Obstructive Sleep Apnea (OSA) syndrome is a common disorder that affects 9 to 24 % of adult middle-aged population,1 characterised by repeated episodes of upper airway obstruction during sleep, intermittent arterial oxygen desaturation, increasing respiratory efforts and sleep disruption.2

Diagnosis suspicion of OSA is of particular relevance, since it has been implicated with high medical 3–6 and peri-operative 7,8 morbidity and mortality and has an effective treatment available.9

Polysomnography is the gold standard test for OSA diagnosis, 10 although its availability, as also the cardiorespiratory sleep study, may be limited. In fact, with the growing recognition of OSA epidemiological relevance and pathophysiological consequences, the medical community has been confronted with a rise in requests for the diagnostic test, with implications regarding resources rationalization. 11 With this in mind, screening tools, such as questionnaires and clinical models, have been developed, trying to combine different OSA risk factors that in a cost-effective way could help clinicians in identifying patients that should be referred or have priority on the waiting list for diagnostic testing.12,13

Berlin Questionnaire (BQ), an outcome from the Sleep in Primary Care Conference, in April 1996 in Berlin, is one of the most recognized screening tools used in this area. 14 It includes 10 items organized in 3 categories concerning snoring and witnessed apneas (5 items), daytime sleepiness (4 items) and high blood pressure (HBP)/obesity (1 item). Patients are also asked to provide information on age, gender, weight, height, neck circumference and ethnicity. Predetermination of high or lower risk for OSA is based on responses to each category of items.14

Netzer and colleagues validated the BQ as a screening tool in primary care, where it demonstrated a high internal validity (Cronbach alpha correlations 0.86-0.92) and performed accurately with a higher sensitivity and specificity in OSA identification (86 and 77 %, respectively, for an apnea/hipopnea index [AHI] > 5 –positive likelihood ratio [LR] of 3.79–, 54 and 97 % for an AHI > 15, 17 and 97 % for an AHI > 30).14

More recently, BQ had been applied in tertiary settings with variable results.15–19

In general cardiovascular patients, namely in a subgroup with atrial fibrillation, BQ showed a good performance (sensitivity of 86 % and specificity of 89 % for an AHI > 5),15 although in another subgroup with resistant HBP, an adapted BQ version presented a lower specificity (sensitivity of 85.5 % and specificity of 65 % for an AHI > 10).16

The same occurred in surgical patients, where BQ demonstrated a high-moderate sensitivity and again a lower specificity (65.6 and 60 % respectively for an AHI > 5, 74.3 and 53.3 % for AHI > 15, 79.5 and 48.6 % for AHI > 30).17

Lastly, low sensitivity and specificity were found in patients undergoing pulmonary rehabilitation (62.5 and 53.8 %, respectively, for an AHI > 10, 67.2 and 52.8 % for an AHI > 15),18 as well as in a sleep clinic population (68 and 49 % respectively for an AHI > 5, 62 and 43 % for an AHI > 10, 57 and 41 % for an AHI > 15). 19 It should be noted that the last study, besides having a retrospective nature, took place in a general sleep clinic population, so BQ has never been applied as OSA screening tool in a specific sleep disordered breathing clinic (SDBC).

In addition, BQ was originally developed in English, being necessary to translate this questionnaire to other languages, so that it can be applied in other countries and contexts.

The mains goals of this study were: 1) development of BQ Portuguese version; 2) assessing the usefulness of BQ in OSA identification, compared with AHI obtained by cardiorespiratory sleep study, in a SDBC. As a secondary objective, authors analyzed the association between the BQ items and AHI.

In face of most SDBC having long waiting lists, it seems attractive to use a screening tool to prioritize patients needing OSA diagnosis test according to probability that they might have a positive result.

When applied to a SDBC population, BQ demonstrated not to be a good test neither for OSA diagnosis nor for severity defining, presenting for different AHI cut-off points a moderate to high sensitivity (72.1 to 88.4 %), a specificity which in whole was low (39.1 to 50 %) and a lower positive LR (1.44 to 1.49).

Our findings reflected, on the one hand, the high frequency of OSA in this population (83.2 %), which influences the screening test operating characteristics, as also the large number of both false positives and false negatives demonstrated by BQ.

Most enrolled subjects (68.4 %) had a BQ high risk score, as a result of population pre selection at the time of consultation referral, which is predominantly composed of snoring, hypertensive and obese subjects, as noted by the high prevalence of positivity in categories 1 (91.6 %) and 3 (66.3 %).

Besides the above, and although 87.7 % of patients with BQ high risk score presented OSA, the same has not happened in individuals at low risk, where the sleep study ruled out OSA in only 26.7 % of subjects, contributing to poor overall agreement between the two tests (68.4 %).

Although polysomnography remains the gold standard, cardiorespiratory sleep study, having demonstrating a high accuracy in OSA diagnosis,22 and like occurred in the original study of Netzer,14 was the test applied.

BQ performance in this SDBC population is in accordance with results obtained in other studies undertaken in tertiary health care setting, particularly in resistant HBP 16 and surgical patients.17

In patients undergoing pulmonary rehabilitation 18 and in the retrospective study that took place in a general sleep clinic, 19 the obtained results, compared with those in our population, were even less satisfactory. Noteworthy, in the last case, the association of the sleep clinic with the psychiatry department of a tertiary hospital, which contributed to the high proportion of patients with positivity in category 2 (74.6 %) and was postulated by the authors as may have influenced BQ performance. To confirm this hypothesis a shortened BQ version was developed, excluding category 2 and item 4 of category 1, in this case because it was poorly answered. Although BQ sensitivity has increased, specificity remained low (80 and 42 % respectively, for an AHI > 5), similar to the obtained in the original questionnaire, corroborating the importance of other factors, in addition to daytime sleepiness, in the clinical picture of this syndrome. In fact, when analyzing the association of category 2 with AHI, the same authors found no significant correlations.

As previously reported, in our population, none of the category 2 items was associated with the AHI and only items, 2 and 5, of category 1, referring to snoring loudness and witnessed apneas, and category 3, referring to HBP/obesity, were significantly associated to the diagnosis gold standard.

The model established by these three BQ items association had a greater ability to discrimínate OSA for all AHI cut-off points, compared to the original BQ, especially for an AHI > 5. The abbreviated BQ version and the modified score confirmed these findings, although it didn't present enough predictive accuracy, so that only these items could be recommended in OSA identification in a SDBC.

A recent meta-analysis of screening tests for OSA 12 corroborated some of these data, having found that BMI, history of hypertension and nocturnal choking are significant test elements in the more accurate prediction models. Besides, it also demonstrated that the Epworth sleepiness scale was the least accurate predictive questionnaire used in this ambit, possibly because excessive daytime sleepiness occurs commonly in obese individuals without OSA, driven by mechanism other than nighttime sleep deprivation.23

Although BQ remains the most accurate questionnaire for predicting OSA diagnosis,12 it is not an appropriate screening tool for a high risk population in a SDBC. BQ was translated into Portuguese language and can thus be used by our country clinicians, in the contexts in which it is validated.

Authors declare that they don't have any conflict of interest.