Poor Children Getting More Antipsychotic Drugs: The Power of a Poor Graph

Children on Medicaid are more likely to be prescribed antipsychotic meds than those with private insurance - and for less severe conditions - that's the front-page news in today's New York Times.  

So I read the story by Duff Wilson with interest, expecting something dramatic in the results of a new study out of Rutgers and Columbia to be published in the Health Affairs Journal.  Like we're pushing the powerful drugs on our most vulnerable citizens a lot more than other kids in comfy homes who have parents and psychologists eager to try other therapeutic solutions.  After all, Medicaid shells out a lot less for therapy than drugs and finding a psychiatrist or psychologist who accepts it can be difficult. 

Sure enough, poor children are 4 times more likely to get antipsychotics. 


Sounds bad, no wonder the group of health professionals, aka Too Many, Too Much, Too Young, yes, the actual name, want to find a better way to help these kids.

Then I saw the actual numbers.  Remember, four times.  Four hundred percent!  Yikes.  But four times more than what?  Would you believe less than 1%?  Turns out almost 1% of privately-insured kids get the mind-altering meds compared to 4% of the Medicaid kids.  Oh yes, I know, these are young and developing minds and bodies, any overuse and abuse of the drugs with serious side effects is problem for concern.  I get it, but..... 

Is this a meaningful difference?  The sample was in the thousands  But still, we're not sure how reliable these figures are.  We know large samples, let alone gigantic ones, often provide statistically significant results that in reality don't mean as much.   Like when we compare thousands of people.  

And then there are the graphics.  Take a look at the graph on the left courtesy of the New York Times:

Here we have an exercise in tweaking the results.  Or how to present your data in the most dramatic manner.   We got a a vertical axis that runs from 0% all the way to 5%.  Looks impressively bad for the poor kids.  But it's deceptive.  And just a piss poor presentation that wouldn't float in Research Methods 101.  I'd take off points for that on a test.  Imagine what the graph would look like if the vertical axis went up to 100%.  We'd barely notice a difference between the two populations. 

And just why are these kids prescribed the drugs?  We read the poor kids are more likely to get them for less severe conditions.   Are they?  Take a closer look at that circle of kids getting the meds:

27% of Medicaid kids vs. 30% of private-insurance kids take them for serious conditions

73% of Medicaid kids vs. 70% of  private-insurance kids for other less severe conditions

So, the to-do about the poor kids getting put on these meds for less severe conditions doesn't look all that meaningful.

Then there's the matter of how many children are diagnosed with these disorders.  Are children on Medicaid more likely to be diagnosed with mental illness?  Yes, according to the article - twice as likely.  A quick search of recent studies suggest the rate might be even higher, even over three times as high.  The current study shows they're four times more likely to be prescribed the drugs.

So it looks like some bias might exist towards prescribing antipsychotics but it doesn't appear to be large.  Do children from families without economic resources, without health insurance deserve competent doctors and therapists?  Absolutely.  Do we need to rethink services to this population?  Absolutely.  But do we have an epidemic of antipsychotic doping among our poorer children?  Clearly not. 

And yet it made the front page of the Sunday New York Times. 

BTW, what about other less powerful drugs?  Are kids prescribed these at the same rate?  Makes me wonder considering we've not heard a study claiming poor kids are morely to get psych meds in general.  The researchers looked at Medicaid and private-insurance claims, apparently millions of claims, surely they had numbers on the other drugs.  Which leads me to suspect there were no significant differences.  Which makes me question if the antipsychotic result was a freak finding. 
Crystal, S., Olfson, M., Huang, C., Pincus, H., & Gerhard, T. (2009). Broadened Use Of Atypical Antipsychotics: Safety, Effectiveness, And Policy Challenges Health Affairs, 28 (5) DOI: 10.1377/hlthaff.28.5.w770


No comments: