Introduction

Based on available general population-based studies, the prevalence of obstructive sleep apnea (OSA) associated with accompanying daytime sleepiness is approximately 11% for adult men and 4% for adult women [1]. It has been suggested that the higher prevalence of OSA in males may result from the fact that women do not show the classic symptomatology or consult for nonspecific symptoms, and thus they may be underdiagnosed. Also, men have a longer, collapsible oropharynx, which is the pharyngeal segment involved in OSA, and a larger, fatter, posterior tongue than women and, hence, more chance of developing OSA [2].

Patients with obstructive sleep apnea usually consult for symptoms such as habitual or loud snoring, apneas, tiredness or fatigue, and excessive daytime sleepiness. Sometimes, the initial complaints are nonspecific symptoms such as insomnia, frequent awakenings at night, nocturia, morning headaches, or neurocognitive impairment [3]. Whereas some studies have pointed out that women complain more frequently of insomnia [4,5,6,7,8,9] and daytime tiredness [7,8,9] compared to men, others have shown discordant results regarding the reporting of snoring [6, 10, 11], apnea [4,5,6, 10], and excessive daytime sleepiness [4, 6, 7].

Several studies have conducted comparative analyses of the prevalence of symptoms between women and men with OSA, with [4, 6] or without [9, 12,13,14] covariate adjustment; other researchers have performed multiple regression analysis to determine whether gender independently predicted OSA symptoms [7, 8, 10, 11, 15]. In order to extend previous observations and clarify some controversies, we aimed to evaluate the influence of gender on reported OSA-related symptoms in a large population of patients referred to the sleep laboratory of our institution.

Methods

Patient selection

We examined the database from Hospital Alemán’s sleep laboratory from January 2013 to January 2016 and preselected 2154 patients older than 17 years who had a polysomnography (PSG) and answered a sleep questionnaire. Since there was male predominance in the initial sample (1454), 700 males were chosen at random to equal the number of females (700). The subjects who had undergone diagnostic PSG, the Berlin questionnaire, a sleep questionnaire, and the Epworth sleepiness scale (ESS) were selected. Exclusion criteria comprised patients who did not answer questions 1 to 7 of the Berlin questionnaire or the ESS, marked more than one answer on any of the questions, had suspected narcolepsy, performed a split-night PSG, or had a PSG with less than 180 min of total sleep time. Of the 1400 patients selected, 316 were excluded for different reasons. Figure 1 shows the study flow chart.

Fig. 1
figure 1

Study flow chart (PSG, polysomnography; SQ, sleep questionnaires; TST, total sleep time)

Full size image

Measurements

All the patients enrolled in this study underwent a diagnostic PSG which included EEG (F4/A1, C4/A1, O2/A1), EOG (two channels), chin EMG, leg EMG (two channels), ECG, airflow by nasal pressure, and an oral thermistor, thoracic and abdominal movements (two channels with piezoelectric sensors), snoring, oxygen saturation (SO2), and body position. The PSGs were registered from 10:30–11:30 p.m. to 5:00–6:00 a.m. Prior to PSG, the patients were given the following instructions: (1) to avoid napping and alcoholic or caffeinated beverages, (2) to continue taking the usual medication, (3) to eat supper between 8 and 9 p.m., and (4) to report to the sleep laboratory between 9 and 10:30 p.m. The PSG analysis was performed manually by four trained physicians following international criteria [16]. OSA was defined as an apnea/hypopnea index (AHI) ≥ 5. On arrival at the sleep lab, the patients completed a Spanish version of the ESS [17] and the Berlin questionnaire [18], and a sleep medical record (see appendix). Body weight and height were recorded in all subjects wearing only light clothes and no shoes. The Spanish version of the Berlin questionnaire was validated against PSG, and it showed to have similar sensitivity and specificity to reports in the original literature [19, 20]. Comorbidities were detected either by self-reporting and/or the medication that the patient was taking. Restless leg syndrome was assumed when patients complained of uncomfortable sensations in their legs before sleeping and reported behaviors to relieve this discomfort. Menopause was considered when it was reported on the sleep questionnaire or when women were older than 50 years [21]. Definitions used for nocturnal and diurnal symptoms associated with OSA are shown in Table 1.

Table 1 Definitions of symptoms associated with obstructive sleep apnea
Full size table

Statistical analysis

A frequency histogram and the Kolmogorov–Smirnov test were used to assess if the study variables had a normal distribution. The Mann–Whitney or the chi-square tests were used to compare differences between women and men.

Multiple logistic regression analysis was performed to assess the relationship between symptoms associated with OSA (present = 1, absent = 0) and various potential explanatory variables. Typical symptoms of OSA (snoring, apnea) were adjusted for age, gender (male = 1, female = 0), BMI, and AHI. In order to predict other less-specific symptoms of OSA (tiredness, excessive diurnal somnolence, insomnia, nocturia, morning headaches, subjective cognitive complaints), logistic regression models included other factors that could potentially explain their presence or absence (mean SO2; minimal SO2; T90; use of sedatives, antidepressants, or analgesics; cardiovascular comorbidities; diabetes; hypothyroidism; rhinitis; pulmonary pathology’ restless leg syndrome; and gastroesophageal reflux symptoms). For the purpose of this investigation, the term comorbidities, used as predictor in logistic regression models, referred to any of the following conditions: hypertension, coronary disease, stroke, heart failure, diabetes, hypothyroidism, rhinitis, asthma, COPD, gastro-esophageal reflux, nocturia, and chronic use of analgesics.

The statistical analysis was carried out with a commercial computer program, MedCalc Statistical Software version 16.8 (MedCalc Software bvba, Ostend, Belgium; http://www.medcalc.org; 2016).

Results

The study sample included 1084 patients, median age was 53 years, 46.5% (504) were women, 72.7% (788) had OSA (AHI ≥ 5), and 31.2% were obese. The characteristics of study population are shown in Table 2.

Table 2 Characteristics of the study population
Full size table

Gender differences in OSA population

Tables 2, 3, and 4 show gender differences in terms of clinical manifestations and PSG features. Women presented older age, lower AHI, and higher prevalence of morbid obesity than men. Seventy-one percent of the women with obstructive sleep apnea were menopausal. Women reported more hypothyroidism and restless legs syndrome than men and complained more often of tiredness, insomnia, morning headache, and subjective cognitive complaints. They also reported taking more sedatives/antidepressants, analgesics, and antacids.

Table 3 Polysomnography
Full size table
Table 4 Symptoms gender differences in obstructive sleep apnea
Full size table

Factors associated with OSA symptoms

Table 5 shows the results of the different logistic regression models used to identify the explanatory variables of OSA-related symptoms.

Table 5 Stepwise multiple logistic analysis
Full size table

Specific OSA symptoms

Snoring (S) and habitual snoring (HS) were significantly associated with male gender (S: coefficient 1.40, OR 4.06, p < 0.001; HS: coefficient 0.85, OR 2.34, p < 0.001), BMI (S: coefficient 0.038, OR 1.04, p 0.039; HS: coefficient 0.076, OR 1.08, p < 0.001), and AHI (S: coefficient 0.067, OR 1.07, p < 0.001; HS: coefficient 0.026, OR 1.03, p < 0.001). Loud snoring had a significant relationship with age (coefficient − 0.0095, OR 0.99, p 0.041), male gender (coefficient 0.76, OR 2.14, p < 0.001), BMI (coefficient 0.031, OR 1.03, p 0.015), and AHI (coefficient 0.025, OR 1.03, p < 0.001). Apnea also showed a significant association with age (coefficient − 0.015, OR 0.98, p 0.004), male gender (coefficient 0.89, OR 2.44, p < 0.001), and AHI (coefficient 0.035, OR 1.03, p < 0.001).

Nonspecific OSA symptoms

Tiredness showed a marked relationship with age (coefficient − 0.014, OR 0.98, p 0.02), female gender (coefficient − 0.55, OR 0.57, p 0.001), insomnia (coefficient 0.83, OR 2.29, p < 0.001), restless leg syndrome (coefficient 0.59, OR 1.80, p 0.001), and ESS (coefficient 0.77 OR 2.16, p < 0.001).

Excessive diurnal somnolence was associated with age (coefficient − 0.0012, OR 0.99, p 0.019), BMI (coefficient 0.033, OR 1.03, p 0.022), AHI (coefficient 0.017, OR 1.02, p < 0.001), mean SO2 (coefficient 0.13, OR 1.14, p 0.007), insomnia (coefficient − 0.42, OR 0.66, p 0.012), and tiredness (coefficient 0.77, OR 2.16, p 0.001).

Sleep onset insomnia was significantly related with age (coefficient − 0.015, OR 0.98, p 0.017), female gender (coefficient − 0.53, OR 0.59, p 0.0035), use of sedatives or antidepressants (coefficient 0.47, OR 1.60, p 0.006), frequent awakenings (coefficient 0.99, OR 2.70, p < 0.001), sleeping movements (coefficient 0.99, 2.69, p < 0.001), restless leg syndrome (coefficient 0.47, OR 1.60, p 0.01), tiredness (coefficient 0.58, OR 1.77, p 0.001), and ESS (coefficient − 0.66, OR 0.52, p 0.001).

Maintenance insomnia showed a meaningful relationship with AHI (coefficient − 0.021, OR 0.95, p 0.003), use of sedatives or antidepressants (coefficient 0.72, OR 2.06, p 0.001), frequent awakenings (coefficient 2.25, OR 9.5, p < 0.001), and tiredness (coefficient 0.70, OR 2.02, p 0.001).

Subjective cognitive complaints correlated with AHI (coefficient − 0.012, OR 0.99, p 0.026), T90 (coefficient 0.018, OR 1.02, p 0.004), insomnia (coefficient 0.47, OR 1.60, p 0.004), sleeping movements (coefficient 0.35, OR 1.42, p 0.02), frequent awakenings (coefficient 0.36, OR 1.44, p 0.02), tiredness (coefficient 0.48, OR 1.62, p < 0.001), ESS (coefficient 0.78, OR 2.17, p < 0.001), consumption of sedatives or antidepressants (coefficient 0.49, OR 1.64, p 0.002), and comorbidities (coefficient 0.44, OR 1.55, p 0.045).

Nocturia presented a marked association with age (coefficient 0.040, OR 1.04, p < 0.001), BMI (coefficient 0.039, OR 1.04, p 0.001), mean SO2 (coefficient − 0.11, OR 0.89, p 0.001), insomnia (coefficient − 0.43, OR 0.65, p 0.02), frequent awakenings (coefficient 1.92, OR 6.8, p < 0.001), and sleeping movements (coefficient 1.19, OR 3.28, p < 0.001).

Morning headaches were associated with age (coefficient − 0.021, OR 0.98, p 0.005), gender (coefficient − 1.15, OR 0.32, p < 0.001), ESS (coefficient 0.75, OR 2.11, p 0.001), insomnia (coefficient 0.83, OR 2.28, p 0.001), and restless leg syndrome (coefficient 0.93, OR 2.52, p < 0.001). Figures 2 and 3 show the influence of gender on the symptoms associated with OSA.

Fig. 2
figure 2

Forest plot showing the probability of reporting OSA symptoms according to gender. Men had more chance of manifesting snoring, habitual snoring, loud snoring, and apnea than women, and less likelihood of referring tiredness, sleep onset insomnia, and morning headache

Full size image
Fig. 3
figure 3

Forest plot showing the lack of association between EDS (excessive daytime sleepiness), maintenance insomnia, subjective cognitive complaints, and nocturia with gender

Full size image

Discussion

In our population that included patients without OSA to avoid leaving out of the analysis patients who could potentially have had airway resistance syndrome (AHI < 5 plus sleepiness), we observed that gender was an independent, explanatory variable of several specific and nonspecific OSA-related symptoms. After adjusting for age, BMI, and AHI, women were less likely to report snoring, habitual or loud snoring, and apneas than males. After controlling for multiple variables, female gender was an independent predictive factor of tiredness, sleep onset insomnia, and morning headaches. On the other hand, the report of excessive daytime sleepiness, nocturia, maintenance insomnia, and subjective cognitive complaints did not show a significant association with gender.

Our observations are supported by early publications. It has been reported that male gender constitutes an independent risk factor for typical symptoms of OSA, such as habitual or loud snoring [10, 22] and witnessed apnea [4, 6, 9, 10], after adjusting for covariables such as age, BMI, alcohol consumption or sedatives, cigarette smoking, and/or AHI. Several potential explanations may account for these findings. We speculated that during the development of the obstructive respiratory event, males could develop a higher intrathoracic negative pressure than females. Thus, the higher forces applied to the vibratory structures of the upper airway could cause more intense and perceptible respiratory sounds. In line with this theory, Itasaka [23] and Stoohs [24] observed that, in subjects with OSA, the intensity of snoring showed a strong positive correlation with inspiratory effort evaluated by esophageal pressure. It has also been described that men had significantly more and longer apneas than their female counterparts [25]. Therefore, women could perceive more easily the breathing pauses than males and so report it to their bedpartner. Third, women may consider their own snoring “unladylike” and therefore be less likely to mention it [2]. In addition, women are more likely to attend clinical appointments on their own, whereas men often do so with their partner [14]. Consequently, information from a partner on snoring and witnessed apneas may be less readily available for women. Finally, the definitions used for habitual or loud snoring are not uniform in published studies and this could add another source of variation in symptom reporting.

Most publications from clinical or community populations have found that women with OSA complain more of tiredness or fatigue than their male counterparts, independently of other confounders [7, 8, 13]. Several hypotheses may be raised to try to explain this observation. Prior studies have indicated that women are frequently more likely to report physical symptoms [26,27,28]. Also, cultural differences may make men less willing than women to admit that they have any of the problems we asked about, or there may be a real gender-based neurophysiologic bias in the way daytime effects of disturbed sleep are perceived. An unexpected finding was the lack of association between tiredness and OSA severity indicators such as AHI, and various measurements of oxygen desaturation. Nevertheless, other investigators have published similar findings [8, 39,40,41,42]. The absence of association between OSA and fatigue suggests that this subjective complaint may be produced by other factors. However, the fact that the sensation of tiredness or fatigue is relieved in some patients who use CPAP adequately is a strong argument for a causal relationship between sleep apnea and this symptomatology [43, 44].

There is contradictory information in the literature regarding the influence of gender on the report of excessive daytime sleepiness (EDS) in subjects with OSA. Although men more frequently had sleepiness than women (see Table 4), in the multiple regression model, the gender was not a predictor of EDS. This finding was maintained even when the cut-off point to define EDS was changed to greater than 9 or 11 (data not shown). In line with our finding, some authors [4, 6] have observed, in models fitted by confounders, that the complaint of excessive daytime sleepiness was not related to gender. A similar finding was reported in a community population study [29]. However, other studies have reported that both women [8] and men [7] with OSA are more likely to report sleepiness. These discrepancies could be related to the way EDS was defined, differences in the variables included in the predictive models of EDS or study population characteristics. Baldwin et al. [7] considered sleepiness as either “feeling frequently or almost always sleepiness during the day” or an ESS greater than 10. While an ESS greater than 10 was related with male gender (adjusted OR = 0.77; CI, 0.66–0.90 for women), the statement “feeling frequently or almost always sleepiness during the day” was not significantly associated with gender (adjusted OR = 1.06; CI, 0.86–1.32). On the other hand, the study by Chervin et al. [8] did not include BMI as a predictor of EDS in the multiple regression analysis and it was described as an independent predictor of EDS [29].

It has been reported that female gender is an independent predictor of insomnia in general and clinical population after adjusting for numerous confounding variables [4, 6, 7, 15]. However, little attention has been paid to the relationship between subtype of insomnia and gender. We found that sleep onset insomnia, but not maintenance insomnia, was independently associated with female gender. This finding could indicate that OSA would not be a cause of initial insomnia. Given that OSA is a disorder that occurs after sleep onset, it would seem more likely that sleep maintenance insomnia would be causally connected with OSA. Indeed, Chung et al. [30] showed that the most common insomnia complaint was frequent awakenings during the night and the least common complaint was difficulty falling asleep. Also, after 2 years of CPAP treatment, OSA patients often improve midnight insomnia but not early or late insomnia [31]. Thus, it has been hypothesized that sleep maintenance insomnia has causal links with OSA, but sleep onset insomnia might be an independent co-occurring sleep disorder [30]. However, in a community-based study [7], different subtypes of insomnia were associated with female gender after controlling for multiple variables including the AHI.

We did not find a relationship between subjective cognitive complaints and gender. Our results are consistent with a paper that assessed the relation between habitual snoring, sleep apnea, and cognitive complaints (concentration and memory problems) in an adult population-based sample [32]. After adjusting for potential confounders such as gender, both concentration and/or memory problems were related to depression, insomnia, unintended sleepiness, habitual snoring, and sleep apnea. Similarly, in a large cohort of adults, the relationship between objective neurocognitive tests and severity of OSA did not show any gender influence on cognitive performance [33].

We did not observe any association between gender and nocturia reports. This is in line with a recent systematic review of factors associated with nocturia in community-based cohorts [34]. Also, the Sleep Health Heart Study evaluated the association between nocturia, cardiovascular morbidity, and OSA and no relationship between nocturia and gender was found after adjusting for several covariates [35].

Female gender constituted a risk factor for morning headaches. In addition, we observed no association between the report of morning headaches and the severity of OSA measured through AHI and other indicators of oxygen desaturation. Such findings are in agreement with previous studies [36, 37]. We speculate that these observations may reflect the fact that migraine and tension headaches are more prevalent conditions in women than in men [38].

Clinical importance and limitation

The fact that obstructive sleep apnea has different clinical presentations in women and men may have important clinical implications. First, reliance on snoring or self-reported apnea may result in more men being referred for sleep study than women. Second, the fact that women present more nonspecific symptoms, such as tiredness or insomnia, could make physicians turn to other diagnostic possibilities like anxiety or depression, rather than OSA. Finally, understanding gender differences will contribute to the better clinical recognition of OSA in women as well as the provision of proper health care and therapeutic practice.

The present study presents limitations due to its retrospective design, selection bias, and the fact that the study site was a tertiary care center with its inherent referral bias and limited generalizability.

In summary, after adjusting for numerous covariables, we found that women with OSA were more likely to report tiredness, sleep onset insomnia, and morning headaches, and less likely to complain of typical OSA symptoms (snoring, apneas) compared to men. We did not observe any gender influence in the self-report of daytime sleepiness, subjective neurocognitive complaints, maintenance insomnia, or nocturia.

Further studies describing the clinical manifestations of different severity levels of OSA in men and women will substantially contribute to its identification and treatment in women across a wide disease spectrum.