Medicaid and mortality

March 21, 2017 at 9:16 pm 3 comments

| Gabriel |

This morning Spotted Toad picked up on the point in Quinones that a lot of pill mills were funded through Medicaid fraud and so he used Medicaid expansion under Obamacare to see if this led to greater drug overdoses in Medicaid expansion states. In fact he found that in the time since the Medicaid expansion, states that participated in the expansion had faster growth in overdose deaths than states that refused Medicaid expansion. That’s interesting, but I never want to base a trend on just two time points. (FWIW, Toad was analyzing the data as the CDC presents it — the analysis below requires a lot more queries). So I queried the CDC data in more granular detail to check if the trend started with Medicaid expansion. (saved query link, just iterate over year to get annual state-level OD deaths).

As it turns out, I was able to replicate Toad’s finding that Medicaid expansion states (blue) have higher rates and faster growth in fatal drug overdoses than Medicaid holdout states (red), but the two groups of states diverged starting in 2010, well before states began implementing Obamacare’s Medicaid expansion. So there may be a real difference between Medicaid expansion states (which are generally Democratic) and Medicaid holdout states (which are Republican), and the difference may even be some aspect of health policy, but it wasn’t Obamacare Medicaid expansion as the divergence starts too early. (It’s worth noting that Toad updated his own post with my graph as soon as I sent it to him).

medicaidod

Here is the data in Stata format (which you can reconstruct yourself from a series of CDC queries).

Here is the code

cd "~/Documents/codeandculture/cdcdrugmortality"
*https://wonder.cdc.gov/controller/saved/D76/D11F702


clear
gen state=""
gen year=.
save drugs19992015.dta, replace

forvalues i=1999/2015 {
 disp "`i'"
 insheet using drugs`i'.txt, clear
 append using drugs19992015.dta, force
 recode year .=`i'
 save drugs19992015, replace
}
drop if state==""

desc
sum

insheet using medicaidholdouts.txt, clear
ren v1 state
save medicaidholdouts.dta, replace

use medicaidholdouts, clear
merge 1:m state using drugs19992015
ren _merge medicaidexpansion 
recode medicaidexpansion 2=1 3=0

list state medicaidexpansion if (state=="Texas" | state=="California") & year==2010

save drugs19992015, replace

collapse (sum)deaths population, by (year medicaidexpansion)
gen cruderate= deaths/population
twoway (line cruderate year if medicaidexpansion==1) (line cruderate year if medicaidexpansion==0) , legend(order(1 "Medicaid expansion" 2 "Holdouts")) ytitle(Weighted Crude Rate of Fatal Drug Overdoses)
graph export medicaidod.png, replace

*have a nice day

And here’s my list of Medicaid holdouts:

Alabama
Florida
Georgia
Idaho
Kansas
Mississippi
Missouri
Nebraska
North Carolina
Oklahoma
South Carolina
South Dakota
Tennessee
Texas
Utah
Virginia
Wisconsin
Wyoming

 

Advertisements

Entry filed under: Uncategorized.

They Meant Us No Harm, But Only Gave Us the Lotus

3 Comments Add your own

  • 1. Michael  |  March 22, 2017 at 6:33 pm

    Can you test if the rates are different? I want to see some confidence intervals.

    Reply
    • 2. gabrielrossman  |  March 23, 2017 at 8:08 am

      This is easier said than done since we’re talking about a ratio based on underlying counts. My best attempt at it is to bootstrap by state then calculate the crude mortality rate. When I do so it looks like most of the years the point estimate for one type of state (expansion vs non-expansion) falls in the CI for the other, but as of 2015 they finally diverge. Of course this doesn’t mean that 2015 is the key year since if anything were ever a case of “the difference between significant and insignificant is not significant” this is it. I tried visualizing it but it’s hard as Stata doesn’t do opacity and I’m not very good at R. If anybody wants to graph it in R either post the code in the comments and I’ll try it or email me and I’ll send you the resampled estimates and point estimates as a CSV.

      Reply
  • […] Gabriel Rossman previously observed, a significant portion of the jump happened between 2010 and 2013: was this ACA expansion states […]

    Reply

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Trackback this post  |  Subscribe to the comments via RSS Feed


The Culture Geeks


%d bloggers like this: