Skip to main content
. Author manuscript; available in PMC: 2022 May 1.
Published in final edited form as: Health Aff (Millwood). 2021 May;40(5):745–753. doi: 10.1377/hlthaff.2020.00649

Exhibit 4.

Effects of Cost Sharing and Early Discharge Due to Cost Sharing on Patient Outcomes


Patient Outcome Baseline Daily Rate Unadjusted Baseline Difference Effect of Exposure to Cost Sharinga Effect of Early Discharge due to Cost Sharinga,c
Proportion of patients experiencing outcome on calendar day 19 in more exposed cohort, % Mean difference in daily rate from calendar day 16–19 between more vs. less exposed cohort, percentage points Difference-in-differences estimate of effect on daily rate,b percentage points Rescaled to cumulative effect on outcome occurring by calendar day 28 associated with one fewer day in a SNF, percentage points

Death 0.22 0.026**** −0.007 −0.14
All-cause hospitalization 0.79 0.049**** 0.019* 0.29
Hospitalization or death 0.99 0.073**** 0.008 0.12
Fall-related hospitalizationd 0.03 0.001 0.002 0.03

Source: Author’s analysis of fee-for-service Medicare claims data

Notes:

SNF= skilled nursing facility

*

p<0.10

**

p<0.05

***

p<0.01

****

p<0.001

a

Estimates reflect changes within nine days after the discharge spike at SNF benefit day 20

b

Estimates for the daily rate of all-cause hospitalization refer to the differential change from days 16–19 to days 22–28 for patients who were more vs. less exposed to cost sharing, omitting days 20–21 because direct transfers from SNFs to hospitals on those days resulting from cost sharing do not reflect unsafe discharges to home (see Exhibit A17 confirming increases in hospitalizations on days 20–21 were coded as facility-to-facility transfers).12 Estimates for death and fall-related hospitalizations, which were not subject to this limitation in interpretability, reflect the differential change from days 16–19 to days 20–28. Day 15 is excluded because we required patients to be in a SNF on day 15, but not other days, to be included in the sample.

c

Approximated by dividing the effect of exposure to cost sharing on the cumulative incidence of the outcome through day 28 (3rd column multiplied by 9 or 7 days, depending on the outcome) by the effect of exposure on the mean number of days spent in a SNF through day 28 (0.45 fewer days).

d

Fall-related injury hospitalizations are identified using the following ICD codes in either the primary or secondary diagnosis field: e880, e881, e882, e884, e885, e888, 800–848, 850–854, 920–924.