Table 4.

Association of Pittsburgh sleep quality assessment index (PSQI) with the single components of MetS

	Linear regression model
	PSQI as a continuous variable
	B	95% CI	P
Abdominal obesity	− .24	− 1.30 to .81	.651
Hypertriglyceridemia	− 1.21	− 2.55 to .12	.074
Low HDL − cholesterol	.11	− 1.09 to 1.31	.855
High blood pressure	.07	− 1.05 to 1.19	901
High fasting blood glucose	− .23	− 1.43 to .96	.700

	Logistic regression model
	PSQI < 5
	OR	95% CI	P
Abdominal obesity	1.36	.72 to 2.57	.339
Hypertriglyceridemia	1.98	.94 to 4.15	.071
Low HDL-cholesterol	1.02	.50 to 2.07	.964
High blood pressure	.78	.41 to 1.48	.451
High fasting blood glucose	1.35	.67 to 2.71	.403

	Logistic regression model
	PSQI ≤ 7
	OR	95% CI	P
Abdominal obesity	1.41	.81 to 2.43	.222
Hypertriglyceridemia	1.60	.79 to 3.22	.190
Low HDL-cholesterol	.92	.49 to 1.71	.791
High blood pressure	1.53	.88 to 2.67	.132
High fasting blood glucose	1.46	.78 to 2.72	.239

All the covariates were entered simultaneously into the regression models. The model included all the variables which differed significantly (P < .050) in univariable analyses in Tables 1, and 2 (i.e. age, sex, use of benzodiazepines, platelet antiaggregants, ACE-inhibitors, and beta-blockers, diagnosis of heart failure, Charlson comorbidity score index, albumin, and hemoglobin levels)