Value-based healthcare in ischemic stroke care: case-mix adjustment models for clinical and patient-reported outcomes

Table 3 Case-Mix Risk Adjustment Models for Mortality, mRS and EQ-5D

	Model 1*: Mortality at 3 months		Model 2*: mRS score at 3 months	Model 3*: EQ-5D index score at 3 months
(Nagelkerke) R²	0.44		0.42		0.37
AUC^a	0.87		0.83		0.78
Variables	Multivariable OR (95% CI)	P-value	Multivariable OR (95% CI)	P-value	ßeta (SE)	P-value
Age	1.07 (1.05–1.09)	< 0.001	1.04 (1.03–1.05)	< 0.001	−0.007 (0.001)	< 0.001
NIHSS on admission	1.17 (1.14–1.20)	< 0.001	1.17 (1.14–1.19)	< 0.001	− 0.020 (0.001)	< 0.001
Heart failure	2.54 (1.30–4.97)	0.007	3.58 (2.14–5.98)	< 0.001	− 0.130 (0.042)	0.002
Previous stroke			1.74 (1.33–2.27)	< 0.001	− 0.063 (0.023)	0.007
Smoking			1.58 (1.18–2.13)	0.003	− 0.055 (0.025)	0.028
Caregiver at discharge			0.67 (0.51–0.86)	0.002	0.057 (0.022)	0.010
Connective tissue disease			2.07 (0.93–4.61)	0.074	−0.119 (0.061)	0.053
Cancer			1.99 (1.35–2.94)	0.001	−0.065 (0.042)	0.122
CCI	1.22 (1.13–1.32)	< 0.001			−0.020 (0.007)	0.003
Onset-to-door time (per 10 min)	1.00 (1.00–1.00)	0.123
Diabetes			2.09 (1.12–3.88)	0.020
Sex (Ref: Female)					0.041 (0.020)	0.036
SES					−0.019 (0.010)	0.053
Nationality (Ref: Native Dutch)					−0.074 (0.045)	0.097

SES Socio-economic status, derived from neighborhood-based ranking, NIHSS National Institute of Health Stroke Scal, CCI Charlson comorbidity index, R² R-squared, 95% CI 95% confidence interval, SE Standard error
*The case-mix models was using backward selection with the Akaike information criterion (AIC) as a cut-off for the p-value. The imputed dataset (10 imputations of original data N = 1022) was used for the development of all three case-mix model development
^aAUC values were calculated for mortality, dichotomized mRS scores (0–2 vs. 3–6) and dichotomized EQ-5D index scores (< 0.65 vs. ≥ 0.65)

ISSN: 1471-2288