Association between Smoking Status and Food and Nutrient Consumption in Japanese: a Large-Scale Cross-Sectional Study

Background: In Japan, in comparison with the rest of the world the death rate of lung cancer is low although the smoking rate is relatively high. This is the so-called “Japanese smoking paradox”. A healthy diet is proposed to attenuate the risk without quitting smoking. We here examined the relationships between smoking status (SS) and the consumption of food and nutrient in Japan. Materials and Methods: Totals of 5,587 men and 2,718 women were divided into three (non-smokers, smokers and heavy smokers) and two (non-smokers and smokers) groups, respectively, according to pack-year, which represents the amount of smoking over a long period. Food and nutrient consumption was estimated with a validated food frequency questionnaire. Using general linear models, food and nutrient consumption was estimated for each group in men and women, separately. Results: In men, SS was positively related to consumption of rice, 3 alcoholic beverages, carbohydrate, alcohol and other 8 foods/nutrients (p< 0.05 for all) and negatively to those of protein animal, fat, fatty acids, dietary fiber, isoflavones and 36 other foods/nutrients (p<0.05 for all). In women, SS was positively associated with intake of 13 foods/nutrients, while being negatively associated with those of rice, energy, dietary fiber, and 14 other foods/nutrients (p<0.05 for all). Conclusions: Our results support lower intake of vegetables and fruits rich in antioxidants, which are thought as preventive factors for many diseases, in smokers.


Introduction
Smoking is a major cause of lung cancer (International Agency for Research on Cancer, 2004).To prevent tobacco epidemic, the WHO Framework Convention on Tobacco Control was adopted in 2003 (World Health Organization, 2003).However, it is suggested numbers of countries will not accomplish a 30% reduction of tobacco use until 2025 (Bilano et al., 2015).Therefore, smoking is needed to cope with the current situation globally.Dietary consumption of fruits and foods containing carotenoids have probably decreased risk of lung cancer (World Cancer Research Fund American Institute for Cancer Research, 2007).Vegetables, fruits and the related nutrients have protective effects against lifestyle related disease (LSRD) such as cardiovascular disease (Nagura et al., 2009) and diabetes mellitus (Yao et al., 2014).In Japan, the percentages of smokers are still higher, while mortality and incidence and schizophrenia (Leon and Diaz, 2005;Shinozaki et al., 2011) are associated with smoking.PS is associated with SS.Thus, an effect on the past or present history of psychiatric disorder is needed to be considered, and PS levels are also needed to be investigated along with SS is investigated.Until now, however, psychiatric disorders and PS have not been taken into account yet (Poisson et al., 2012;Vlassopoulos et al., 2013).Therefore, in the present study, we examined the association between SS and dietary consumption of foods and nutrients in a large-scale cross-sectional study in Japan profoundly, excluding the subjects with psychiatric disorders and considering self-reported psychological stress (SRPS) as one of confounding factors.

Subjects
The examinees of a health checkup at a local health center were invited to participate as study subjects from July 2005 to March 2007.The aim of the study was explained face-to-face and written informed consent for participation in the study was obtained from 12,066 subjects.Of those subjects, after 3,761 (31.2%) were excluded, 5,587 (46.3%) men and 2,718 (22.5%) women were included in the analyses.Details of the study design and the selection of study subjects are presented in our previous studies (Mochizuki et al., 2011;Endoh et al., 2015).In brief, the study exclusion criteria were 1) age younger than 29 years old or older than 61 years old, 2) missing data on SRPS, 3) past or present history of depression or being treated with an anti-depressant or a tranquilizer, 4) total energy intake and body mass index (BMI) greater or less than 2 standard deviations from the mean, 5) no data on lifestyle factors, 6) less than 1 or over 65 hours of sleeping time per day, and 7) past or present disease history of stroke, myocardial infarction, or cancer.This study was conducted according to the guidelines of the Declaration of Helsinki and all procedures involving subjects were approved by the Ethics Committee of the University of Shizuoka.
A validated food frequency questionnaire (FFQ) (Kobayashi et al., 2011;Kobayashi et al., 2012) including 58 food items was used.Based on each food consumption, intakes of total energy and 42 nutrients were calculated, and then individual food and nutrient consumption was adjusted for total energy intake (mg, μg or g/1000 kcal/ DOI:http://dx.doi.org/10.7314/APJCP.2015.16.15.6527Association between Smoking Status and Diet in Japan

Statistical analyses
Statistical analysis was performed for men and women, separately.Characteristics of subjects were analyzed using general linear models for continuous variables, and chisquared test or Fisher's exact test for categorical variables.Using general linear models, adjusted mean intakes of food and nutrient were calculated by stratification of SS.The following variables were used as confounding factors: age (years), BMI (kg/m 2 ), dietary energy intake (kcal/day), alcohol consumption (g/day), sleeping time (hours/day), habitual exercise (yes or no), past or present history of hypertension and diabetes mellitus (yes or no), and SRPS.All p values were based on two-tailed tests.Statistical analyses were conducted using PASW Statistics for Windows, version 18.0 (SPSS Inc., Chicago, IL, USA); p values < 0.05 were statistically considered significant.

Results
Table 1 shows the characteristics of study subjects, according to SS.In men, age, BMI, energy, alcohol, sleeping time, percentages of history of hypertension and diabetes mellitus, habitual exercise, and SRPS were significantly different among the 3 groups (p<0.05 for all).In women, age, total energy intake, and alcohol were significantly different between the 2 groups (p<0.05 for all).
Table 3 shows adjusted means of nutrients in men and women, respectively, according to SS.In men, SS is positively associated with consumption of, carbohydrate, alcohol, energy percent of carbohydrate and alcohol, niacin, and sodium, and negatively with those of animal protein, fat, animal fat, vegetable fat, energy percent of protein and fat, saturated fatty acid (SFA), monounsaturated fatty acid (MUFA), polyunsaturated fatty acid (PUFA), n-3 PUFA, n-6 PUFA, soluble, insoluble, and total dietary fiber, daizein, genistein, retinol, carotene, retinol equivalent, vitamin D, E, K, B1, B2, B6, and B12, folate, pantothenic acid, potassium, calcium, magnesium, phosphorus, iron and zinc (p<0.05 for all).In women, SS is positively associated with energy percent of alcohol, n-6 PUFA/n-3 PUFA, dietary intake of niacin, and negatively with those of energy, soluble, insoluble and total dietary fiber, pantothenic acid, vitamin C, calcium copper and manganese (p<0.05 for all).

Discussion
Compared with previous studies, we showed thorough relationships between SS and dietary intakes of foods and nutrients in men.SS was positively associated with the consumption of rice, soda and some alcoholic beverages and negatively associated with those of soy products, some vegetables, fruits, vitamins and minerals.
Until today, the association between SS and consumption of foods and nutrients has not fully been investigated in Japanese (Kato et al., 1989;Dyer et al., 2003;Sugiura et al., 2009).SS was negatively associated with consumption of 17 foods, including 6 vegetables, soy products and citrus fruits and 28 nutrients, including PUFAs, dietary fibers, isoflavones, 12 vitamins and 6 minerals.It is widely known that vegetables and fruit and the related nutrients, i.e. vitamins have a protective role for LSRDs (Wang et al., 2014).Nevertheless our results suggested that SS is negatively associated with the consumption of foods and nutrients which have protective roles, but our study have 3 strengths.First, regarding "Japanese smoking paradox", our study results will be helpful to investigate the association between dietary consumption and risk of lung cancer in Japanese, even if we did not show the risk.Smoking status is one of important confounding factors according to Rothman et al. (2008) because consumption of foods and nutrients is associated with SS.Second, a dose-response relationship of SS by pack-year and consumption of foods and nutrients was enlarged.Third, psychiatric disorders were considered for relationships because of being related to smoking (Wilhelm et al., 2003;Leon and Diaz, 2005;Shinozaki et al., 2011).Psychiatric disorders are also increasing globally (Murray et al., 2012) as an important public health issue, but have not been considered previously (Poisson et al., 2012;Vlassopoulos et al., 2013).
We showed that SS was also positively associated with alcohol consumption in men and women.Drinking in addition to smoking is also a risk factor for LSRDs.Unhealthy lifestyle such as smoking, drinking, no or less exercise and obesity are associated each other (Heikkila et al., 2013).We have found that the interaction between drinking and SRPS in relation to diet in Japanese (Endoh et al., 2015).Physical activity is also proposed to decrease the risk of lung cancer (World Cancer Research Fund American Institute for Cancer Research, 2007).However, it is needed to examine the association between exercise and diet because of being not fully understood especially in Japanese.
We assessed the SS by pack year, which a scale to determine the amount a person has smoked during long time.There are two reasons to choose pack year as SS.Lifestyle factors such as smoking and diet are developed in long-term.Pack year is associated with age since it is linked to the amount of tobacco during a long period of time.However, the effect of age was controlled to some extent because consumption of foods and nutrients were adjusted for age.A dose-response relationship between smoking and dietary intakes of foods and nutrients has not been demonstrated previously (Kato et al., 1989;Dyer et al., 2003;Sugiura et al., 2009).
The present study had several limitations.First, a cause-effect relationship between SS and diet could not be clearly demonstrated in our cross-sectional study, but we excluded the subjects with psychiatric disorders, which are related to SS.In contrast to previous studies, we also assessed SS by pack-year to investigate a doseresponse relationship.These were not considered in previous studies.Second, we did not collect information about socioeconomic status, including household income, occupation, job position, and marital status (Galobardes et al., 2001).Third, we assessed SRPS with a single question.In our previous study, however, we have demonstrated that significant higher proportion of badly lifestyle (i.e., drinker, smoker and non-exerciser) was among the subjects with the higher level of SRPS (Endoh et al., 2015).Finally, we did not consider the starting age of smoking.Considering current SS is important because smoking cessation increases BMI (John et al., 2005).Regarding SS, however, we assessed SS by pack year and could analyze the linear relation between SS and dietary intake of food and nutrients.
Our results imply that SS is negatively associated with the consumption of foods and nutrients which are rich in antioxidants and have protective roles, and positively with dietary factors for increased risks of various diseases such as cancer, especially in men.Further studies are needed to clarify "Japanese smoking paradox".

Table 1 . Characteristics of Subjects According to Smoking Status a
*SRPS; Self-reported psychological stress; a Based on the general linear model for continuous variables