Waterpipe (hookah) tobacco use in pregnancy: Use, preferences, and perceptions of flavors

Abstract

Objective:

Waterpipe tobacco (WPT; hookah) use is common in pregnant and reproductive-age women. Sweet flavors contribute to the appeal of WPT and are a potential regulatory target. This study investigated use, preferences, and perceptions of WPT flavors in pregnant WPT users, and the impact of flavor preferences on preconception/prenatal WPT use and exposure biomarkers.

Methods.

58 pregnant WPT users (mean age=27 years) completed a detailed interview regarding their WPT flavors use, preferences, and perceptions. Biomarkers of nicotine and carcinogen exposure (e.g., cotinine, benzene, butadiene) were also collected.

Results:

55% of participants were dual/poly WPT users (i.e., reported use of one or more other tobacco products in addition to WPT). Pregnant WPT users reported nearly exclusive use of flavored WPT, with greater use of menthol/mint (68%) followed by fruit flavors (48%) (P<.001), and greater preferences for fruit followed by menthol/mint flavors (Ps<.05). Harm perceptions did not differ among flavors. Compared to dual/poly WPT users, WPT-only users reported more total WPT use events, greater use of and preference for menthol/mint flavored WPT (Ps<.001), and decreased exposure biomarkers (Ps≤.040). Preference for menthol/mint and fruit flavors predicted more flavored WPT use events during preconception and pregnancy; preference for menthol/mint predicted detectable cotinine and benzene levels but not butadiene.

Conclusions:

This is the first study of WPT flavor use, preferences, and perceptions in pregnant women. Use of and preference for menthol/mint and fruit WPT flavors in this vulnerable population could be considered in regulating WPT flavors to protect the health of women and children.

INTRODUCTION

Waterpipe (hookah) tobacco (WPT) smoking is a centuries-old tradition that evolved in the Middle East and India but has become an increasing public health problem in the US and worldwide. While cigarette use has declined significantly in the US,[1] use of other tobacco products including WPT is increasing, particularly among young adults.[2–7] There is also evidence for increasing WPT use and social acceptability among reproductive age women.[8–11] However, despite well-known causal effects of tobacco use in pregnancy on maternal, fetal, and infant morbidity, only a small number of studies have addressed WPT use in pregnant women, a uniquely vulnerable population.[1, 12] In several studies from Lebanon, Iran, and Jordan, prevalence of WPT use in pregnancy ranged from 5.4–8.7%.[13–18] In the Population Assessment of Tobacco and Health (PATH) study in the US, WPT was the third most prevalent tobacco product used by pregnant women after cigarettes and e-cigarettes.[11] Prevalence of WPT use in pregnancy was 2.5% in the overall sample but 12.4% among pregnant cigarette smokers.

Because WPT involves burning charcoal to heat (flavored) tobacco), WPT users are exposed to toxic combustion products as well as nicotine and other toxicants.[19] In laboratory studies, relative to a single cigarette, WPT smokers were exposed to 1.7 times the nicotine, 3.7 times the carbon monoxide (CO), and 50 times the smoke.[20, 21] WPT use is also associated with exposure to volatile organic compounds (VOCs), including 1,3-butadiene, acrylonitrile, acetaldehyde, and benzene, that have the highest cancer risk indices.[22] WPT use (vs. cigarettes) is associated with particular increases in benzene,[23, 24] a carcinogen linked to leukemia, the most common cancer in children and teens.[25] Finally, WPT use is associated with health outcomes similar to those associated with cigarettes, including respiratory disease and oral and lung cancer.[26–29]

The issue of health risks and toxicity from WPT use is even more salient among pregnant women due to impact of maternal use on both mother and the developing fetus. Decades of research have supported causal links between maternal prenatal cigarette use and obstetric complications (e.g., premature delivery, placental abruption), as well as infant morbidity and mortality (e.g., low birth weight, sudden infant death syndrome).[1, 12] A small number of studies have investigated WPT use and health outcomes among pregnant women in the Middle East (e.g., Lebanon, Iran, Jordan) and Southeast Asia (i.e., Cambodia).[16, 29–31] WPT-exposed infants were 3.7 times more likely to have respiratory problems and were 2.4 times more likely to be born low birthweight (weight less than 2500g) than unexposed infants.[26] Low birth weight infants are at increased risk for admittance to neonatal intensive care and for poorer long-term outcomes.[32] WPT-exposed children were also at increased risk for asthma.[33]

Despite adverse health outcomes of WPT use, it is often perceived as less harmful than cigarette use.[34] Decreased harm perceptions and increased social acceptability of WPT may be due to the social and more sporadic nature of its use, flavorings, and misperception that harmful toxicants from tobacco are “filtered” by the water.[34–36] In addition, there is a dearth of regulations and poor compliance with existing regulations to control the use and spread of WPT.[37] A review of tobacco control legislation in 62 countries found that WPT-specific language was absent from such legislation in the majority of countries.[37] In the US, the Food and Drug Administration (FDA) asserted regulatory authority over WPT in 2016.[38] To facilitate the development of a WPT-specific regulatory framework to protect vulnerable populations such as pregnant women, it is critical to identify WPT characteristics that are associated with higher use and are amenable to regulation.

One reason for the increasing global appeal of WPT is the introduction of maassel, a sweetened, flavored tobacco product made up of shredded tobacco, glycerol, flavors (e.g., candy, fruit), and other additives.[9, 39, 40] Maassel simplified the WPT preparation process—allowing for increased availability, flavor variety, and palatability of WPT. Thus, maassel and WPT flavors represent an important potential regulatory target. Pregnant women may be especially vulnerable to the appeal of WPT flavors due to alterations in taste, cravings, and nausea during pregnancy.[41] We recently showed higher use of and preference for fruit and menthol/mint vs. unflavored e-cigarettes in pregnancy and preconception, and links between preferences for flavored e-cigarettes and lifetime use of e-cigarettes by pregnant women.[42] We also showed higher preference for sweet WPT flavors in reproductive age women.[43]

In the present study, we extend this work to investigate use, preferences, and perceptions of PT flavors in pregnant users, including WPT-only users and WPT users who use other tobacco products (dual/poly WPT users). Our aims were to investigate: (a) flavored WPT use over preconception and first trimester in pregnant WPT-only and dual/poly WPT users, (b) preferences for and perceptions of WPT flavors in pregnant WPT-only and dual/poly users, and (c) the impact of flavor preferences on use and biomarkers of exposure, including nicotine, CO, and VOCs. Based on our prior work [42, 43], we hypothesized that fruit, sweet, and menthol/mint flavors would be used most frequently and preferred by pregnant WPT users. Further, we hypothesized that preferences for sweet and menthol/mint flavors would be associated with greater use of WPT and dual/poly WPT use and higher levels of biomarkers of exposure.

METHODS

Overview

The present study involved a cross-sectional analysis of baseline data from the first cohort of an ongoing longitudinal study of pregnant WPT users. This cohort included 58 pregnant WPT users (26 WPT-only, 32 dual/poly WPT users). All study procedures were approved by local Institutional Review Boards; all participants provided written informed consent. Participants completed a baseline interview and biomarker collection session in early pregnancy. To minimize potential biases, (a) study interviews and data analyses were conducted by different individuals, (b) interview staff received training based on project-specific protocols, (c) 36% of audiotapes from the baseline interview were reviewed for quality control, and (d) biomarkers of use/exposure were used in addition to maternal report data. All participants were compensated $50 for their participation in the baseline session; 36 of the 58 participants (62%) received a $50 bonus for keeping their scheduled appointment.

Screening and Recruitment

Pregnant women responded to invitations (face-to-face contact, brochures, paper and electronic advertisements) to the “Let’s Talk about Hookah” study offered at multiple obstetric offices and community sites within Rhode Island. Although this was a convenience sample of 58 pregnant WPT users in southeastern New England (98% residents of Rhode Island; 2% Massachusetts), recruitment sites were chosen to cover a breadth of pregnant women and to reach women of diverse racial/ethnic identities and a range of socioeconomic status. Face-to-face recruitment was conducted by research staff not involved in prenatal care. One hundred and four women who expressed interest in the study were screened by telephone between August 2017 and August 2018 to determine whether they met the eligibility criteria: (a) English-speaking, (b) aged 18 – 40 years, (c) WPT use at least once during pregnancy or preconception (i.e., 3 months prior to conception), (d) a singleton pregnancy and (e) no severe medical (pre-eclampsia) or psychiatric (bipolar or psychotic disorder) conditions during pregnancy. Nine women were no longer interested when called and 14 were ineligible; 21 women did not attend their baseline interview and could not be rescheduled, leaving 60 participants who consented and enrolled. Two participants were excluded following the baseline interview due to no WPT use during preconception/pregnancy and a bipolar diagnosis. Of the 58 participants, 56 were enrolled via face-to-face recruitment efforts in obstetric offices; two were enrolled from posted advertisements or friends.

Interview Session

Participants completed the baseline interview session during the 1^st or 2^nd trimester (M=14, SD = 4 weeks gestation; range = 7 – 25). The interview assessed (a) demographic information (e.g., age, ethnicity) and (b) general health, pregnancy, and tobacco history. We used the Timeline Follow Back (TLFB) method to assess (c) tobacco and other substance use[44–46] and the Tobacco Flavors Interview[47] to assess (d) flavor preferences.

Measures

Timeline Followback (TLFB)

The TLFB is a structured calendar-based assessment of daily substance use that uses key dates (e.g., holidays, personal events) to cue memory and increase accuracy of recall.[44–46] The TLFB was used to assess daily WPT use and other tobacco/nicotine product use (i.e., cigarettes, e-cigarettes, cigars/cigarillos/little cigars), alcohol and other substance use, and exposure to second-hand WPT and cigarette smoke. The TLFB interview covered each day of pregnancy up to the interview day and the 3 months prior to conception. For each episode of WPT use, mothers were asked about the brand, flavors, amount of use (i.e., quantity and duration), and social context (e.g., where and with whom) of their use. The data collected from the TLFB was used to determine (a) the number of WPT-only and dual/poly WPT users (How many times did you use [TOBACCO PRODUCT] in the past x months?) and (b) the number of flavor use events (Did you smoke hookah that was flavored? If yes: What flavor or flavors did you smoke?). Participants who reported use of any non-WPT tobacco product (i.e., cigarettes, e-cigarettes, cigars/cigarillos/little cigars) during preconception or pregnancy were categorized as dual/poly users. Participants who reported use of flavored WPT during preconception or pregnancy were categorized as using any one of ten flavor categories (see below; e.g., menthol/mint, fruit) during that WPT use event.

Tobacco Flavors Interview: Overview

The Tobacco Flavors Interview involves use of a portable flip chart presentation easel binder.[47] The interview includes a series of questions regarding use, preferences, perceptions, and intentions to use flavored WPT across ten flavor categories (menthol/mint, clove/spice, fruit, chocolate, alcohol [e.g., margarita], coffee, or other beverages [e.g., lemonade], candy or other sweets [e.g., cotton candy, butterscotch], and unflavored) consistent with the flavor categories from the PATH study Wave I.[48] The presentation easel binder includes photos of WPT products, flavor images (e.g., fruit, alcohol), and response choices (e.g., 1–7 scale with anchors of extremely dislike to extremely like).

Tobacco Flavors Interview: Measures

All measures were pilot tested in a sample of 630 reproductive-aged women.[43] Higher ratings indicate more positive preferences for and perceptions of harm for the specific flavor.

Preferences for WPT flavorings.

Preferences for WPT flavors were measured using four items reported on a 7-point Likert scale (extremely dislike—extremely like, extremely unattractive—extremely attractive, extremely unpleasant-extremely-pleasant, not at all interested—very interested).

General and Pregnancy-Specific Risk Perceptions of WPT.

Participants were asked about their perceptions of general (“for your health”) and pregnancy-related (“for pregnant women to use,” and “to the fetus”) risks across all WPT flavors using 3 items reported on a 7-point scale (not at all harmful—very harmful) adapted from published measures.[49, 50]

Biomarkers of WPT/Tobacco Smoke Exposure

Cotinine.

Saliva cotinine is a reliable biomarker for nicotine exposure.[51] Maternal saliva samples were collected at the interview session then frozen until analysis by University of California, San Francisco (UCSF) Clinical Pharmacology Research Laboratory using gas chromatography (GC) with nitrogen-phosphorus (N-P) detection, modified for use with a capillary column.[52, 53] Limit of quantification (LOQ) for cotinine was 10 ng/mL.

Breath

CO concentrations were measured using the Micro+™ basic Smokerlyzer (coVita, Santa Barbara, CA, USA) as a quantitative measure of exposure to combustion products.[54]

Urine Mercapturic Acid Metabolites of VOCs.

Maternal urine samples were collected at the interview session for determination of mercapturic acid metabolites of acrolein (3-hydroxypropylmercapturic acid, 3-HPMA), acrylamide (2-carbamoylethylmercapturic acid, AAMA), acrylonitrile (2-cyanoethylmercapturic acid, CNEMA), benzene (phenylmercapturic acid, PMA), crotonaldehyde (3-hydroxy-1-methylpropylmercapturic acid, HMPMA), ethylene oxide (2-hydroxyethylmercapturic acid, HEMA), propylene oxide (2-hydroxypropylmercapturic acid, 2-HPMA), 1,3-butadiene (sum of isomers 1-hydroxy-3-buten-2-yl-mercapturic acid and 2-hydroxy-3-buten-1-yl-mercapturic acid, abbrev. MHBMA-1+2, and 4-hydroxy-2-buten-1-yl-mercapturic acid, abbrev. MHBMA-3), and methylating agents (methylmercapturic acid, MMA)). Samples were frozen until analysis by UCSF using liquid chromatography with tandem mass spectrometry (LC-MS/MS).[23] LOQs ranged from 0.1 to 5 ng/mL.

Data Management and Statistical Analysis

All data analyses were conducted using Stata 15.[55] Differences in characteristics and biomarkers between WPT-only and dual/poly users were examined using a series of bivariate analyses (chi-square, t-tests) (Table 1). Preferences and perceptions of WPT flavors were significantly skewed and therefore evaluated using Kruskal-Wallis analysis of variance, followed by Dunn’s tests with a Bonferroni adjustment. Differences between flavor preferences and perceptions by user type were assessed using a two-sample Wilcoxon rank sum test (Table 2). Because a ceiling effect was observed for the perceptions of pregnancy harm (67% endorsed “very harmful” for mint/menthol and fruit-flavors), this scale was omitted from all regression analyses but included in Table 2 for descriptive purposes. Biomarker levels were positively skewed and were log transformed prior to descriptive analyses. Biomarker values below the limit of quantification (BLQ) were imputed using standard methods (BLQ/√2).[56], and those in urine were creatinine-normalized to adjust for differences in urine dilution.[57]

Table 1.

Sample Characteristics and Biomarker Levels by WPT Use (N = 58).

	Overall	WPT-Only Users	Dual/Poly Users	p
	N = 58 % (n)	n = 26 % (n)	n = 32 % (n)
Maternal Characteristics
Age (years), M (SD)	27 (5)	27 (5)	27 (5)	.636
Non-White Racea	72% (42)	81% (21)	66% (21)	.199
Latina Ethnicity	53% (31)	73% (19)	38% (12)	.007
≤ High School education	50% (29)	38% (10)	59% (19)	.113
Income <$30,000/year	38% (22)	31% (8)	44% (14)	.311
Unemployed	40% (23)	42% (11)	38% (12)	.710
Unmarried	83% (10)	73% (19)	91% (29)	.078
Unplanned pregnancy	69% (40)	69% (18)	69% (22)	.969
Parity, M (SD)	1 (1)	1 (1)	1 (1)	.316
Pregnancy Tobacco & Substance Use/Exposure
Any Cigarette	33% (19)	0% (0)	59% (19)	NA
Any e-Cigarette	19% (11)	0% (0)	34% (11)	NA
Any Cigar	36% (21)	0% (0)	66% (21)	NA
Any Alcohol	88% (51)	88% (23)	88% (28)	.911
Any Marijuana	52% (30)	23% (6)	75% (24)	<.001
Secondhand WPT exposure	67% (39)	69% (18)	66% (21)	.085
Secondhand smoke exposure	47% (27)	31% (8)	59% (19)	.030
Biomarkers
	n = 57	ncotinine/CO = 25/26	ncotinine/CO = 32/31
Saliva Cotinine (GM, % BLQ, ng/ml)	1.52 (77%)	0.71 (96%)	2.77 (63%)	.002
Breath CO (ppm, % <3 ppm)	1.82 (84%)	1.41 (96%)	2.26 (74%)	.092
Urine VOCs (GM, % BLQ, ng/mg)	n = 31	n = 16	n = 15
Benzene (PMA)	.12 (61%)	0.09 (69%)	0.14 (53%)	.224
Butadiene (MHBMA 1+2, −3)	.08 (68%)	0.08 (63%)	0.07 (73%)	.501
Ethylene Oxide (HEMA)	1.46 (6%)	1.06 (6%)	2.1 (0%)	.075
Methylating Agents (MMA)	11.68 (16%)	8.84 (25%)	15.7 (7%)	.218
Acrylonitrile (CNEMA)	1.52 (55%)	0.52 (69%)	4.8 (40%)	.007
Acrolein (3-HPMA)	231.25 (0%)	175.3 (0%)	310.81 (0%)	.040
Propylene Oxide (2-HPMA)	21.63 (3%)	15.4 (6%)	31.11 (0%)	.039
Acrylamide (AAMA)	54.76 (0%)	41.1 (0%)	74.33 (0%)	.033
Crotonaldehyde (HPMMA)	132.98 (0%)	115.5 (0%)	154.63 (0%)	.191

Note. WPT=Waterpipe Tobacco. GM=Geometric Mean. BLQ=Below the limit of quantification for the assay. CO=Breath Carbon Monoxide levels. VOC=Volatile Organic Compounds; NA= not applicable. N=31 for VOCs (N=16 WPT-only users; N=15 Dual/poly users). Bold values indicate statistically significant differences between WPT-only and dual/poly WPT users.

^aNon-White race includes 21% (n = 12) African American, 1.5% (n = 1) Asian, 1.5% (n = 1) American Indian/Alaska Native, 38% (n = 22) other, and 10% (n = 6) mixed/unknown.

Table 2.

Mean flavor preferences and perceptions by WPT use among pregnant women (N = 58).

	Fruit	Menthol/Mint	Candy	Chocolate	Other Sweets	Alcohol	Coffee	Other Beverages	Spice	Tobacco
Preferences
Liking
All	5.52	5.00	3.75	2.69	3.45	2.86	2.07	2.83	2.24	2.19
WPT-Only	5.31	5.62*	3.31	2.38	3.00	2.92	1.73	2.77	2.38	2.19
Dual/Poly	5.69	4.50*	4.09	2.94	3.81	2.81	2.34	2.88	2.13	2.19
Attractiveness
All	5.09	4.66	3.55	2.62	3.31	2.78	2.14	2.88	2.21	2.16
WPT-Only	4.88	5.38*	3.31	2.35	2.81	2.88	1.77	2.81	2.31	2.35
Dual/Poly	5.25	4.06*	3.75	2.84	3.72	2.69	2.44	2.94	2.13	2.00
Pleasant
All	5.03	4.78	3.59	2.59	3.31	2.81	2.05	3.00	2.16	2.03
WPT-Only	4.81	5.42*	3.19	2.19	2.73	2.77	1.65	2.81	2.35	2.23
Dual/Poly	5.22	4.25*	3.91	2.91	3.78	2.84	2.38	3.16	2.00	1.88
Interest
All	4.93	4.74	3.53	2.48	3.26	2.62	1.93	2.91	2.07	1.98
WPT-Only	4.65	5.38*	3.19	2.00	2.92	2.65	1.65	2.88	2.15	2.12
Dual/Poly	5.16	4.22*	3.81	2.88	3.53	2.59	2.16	2.94	2.00	1.88
Perceptions
General Harm
All	4.91	5.03	4.91	4.98	4.91	5.22	5.09	4.93	5.05	5.29
WPT-Only	4.85	4.85	4.92	5.12	4.92	5.19	5.19	4.96	4.92	5.54
Dual/Poly	4.97	5.19	4.91	4.88	4.91	5.25	5.00	4.91	5.16	5.09
Pregnancy Harm
All	6.17	6.21	6.12	6.14	6.12	6.36	6.26	6.19	6.17	6.36
WPT-Only	6.27	6.31	6.27	6.27	6.27	6.54	6.46	6.38	6.27	6.54
Dual/Poly	6.09	6.13	6.00	6.03	6.00	6.22	6.09	6.03	6.09	6.22
Fetal Harm
All	6.17	6.21	6.19	6.17	6.19	6.28	6.19	6.22	6.21	6.28
WPT-Only	6.35	6.42	6.35	6.31	6.31	6.46	6.31	6.38	6.38	6.46
Dual/Poly	6.03	6.03	6.06	6.06	6.09	6.13	6.09	6.09	6.06	6.13

Note. WPT=Waterpipe Tobacco. Use groups are based on WPT use regardless of flavor (n=26 WPT-only users; n=32 dual/poly WPT users). Preferences and perceptions were rated on a 1 to 7 scale with higher values indicating greater flavor preference or greater perceptions of the harmfulness. Bold font indicates statistically significant differences between WPT-only and dual/poly WPT users by Wilcoxon test.

^*P≤.05.

Regression analyses were used to determine the association between preferences, perceptions, and user type (0=WPT-only; 1=dual/poly WPT use) on total number of flavored WPT use events during preconception or pregnancy. Based on prior research on non-cigarette tobacco use,[58] age (continuous) and ethnicity (1 = Latina; 0 = non-Latina) were included as covariates. Number of menthol/mint and fruit-flavored WPT use events were skewed and clustered at zero, with the variance surrounding the means indicating a high degree of overdispersion; therefore, negative binomial regression analysis with robust standard errors was used. Preferences (liking, attractiveness, pleasant, interest) exhibited a high degree of multicollinearity for menthol/mint (variance inflation factor [VIF] ≥8.20) and fruit (VIF ≥3.79) and were averaged to create a mean preference score.

Logistic regression analyses were used to predict biomarkers (i.e., cotinine, benzene, butadiene) that were detectable in the sample (1=yes; 0=BLQ) from preferences, perceptions general harm, harm to the fetus), and total number of flavored WPT use events during preconception or pregnancy. (Age and ethnicity were included as covariates in these models.) Cotinine was selected as the primary proximate metabolite of nicotine intake;[59] benzene and butadiene were selected due to their association with WPT in particular and causal links with hematological malignancies.[24, 25, 60] Missing biomarker data was omitted from analyses: (N=1 for saliva cotinine and CO; N=27 for urine VOCs).

RESULTS

Sample Characteristics

Characteristics of the overall sample and stratified by WPT-only and dual/poly WPT users are provided in Table 1. The sample included 58 pregnant WPT users (M_age= 27; 19–39 years; 72% (n=42) non-White race; 53% (n=31) Latino ethnicity; 83% unmarried; 50% ≤high school education). The sample was not representative of pregnant women in Rhode Island (M_age= 30; 15–45 years; 21% non-White race; 26% Latino; 44% unmarried; 33% ≤high school education).[61] Dual/poly WPT users reported using cigarettes (n=8; 25% of dual/poly users), e-cigarettes (n=3; 9%), cigars (n=9; 28%), cigarettes + cigars (n=4; 13%), e-cigarettes + cigars (n=1, 3%), or cigarettes + e-cigarettes + cigars (n=7; 22%) during pregnancy or preconception. Dual/poly WPT users were less likely to identify as Latina ethnicity compared to WPT-only users (38% vs. 73% identified as Latina ethnicity, respectively). Dual/poly users were more likely to report marijuana use (P<.001) and secondhand cigarette smoke exposure (P=.030) versus WPT-only users.

Biomarkers of Nicotine Intake, Carbon Monoxide, and VOCs

Table 1 provides biomarkers of nicotine intake, CO concentration, and carcinogen exposure for the overall sample by user type. Levels of biomarkers were low, with more than >50% below the limit of quantification for cotinine, and mercapturic acid metabolites of benzene, butadiene, and acronitrile; 84% of CO values were below the cut-off for active smoking (≤ 3 ppm[62]). Dual/poly users showed increased levels of cotinine, acrylonitrile, acrolein, propylene-oxide, and acrylamide vs. WPT-only users (Ps ≤.040). Dual/poly users were also more likely to show positive CO readings [26% (n=8/31) vs 4% (n=½6)], χ²(1) = 5.13, P=.024.

Prevalence and Patterns of WPT Use in Preconception and First Trimester

Prevalence of WPT use was 81% (N=47) during pregnancy and 98% (N=57) during preconception. Participants reported using WPT an average of 8 times (SD = 13; range = 0 – 69) during first trimester and 15 times (SD = 28; range = 0 – 167) during preconception. On average, 26% used WPT weekly, 22% used < weekly but ≥ monthly, and 52% used < monthly. Sixty-eight percent of pregnant WPT users reported using menthol/mint flavors, 48% reported using fruit flavors, 33% reported using mint + fruit, and 5% reported using candy flavors. Figure 1 shows the proportion of flavors used across all WPT use events from the TLFB. Of the 1,376 WPT use events (904 preconception/472 first trimester), 68% included menthol/ flavored WPT, 14% fruit, 14% mint + fruit, 2% mint + coffee, and 1% other flavors (i.e., candy, fruit + candy, mint + chocolate, fruit + chocolate, coffee, alcohol, other beverages). Patterns were similar during preconception and 1^st trimester. Marijuana was mixed with WPT in 59 (4%) of events. WPT-only users reported more preconception (M = 24, SD = 8 vs. M = 9, SD = 16; P = .024) and 1^st trimester WPT use events (M=11, SD=17 vs. M=6, SD=8; P = .011) compared to dual/poly users. WPT-only users also reported more events using menthol/mint flavors compared to dual/poly users (M=30, SD=55 vs. M=5, SD=12; P<.001). There were no significant differences between WPT-only and dual/poly users for events with fruit flavors (M=3, SD=7 vs. M=4, SD=10; P=.366) or use of WPT mixed with marijuana (M=0.04, SD=0.20 vs. M=2, SD=9; P=.178).

Figure 1.

Proportion of flavors used across all WPT use events (N=1,376) during preconception and pregnancy. NOTE: Other flavors included: candy, fruit + candy, mint + chocolate, fruit + chocolate, coffee, alcohol, other beverages.

Preferences and Perceptions of Flavored WPT

Table 2 documents preference and perceptions for each WPT flavor for the overall sample and stratified by WPT-only and dual/poly users. Mean preferences were significantly different between flavors for each preference scale (liking, attractiveness, pleasant, interest), Ps <.001). Fruit and menthol/mint were most preferred, followed by candy and other sweets; tobacco, spice, and coffee were least preferred. Post-hoc tests indicated that fruit preferences were higher (Ps<.05) than all other flavors except menthol/mint; menthol/mint preferences were higher (Ps<.05) than all flavors except fruit and candy; preferences for candy were higher (Ps<.05) than coffee, spice, and tobacco flavors. Perceptions of harm to general health, pregnancy, and fetal health did not differ among flavors (Ps>.950). WPT-only users reported higher preferences (liking, attractiveness, pleasant, interest) for menthol/mint flavors (Ps <.05), while harm perceptions were uniformly elevated for WPT-only and dual/poly users for all three harm scales (general, pregnancy, fetal).

Preferences and Perceptions as Predictors of Flavored WPT Events

Regression models investigating associations among preferences, harm perceptions, and use of flavored WPT are presented in Table 3. Analyses were restricted to menthol/mint and fruit flavors because these flavors were used by 68% and 48% of participants in 68% and 14% of WPT events, respectively. Preferences for both menthol/mint and fruit-flavored WPT were associated with more flavored WPT use events across preconception and pregnancy (Ps ≤.007). WPT-only users and participants of Latina ethnicity reported greater menthol/mint flavored WPT use events compared to non-Latina participants and dual/poly users, respectively (Ps<.001).

Table 3.

Associations between flavor preferences, harm perceptions, and use of flavored WPT during preconception and pregnancy (N = 58).

	Menthol/Mint†		Fruit‡
Predictor	IRR (95% CI)	P	IRR (95% CI)	P
Age	0.98 (0.90, 1.07)	.723	1.08 (0.97, 1.20)	.152
Latina Ethnicitya	4.01 (1.88, 8.54)	<.001	0.74 (0.28, 1.97)	.546
Flavor Preferences	1.77 (1.39, 2.25)	<.001	1.96 (1.20, 3.19)	.007
Flavor Perceptions
General Harm	0.90 (0.67, 1.22)	.505	1.17 (0.85, 1.62)	.346
Fetal Harm	0.95 (0.73, 1.26)	.740	0.83 (0.58, 1.19)	.312
Dual/Poly WPT Useb	0.25 (0.12, 0.52)	<.001	1.09 (0.40, 2.93)	.872

Note. WPT=Waterpipe Tobacco. Negative binomial regression with robust variance estimation was used for all count variables (menthol/mint and fruit-flavored WPT events over preconception and 1^st trimester). Age, preferences, and perceptions (general harm, fetus) are continuous variables; ethnicity (1 = Latina; 0 = non-Latina) and use status (1=dual/poly WPT user; 0=WPT-only user) were dummy coded. IRR = incidence rate ratio; CI = confidence interval.

^aReference category is non-Latina (0).

^bReference category is WPT-only users (0).

^†Overdispersion parameter: α =1.84 (95% CI = 1.27 to 2.66)

^‡Overdispersion parameter: α =2.70 (95% CI = 1.57 to 4.64)

Preferences and Perceptions as Predictors of Biomarkers of Exposure

Regression models investigating associations among preferences, harm perceptions, and biomarkers are presented in Table 4. These models predicted detection of cotinine, benzene, and butadiene, given prior evidence for associations with WPT use.[24] Higher preference scores for menthol/mint flavors were associated with >2x increased likelihood of detection of cotinine or benzene (Ps≤.038); however, more menthol/mint flavored WPT use events and higher perceptions of fetal harm from menthol/mint flavored WPT were associated with decreased likelihood of cotinine and benzene detection (Ps≤.038). Finally, each additional fruit-flavored WPT use event was associated with a nearly 2-fold increased likelihood of butadiene detection (P=.038). No other statistically significant associations between flavored WPT use events or harm perceptions and biomarkers were found (Ps>.05). Latina ethnicity was associated with increased detection of butadiene, but decreased cotinine detection (Ps ≤.043).

Table 4.

Associations between flavor preferences, perceptions, use, and biomarkers of use during preconception and pregnancy (Ns=31–57)

	Cotinine (N=57)				Benzene (N=31)				Butadiene (N=31)
Specific Flavor	Menthol/Mint		Fruit		Menthol/Mint		Fruit		Menthol/Mint		Fruit
Predictor	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P
Age	1.11 (0.96, 1.29)	.147	1.05 (0.92, 1.20)	.427	1.39 (0.76, 2.56)	.286	1.40 (0.89, 2.18)	.142	1.00 (0.82, 1.21)	.960	1.08 (0.92, 1.27)	.355
Latina ethnicity	0.07 (0.01, 0.59)	.014	0.17 (0.03, 0.95)	.043	0.62 (0.07, 5.28)	.664	1.55 (0.17, 13.94)	.696	1.67 (0.16, 17.11)	.665	6.43 (1.14, 36.31)	.035
Flavor Preferences	2.12 (1.10, 4.09)	.025	0.92 (0.61, 1.40)	.740	1.91 (1.04, 3.53)	.038	0.82 (0.48, 1.41)	.470	1.53 (0.63, 3.70)	.350	1.13 (0.77, 1.66)	.539
Flavor Perceptions
General Harm	0.72 (0.36, 1.45)	.360	0.96 (0.55, 1.70)	.897	1.36 (0.78, 2.35)	.279	1.14 (0.68, 1.95)	.632	0.97 (0.51, 1.84)	.921	0.66 (0.28, 1.55)	.338
Fetal Harm	0.40 (0.17, 0.95)	.038	0.74 (0.43, 1.29)	.291	0.32 (0.10, 0.98)	.045	0.50 (0.24, 1.07)	.075	0.67 (0.33,1.39)	.285	0.77 (0.42,1.40)	.384
Flavored WPT use events	0.80 (0.69, 0.92)	.002	1.04 (0.99, 1.10)	.131	0.98 (0.93, 1.04)	.517	1.07 (0.92, 1.23)	.383	1.01 (0.95, 1.07)	.836	1.64 (1.03, 2.62)	.038

Note. WPT=Waterpipe Tobacco. Logistic regression with robust variance estimation was used for all analyses. Age, preferences, perceptions, and number of WPT use events in which menthol/mint OR fruit flavors were used are continuous variables; ethnicity (1 = Latina; 0 = non-Latina) was dummy coded. Specific flavors (menthol/mint and fruit) indicate the flavor of focus for each regression analysis (i.e., perceptions and perceptions of menthol/mint [fruit] flavored WPT were included in the models predicting menthol/mint [fruit] flavored WPT use events). OR = odds ratio; CI, confidence interval.

^aReference category is non-Latina (0).

DISCUSSION

We investigated use, preference, and perceptions of WPT flavorings among pregnant women. Pregnant WPT users reported nearly exclusive use of flavored tobacco; the majority of preconception/prenatal WPT use episodes involved menthol/mint flavored WPT, followed by fruit flavors. Preferences were highest for fruit, menthol/mint, and sweet-flavored WPT. Harm perceptions were high, but did not differ by flavor. Preferences for fruit and menthol/mint flavors predicted preconception/prenatal use of flavored WPT, whereas harm perceptions did not. Compared to dual/poly WPT use, WPT-only users used more WPT, were more likely to use menthol/mint flavors, were less likely to use fruit flavors, and had lower rates of detectable WPT/tobacco smoke exposure biomarkers.

High rates of use of mint and sweet flavors support our hypotheses and complement studies of WPT use and other tobacco product (i.e. e-cigarette) use in non-pregnant populations.[58, 63] However, in the present study, menthol/mint was used more frequently than fruit or other flavors—particularly among WPT-only users, whereas most prior studies with non-pregnant samples have shown greatest use of fruit flavors, with menthol/mint used less frequently. For example, in a sample of ~75,000 US adults, order of prevalence of WPT flavors use was: fruit (74%), menthol/mint (19%), and candy and other sweets (17%).[58] In prior studies of both pregnant and non-pregnant e-cigarette users, however, rates of use of fruit and menthol/mint flavored e-cigarettes were similar [42, 58, 64–66].

Intriguingly, although pregnant women reported greater use of menthol/mint than fruit flavors, they endorsed similar preferences for both fruit and menthol/mint, followed by candy and other sweet flavors, which were preferred over coffee, spice, and unflavored WPT. Further, we found that greater preferences for both fruit and menthol/mint flavors predicted flavored WPT use during preconception/pregnancy. Results complement findings from discrete-choice experiments, in which participants—especially women—were more likely to choose various fruit- over unflavored WPT. [67, 68] Results also complement findings from studies of e-cigarette and other tobacco product flavor preferences in adults, young adults, youth and pregnant women. Sweet flavors were preferred among youth and young adult nonsmokers and were associated with use and intentions to use, whereas menthol/mint was preferred among adults [64–66]. In a study of e-cigarette flavor preferences among pregnant women, fruit and candy-flavored e-cigarettes were preferred over menthol/mint flavors [42].

Although preferences and use episodes differed by flavor, harm perceptions for general, pregnancy, and fetal health did not. Perceptions of pregnancy and fetal harm were uniformly high (average of >6.1 on a 7-point scale) across all flavors (although fetal harm perceptions showed some evidence of variability despite high averages). Further, harm perceptions showed no associations with episodes of (flavored) preconception/prenatal WPT use. Prior WPT studies have shown decreased harm perceptions for (flavored) WPT vs. cigarettes, but have not investigated harm perceptions across WPT flavors.[36, 64] However, prior studies of e-cigarette flavorings revealed decreased harm perceptions for fruit and sweet-flavors vs. tobacco flavors.[66] Results from this study extend prior research by our group using similar methods to the present study—one in pregnant women focused on e-cigarettes and one in reproductive-age women focused on WPT. In both studies, harm perceptions did not differ by flavor and were not associated with e-cigarette/WPT use.[42, 43] Lack of differences in harm perceptions across flavors may relate to our focus on pregnant women and widespread public health and education campaigns regarding the harmful effects of tobacco use during pregnancy.

To our knowledge, the present study is the first to measure biomarkers of WPT/tobacco smoke exposure in pregnant WPT users. Although overall levels of exposure biomarkers were low, preference for menthol/mint flavored WPT was associated with double the likelihood of detection of cotinine or benzene. Further, use of fruit flavors was associated with increased likelihood of butadiene detection, while use of menthol/mint flavors was associated with decreased cotinine detection. In addition, consistent with prior literature, dual/poly WPT users showed higher levels of cotinine and certain VOCs versus WPT-only users. Results should be interpreted with caution given the smaller sample size for biomarker analyses and because biomarkers were assessed at the time of the interview, not in relation to recent episodes of WPT use. Given the sporadic nature of WPT use for most participants and the relatively short elimination half-life of the biomarkers measured (hours), the present study likely underestimates maternal/fetal exposure to toxic chemicals.

Regulatory Implications

As regulatory policies regarding WPT are formulated in the US and other countries,[38, 69] it may be relevant to consider pregnant women as a uniquely vulnerable population. Labeling requirements for WPT and devices, and warnings in WPT establishments advising pregnant women not to use WPT are warranted immediately.[20, 21, 40, 70] Education for prenatal providers regarding how to screen and counsel women who use WPT during pregnancy is also needed.

Our study highlights the importance of comprehensive flavored tobacco product regulation. To date, in the US, only a handful of municipalities in California have banned the sale of all flavors in all tobacco products,[71, 72] and no states have done so, although the California legislature is currently considering a proposal to end flavored tobacco sales statewide (Senate Bill 38) [73]. Outside the US, Canada has taken the lead in regulating flavored tobacco products, with five provinces/territories (Ontario, Quebec, New Brunswick, Prince Edward Island, and Newfoundland and Labrador) adopting flavored tobacco bans that include WPT [74]. Continued research on WPT use, preferences, and biomarkers during pregnancy and in reproductive age women may inform the development of regulatory policies regarding WPT flavors.[75] Research on the unique impact of flavorings, WPT, and flavored WPT on human fetal development, which is urgently needed,[76] may also play a role in influencing the regulatory approach to the continued sale of flavored WPT.

Limitations

Several limitations should be considered when interpreting our findings. First, we analyzed baseline data from our first cohort in an ongoing longitudinal study. Although we used TLFB methods to cue recall, all data were collected from a single interview and thus cannot support causal inferences. Second, our study sampled a small (N=58) convenience sample of low-income, racially/ethnically diverse pregnant women from southern New England (Rhode Island and Massachusetts). Selection bias is thus possible, and results may not generalize to the local population or other US regions or countries. Further, including only English-speaking women may have limited the generalizability of findings related to Latina ethnicity. Inclusion of women who were recruited by both active (approached in clinic) and passive (advertisements) means may also limit the generalizability of the sample. Nationally, internationally, or locally-representative studies with larger samples are needed to better elucidate WPT/flavors use and preferences in pregnancy. Finally, we assessed use and preference of multiple broad categories of WPT flavors (e.g., fruit) but did not assess specific flavors (e.g., apple). Future research should investigate specific flavors that may be appealing to pregnant mothers.

Conclusions

Given increasing use of WPT by pregnant and reproductive age women and known maternal and fetal toxicity of tobacco, it is critical to understand characteristics of WPT that are associated with maternal use and are amenable to regulation. In this study, pregnant WPT users reported nearly exclusive use of flavored WPT, with menthol/mint and fruit-flavored WPT used most frequently and most preferred. Further, preferences for fruit and menthol/mint flavors were associated with greater WPT use and exposure biomarkers despite high levels of harm perception for all flavors. Results from this uniquely vulnerable population highlight the need to implement warning labels for pregnant women on WPT, devices, and WPT establishments, and regulate sweet and mint/menthol WPT flavors.

WHAT THIS PAPER ADDS.

What is already known on this topic?

• Waterpipe (hookah) tobacco use is a growing public health problem among pregnant and reproductive-age women.

• Given known impacts of prenatal tobacco use on maternal and infant morbidity, pregnant WPT users represent a uniquely vulnerable population.

• Sweet flavors contribute to the appeal of waterpipe tobacco (WPT) and are a potential regulatory target.

What important gaps in knowledge exist on this topic?

• Little research has addressed WPT use in pregnancy.

• Little is known regarding characteristics of WPT use by pregnant women that are amenable to regulatory efforts to protect the health of women and children.

What this paper adds?

• Pregnant WPT users reported greatest use of and preferences for menthol/mint and fruit flavors.

• Preferences for menthol/mint and fruit flavored WPT predicted flavored WPT use and biomarkers of exposure.

• Findings in this vulnerable population highlight the need to regulate both sweet and mint/menthol flavors in WPT.