: 629
Joined: Wed Feb 01, 2017 12:51 pm

Re: Combining risks and benefits - summing it up.

Post by Fiver »

good idea! ...on it.
User avatar
Stavia
Contributor
Contributor
Posts: 5255
Joined: Tue Apr 29, 2014 6:47 pm
Location: Middle Earth

Re: Combining risks and benefits - summing it up.

Post by Stavia »

Tom wrote:good idea! ...on it.
let us know Tom!! enjoy your data delving
Post Reply