Table 1.
GEE Model* Estimated 30-Day Readmission Rates and 30-Day Mortality Rates at the Hospital Level after Patients’ Discharge from COPD Hospitalization by the HRRP from December 2006 to November 2017 in the United States
HRRP Periods | Period minus Announcement Period*† [% (95% CI)] | P Value‡ | ||
---|---|---|---|---|
30-d readmission rate | Preannouncement | December 2006 to July 2008 | 0.46 (0.41 to 0.5) | <0.0001 |
August 2008 to March 2010 | 0.32 (0.28 to 0.36) | <0.0001 | ||
Announcement | January 2013 to August 2014 | −1.75 (−1.8 to 1.71) | <0.0001 | |
Implementation | October 2014 to April 2016 | −1.53 (−1.58 to 1.48) | <0.0001 | |
May 2016 to November 2017 | −1.34 (−1.39 to 1.29) | <0.0001 | ||
30-d mortality rate | Preannouncement | December 2006 to July 2008 | 0.33 (0.29 to 0.36) | <0.0001 |
August 2008 to March 2010 | 0.1 (0.07 to 0.13) | <0.0001 | ||
Announcement | January 2013 to August 2014 | 0.12 (0.09 to 0.16) | <0.0001 | |
Implementation | October 2014 to April 2016 | 0.22 (0.18 to 0.26) | <0.0001 | |
May 2016 to November 2017 | 0.71 (0.67 to 0.75) | <0.0001 |
Definition of abbreviations: CI = confidence interval; COPD = chronic obstructive pulmonary disease; GEE = generalized estimating equation; HRRP = Hospital Readmissions Reduction Program.
GEE model was used to account for the hospital cluster effect among multiple periods. We compared each periods’ mortality rate with that of the announcement period (each period minus the announcement period).
Announcement period from April 2010 to November 2011 was used as a pivot for comparisons among periods.
P value shows the significance of the differences. The P value was corrected by the Bonferroni procedure to account for family-wise error.