Admission Data Predict High Hospital Readmission Risk

Article Figures & Data

Download figure
Open in new tab
Figure 1.
Charlson score distribution at admission.
Download figure
Open in new tab
Figure 2.
Readmission risk by Charlson score.
Download figure
Open in new tab
Figure 3.
Charlson score and polypharmacy dichotomies predict less than 30-day readmission. The graph shows the receiver operating characteristics curve for the model. Area under the curve = 0.8505.

Table 1. Characteristics of the Study Sample and Readmission Risk

Characteristics	Admissions, n (%)	Readmission Risk (%)	Glimmix P Value
Age (years)
18–49	346 (36.1)	13.6
50–64	341 (35.6)	15.5
≥65	271 (28.3)	13.7	.6
Sex
Female	608 (63.5)	14.0
Male	350 (36.5)	14.9	1.0
Insurance status
Medicaid	351 (36.6)	16.0
Medicare	246 (25.7)	14.2
Commercial	225 (23.5)	14.7
Self-pay	136 (14.2)	9.6	.5
ED use in the previous 6 months
Yes	497 (51.9)	16.1
No	461 (48.1)	12.4	.3
Smoking
Yes	298 (31.1)	14.1
No	660 (68.9)	14.4	.55
Polypharmacy
Yes	649 (67.8)	17.7
No	309 (32.3)	7.1	.0003
RN-rated cognitive issue
Yes	75 (7.8)	21.3
No	883 (92.2)	13.7	.11
Self-rated financial issue
Yes	109 (11.4)	11.0
No	848 (88.6)	14.7	.4
Self-rated social support issue
Yes	52 (5.4)	5.8
No	906 (94.6)	14.8	.14
Admission for HF, PNA, or COPD
Yes	440 (45.9)	18.6
No	518 (54.1)	10.6	.003
30-Day readmission risk	958 (100.0)	14.3

COPD, chronic obstructive pulmonary disease; ED, emergency department; HF, heart failure; PNA, pneumonia; RN, registered nurse.

Table 2. Summary of Two Glimmix Logistical Models of 30-Day Readmission Risk

There were two Charlson score groups: 0–4 and ≥5. Polypharmacy means the use of ≥6 medications.
COPD, chronic obstructive pulmonary disease; HF, heart failure; PNA, pneumonia.
* Model 2 readmission logit = β₀ + β₁ Charlson_i + β₂ Polypharmacy_i + ϵ_i; β₀ (standard error [SE]) = 1.4475 (0.1610); β₁ (SE) = 0.5446 (0.2170); β₂ (SE) = 0.7924 (0.2638). ϵ_i Differs for each patient.

Table 4. Data Clustering of 958 Admissions Among 568 Patients