Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Mortality reduction from quitting smoking in Hong Kong: population-wide proportional mortality study

Mortality reduction from quitting smoking in Hong Kong: population-wide proportional mortality study Abstract Background The effects of smoking cessation might be different in different populations. Proportional mortality studies of all deaths, relating the certified cause to retrospectively determined smoking habits, have helped assess the hazards of smoking in Hong Kong, and further analyses can help assess the effects of prolonged cessation (although not of recent cessation, as life-threatening disease can itself cause cessation, particularly in old age). Methods The LIMOR study sought the certified causes of all deaths in 1998, and interviewed 81% of families at death registries to determine the decedent’s smoking history. Cases were deaths from pre-defined diseases of interest (N = 15 356); controls were deaths from pre-defined non-smoking-related diseases (N = 5023). Case vs control odds ratios for ex-smokers vs smokers were calculated by age-, sex- and education-standardized logistic regression. These are described as mortality rate ratios (RRs), with a group-specific confidence interval (CI). Results For the aggregate of all deaths from any of the diseases of interest at ages 35-69 years, the RRs for current smoking, quitting 0-4, 5-9 or 10+ years ago and never-smoking were, respectively, RR = 1 (95% CI 0.86-1.17), 0.91 (0.73-1.14), 0.71 (0.49-1.02), 0.66 (0.50-0.87) and 0.43 (0.37-0.48). Younger age of quitting (25-44 or 45-64) appeared to be associated with greater protection: RR = 0.58 (0.38-0.88) and 0.71 (0.54-0.93), respectively. These patterns were less clear at older ages, particularly for death from emphysema. Conclusions Longer durations of smoking cessation are associated with progressively lower mortality rates from the diseases of interest. For sustainable monitoring of tobacco-attributed mortality, approximate years since last smoked should be recorded during death registration. Smoking cessation, proportional mortality study, epidemiology, death certificate, Hong Kong China, health benefit Key Messages There is limited evidence from low- and middle-income countries (LMICs) about the effects of cessation and, in populations, comparison of current mortality rates in smokers and ex-smokers would underestimate the benefits of quitting. Proportional mortality studies could provide a timely low-cost alternative to prospective cohort studies, at least for assessing the effects of long-term cessation, in addition to the harms of smoking. Taking the advantage of a proportional mortality study in Hong Kong, which is the most urbanized and Westernized Chinese city and where the tobacco epidemic reached its peak about 20 years earlier than in the Chinese mainland we examined the effects of cessation by comparing the proportions of ex-smokers and current-smokers among those dying from particular diseases. The benefits of quitting and of longer durations of smoking cessation, associated with progressively lower mortality rates of the diseases of interest, can be observed by using the proportional mortality study design. Introduction Although smoking-attributed mortality is decreasing steadily among men in most developed countries, it is still increasing in many developing countries.1,2 In China, with 20% of the world’s population consuming 40% of the world’s cigarettes, smoking-attributed deaths will continue to increase over the next few decades, unless there is widespread cessation.3 Prospective studies from developed countries provide strong quantitative evidence that stopping smoking works.4–7 For example, UK smokers who stop before age 40 (preferably well before 40) avoid over 90% of the excess mortality rates among continuing smokers.8 As yet, there is limited evidence from low- and middle-income countries (LMICs) about the effects of cessation,9 and in populations where the risks among smokers are still rising, comparison of current mortality rates among smokers and ex-smokers would underestimate the benefits of quitting. As the epidemic of death from tobacco is at a more advanced stage in Hong Kong than in the Chinese mainland studies in Hong Kong could be particularly informative. The ideal would be a large prospective study that carefully limits the effects of reverse causality, whereby life-threatening disease may make smokers stop [for chronic obstructive pulmonary disease (COPD), this can happen many years before death, artificially reducing the death rate among the smokers and increasing it substantially among the ex-smokers]. However, cohort studies are difficult to conduct, expensive and take many years to deliver results,10–13 so retrospective studies are also needed, although researchers have to consider carefully how reverse causality might distort their findings, particularly for COPD. Proportional mortality studies that involve only dead subjects (comparing the proportions of ever-smokers and never-smokers among those dying from particular diseases) could provide a timely low-cost alternative, at least for assessing the effects of long-term cessation. Proportional mortality studies have been used to investigate the harms of smoking, mostly in developing countries, but the benefits of quitting were not reported,14–19 partly for fear of producing results that are substantially biased by reverse causality, and some reports did not mention whether quitting was asked about. The Hong Kong Lifestyle and Mortality (HK LIMOR) study sought from family members information about the previous smoking habits of all who died in Hong Kong in 1998, including information on how long ago the dead person had stopped smoking; but the main report combined the ex-smokers with those who had continued smoking until the last year of their life (so it compared those who had ever smoked vs those who had never smoked).20 We now report its findings according to the duration of quitting. Methods The study methods and results on smoking and various other factors associated have been reported previously (Supplementary Part I, available at IJE online).20–27 We have found the definition of dead cases and controls in the most recent proportional mortality case-control study most appropriate for the HK LIMOR study,9 with cases being deaths from diseases that could be caused by smoking, and controls being deaths from all non-smoking related causes (Supplementary Table S1, available at IJE online).19 Details of the assessment of quitting are in Supplementary Part II (Supplementary materials, available at IJE online). Present analysis only used deaths with additional quitting data collected in the second half of all deaths in 1998. To compare our previous results for all deaths on the harms of smoking (dead cases vs living controls),20 we also included the results comparing dead cases with dead controls (Supplementary Table S2, available at IJE online). Statistical analysis Case vs control odds ratios for the quitting status were calculated by unconditional logistic regression, adjusted for 5-year age groups, sex and education, and are described as mortality rate ratios (RRs, calculated as odds ratios). To assess the association of stopping smoking with smoking-attributed mortality by the age of quitting, the RR was calculated for never smokers, and quitters who stopped smoking at the ages of 25-44 and 45-64 years, compared with current smokers. The group-specific confidence interval (CI) for the current smokers’ RR of 1.00 was calculated to reflect the variance of the log risk in the current smokers, using Plummer’s method.28 To select the appropriate age range, mortality RRs were calculated for main cause-specific deaths by smoking history (ever vs never smokers) in all deaths (Supplementary Tables S3 and S4, available at IJE online). Smoking was expected to cause few deaths before age 35, and cause of death information can be unreliable in old age, so subjects aged 35-84 at death were included. As most of those killed by smoking would otherwise have survived beyond age 70, but a minority would have died by 70, we stratified the age into two groups (middle age: 35-69 or old age: 70-84 years) for all the analyses. All analyses were conducted using R 3.3.1. Results Mortality by duration of quitting There were 9772 male and 5584 female cases, and 2503 male and 2520 female controls. The cases and controls had similar demographic characteristics, with a mean [standard deviation (SD)] age of 70.2 (10.7) and 68.2 (12.0) years, respectively (Supplementary Table S5, available at IJE online). Table 1 shows that in middle age (35-69 years), all the RRs among long-term quitters (vs current smokers) were lower: RR 0.50 (95% CI 0.33-0.74) for lung cancer, 0.92 (0.66-1.28) for smoking-related cancer, 0.40 (0.27-0.61) for cardiovascular disease (CVD) and 0.81 (0.49-1.33) for COPD. But in old age (70-84 years), the RR for COPD was higher (1.40, 1.11-1.76). Quitters who had stopped for 5-9 years also had lower lung cancer, smoking-related cancer and CVD mortality risks, but higher COPD mortality risks compared with current smokers. New quitters who had stopped for less than 5 years had lower CVD mortality risks: RR 0.56 (95% CI 0.41-0.78) in middle age and 0.90 (0.68-1.18) in old age. However, the RRs for lung cancer and COPD among new quitters (vs current smokers) in old age were higher, probably because of ill-quitter effect. Table 1. Duration of smoking cessation and mortality from four smoking-related diseases—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers. Deaths from selected diseases vs deaths from control diseases, at ages 35-69 or 70-84 years (both sexes) Current smokers Duration of quitting Trend P-value* Never smokers <5 years 5-9 years ≥10 years Lung cancer Age 35-69 1.00 (231) 1.23 (141) 0.66 (28) 0.50 (40) <0.001 0.22 (187) 0.82-1.21 0.95-1.60 0.40-1.08 0.33-0.74 0.18-0.28 Age 70-84 1.00 (228) 1.21 (132) 0.79 (47) 0.62 (121) 0.001 0.19 (174) 0.83-1.21 0.93-1.59 0.53-1.17 0.49-0.78 0.16-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 1.11 (151) 0.72 (36) 0.92 (84) 0.55 0.58 (471) 0.83-1.20 0.88-1.47 0.45-1.15 0.66-1.28 0.50-0.68 Age 70-84 1.00 (166) 0.93 (73) 0.66 (29) 0.67 (95) 0.011 0.60 (318) 0.81-1.23 0.69-1.27 0.42-1.05 0.52-0.86 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.56 (66) 0.59 (28) 0.40 (37) <0.001 0.46 (338) 0.82-1.22 0.41-0.78 0.36-0.98 0.27-0.61 0.38-0.54 Age 70-84 1.00 (266) 0.90 (111) 0.72 (49) 0.83 (197) 0.15 0.76 (785) 0.83-1.20 0.68-1.18 0.48-1.06 0.67-1.02 0.67-0.86 COPD Age 35-69 1.00 (66) 0.72 (24) 1.35 (19) 0.81 (22) 0.79 0.26 (46) 0.75-1.34 0.46-1.14 0.76-2.42 0.49-1.33 0.18-0.38 Age 70-84 1.00 (104) 1.34 (67) 1.59 (42) 1.40 (134) 0.050 0.26 (106) 0.79-1.27 0.97-1.85 1.05-2.41 1.11-1.76 0.21-0.33 Current smokers Duration of quitting Trend P-value* Never smokers <5 years 5-9 years ≥10 years Lung cancer Age 35-69 1.00 (231) 1.23 (141) 0.66 (28) 0.50 (40) <0.001 0.22 (187) 0.82-1.21 0.95-1.60 0.40-1.08 0.33-0.74 0.18-0.28 Age 70-84 1.00 (228) 1.21 (132) 0.79 (47) 0.62 (121) 0.001 0.19 (174) 0.83-1.21 0.93-1.59 0.53-1.17 0.49-0.78 0.16-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 1.11 (151) 0.72 (36) 0.92 (84) 0.55 0.58 (471) 0.83-1.20 0.88-1.47 0.45-1.15 0.66-1.28 0.50-0.68 Age 70-84 1.00 (166) 0.93 (73) 0.66 (29) 0.67 (95) 0.011 0.60 (318) 0.81-1.23 0.69-1.27 0.42-1.05 0.52-0.86 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.56 (66) 0.59 (28) 0.40 (37) <0.001 0.46 (338) 0.82-1.22 0.41-0.78 0.36-0.98 0.27-0.61 0.38-0.54 Age 70-84 1.00 (266) 0.90 (111) 0.72 (49) 0.83 (197) 0.15 0.76 (785) 0.83-1.20 0.68-1.18 0.48-1.06 0.67-1.02 0.67-0.86 COPD Age 35-69 1.00 (66) 0.72 (24) 1.35 (19) 0.81 (22) 0.79 0.26 (46) 0.75-1.34 0.46-1.14 0.76-2.42 0.49-1.33 0.18-0.38 Age 70-84 1.00 (104) 1.34 (67) 1.59 (42) 1.40 (134) 0.050 0.26 (106) 0.79-1.27 0.97-1.85 1.05-2.41 1.11-1.76 0.21-0.33 Numbers of cases (i.e. deaths from the disease of interest) are given in brackets after each RR (in bold). Numbers of controls in current/ex < 5/ex 5-9/ex 10 years/never smokers are 194/96/36/65/744 at ages 35-69 years, and 200/95/54/175/826 at ages 70-84 years. CI, group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. Other smoking-related cancers: upper aerodigestive, stomach, liver, pancreas, cervix, urinary and myeloid leukaemia (ICD-9 140-51, 155-7, 160-1, 179-80, 184, 188-9, 205). Cardiovascular disease: stroke and ischaemic heart disease (ICD-9 430-8, 410-4, 440-8). COPD: chronic obstructive pulmonary disease (ICD-9 415-7, 490-6); heterogeneity by sex was P < 0.05 only for COPD. * Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, each would have yielded P < 0.05. Table 1. Duration of smoking cessation and mortality from four smoking-related diseases—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers. Deaths from selected diseases vs deaths from control diseases, at ages 35-69 or 70-84 years (both sexes) Current smokers Duration of quitting Trend P-value* Never smokers <5 years 5-9 years ≥10 years Lung cancer Age 35-69 1.00 (231) 1.23 (141) 0.66 (28) 0.50 (40) <0.001 0.22 (187) 0.82-1.21 0.95-1.60 0.40-1.08 0.33-0.74 0.18-0.28 Age 70-84 1.00 (228) 1.21 (132) 0.79 (47) 0.62 (121) 0.001 0.19 (174) 0.83-1.21 0.93-1.59 0.53-1.17 0.49-0.78 0.16-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 1.11 (151) 0.72 (36) 0.92 (84) 0.55 0.58 (471) 0.83-1.20 0.88-1.47 0.45-1.15 0.66-1.28 0.50-0.68 Age 70-84 1.00 (166) 0.93 (73) 0.66 (29) 0.67 (95) 0.011 0.60 (318) 0.81-1.23 0.69-1.27 0.42-1.05 0.52-0.86 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.56 (66) 0.59 (28) 0.40 (37) <0.001 0.46 (338) 0.82-1.22 0.41-0.78 0.36-0.98 0.27-0.61 0.38-0.54 Age 70-84 1.00 (266) 0.90 (111) 0.72 (49) 0.83 (197) 0.15 0.76 (785) 0.83-1.20 0.68-1.18 0.48-1.06 0.67-1.02 0.67-0.86 COPD Age 35-69 1.00 (66) 0.72 (24) 1.35 (19) 0.81 (22) 0.79 0.26 (46) 0.75-1.34 0.46-1.14 0.76-2.42 0.49-1.33 0.18-0.38 Age 70-84 1.00 (104) 1.34 (67) 1.59 (42) 1.40 (134) 0.050 0.26 (106) 0.79-1.27 0.97-1.85 1.05-2.41 1.11-1.76 0.21-0.33 Current smokers Duration of quitting Trend P-value* Never smokers <5 years 5-9 years ≥10 years Lung cancer Age 35-69 1.00 (231) 1.23 (141) 0.66 (28) 0.50 (40) <0.001 0.22 (187) 0.82-1.21 0.95-1.60 0.40-1.08 0.33-0.74 0.18-0.28 Age 70-84 1.00 (228) 1.21 (132) 0.79 (47) 0.62 (121) 0.001 0.19 (174) 0.83-1.21 0.93-1.59 0.53-1.17 0.49-0.78 0.16-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 1.11 (151) 0.72 (36) 0.92 (84) 0.55 0.58 (471) 0.83-1.20 0.88-1.47 0.45-1.15 0.66-1.28 0.50-0.68 Age 70-84 1.00 (166) 0.93 (73) 0.66 (29) 0.67 (95) 0.011 0.60 (318) 0.81-1.23 0.69-1.27 0.42-1.05 0.52-0.86 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.56 (66) 0.59 (28) 0.40 (37) <0.001 0.46 (338) 0.82-1.22 0.41-0.78 0.36-0.98 0.27-0.61 0.38-0.54 Age 70-84 1.00 (266) 0.90 (111) 0.72 (49) 0.83 (197) 0.15 0.76 (785) 0.83-1.20 0.68-1.18 0.48-1.06 0.67-1.02 0.67-0.86 COPD Age 35-69 1.00 (66) 0.72 (24) 1.35 (19) 0.81 (22) 0.79 0.26 (46) 0.75-1.34 0.46-1.14 0.76-2.42 0.49-1.33 0.18-0.38 Age 70-84 1.00 (104) 1.34 (67) 1.59 (42) 1.40 (134) 0.050 0.26 (106) 0.79-1.27 0.97-1.85 1.05-2.41 1.11-1.76 0.21-0.33 Numbers of cases (i.e. deaths from the disease of interest) are given in brackets after each RR (in bold). Numbers of controls in current/ex < 5/ex 5-9/ex 10 years/never smokers are 194/96/36/65/744 at ages 35-69 years, and 200/95/54/175/826 at ages 70-84 years. CI, group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. Other smoking-related cancers: upper aerodigestive, stomach, liver, pancreas, cervix, urinary and myeloid leukaemia (ICD-9 140-51, 155-7, 160-1, 179-80, 184, 188-9, 205). Cardiovascular disease: stroke and ischaemic heart disease (ICD-9 430-8, 410-4, 440-8). COPD: chronic obstructive pulmonary disease (ICD-9 415-7, 490-6); heterogeneity by sex was P < 0.05 only for COPD. * Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, each would have yielded P < 0.05. Figure 1 shows declining trends of RRs for all deaths of interest from current smoking to quitting for less than 5, 5-9 and 10+ years, and never smoking: RR 1.00 (95% CI 0.86-1.17), 0.91 (0.73-1.14), 0.71 (0.49-1.02), 0.66 (0.50-0.87) and 0.43 (0.37-0.48) in middle age (P for linear trend: 0.006); and 1.00 (0.86-1.16), 1.06 (0.85-1.32), 0.86 (0.64-1.16), 0.87 (0.74-1.03) and 0.55 (0.50-0.61) in old age (P for linear trend: 0.19). Figure 1 View largeDownload slide Duration of smoking cessation and mortality from all deaths of interest—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers Deaths from selected diseases vs deaths from control diseases, at 35–69 or 70–84 years (both sexes). Numbers of cases (ie, deaths from the disease of interest) are given in brackets after each RR. Number of controls in current/ ex <5/ ex 5–9/ ex ≥10 years/ never smokers are 194/96/36/65/744 at ages 35–69 years, and 200/95/54/175/826 at ages 70–84 years. CI: group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variances of the log risk in only that one group. *Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, both would have yielded p<0.001. Figure 1 View largeDownload slide Duration of smoking cessation and mortality from all deaths of interest—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers Deaths from selected diseases vs deaths from control diseases, at 35–69 or 70–84 years (both sexes). Numbers of cases (ie, deaths from the disease of interest) are given in brackets after each RR. Number of controls in current/ ex <5/ ex 5–9/ ex ≥10 years/ never smokers are 194/96/36/65/744 at ages 35–69 years, and 200/95/54/175/826 at ages 70–84 years. CI: group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variances of the log risk in only that one group. *Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, both would have yielded p<0.001. Mortality by age of quitting Figure 2 also shows a clear declining trend of mortality risk, from current smoking to quitting at the ages of 45-64 and 25-44 years and never smoking, for all deaths of interest in middle age: 1.00 (0.86-1.17), 0.71 (0.54-0.93), 0.58 (0.38-0.88) and 0.43 (0.37-0.49), (P for linear trend: 0.003). However, subjects in old age (70-84 years) who stopped at younger age (25-44 years) had higher mortality risks: 1.47 (0.51-4.22), but this was based on only 31 subjects (27 cases and four controls), so it is not statistically reliable. Figure 2 View largeDownload slide Age at smoking cessation and mortality from all deaths of interest—mortality rate rations (RR, 95% CI) comparing ex- or never-somkers vs current smokers To limit reverse causality, analyses exclude those who stopped <5 years ago, or at age>65. They compare deaths from selected diseases vs deaths from control diseases, at ages 35–69 or 70–84 years (both sexes) Numbers of cases (ie, deaths from the disease of interest) are given in brackets after each RR. Numbers of controls in current/ ex 45–64/ ex 25–44 years/ never smokers are 194/137/41/744 at ages 35–69 years, and 200/113/4/826 at ages 70–84 years. CI: group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. *Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, both would have yielded p<0.001. Figure 2 View largeDownload slide Age at smoking cessation and mortality from all deaths of interest—mortality rate rations (RR, 95% CI) comparing ex- or never-somkers vs current smokers To limit reverse causality, analyses exclude those who stopped <5 years ago, or at age>65. They compare deaths from selected diseases vs deaths from control diseases, at ages 35–69 or 70–84 years (both sexes) Numbers of cases (ie, deaths from the disease of interest) are given in brackets after each RR. Numbers of controls in current/ ex 45–64/ ex 25–44 years/ never smokers are 194/137/41/744 at ages 35–69 years, and 200/113/4/826 at ages 70–84 years. CI: group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. *Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, both would have yielded p<0.001. Table 2 shows mortality RRs for the other four disease groupings (causable by smoking) by age of quitting vs current smoking. Subjects in middle age (35-69 years) who stopped smoking at younger age (25-44 years) had non-significantly (with overlapping 95% CIs) lower mortality risks than those who stopped smoking at older age (45-64 years) for lung cancer (RR 0.41, 95% CI 0.22-0.78 vs 0.61, 0.42-0.88), CVD (0.36, 0.18-0.70 vs 0.51, 0.35-0.73), and all deaths of interest (0.58, 0.38-0.88 vs 0.71, 0.54-0.93, Figure 2), respectively. Few quitters stopped before age 45, so the estimates of quitting at the age of 25-44 years in old age are not statistically reliable. Among subjects who had stopped at age 45-64 years, the RRs for lung cancer, smoking-related cancer and CVD were lower, but those for COPD were still substantially higher than current smokers: RR 1.12 (95% CI 0.73-1.72) in middle age and 1.59 (1.20-2.11) in old age. Table 2. Age at smoking cessation and mortality from four smoking-related diseases—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers. To limit reverse causality, analyses exclude those who stopped < 5 years ago, or at age> 65. They compare deaths from selected diseases vs deaths from control diseases, at ages 35-69 or 70-84 years (both sexes) Current smokers Age at quitting Trend P-value* Never smokers 45-64 years 25-44 years Lung cancer Age 35-69 1.00 (231) 0.61 (53) 0.41 (14) < 0.001 0.22 (187) 0.83-1.21 0.42-0.88 0.22-0.78 0.17-0.27 Age 70-84 1.00 (228) 0.66 (86) 0.78 (4) 0.015 0.19 (174) 0.83-1.21 0.50-0.88 0.19-3.16 0.15-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 0.84 (81) 0.84 (38) 0.35 0.58 (471) 0.83-1.20 0.60-1.17 0.52-1.36 0.50-0.69 Age 70-84 1.00 (166) 0.71 (69) 1.00 (4) 0.08 0.60 (318) 0.81-1.23 0.53-0.97 0.25-4.07 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.51 (53) 0.36 (12) < 0.001 0.46 (338) 0.82-1.21 0.35-0.73 0.18-0.70 0.39-0.55 Age 70-84 1.00 (266) 0.90 (133) 2.08 (10) 0.82 0.76 (785) 0.83-1.20 0.69-1.15 0.65-6.66 0.66-0.87 COPD Age 35-69 1.00 (66) 1.12 (36) 0.60 (5) 0.78 0.26 (46) 0.75-1.32 0.73-1.72 0.23-1.60 0.18-0.38 Age 70-84 1.00 (104) 1.59 (92) 2.83 (6) 0.008 0.26 (106) 0.79-1.27 1.20-2.11 0.74-10.8 0.20-0.33 Current smokers Age at quitting Trend P-value* Never smokers 45-64 years 25-44 years Lung cancer Age 35-69 1.00 (231) 0.61 (53) 0.41 (14) < 0.001 0.22 (187) 0.83-1.21 0.42-0.88 0.22-0.78 0.17-0.27 Age 70-84 1.00 (228) 0.66 (86) 0.78 (4) 0.015 0.19 (174) 0.83-1.21 0.50-0.88 0.19-3.16 0.15-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 0.84 (81) 0.84 (38) 0.35 0.58 (471) 0.83-1.20 0.60-1.17 0.52-1.36 0.50-0.69 Age 70-84 1.00 (166) 0.71 (69) 1.00 (4) 0.08 0.60 (318) 0.81-1.23 0.53-0.97 0.25-4.07 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.51 (53) 0.36 (12) < 0.001 0.46 (338) 0.82-1.21 0.35-0.73 0.18-0.70 0.39-0.55 Age 70-84 1.00 (266) 0.90 (133) 2.08 (10) 0.82 0.76 (785) 0.83-1.20 0.69-1.15 0.65-6.66 0.66-0.87 COPD Age 35-69 1.00 (66) 1.12 (36) 0.60 (5) 0.78 0.26 (46) 0.75-1.32 0.73-1.72 0.23-1.60 0.18-0.38 Age 70-84 1.00 (104) 1.59 (92) 2.83 (6) 0.008 0.26 (106) 0.79-1.27 1.20-2.11 0.74-10.8 0.20-0.33 Numbers of cases (i.e. deaths from the disease of interest) are given in brackets after each RR (in bold). Numbers of controls in current/ex 45–64/ex 25–44 years/never smokers are 194/137/41/744 at ages 35–69 years, and 200/113/4/826 at ages 70–84 years. CI, group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. Other smoking-related cancers: upper aerodigestive, stomach, liver, pancreas, cervix, urinary and myeloid leukaemia (ICD-9 140-51, 155-7, 160-1, 179-80, 184, 188-9, 205). Cardiovascular disease: stroke and ischaemic heart disease (ICD-9 430-8, 410-4, 440-8). COPD: chronic obstructive pulmonary disease (ICD-9 415-7, 490-6); heterogeneity was P < 0.05 by sex only for COPD. * Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, each would have yielded P < 0.01. Table 2. Age at smoking cessation and mortality from four smoking-related diseases—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers. To limit reverse causality, analyses exclude those who stopped < 5 years ago, or at age> 65. They compare deaths from selected diseases vs deaths from control diseases, at ages 35-69 or 70-84 years (both sexes) Current smokers Age at quitting Trend P-value* Never smokers 45-64 years 25-44 years Lung cancer Age 35-69 1.00 (231) 0.61 (53) 0.41 (14) < 0.001 0.22 (187) 0.83-1.21 0.42-0.88 0.22-0.78 0.17-0.27 Age 70-84 1.00 (228) 0.66 (86) 0.78 (4) 0.015 0.19 (174) 0.83-1.21 0.50-0.88 0.19-3.16 0.15-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 0.84 (81) 0.84 (38) 0.35 0.58 (471) 0.83-1.20 0.60-1.17 0.52-1.36 0.50-0.69 Age 70-84 1.00 (166) 0.71 (69) 1.00 (4) 0.08 0.60 (318) 0.81-1.23 0.53-0.97 0.25-4.07 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.51 (53) 0.36 (12) < 0.001 0.46 (338) 0.82-1.21 0.35-0.73 0.18-0.70 0.39-0.55 Age 70-84 1.00 (266) 0.90 (133) 2.08 (10) 0.82 0.76 (785) 0.83-1.20 0.69-1.15 0.65-6.66 0.66-0.87 COPD Age 35-69 1.00 (66) 1.12 (36) 0.60 (5) 0.78 0.26 (46) 0.75-1.32 0.73-1.72 0.23-1.60 0.18-0.38 Age 70-84 1.00 (104) 1.59 (92) 2.83 (6) 0.008 0.26 (106) 0.79-1.27 1.20-2.11 0.74-10.8 0.20-0.33 Current smokers Age at quitting Trend P-value* Never smokers 45-64 years 25-44 years Lung cancer Age 35-69 1.00 (231) 0.61 (53) 0.41 (14) < 0.001 0.22 (187) 0.83-1.21 0.42-0.88 0.22-0.78 0.17-0.27 Age 70-84 1.00 (228) 0.66 (86) 0.78 (4) 0.015 0.19 (174) 0.83-1.21 0.50-0.88 0.19-3.16 0.15-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 0.84 (81) 0.84 (38) 0.35 0.58 (471) 0.83-1.20 0.60-1.17 0.52-1.36 0.50-0.69 Age 70-84 1.00 (166) 0.71 (69) 1.00 (4) 0.08 0.60 (318) 0.81-1.23 0.53-0.97 0.25-4.07 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.51 (53) 0.36 (12) < 0.001 0.46 (338) 0.82-1.21 0.35-0.73 0.18-0.70 0.39-0.55 Age 70-84 1.00 (266) 0.90 (133) 2.08 (10) 0.82 0.76 (785) 0.83-1.20 0.69-1.15 0.65-6.66 0.66-0.87 COPD Age 35-69 1.00 (66) 1.12 (36) 0.60 (5) 0.78 0.26 (46) 0.75-1.32 0.73-1.72 0.23-1.60 0.18-0.38 Age 70-84 1.00 (104) 1.59 (92) 2.83 (6) 0.008 0.26 (106) 0.79-1.27 1.20-2.11 0.74-10.8 0.20-0.33 Numbers of cases (i.e. deaths from the disease of interest) are given in brackets after each RR (in bold). Numbers of controls in current/ex 45–64/ex 25–44 years/never smokers are 194/137/41/744 at ages 35–69 years, and 200/113/4/826 at ages 70–84 years. CI, group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. Other smoking-related cancers: upper aerodigestive, stomach, liver, pancreas, cervix, urinary and myeloid leukaemia (ICD-9 140-51, 155-7, 160-1, 179-80, 184, 188-9, 205). Cardiovascular disease: stroke and ischaemic heart disease (ICD-9 430-8, 410-4, 440-8). COPD: chronic obstructive pulmonary disease (ICD-9 415-7, 490-6); heterogeneity was P < 0.05 by sex only for COPD. * Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, each would have yielded P < 0.01. Discussion The purpose of the HK LIMOR study was to estimate the smoking-attributed mortality in Hong Kong, where the tobacco epidemic has reached a fairly advanced stage,20 and the present analysis is first to investigate the health benefits of quitting using proportional mortality study design. We found lower RRs of cause-specific mortality associated with longer duration of quitting and quitting at younger age. These benefits have not been reported in previous proportional mortality studies, probably because the information on quitting (or ex-smokers) was unavailable. Reverse causality, as in many cohort studies, was observed in the present analysis for COPD, which is most subject to ill-quitter effect.29 COPD typically develops among smokers with slowly worsening respiratory symptoms over many years, during which some may quit smoking when necessitated by their worsening condition. Thus, ill-quitter effect would be more common for COPD than cardiovascular disease, especially among older people with longer duration of smoking and symptoms, and lower ability to recover.30 Indeed, in the present analysis, reverse causality was observed for COPD in middle-aged subjects who had quit for 5-9 years, and in old age regardless of duration of quitting. To clarify this issue, collecting information on the reason for quitting is essential and recommended. To study the benefits of quitting in proportional mortality studies, differential misclassification of quitting between deaths and healthy subjects may occur using living controls. Indeed, in the HK LIMOR study, we found that the proportion of ex-smokers (or quitters) in all the deaths was higher than that in healthy subjects; in particular, the proportion in dead cases (deaths from diseases causable by smoking) was much higher than that in living controls.9 It is plausible that dead subjects were more likely to have quit due to ill health than healthy living subjects. We took advantage of the strength of the definition of dead cases and dead controls in the most recent mortality case-control study, which is that any ill-quitter effect in dead cases should be similar to that in dead controls, to examine the benefits of quitting. As, however, quitting must have reduced some risks from certain deaths in the dead control group (for example, some of those from diabetes and colorectal cancer),31 the present analyses may have slightly underestimated the benefits of quitting. Indeed, in old age (70-84 years), the RRs among ex-smokers are slightly smaller than those in our previous prospective cohort study of older Chinese in Hong Kong.12 However, in middle age (35-69 years), the lower RRs (vs current smokers) among all the quitters in Table 1 were consistently observed for all deaths of interest and CVD, particularly among new quitters who had stopped smoking for less than 5 years (reduction of CVD mortality: 44%, 95% CI 22-59%), which are consistent with the U.S. Surgeon General's Reports on Smoking and Tobacco Use (USSG) Report 2010, that rapid risk reduction of vascular mortality by about 50% can be observed after quitting for 1 year. To help monitor the tobacco epidemic, the methods of proportional mortality study should be used in any other populations where death registration is organized centrally. As interviewing was done in death registries, the HK LIMOR study had a high coverage of all deaths (81%), and a short time interval between death and interview, reducing recall error. Moreover, the analytical strategies in the present study using deaths from smoking-related causes as cases and those from non-smoking related causes as controls, could be used to assess the benefits of quitting, in addition to the harms of smoking. Several limitations should be considered. First, the definitions of cases and controls in previous proportional mortality studies varied. For example, some diseases such as breast cancer and colorectal cancer are considered to be causally related to smoking by the 2014 USSG Report, but were defined as controls in our study. Smoking was associated with reduced risk of Parkinson’s disease, ulcerative colitis and endometrial cancer, but we analysed them as cases.31 Nonetheless, there is no consensus of the definitions of cases and controls in mortality case-control studies.9 Further research is warranted, by collaborating all proportional mortality studies in the world to determine the most appropriate definition for studying smoking or other factors.32 The HK LIMOR study was conducted in 1998 to investigate the relation between smoking and mortality. The results might not be generalized to the present situation in Hong Kong, but should be relevant to show the great benefits of stopping smoking in the Chinese mainland. A new mortality case-control study in Hong Kong and in the Chinese mainland (or any other LMICs) is warranted. To monitor the tobacco epidemic routinely, a new method of the proportional mortality study for collecting data is recommended, as in the South Africa’s and the Tianjin’s proportional mortality studies, that the smoking status of the deceased is routinely recorded on the new death certificate for long-term sustainable monitoring.16,33 The tobacco epidemic in Hong Kong (the most urbanized and Westernized Chinese city) reached its peak about 20 years earlier than in the Chinese mainland, but about 20 years later than in developed Western countries.12 Hong Kong has been a forewarning model for the Chinese mainland.20 The mortality relative risk has already reached 3 in the UK, USA and Australia, which could be followed by the Chinese mainland and other LMICs undergoing rapid economic development in the next few decades. Stopping smoking is one of the most practicable ways to avoid a large proportion of smoking-attributable deaths, particularly premature deaths (age 35-69).34 Quitting, however, is uncommon in the Chinese mainland and other LMICs. Ex-smokers are far fewer than current smokers in most LMICs, particularly those who stopped smoking at young age (before 45 years). However, with increasing smoking cessation from more effective tobacco-control measures in developing regions,2 the applicability of proportional mortality studies should expand. Moreover, many higher-income countries also have no or limited prospective data showing the harms of smoking and the benefits of quitting. Proportional mortality studies can provide evidence that smoking kills, and our present study has shown that this study design could also provide evidence that stopping smoking works in populations with reasonably reliable underlying causes of deaths, by adding simple questions about smoking and the duration of quitting at death. Evidence of the health benefits of quitting is needed to support strong tobacco control policies and provide smoking cessation services, and to motivate people to quit before illnesses occur. In conclusion, the health benefits of quitting can be observed by using the proportional mortality study design. In populations with reliable underlying causes of deaths in middle age, this study design is recommended to estimate the benefits of quitting and the harms of smoking, with timeliness and low costs. One simple question about smoking (never, ex- and current smoking 10 years ago, or even better including the duration of quitting at death) should be added to death notification forms (at least be recorded during death registration) to monitor the tobacco epidemic. Author Contributions T.H.L. and S.Y.H. designed and conducted the study in consultation with R.P., and S.Y.H. is the guarantor for the paper; Z.M.M. analysed the data, wrote the first draft and has checked the accuracy and completeness of the references; all authors revised the text critically for important intellectual content, and contributed to final approval of the paper. Supplementary Data Supplementary data are available at IJE online. Acknowledgments The chief acknowledgments are to the relatives who provided information for this study, the research staff, and the Immigration Department and Department of Health (Dr KH Mak) of the Government of the Hong Kong Special Administrative Region. We thank Paul McGale for advice on data analysis, and Lin Xu for revising the manuscript. Conflict of interest: None declared. Funding This work was supported by the Hong Kong Health Services Research Committee (631012) and the Hong Kong Council on Smoking and Health, plus direct support to CTSU by the Imperial Cancer Research Fund and the Medical Research Council. The sponsors of the study had no role in study design, data collection, data analysis, data interpretation or writing of the report. References 1 Jha P , Peto R. Global effects of smoking, of quitting, and of taxing tobacco . N Engl J Med 2014 ; 370 : 60 – 68 . Google Scholar CrossRef Search ADS PubMed 2 World Health Organization . Monitoring Tobacco Use and Prevention Policies . Geneva : World Health Organization , 2017 . 3 Chen ZM , Peto R , Zhou MG et al. Contrasting male and female trends in tobacco-attributed mortality in China: evidence from successive nationwide prospective cohort studies . Lancet 2015 ; 386 : 1447 – 56 . Google Scholar CrossRef Search ADS PubMed 4 Pirie K , Peto R , Reeves GK , Green J , Beral V ; Million Women Study Collaborators. The 21st century hazards of smoking and benefits of stopping: a prospective study of one million women in the UK . Lancet 2013 ; 381 : 133 – 41 . Google Scholar CrossRef Search ADS PubMed 5 Thun MJ , Carter BD , Feskanich D et al. 50-year trends in smoking-related mortality in the United States . N Engl J Med 2013 ; 368 : 351 – 64 . Google Scholar CrossRef Search ADS PubMed 6 Sakata R , McGale P , Grant E , Ozasa K , Peto R , Darby S. Impact of smoking on mortality and life expectancy in Japanese smokers:a prospective cohort study . BMJ 2012 ; 345 : e7093 . Google Scholar CrossRef Search ADS PubMed 7 Jha P , Ramasundarahettige C , Landsman V et al. 21st-century hazards of smoking and benefits of cessation in the United States . N Engl J Med 2013 ; 368 : 341 – 50 . Google Scholar CrossRef Search ADS PubMed 8 Doll R , Peto R , Boreham J , Sutherland I. Mortality in relation to smoking: 50 years' observations on male British doctors . BMJ 2004 ; 328 : 1519 . Google Scholar CrossRef Search ADS PubMed 9 Mai ZM. Quitting smoking and mortality: a population-based mortality case-control study in Hong Kong . School of Public Health, Li Ka Shing Faculty of Medicine, University of Hong Kong , 2016 . 10 Lam TH , He Y , Shi QL et al. Smoking, quitting, and mortality in a Chinese cohort of retired men . Ann Epidemiol 2002 ; 12 : 316 – 20 . Google Scholar CrossRef Search ADS PubMed 11 Jiang CQ , Xu L , Lam TH , Lin JM , Cheng KK , Thomas GN. Smoking cessation and carotid atherosclerosis: the Guangzhou Biobank Cohort Study—CVD . J Epidemiol Community Health 2010 ; 64 : 100 – 0–9 . Google Scholar CrossRef Search ADS 12 Lam TH , Xu L , Schooling CM , Chan WM , Lee SY , Leung GM. Smoking and mortality in a prospective cohort study of elderly Chinese in Hong Kong . Addiction 2015 ; 110: 502 – 10 . Google Scholar CrossRef Search ADS PubMed 13 He Y , Jiang B , Li LS et al. Changes in smoking behavior and subsequent mortality risk during a 35-year follow-up of a cohort in Xi'an, China . Am J Epidemiol 2014 ; 179 : 1060 – 70 . Google Scholar CrossRef Search ADS PubMed 14 Liu BQ , Peto R , Chen ZM et al. Emerging tobacco hazards in China: 1. Retrospective proportional mortality study of one million deaths . BMJ 1998 ; 317 : 1411 – 22 . Google Scholar CrossRef Search ADS PubMed 15 Gajalakshmi V , Peto R , Kanaka TS , Jha P. Smoking and mortality from tuberculosis and other diseases in India: retrospective study of 43 000 adult male deaths and 35 000 controls . Lancet 2003 ; 362 : 507 – 15 . Google Scholar CrossRef Search ADS PubMed 16 Sitas F , Urban M , Bradshaw D , Kielkowski D , Bah S , Peto R. Tobacco attributable deaths in South Africa . Tob Control 2004 ; 13 : 396 – 99 . Google Scholar CrossRef Search ADS PubMed 17 Jha P , Jacob B , Gajalakshmi V et al. A nationally representative case–control study of smoking and death in India . N Engl J Med 2008 ; 358 : 1137 – 47 . Google Scholar CrossRef Search ADS PubMed 18 Alam DS , Jha P , Ramasundarahettige C et al. Smoking-attributable mortality in Bangladesh: proportional mortality study . Bull World Health Organ 2013 ; 91 : 757 – 64 . Google Scholar CrossRef Search ADS PubMed 19 Sitas F , Egger S , Bradshaw D et al. Differences among the coloured, white, black, and other South African populations in smoking-attributed mortality at ages 35–74 years: a case-control study of 481 640 deaths . Lancet 2013 ; 382 : 685 – 93 . Google Scholar CrossRef Search ADS PubMed 20 Lam TH , Ho SY , Hedley AJ , Mak KH , Peto R. Mortality and smoking in Hong Kong: case-control study of all adult deaths in 1998 . BMJ 2001 ; 323 : 361 . Google Scholar CrossRef Search ADS PubMed 21 Lam TH , Ho SY , Hedley AJ , Mak KH , Leung GM. Leisure time physical activity and mortality in Hong Kong: case-control study of all adult deaths in 1998 . Ann Epidemiol 2004 ; 14 : 391 – 98 . Google Scholar CrossRef Search ADS PubMed 22 McGhee S , Ho SY , Schooling CM et al. Mortality associated with passive smoking in Hong Kong . BMJ 2005 ; 330 : 287 – 88 . Google Scholar CrossRef Search ADS PubMed 23 Ho SY , Schooling CM , Hui LL , McGhee SM , Mak KH , Lam TH. Soy consumption and mortality in Hong Kong: proxy-reported case-control study of all older adult deaths in 1998 . Prev Med 2006 ; 43 : 20 – 26 . Google Scholar CrossRef Search ADS PubMed 24 Schooling CM , Ho SY , Leung GM et al. Diet synergies and mortality—a population-based case-control study of 32 462 Hong Kong Chinese older adults . Int J Epidemiol 2006 ; 35 : 418 – 26 . Google Scholar CrossRef Search ADS PubMed 25 Ou CQ , Hedley AJ , Chung RY et al. Socioeconomic disparities in air pollution-associated mortality . Environ Res 2008 ; 107 : 237 – 44 . Google Scholar CrossRef Search ADS PubMed 26 Wang MP , Thomas GN , Ho SY , Lai HK , Mak KH , Lam TH. Fish consumption and mortality in Hong Kong Chinese—the LIMOR Study . Ann Epidemiol 2011 ; 21 : 164 – 69 . Google Scholar CrossRef Search ADS PubMed 27 Thomas GN , Wang MP , Ho SY , Mak KH , Cheng KK , Lam TH. Adverse lifestyle leads to an annual excess of 2 million deaths in China . PLoS ONE 2014 ; 9 : e89650 . Google Scholar CrossRef Search ADS PubMed 28 Plummer M. Improved estimates of floating absolute risk . Stat Med 2004 ; 23 : 93 – 104 . Google Scholar CrossRef Search ADS PubMed 29 IARC . Handbooks of Cancer Prevention, Tobacco Control, Vol. 11 . Reversal of Risk After Quitting Smoking . Lyon : International Agency for Research on Cancer , 2007 . 30 Godtfredsen N , Lam TH , Hansel T et al. COPD-related morbidity and mortality after smoking cessation: status of the evidence . Eur Respir J 2008 ; 32 : 844 – 53 . Google Scholar CrossRef Search ADS PubMed 31 U.S. Department of Health and Human Services . The Health Consequences of Smoking—50 Years of Progress . Atlanta GA : Office of the Surgeon General , 2014 . 32 Zaridze D , Brennan P , Boreham J et al. . Alcohol and cause-specific mortality in Russia: a retrospective case–control study of 48 557 adult deaths . Lancet 2009 ; 373 : 2201 – 14 . Google Scholar CrossRef Search ADS PubMed 33 Jiang GH , Zhang H , Li W et al. Study on smoking-attributed mortality by using all causes of death surveillance system in Tianjin . Zhonghua Liu Xing Bing Xue Za Zhi 2016 ; 37 : 381 – 83 . Google Scholar PubMed 34 Peto R , Lopez AD , Norheim OF. Halving premature death . Science 2014 ; 345 : 1272 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018; all rights reserved. Published by Oxford University Press on behalf of the International Epidemiological Association This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png International Journal of Epidemiology Oxford University Press

Mortality reduction from quitting smoking in Hong Kong: population-wide proportional mortality study

Download PDF
8 pages

Loading next page...
 
/lp/ou_press/mortality-reduction-from-quitting-smoking-in-hong-kong-population-wide-cWavJLQk2i

References (32)

Publisher
Oxford University Press
Copyright
© The Author(s) 2018; all rights reserved. Published by Oxford University Press on behalf of the International Epidemiological Association
ISSN
0300-5771
eISSN
1464-3685
DOI
10.1093/ije/dyx267
Publisher site
See Article on Publisher Site

Abstract

Abstract Background The effects of smoking cessation might be different in different populations. Proportional mortality studies of all deaths, relating the certified cause to retrospectively determined smoking habits, have helped assess the hazards of smoking in Hong Kong, and further analyses can help assess the effects of prolonged cessation (although not of recent cessation, as life-threatening disease can itself cause cessation, particularly in old age). Methods The LIMOR study sought the certified causes of all deaths in 1998, and interviewed 81% of families at death registries to determine the decedent’s smoking history. Cases were deaths from pre-defined diseases of interest (N = 15 356); controls were deaths from pre-defined non-smoking-related diseases (N = 5023). Case vs control odds ratios for ex-smokers vs smokers were calculated by age-, sex- and education-standardized logistic regression. These are described as mortality rate ratios (RRs), with a group-specific confidence interval (CI). Results For the aggregate of all deaths from any of the diseases of interest at ages 35-69 years, the RRs for current smoking, quitting 0-4, 5-9 or 10+ years ago and never-smoking were, respectively, RR = 1 (95% CI 0.86-1.17), 0.91 (0.73-1.14), 0.71 (0.49-1.02), 0.66 (0.50-0.87) and 0.43 (0.37-0.48). Younger age of quitting (25-44 or 45-64) appeared to be associated with greater protection: RR = 0.58 (0.38-0.88) and 0.71 (0.54-0.93), respectively. These patterns were less clear at older ages, particularly for death from emphysema. Conclusions Longer durations of smoking cessation are associated with progressively lower mortality rates from the diseases of interest. For sustainable monitoring of tobacco-attributed mortality, approximate years since last smoked should be recorded during death registration. Smoking cessation, proportional mortality study, epidemiology, death certificate, Hong Kong China, health benefit Key Messages There is limited evidence from low- and middle-income countries (LMICs) about the effects of cessation and, in populations, comparison of current mortality rates in smokers and ex-smokers would underestimate the benefits of quitting. Proportional mortality studies could provide a timely low-cost alternative to prospective cohort studies, at least for assessing the effects of long-term cessation, in addition to the harms of smoking. Taking the advantage of a proportional mortality study in Hong Kong, which is the most urbanized and Westernized Chinese city and where the tobacco epidemic reached its peak about 20 years earlier than in the Chinese mainland we examined the effects of cessation by comparing the proportions of ex-smokers and current-smokers among those dying from particular diseases. The benefits of quitting and of longer durations of smoking cessation, associated with progressively lower mortality rates of the diseases of interest, can be observed by using the proportional mortality study design. Introduction Although smoking-attributed mortality is decreasing steadily among men in most developed countries, it is still increasing in many developing countries.1,2 In China, with 20% of the world’s population consuming 40% of the world’s cigarettes, smoking-attributed deaths will continue to increase over the next few decades, unless there is widespread cessation.3 Prospective studies from developed countries provide strong quantitative evidence that stopping smoking works.4–7 For example, UK smokers who stop before age 40 (preferably well before 40) avoid over 90% of the excess mortality rates among continuing smokers.8 As yet, there is limited evidence from low- and middle-income countries (LMICs) about the effects of cessation,9 and in populations where the risks among smokers are still rising, comparison of current mortality rates among smokers and ex-smokers would underestimate the benefits of quitting. As the epidemic of death from tobacco is at a more advanced stage in Hong Kong than in the Chinese mainland studies in Hong Kong could be particularly informative. The ideal would be a large prospective study that carefully limits the effects of reverse causality, whereby life-threatening disease may make smokers stop [for chronic obstructive pulmonary disease (COPD), this can happen many years before death, artificially reducing the death rate among the smokers and increasing it substantially among the ex-smokers]. However, cohort studies are difficult to conduct, expensive and take many years to deliver results,10–13 so retrospective studies are also needed, although researchers have to consider carefully how reverse causality might distort their findings, particularly for COPD. Proportional mortality studies that involve only dead subjects (comparing the proportions of ever-smokers and never-smokers among those dying from particular diseases) could provide a timely low-cost alternative, at least for assessing the effects of long-term cessation. Proportional mortality studies have been used to investigate the harms of smoking, mostly in developing countries, but the benefits of quitting were not reported,14–19 partly for fear of producing results that are substantially biased by reverse causality, and some reports did not mention whether quitting was asked about. The Hong Kong Lifestyle and Mortality (HK LIMOR) study sought from family members information about the previous smoking habits of all who died in Hong Kong in 1998, including information on how long ago the dead person had stopped smoking; but the main report combined the ex-smokers with those who had continued smoking until the last year of their life (so it compared those who had ever smoked vs those who had never smoked).20 We now report its findings according to the duration of quitting. Methods The study methods and results on smoking and various other factors associated have been reported previously (Supplementary Part I, available at IJE online).20–27 We have found the definition of dead cases and controls in the most recent proportional mortality case-control study most appropriate for the HK LIMOR study,9 with cases being deaths from diseases that could be caused by smoking, and controls being deaths from all non-smoking related causes (Supplementary Table S1, available at IJE online).19 Details of the assessment of quitting are in Supplementary Part II (Supplementary materials, available at IJE online). Present analysis only used deaths with additional quitting data collected in the second half of all deaths in 1998. To compare our previous results for all deaths on the harms of smoking (dead cases vs living controls),20 we also included the results comparing dead cases with dead controls (Supplementary Table S2, available at IJE online). Statistical analysis Case vs control odds ratios for the quitting status were calculated by unconditional logistic regression, adjusted for 5-year age groups, sex and education, and are described as mortality rate ratios (RRs, calculated as odds ratios). To assess the association of stopping smoking with smoking-attributed mortality by the age of quitting, the RR was calculated for never smokers, and quitters who stopped smoking at the ages of 25-44 and 45-64 years, compared with current smokers. The group-specific confidence interval (CI) for the current smokers’ RR of 1.00 was calculated to reflect the variance of the log risk in the current smokers, using Plummer’s method.28 To select the appropriate age range, mortality RRs were calculated for main cause-specific deaths by smoking history (ever vs never smokers) in all deaths (Supplementary Tables S3 and S4, available at IJE online). Smoking was expected to cause few deaths before age 35, and cause of death information can be unreliable in old age, so subjects aged 35-84 at death were included. As most of those killed by smoking would otherwise have survived beyond age 70, but a minority would have died by 70, we stratified the age into two groups (middle age: 35-69 or old age: 70-84 years) for all the analyses. All analyses were conducted using R 3.3.1. Results Mortality by duration of quitting There were 9772 male and 5584 female cases, and 2503 male and 2520 female controls. The cases and controls had similar demographic characteristics, with a mean [standard deviation (SD)] age of 70.2 (10.7) and 68.2 (12.0) years, respectively (Supplementary Table S5, available at IJE online). Table 1 shows that in middle age (35-69 years), all the RRs among long-term quitters (vs current smokers) were lower: RR 0.50 (95% CI 0.33-0.74) for lung cancer, 0.92 (0.66-1.28) for smoking-related cancer, 0.40 (0.27-0.61) for cardiovascular disease (CVD) and 0.81 (0.49-1.33) for COPD. But in old age (70-84 years), the RR for COPD was higher (1.40, 1.11-1.76). Quitters who had stopped for 5-9 years also had lower lung cancer, smoking-related cancer and CVD mortality risks, but higher COPD mortality risks compared with current smokers. New quitters who had stopped for less than 5 years had lower CVD mortality risks: RR 0.56 (95% CI 0.41-0.78) in middle age and 0.90 (0.68-1.18) in old age. However, the RRs for lung cancer and COPD among new quitters (vs current smokers) in old age were higher, probably because of ill-quitter effect. Table 1. Duration of smoking cessation and mortality from four smoking-related diseases—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers. Deaths from selected diseases vs deaths from control diseases, at ages 35-69 or 70-84 years (both sexes) Current smokers Duration of quitting Trend P-value* Never smokers <5 years 5-9 years ≥10 years Lung cancer Age 35-69 1.00 (231) 1.23 (141) 0.66 (28) 0.50 (40) <0.001 0.22 (187) 0.82-1.21 0.95-1.60 0.40-1.08 0.33-0.74 0.18-0.28 Age 70-84 1.00 (228) 1.21 (132) 0.79 (47) 0.62 (121) 0.001 0.19 (174) 0.83-1.21 0.93-1.59 0.53-1.17 0.49-0.78 0.16-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 1.11 (151) 0.72 (36) 0.92 (84) 0.55 0.58 (471) 0.83-1.20 0.88-1.47 0.45-1.15 0.66-1.28 0.50-0.68 Age 70-84 1.00 (166) 0.93 (73) 0.66 (29) 0.67 (95) 0.011 0.60 (318) 0.81-1.23 0.69-1.27 0.42-1.05 0.52-0.86 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.56 (66) 0.59 (28) 0.40 (37) <0.001 0.46 (338) 0.82-1.22 0.41-0.78 0.36-0.98 0.27-0.61 0.38-0.54 Age 70-84 1.00 (266) 0.90 (111) 0.72 (49) 0.83 (197) 0.15 0.76 (785) 0.83-1.20 0.68-1.18 0.48-1.06 0.67-1.02 0.67-0.86 COPD Age 35-69 1.00 (66) 0.72 (24) 1.35 (19) 0.81 (22) 0.79 0.26 (46) 0.75-1.34 0.46-1.14 0.76-2.42 0.49-1.33 0.18-0.38 Age 70-84 1.00 (104) 1.34 (67) 1.59 (42) 1.40 (134) 0.050 0.26 (106) 0.79-1.27 0.97-1.85 1.05-2.41 1.11-1.76 0.21-0.33 Current smokers Duration of quitting Trend P-value* Never smokers <5 years 5-9 years ≥10 years Lung cancer Age 35-69 1.00 (231) 1.23 (141) 0.66 (28) 0.50 (40) <0.001 0.22 (187) 0.82-1.21 0.95-1.60 0.40-1.08 0.33-0.74 0.18-0.28 Age 70-84 1.00 (228) 1.21 (132) 0.79 (47) 0.62 (121) 0.001 0.19 (174) 0.83-1.21 0.93-1.59 0.53-1.17 0.49-0.78 0.16-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 1.11 (151) 0.72 (36) 0.92 (84) 0.55 0.58 (471) 0.83-1.20 0.88-1.47 0.45-1.15 0.66-1.28 0.50-0.68 Age 70-84 1.00 (166) 0.93 (73) 0.66 (29) 0.67 (95) 0.011 0.60 (318) 0.81-1.23 0.69-1.27 0.42-1.05 0.52-0.86 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.56 (66) 0.59 (28) 0.40 (37) <0.001 0.46 (338) 0.82-1.22 0.41-0.78 0.36-0.98 0.27-0.61 0.38-0.54 Age 70-84 1.00 (266) 0.90 (111) 0.72 (49) 0.83 (197) 0.15 0.76 (785) 0.83-1.20 0.68-1.18 0.48-1.06 0.67-1.02 0.67-0.86 COPD Age 35-69 1.00 (66) 0.72 (24) 1.35 (19) 0.81 (22) 0.79 0.26 (46) 0.75-1.34 0.46-1.14 0.76-2.42 0.49-1.33 0.18-0.38 Age 70-84 1.00 (104) 1.34 (67) 1.59 (42) 1.40 (134) 0.050 0.26 (106) 0.79-1.27 0.97-1.85 1.05-2.41 1.11-1.76 0.21-0.33 Numbers of cases (i.e. deaths from the disease of interest) are given in brackets after each RR (in bold). Numbers of controls in current/ex < 5/ex 5-9/ex 10 years/never smokers are 194/96/36/65/744 at ages 35-69 years, and 200/95/54/175/826 at ages 70-84 years. CI, group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. Other smoking-related cancers: upper aerodigestive, stomach, liver, pancreas, cervix, urinary and myeloid leukaemia (ICD-9 140-51, 155-7, 160-1, 179-80, 184, 188-9, 205). Cardiovascular disease: stroke and ischaemic heart disease (ICD-9 430-8, 410-4, 440-8). COPD: chronic obstructive pulmonary disease (ICD-9 415-7, 490-6); heterogeneity by sex was P < 0.05 only for COPD. * Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, each would have yielded P < 0.05. Table 1. Duration of smoking cessation and mortality from four smoking-related diseases—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers. Deaths from selected diseases vs deaths from control diseases, at ages 35-69 or 70-84 years (both sexes) Current smokers Duration of quitting Trend P-value* Never smokers <5 years 5-9 years ≥10 years Lung cancer Age 35-69 1.00 (231) 1.23 (141) 0.66 (28) 0.50 (40) <0.001 0.22 (187) 0.82-1.21 0.95-1.60 0.40-1.08 0.33-0.74 0.18-0.28 Age 70-84 1.00 (228) 1.21 (132) 0.79 (47) 0.62 (121) 0.001 0.19 (174) 0.83-1.21 0.93-1.59 0.53-1.17 0.49-0.78 0.16-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 1.11 (151) 0.72 (36) 0.92 (84) 0.55 0.58 (471) 0.83-1.20 0.88-1.47 0.45-1.15 0.66-1.28 0.50-0.68 Age 70-84 1.00 (166) 0.93 (73) 0.66 (29) 0.67 (95) 0.011 0.60 (318) 0.81-1.23 0.69-1.27 0.42-1.05 0.52-0.86 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.56 (66) 0.59 (28) 0.40 (37) <0.001 0.46 (338) 0.82-1.22 0.41-0.78 0.36-0.98 0.27-0.61 0.38-0.54 Age 70-84 1.00 (266) 0.90 (111) 0.72 (49) 0.83 (197) 0.15 0.76 (785) 0.83-1.20 0.68-1.18 0.48-1.06 0.67-1.02 0.67-0.86 COPD Age 35-69 1.00 (66) 0.72 (24) 1.35 (19) 0.81 (22) 0.79 0.26 (46) 0.75-1.34 0.46-1.14 0.76-2.42 0.49-1.33 0.18-0.38 Age 70-84 1.00 (104) 1.34 (67) 1.59 (42) 1.40 (134) 0.050 0.26 (106) 0.79-1.27 0.97-1.85 1.05-2.41 1.11-1.76 0.21-0.33 Current smokers Duration of quitting Trend P-value* Never smokers <5 years 5-9 years ≥10 years Lung cancer Age 35-69 1.00 (231) 1.23 (141) 0.66 (28) 0.50 (40) <0.001 0.22 (187) 0.82-1.21 0.95-1.60 0.40-1.08 0.33-0.74 0.18-0.28 Age 70-84 1.00 (228) 1.21 (132) 0.79 (47) 0.62 (121) 0.001 0.19 (174) 0.83-1.21 0.93-1.59 0.53-1.17 0.49-0.78 0.16-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 1.11 (151) 0.72 (36) 0.92 (84) 0.55 0.58 (471) 0.83-1.20 0.88-1.47 0.45-1.15 0.66-1.28 0.50-0.68 Age 70-84 1.00 (166) 0.93 (73) 0.66 (29) 0.67 (95) 0.011 0.60 (318) 0.81-1.23 0.69-1.27 0.42-1.05 0.52-0.86 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.56 (66) 0.59 (28) 0.40 (37) <0.001 0.46 (338) 0.82-1.22 0.41-0.78 0.36-0.98 0.27-0.61 0.38-0.54 Age 70-84 1.00 (266) 0.90 (111) 0.72 (49) 0.83 (197) 0.15 0.76 (785) 0.83-1.20 0.68-1.18 0.48-1.06 0.67-1.02 0.67-0.86 COPD Age 35-69 1.00 (66) 0.72 (24) 1.35 (19) 0.81 (22) 0.79 0.26 (46) 0.75-1.34 0.46-1.14 0.76-2.42 0.49-1.33 0.18-0.38 Age 70-84 1.00 (104) 1.34 (67) 1.59 (42) 1.40 (134) 0.050 0.26 (106) 0.79-1.27 0.97-1.85 1.05-2.41 1.11-1.76 0.21-0.33 Numbers of cases (i.e. deaths from the disease of interest) are given in brackets after each RR (in bold). Numbers of controls in current/ex < 5/ex 5-9/ex 10 years/never smokers are 194/96/36/65/744 at ages 35-69 years, and 200/95/54/175/826 at ages 70-84 years. CI, group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. Other smoking-related cancers: upper aerodigestive, stomach, liver, pancreas, cervix, urinary and myeloid leukaemia (ICD-9 140-51, 155-7, 160-1, 179-80, 184, 188-9, 205). Cardiovascular disease: stroke and ischaemic heart disease (ICD-9 430-8, 410-4, 440-8). COPD: chronic obstructive pulmonary disease (ICD-9 415-7, 490-6); heterogeneity by sex was P < 0.05 only for COPD. * Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, each would have yielded P < 0.05. Figure 1 shows declining trends of RRs for all deaths of interest from current smoking to quitting for less than 5, 5-9 and 10+ years, and never smoking: RR 1.00 (95% CI 0.86-1.17), 0.91 (0.73-1.14), 0.71 (0.49-1.02), 0.66 (0.50-0.87) and 0.43 (0.37-0.48) in middle age (P for linear trend: 0.006); and 1.00 (0.86-1.16), 1.06 (0.85-1.32), 0.86 (0.64-1.16), 0.87 (0.74-1.03) and 0.55 (0.50-0.61) in old age (P for linear trend: 0.19). Figure 1 View largeDownload slide Duration of smoking cessation and mortality from all deaths of interest—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers Deaths from selected diseases vs deaths from control diseases, at 35–69 or 70–84 years (both sexes). Numbers of cases (ie, deaths from the disease of interest) are given in brackets after each RR. Number of controls in current/ ex <5/ ex 5–9/ ex ≥10 years/ never smokers are 194/96/36/65/744 at ages 35–69 years, and 200/95/54/175/826 at ages 70–84 years. CI: group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variances of the log risk in only that one group. *Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, both would have yielded p<0.001. Figure 1 View largeDownload slide Duration of smoking cessation and mortality from all deaths of interest—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers Deaths from selected diseases vs deaths from control diseases, at 35–69 or 70–84 years (both sexes). Numbers of cases (ie, deaths from the disease of interest) are given in brackets after each RR. Number of controls in current/ ex <5/ ex 5–9/ ex ≥10 years/ never smokers are 194/96/36/65/744 at ages 35–69 years, and 200/95/54/175/826 at ages 70–84 years. CI: group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variances of the log risk in only that one group. *Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, both would have yielded p<0.001. Mortality by age of quitting Figure 2 also shows a clear declining trend of mortality risk, from current smoking to quitting at the ages of 45-64 and 25-44 years and never smoking, for all deaths of interest in middle age: 1.00 (0.86-1.17), 0.71 (0.54-0.93), 0.58 (0.38-0.88) and 0.43 (0.37-0.49), (P for linear trend: 0.003). However, subjects in old age (70-84 years) who stopped at younger age (25-44 years) had higher mortality risks: 1.47 (0.51-4.22), but this was based on only 31 subjects (27 cases and four controls), so it is not statistically reliable. Figure 2 View largeDownload slide Age at smoking cessation and mortality from all deaths of interest—mortality rate rations (RR, 95% CI) comparing ex- or never-somkers vs current smokers To limit reverse causality, analyses exclude those who stopped <5 years ago, or at age>65. They compare deaths from selected diseases vs deaths from control diseases, at ages 35–69 or 70–84 years (both sexes) Numbers of cases (ie, deaths from the disease of interest) are given in brackets after each RR. Numbers of controls in current/ ex 45–64/ ex 25–44 years/ never smokers are 194/137/41/744 at ages 35–69 years, and 200/113/4/826 at ages 70–84 years. CI: group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. *Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, both would have yielded p<0.001. Figure 2 View largeDownload slide Age at smoking cessation and mortality from all deaths of interest—mortality rate rations (RR, 95% CI) comparing ex- or never-somkers vs current smokers To limit reverse causality, analyses exclude those who stopped <5 years ago, or at age>65. They compare deaths from selected diseases vs deaths from control diseases, at ages 35–69 or 70–84 years (both sexes) Numbers of cases (ie, deaths from the disease of interest) are given in brackets after each RR. Numbers of controls in current/ ex 45–64/ ex 25–44 years/ never smokers are 194/137/41/744 at ages 35–69 years, and 200/113/4/826 at ages 70–84 years. CI: group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. *Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, both would have yielded p<0.001. Table 2 shows mortality RRs for the other four disease groupings (causable by smoking) by age of quitting vs current smoking. Subjects in middle age (35-69 years) who stopped smoking at younger age (25-44 years) had non-significantly (with overlapping 95% CIs) lower mortality risks than those who stopped smoking at older age (45-64 years) for lung cancer (RR 0.41, 95% CI 0.22-0.78 vs 0.61, 0.42-0.88), CVD (0.36, 0.18-0.70 vs 0.51, 0.35-0.73), and all deaths of interest (0.58, 0.38-0.88 vs 0.71, 0.54-0.93, Figure 2), respectively. Few quitters stopped before age 45, so the estimates of quitting at the age of 25-44 years in old age are not statistically reliable. Among subjects who had stopped at age 45-64 years, the RRs for lung cancer, smoking-related cancer and CVD were lower, but those for COPD were still substantially higher than current smokers: RR 1.12 (95% CI 0.73-1.72) in middle age and 1.59 (1.20-2.11) in old age. Table 2. Age at smoking cessation and mortality from four smoking-related diseases—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers. To limit reverse causality, analyses exclude those who stopped < 5 years ago, or at age> 65. They compare deaths from selected diseases vs deaths from control diseases, at ages 35-69 or 70-84 years (both sexes) Current smokers Age at quitting Trend P-value* Never smokers 45-64 years 25-44 years Lung cancer Age 35-69 1.00 (231) 0.61 (53) 0.41 (14) < 0.001 0.22 (187) 0.83-1.21 0.42-0.88 0.22-0.78 0.17-0.27 Age 70-84 1.00 (228) 0.66 (86) 0.78 (4) 0.015 0.19 (174) 0.83-1.21 0.50-0.88 0.19-3.16 0.15-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 0.84 (81) 0.84 (38) 0.35 0.58 (471) 0.83-1.20 0.60-1.17 0.52-1.36 0.50-0.69 Age 70-84 1.00 (166) 0.71 (69) 1.00 (4) 0.08 0.60 (318) 0.81-1.23 0.53-0.97 0.25-4.07 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.51 (53) 0.36 (12) < 0.001 0.46 (338) 0.82-1.21 0.35-0.73 0.18-0.70 0.39-0.55 Age 70-84 1.00 (266) 0.90 (133) 2.08 (10) 0.82 0.76 (785) 0.83-1.20 0.69-1.15 0.65-6.66 0.66-0.87 COPD Age 35-69 1.00 (66) 1.12 (36) 0.60 (5) 0.78 0.26 (46) 0.75-1.32 0.73-1.72 0.23-1.60 0.18-0.38 Age 70-84 1.00 (104) 1.59 (92) 2.83 (6) 0.008 0.26 (106) 0.79-1.27 1.20-2.11 0.74-10.8 0.20-0.33 Current smokers Age at quitting Trend P-value* Never smokers 45-64 years 25-44 years Lung cancer Age 35-69 1.00 (231) 0.61 (53) 0.41 (14) < 0.001 0.22 (187) 0.83-1.21 0.42-0.88 0.22-0.78 0.17-0.27 Age 70-84 1.00 (228) 0.66 (86) 0.78 (4) 0.015 0.19 (174) 0.83-1.21 0.50-0.88 0.19-3.16 0.15-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 0.84 (81) 0.84 (38) 0.35 0.58 (471) 0.83-1.20 0.60-1.17 0.52-1.36 0.50-0.69 Age 70-84 1.00 (166) 0.71 (69) 1.00 (4) 0.08 0.60 (318) 0.81-1.23 0.53-0.97 0.25-4.07 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.51 (53) 0.36 (12) < 0.001 0.46 (338) 0.82-1.21 0.35-0.73 0.18-0.70 0.39-0.55 Age 70-84 1.00 (266) 0.90 (133) 2.08 (10) 0.82 0.76 (785) 0.83-1.20 0.69-1.15 0.65-6.66 0.66-0.87 COPD Age 35-69 1.00 (66) 1.12 (36) 0.60 (5) 0.78 0.26 (46) 0.75-1.32 0.73-1.72 0.23-1.60 0.18-0.38 Age 70-84 1.00 (104) 1.59 (92) 2.83 (6) 0.008 0.26 (106) 0.79-1.27 1.20-2.11 0.74-10.8 0.20-0.33 Numbers of cases (i.e. deaths from the disease of interest) are given in brackets after each RR (in bold). Numbers of controls in current/ex 45–64/ex 25–44 years/never smokers are 194/137/41/744 at ages 35–69 years, and 200/113/4/826 at ages 70–84 years. CI, group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. Other smoking-related cancers: upper aerodigestive, stomach, liver, pancreas, cervix, urinary and myeloid leukaemia (ICD-9 140-51, 155-7, 160-1, 179-80, 184, 188-9, 205). Cardiovascular disease: stroke and ischaemic heart disease (ICD-9 430-8, 410-4, 440-8). COPD: chronic obstructive pulmonary disease (ICD-9 415-7, 490-6); heterogeneity was P < 0.05 by sex only for COPD. * Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, each would have yielded P < 0.01. Table 2. Age at smoking cessation and mortality from four smoking-related diseases—mortality rate ratios (RR, 95% CI) comparing ex- or never-smokers vs current smokers. To limit reverse causality, analyses exclude those who stopped < 5 years ago, or at age> 65. They compare deaths from selected diseases vs deaths from control diseases, at ages 35-69 or 70-84 years (both sexes) Current smokers Age at quitting Trend P-value* Never smokers 45-64 years 25-44 years Lung cancer Age 35-69 1.00 (231) 0.61 (53) 0.41 (14) < 0.001 0.22 (187) 0.83-1.21 0.42-0.88 0.22-0.78 0.17-0.27 Age 70-84 1.00 (228) 0.66 (86) 0.78 (4) 0.015 0.19 (174) 0.83-1.21 0.50-0.88 0.19-3.16 0.15-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 0.84 (81) 0.84 (38) 0.35 0.58 (471) 0.83-1.20 0.60-1.17 0.52-1.36 0.50-0.69 Age 70-84 1.00 (166) 0.71 (69) 1.00 (4) 0.08 0.60 (318) 0.81-1.23 0.53-0.97 0.25-4.07 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.51 (53) 0.36 (12) < 0.001 0.46 (338) 0.82-1.21 0.35-0.73 0.18-0.70 0.39-0.55 Age 70-84 1.00 (266) 0.90 (133) 2.08 (10) 0.82 0.76 (785) 0.83-1.20 0.69-1.15 0.65-6.66 0.66-0.87 COPD Age 35-69 1.00 (66) 1.12 (36) 0.60 (5) 0.78 0.26 (46) 0.75-1.32 0.73-1.72 0.23-1.60 0.18-0.38 Age 70-84 1.00 (104) 1.59 (92) 2.83 (6) 0.008 0.26 (106) 0.79-1.27 1.20-2.11 0.74-10.8 0.20-0.33 Current smokers Age at quitting Trend P-value* Never smokers 45-64 years 25-44 years Lung cancer Age 35-69 1.00 (231) 0.61 (53) 0.41 (14) < 0.001 0.22 (187) 0.83-1.21 0.42-0.88 0.22-0.78 0.17-0.27 Age 70-84 1.00 (228) 0.66 (86) 0.78 (4) 0.015 0.19 (174) 0.83-1.21 0.50-0.88 0.19-3.16 0.15-0.23 Other smoking-related cancer Age 35-69 1.00 (284) 0.84 (81) 0.84 (38) 0.35 0.58 (471) 0.83-1.20 0.60-1.17 0.52-1.36 0.50-0.69 Age 70-84 1.00 (166) 0.71 (69) 1.00 (4) 0.08 0.60 (318) 0.81-1.23 0.53-0.97 0.25-4.07 0.51-0.71 Cardiovascular disease Age 35-69 1.00 (231) 0.51 (53) 0.36 (12) < 0.001 0.46 (338) 0.82-1.21 0.35-0.73 0.18-0.70 0.39-0.55 Age 70-84 1.00 (266) 0.90 (133) 2.08 (10) 0.82 0.76 (785) 0.83-1.20 0.69-1.15 0.65-6.66 0.66-0.87 COPD Age 35-69 1.00 (66) 1.12 (36) 0.60 (5) 0.78 0.26 (46) 0.75-1.32 0.73-1.72 0.23-1.60 0.18-0.38 Age 70-84 1.00 (104) 1.59 (92) 2.83 (6) 0.008 0.26 (106) 0.79-1.27 1.20-2.11 0.74-10.8 0.20-0.33 Numbers of cases (i.e. deaths from the disease of interest) are given in brackets after each RR (in bold). Numbers of controls in current/ex 45–64/ex 25–44 years/never smokers are 194/137/41/744 at ages 35–69 years, and 200/113/4/826 at ages 70–84 years. CI, group-specific confidence interval for the age-, sex- and education-adjusted RR, reflecting the variance of the log risk in only that one group. Other smoking-related cancers: upper aerodigestive, stomach, liver, pancreas, cervix, urinary and myeloid leukaemia (ICD-9 140-51, 155-7, 160-1, 179-80, 184, 188-9, 205). Cardiovascular disease: stroke and ischaemic heart disease (ICD-9 430-8, 410-4, 440-8). COPD: chronic obstructive pulmonary disease (ICD-9 415-7, 490-6); heterogeneity was P < 0.05 by sex only for COPD. * Trend test in smokers, excluding never smokers; if trend tests in this table included never smokers, each would have yielded P < 0.01. Discussion The purpose of the HK LIMOR study was to estimate the smoking-attributed mortality in Hong Kong, where the tobacco epidemic has reached a fairly advanced stage,20 and the present analysis is first to investigate the health benefits of quitting using proportional mortality study design. We found lower RRs of cause-specific mortality associated with longer duration of quitting and quitting at younger age. These benefits have not been reported in previous proportional mortality studies, probably because the information on quitting (or ex-smokers) was unavailable. Reverse causality, as in many cohort studies, was observed in the present analysis for COPD, which is most subject to ill-quitter effect.29 COPD typically develops among smokers with slowly worsening respiratory symptoms over many years, during which some may quit smoking when necessitated by their worsening condition. Thus, ill-quitter effect would be more common for COPD than cardiovascular disease, especially among older people with longer duration of smoking and symptoms, and lower ability to recover.30 Indeed, in the present analysis, reverse causality was observed for COPD in middle-aged subjects who had quit for 5-9 years, and in old age regardless of duration of quitting. To clarify this issue, collecting information on the reason for quitting is essential and recommended. To study the benefits of quitting in proportional mortality studies, differential misclassification of quitting between deaths and healthy subjects may occur using living controls. Indeed, in the HK LIMOR study, we found that the proportion of ex-smokers (or quitters) in all the deaths was higher than that in healthy subjects; in particular, the proportion in dead cases (deaths from diseases causable by smoking) was much higher than that in living controls.9 It is plausible that dead subjects were more likely to have quit due to ill health than healthy living subjects. We took advantage of the strength of the definition of dead cases and dead controls in the most recent mortality case-control study, which is that any ill-quitter effect in dead cases should be similar to that in dead controls, to examine the benefits of quitting. As, however, quitting must have reduced some risks from certain deaths in the dead control group (for example, some of those from diabetes and colorectal cancer),31 the present analyses may have slightly underestimated the benefits of quitting. Indeed, in old age (70-84 years), the RRs among ex-smokers are slightly smaller than those in our previous prospective cohort study of older Chinese in Hong Kong.12 However, in middle age (35-69 years), the lower RRs (vs current smokers) among all the quitters in Table 1 were consistently observed for all deaths of interest and CVD, particularly among new quitters who had stopped smoking for less than 5 years (reduction of CVD mortality: 44%, 95% CI 22-59%), which are consistent with the U.S. Surgeon General's Reports on Smoking and Tobacco Use (USSG) Report 2010, that rapid risk reduction of vascular mortality by about 50% can be observed after quitting for 1 year. To help monitor the tobacco epidemic, the methods of proportional mortality study should be used in any other populations where death registration is organized centrally. As interviewing was done in death registries, the HK LIMOR study had a high coverage of all deaths (81%), and a short time interval between death and interview, reducing recall error. Moreover, the analytical strategies in the present study using deaths from smoking-related causes as cases and those from non-smoking related causes as controls, could be used to assess the benefits of quitting, in addition to the harms of smoking. Several limitations should be considered. First, the definitions of cases and controls in previous proportional mortality studies varied. For example, some diseases such as breast cancer and colorectal cancer are considered to be causally related to smoking by the 2014 USSG Report, but were defined as controls in our study. Smoking was associated with reduced risk of Parkinson’s disease, ulcerative colitis and endometrial cancer, but we analysed them as cases.31 Nonetheless, there is no consensus of the definitions of cases and controls in mortality case-control studies.9 Further research is warranted, by collaborating all proportional mortality studies in the world to determine the most appropriate definition for studying smoking or other factors.32 The HK LIMOR study was conducted in 1998 to investigate the relation between smoking and mortality. The results might not be generalized to the present situation in Hong Kong, but should be relevant to show the great benefits of stopping smoking in the Chinese mainland. A new mortality case-control study in Hong Kong and in the Chinese mainland (or any other LMICs) is warranted. To monitor the tobacco epidemic routinely, a new method of the proportional mortality study for collecting data is recommended, as in the South Africa’s and the Tianjin’s proportional mortality studies, that the smoking status of the deceased is routinely recorded on the new death certificate for long-term sustainable monitoring.16,33 The tobacco epidemic in Hong Kong (the most urbanized and Westernized Chinese city) reached its peak about 20 years earlier than in the Chinese mainland, but about 20 years later than in developed Western countries.12 Hong Kong has been a forewarning model for the Chinese mainland.20 The mortality relative risk has already reached 3 in the UK, USA and Australia, which could be followed by the Chinese mainland and other LMICs undergoing rapid economic development in the next few decades. Stopping smoking is one of the most practicable ways to avoid a large proportion of smoking-attributable deaths, particularly premature deaths (age 35-69).34 Quitting, however, is uncommon in the Chinese mainland and other LMICs. Ex-smokers are far fewer than current smokers in most LMICs, particularly those who stopped smoking at young age (before 45 years). However, with increasing smoking cessation from more effective tobacco-control measures in developing regions,2 the applicability of proportional mortality studies should expand. Moreover, many higher-income countries also have no or limited prospective data showing the harms of smoking and the benefits of quitting. Proportional mortality studies can provide evidence that smoking kills, and our present study has shown that this study design could also provide evidence that stopping smoking works in populations with reasonably reliable underlying causes of deaths, by adding simple questions about smoking and the duration of quitting at death. Evidence of the health benefits of quitting is needed to support strong tobacco control policies and provide smoking cessation services, and to motivate people to quit before illnesses occur. In conclusion, the health benefits of quitting can be observed by using the proportional mortality study design. In populations with reliable underlying causes of deaths in middle age, this study design is recommended to estimate the benefits of quitting and the harms of smoking, with timeliness and low costs. One simple question about smoking (never, ex- and current smoking 10 years ago, or even better including the duration of quitting at death) should be added to death notification forms (at least be recorded during death registration) to monitor the tobacco epidemic. Author Contributions T.H.L. and S.Y.H. designed and conducted the study in consultation with R.P., and S.Y.H. is the guarantor for the paper; Z.M.M. analysed the data, wrote the first draft and has checked the accuracy and completeness of the references; all authors revised the text critically for important intellectual content, and contributed to final approval of the paper. Supplementary Data Supplementary data are available at IJE online. Acknowledgments The chief acknowledgments are to the relatives who provided information for this study, the research staff, and the Immigration Department and Department of Health (Dr KH Mak) of the Government of the Hong Kong Special Administrative Region. We thank Paul McGale for advice on data analysis, and Lin Xu for revising the manuscript. Conflict of interest: None declared. Funding This work was supported by the Hong Kong Health Services Research Committee (631012) and the Hong Kong Council on Smoking and Health, plus direct support to CTSU by the Imperial Cancer Research Fund and the Medical Research Council. The sponsors of the study had no role in study design, data collection, data analysis, data interpretation or writing of the report. References 1 Jha P , Peto R. Global effects of smoking, of quitting, and of taxing tobacco . N Engl J Med 2014 ; 370 : 60 – 68 . Google Scholar CrossRef Search ADS PubMed 2 World Health Organization . Monitoring Tobacco Use and Prevention Policies . Geneva : World Health Organization , 2017 . 3 Chen ZM , Peto R , Zhou MG et al. Contrasting male and female trends in tobacco-attributed mortality in China: evidence from successive nationwide prospective cohort studies . Lancet 2015 ; 386 : 1447 – 56 . Google Scholar CrossRef Search ADS PubMed 4 Pirie K , Peto R , Reeves GK , Green J , Beral V ; Million Women Study Collaborators. The 21st century hazards of smoking and benefits of stopping: a prospective study of one million women in the UK . Lancet 2013 ; 381 : 133 – 41 . Google Scholar CrossRef Search ADS PubMed 5 Thun MJ , Carter BD , Feskanich D et al. 50-year trends in smoking-related mortality in the United States . N Engl J Med 2013 ; 368 : 351 – 64 . Google Scholar CrossRef Search ADS PubMed 6 Sakata R , McGale P , Grant E , Ozasa K , Peto R , Darby S. Impact of smoking on mortality and life expectancy in Japanese smokers:a prospective cohort study . BMJ 2012 ; 345 : e7093 . Google Scholar CrossRef Search ADS PubMed 7 Jha P , Ramasundarahettige C , Landsman V et al. 21st-century hazards of smoking and benefits of cessation in the United States . N Engl J Med 2013 ; 368 : 341 – 50 . Google Scholar CrossRef Search ADS PubMed 8 Doll R , Peto R , Boreham J , Sutherland I. Mortality in relation to smoking: 50 years' observations on male British doctors . BMJ 2004 ; 328 : 1519 . Google Scholar CrossRef Search ADS PubMed 9 Mai ZM. Quitting smoking and mortality: a population-based mortality case-control study in Hong Kong . School of Public Health, Li Ka Shing Faculty of Medicine, University of Hong Kong , 2016 . 10 Lam TH , He Y , Shi QL et al. Smoking, quitting, and mortality in a Chinese cohort of retired men . Ann Epidemiol 2002 ; 12 : 316 – 20 . Google Scholar CrossRef Search ADS PubMed 11 Jiang CQ , Xu L , Lam TH , Lin JM , Cheng KK , Thomas GN. Smoking cessation and carotid atherosclerosis: the Guangzhou Biobank Cohort Study—CVD . J Epidemiol Community Health 2010 ; 64 : 100 – 0–9 . Google Scholar CrossRef Search ADS 12 Lam TH , Xu L , Schooling CM , Chan WM , Lee SY , Leung GM. Smoking and mortality in a prospective cohort study of elderly Chinese in Hong Kong . Addiction 2015 ; 110: 502 – 10 . Google Scholar CrossRef Search ADS PubMed 13 He Y , Jiang B , Li LS et al. Changes in smoking behavior and subsequent mortality risk during a 35-year follow-up of a cohort in Xi'an, China . Am J Epidemiol 2014 ; 179 : 1060 – 70 . Google Scholar CrossRef Search ADS PubMed 14 Liu BQ , Peto R , Chen ZM et al. Emerging tobacco hazards in China: 1. Retrospective proportional mortality study of one million deaths . BMJ 1998 ; 317 : 1411 – 22 . Google Scholar CrossRef Search ADS PubMed 15 Gajalakshmi V , Peto R , Kanaka TS , Jha P. Smoking and mortality from tuberculosis and other diseases in India: retrospective study of 43 000 adult male deaths and 35 000 controls . Lancet 2003 ; 362 : 507 – 15 . Google Scholar CrossRef Search ADS PubMed 16 Sitas F , Urban M , Bradshaw D , Kielkowski D , Bah S , Peto R. Tobacco attributable deaths in South Africa . Tob Control 2004 ; 13 : 396 – 99 . Google Scholar CrossRef Search ADS PubMed 17 Jha P , Jacob B , Gajalakshmi V et al. A nationally representative case–control study of smoking and death in India . N Engl J Med 2008 ; 358 : 1137 – 47 . Google Scholar CrossRef Search ADS PubMed 18 Alam DS , Jha P , Ramasundarahettige C et al. Smoking-attributable mortality in Bangladesh: proportional mortality study . Bull World Health Organ 2013 ; 91 : 757 – 64 . Google Scholar CrossRef Search ADS PubMed 19 Sitas F , Egger S , Bradshaw D et al. Differences among the coloured, white, black, and other South African populations in smoking-attributed mortality at ages 35–74 years: a case-control study of 481 640 deaths . Lancet 2013 ; 382 : 685 – 93 . Google Scholar CrossRef Search ADS PubMed 20 Lam TH , Ho SY , Hedley AJ , Mak KH , Peto R. Mortality and smoking in Hong Kong: case-control study of all adult deaths in 1998 . BMJ 2001 ; 323 : 361 . Google Scholar CrossRef Search ADS PubMed 21 Lam TH , Ho SY , Hedley AJ , Mak KH , Leung GM. Leisure time physical activity and mortality in Hong Kong: case-control study of all adult deaths in 1998 . Ann Epidemiol 2004 ; 14 : 391 – 98 . Google Scholar CrossRef Search ADS PubMed 22 McGhee S , Ho SY , Schooling CM et al. Mortality associated with passive smoking in Hong Kong . BMJ 2005 ; 330 : 287 – 88 . Google Scholar CrossRef Search ADS PubMed 23 Ho SY , Schooling CM , Hui LL , McGhee SM , Mak KH , Lam TH. Soy consumption and mortality in Hong Kong: proxy-reported case-control study of all older adult deaths in 1998 . Prev Med 2006 ; 43 : 20 – 26 . Google Scholar CrossRef Search ADS PubMed 24 Schooling CM , Ho SY , Leung GM et al. Diet synergies and mortality—a population-based case-control study of 32 462 Hong Kong Chinese older adults . Int J Epidemiol 2006 ; 35 : 418 – 26 . Google Scholar CrossRef Search ADS PubMed 25 Ou CQ , Hedley AJ , Chung RY et al. Socioeconomic disparities in air pollution-associated mortality . Environ Res 2008 ; 107 : 237 – 44 . Google Scholar CrossRef Search ADS PubMed 26 Wang MP , Thomas GN , Ho SY , Lai HK , Mak KH , Lam TH. Fish consumption and mortality in Hong Kong Chinese—the LIMOR Study . Ann Epidemiol 2011 ; 21 : 164 – 69 . Google Scholar CrossRef Search ADS PubMed 27 Thomas GN , Wang MP , Ho SY , Mak KH , Cheng KK , Lam TH. Adverse lifestyle leads to an annual excess of 2 million deaths in China . PLoS ONE 2014 ; 9 : e89650 . Google Scholar CrossRef Search ADS PubMed 28 Plummer M. Improved estimates of floating absolute risk . Stat Med 2004 ; 23 : 93 – 104 . Google Scholar CrossRef Search ADS PubMed 29 IARC . Handbooks of Cancer Prevention, Tobacco Control, Vol. 11 . Reversal of Risk After Quitting Smoking . Lyon : International Agency for Research on Cancer , 2007 . 30 Godtfredsen N , Lam TH , Hansel T et al. COPD-related morbidity and mortality after smoking cessation: status of the evidence . Eur Respir J 2008 ; 32 : 844 – 53 . Google Scholar CrossRef Search ADS PubMed 31 U.S. Department of Health and Human Services . The Health Consequences of Smoking—50 Years of Progress . Atlanta GA : Office of the Surgeon General , 2014 . 32 Zaridze D , Brennan P , Boreham J et al. . Alcohol and cause-specific mortality in Russia: a retrospective case–control study of 48 557 adult deaths . Lancet 2009 ; 373 : 2201 – 14 . Google Scholar CrossRef Search ADS PubMed 33 Jiang GH , Zhang H , Li W et al. Study on smoking-attributed mortality by using all causes of death surveillance system in Tianjin . Zhonghua Liu Xing Bing Xue Za Zhi 2016 ; 37 : 381 – 83 . Google Scholar PubMed 34 Peto R , Lopez AD , Norheim OF. Halving premature death . Science 2014 ; 345 : 1272 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018; all rights reserved. Published by Oxford University Press on behalf of the International Epidemiological Association This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

International Journal of Epidemiology – Oxford University Press

Published: Feb 8, 2018

There are no references for this article.