This study aimed to elucidate the relations between ovarian hormones and smoking behavior. Our study is strengthened by the direct and multivariate assessment of hormone levels at a time when they were temporally congruent with the assessment of smoking behavior, which was also defined in multiple ways. Topography variables were reduced to three meaningful factors: number of puffs, flow rate, and puff intensity. The three factors appear to assess different aspects of smoking behavior, and they were differentially related to the ovarian hormone variables assessed.
The results of this study suggest that both the change in hormone levels over time and
the relative ratio of progesterone to estradiol appear to influence smoking behavior, but each to a limited extent. Our results may help to explain inconsistent findings in previous studies and suggest potential mechanisms by which hormones influence nicotine addiction. Decreasing estradiol and decreasing progesterone were also associated with greater puff intensities in the current study. These results are consistent with a previous study by Snively and colleagues that found increased smoking during the late luteal phase (Snively, et al., 2000
), which is characterized by both decreasing estradiol and progesterone.
Although variance in smoking behaviors accounted for by ovarian hormone measures seen in this study was small (4–6%), gender differences in nicotine dependence and ovarian hormone function remain an important area of investigation. According to Cohen (1992)
=.10 represents a small effect size, whereas .3 and .5 represent medium and large effect sizes, respectively. Rosenthal (1990)
has asserted that even small associations between variables are often meaningful in a practical context. Examples include the “small” associations between smoking and mortality, aspirin use and myocardial infarction, and gender and chronic heart disease (Rutledge & Loh, 2004
Consistent with the notion that estradiol may increase sensitivity to nicotine, whereas progesterone is protective (Lynch & Sofuoglu, 2010
), high levels of progesterone relative to estradiol (i.e., P:E ratio) were associated with a lower number of puffs and a smaller mass of cigarettes smoked. Conversely, low levels of progesterone relative to estradiol were associated with a higher number of puffs and a greater mass of cigarettes smoked. Notably, this effect cannot be explained by the effect of progesterone alone on smoking because the absolute level of progesterone was not significantly associated with smoking. Instead, progesterone appears to reduce smoking behavior only when it is not opposed by estradiol. These findings are consistent with animal and human studies demonstrating diminished interest in nicotine following progesterone administration in the context of low levels of estradiol (Lynch, 2009
; Sofuoglu & Mooney, 2009
) and diminished relapse rates among female smokers who quit during the luteal phase compared to the follicular phase (Allen, et al., 2008
). In this study, the P:E ratio was more strongly correlated with smoking behavior than the E:P ratio, which suggests that the level of progesterone relative to estradiol rather than the reverse should be examined in future studies of smoking behavior.
Results of this study suggest that the best predictors of smoking behavior are the ratio of progesterone to estradiol and the change in ovarian hormones over time. However, the mechanism by which decreasing ovarian hormones influence the volume and duration of cigarette puffs is unclear, as is how a low ratio of progesterone to estradiol influences smoking behavior. Future studies should replicate these findings and explore potential mechanisms by which changes in hormone levels and the ratio of progesterone to estradiol differentially influence smoking behavior.
In the current study, the smoking measures were collected during an adlib
smoking session without any alternative reinforcers. The addition of an alternative reinforcer to smoking (e.g., money) may increase the effect size of steroid hormones on smoking behavior in future studies. Moreover, ovarian hormones may affect smoking behaviors in some women more than others, which could explain the small effect sizes detected in the current study. For example, it is well established that some women are differentially sensitive to the effects of ovarian hormones (Rubinow, Schmidt, & Roca, 1998
). Moreover, previous research suggests an increased desire to smoke and to relieve negative affect during the late luteal phase only among women with premenstrual symptoms (Allen, et al., 1999
). The effects of hormones on smoking may therefore be greater in women who experience premenstrual symptoms. Future studies may identify subpopulations of female smokers acutely sensitive to the effects of ovarian hormones.
This study had certain weaknesses. First, the controlled laboratory environment in which the study was conducted and the inclusion of treatment-seeking women limits the generalizability of the results. Second, estradiol and progesterone were measured relatively infrequently. The hormone change indexes created in this study represent the change in hormone levels between the first and second study visits, which were approximately 11 days apart. Hormones change significantly over the course of 11 days in normally cycling women, and thus, the hormone change index was a very rough indicator of change over time. More refined (e.g., daily) hormonal assessment is preferred, but feasibility may be diminished. Recent advances in hormone assay techniques allow for ovarian hormone detection in saliva rather than plasma. Salivary sampling methods are not only easier and more acceptable to women, but they also detect the unbound or “biologically available” portion of ovarian hormones, allowing for more powerful analyses of associations between hormones and behavioral symptoms (Edler, Lipson, & Keel, 2007
; Shirtcliff et al., 2000
). Fourth, the effects of the cue reactivity assessment that immediately preceded the adlib
smoking procedure are unknown. However, all subjects received the same cue reactivity procedure, regardless of menstrual phase, and thus, it is unlikely that the cue reactivity session had a meaningful effect on the associations between hormones and smoking. Finally, a more direct test of the hypothesis that ovarian hormones influence smoking behavior in women could come from future experimental studies wherein exogenous estradiol and progesterone are administered to female smokers and the subsequent effects on smoking are examined. Future studies could also examine whether nicotine consumption per se or smoking behavior more generally are affected by changing hormone levels by examining differences in behavior when subjects are given denicotinized versus regular nicotine-containing cigarettes.
Advances in this line of research may lead to novel pharmacological treatments for nicotine dependence. For example, women who are less likely to achieve abstinence may benefit from a combination treatment such as varenicline plus transdermal progesterone. Alternatively, reducing the cyclic fluctuations in ovarian hormones with hormonal contraceptives may attenuate the exacerbation of smoking behavior seen in the late luteal phase. An improved understanding of the association between ovarian hormone levels and smoking could be used to enhance behavioral smoking interventions. By helping women to identify predictable biological triggers for smoking, behavioral strategies could be implemented to prepare women for periods of increased urges to smoke.
In conclusion, the ratio of progesterone to estradiol and the change in hormone levels over time were the strongest hormonal predictors of smoking behavior. Ovarian hormone function remains a primary candidate for explaining gender differences in nicotine addiction. Future studies are needed to elucidate the neurobiological mechanisms by which hormones influence smoking and to determine the utility of ovarian hormone manipulation in augmenting women’s response to smoking cessation treatment.