Risk-adjustment models for heart failure patients’ 30-day mortality and readmission rates: the incremental value of clinical data abstracted from medical charts beyond hospital discharge record

Table 2 Odds ratios (ORs) of 30-day mortality and readmission for each of the variables retained in multivariable risk-adjustment models based on HDRs only (#1) and HDR plus clinical data (#2)

Characteristics	Model #1			Model #2
Characteristics	OR	95 % CI	P	OR	95 % CI	p
30-day mortality
From HDRs
Gender
Male	1.00			1.00
Female	0.58	0.41–0.83	<0.01	0.63	0.37–1.06	0.08
Age^a	1.00	1.00–1.00	<0.001	1.00	1.00–1.00	<0.001
Cerebrovascular disease
No	1.00			1.00
Yes	1.89	1.53–2.35	<0.001	1.96	1.33–2.87	<0.01
From medical charts
Systolic blood pressure^b	–			0.05	0.02–0.13	<0.001
Heart rhythm
Sinus rhythm with heart rate < 90 bpm^c	–			1.00
Sinus rhythm with heart rate ≥ 90 bpm^c	–			2.94	1.87–4.61	<0.001
Atrial fibrillation	–			1.78	1.31–2.41	<0.001
Pacemaker rhythm	–			1.21	0.29–5.11	0.79
Serum creatinine^d	–			0.26	0.12–0.54	<0.001
30-day readmission
From HDRs
Gender
Male	1.00			1.00
Female	0.95	0.66–1.36	0.78	0.96	0.67–1.38	0.82
Age^a	1.00	1.00–1.00	<0.01	1.00	1.00–1.00	<0.01
Length of stay^b	1.41	1.14–1.74	<0.01	1.38	1.12–1.71	<0.01
Previous acute myocardial infarction
No	1.00			1.00
Yes	2.05	1.11–3.78	0.02	1.90	1.05–3.44	0.03
From medical charts
Systolic blood pressure^b	–			0.33	0.19–0.59	<0.001

Abbreviations: CI confidence interval
^aThis variable was cubed to achieve normality
^bThis variable was log-transformed to achieve normality
^cThis cutoff was chosen using ROC analysis to minimize the number of false positive and false negatives
^dThis variable was transformed using the reciprocal of square root to achieve normality. We also included a squared term in the regression to achieve a good model specification (OR = 357.41, 95 % CI = 36.36–3513.66, p < 0.001)

ISSN: 1472-6963