Allen and her colleagues [1] report that the timing of the menstrual cycle influences women's success in smoking cessation. The observed effect is puzzling, because it is the opposite of the direction expected (women quitting in the follicular phase did better), and because it was not associated with measured variations in mood, which was the expected mechanism. It may be that hormonal status affects quitting processes by other mechanisms as well. Obviously, the findings have practical implications for when women should attempt to quit. A caveat is that asking smokers to wait may cause some to lose interest, as suggested by the dropout rate in this study. Nevertheless, although the menstrual cycle effect was relatively small, smoking cessation is so important, and so difficult, that smokers must exploit every possible advantage. Allen et al.'s findings may also help to explain variations in women's success in smoking cessation, which is often observed to be lower than men's [2].

Allen et al.'s findings have implications beyond menstrual cycle effects. More broadly, the findings illustrate the importance of the smoker's internal state at the time of quitting or at the time of lapsing. Such variations have, historically, received too little attention. Research into smoking cessation and relapse has tended to focus upon individual differences—identifying individual characteristics associated with smoking cessation success (e.g. [3]). For example, heavier and more dependent smokers [4] and those with certain genetic polymorphisms [5] have higher failure rates. Such findings can be theoretically and clinically important. Moreover, the focus on individual difference is consistent with a larger tendency in behavioral science to put stable personal characteristics of the individual at the center of our explanatory frameworks; explanations from personality, psychiatric diagnosis and so on are seen as powerful and useful.

However, this focus on stable individual differences can lead to relative neglect of more dynamic and more immediate influences on behavior. In the study of lapses, the case for these ‘local’ influences is particularly strong, because each lapse is a particular episode occurring at a particular time and place. Whatever power they have in explaining who lapses, stable individual differences cannot explain when they lapse. Indeed, a large body of evidence documents the effect of the proximal environment on lapse risk; for example, lapses tend to occur when smokers are drinking, exposed to smoking cues and/or emotionally distressed [6,7]. Looking beyond the immediate setting in which a lapse occurs, other data from Allen et al.'s study [8] show that craving increased in the days preceding a lapse, and Shiffman & Waters [9] showed that negative affect rose in the hours before subjects' initial lapses. Importantly, both these analyses focus on within-person changes in state: unlike analyses that suggest that people who have higher cravings or more negative affect are prone to relapse, these analyses tell us that, for an individual smoker, it is when craving or negative affect increases that a lapse is likely to occur.

This more dynamic view of the lapse and relapse process does not neglect individual differences, but puts more emphasis upon how variations in the smoker's internal state (mood, craving, fatigue, effects of other drugs, physiological or hormonal status) and in the immediate environment (the presence of cues, the availability of cigarettes) promote or prevent lapsing [10,11]. In this sense, it focuses upon understanding why a smoker of a particular kind lapsed at a particular time, as well as upon why particular smokers lapse where others do not.

Such a dynamic approach to relapse has daunting consequences for theory, treatment and research methods. Theoretically, dynamic explanations are typically more complex: they trade on variation over time, and often suggest complex interactions leading to precipitous or ‘catastrophic’ processes [12], rather than simple main effect associations. The implications for treatment can be daunting when one considers that a smoker's abstinence can be undone by unpredictable, even random, events. This implies that treatment cannot simply change the person in some simple way, but must imbue the person with the ability to respond more effectively to a wide range of situational challenges. Finally, the study of dynamic phenomena requires research methods capable of capturing dynamic swings in subject states and situations. Whereas individual differences can be measured once, at an arbitrary time of the investigator's choosing, dynamic effects must be assessed repeatedly over time, often in the natural environment—the approach of Ecological Momentary Assessment [13]. In the Allen et al. study, the diary data did not show variation in mood according to the phase of the smokers' menstrual cycle, but this may reflect limitations in the methods, including the use of paper diaries, which are subject to back-filling [14] and thus may reintroduce recall bias and fail to capture actual real-time variations in state.

The Allen et al. study does more than implicate menstrual cycle effects in quitting and relapse—it reminds us that, in their struggle to quit smoking, smokers are buffeted by volatile changes in their physiological and emotional states as well as the situations in which they find themselves. More attention to such dynamic processes may help us to more clearly understand—and ultimately prevent—relapse.

Declaration of interest

The author is a co-founder of invivodata, inc., which provides electronic diaries for clinical trials.