Impact of Smoking Cessation Advice on Future Smoking Behavior, Morbidity, and Mortality: Up to 40 Years of Follow-up of the First Randomized Controlled Trial of a General Population Sample FREE

While there is evidence from cohort studies that people who stop smoking have a longer life expectancy relative to their nonabstaining counterparts,¹ interpretation of these observational data is often complicated by confounding. That is, the characteristics of people who quit could be different in relevant, unmeasured ways from those who do not. Extended follow-up of participants in a randomized controlled trial, which should circumvent this problem, is scarce and limited to samples whose members already had a smoking-related disease.^2- 3 Ten-year⁴ and 20-year⁵ follow-ups of study members in the Whitehall Smoking Cessation Study—to our knowledge, the only trial of an apparently healthy population with future mortality ascertainment—found little apparent benefit for life expectancy despite a reduction in cough, phlegm production, and rate of pulmonary decline at the 3-year follow-up.⁶ We present herein the results for smoking behavior after 3 decades of follow-up and mortality risk after 4 decades. With most of the trial participants now deceased, this provides an unusual opportunity to compare the gain in life years with what would be predicted from cohort studies.

Participants were drawn from men who took part in the Whitehall study of male civil servants in London, England, a cohort study established between 1967 and 1970 to identify risk factors for cardiovascular disease. Described in detail elsewhere,⁷ the study recruited 16 016 men aged 40 to 59 years. Using an early version of the Framingham algorithm, we targeted 1500 cigarette smokers with high cardiovascular disease risk for recruitment in the trial. After exclusions, chiefly for those receiving medical care, 1445 entered the trial and were randomized to either the intervention (714 men) or “normal care” arm (731 men).

Men in the intervention group were invited for a 15-minute consultation with one of the study's clinical staff at which the risks of smoking were outlined at length, with some discussion of the practicalities of smoking cessation. In a follow-up appointment 1 week later, each member of the intervention group was given a smoking record card to be completed daily over the following 3 weeks and mailed back; a supportive response from study personnel followed. Further 15-minute interviews took place at the research center at 10 weeks and 6 months.

The reference group of men receiving “normal care,” who did not participate in the 6-month antismoking intervention, were told at baseline they were continuing to participate in the main observational study. In 1997, approximately 30 years after the baseline screening, surviving study members were mailed a questionnaire with enquiries about their smoking behavior and health.⁸

Baseline characteristics of men in each arm of the trial were very similar (results available on request).⁶ At the 1-year follow-up, the prevalence of self-reported abstinence from smoking in the intervention group (39%) was substantially higher than in the normal care group (9%).⁶ Corresponding figures for a 3-year follow-up were 36% and 14%, respectively.⁶ At resurvey almost 30 years following the intervention, most survivors in both the intervention and control groups had stopped smoking (81% vs 79%, respectively) (eTable).

The Table shows that after 40 years of mortality surveillance the hazard ratio for overall mortality was slightly lower for the intervention than for the control group. Although this difference was not statistically significant, it corresponds with an estimated 0.4 life-years gained. From the British Doctors study¹ it is possible to estimate that an intervention that yielded an increased cessation rate at the 12-month follow-up of 30% (39% vs 9%) in smokers with the age distribution represented in this study would be expected to yield approximately 0.8 life-years.⁹ The Table also shows lower hazard ratios for many specific causes of death in the intervention group compared with the control group, though none were statistically significant at conventional levels. As might be expected, the largest putative effect size was for lung cancer.

Although this study had low power to detect a long-term effect on mortality of the size predicted from the British Doctors study,¹ to our knowledge, these data provide the best estimate of such an intervention to date. This estimated gain of 0.4 life-years was approximately half what might be expected from the British Doctors cohort study.¹ In addition to the low sample size, several factors may have militated against showing an effect of the kind predicted. First, there was no biochemical verification of abstinence, so at least some of the “effect” may have resulted from misreporting by smokers in the intervention group who did not want to admit that they were not able to stop.¹⁰ Second, brief advice without pharmacological support tends to lead to cessation in lighter, less nicotine-dependent smokers, with heavier smokers requiring more support.¹¹ These light smokers would have been exposed to less-than-average risk from their smoking.