Validation of the Slovene Version of the Stop-Bang Questionnaire in a Primary Practice Setting

Obstructive sleep apnea (OSA) is the most common sleep related respiratory disorder (1) and a standalone risk factor for various clinical conditions, such as hypertension, stroke, depression and diabetes (2). Furthermore, OSA constitutes a notable contributor to motor vehicle accidents (1). It is linked to a rise in overall mortality rates, especially attributed to coronary artery disease (1, 3).

In order to establish a diagnosis, a sleep study is required. There are several types of sleep study available, and the gold standard is a traditional laboratory type 1 polysomnography (PSG). Increasingly, however, homebased type 3 polygraphy (PG) is being used as it is easier to perform, cheaper, and more mobile (4, 5). The various types of sleep studies are listed in Table 1.

The result of such a study is expressed by the apnea-hypopnea index (AHI) for PSG and respiratory event index (REI) for PG, which indicate the average number of apneas and hypopneas per hour of sleep or bedrest, respectively (6). Based on the number of these events, OSA is categorised as mild (5≤AHI/REI<15), moderate (15≤AHI/REI<30) and severe (AHI/REI≥30) (7).

Estimates suggest that nearly 1 billion adults aged 30– 69 years worldwide could have OSA (8). Roughly 80% of individuals experiencing moderate-to-severe OSA are believed to remain undiagnosed (7).

Sleep studies are time consuming, labour intensive, and can be costly (9, 10). For this purpose several risk stratification questionnaires have been developed in order to assess the pretest probability of OSA (5). Such screening methods have become important, especially in primary care (11).

The STOP-BANG questionnaire (SBQ) was developed as a preoperative screening tool for OSA (12). Due to its practicality and high sensitivity, the SBQ has been validated in surgical and sleep clinic settings worldwide. However, its validity has been explored to a much lesser extent in the general population and in primary care settings (13–15).

We have previously published the details pertaining to the translation, adaptation, test-retest reliability, and internal consistency as well as the validation of the Slovene SBQ in a sleep laboratory setting (16).

The aim of the current study was to validate the Slovene version of the SBQ in a primary practice setting.

A total of 153 participants were included, of whom 75 (49.0%) were diagnosed with OSA based on a manual REI≥5. The detailed classification into OSA severity levels and descriptive statistics of the sample are provided in Table 2.

The correlation between the SBQ and manual REI, as assessed by the Pearson correlation coefficient, was significant (p<0.00), and the details are presented in Figure 1.

Figure 1.

The sensitivity, specificity, positive (PPV) and negative predictive values (NPV) at different SBQ cutoff values and for different severities of OSA are given in Table 3.

ROC analysis revealed AUC values of 0.823 (95% CI: 0.758– 0.888) for any OSA (REI≥5), 0.819 (95% CI: 0.742–0.869) for moderate to severe OSA (REI≥15), and 0.847 (95% CI: 0.695–0.999) for severe OSA (REI≥30). The receiver operating characteristic curves are shown in Figure 2.

Figure 2.

The Slovene translation of the SBQ showed good correlation with and diagnostic accuracy for OSA in the primary practice setting. At the standard cutoff of ≥3, the SBQ demonstrated an area under the curve of 0.82 for any OSA (REI≥5), 0.82 for moderate and severe OSA (REI≥15) and 0.85 for severe OSA (REI≥30).

The prevalence of OSA (REI≥5) in our primary practice setting was notably high, at 49.0%, with 21.5% exhibiting moderate to severe OSA (REI≥15). Two key factors that may account for this high rate of OSA in primary practice are the age of the patients (with an average age of 49.7 years ±13.1) and the presence of comorbidities. Notably, OSA is more frequently observed in older individuals (19), it often coexists with other chronic diseases (20) and it is older individuals with chronic illness who are more frequent visitors to primary practice clinics (21). Muñoz-Gómez, who validated the Spanish version of the SBQ, using type 3 PG, for primary practice setting, found that 61.5% of the participants had OSA (REI≥5) with 38.8% having moderate and severe OSA (REI≥15). Even higher rates of OSA among patients were reported by Bailes and Fichten and their colleagues, who recruited patients older than 45 years in primary practice. The average age of patients in this study was 57.5 (±11.5) years, with 75% of patients having an AHI≥10, as assessed by PSG (22).

There is no data for the prevalence of OSA in the Slovene general population. A previous study showed an OSA prevalence of 69.6% (REI≥5) in a Slovene sleep clinic, with 47.2% having moderate and severe OSA (REI≥15) (16). The differences in prevalence and severity of OSA in the two environments are to be expected, as there is a gatekeeper system in place in Slovenia and thus all the patients referred to a sleep clinic are in a sense pre-screened.

We found a good correlation between the Slovene version of the SBQ in the primary practice setting and REI. The sensitivity, specificity, NPV and PPV were also impressive. Sensitivity was, however, somewhat lower than in the sleep lab. These differences could be attributed to the small sample sizes, especially in the primary practice setting where the disease prevalence was lower and cases of OSA were on average milder.

The diagnostic ability of the Slovene SBQ, at a cutoff of ≥3, for any OSA (REI≥5) assessed by ROC analysis yielded a high AUC of 0.82, this being even higher than the 0.76 in the Slovene sleep clinic (16). Muñoz-Gómez, who validated the Spanish version of the SBQ for primary practice setting, found an AUC of 0.69 for any OSA (23) and a metaanalysis of sleep clinics showed an AUC of 0.74 (24). The AUC improved for the detection of moderate to severe OSA with the Spanish version in the primary practice setting with an AUC of 0.77 (23), whereas it stayed more or less the same with our version.

Youden’s index, which provides the optimal threshold value based on the longest vertical distance from the diagonal line to the ROC curve, was 3 for all OSA severity levels. This threshold value aligns with the classic threshold recommended by Chung (12).

Many studies have been published validating the SBQ in sleep laboratory setting (25), including a Slovene version (16). There have also been studies validating the SBQ in the general population (25), however, publications validating the SBQ in primary practice setting, the most ubiquitous medical setting, are relatively rare. Compounding the problem, there have also been articles claiming to validate the SBQ in a primary practice setting even though only patients suspected of having a sleep disorder were included and then referred to a sleep clinic for a sleep study (11), something we would consider a pre-screened sleep clinic population.

In our study, we made a methodological restriction by setting the maximum age of the participants at 70 years. While this probably had some impact on the representativeness of our sample compared to the broader family medicine clinic demographic, it was strategically employed to optimise participant engagement. We anticipated that this age limit would allow for a more seamless integration of participants into the study, particularly in terms of understanding instructions, correctly completing questionnaires, and proficiently utilising the designated equipment. Furthermore, compliance with continuous positive airway pressure (CPAP) therapy decreases with age, with the adherence of patients decreasing significantly from the age of 65-69 years, and decreasing further with increasing age (26).

The study was further hindered by the COVID-19 pandemic, extending the patient recruitment period from the initially planned one year to four years. This included a two-year hiatus, after which the pace of patient recruitment was slower than anticipated. In addition, the lack of funding for patient recruitment was a significant constraint. Another limitation was the decision to use type 3 PG instead of the gold standard, that is type 1 PSG. Type 3 PG is now routinely used in clinical practice for the diagnosis of OSA, as it has been shown to be a reliable, cost-effective and simpler alternative to PSG (26, 27). However, we were conscious of the limitations of type 3 PG and therefore excluded patients who were regularly taking sedatives, opioids and tranquilizers, as well as patients with heart failure, neuromuscular disease, COPD stage D, and so on, in whom central or mixed types of apnea and other sleep disorders are more common and for whom a type 1 PSG would be preferable according to the AASM guidelines. Our study was not the first to utilize PG for the validation of the SBQ questionnaire, as Reis et al. (27) and Muñoz-Gómez et al. (23) also did so. We are also of the opinion PSG is too complex and impractical for use in a primary practice setting.

With this study, we have confirmed the validity of the Slovene translation of the SBQ as a reliable instrument for OSA risk stratification in the primary practice setting.