Anderson and Steinberg developed a model for predicting hospital readmissions in a patient receiving Medicare. The authors are from the Johns Hopkins University.
Patient selection: Medicare recipient being evaluated at the time of hospital discharge
Outcome: readmission to an acute care hospital within the next 60 days
Parameters from multivariate analysis:
(1) age in years
(2) gender
(3) race
(4) supplemental Medicaid coverage
(5) number of hospital discharges in the 60 days prior to current admission
(6) current admission for self-limited, non-chronic disease
(7) surgery performed during current admission
(8) reimbursement to the hospital in US dollars
(9) location of current hospital
(10) number of beds in current hospital
|
Parameter
|
Finding
|
Points
|
|
gender
|
female
|
0
|
|
|
male
|
1
|
|
race
|
White
|
0
|
|
|
non-White
|
1
|
|
supplemental Medicaid coverage
|
no
|
0
|
|
|
yes
|
1
|
|
reasons for current admission
|
not self-limited or chronic
|
0
|
|
|
self-limited, non-chronic
|
1
|
|
surgery performed
|
no
|
0
|
|
|
yes
|
1
|
|
location
|
urban
|
1
|
|
|
non-urban
|
0
|
|
hospital beds
|
>= 100
|
0
|
|
|
< 100
|
1
|
X =
= (-0.0048 * (age in years)) + (0.10908 * (points for gender)) - (0.14732 * (points for race)) + (0.21179 * (points for supplemental Medicaid)) + (0.46509 * (number of hospital discharges)) - (0.34245 * (current admission)) - (0.33536 * (surgery)) + (0.00005 * (reimbursement)) - (0.19647 * (points for hospital location)) + (0.16399 * (points for hospital size in beds)) - 0.1684
probability of hospital readmission to an acute care hospital within the next 60 days =
= 1 / (1 + EXP((-1)*X))
Limitations:
• Some variables like hospital reimbursment need to be adjusted for the interval change since 1985.
• The significance of bed size may have changed since 1985.
