. Author manuscript; available in PMC: 2016 May 1.

Published in final edited form as: Am J Psychiatry. 2015 Jan 13;172(5):460–465. doi: 10.1176/appi.ajp.2014.14040480

TABLE 1.

Multivariate¹ Cox regression models for time to severe dementia

			Psychosis Cluster		Affective Cluster		Agitation/Aggression		Apathy/Indifference		Neuropsychiatric Inventory Significance²
	Variable		Hazard Ratio	P-value	Hazard Ratio	P-value	Hazard Ratio	P-value	Hazard Ratio	P-value	Hazard Ratio	P-value
Unadjusted, bivariate value³			2.024	0.018	1.387	0.191	2.321	0.009	1.176	0.604	--^* (1.214/2.129)	0.025^* (0.560/0.008)
ADJUSTED MODEL	Alzheimer’s dementia onset age		0.290	<0.001	0.313	<0.001	0.351	0.001	0.279	<0.001	0.295	<0.001
	Alzheimer’s dementia onset age squared		1.008	<0.001	1.007	<0.001	1.007	0.001	1.008	<0.001	1.008	<0.001
	Female		1.949	0.031	1.888	0.038	1.885	0.039	1.998	0.025	1.852	0.047
	Education⁴		1.791	0.79	1.910	0.048	1.934	0.041	1.888	0.56	0.756	0.095
	Apolipoprotein E – e4 Carrier		0.940	0.823	1.076	0.767	1.109	0.706	1.070	0.808	1.106	0.711
	Global medical health rating		1.554	0.140	1.585	0.128	1.511	0.174	1.527	0.146	1.737	0.074
	Dementia duration at baseline		0.811	0.026	0.827	0.042	0.753	0.006	0.821	0.038	0.763	0.005
	Psychosis Cluster		2.007	0.028
	Affective Cluster				1.512	0.119
	Agitation/Aggression						2.946	0.004
	Apathy/Indifference								1.552	0.172
	Neuropsychiatric inventory significance²
		Trichotomized									--	0.002
		Mild symptom(s)									1.077	0.832
		Clinically significant symptom(s)									2.682	0.001

Cox regression models were controlled for age of dementia onset, dementia duration at baseline, gender, education level, global medical health rating, and apolipoprotein E – e4 status;

Neuropsychiatric inventory examined as a trichotomous variable (none vs. mild vs. clinically significant);

Bivariate unadjusted model for each neuropsychiatric inventory symptom or cluster;

⁴

Reference category is greater than or equal to high school;

Data is shown as trichotomized comparison followed in parentheses by mild symptoms vs. none / clinically significant symptoms vs. none