Sluggish Future: No Longer Fresh Over at Finance and Development

Must-Read: The last, italicized, clause in the quote from Ross Douthat below is a lie:

Ross Douthat: Is Obamacare a Lifesaver?: "Now that the Republican Party has beclowned itself on health care... Obamacare repeal... in rubble...

...every G.O.P. policy person who ever championed a replacement plan is out wandering in sackcloth and ashes, wailing, “The liberals were right about my party, the liberals were right about my party,” beneath a harsh uncaring heaven... now, in these hours of right-wing self-abnegation, it’s worth raising once again the most counterintuitive and frequently scoffed-at point that conservatives have made about Obamacare:

It probably isn’t saving many lives.

One of the most powerful arguments in the litany that turned moderate Republican lawmakers to jelly was that they were voting to “make America sick again”.... Tens of thousands of people, Democrats warned, would die if Paul Ryan’s stingy replacement took its place.... This argument was still most likely false..... The link between health insurance and actual health has always been a lot murkier than most champions of universal coverage admit... little evidence that giving people insurance actually makes them healthier. Recent data... is mixed: A study of Mitt Romney’s Massachusetts insurance expansion showed health benefits for the newly insured... but a study of Oregon’s pre-Obamacare Medicaid expansion found that the recipients’ physical health did not improve...

Look at this summary table from the Oregon Medicaid Study:

Oregon Medicaid Study Summary Table

  1. The bottom red circle tells us: relative to the control group, an extra 5% of those who won the Medicaid lottery are taking diabetes meds.

  2. The top red circle tells us: relative to the control group, 1% fewer of the "got Medicaid" group have elevated blood glucose levels.

12% rather than 6% being treated for diabetes; 4% rather than 5% with high glucose levels. Both treatment likelihood and health indicators are better for the "got Medicaid" group.

That is a clinically-significant improvement in the health of the population that got on Medicaid: for first-line anti-diabetes drugs to knock about 1/5 of those with the diagnosis back into the normal range is what you would expect.

The problem is the sample size: only 6000. In a population that small, statistical noise means that you could not reject the hypothesis that the effect was four times as large--that the drugs knocked 4% of the "got Medicaid" group into the healthy range. But also if you started out believing that diabetes drugs were counterproductive—that they caused blood glucose levels to increase—you could not reject that hypothesis either.

The proper way to describe this is that while the effects on emotional health and financial stability were as expected clinically and were statistically significant, the effects on blood glucose (and hypertension, and high cholesterol) were as expected clinically but were not statistically significant.

If Douthat had said that the improvements in physical health were not statistically significant, I would accuse him of misleading his audience by not also noting that the improvements in physical health were in line with clinical expectations for treatment success.

But Douthat claims, instead, that: "a study of Oregon’s pre-Obamacare Medicaid expansion found that the recipients’ physical health did not improve". That is simply a lie. I don't know whether he is knowingly or unknowingly spreading this lie. But his ultimate source for the claim was looking at this table. And lied.

Is it too much to ask the New York Times to at least pretend to care about quality control here?

Comments