. 2019 Mar 8;2(3):e190348. doi: 10.1001/jamanetworkopen.2019.0348

Table 2. Performance of Readmission Scores Using Cutoff Values Among All Patients Across 3 Hospitals^a.

Score	Patients Flagged, %	Readmissions Detected, %^b	Improved Efficiency, %^c	P Value^d
mLACE, high vs low	34.4	54.7	31.4	<.001
B score at discharge to detect 54.7% readmitted^e	23.6	54.7	31.4	<.001
HOSPITAL, medium/high vs low	25.1	40.2	54.9	<.001
B score at discharge to detect 40.2% readmitted^e	16.2	40.2	54.9	<.001
HOSPITAL, high vs medium/low	5.0	11.2	32.0	<.001
B score at discharge to detect 11.2% readmitted^e	3.4	11.2	32.0	<.001
Maxim/RightCare^f	41.5	57.7	25.5	<.001
B score 48 h after admission to detect 57.7% readmitted	30.9		25.5	<.001
B score at discharge to detect 57.7% readmitted	20.6		50.4	<.001
B score ≥0.167 at discharge, top 10%	10.0	28.6	37.5^g	NA
B score ≥0.217 at discharge, top 5%	5.0	16.4	43.1^g	NA

Abbreviations: B score, Baltimore score; NA, not applicable.

^{^a}

The modified LACE (mLACE) score was calculated at discharge using 4 items: length of stay, acuity of admission, comorbidities, and emergency department visits 6 months before hospital admission. The Maxim/RightCare score was calculated 48 hours after admission using a proprietary formula. The HOSPITAL score was calculated at discharge using 7 variables (hemoglobin level, discharge from an oncology service, sodium level, procedure during the index admission, index type of admission, number of admissions during the past 12 months, and length of stay) to categorize patients as low, medium, or high risk.

^{^b}

Sensitivity for identifying readmissions.

^{^c}

Improved efficiency was computed as (p₁–p₂)/p₁, where p₁ represents the proportion of patients flagged as high risk based on mLACE; p₂ indicates the proportion of patients with B score greater than or equal to the cutoff value among all patients.

^{^d}

Comparing the percentage of patients flagged at a given sensitivity using the McNemar test.

^{^e}

The B score being a rank classifier identifies patients by relative risk. Therefore, cutoffs were identified to match the proportion of patients identified by standard scores (sensitivity) or patients identified in the top 5% or 10% of risk. The B score was matched to the percentage of readmissions predicted by standard scores at standard cutoff values for high vs low risk (mLACE, HOSPITAL, and Maxim/RightCare scores).

^{^f}

Collected at hospital 2 only and measured at 48 hours after admission (n = 2291).

^{^g}

Indicates percentage readmitted.

Table 2. Performance of Readmission Scores Using Cutoff Values Among All Patients Across 3 Hospitalsa.

Table 2. Performance of Readmission Scores Using Cutoff Values Among All Patients Across 3 Hospitals^a.