Prevalence and Diagnostic Approach to Sleep Apnea in Hemodialysis Patients: A Population Study

Table 3

Multivariate logistic models for prediction of obstructive sleep apnea.

Factors	Model 1		Model 2		ROC area	Hosmer-Lemeshow
	OR	95% CI	OR	95% CI
Gender (f)	0.72	0.12–4.20
Age (y)	1.10	1.02–1.19	1.12	1.02–1.22
Neck circumference (cm)	1.65	1.16–2.36	1.63	1.13–2.34
Hypertension	147	2–11613	379	2–87016
BMI (kg/m^2)	0.87	0.68–1.10
Daytime sleepiness	0.58	0.08–4.32
Snoring	3.91	0.56–27.2	6.57	0.71–60.71
Unrefreshing sleep	0.26	0.04–1.78	0.10	0.01–1.11	0.900	0.010
eKt/V			0.08	0.00–14.09	0.927	0.259
Time on RRT			1.06	0.92–1.21	0.927	0.387

Model 1: prediction model with classical risk factors ().
Model 2: specific prediction model: significant factors of Model 1 () + stepwise addition of hemodialysis characteristics to maximize discriminatory accuracy (ROC area) and goodness of fit (Hosmer-Lemeshow test).
OR: Odds Ratio and 95% CI in the final cumulative model.
ROC area: area under the ROC curve of the model when the factor is added.
Hosmer-Lemeshow test: goodness of fit of the model when the factor is added.