The rising prevalence of sleep apnea (SA) asks for simple but accurate tools to identify patients with this disorder. Detection and treatment, especially for patients with cardiovascular comorbidities, enables the prevention of various serious consequences. Some easy-to-use questionnaires have been developed to infer SA presence from the existence of related symptoms (e.g., daytime sleepiness), demographic characteristics (e.g., old age), and comorbidities (e.g., hypertension). Can these questionnaires perform on their own as a reliable screening tool? A recently published paper demonstrates how this seemingly ideal solution may not be sufficient.
Traaen et al. aimed to characterize SA in patients with paroxysmal atrial fibrillation (PAF). They prospectively studied 579 PAF patients (72.9% men, mean age 59.9±9.6 years) comparing two methods—ambulatory sleep recordings and sleep questionnaires. The recordings were performed at the patients’ home for two consecutive nights, using home sleep apnea testing (HSAT). The apnea-hypopnea index (AHI) was defined as the number of apneas and hypopneas per hour of recording time, and apneas were divided into obstructive or central depending on whether or not, respectively, an increased respiratory effort was associated with the apnea.
SA related symptoms and demographic characteristics were measured by three commonly used questionnaires:
- Epworth sleepiness scale (ESS): to assess daytime sleepiness
- Berlin questionnaire (BQ): to assess snoring and cessation of breathing, daytime sleepiness, history of hypertension and BMI
- STOP-Bang questionnaire (SBQ): to assess snoring, tiredness, observed apnea, high blood pressure, BMI, age, neck circumference, and gender
A total of 479 (82.7%) patients had mild SA (AHI>5), whereas moderate-severe SA (AHI>15) was diagnosed in 244 patients (42.1%). The type of SA was predominantly obstructive, with a median AHI of 12.1 (6.7–20.6) (range 0.4–85.8). SA severity was significantly associated with increased age, BMI, waist and neck circumference, and body and visceral fat. However, despite this link with some of SA associated components, none of the questionnaires discriminated well between patients with or without SA.
Although increased sleepiness is a common symptom of SA, there was no association between ESS and AHI (R2 = 0.003, p = 0.367, see Figure). Even patients with the most severe SA, with hypopneas lasting up to 112s, did not complain of daytime sleepiness. Similarly, both the BQ and the SBQ predicted a lower prevalence of moderate-severe SA compared to diagnostic testing and performed on SA prediction with low sensitivity (46% to 84%) and specificity (46% to 65%).
Study results suggest that the commonly used questionnaires are not effective for SA detection in AF population; while SA prevalence in this population was 82.7%, the questionnaires predicted a much lower prevalence of 42.5% to 66.9%. When looking at the general population, such results imply that AF is a better predictor for SA than these questionnaires. This connection between AF and SA, first demonstrated in 1983 by Guilleminault et al., is independent of other risk factors and mediated through activation of the central nervous system, involving processes of intermittent hypoxemia and hemodynamic changes. Later studies suggested the pathway between the two diseases is complex and may be bi-directional.
Importantly, the results of this study are not unique to the AF population but reflect a general problem with the common questionnaires. In a 2018 meta-analysis of 39 studies comprising 18,068 subjects, Amra et al., tested the accuracy of SA screening questionnaires against polysomnography (PSG). They found that the accuracy of the questionnaires was highly variable between studies and that different questionnaires were better at predicting different levels of SA, together making them hard to be generally trusted. Even in cases of better sensitivity, the specificity was low leading to a high number of false-positive subjects.
Current guidelines from the European Society of Cardiology recommend screening for symptoms of SA in patients with cardiovascular diseases. Since the existing questionnaires are not consistently reliable in their selection performance and yield different estimations in different clinical populations, it is still essential to rely on diagnostic testing for SA in cardiovascular patients, specifically in AF patients referred for ablation that may not present symptoms of SA and would significantly benefit from its treatment.
Traaen GM, Overland B, Aakeroy L, et al. Prevalence, risk factors, and type of sleep apnea in patients with paroxysmal atrial fibrillation. IJC Heart Vasc. 2020;(26):00447.
Guilleminault C, Connolly SJ, Winkle RA. Cardiac arrhythmia and conduction disturbances during sleep in 400 patients with sleep apnea syndrome. Am. J. Cardiol. 1983;52(5):490-494.
Amra B, Rahmati B, Soltaninejad F, Feizi A. Screening questionnaires for obstructive sleep apnea: an updated systematic review. Oman Med J. 2018;33(3):184.