Prior to the ready availability of polysomnography (PSG), OSA was often suspected on a clinical basis from the major signs and symptoms of what now constitutes the OSA syndrome , as a continuum has been shown to exist between snoring, obesity, and OSA [14, 15].

In 1989, Palomake et al. suggested that snoring alone was a risk factor for stroke [16]. In their study of 167 men with stroke, 36 % suffered stroke in sleep . Stepwise multiple logistic regression analysis found snoring to be the only potential risk factor significantly related to stroke in sleep. In 1991, it was reported that the odds ratio (OR) of snoring increased when the combination of obesity and sleepiness (classical signs/symptoms of OSA) was also present [17]. Retrospective analysis of pioneering studies such as these led to the routine use of PSG in similar research, as a significant number of these subjects were suspected to have had OSA.

The foundation of PSG is electroencephalography (EEG). In 1875, Caton performed the first animal studies and in 1929 Berger published recordings from humans [18–21]. By 1936, there were only six EEG laboratories in the USA, including our facility at the University of Iowa, under the direction of my (MED) mentor John Knott [21]. The first continuously recorded all-night EEG sleep studies by Loomis et al. in 1937 were followed by the discovery of rapid eye movement (REM) sleep by Aserinsky and Kleitman in 1953, which showed the utility of the electrooculogram (EOG) [22, 23]. In 1967, Jouvet showed the association of REM sleep with hypotonia justifying the present use of the electromyogram (EMG) [24]. In 1968, the combination of EEG, EOG, and EMG allowed for formal PSG and the first published standardized technique for scoring sleep stages by Rechtschaffen and Kales [25]. In 1978, utilizing the PSG, Guilleminault et al. defined the term apnea index (the average number of apneas and hypopneas per hour of sleep), to precisely define the presence and severity of OSA [26]. In 2007, the American Academy of Sleep Medicine published the first standard definition for hypopnea in the pediatric population, thus allowing for the routine reporting of the apnea–hypopnea index (AHI; the average number of apneas plus hypopneas per hour of sleep) in children [27]. Today, the AHI is “…the key measure used for case identification, for quantifying disease severity, and for defining disease prevalence in normal and clinical populations” [28] (Table 31.1).

Transient ischemic attacks (TIAs; “ministrokes”) are focal neurological deficits resolving within 24 h [3]. Case reports of patients with histories of TIAs during sleep, obesity, and snoring, after the treatment of their PSG-documented OSA led to no further recurrence of TIA, indirectly suggest a cause-and-effect relationship between OSA and stroke, as 15 % of strokes are preceded by TIA (90-day risk; up to 17.3 %) [3, 39–42]. Nevertheless, case-controlled study reports have been inconsistent and OSA is not considered a TIA risk factor [43, 44].

Large population and clinically based cohort studies, despite using variable populations, PSG methodologies, OSA definitions, and statistical analyses, and often examining stroke risk in combination with other medical problems, suggest that adults with OSA and an AHI ≥ 20 are at stroke risks (Table 31.2).

In 1996, we published a 4-year follow-up on all stroke and sex- and age-matched control subjects from a 1992 prevalence study of OSA in stroke, with the exception of three controls who moved, leaving no contact information [9, 10]. All stroke subjects who died had OSA ( N = 5), of whom only one used continuous PAP (CPAP; death from urosepsis). Only one male control died; without OSA, from prostatic carcinoma. Respectively, stroke subjects dead versus alive had mean AHIs of 41.3 and 22.1, suggesting the diagnosis and severity of OSA in stroke were associated with greater long-term mortality.

A 1990, prospective, treatment versus non-treatment study by Partinen et al. has been used to suggest a cause-and-effect relationship between OSA and stroke [51, 52]. Over 7 years, they followed 198 OSA subjects treated with either weight loss ( N = 127) or tracheostomy ( N = 71); 5.2 % of the weight loss group had stroke (17.3 % died, 11 % from vascular etiologies), whereas only 1.2 % with tracheostomy suffered stroke (2.8 % died).

In 2005, Martinez-Garcia et al. showed, of 51 patients with recent stroke and minimal AHI of 20, only 15 (29.4 %) tolerated CPAP after 1 month [53]. In those who could not tolerate CPAP, the probability of a new vascular event increased fivefold (OR 5.09) .

In 2006, Palombini et al. studied 32 patients with recent stroke [54]. Only 22 % tolerated CPAP for 8 weeks; difficulties with CPAP use as perceived by the patient and family members, facial weakness, motor impairment, and increased difficulties and discomfort using a full-face mask. Brown et al. found stroke patients took longer to put on ( P< 0.01) and remove ( P< 0.01) traditional headgear as compared to a one-piece system [55]. Palombini et al. stated, “Better education and support of patients and families, and special training session in rehabilitation services, will be needed to improve compliance” [54]. Disler et al. showed, in a supportive setting, patients with stroke and OSA, despite moderately severe motor and cognitive functional independence measurement scores, tolerated CPAP with normalization of oxygen saturations [56].