Thread by @eugelin06

1/ Super excited about our research on Medicare coverage delays among the uninsured and their effect on long-term PD use!

Published in the current issue of @MedicalCareLWW with @gchertow, Jay Bhattacharya, and Darius Lakdawalla. @SchaefferCenter
journals.lww.com/lww-medicalcar…

2/ We used a regression discontinuity design (a #causalinference method in observational studies to reduce selection bias) to identify the effect of this esoteric, but important, @CMSGov coverage policy

3/ The policy is strange and creates strange differences in the uninsured: most uninsured patients receive Medicare on day 1 of month 4 of dialysis (i.e., if you start dialysis June 1-30, you get Medicare Sept 1)

4/ But those using home dialysis in the beginning get Medicare coverage on day 1 AND

pre-dialysis coverage to day 1 of the month of start (i.e., if those starting June 15, get Medicare coverage retroactively to June 1)

5/ In other words, starting home dialysis at the end of the month yields an extra 30 days of pre-dialysis Medicare coverage vs. starting home dialysis at the beginning.

What is pre-dialysis coverage good for? PD catheter placements

6/ IF the dialysis start day is random, then we can analyze the natural experiment to test the effect of this policy (or conversely, the effect of eliminating this policy)

This assumption isn't trivial - there is an obvious incentive to delay starts to the end of the month

7/ Surprisingly, it appears random. Here is the actual distribution of start days compared to random. Critically, we don't see a spike in starts at the end of the month (what you'd see if people delayed)

Observed characteristics are also similar by start day

8/ We also observe no difference by profit status in the 1st 3 days vs. last 3 days of the month, something we'd expect if facilities tried to game dialysis start days

9/ Again, IF dialysis start day is random, we can simulate an RCT and control for UNobservable characteristics using the start day

Unfortunately, we can't prove this assumption, but hopefully you agree the evidence is suggestive

10/ This policy also creates a discontinuity that allows us to estimate the policy effect. In short, at the beginning of the month, patients receive 0 days of pre-dialysis coverage, but over time, they receive more until the month resets

11/ Unsurprisingly, we find increased PD use at the end of the month at the start of dialysis. What's interesting is that this persists through 1 year. The difference isn't trivial either, we're talking an absolute difference of 4%

12/ We also find that this difference worsened after the PPS and after Medicaid expansion - so this isn't an anachronism that has been solved with PD incentives, the disparity has worsened.

13/ Again, we see large differences that persist through 1 year of dialysis

14/ Why does this happen? Generally, we wait 2 weeks for PD catheters to heal - this means you need 2+ weeks of pre-dialysis coverage. Therefore, starting dialysis at the beginning of the month makes PD financially unattractive.

15/ But wait, can't patients just switch to PD later? In fact, you could start HD on 6/1, get a PD catheter on 6/1, switch to PD on 6/15, and you would get Medicare coverage of everything!

Our study suggests that this doesn't happen, which is why differences persist through 1yr

16/ Dialysis modality choice is really sticky unfortunately. Switch rates are abysmally low after Month 1.

17/ What is the solution? Cover the 30 days before dialysis! In fact, we used Medicaid (a socioeconomically disadvantaged population with limited medical coverage) as a control group and show that there is no similar discontinuity

18/ Here is the same graph for long-term PD differences in the Medicaid population:

19/ Tl;dr: the Medicare waiting period for the uninsured leads to large decreases in PD use. Limited insurance (e.g., Medicaid expansion or something else) could solve the problem.

Also, no one switches PD after they start HD. We need to do a better job educating early.