Sex Differences in the Association of E-cigarette and Cigarette Use and Dual Use With Self-Reported Hypertension Incidence in US Adults

Abstract

Background

While there is some evidence and conceptual plausibility that tobacco product use is associated with hypertension incidence and that this association varies by sex, extant longitudinal research had been conducted prior to the emergence of e-cigarette and dual e-cigarette and cigarette use.

Aims and Methods

Data were analyzed from the US Population Assessment of Tobacco and Health study for adults with no lifetime history of hypertension at wave 1 (2013–2014) who completed waves 2–4 follow-up surveys (2014–2018; n = 16 434). Sex-stratified weighted covariate-adjusted multivariable Cox regression models were used to examine the association between established current e-cigarette or cigarette exclusive or dual-use (as a time-varying and time-lagged regressor) and subsequent self-reported hypertension onset.

Results

Weighted cumulative hypertension incidence by wave 4 varied by waves 1–3 e-cigarette, cigarette, and dual use status in females (nonuse [incidence: 9.9%], exclusive e-cigarette use [11.8%], exclusive cigarette use [14.8%], dual-use [12.4%]; p = .003 for omnibus differences among all groups) but not males (nonuse [12.6%], exclusive e-cigarette use [9.7%], exclusive cigarette use [13.7%], dual-use [9.3%]; p = .231). Among females, exclusive cigarette (vs. no) use (hazard ratio: 1.69, 95%CI 1.21 to 2.34; p = .002), but not exclusive e-cigarette or dual-use, was significantly associated with subsequent hypertension. Dose–response models were suggestive that consistent exclusive e-cigarette or dual-use versus nonuse across multiple may be associated with hypertension among females, but results were nonsignificant.

Conclusions

The association of e-cigarette, cigarette, and dual use with hypertension may differ by sex, whereby exclusive cigarette use could be a prospective risk factor for subsequent self-reported hypertension in US adult females.

Implications

This nationally representative cohort study provides the very first evidence of whether there are prospective associations of established e-cigarette and cigarette use and dual use with future hypertension onset among US adult females and males. We found that exclusive cigarette smoking was associated with an increased risk of incident hypertension among females, but not males. We observed a trend of a dose–response relationship between e-cigarette use and risk of incident hypertension among female exclusive e-cigarette users or dual e-cigarette and cigarette users. Our study will contribute to understanding the chronic health risks of vaping to prevent the potential long-term e-cigarette use-related health burden.

Introduction

High blood pressure and tobacco smoking are among the top two leading risk factors contributing to preventable mortality worldwide¹ in part because they independently and synergistically increase the risk of cardiovascular diseases (CVD).² Additionally, there is a complex association between smoking and BP, with limited evidence that smoking may increase risk of incident hypertension.^3,4 However, the association of tobacco product use with hypertension incidence is unclear.

First, virtually all of the epidemiologic evidence on the smoking-hypertension association was conducted prior to recent changes in the tobacco product landscape. The prevalence of current cigarette smoking among US adults declined from 20.9% in 2005 to 13.7% in 2018,⁵ but the prevalence of use of e-cigarettes, which did not arrive in US markets until 2007, was 3.2% in 2018.⁶ Dual use of cigarettes and e-cigarettes is one of the most prevalent combinations among multiple tobacco product users in the United States,⁷ raising new questions about the association between e-cigarette and cigarette use and dual use status and risk of developing hypertension. The toxicants in e-cigarette aerosol, such as nicotine, carbonyls, ultrafine particulates, and heavy metals, are known to be associated with BP elevation.⁸ Nicotine inhalation was found to alter both pulmonary and systemic BP by enhanced activation of mitogen-activated protein kinase pathways and renin-angiotensin-aldosterone system in mice.⁹ Recent clinical trials showed that exposure to e-cigarette’s aerosol for several minutes acutely increases both systolic blood pressure (SBP) and diastolic blood pressure (DBP).¹⁰ A prior cross-sectional study provided initial epidemiological evidence showing the use of e-cigarettes fairly regularly (i.e. “established use”) was associated with an increased odds of self-reported hypertension in adults.¹¹ Whether e-cigarette use (either alone or in dual use with cigarettes) is associated with hypertension over longer periods of follow-up is unknown.¹²

Second, whether there are sex differences in the association between e-cigarette and cigarette use and dual use with hypertension is unknown. Studies have shown that females metabolize nicotine more rapidly than males because of the effect of estrogen on the activity of enzyme CYP2A6.¹³ In addition, estrogen engages several mechanisms (e.g. sympathetic nervous system and angiotensin) that protect against hypertension.¹⁴ While evidence indicates that the role of tobacco product use in hypertension risk could differ by sex, this hypothesis has not been previously investigated.

This cohort study examined sex differences in the association between e-cigarette, cigarette, and dual use and subsequent onset of self-reported diagnosed hypertension among US adults over 4 waves of the nationally representative Population Assessment of Tobacco and Health (PATH) survey study. To our knowledge, this is the first study to estimate the longitudinal association between product use and hypertension onset across established e-cigarette, cigarette, and dual users.

Methods

Study Population

The PATH study, a US nationally representative population-based study, is a collaboration between the National Institute on Drug Abuse (NIDA), National Institutes of Health (NIH), and the Center for Tobacco Products (CTP), Food and Drug Administration (FDA). The study sampled over 150 000 mailing addresses across the United States to create a national sample of tobacco users and nonusers. The nationally representative cohort was assembled by a four-stage stratified area probability sampling method. Longitudinal weights were introduced to compensate for different probabilities of selection, nonresponse, possible deficiencies in the sampling frame, and attrition. The national survey data were collected using the Audio Computer-Assisted Self-Interviewing, Computer-Assisted Personal Interviewing administered questionnaires, and various paper data collection forms. Detailed information regarding study design and methodology have been published.¹⁵

Baseline data (wave 1) collection occurred between September 2013 and December 2014, resulting in a baseline cohort containing 32 320 adult respondents (age ≥18 years). Sequential annual waves of data were collected from October 2014 to October 2015 (wave 2), from October 2015 to October 2016 (wave 3), and from December 2016 to January 2018 (wave 4).

The study period for the current report was from wave 1 (baseline) through wave 4. Of the 32 320 baseline participants, 7127 were excluded because they self-reported ever having diagnosed hypertension at baseline. Among the remaining baseline sample, 8759 were excluded because they lacked follow-up data at wave 2, 3, or 4. The primary analytic sample (cohort 1) included 16 434 participants who were hypertension-free at baseline and completed interviews for all follow-up waves 2–4 (Figure 1). Within the study period, participants were either identified as hypertension incident cases or right censored at Wave 4.

Figure 1.

Study flowchart. *Primary analyses were conducted by fitting weighted models using cohort 1 with longitudinal weights for baseline respondents who completed an interview for all follow-up waves (waves 2, 3, and 4). An alternative cohort (cohort 2) for sensitivity analyses was generated in the midway with interval censored data. Details were presented in Supplementary Figure S3.

Measures

Sex Assessment

Sex was assessed based on the self-reported answer to the question “What is your sex?” at baseline for all respondents. Participants who responded affirmatively “male” or “female” were identified correspondingly. Those who responded “don’t know” or “refused” were classified as missing values.

E-cigarette, Cigarette, and Dual Use and Other Tobacco Use Exposure Variables

At each wave for waves 1–3, current established e-cigarette users were defined following previous research¹⁶ as reporting ever using e-cigarettes at least “fairly regularly” and currently use them “every day” or “some days”. Current established cigarette use was defined as smoking at least 100 cigarettes in their lifetime and currently smoking every day or some days at each wave for waves 1–3. We then derived a composite time-varying variable—established current e-cigarette, cigarette, and dual use—with four categories: (1) e-cigarette users, (2) cigarette users, (3) dual users, and (4) nonusers of e-cigarettes or cigarettes.

Additionally, we combined the established current users of other tobacco products (i.e. traditional cigar, hookah, cigarillo, filtered cigar, cigar, blunt, snus, pipe, smokeless tobacco, or dissolvable tobacco) as a single covariate “other tobacco product use” as a time-varying binary covariate (yes/no) for waves 1–3. A time-invariant covariate “lifetime cigarettes exposure (pack-year)” was created to measure the amount a person has smoked over a long period of time. It was calculated by multiplying the number of packs of cigarettes ever smoked per day by the number of years the person has smoked (pack-year range: 0–100 for 99.7% of participants) based on the related data at baseline.

Self-Reported Diagnosed Hypertension

At wave 1, respondents were asked “Has a doctor, nurse, or other health professionals ever told you that you had high blood pressure?”, with those responding affirmatively classified as baseline hypertension and were excluded from the sample. At waves 2–4, they were asked “In the past 12 months, has a doctor, nurse, or other health professional told you that you had high blood pressure?”, which was used to classify a time-varying binary outcome variable (yes or no) at each follow-up wave. A total of 83 participants were found to have inconsistent responses during follow-ups (i.e. reporting never having hypertension at wave 4 but reporting being diagnosed as hypertension at wave 2 or 3). These participants were treated as “not having hypertension” in primary analyses. Alternative methods of handling inconsistent responses were conducted in sensitivity analyses (details were described in the sensitivity analyses section of Supplementary Methods).

Covariates

Factors previously associated with hypertension onset were included as priori covariates.¹¹ In addition to the aforementioned tobacco exposure covariates, variables adjusted for in multivariable analyses included age, race, ethnicity, education level, body mass index (BMI), physical activity (PA), heavy alcohol use, disease-related covariates (i.e. CVD, hypercholesterolemia, and diabetes mellitus), and family history of hypertension. All adjusted covariates were employed as categorical variables in a time-varying or time-invariant manner correspondent with their assessment frequency across waves (See Supplementary Table S1 for information on time-varying and time-invariant covariates; See covariates section of Supplementary Methods for detailed information on defining each covariate).

Some potential confounders (i.e. secondhand smoke exposure, substance use, pregnancy history, marital status, income, poverty level, and insurance type) were examined but not included in multivariable analyses because of the high missing data rate or lack of significance in the backward elimination model selection procedure (see results in Supplementary Table S2).

Statistical Analyses

We used weighted frequency distributions and the Rao-Scott modified likelihood ratio test to examine the difference among groups. Interactions between sex and e-cigarette, cigarette, and dual use were examined to determine effect modification by sex. Sex-stratified weighted Cox regression models (model 1) were then used to test e-cigarette, cigarette, and dual use at waves 1–3 as having a time-varying and time-lagged association with self-reported hypertension at waves 2–4 among baseline lifetime hypertension-free participants.^17,18 This approach incorporated all waves of data on e-cigarette, cigarette, and dual use occurring before the onset of hypertension for every participant. Follow-up hypertension onset during waves 2–4 were regressed on wave 1 e-cigarette, cigarette, and dual use, follow-up hypertension onset during waves 3–4 were regressed on wave 2 e-cigarette, cigarette, and dual use, and so on. When a hypertension case occurred, e-cigarette, cigarette, and dual use at that wave and ensuing waves were not additionally incorporated into the model estimates. We adjusted for the eight time-varying and four time-invariant covariates in the multivariable models described above (see variables in Supplementary Table S3 and covariates information in Supplementary Table S1).

In previous studies, a dose–response relationship was reported between cigarette smoking and risk of cardiovascular event.^19,20 To further explore dose–response relationship of e-cigarette, cigarette, and dual use with hypertension, sex-stratified weighted Cox regression models with time-varying, time-lagged, and time-cumulative regressors (model 2) were conducted to examine the cumulative effects of e-cigarette, cigarette, and dual use across multiple waves on hypertension onset. This approach is derived from the aforementioned Cox model (model 1), with the same regression procedure but only one modification on the regressor.¹⁷ Instead of regressing on whether the participant was a current established e-cigarette, cigarette, and dual user in a single preceding wave (model 1), model 2 regressed on the graded consistency of e-cigarette, cigarette, and dual use in all available preceding waves in a dose–response manner. The cumulative proportion of waves of established e-cigarette, cigarette, and dual use were calculated in all preceding waves. If the participant consistently reported established e-cigarette, cigarette, and dual use in all preceding waves, the regressor was defined as “consistent use” with the highest cumulative dose grade (i.e. 100% of the waves preceding hypertension onset involved use). If the participant reported established e-cigarette, cigarette, and dual use in some (but not all) preceding waves, the regressor was defined as “intermittent use” with the intermediate cumulative dose grade (i.e. >0% and <100% of the waves preceding hypertension onset involved use). Finally, if the participant reported no e-cigarette, cigarette, and dual use in all preceding waves, the regressor was defined as “no use” with the lowest cumulative dose grade (i.e. 0% of the waves preceding hypertension onset involved use). The multivariable models were fitted and adjusted for all covariates listed previously. Potential interactions between sex and e-cigarette, cigarette, and dual use were tested.

All weighted Cox regression models were conducted using SAS v9.4 (SAS Institute Inc., Cary, NC). Reverse Kaplan–Meier plots were produced using survival²¹ and survminer²² packages in R v4.0.3²³ based on the unadjusted crude models. We used the Taylor series linearization method with variance strata and cluster identifiers for complex sample variance estimation. The Efron method was used for handling ties. Hazard ratios (HRs) and 95% confidence intervals (95%CIs) were calculated to measure associations of e-cigarette, cigarette, and dual use with subsequent self-reported hypertension. Missing data were handled as listwise deletions due to the low missing rate (0%–3.6%) (Supplementary Table S1 reports missing data for each variable). Alternative methods of handling missing data (e.g. weighted hot-deck imputation) were used to conduct sensitivity analyses (details were described in sensitivity analyses section of Supplementary Methods). Statistical significance was set at 5% for two-sided tests. Holm–Sidak methods were used for controlling multiple testing error rates (details were described in multiple comparisons section of Supplementary Methods).

Results

Sample Characteristics

The analytic cohort (16 434 adults; 8792 females [weighted %: 53.4; 95%CI 52.2 to 54.5]) was sociodemographically diverse (Supplementary Table S3). One thousand and seven hundred fifty-seven (weighted %: 11.5; 95%CI 10.7 to 12.4) individuals self-reported hypertension onset during the four-waves (September 2013 to January 2018) follow-up. Females were less likely to report hypertension onset or being diagnosed with any CVD, and less likely to be overweight (25.0 ≤BMI ≤29.9 kg/m²), tobacco products users, less educated (≤high school), or have long lifetime cigarettes exposure (≥7.5 pack-years). Conversely, females were more likely to report having a family history of hypertension, being diagnosed with diabetes mellitus, and being less physically active (<150 min MVPA/week). Comparisons grouped by participants who self-reported hypertension onset or not across follow-up waves can be found in Supplementary Table S4.

Descriptive Analyses of Tobacco Use Exposure

The mean per-wave weighted prevalence across waves 1–3 was 1.3% (range: 1.0%–1.5%) for exclusive e-cigarette use, 16.9% (range: 16.8%–19.4%) for exclusive cigarette use, 1.9% (range: 1.6%–2.1%) for dual use, and 79.9% (range: 79.5%–80.6%) for nonuse (see wave specific prevalence table in Supplementary Figure S1). The e-cigarette, cigarette, and dual use prevalence changed by waves (Supplementary Figure S1). Most baseline exclusive e-cigarette users, exclusive cigarette users, and nonusers did not change their use status through wave 3. However, the majority (weighted %: 46.3%) of baseline dual users became exclusive cigarette users at wave 3 (Supplementary Figure S2).

Association Between E-cigarette, Cigarette, and Dual Use and Subsequent Self-Reported Hypertension

The estimated unadjusted cumulative probabilities of subsequent hypertension onset by wave 4 were shown with hazard curves in Figure 2. The cumulative incidence of self-reported hypertension by wave 4 significantly varied by waves 1–3 e-cigarette, cigarette, and dual use in females (nonuse [incidence: 9.9%], exclusive e-cigarette use [11.8%], exclusive cigarette use [14.8%], dual-use [12.4%], p = .003 for the test of omnibus differences among groups) but not males (nonuse [incidence: 12.6%], exclusive e-cigarette use [9.7%], exclusive cigarette use [13.7%], dual-use [9.3%], p = .231 for the difference among groups). The effect modification analyses showed significant interaction between the female sex and e-cigarette, cigarette, and dual use (see interaction effect estimates in Supplementary Table S5).

Figure 2.

Estimated hazard curves for self-reported diagnosed hypertension onset by e-cigarette, cigarette, and dual use statues in preceding waves. The reverse Kaplan–Meier plots with hazard curves shows the estimated weighted unadjusted cumulative probabilities of subsequent hypertension onset from wave 2 through wave 4. The horizontal axis depicts 4 annual waves assessment. The vertical axis depicts the estimated weighted unadjusted cumulative incidences of self-reporting hypertension at each follow-up wave; the estimates were reported by dual use, exclusive e-cigarette use, exclusive cigarette use and non-use at the preceding waves. For dual users, exclusive e-cigarette users, exclusive cigarette users, and nonusers (as reference) in the preceding waves, the cumulative estimated incidences of self-reporting hypertension at the final assessment (wave 4) were 9.3%, 9.7%, 13.7%, and 12.6% (reference) among males (A), and were 12.4%, 11.8%, 14.8%, and 9.9% (reference) among females (B).

In the sex-stratified weighted multivariable models (model 1, see Table 1), the association of exclusive cigarette (vs. no) use in waves 1–3 with subsequent hypertension onset during waves 2–4 among females was significant (hazard ratio [HR] 1.65, 95%CI 1.18 to 2.32). The HRs of subsequent hypertension onset among female exclusive e-cigarette users (vs. nonusers; HR 1.47, 95%CI 0.90 to 2.41) and dual users (vs. nonusers; HR 1.41, 95%CI 0.91 to 2.21) were nonsignificant. For males, neither exclusive e-cigarette, exclusive cigarette, nor dual use (vs. nonuse) were associated with subsequent hypertension onset in multivariable models. Covariates such as age and family history of hypertension were associated with hypertension onset in the multivariable models (see results of covariates in Supplementary Tables S6 and S7).

Table 1.

Associations of Established Current E-cigarette, Cigarette, and Dual Use and Subsequent Self-Reported Hypertension Onset (Model 1)^a

Time-varying regressors,	Associations with subsequent hypertension development, waves 2–4
E-cigarette, cigarette, and dual use b	Male (n = 7642)			Female (n = 8792)
Waves 1–3	Hazard ratio (95%CI)	P-value	Padjustedc	Hazard ratio (95%CI)	P-value	Padjustedc
Exclusive e-cigarette (vs. no) use	1.06 (0.58, 1.94)	.86	1.00	1.54 (0.96, 2.48)	.11	.69
Exclusive cigarette (vs. no) use	1.12 (0.84, 1.49)	.44	1.00	1.69 (1.21, 2.34)	.002	.02
Dual (vs. no) use	1.02 (0.65, 1.59)	.93	1.00	1.56 (1.04, 2.34)	.04	.34

Sex-stratified weighted multivariable Cox regression hazards model that included time-varying e-cigarette, cigarette, and dual use regressor and adjusted for eight time-varying covariates (i.e. education level, other tobacco products use, body mass index, physical activity, heavy drinker, diabetes mellitus, hypercholesterolemia, and cardiovascular diseases) and time-invariant covariates (i.e. baseline age, race/ethnicity, lifelong cigarette smoking, and family history of hypertension at wave 4). All adjusted covariates were employed as categorical variables. Results of covariates are presented in Supplementary Tables S5 and S6.

E-cigarette, cigarette, and dual use regressor is a time-varying and time-lagged variable. E-cigarette, cigarette, and dual use status at waves 1–3 were used to examine the association with self-reported hypertension at waves 2–4 among baseline lifetime hypertension-free participants (n = 16 434). At wave 1, N (no use) = 10 606, N (e-cigarette use) = 300, N (cigarette use) = 4977, and N (dual use) = 488. The numbers dynamically changed across the subsequent waves. The wave-specific e-cigarette, cigarette, and dual use prevalence and the dynamic changes across waves were shown in Supplementary Figures S1 and S2.

P_adjusted shows the statistical significance after the Holm–Sidak corrections for multiple testing in Kaplan–Meier plots, models 1 and 2. Details were described in multiple comparisons section of Supplementary Methods.

Results of model 2 estimated possible dose–response effect of consistency of use across waves 1–3 in multivariable models (see Table 2). Compared to cases of consistent nonuse of either cigarettes or e-cigarettes across waves 1–3, exclusive cigarette use on an intermittent basis (i.e. use on some but not all exposure waves) or on a consistent basis (i.e. use on all exposure waves) were significantly associated increased likelihood of subsequent hypertension onset by wave 4. Exclusive e-cigarette use or dual use on intermittent and consistent bases showed a dose-response relationship with subsequent hypertension incidence in females: The estimates decreased for intermittent use, and increased for consistent use, compared with the corresponding overall estimates in model 1. However, the associations were not significant. For males, neither intermittent nor consistent exclusive or dual use of e-cigarettes or cigarettes were significantly associated with subsequent hypertension incidence. The effect modification analyses showed significant interactions only between sex and consistent cigarette use (see interaction effect estimates in Supplementary Table S5).

Table 2.

Associations of E-cigarette, Cigarette, and Dual Use Consistency and Subsequent Self-Reported Hypertension Onset (Dose–Response Effect, Model 2)^a

Time-varying regressors,	Associations with subsequent hypertension development, waves 2–4
E-cigarette, cigarette, and dual use consistency b	Male (n = 7642)			Female (n = 8792)
from wave 1 through 3	Hazard ratio (95%CI)	P-value	Padjustedc	Hazard ratio (95%CI)	P-value	Padjustedc
Exclusive e-cigarette use
Intermittent (vs. no) use	0.89 (0.48, 1.64)	.70	1.00	0.89 (0.43, 1.83)	.74	1.00
Consistent (vs. no) use	1.17 (0.56, 2.46)	.67	1.00	1.84 (0.96, 3.52)	.07	.52
Exclusive cigarette use
Intermittent (vs. no) use	1.05 (0.70, 1.59)	.81	1.00	2.07 (1.09, 3.94)	.03	.26
Consistent (vs. no) use	1.10 (0.77, 1.58)	.59	1.00	1.69 (1.21, 2.36)	.002	.02
Dual use
Intermittent (vs. no) use	0.71 (0.40, 1.26)	.24	.94	0.86 (0.44, 1.68)	.66	1.00
Consistent (vs. no) use	1.18 (0.65, 2.14)	.59	1.00	1.71 (1.00, 2.93)	.05	.40

Model 2 regressed on the graded consistency of e-cigarette, cigarette, and dual use in all available exposure waves in a dose-response manner. Sex-stratified weighted multivariable Cox regression hazards model that included three simultaneous time-varying E-cigarette, cigarette, and dual use consistency regressors and adjusted for eight time-varying covariates (i.e. education level, other tobacco products use, body mass index, physical activity, heavy drinker, diabetes mellitus, hypercholesterolemia, and cardiovascular diseases) and time-invariant covariates (i.e. baseline age, race/ethnicity, lifelong cigarette smoking, and family history of hypertension at wave 4). All adjusted covariates were employed as categorical variables.

Three use consistency regressors (i.e. exclusive e-cigarette, exclusive cigarette, and dual use) are time-varying, time-lagged, and time-cumulative variables. Use consistency status at waves 1–3 were used to examine the association with self-reported hypertension at waves 2–4 among baseline lifetime hypertension-free participants (n = 16 434). Intermittent use = established e-cigarette, cigarette, and dual use was reported in at least one (but not all) preceding wave; consistent use = established e-cigarette, cigarette, and dual use was reported in all preceding waves. When model 2 regressed on wave 1 data, there was no intermittent use but only consistent use status because the total number of preceding waves was one.

P_adjusted shows the statistical significance after the Holm–Sidak corrections for multiple testing in Kaplan–Meier plots, models 1 and 2. Details were described in multiple comparisons section of Supplementary Methods.

Sensitivity Analyses

Sensitivity analyses found that the primary results were not substantially changed if alternative approaches to handling missing data (Supplementary Table S8) were applied or if the results were retested treating inconsistent responses as incident cases or missing values in the analytic models (Supplementary Table S9). The sensitivity analyses using the alternative cohort (cohort 2 including 1090 interval censored data shown in Supplementary Figure S3; see Supplementary Supplementary Table S10 for information on how cohort 2 differed from the analytic sample on several characteristics) did produce slightly more conservative yet similarly significant estimates (Supplementary Table S11) to primary analyses.

Discussion

This nationally representative cohort study provides the first test of whether there are prospective associations of established e-cigarette and cigarette use and dual use with future hypertension onset among US adult females and males. Results found that exclusive cigarette smoking, but not exclusive e-cigarette use or dual e-cigarette and cigarette use, was associated with an increased risk of incident hypertension among females. Neither e-cigarette nor cigarette use when used exclusively or in tandem as part of a dual-use pattern were associated with subsequent hypertension in males.

Several prior small trials focusing on the acute effect (minutes to hours) of e-cigarette use on BP showed inconsistent results.^10,24 Combining the results from these studies, a recent meta-analysis concluded that there are significant acute increases in BP after e-cigarette use,²⁴ though the pooled average acute elevation was only 2.0 mmHg for both SBP and DBP. Although the chronic effect of e-cigarette use on BP is potentially supported by some biological mechanism studies,^25–27 the direct epidemiologic evidence from population study had been lacking prior to the current investigation. Using the PATH wave 3 data, our previous cross-sectional study first reported that a higher odds of hypertension prevalence was associated with each established e-cigarette, cigarette, and dual use status (i.e. exclusive e-cigarette, exclusive cigarette, or dual-use) versus nonuse.¹¹ In the current study, we found no significant prospective association between exclusive e-cigarette or dual e-cigarette and cigarette use with subsequent hypertension onset after statistical corrections to control type-I error due to multiple tests. The relatively short period of exposure for comparisons and the relatively small sample sizes of exclusive e-cigarette or dual users, which are two of the limitations of this study, precluded inferences regarding whether these are true null associations or perhaps type II errors due to insufficient statistical power. This issue is particularly relevant in interpreting the relations for exclusive e-cigarette use or dual use in females, which show nonsignificant positive associations (see minimum detectable effects and expected sample sizes in Supplementary Table S12). These nonsignificant positive HR estimates are suggestive of associations of hypertension onset with prolonged exclusive e-cigarette use or dual use in females and warrants further investigation. However, each of the HR estimates in males are near or below 1.0 and are thus unlikely to be type II error cases whereby true positive associations were not detected due to power.

There are no comparable studies that have examined associations of tobacco product use with hypertension in the current context in which e-cigarette use is appreciably prevalent, but some prior cross-sectional studies reported inconsistent chronic effects of cigarette use on BP change.²⁸ A recent longitudinal study using a 4786-individual cohort examined associations of smoking with BP measures by sex and showed that the increase in pulse pressure (PP) during 30-year follow-up that was associated with smoking was present among females but not males⁴; but no associations for SBP or DBP were observed.

Although mechanisms responsible for sex difference in hypertension were well studied,²⁹ the reason for a female-specific association of smoking with hypertension is not clear. Evidence indicates that, compared with males, females may have lower odds of successful smoking cessation.³⁰ However, heavy cigarette consumption and chronicity of smoking seems not enough to explain the sex difference because lifelong cigarette smoking (pack-years) was considered and adjusted for in models, and evidence suggests that males tend to smoke more cigarettes per day than females.³¹ Additionally, the cumulative exposure analysis found that both intermittent and consistent exclusive smoking across multiple survey waves was associated with increased hypertension incidence, with no evidence that the increased risk was stronger for consistent smoking than it was for intermittent smoking. Another explanation is that the results are explained by residual confounding that is more pronounced in females than males, but we are unaware of any plausible confound outside of the variables such as education level, BMI, diabetes mellitus, PA, family history of hypertension, and others controlled for here that might operate in such a sex-specific manner.

It is also possible that females might be vulnerable to the effects of smoking on hypertension. Females metabolize nicotine more rapidly,¹³ and thus experience stronger cigarette cravings than males in response to stress³² and experience more negative affect when stopping smoking.³³ Given that stress and negative affect are known risk factors for hypertension, it is possible that the prominent linkage of stress and distress with smoking in women could amplify the extent to which smoking may increase hypertension risk in females. Future studies are needed to address this putative mechanism and other possible mechanisms of the sex difference in the association between cigarette use and hypertension.

Limitations

First, the outcome variable in this study (i.e. diagnosis of hypertension) is based on self-reports by the respondent. Although recall bias was reduced by repeated assessment, potential misclassification still exists. In addition, we have limited information to further subgroup hypertension into elevated SBP and elevated DBP, which are considered as two independent risk factors of CVD.³⁴ Second, when assessing cardiovascular endpoints, an unfortunate constraint of the PATH study is its lack of data pertaining to dietary factors (e.g. sodium intake), and detailed assessment pertaining to stress and depression. They are particularly relevant for hypertension.^35,36 Third, compared to studies with long-term cigarette smoking exposure (e.g. 30 years⁴) for hypertension onset, the duration of follow-up (September 2013 to January 2018; maximum 52 months) in this study is relatively short for a cumulative effect of any long-term exposure to e-cigarettes on hypertension outcome. The short duration of e-cigarette exposure may be insufficient for either a young exclusive vaper to develop hypertension, or for an older vaper who was a former smoker to have significant cumulative effect over long-term smoking exposure. Fourth, the relatively small sample size of dual users made it not allowable to further distinguish smokers who partially switched to e-cigarettes from smokers who added e-cigarettes for more enjoyment. Thus, the effect of dual use observed among some sub-groups in this study is mixed. Finally, the e-cigarette marketplace has been evolving rapidly since the time of the initiation of PATH study.³⁷ The PATH study is an ongoing study. Future study is needed to re-evaluate associations with health outcomes as e-cigarette technology continues to evolve and duration of follow-up allows more power.

Conclusions

The association of e-cigarette, cigarette, and dual use with hypertension may differ by sex, whereby exclusive cigarette use could be a prospective risk factor for subsequent self-reported hypertension in US adult females but not males. Because of statistical power limitations and the relatively short period of e-cigarette exposure in this study, further research on the association of e-cigarette use or dual use with hypertension in females is warranted.

Supplementary Material

A Contributorship Form detailing each author’s specific involvement with this content, as well as any supplementary data, are available online at https://academic.oup.com/ntr.

Funding

This study was supported by the University of South California Tobacco Center of Regulatory Science Pilot Award (no. 138628897) from the National Cancer Institute of the National Institutes of Health under Award Number U54CA180905 (AML, HS, DL, and DJO), by the University of Rochester CTSA award number UL1 TR002001 from the National Center for Advancing Translational Sciences of the National Institutes of Health (DL), and by the University of Rochester Infection and Immunity: From Molecules to Populations (IIMP) award number BWF-1014095 from the Burroughs Welcome Fund of Institutional Program Unifying Population and Laboratory Based Sciences (HS). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or the Food and Drug Administration (FDA).

Acknowledgments

None.

Declaration of Interests

The authors have no conflicts of interest to disclose.

Author Contributions

HS, DL, and DJO: Conceived and designed the study. HS, DL, DJO, and AML: Analyzed the data. HS, DL, DJO, AML, and QW: Wrote and edited the manuscript.

Data Availability

The PATH waves 1–4 data are de-identified open-source data that are publicly available from the National Addiction and HIV Data Archive Program (NAHDAP) website (https://www.icpsr.umich.edu/icpsrweb/NAHDAP/studies/36498/datadocumentation). Before the start of data collection, the PATH study received approval from the Westat Institutional Review Board.