https://orcid.org/0000-0002-4643-6809
Abstract
Reducing cigarettes per day may lower the risk of lung cancer compared with continuing to smoke at the same intensity. Other changes in smoking behaviors, such as increasing cigarette consumption or quitting for a period and relapsing, may also affect lung cancer risk.
We examined changes in smoking status and cigarettes per day among 24 613 Finnish male smokers aged 50-69 years who participated in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study. Longitudinal data on smoking were collected during study follow-up visits 3 times a year (approximately every 4 months) between 1985 and 1993. Incident lung cancer patients through 2012 were identified by the Finnish Cancer Registry. Risk ratios (RRs) and 95% confidence intervals (CIs) were estimated using Cox proportional hazards regression.
Compared with smoking 20 cigarettes per day continuously across the intervention period, reducing an average of 5 cigarettes per day per year while smoking was associated with a 20% lower risk of lung cancer (95% CI = 0.71 to 0.90). A substantially lower risk of lung cancer was also observed when participants smoked at 50% (RR = 0.72, 95% CI = 0.57 to 0.90) and 10% (RR = 0.55, 95% CI = 0.36 to 0.83) of study visits, relative to smoked at 100% of study visits.
Smokers may lower their risk of lung cancer by reducing smoking intensity (cigarettes per day while smoking) and the time they smoke. However, quitting smoking completely is the most effective way for smokers to reduce their risk of lung cancer.
Lung cancer is the leading cause of cancer death globally (1), and nearly two-thirds of the estimated 1.8 million lung cancer deaths are attributed to cigarette smoking (2,3). Extensive epidemiologic evidence summarized in the International Agency for Research on Cancer’s 11th Handbook of Cancer Prevention and the 2020 US Surgeon General’s Report concludes that smoking cessation reduces the risk of lung cancer incidence and mortality (4,5). There is also some evidence from population-based cohort studies that reducing cigarettes per day lowers the risk of lung cancer compared with continuing to smoke at the same intensity (6-9).
A recent meta-analysis (10) showed an overall 28% reduction in the risk of lung cancer for individuals who reduced the number of cigarettes per day by at least 50% compared with individuals who continued to smoke heavily (defined as ≥15 cigarettes per day). Similar results were found in analyses that used different definitions of smoking reduction, for example, reducing from heavy smoking (≥20 cigarettes per day) to moderate smoking (10-19 cigarettes per day) (10). However, it is unclear if other changes in smoking behaviors, such as increasing cigarette consumption or quitting smoking for a period and relapsing, may also affect lung cancer risk as well as other smoking-related diseases.
Additionally, most of the prospective cohort studies examining changes in smoking behaviors and lung cancer incidence collected smoking data at only a few timepoints (eg, at baseline and during a second follow-up examination) (6,9,11,12) or relied on participants to recall their smoking behaviors at different periods of their lives (13). However, descriptive studies of smoking patterns (14,15) show that smokers often undergo multiple transitions between smoking, abstinence, relapse, and changes in smoking intensity, with some of these attempts resulting in permanent smoking cessation (14).
Here, we leveraged the prospective longitudinal data on cigarette smoking collected in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention (ATBC) Study, a cancer prevention trial conducted in Finland between 1985 and 1993. The objective of the trial was to evaluate the effects of alpha-tocopherol and beta-carotene supplementation on the incidence of lung cancer and other cancers among 29 133 male smokers (16). As part of the study, data on cigarette smoking were collected 3 times per year during the intervention period. With more than 20 years of postintervention follow-up, the ATBC Study provides a unique opportunity to examine whether changes in smoking behaviors modify the risk of subsequent cancer incidence.
Methods
Participants
This analysis is set within the ATBC Study, a randomized, double-blinded, placebo-controlled trial conducted in Finland between 1985 and 1993, which enrolled 29 133 male current smokers aged 50-69 years. The ATBC Study protocol was approved by the institutional review boards at the US National Cancer Institute and the Finnish National Public Health Institute. Detailed information including procedures for informed consent are available at the ATBC Study website (https://atbcstudy.cancer.gov).
The ATBC Study recruitment started in April 1985 and continued until June 1988. All men completed baseline questionnaires on sociodemographic characteristics, medical conditions, smoking, and occupational history. Follow-up consisted of 3 study visits per year to the local field center during which the men were asked about their health and smoking behaviors since the last study visit. The measurement interval for the longitudinal assessment of cigarette smoking was approximately 4 months (see Figure 1) from baseline and ended in April 1993. For 80% (n = 19 744) of men, data on smoking were available for at least 14 study visits (approximately 5 years). Because we used a time-to-event analysis where the study outcome was time from the end of the intervention to either cancer diagnosis or the end of follow-up for all participants, men who died or were diagnosed with cancer during the intervention period (n = 4520) were excluded. This approach may help mitigate reverse causation, because deaths occurring in the first years of the intervention are more likely to be associated with prevalent diseases, which may in turn influence smoking behavior (17). For this analysis, 24 613 men were followed from the end of the intervention period in 1993 to the first lung cancer diagnosis, death, or December 31, 2012, whichever occurred first.
Longitudinal assessment of cigarette smoking in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study from 1985 to 1993.
Cigarette smoking
At baseline, all men smoked 5 or more cigarettes per day. Smoking status and cigarettes per day were then reassessed at each study visit using 2 questions: “Have you smoked since your last visit?” (No; Yes, but I have now stopped; Yes, continuously) and “How much do you smoke daily on an average at present?” Men who stopped smoking since their last visit were assigned the “not smoking” status, whereas those who responded “Yes, but I have now stopped” were classified as “trying to quit.” The status of “still smoking” was assigned to men who reported having smoked since their last visit.
Lung cancer incidence
Incident cases of lung cancer were identified through linkage with the Finnish Cancer Registry, which provides accurate and nearly complete national cancer data overall (96% completeness for all-solid tumors and 86% completeness for nonsolid tumors) (18). In addition, medical records of cancer patients identified through April 1999 were reviewed by 1 or 2 study physicians or oncologists to confirm the diagnosis. After April 1999, only a subset of medical records was reviewed annually. Overall, 59% of all medical records were reviewed. There were 3013 participants diagnosed with incident lung cancer during a mean (SD) follow-up of 12.8 (6.4) years. Among them, 399 (13.2%) men had between 2 and 4 cancer diagnoses.
Statistical analysis
The objective of the analysis was to examine whether changes across the 3 smoking statuses and in smoking intensity (defined by the number of cigarettes per day while smoking) over the intervention period were associated with changes in subsequent lung cancer risk. We used a 2-stage analysis to estimate these relationships. We first estimated the proportion of study visits for each smoking status and the probabilities of transitioning between these statuses (eg, from smoking to not smoking and from smoking to trying to quit) over successive study visits; these estimates were computed on an individual basis. The smoking intensity variables were summarized by fitting a linear regression with time (study visits) as the independent variable and cigarettes per day as the dependent variable; these regression models were individually fitted using data from visits in which smoking was reported. The estimated intercept and slope correspond to the baseline number of cigarettes smoked per day and the average rate of change in smoking intensity per intervention year while smoking, respectively.
In the second stage of the analysis, we used Cox proportional hazards regression to estimate the relationship between the derived smoking behavior summaries and lung cancer incidence. Specifically, we can distinguish between effects of the proportion of smoking visits (and transitions between smoking statuses) and the dynamics of smoking intensity on lung cancer risk. Models were adjusted for the vitamin supplementation group (alpha-tocopherol alone, beta-carotene alone, alpha-tocopherol plus beta-carotene; reference group: placebo) (19), age at the end of the intervention, education (high school or less, college, and college graduate), and age at smoking initiation (20-22). Sensitivity analyses included a time-dependent coefficient Cox proportional hazards model to examine potential differences in risk ratios (RRs) for years 0-10 (1993-2002) and for greater than 10 years (2003-2012) and potential interactions for vitamin supplementation group and smoking variables.
Using the results of the Cox models, we estimated the risk of lung cancer for each participant relative to a reference individual who smoked 20 cigarettes per day at baseline, smoked during all the study visits, and did not change smoking intensity across the intervention period. We compared smoking trajectories for participants at extreme ranges of risk. Further, we calculated risk ratios of lung cancer for different scenarios of smoking trajectories (eg, baseline smoking intensity of 20 cigarettes per day, smoking 50% of the study visits, and increasing smoking intensity at a rate of 5 cigarettes per day per intervention year). For these calculations, we created a dichotomous variable for smoking by grouping men who were trying to quit with those who were still smoking. All statistical tests were 2-sided, and a P value of less than .05 was used to define statistical significance.
Results
Table 1 displays the participant’s characteristics by categories of smoking intensity at baseline. Overall, the average age of smoking initiation was 20 years, and 67.1% (n = 16 513) of men smoked between 5 and 20 cigarettes per day. Compared with men who smoked 5-15 cigarettes per day at baseline, those who smoked more than 20 cigarettes per day started smoking at a somewhat younger age, had a higher educational level (college, college graduate), and were more likely to report a family history of lung cancer.
Baseline participant characteristics by categories of smoking intensity (1985-1993)
| Characteristic | 5-15 cigarettes per day (n = 8207) | 16-20 cigarettes per day (n = 8306) | ≥21 cigarettes per day (n = 8100) |
|---|---|---|---|
| Mean age (SD), y | 64.0 (5.2) | 62.9 (4.9) | 61.9 (4.7) |
| Mean age (SD) of smoking initiation, y | 20.5 (5.5) | 19.4 (4.4) | 18.8 (4.2) |
| Education, % | |||
| High school or less | 33.8 | 35.0 | 31.7 |
| College | 61.5 | 61.8 | 63.1 |
| College graduate | 4.7 | 3.2 | 5.2 |
| Lung cancer participants, % | 9.1 | 12.3 | 15.4 |
| Family history of lung cancer, % | 9.5 | 11.6 | 12.3 |
| Characteristic | 5-15 cigarettes per day (n = 8207) | 16-20 cigarettes per day (n = 8306) | ≥21 cigarettes per day (n = 8100) |
|---|---|---|---|
| Mean age (SD), y | 64.0 (5.2) | 62.9 (4.9) | 61.9 (4.7) |
| Mean age (SD) of smoking initiation, y | 20.5 (5.5) | 19.4 (4.4) | 18.8 (4.2) |
| Education, % | |||
| High school or less | 33.8 | 35.0 | 31.7 |
| College | 61.5 | 61.8 | 63.1 |
| College graduate | 4.7 | 3.2 | 5.2 |
| Lung cancer participants, % | 9.1 | 12.3 | 15.4 |
| Family history of lung cancer, % | 9.5 | 11.6 | 12.3 |
Baseline participant characteristics by categories of smoking intensity (1985-1993)
| Characteristic | 5-15 cigarettes per day (n = 8207) | 16-20 cigarettes per day (n = 8306) | ≥21 cigarettes per day (n = 8100) |
|---|---|---|---|
| Mean age (SD), y | 64.0 (5.2) | 62.9 (4.9) | 61.9 (4.7) |
| Mean age (SD) of smoking initiation, y | 20.5 (5.5) | 19.4 (4.4) | 18.8 (4.2) |
| Education, % | |||
| High school or less | 33.8 | 35.0 | 31.7 |
| College | 61.5 | 61.8 | 63.1 |
| College graduate | 4.7 | 3.2 | 5.2 |
| Lung cancer participants, % | 9.1 | 12.3 | 15.4 |
| Family history of lung cancer, % | 9.5 | 11.6 | 12.3 |
| Characteristic | 5-15 cigarettes per day (n = 8207) | 16-20 cigarettes per day (n = 8306) | ≥21 cigarettes per day (n = 8100) |
|---|---|---|---|
| Mean age (SD), y | 64.0 (5.2) | 62.9 (4.9) | 61.9 (4.7) |
| Mean age (SD) of smoking initiation, y | 20.5 (5.5) | 19.4 (4.4) | 18.8 (4.2) |
| Education, % | |||
| High school or less | 33.8 | 35.0 | 31.7 |
| College | 61.5 | 61.8 | 63.1 |
| College graduate | 4.7 | 3.2 | 5.2 |
| Lung cancer participants, % | 9.1 | 12.3 | 15.4 |
| Family history of lung cancer, % | 9.5 | 11.6 | 12.3 |
As shown in panels A to C of Figure 2, we observed substantial interindividual variation in smoking trajectories among men who reported smoking throughout the intervention period. Figure 2, A, shows examples of smoking trajectories of men who reported fewer cigarettes per day at the last study visit than at baseline. Figure 2, B, shows examples of men who smoked at approximately the same intensity across the intervention period, and Figure 2, C, shows examples of those who smoked more cigarettes per day at the last study visit than at baseline. Overall, the average smoking intensity was 20.3 cigarettes per day at baseline. Men who reported smoking throughout the intervention period reduced their smoking intensity by an average of 0.41 cigarettes per day per intervention year, that is, approximately 2 cigarettes per day over a 5-year period (5 times 0.41).
Examples of smoking trajectories from participants who reported smoking throughout the intervention period (1985-1993).
Results from Cox regression models showed that the proportion of study visits in which men smoked, the number of cigarettes per day smoked at baseline, and the rate of change in smoking intensity over the intervention period were each associated with changes in subsequent risk of lung cancer.
Compared with not smoking, the risk of lung cancer was 14% (RR = 1.14, 95% confidence interval [CI] = 1.04 to 1.25) higher for every 20% increase in the proportion of visits in which men smoked. With respect to baseline smoking intensity, the risk of lung cancer increased by 13% (RR = 1.13, 95% CI = 1.07 to 1.19) for each 5 cigarettes per day increase in this intensity. Regarding the rate of change in smoking intensity during the intervention period, the risk of lung cancer increased by 25% (RR = 1.25, 95% CI = 1.11 to 1.40) for each increase of 5 cigarettes per day per intervention year while smoking (Table 2). The association between the proportion of study visits in which men smoked and the risk of lung cancer strengthened slightly (RR = 1.17, 95% CI = 1.01 to 1.36) when we additionally included the transition probabilities between smoking statuses in the Cox model (Supplementary Table 1, available online). We note that the effects for the transition probabilities were small with 95% confidence intervals containing 1.
Lung cancer risk among participants in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study
| Explanatory variables | RR (95% CI) |
|---|---|
| Smoking behavior | |
| Not smoking | 1.00 (Referent) |
| Trying to quita | 1.09 (0.80 to 1.47) |
| Smokingb | 1.14 (1.04 to 1.25) |
| Baseline cigarettes per dayc | 1.13 (1.07 to 1.19) |
| Rate of change cigarettes per dayd | 1.25 (1.11 to 1.40) |
| Intervention | |
| Placebo | 1.00 (Referent) |
| Alpha-tocopherol | 1.06 (0.78 to 1.43) |
| Beta-carotene | 1.15 (0.85 to 1.55) |
| Alpha-tocopherol plus beta-carotene | 1.09 (0.80 to 1.49) |
| Education | |
| High school or less | 1.00 (Referent) |
| College | 0.73 (0.59 to 0.91) |
| College graduate | 0.40 (0.20 to 0.78) |
| Age, y | 1.15 (1.12 to 1.17) |
| Age at smoking initiation, y | 0.91 (0.88 to 0.94) |
| Explanatory variables | RR (95% CI) |
|---|---|
| Smoking behavior | |
| Not smoking | 1.00 (Referent) |
| Trying to quita | 1.09 (0.80 to 1.47) |
| Smokingb | 1.14 (1.04 to 1.25) |
| Baseline cigarettes per dayc | 1.13 (1.07 to 1.19) |
| Rate of change cigarettes per dayd | 1.25 (1.11 to 1.40) |
| Intervention | |
| Placebo | 1.00 (Referent) |
| Alpha-tocopherol | 1.06 (0.78 to 1.43) |
| Beta-carotene | 1.15 (0.85 to 1.55) |
| Alpha-tocopherol plus beta-carotene | 1.09 (0.80 to 1.49) |
| Education | |
| High school or less | 1.00 (Referent) |
| College | 0.73 (0.59 to 0.91) |
| College graduate | 0.40 (0.20 to 0.78) |
| Age, y | 1.15 (1.12 to 1.17) |
| Age at smoking initiation, y | 0.91 (0.88 to 0.94) |
Risk ratio corresponds to a change of 20% in the proportion of visits when men were trying to quit. CI = confidence interval; RR = risk ratio.
Risk ratio corresponds to a change of 20% in the proportion of visits when men smoked.
Risk ratio corresponds to an increase of 5 cigarettes per day in the baseline smoking intensity.
Risk ratio corresponds to an increase in smoking intensity at a rate of 5 cigarettes per day per intervention year.
Lung cancer risk among participants in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study
| Explanatory variables | RR (95% CI) |
|---|---|
| Smoking behavior | |
| Not smoking | 1.00 (Referent) |
| Trying to quita | 1.09 (0.80 to 1.47) |
| Smokingb | 1.14 (1.04 to 1.25) |
| Baseline cigarettes per dayc | 1.13 (1.07 to 1.19) |
| Rate of change cigarettes per dayd | 1.25 (1.11 to 1.40) |
| Intervention | |
| Placebo | 1.00 (Referent) |
| Alpha-tocopherol | 1.06 (0.78 to 1.43) |
| Beta-carotene | 1.15 (0.85 to 1.55) |
| Alpha-tocopherol plus beta-carotene | 1.09 (0.80 to 1.49) |
| Education | |
| High school or less | 1.00 (Referent) |
| College | 0.73 (0.59 to 0.91) |
| College graduate | 0.40 (0.20 to 0.78) |
| Age, y | 1.15 (1.12 to 1.17) |
| Age at smoking initiation, y | 0.91 (0.88 to 0.94) |
| Explanatory variables | RR (95% CI) |
|---|---|
| Smoking behavior | |
| Not smoking | 1.00 (Referent) |
| Trying to quita | 1.09 (0.80 to 1.47) |
| Smokingb | 1.14 (1.04 to 1.25) |
| Baseline cigarettes per dayc | 1.13 (1.07 to 1.19) |
| Rate of change cigarettes per dayd | 1.25 (1.11 to 1.40) |
| Intervention | |
| Placebo | 1.00 (Referent) |
| Alpha-tocopherol | 1.06 (0.78 to 1.43) |
| Beta-carotene | 1.15 (0.85 to 1.55) |
| Alpha-tocopherol plus beta-carotene | 1.09 (0.80 to 1.49) |
| Education | |
| High school or less | 1.00 (Referent) |
| College | 0.73 (0.59 to 0.91) |
| College graduate | 0.40 (0.20 to 0.78) |
| Age, y | 1.15 (1.12 to 1.17) |
| Age at smoking initiation, y | 0.91 (0.88 to 0.94) |
Risk ratio corresponds to a change of 20% in the proportion of visits when men were trying to quit. CI = confidence interval; RR = risk ratio.
Risk ratio corresponds to a change of 20% in the proportion of visits when men smoked.
Risk ratio corresponds to an increase of 5 cigarettes per day in the baseline smoking intensity.
Risk ratio corresponds to an increase in smoking intensity at a rate of 5 cigarettes per day per intervention year.
Compared with placebo, there were no effects of the vitamin supplementation (alpha-tocopherol, beta-carotene, or alpha-tocopherol plus beta-carotene) on lung cancer risk (Table 2; Supplementary Table 1, available online), and there was no evidence for interactions between the intervention group and smoking variables in the sensitivity analysis (data not shown).
Results from the time-varying coefficient model (Supplementary Table 2, available online) showed that the effects of smoking behavior during the intervention period on lung cancer risk were similar when distinguishing between the early and later follow-up (<10 and >10 years of follow-up).
Using the results of the Cox models, we estimated risk ratios of lung cancer for each participant and ranked them from the top (highest risk) to the bottom (lowest risk) of the distribution. As shown in the examples of smoking trajectories in panels A and B of Figure 3, men with the highest risk of lung cancer consistently smoked many cigarettes per day at all study visits (Figure 3, A). In contrast, men with the lowest risk of lung cancer stopped smoking and remained abstinent throughout the intervention period (Figure 3, B).
Examples of smoking trajectories among participants in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study from the highest (A) to lowest (B) risk of lung cancer. RR = risk ratio.
In general, we observed a dose-response relationship with increased lung cancer risk ratios associated with higher intensity of smoking (Figure 4). For example, compared with smoking 20 cigarettes per day from baseline for 100% of study visits and holding this smoking intensity constant, increasing smoking intensity at an average rate of 5 cigarettes per day per intervention year was associated with a 25% increase in the risk of lung cancer (RR = 1.25, 95% CI = 1.11 to 1.40). If the baseline smoking intensity was 40 cigarettes per day, the risk associated with smoking 100% of the visits in addition to increasing smoking intensity at an average rate of 5 cigarettes per day per intervention year would be 2 times (RR = 2.02, 95% CI = 1.57 to 2.62) the risk of smoking 20 cigarettes per day at baseline.
Relative risks (RRs) for lung cancer associated to changes in smoking trajectories in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study. CI = confidence interval.
In contrast, compared with the same smoking intensity of 20 cigarettes per day from baseline for 100% of study visits, reducing smoking intensity at an average rate of 5 cigarettes per day per intervention year was associated with a 20% (RR = 0.80, 95% CI = 0.71 to 0.90) reduction in lung cancer risk. Substantial changes in lung cancer risk were also seen by reducing the proportion of study visits in which men smoked. Compared with smoking 20 cigarettes per day from baseline for 100% of study visits, the risk of lung cancer was 28% (RR = 0.72, 95% CI = 0.57 to 0.90) and 45% (RR = 0.55, 95% CI = 0.36 to 0.83) lower when participants smoked only 50% and 10% of the study visits, respectively (Figure 4).
Discussion
In this prospective cohort study of more than 24 000 male Finnish cigarette smokers and over an average postintervention follow-up of 12.8 years, we found that changes in the rate of cigarettes per day and in the proportion of study visits in which men smoked during the intervention were each associated with changes in the risk of lung cancer. Compared with smoking 20 cigarettes per day from baseline for 100% of the study visits, reducing smoking intensity at an average rate of 5 cigarettes per day per intervention year was associated with a 20% reduction in the risk of lung cancer. Furthermore, compared with smoking 20 cigarettes per day consistently from baseline throughout the intervention period, reducing the proportion of time when participants reported smoking to 50% and to 10% was associated with a 28% and 45% lower risk of lung cancer, respectively. To our knowledge, this is the first study to report independent effects of prospectively assessed changes in the amount of time smoking and changes in smoking intensity on lung cancer incidence.
These results are consistent with findings of previous population-based cohort studies conducted in Denmark (23) and South Korea (7,9) that reported a decrease in lung cancer risk for those who reduced cigarettes per day by more than 50%. In the Danish study (23), men and women aged 20-93 years were categorized as heavy smokers (≥15 cigarettes per day) and light smokers (1-14 cigarettes per day). Smoking reduction was defined as having reported smoking heavily (≥15 cigarettes per day) at the first examination (1964) and reported smoking at least 50% less without quitting at the second examination (1998). Compared with participants who smoked at least 15 cigarettes per day persistently, smoking reduction was associated with a 27% lower risk of lung cancer (hazard ratio [HR] = 0.73, 95% CI = 0.36 to 0.69).
In the Korean National Prospective Occupational cohort study (7), all participants were male civil servants aged 30-58 years. Participants were classified as heavy (≥20 cigarettes per day), moderate (10-19 cigarettes per day), and light smokers (<10 cigarettes per day) at the first examination (1990), and reduction of smoking was defined as any change from heavy to moderate, heavy to light, and moderate to light at the second examination (1992). Compared with heavy smokers who did not reduce, the risk of lung cancer was 28% (HR = 0.72, 95% CI = 0.59-0.89) and 55% (HR = 0.45, 95% CI = 0.36 to 0.55) lower among those who reduced their smoking from heavy to moderate and from moderate to light, respectively. Reducing from moderate smoking to light smoking was also associated with lower risk of lung cancer compared with moderate smokers who did not reduce.
In a subsequent analysis using data from the Korean National Health Insurance Service-National Health Screening Cohort, Choi et al. (9) examined the effect of change in smoking habits on the risk of all-cancer incidence, smoking-related cancer, and lung cancer among 143 071 men aged 40 years or older. Smoking reduction was defined as any changes from heavy (≥20 cigarettes per day) to moderate (10-19 cigarettes per day), heavy to light (<10 cigarettes per day), or moderate to light smoking between the first (2002 and 2003) and the second examination (2004 and 2005). There was no statistically significant reduction in the risk of lung cancer associated with changing from heavy to moderate smoking (HR = 0.82, 95% CI = 0.59 to 1.15) or from heavy to light smoking (HR = 1.09, 95% CI = 0.59 to 2.00). However, compared with persistent heavy smokers, moderate smokers who reduced their smoking to less than 10 cigarettes per day had lower risks of all cancer (HR = 0.82, 95% CI = 0.72 to 0.94), smoking-related cancer (HR = 0.74, 95% CI = 0.59 to 0.93), and lung cancer (HR = 0.55, 95% CI = 0.38 to 0.79).
Unique contributions of our study include prospective assessments of cigarette smoking and a long postintervention follow-up of nearly 20 years that allowed us to capture the long-term effect of changes in smoking on lung cancer incidence. We developed a novel method for summarizing longitudinal smoking behavior that distinguishes between the proportion of time smoking and the exposure intensity while smoking. Our study also had several limitations. First, all the ATBC Study participants were male Finnish smokers aged 50-69 years with a long-term smoking history; more longitudinal studies are needed to understand the impact of changes in cigarette smoking on lung cancer incidence in other sociodemographic groups, including cohorts of younger men and women. Second, the assessment of smoking was based on self-reports, and the responses were not verified by biochemical measurements. However, self-reported smoking in a random sample of Finnish population was found to be reliable at the time of data collection; among regular smokers, 97.2% of men had serum cotinine concentrations in the active smoking range (≥10 ng/ml) (24). Third, it is possible that changes in smoking status occurred in the postintervention period, which would have influenced the risk of lung cancer. Both results from experimental and nonexperimental studies have shown that smoking reduction is associated with an increased likelihood of future smoking cessation (25,26). Nevertheless, we observed similar associations at both time intervals of follow-up in our study, suggesting that longitudinal smoking data measured between 1985 and 1993 are associated with both smoking behaviors after the intervention period and with subsequent lung cancer risk.
In conclusion, we observed evidence that smokers may meaningfully lower their risk of lung cancer by reducing the number of cigarettes they smoke per day and the proportion of time they smoke. However, our data reinforce the fact that quitting smoking completely is the most effective way for smokers to reduce their risk of lung cancer.
Data availability
The data described in this manuscript are available upon reasonable request, pending proposal approval and completion of a Data Transfer Agreement. More information can be found at the ATBC Study website (https://atbcstudy.cancer.gov/ptsa/).
Author contributions
Daniela Sarahí Gutiérrez-Torres, ScD (Methodology; Visualization; Writing—original draft; Writing—review & editing), Sungduk Kim, PhD (Formal analysis; Methodology; Validation; Writing—review & editing), Demetrius Albanes, MD (Writing—review & editing), Stephanie J. Weinstein, PhD (Writing—review & editing), Maki Inoue-Choi, PhD (Writing—review & editing), Paul S. Albert, PhD (Methodology; Supervision; Validation; Writing—review & editing), and Neal D. Freedman, PhD (Conceptualization; Methodology; Supervision; Writing—review & editing).
Funding
This work was supported by the Intramural Research Program of the National Cancer Institute.
Conflicts of interest
The authors have no potential conflicts of interest to disclose.
Acknowledgements
We thank Dr Michele Bloch for her valuable comments on our manuscript. The interpretation and reporting of these data are the sole responsibility of the authors and do not reflect the official policy of the Department of Health and Human Service, National Institutes of Health, and National Cancer Institute. The funder did not play a role in the design of the study; the collection, analysis, and interpretation of the data; the writing of the manuscript; and the decision to submit the manuscript for publication.
Results of this manuscript have been presented in part as a poster presentation at the American Association for Cancer Research 2023 Annual Meeting (Abstract 6462 https://doi.org/10.1158/1538-7445.AM2023-6462).
References
Author notes
These co-first authors contributed equally to this work.
These co-senior authors contributed equally to this work.
