Generate basic statistics on SDH distribution for MU maternal cohort

gpcnetwork / r21_drivers_mu

All material about NIH/NIMDA funded project on "DRIVERs: Data systems Research to Identify driVers of Ethnic & Racial Inequities in Maternal Mortality"

0 stars 0 forks source link

Generate basic statistics on SDH distribution for MU maternal cohort #1

Open sxinger opened 3 months ago

sxinger commented 3 months ago

DEID_SDOH_DB.SSDH.SSDH_SEL: table with selected structural social determinants of health

spinkac commented 2 months ago

@sxinger I have the MU data, including SDOH ready to do some beginning logistic regressions. How would you like to group the predictor variables like race/Hispanic/ruca? Should age be grouped (as we did earlier) or treated a continuous and mapped with linear/quadratic/cubic, etc? I ran a couple of models out of curiosity, and found that medicaid and age are highly significant, but Hispanic was not and race was not if SES was included.

sxinger commented 2 months ago

@spinkac

How would you like to group the predictor variables like race/Hispanic/ruca?

For RACE, we can group the very rare races together into OT category For RUCA, we can follow this grouping rule: https://depts.washington.edu/uwruca/ruca-maps.php

Should age be grouped (as we did earlier) or treated a continuous and mapped with linear/quadratic/cubic, etc?

I think we can experiment a bit on this, and see which way would give us a better fitted model.

I ran a couple of models out of curiosity, and found that medicaid and age are highly significant, but Hispanic was not and race was not if SES was included.

Sounds like a good start. Can you also include ADI?

spinkac commented 1 month ago

Things to do:

[ ] Add BMI and Smoking Status (VITAL/OBS_CLIN)
[ ] Do Univariate analysis of ACS variables to see which ones correlate with morbidity
[ ] Look to see if disparities are increasing/decreasing/no change over time. Maybe look at either 5 year window or consider linear effect of time on disparity per racial group