Table 1. Assessment of the quality of the studies.

Studies included in the meta-analysis
Study	Was the research question stated clearly?	Were the patients with stroke recruited in an acceptable way?	Is the control group a non-clinical population?	Is the control group adequately described?	Was sleep assessed and reported comprehensively?*	What confounding variables have the authors accounted for?
Arzt et al. 2010 [21]	Yes	Yes	Yes	Yes1	Yes	BMI (m ± sd)
Terzoudi et al. 2009 [26]	Yes	Yes	No	Yes	Yes	Barthel Index, m score
Gottselig et al. 2002 [27]	Yes	Yes	No	Yes	Yes	/
Müller et al. 2002 [28]	Yes	Yes	No	Yes	Yes	/
Bassetti and Aldrich 2001 [29]	Yes	Yes	No	Yes1	Yes	Scandinavian Stroke Scale (m ± sd)
Santamaria et al. 2000 [30]	Yes	Yes	Yes	Yes	Yes	n for hypertension, diabetes, smoking
Mohsenin and Valor 1995 [33]	Yes	Yes	No	No	Yes	BMI (m ± sd); hypertension %; smoking; snoring %; cardiovascular disorder
Hudgel et al. 1993 [34]	Yes	Yes	Yes	No	Yes	BMI can be calculated
Giubilei et al. 1992 [32]	Yes	Yes	No	No	Yes	/
Studies excluded in the meta-analysis (but included in systematic review)
Coelho et al. 2010 [22]	Yes	No	No	No	Yes	hypertension %; polyneuropathy %; anemia %; antidepressants %; fatigue %
Siccoli et al. 2008 [25]	Yes	Yes	No	No	No3	m and range for BDI
Bliwise et al. 2002 [23]	Yes	Yes	No	Yes2	Yes	/
Vock et al. 2002 [31]	Yes	Yes	No	Yes	No3	/
Pinto et al. 2000 [35]	Yes	Yes	Yes	Yes	No3	/
Yokoyama et al. 1996 [24]	Yes	Yes	Yes	No	No4	/

* For comprehensively is meant that the study reported means with all data necessary to conduct meta-analytic calculations. 1 = no information about clinical assessment reported. 2 = control group characterized but comprising two groups of patients with Alzheimer and Parkinson diseases. 3 = not possible to calculate the effect size. 4 = sleep variables not considered in other studies, thus not possible to conduct meta-analytic calculations for only one study.