Skip to main content

Table 2 Prediction of 30-day readmission based on PREADM-H model variables (Derivation cohort, N = 24,599)

From: Identifying patients at highest-risk: the best timing to apply a readmission predictive model

Variables

OR

(95% CI)

P value

Chronic condition

 CHFP

1.16

(1.07–1.25)

< 0.001

 COPDP

1.19

(1.10–1.29)

< 0.001

 CRFP

1.18

(1.09–1.27)

< 0.001

 MalignancyP

1.02

(0.93–1.13)

0.658

 ArrhythmiaP

1.03

(0.96–1.11)

0.446

 DisabilityP

1.30

(1.21–1.40)

< 0.001

Oncology (treatment phase)H

1.13

(1.00–1.26)

0.041

 Body mass indexP

0.99

(0.98–0.99)

< 0.001

 No. hospital admissions in the past yearP

1.13

(1.10–1.15)

< 0.001

 No. primary care and specialist visits in the past yearP

1.00

(1.00–1.00)

0.522

 Residing in hospital’s catchment areaP

0.65–1.62

 

0.06–1.00

 No. days from last hospitalizationPH

1.00

(1.00–1.00)

< 0.001

 Index admission LOSH > =5 d

1.26

(1.18–1.35)

< 0.001

Procedure (any ICD-9 codes) during hospital stayH

0.84

(0.71–0.98)

0.026

 Index admission typeH: urgent

2.09

(1.65–2.66)

< 0.001

 Low hemoglobin level at dischargeH (< 12 g/dL)

1.28

(1.19–1.38)

< 0.001

 Low sodium level at dischargeH (< 13 mEq/L)

1.26

(1.15–1.38)

< 0.001

 Model Performance

(top 10%)

(top 20%)

 

 PPV

43.0 (40.0–46.0)

36.1 (34.1–38.1)

 

 Sensitivity

21.1 (19.4–22.9)

37.5 (35.4–39.6)

 

 Specificity

92.9 (92.4–93.5)

84.5 (83.7–85.3)

 

 C-stat (validation cohort, n = 10,557)

0.68 (95% CI: 0.67–0.70)

  1. Abbreviations OR Odds ratio, CI Confidence interval, CHF Congestive heart failure, COPD Chronic obstructive pulmonary disease, CRF Chronic renal failure, LOS Length of stay, C-stat Model’s discrimination and calibration, PPV Positive predictive value
  2. P: variables from PREADM model
  3. H: variables from HOSPITAL model
  4. PH: variables from PREADM-H model
  5. PREADM-H: Preadmission Readmission Detection Model + Hospital model