study design question

[c***@gmail.com]

Hi,
I would like to design a study that takes a large dataset and looks at mortality as an endpoint, in the regression analysis, I would like to control for patient specific covariates such as smoking status, etc. The difficult part of my analysis is that I would like to break down the factors contributing to mortality yearly across a 10 year span.

Any help would be appreciated.

[Ki]

Post by c***@gmail.com
Hi,
I would like to design a study that takes a large dataset and looks at mortality as an endpoint, in the regression analysis, I would like to control for patient specific covariates such as smoking status, etc. The difficult part of my analysis is that I would like to break down the factors contributing to mortality yearly across a 10 year span.
Any help would be appreciated.

I think it really requires more information to get useful answers. Like is this a longitudinal data collected for 10 years? or this is a data collected for 10 years from different people with no repeated measure? What is your sample size? How many covariates do you have? Missing data? Is this 'large dataset' a nationally representative data? Or is this a clinical data from a hospital network?

If you are more specific, you are more likely to get better answer.

[David Marso]

Post by Ki

Post by c***@gmail.com
Hi,
I would like to design a study that takes a large dataset and looks at mortality as an endpoint, in the regression analysis, I would like to control for patient specific covariates such as smoking status, etc. The difficult part of my analysis is that I would like to break down the factors contributing to mortality yearly across a 10 year span.
Any help would be appreciated.

I think it really requires more information to get useful answers. Like is this a longitudinal data collected for 10 years? or this is a data collected for 10 years from different people with no repeated measure? What is your sample size? How many covariates do you have? Missing data? Is this 'large dataset' a nationally representative data? Or is this a clinical data from a hospital network?
If you are more specific, you are more likely to get better answer.

INDEED...
I was going to ask WHAT makes this 'The difficult part of your analysis...'
Please define the issues and be specific about what you mean by
"I would like to break down the factors contributing to mortality yearly across a 10 year span."

Have you considered hiring a statistical consultant with subject knowledge expertise to help guide you between Scylla and Charybdis?

Since you are in the 'design phase' you would do well to CYA so you don't waste a lot of your time and other's time engaging in fruitless pursuit.

[Rich Ulrich]

Post by Ki

Post by c***@gmail.com
Hi,
I would like to design a study that takes a large dataset and looks at mortality as an endpoint, in the regression analysis, I would like to control for patient specific covariates such as smoking status, etc. The difficult part of my analysis is that I would like to break down the factors contributing to mortality yearly across a 10 year span.
Any help would be appreciated.

I think it really requires more information to get useful answers. Like is this a longitudinal data collected for 10 years? or this is a data collected for 10 years from different people with no repeated measure? What is your sample size? How many covariates do you have? Missing data? Is this 'large dataset' a nationally representative data? Or is this a clinical data from a hospital network?
If you are more specific, you are more likely to get better answer.

Right. Also -

My first thought is that the data would be a censored life-table -
Not everyone suffers "mortality" but there is 10 years of data.

What sounds particularly strange is, "factors contributing to
mortality yearly across a 10 year span." If there are to be
different factors, I imagine that there almost has to be different
causes-of-death, i.e., which cancer, heart disease, etc. - and
COD would be available, and should be looked at first.