The Berlin Questionnaire in pregnancy predominantly identifies obesity

Abstract

Study Objectives:

Obstructive sleep apnea (OSA) is common in pregnant women and is a risk factor for poor perinatal outcomes. The Berlin Questionnaire (BQ) is a validated OSA screening tool that is often used in pregnancy. However, its poor performance in this population is likely attributed to the scoring paradigm that primarily identifies obesity. Moreover, the associations between the BQ and pregnancy outcomes are often those same outcomes that are obesity related. Therefore, this study examined associations between each of the 3 BQ domains, independently and jointly, in relation to gestational diabetes (GDM) and hypertensive disorders of pregnancy (HDP).

Methods:

Pregnant third-trimester women were recruited from a tertiary medical center and completed the BQ, which includes 3 independent domains: snoring, sleepiness, and obesity/high blood pressure. Medical records were accessed for diagnoses of GDM and HDP.

Results:

Of the 1,588 pregnant women, 44% had a positive BQ score. Women with a positive score for domains of snoring exclusively, sleepiness exclusively, or their combination did not have an increased risk for GDM or HDP. However, women without snoring or sleepiness, but with a positive score on the body mass index/blood pressure domain had increased odds of GDM (odds ratio 2.0, 95% confidence interval 1.3–3.3) and HDP (odds ratio 2.9, 95% confidence interval 1.6–5.5). Furthermore, any positive score in domain combinations that included body mass index/blood pressure had increased odds of GDM and HDP compared with negative scores in all domains. In addition, in the presence of obesity without hypertension, snoring, or sleepiness, the odds of GDM and HDP were similarly increased.

Conclusions:

The poor performance of the BQ in screening for OSA risk may be attributed to its predominant reliance on identification of obesity.

Citation:

O’Brien LM, Levine RS, Dunietz GL. The Berlin Questionnaire in pregnancy predominantly identifies obesity. J Clin Sleep Med. 2021;17(8):1553–1561.

BRIEF SUMMARY

Current Knowledge/Study Rationale: The Berlin Questionnaire is often used to screen for obstructive sleep apnea risk in pregnant women, yet its performance in this population is poor. We hypothesized that its scoring paradigm renders the Berlin questionnaire as a proxy for maternal obesity rather than obstructive sleep apnea.

Study Impact: We have demonstrated that the body mass index/blood pressure domain is the primary driver of relationships between positive Berlin Questionnaire scores and gestational diabetes/hypertensive disorders of pregnancy. These findings are similar to associations between maternal obesity and gestational diabetes/hypertensive disorders of pregnancy, which may explain why the Berlin Questionnaire is a poor predictor of obstructive sleep apnea risk in this population.

INTRODUCTION

There is now robust data to demonstrate an independent association between sleep-disordered breathing during pregnancy and poor maternal outcomes, particularly maternal hypertension and diabetes. ^1,2 Such relationships have been reported in numerous studies utilizing self-reported measures such as presence of habitual snoring, ^3–5 diagnosis codes from medical records, ^6–8 and an objective measure of obstructive sleep apnea (OSA). ⁹ Although polysomnography is the gold standard for identification of OSA, sleep studies are expensive, labor-intensive, and not readily available. Screening tools are therefore necessary to identify women at high risk because OSA can be treated in pregnancy and could improve maternal and fetal outcomes.

Screening tools for OSA that have been specifically developed for pregnant women are lacking. Most studies identify OSA risk in pregnancy either by the presence of snoring ^3,5 or screening tools that have been validated in the general or sleep clinic populations. ^10–12 Nonetheless, the Berlin questionnaire (BQ), ¹³ a tool developed in primary care patients, is commonly used in pregnancy. ^{10,11,14–22} The sensitivity and specificity of the BQ has been reported to range from 37 to 86% and from 53 to 84%, respectively, among primary care patients. ^13,23–26 However, despite its widespread use in pregnancy, the utility of the BQ is questionable in pregnant women. ²⁷ This could be due to the inclusion of key OSA risk factors, such as daytime sleepiness or significant weight gain, that represent common physiologic phenomena in pregnant women.

A common risk factor for OSA is obesity. Indeed, the prevalence of OSA symptoms increases with higher body mass index (BMI). ^28,29 Obesity is particularly relevant to pregnancy, because not only is the prevalence of excessive weight high in reproductive-age women, ³⁰ the vast majority of pregnant women, regardless of their preconception BMI, gain significantly more than 10% body weight.

Among pregnant women, the scoring algorithm of the BQ may be measuring obesity rather than OSA risk, as 1 of the 3 BQ domains assesses obesity. Obesity is a well-known OSA risk factor, and both obesity and OSA share similar associations with adverse pregnancy outcomes: gestational hypertension, preeclampsia, gestational diabetes, caesarean section, and fetal growth. ^2,3,31–33 Therefore, the goal of the current study was to highlight the subdomain of obesity within the BQ as a strong predictor of gestational diabetes and hypertensive disorders of pregnancy.

METHODS

Study design and population

This cohort of pregnant women was recruited from prenatal clinics within a large tertiary medical center. Women were eligible to enroll in the study if they were at least 18 years old and at least 28 weeks pregnant with a singleton gestation. Written informed consent was obtained and this study was approved by the Institutional Review Board.

Women completed brief screening tools about their sleep, including the BQ, ¹³ which is the focus of the present study. Following delivery, medical records were accessed to ascertain several key variables, including participant demographics, prepregnancy and current BMI, parity, presence of a diagnosis of gestational hypertension, pre-eclampsia, and/or gestational diabetes. In addition, delivery information was obtained. All women received a $10 gift card for their participation.

Measures

Berlin Sleep Questionnaire

The BQ effectively assesses OSA risk in adults. ¹³ The instrument has been used extensively in primary care populations and has been validated against polysomnography. ^13,23,24 A positive BQ has been shown to predict the presence of OSA with a sensitivity of 0.86, a specificity of 0.77, a positive predictive value of 0.89, and a likelihood ratio of 3.79, ¹³ although these metrics differ by population. ³⁴ The BQ is comprised of three domains: snoring, sleepiness, and BMI/blood pressure (BP). The snoring domain is scored as positive with frequent symptoms (> 3 or 4 times per week) in 2 or more questions about snoring; the sleepiness domain is scored positive in the presence of frequent (> 3 or 4 times per week) wake time sleepiness, drowsy driving, or both. The BMI/BP domain is scored as positive with either a history of high blood pressure (chronic hypertension) or a BMI > 30 kg/m² (or both). A positive response in 2 of the 3 domains identifies those at high risk for OSA. Furthermore, because 1 of the domains of the BQ includes either high BP or a BMI > 30 kg/m², this domain was further examined as 2 separate subdomains: obesity or chronic hypertension.

Habitual snoring

The single question item regarding habitual snoring was analyzed both separately and as part of the scoring paradigm for the BQ (see above). Habitual snoring has been strongly and reliably associated with the polysomnogram-derived apnea-hypopnea index (AHI) or respiratory disturbance index, ^35,36 and in women, a report of “often” or “usually (always or almost always)” snoring is associated with odds ratios of 3.8 and 16.3, respectively, for objectively confirmed OSA. ³⁵

Pregnancy outcomes

Medical records were reviewed for diagnoses of gestational hypertension/pre-eclampsia (considered as hypertensive disorders of pregnancy; HDP) as well as gestational diabetes mellitus. Clinical diagnoses were obtained from medical coding using the International Classification of Diseases, Ninth Revision (ICD-9).

Covariates

Demographic, health, and medical data were evaluated as potential predictors of gestational diabetes mellitus (GDM) and HDP. Demographic covariates included race/ethnicity (Caucasian, African American, or Asian and other—Hispanic, Native American, American Indian, or multiracial), maternal education (“less than high school”, “high school”, “some college”, “Bachelor’s degree or higher”), and maternal age at enrollment. Health and medical information included maternal smoking (yes/no), parity (dichotomized), and history of HDP (yes/no).

Statistical analysis

We calculated the frequencies and proportions of maternal characteristics in women who screened positive or negative with the BQ using descriptive statistics procedures, Pearson’s chi-square, and t tests.

To investigate the associations of the individual domains, independently and jointly, we created 3 variables that corresponded to the 3 domains: snoring, sleepiness, and BMI/BP. Each of these 3 variables included 5 categories representing all possible combinations of responses. For example, the 5 categories of the sleepiness variable represented the following possible responses: 0 = negative across all 3 domains, 1 = positive in the sleepiness domain and negative in the other 2, 2 = positive in both sleepiness and snoring domains and negative in BMI/BP, 3 = positive in both sleepiness and BMI/BP domains and negative in snoring domain, 4 = positive in all 3 domains; and 5 = negative in sleepiness domain and positive in at least 1 of the other domains ( Table 1 ).

Table 1.

Description of categories of three variables corresponding to Berlin Questionnaire domains.

Variables Category	n	Berlin Questionnaire Domains
		Sleepiness	Snoring	BMI/BP
Sleepiness
Negative for all domains	346
Sleepiness only	108	X
Sleepiness + snoring	119	X	X
Sleepiness +BMI/BP	193	X		X
Sleepiness + Snoring + BMI/BP	242	X	X	X
Snoring OR BMI/BP OR Snoring + BMI/BP	486		X
				X
			X	X
Snoring
Negative for all domains	346
Snoring	202		X
Snoring + Sleepiness	119		X
Snoring + BMI/BP	150		X	X
Snoring + Sleepiness + BMI/BP	242	X	X	X
Sleepiness OR BMI/BP OR Sleepiness + BMI/BP	623	X
				X
		X		X
BMI/BP
Negative for all domains	346
BMI/BP	228			X
BMI/BP + Sleepiness	193	X		X
BMI/BP + Snoring	150		X	X
BMI/BP + Sleepiness + Snoring	242	X	X	X
Sleepiness OR Snoring OR Sleepiness + Snoring	429	X
			X
		X	X

Each of the 3 domains (sleepiness, snoring, BMI/BP) includes 5 categories that represent all possible combinations of responses to the Berlin Questionnaire domains. BMI = body mass index, BP = blood pressure.

A bar chart was used to compare the distribution of the pregnancy outcomes, GDM and HDP, among women who scored positive in each of the domains (sleepiness, snoring, BMI/BP). Furthermore, since the latter domain includes a combination of BMI and chronic hypertension, the individual contributions of each of these variables were investigated by repeating the above analyses in 2 subsamples: 1 that included only women with obesity (BMI ≥ 30 kg/m²) without reported chronic hypertension and a second that included only women of normal weight (BMI < 30 kg/m²) with chronic hypertension. In addition, we used the single-item snoring frequency question within the snoring domain to examine the proportions of GDM and HDP among women who reported frequent snoring in pregnancy. In a subsequent step, we calculated frequencies and proportions of pregnancy outcomes across cumulative BQ domains.

Logistic regression analysis was performed to examine bivariate associations of the 3 BQ domains, independently and jointly, with GDM and HDP. In adjusted odds ratio models, results controlled for maternal age, race, education, smoking, history of hypertensive disorders of pregnancy and parity. A P value < 0.05 was considered statistically significant. All data obtained were double-entered into a database to ensure accuracy and analyzed with R 3.3.2. (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Overall, 1,588 pregnant women are included in the analysis, with 44.3% screening positive for OSA by the BQ. Note that responses of “don’t know” to the snoring item were excluded (n = 37). Analyzing as negative or positive responses did not change the results. Maternal characteristics by the BQ screening results are presented in Table 2 . Women with a positive BQ screen attained less education, had a higher mean BMI, were more likely to smoke, and had a history of HDP. Compared with women diagnosed with HDP, those diagnosed with GDM were more likely to obtain positive screening results in most domains of the BQ and in the single snoring frequency question within the snoring domain. In contrast, within the BP domain, women with a diagnosis of HDP were more likely to screen positive than women with a diagnosis of GDM ( Figure 1 ).

Table 2.

Maternal characteristics by screening results for the Berlin Questionnaire.

Maternal Characteristics	n	Positive BQ	Negative BQ	P
Maternal age at enrollment (years)	1,588	29.8 ± 5.9	29.7 ± 5.7	0.9
Maternal racial background
Caucasian	1,124	510 (45%)	614 (55%)	< 0.01
African American	230	109 (47%)	121 (53%)
Asian/Other	231	83 (36%)	148 (64%)
Maternal education
Less than high school	137	66 (48%)	71 (52%)	< 0.01
High school	328	172 (52%)	156 (48%)
Some college	336	171 (51%)	165 (49%)
Bachelors or higher	759	280 (37%)	479 (63%)
Prepregnancy BMI (kg/m2)	1,588	30.2 ± 8.1	23.6 ± 5.2	< 0.01
Parity	1,588	1.7 ± 1.1	1.62 ± 1.2	0.09
Current smoker	191	108 (57%)	83 (43%)	< 0.01
Gestational week at enrollment (weeks)	1,588	34.3 ± 3.7	34.1 ± 3.7	0.03
History of HDP	79	44 (56%)	35 (44%)	0.05

Continuous variables were compared using t-tests and proportions were compared using chi-square tests. BMI = body mass index, BP = blood pressure, BQ = Berlin Questionnaire, HDP = hypertensive disorders of pregnancy.

Figure 1. Frequencies of pregnancy outcomes by domains of the Berlin Questionnaire.

Note that the domain of BMI/BP is shown separately as its two component parts. BMI = body mass index, BP = blood pressure, GDM = gestational diabetes mellitus.

Pregnant women who scored positive on only 1 of the 3 BQ domains had similar frequencies of GDM compared to HDP ( Table 3 ). In contrast, the distribution of GDM and HDP was significantly different for women with a positive score in any combination of 3 domains ( Table 3 ). Of note, the inclusion of the BMI/BP domain (or just the BMI without the BP) with any other domain drove the observed increase in GDM frequency and a positive response to the presence of high blood pressure was associated with HDP ( Figure 2 ). In a subsample restricted to obese and nonhypertensive pregnant women only (ie, none of whom had a positive score in either the sleepiness or snoring domains), GDM frequency was twice the HDP frequency ( Table 3 ). Similarly, of the n = 12 nonobese but hypertensive pregnant women without a positive score in the sleepiness or snoring domains, there were n = 3 women with GDM ( Table 3 ).

Table 3.

Pregnancy outcomes across the Berlin Questionnaire domains: individual domains and combinations.

Berlin Questionnaire Domains	n	Gestational Diabetes	Hypertensive Disorders of Pregnancy	P
Sleepiness only	202	17 (8.4%)	12 (5.9%)	0.33
Snoring only	108	19 (17.6%)	15 (13.9%)	0.46
BMI/BP only	228	46 (20.2%)	32 (14.0%)	0.14
Sleepiness + Snoring	119	9 (7.6%)	19 (15.9%)	0.05
Sleepiness + BMI/BP	193	34 (17.6%)	18 (9.3%)	0.02
Snoring + BMI/BP	150	46 (30.7%)	26 (17.3%)	0.01
Sleepiness + Snoring + BMI/BP	242	65 (26.8%)	43 (17.7%)	0.02
Restricted to: women with obesity (BMI ≥ 30 kg/m2) and without hypertension
Sleepiness	0	-	-	-
Snoring	0	-	-	-
BMI	202	41 (20.3%)	21 (10.4%)	0.01
Restricted to: women without obesity (BMI < 30 kg/m2) and with hypertension
Sleepiness	0	-	-	-
Snoring	0	-	-	-
BP	12	3 (25%)	5 (42%)	0.38†

Continuous variables were compared using t-tests and proportions were compared using chi-square tests. †Fisher’s Exact Test. BMI = body mass index, BQ = Berlin Questionnaire, BP = blood pressure.

Figure 2. Proportion of women with gestational diabetes and hypertensive disorders of pregnancy in relation to the domains of the Berlin Questionnaire.

Note that the domain of BMI/BP is also shown separately as its two component parts (BMI only as well as BP only). The 3 far right columns (all nonobese) have small denominators between n = 8 and 14). BMI = body mass index, BP = blood pressure, GDM = gestational diabetes mellitus.

The associations of the BQ domains, independently and jointly, are presented in Table 4 . In adjusted models, women with a positive score in the BQ domains of snoring alone, sleepiness alone or their combination did not have an increased risk for GDM or HDP. However, women absent of snoring or sleepiness, but with a positive score in the BMI/BP domain, had significantly increased odds of GDM and HDP odds ratio (OR) 2.0 (95% confidence interval [CI] 1.3, 3.3) and OR 2.9 (95% CI 1.6, 5.5), respectively. Pregnant women with a positive score in any domain combinations that included BMI/BP had increased odds of GDM and HDP compared with women with a negative score in all domains. For example, women with a positive score in the BMI/BP alone, BMI/BP and sleepiness, BMI/BP and snoring, and an intersection of all 3 domains, had increased HDP odds compared with women who were negative on all domains, OR 2.9 (95%CI 1.6–5.5), OR 2.2 (95%CI 1.1–4.4), OR 2.9 (95%CI 1.5–5.7), and OR 4.6 (95%CI 2.6–8.6), respectively. Of note, women absent of a positive score in the BMI/BP domain but with a positive score in the other 2 domains or their combination had similar odds of GDM and HDP as those who were negative on all domains.

Table 4.

Associations of the BQ domains and their combinations with pregnancy outcomes.

BQ Domains	n	Gestational Diabetes (n = 278) Odds Ratios (95% CI)		Hypertensive Disorders of Pregnancy (n = 189) Odds Ratios (95% CI)
		Unadjusted Models	Adjusted Models	Unadjusted Models	Adjusted Models
Sleepiness
Negative for all domains	346	Reference	Reference	Reference	Reference
Sleepiness only	202	0.7 (0.4, 1.3)	0.7 (0.4, 1.4)	1.8 (0.5, 2.4)	1.2 (0.5, 2.5)
Sleepiness + Snoring	119	0.6 (0.3, 1.2)	0.6 (0.3, 1.2)	1.6 (0.7, 3.4)	1.5 (0.6, 3.5)
Sleepiness + BMI/BP	193	2.3 (1.4, 3.6)	2.3 (1.4, 3.7)	2.6 (1.4, 4.8)	2.2 (1.1, 4.4)
Sleepiness + Snoring + BMI/BP	242	2.6 (1.7, 4.1)	2.4 (1.5, 3.7)	4.5 (2.6, 8.0)	4.5 (2.5, 8.4)
Snoring OR BMI/BP OR Snoring + BMI/BP	486	1.8 (1.2, 2.7)	1.8 (1.2, 2.7)	3.0 (1.8, 5.2)	2.4 (1.4, 4.4)
Snoring
Negative for all domains	346	Reference	Reference	Reference	Reference
Snoring	202	1.3 (0.7, 2.5)	1.2 (0.6, 2.2)	1.4 (0.5, 3.1)	0.9 (0.3, 2.5)
Snoring + Sleepiness	119	0.6 (0.3, 1.2)	0.6 (0.3, 1.2)	1.6 (0.7, 3.4)	1.5 (0.6, 3.5)
Snoring + BMI/BP	150	2.1 (1.3, 3.5)	1.9 (1.1, 3.2)	3.8 (2.0, 7.1)	2.8 (1.5, 5.6)
Snoring + Sleepiness + BMI/BP	242	2.6 (1.7, 4.1)	2.4 (1.5, 3.7)	4.5 (2.6, 8.0)	4.6 (2.6, 8.5)
Sleepiness OR BMI/BP OR Sleepiness + BMI/BP	623	1.6 (1.1, 2.3)	1.6 (1.1, 2.5)	2.4 (1.4, 4.1)	2.1 (1.2, 3.8)
BMI and BP
Negative for all domains	346	Reference	Reference	Reference	Reference
BMI/BP	228	1.8 (1.2, 2.9)	2.0 (1.3, 3.3)	3.3 (1.9, 6.1)	2.9 (1.6, 5.5)
BMI/BP + Sleepiness	193	2.3 (1.4, 3.6)	2.3 (1.4, 3.8)	2.6 (1.4, 4.8)	2.2 (1.1, 4.4)
BMI/BP + Snoring	150	2.1 (1.3, 3.5)	1.9 (1.2, 3.3)	3.8 (2.0, 7.1)	2.9 (1.5, 5.7)
BMI/BP + Sleepiness + Snoring	242	2.6 (1.7, 4.1)	2.4 (1.5, 3.8)	4.5 (2.6, 8.0)	4.6 (2.6, 8.6)
Sleepiness OR Snoring OR Sleepiness + Snoring	429	0.8 (0.5, 1.3)	0.8 (0.5, 1.3)	1.3 (0.7, 2.5)	1.2 (0.6, 2.3)

GDM odds ratios adjusted for: race, history of HDP, age. HDP odds ratios adjusted for education, race, smoking, parity, history of HDP. CI = 95% confidence interval, BMI = body mass index, BP = blood pressure, GDM = gestational diabetes, HDP = hypertensive disorders of pregnancy.

The single question item regarding habitual snoring (snoring at least 3–4 nights per week) was significantly associated with HDP (adjusted OR 1.84, 95%CI 1.37–2.48) but not GDM (adjusted OR 1.25, 95%CI 0.93–1.67) when adjusted for the same covariates as the earlier models.

DISCUSSION

The BQ is a validated screening tool in the general population for the identification of OSA risk. Although often used in pregnancy, it has shown little clinical utility for the presence of objectively defined OSA. ^19,27,37 Presence of OSA is an independent risk factor for morbidities such as GDM and HDP. ⁹ Despite the known associations between OSA and GDM and HDP, this study demonstrates that associations with pregnancy outcomes are predominantly driven by the BMI/BP domain of the BQ rather than snoring or sleepiness. These findings suggest that the BQ, when used in pregnancy, is likely representing excessive maternal weight.

While the BQ has long been used to screen for high OSA risk in the nonpregnant population, its sensitivity and specificity for OSA are not uniform across different populations. Sensitivity has been reported as 33–94% and specificity as 2–95%, depending on whether studies were community based, from a sleep clinic population, or from a subset of patients. ^{26,34,38–40} Moreover, different studies have used and differing severity thresholds of objectively confirmed OSA as well as investigated the role of sex. Indeed, the BQ appears to demonstrate poor metrics for identification of mild OSA and for OSA in women. ^23,41 In a study of female nurses, the sensitivity of the BQ to detect those with a respiratory disturbance index of at least 5 was only 33% (with a corresponding specificity of 83%), and, of note, the domain of BMI/BP performed the best for prediction of sleep apnea. ⁴¹ In pregnancy, the BQ also performs poorly for identification of OSA, ^27,37 although its sensitivity and specificity differ by trimester with better prediction of OSA in the third trimester. ¹⁴ Similar to nonpregnant populations, the presence of obesity and hypertension have been reported as the strongest predictors of objectively defined OSA in pregnant women. ³⁷

In the pregnant population, the BQ has also been used to investigate OSA risk in relation to pregnancy outcomes such as GDM, HDP, cesarean section, fetal growth problems, and preterm birth. ^{10,11,20,21,42} However, most studies do not identify which of the 3 domains have contributed to a positive BQ screen; rather, they have considered a positive score on any 2 domains. Of note, obesity is an established risk factor for OSA and has been independently associated with the same pregnancy outcomes. ³⁰ Indeed, a recent study that examined pathways between maternal obesity, habitual snoring, and HDP demonstrated that HDP is predominantly driven by maternal obesity. ³²

While weight gain is expected in the third trimester of pregnancy, an individual participant meta-analysis of 39 cohorts from Europe, North America, and Oceania (sample size of 265,279 women) reported that almost 30% of women enter pregnancy overweight or obese. ⁴³ Moreover, one-third of normal weight women were found to have excessive gestational weight gain, with the proportion being even higher in the women who were overweight and obese (66% and 57%, respectively). ⁴⁴ Thus, a significant proportion of women will likely screen positive for the BMI/BP domain of the BQ by virtue of BMI alone.

In the current study, a positive response exclusively to the snoring domain only was not associated with GDM or HDP. It should be noted that this association was examined within a cohort of women who screened negative for both of the other 2 domains, ie, daytime sleepiness and BMI/BP. The snoring domain together with a positive response to the BMI/BP domain, with or without sleepiness was, however, associated with GDM and HDP, as was a positive response to a single item about snoring frequency (the latter would, however, have included women with high BMIs). Similarly, a positive response exclusively to the sleepiness domain was not associated with GDM or HDP but when sleepiness was combined with BMI/BP, with or without a positive snoring domain, there were significant associations with GDM and HDP. While sleepiness is often considered a symptom of OSA, in pregnant women it can also be attributed to the physiological changes of pregnancy. ⁴⁵ Importantly, it is evident from our findings that all significant associations of the BQ domains with GDM or HDP included a positive screen on the BMI/BP domain.

The role of obesity is further highlighted in several studies that have used the BQ in pregnancy. In a large sample of second trimester Thai women, those with high blood pressure at enrollment were excluded, and thus all women who endorsed the BMI/BP domain had a BMI ≥ 30 ⁴⁶ ; almost one-quarter of high risk (ie, positive BQ) women were in this category. In a study of third trimester women, the majority (82%) who screened positive with the BQ had a BMI ≥ 30 kg/m² and maternal obesity at the time of screening increased the odds of a positive BQ 3-fold. ¹⁷ Unsurprisingly, a cohort of Peruvian women that were screened with the BQ also showed that the prevalence of a positive screen was 2%, 8%, and 26% for study participants who were lean, overweight, and obese, respectively. ⁴⁷ Moreover, compared to women of normal weight, women with obesity (≥ 30 kg/m²) had a 13-fold higher odds of a positive BQ screen (adjusted OR  = 13.2, 95%CI: 6.3–28.0). These findings are similar to a large study of Hispanic women, where the adjusted relative risk of screening positive on the BQ by BMI was 9-fold higher (95%CI 4.7–17.4) for women with obesity compare to women who were lean. ¹⁶ Of relevance, in the latter study, an evaluation of the presence of obesity as a single variable demonstrated a similar relative risk for HDP than did a positive BQ screen (adjusted relative risk 1.54, 95%CI 1.24–1.92 and adjusted relative risk 1.90, 95%CI 1.52–2.37, respectively), with presence of chronic hypertension alone being even stronger (adjusted relative risk 2.9, 95%CI 2.34–3.59). ¹⁶ Thus, many women who endorsed domain 3 in the aforementioned study therefore already had major risk factors for the perinatal outcomes under study (GDM, HDP, cesarean section, preterm birth, small- and large-for-gestational age) regardless of OSA.

An additional concern is that application of criteria for domain 3 (BMI/BP) is not consistent across studies. Some studies include only the presence of chronic hypertension and/or obesity, ^16,46 whereas others required pre-eclampsia and/or obesity. ^10,11 Use of the BQ to predict perinatal outcomes, such as HDP, thus becomes problematic with differential inclusion criteria for a positive screen and limits comparisons. While inclusion of chronic hypertension does of course increase the odds of HDP, so does obesity; the contribution of these 2 major risk factors cannot then be teased apart. Importantly, if presence of pre-eclampsia is used as a criterion for domain 3 in the BQ, an overall positive BQ screen cannot then be used to predict presence of pre-eclampsia as a maternal outcome of OSA.

This study has several strengths. This study examined each of the BQ domains independently and in combination with other domains in relation to HDP and GDM and quantified their contribution to the aforementioned outcomes. Second, the large sample size and diverse population of low-risk and high-risk pregnant women attending a tertiary referral center support statistical power and increases external validity of these findings. Finally, this study utilized ICD-9 codes to assess each of the pregnancy outcomes examined in relation to BQ domains. The lack of polysomnographic data to correlate with the BQ could be considered a limitation. However, this study did not aim to determine the utility of the BQ as a screening tool for OSA, rather it was to tease out which of its domains is most predictive of GDM and HDP. Moreover, multiple studies have already demonstrated that the BQ is not helpful for assessment of the presence of OSA. ^19,27,37

CONCLUSIONS

This study has demonstrated that use of the BQ during pregnancy is likely measuring excessive maternal weight rather than OSA; this could explain its poor performance in pregnant women. The single item question about habitual snoring embedded within the snoring domain of the BQ demonstrated significant associations with GDM and HDP. In the absence of a reliable and simple OSA screening tool validated in the pregnant population, use of such a single item question appears to be a more efficient approach to screen women for OSA than the BQ.

DISCLOSURE STATEMENT

All authors have seen and approved the manuscript. This study was funded by the National Institutes of Health National Heart, Lung, and Blood Institute (HL089918) (to L.M.O.); a T32 Grant from the National Institute of Neurological Disorders and Stroke (NIH/NINDS T32 NS007222) (to GLD); and by a grant from National Institute of Child Health and Human Development (NIH/NICHD F32HD091938). The authors report no conflicts of interest.

ABBREVIATIONS

AHI

apnea-hypopnea index

BMI

body mass index

BP

blood pressure

BQ

Berlin Questionnaire

GDM

gestational diabetes mellitus

HDP

hypertensive disorders of pregnancy

ICD

International Classification of Diseases

OR

odds ratio

OSA

obstructive sleep apnea

95% CI

95% confidence intervals