Association of cigarette smoking from adolescence to middle-age with later total and cardiovascular disease mortality: The Harvard Alumni Health Study

Introduction

Evidence regarding the future health impact of smoking at younger ages is usually based on retrospective reporting of this health behaviour, which demonstrates only moderate agreement with contemporaneous reporting.¹ Studies of subsequent mortality risk with prospectively measured smoking in adolescence or early adulthood are rare and tend to focus on total mortality.^2-3 Only one⁴ has considered cause-specific mortality, in particular cardiovascular disease (CVD). Additionally, the single baseline measure of smoking in adolescence utilised in these studies is unlikely to completely capture lifetime risk, for example there is evidence that mortality risk in lifelong smokers vs. lifelong non-smokers is far greater than that estimated for current smokers vs. non-smokers at any individual time point.²

Cigarette smoking in adolescence is common and continues to rise. Policy makers require reliable estimates of risk disease burden. Accordingly, we have explored all-cause, CVD and cancer mortality associations with cigarette smoking in early adulthood through to older age in a large cohort of male US college students. To our knowledge, this duration of lifetime prospective measurement of cigarette smoking is unrivalled in the literature.

Methods

The Harvard Alumni Health Study is an ongoing cohort study of male alumni from Harvard University who entered college between 1916-1950. Cigarette smoking status was obtained at baseline during a medical examination and comparable smoking data were also reported in follow-up surveys in 1962/1966, 1977 and 1988. We explored baseline smoking associations with mortality to end of 1998 from (i) all causes, (ii) CVD, (iii) cancers related to smoking, and (iv) cancers not related to smoking, using Cox proportional hazards regression. We also examined associations of all-cause mortality with continuing smoking status based on smoking status collected at baseline and in 1962/66, 1977 and 1988. Men were defined as continuing non-smokers (non-smoker at all four time-points), continuing smokers (current smokers at all four time-points), or quitters (current smoker at baseline and non-smoker by 1988 with only one change in status). Men who changed their smoking status more than once or took up smoking during follow-up were omitted from these analyses. There were too few deaths to explore cause-specific mortality in this context.

Results

Of 33,415 men in the original cohort, 28,236 (84.5%) had data on cigarette smoking at baseline (mean age 18). Baseline characteristics of men included and excluded from analyses were very similar. Overall 10,253 (36.3%) men smoked at baseline, although the proportion of smokers varied according to the decade of interview: around a quarter of men interviewed in the 1910s were cigarette smokers and this proportion increased steadily to 40.6% in the 1940s before dropping to 34.3% in the 1950s. After a median follow-up period of 53.2 (range: 0.3 to 83.5) years, 13,704 (48.5%) men had died. Men reporting that they smoked cigarettes at baseline experienced a 30% (95% confidence interval (CI): 26%, 35%) increase in mortality from all causes (Table) compared with those who were non-smokers. Mortality for specific causes was also raised in men who were smokers at baseline. As anticipated, this was most marked for smoking-related cancers (Hazard ratio (HR); 95% CI: 1.91; 1.72, 2.12) and there was also a clear 20% (14%, 27%) increase in CVD mortality in men who smoked in early adulthood.

Table 1. Hazard ratio (95% confidence interval) for mortality in relation to baseline and continuing smoking status (The Harvard Alumni Health Study).

	Men with baseline smoking data (N=28,236)		Men with continuing smoking data (N=5,785)
	N (alive / died)	HR* (95% CI)	N (alive / died)	HR* (95% CI)
All cause
Baseline smoking
Non-smoker	9,419 / 8,564	1.00 (ref)	2,782 / 734	1.00 (ref)
Cigarette smoker	5,113 / 5,140	1.30 (1.26, 1.35)	1,767 / 502	1.36 (1.21, 1.52)
p		<0.001		<0.001
Continuing smoking†
Continuing non-smoker			2,782 / 734	1.00 (ref)
Continuing smoker			190 / 63	2.11 (1.63, 2.74)
Quit during follow-up			1,577 / 439	1.29 (1.15, 1.46)
P				<0.001
CVD ‡
Baseline smoking
Non-smoker	14,290 / 3,693	1.00 (ref)
Cigarette smoker	8,287 / 1,966	1.20 (1.14, 1.27)
p		<0.001
Smoking related cancer‡,§
Baseline smoking
Non-smoker	17,222 / 761	1.00 (ref)
Cigarette smoker	9,562 / 691	1.91 (1.72, 2.12)
p		<0.001
Non-smoking related cancer‡,§
Baseline smoking
Non-smoker	16,750 / 1,233	1.00 (ref)
Cigarette smoker	9,617 / 636	1.10 (1.00, 1.21)
p		0.06

Abbreviations: HR: hazard ratio; CI: confidence interval; CVD: cardiovascular disease;

^*Adjusted for age at baseline examination, year of baseline examination; additional adjustments for height, BMI, blood pressure, exercise, alcohol at baseline and socioeconomic status in 1988 had little impact on these associations;

^†Based on smoking at (i) baseline, (ii) 1962/66, (iii) 1977, and (iv) 1988;

^‡There were too few deaths to explore associations with continuing smoking habits for this cause;

^§Cancers considered to be related to smoking were: lung, oral cavity, nasopharynx, oropharynx, hypopharynx, nasal cavity and paranasal sinuses, larynx, oesophagus, stomach, pancreas, liver, kidney (body and pelvis), ureter, urinary bladder, uterine cervix and myeloid leukaemia; all other cancers were considered not to be related to smoking.

Analysis of continuing smoking status was based on 5,785 men with complete data on smoking at baseline, 1962/66 (mean age 42), 1977 (mean age 55), and 1988 (mean age 66). At baseline, 2,269 (39.2%) of these men were cigarette smokers and the impact of smoking was very similar to that in the full cohort. Among baseline smokers, 1,303 (57.4%) were still smoking in 1966, 573 (25.3%) in 1977, and by 1988 only 253 (11.1%) were continuing smokers. Mortality in continuing smokers was over double that in continuing non-smokers (HR; 95% CI: 2.11; 1.63, 2.74), while mortality in men who smoked at baseline but subsequently quit was higher than in continuing non-smokers but markedly lower than in those who continued to smoke (1.29; 1.15, 1.46).

Discussion

The negative health effects of cigarette smoking are well understood, yet smoking rates in young people are rising. The link between adolescent smoking and later mortality has been little examined, particularly in the context of CVD mortality. Although adolescent smoking patterns in our cohort may differ from those of contemporary adolescents, our results indicate that cigarette smoking reported directly in early adulthood has a negative impact on mortality, particularly due to CVD and smoking-related cancers, over 50 years later. This is consistent with previous evidence.^2-4 Changing smoking rates by baseline year track the increasing popularity of cigarette smoking during the first half of the 20^th century, followed by a decline in uptake as negative health effects became more widely understood in 1960s. It is interesting that almost 90% of baseline smokers had quit by 1988, reflecting increasing public health information during this period. The beneficial effects of quitting smoking, as seen previously,⁵ are clear, with the excess mortality in those who quit much reduced in comparison to those who continued to smoke. Health education should be targeted at preventing smoking uptake in young people and encouraging current smokers to quit.