A total of 6,296 men were recruited from different regions of China: the center (Hubei, not= 1151), south (Guangdong, not= 1038), north (Hebei, not= 1093), northeast (Shanxi, not= 1127), southwest (Guizhou, not= 921), and the east (Jiangsu, not= 966). After excluding 217 who did not complete questionnaires and 114 who did not provide information on smoking or alcohol exposure, 5,965 men were finally included in the analysis. The characteristics of the participants are shown in Table 1. The mean age was 55.93 ± 0.12 years, and men aged 40-49, 50-59, 60-69, 70-79 were 1,789 (30.0%), 1977 (33.1%), 1685 (28.2%), 514 (8.6%), respectively. There are 1,884 (31.6%) never smokers, 3,389 (56.8%) current smokers and 692 (11.6%) former smokers. The distribution of drinking status was as follows: 1,664 (27.9%) were non-drinkers, 2,188 (36.7%) were occasional drinkers, 1,798 (30.1%) were frequent drinkers and 315 (5.3%) were former drinkers.
There were 1297 (21.7%) participants who scored ≥ 5 on the BDI-SF and were considered to have depressive symptoms. Of 1297 men with depressive symptoms, 567 (43.7%), 595 (45.9%) and 135 (10.4%) had mild, moderate and severe depressive symptoms, respectively. Living in town (P= 0.001), no spouse (P= 0.004), and having a comorbidity (P
Association of smoking with depressive symptoms
Compared to never smokers (19.1%), the prevalence of depressive symptoms was significantly higher among current smokers (23.0%, P= 0.001) and former smokers (22.8%, P= 0.034; Table 1). While no significant differences were observed between current and former smokers (P= 0.917). After adjusting for confounders including residence, marital status, comorbidity, and alcohol use, the difference remained. Compared with never smokers, current and former smokers had 1.32 times (95% CI 1.15-1.53) and 1.25 times (95% CI 1.01-1 .55) a probability of suffering from depressive symptoms (Table 1). After stratification by age, the relationship was observed in the age subgroups 40-59 and 60-79, except that a similar prevalence of depressive symptoms was observed between those who had never smoked and those who had ever smoked (OR = 1.06, 95% CI 0.79-1.41) from the 40-59 age subgroup (Supplementary File 2). However, the severity of depressive symptoms was similar in men with different prior smoking status (P= 0.427) and after (P= 0.633) adjusting for confounders (Supplemental File 1).
Daily cigarette consumption
Among 3,389 current smokers, 747 (22.0%) smoked 20 cigarettes/day (Table 1). Compared to non-smokers (19.1%), men who smoked P= 0.092; Fig. 1A) in men who smoked > 20 cigarettes/day (20.7%), compared to non-smokers. The relationship was observed in the 60-79 age subgroup (Supplementary File 2). For the 40-59 age subgroup, the pattern of association remained in general, as the odds of depressive symptoms tended to be higher, although not statistically significant, in men who smoked 2= 0.697, P= 0.038): OR = 1.053 + 0.556 × (daily cigarette consumption)-0.179 × (daily cigarette consumption)2. Among current smokers, the risk of depressive symptoms was negatively associated with daily cigarette smoking (β = -0.17, P= 0.001), which was 15.6% lower (OR = 0.84, 95% CI 0.76 to 0.94) with daily cigarette consumption increasing by one level after adjusting for confounders ( table 1).
Compared to former smokers (22.8%), men who smoked P= 0.950) and > 20 cigarettes/day (P= 0.321) were similar to those of former smokers (Fig. 2A). The relationship was observed in both the 40-59 and 60-69 age subgroups (Supplementary File 2). However, the severity of depressive symptoms was similar among smokers with different levels of cigarette smoking without (P= 0.373) and with (P= 0.845) adjusting for confounders (Supplemental File 1).
Association of alcohol consumption with depressive symptoms
The prevalence of depressive symptoms was similar in never (21.0%), occasional (23.4%, P= 0.075) and frequent drinkers (19.3%, P= 0.204; Table 1). The prevalence of depressive symptoms was higher among former drinkers (27.6%, P= 0.010), however, it was not significant (OR = 1.31, 95% CI 0.99–1.74, Table 1) after adjusting for confounders. The relationship remained generally in the age subgroups of 40–59 and 60–79, except that a lower likelihood of depressive symptoms was observed among frequent drinkers (OR = 0.72, CI 95% 0.55 to 0.95) than among 60 to 79 year olds who have never drunk. subgroup (supplementary file 3). Severity of depressive symptoms was not associated with prior drinking status (P= 0.143) and after (P= 0.246) adjusting for confounders (Supplementary File 1).
Weekly alcohol consumption
Of 3,986 current drinkers, 2,052 (51.5%) drank 280 g/week (Table 1). Compared to people who had never drank (21.0%), those who drank P= 0.920; Figure 1B). While men who drank >280g/week (16.0%) had a 29.4% (OR=0.71, 95% CI 0.57–0.88; Fig. 1B) lower risk of symptoms depressive than those who had never drunk. The relationship remained generally within the 40-59 and 60-79 age subgroups (Supplementary File 3). The inverted J-shaped relationship between alcohol consumption and the prevalence of depressive symptoms has been illustrated in Fig. 1B and can be formulated by the following equation (R2= 0.990, P= 0.033): OR = 1.007 + 0.243 × (weekly alcohol consumption)-0.115 × (weekly alcohol consumption)2. Among current drinkers, the likelihood of depressive symptoms was negatively associated with weekly alcohol consumption (β = -0.22, P
Compared to former drinkers (27.6%), a similar prevalence of depressive symptoms was observed in men who drank 280 g/week (16.0%) compared to former drinkers (Fig. 2B). And the relationship was observed in both the 40-59 and 60-79 age subgroups (Supplementary File 3). Severity of depressive symptoms was not associated with weekly alcohol consumption without (P= 0.735) and with (P= 0.962) adjusting for confounders (Supplementary File 1).
No interaction was observed between smoking and alcohol consumption regarding depressive symptoms
Participants were classified into 12 groups according to smoking status (3 levels) and alcohol consumption status (4 levels). Analysis of variance for the factorial design was applied to compare BDI-SF scores among the fully crossover groups. The interactive effect of smoking and alcohol consumption was not significant (P= 0.962). It was further explored by adding the product term of smoking and alcohol consumption in the binary logistic regression model with the existence of depressive symptoms as the dependent variable, and the product term was not significant (P= 0.432, Table 1). It showed that despite the associations of depressive symptoms with smoking and alcohol consumption, they were not interacting.