Depressive symptoms in people with chronic physical conditions: prevalence and risk factors in a Hong Kong community sample

Table 3 Multilevel regression analysis to determine the strongest predictors of depressive symptoms in those with chronic conditions (N = 6,195)

Model (PHQ-9)	Cumulative R²	R² Change	-Log-likelihood	P value of log-likelihood Change
Model 1: Demographic variables ^a	0.36	0.36	16252	<0.001
Model 2: Demographic factors + all remaining variables ^b
Model 3: model 2 + life stress (excluding health problem)	0.37	0.01	15809	<0.001
Model 4: model 2 + life stress ^c	0.42	0.05	15550	<0.001
Model 5: model 4 + number of chronic conditions	0.43	0.01	15483	<0.001
Model 6: model 4 + family satisfaction	0.43	0.00	15441	<0.001
Model (PHQ-6)
Model 1: Demographic variables ^a	0.28	0.28	13268	<0.001
Model 2: Demographic factors + all remaining variables ^b
Model 3: model 2 + life stress (excluding health problem)	0.29	0.01	12865	<0.001
Model 4: model 2 + life stress ^c	0.34	0.05	12625	<0.001
Model 5: model 4 + number of chronic conditions	0.35	0.01	12582	<0.001
Model 6: model 4 + family satisfaction	0.35	0.00	12558	<0.001

^a Demographic variables included age, gender, marital status, body mass index.
^b All remaining variables included drinking, physical activity, and neighborhood cohesion. In model 2, number of chronic conditions was selected secondly into the model and family satisfaction the third. In next steps, neighborhood cohesion, physical activity, and drinking were also selected but with negligible R² Changes (all < 0.002), which were not presented.
^c The 19-item life stress scale, which includes items on health problems, replaced the 17-item scale in the model, explaining 5% more variance in depressive symptom scores compared with the 17-item scale.

ISSN: 1471-244X