A flexible approach to measure care coordination based on patient-sharing networks

Table 1 Results of the generalized linear mixed model (second year)

	Variable	Model 1 (baseline)	Model 2 (incl. CD)	Model 3 (incl. FCD)
Elixhauser-subscales	Congestive heart failure	0.14*	0.15*	0.16*
	Cardiac arrythmias	0.14*	0.15**	0.14*
	Valvular disease	-0.29*	-0.31**	-0.36**
	Vascular disorders	0.28***	0.28***	0.32***
	Complicated hypertension	-0.15	-0.15	-0.19*
	Pulmonary diseases	0.10**	0.12**	0.11*
	Complicated Diabetes	0.13*	0.13	0.14*
	Renal failure	-0.14	-0.14	-0.19*
	Fluid and electrolyte	0.43***	0.46***	0.45***
	Alcohol Abuse	0.42***	0.45***	0.45***
	Drug Abuse	0.50***	0.55***	0.52***
	Depression	-0.11**	-0.11**	-0.10**
Other control variables	Hospitalization	1.67***	1.64***	1.62***
	Antipsychotics	0.35***	0.40***	0.39***
	Care level 0	0.37***	0.38***	0.40***
	Care level 1	0.17**	0.18**	0.19**
	Nursing home	0.18	0.18	0.22*
	Age	-0.02***	-0.02***	-0.02***
	Neurologist contacts	-0.02**	-0.01**	-0.01*
Coordination metrics	Care density		0.00
	Care density dummy		0.10**
	Fragmented Care Density			0.61***
Other model parameters	Intercept	-2.85***	-3.11***	-1.31***
Other model parameters	Patient-level variance	0.44***	0.42***	0.42***
Global Fit Statistics	-2 Log Likelihood	34,590.69	34,580.26	34,537.39
	AIC	34,680.69	34,674.26	34,629.39
	BIC	35,038.57	35,048.05	34,995.23

*p < .05, ***p < .01, ***p < .001, Cells contain estimated regression coefficients if not stated otherwise. Care density dummy is a dummy indicator that encodes whether care density can be calculated (i.e. whether the patient was treated by at least two outpatient providers). All non-significant control variables were omitted from this table. The full table is provided in the supplement

ISSN: 1471-2288