THE AUTHORS REPLY

We thank Aubin et al. (1) for their comments. As we emphasized in our article (2), we fully agree with Aubin et al. as to the importance of smoking cessation to public health. Indeed, our recent study demonstrated that even nondaily cigarette smokers benefit from smoking cessation (3).

Unfortunately, rather than quit, people may decrease the number of cigarettes they smoke per day (CPD) over their lifetime. Relatively few studies have investigated the impact of changing CPD on the risk of mortality (2, 4–7). Existing studies have had limitations, such as small numbers of deaths or examination of changes over only a short time period. In our study, we took advantage of the large cohort size and availability of detailed information on smoking at different ages to examine the association of changes in CPD between ages 25–29 years and 50–59 years with subsequent mortality risk. We fully acknowledge that our study was limited by participants’ recalling their prior smoking, although we noted the internal consistency of our findings—for example, participants who reported reducing their CPD by a larger extent had lower mortality risks than participants who reported reducing their CPD to a lesser extent, with still lower risks being observed among participants who had quit smoking (2).

In addition to our main analysis, we identified people who had changed their CPD between the 1995–1996 and 2004–2005 questionnaires. Although that analysis benefited from participants’ contemporaneously recalling their smoking at each assessment point, it was limited by determining changes in CPD over just a decade, as opposed to a period of 25–35 years in our primary analysis. More importantly, our statistical power was substantially lower for those analyses, as they required participants to be current smokers in 2004–2005 (18,117 participants; 4,283 subsequent deaths), as opposed to our primary analysis, which included 51,633 participants (11,593 deaths) who smoked at both ages 25–29 years and ages 50–59 years. For these reasons, we presented results for reported changes in CPD between 1995–1995 and 2004–2005 in a supplemental table rather than as part of our primary analysis. Although statistical power in the supplemental analysis was modest, we observed findings that were parallel to those of our primary presentation. Participants reporting an increase in CPD had generally higher mortality risks than those who smoked a consistent amount, and participants reporting a large reduction in CPD (from >30 CPD to >0–10 CPD) had a lower mortality risk than participants who continued smoking >30 CPD, although the observed reduction in risk was substantially less than that observed for smoking cessation (2).

Large changes in CPD over a long duration of time can potentially have a large impact on the total number of cigarettes that people smoke during their lifetime. For example, a pack-a-day (20-CPD) smoker at age 25 years will smoke about 219,000 cigarettes by the age of 55 if they continue to smoke 20 CPD but 110,000 cigarettes by the age of 55 if they reduce their smoking to 10 CPD. Yet, a smaller reduction in CPD will have much less impact on the total number of cigarettes a person smokes over their lifetime, as would changing CPD for just a few years rather than the 25–35 years examined in our article. The impact of these two factors—the magnitude of CPD change and the duration of CPD change—on disease risk and mortality is an important yet unresolved research area. Quantifying the shape of the association, including whether changes in CPD relate linearly or log-linearly to mortality, will also require future research, ideally in large studies with serial measures of cigarette use. We are actively working to identify such cohorts ourselves and are hopeful that this dialogue will inspire other investigators to conduct such studies.