Shimane J. Med. Sci., Vol.29 pp.

71-78, 2013

Cumulative Number of Cigarettes Smoked Is an Effective Marker

to Predict Future Diabetes
Takeyasu KAKAMU1）2）, Tsuyoshi TANABE1）, Shigeto MORIWAKI1）, Hiroki AMANO1）,
Mikiko KOBAYASHI-MIURA1） and Yasuyuki FUJITA1）
1)
Department of Public Health, Shimane University Faculty of Medicine, Shimane, Japan
2)
Department of Hygiene and Preventive Medicine, Fukushima Medical University School of Medicine
（Received December 3, 2012; Accepted December 4, 2012）

To estimate a cutoff level for cumulative number evidence shows that smoking exacerbates some as-
of cigarettes smoked critical for diabetes prevention, pects of diabetes. The development of complications
we examined health-check data for 121 male resi- in patients with diabetes is largely dependent on
dents in Shimane Prefecture, Japan, from 1998 to the effectiveness of its management. In particular,
2005. We used the Brinkman Index（BI）calculated smokers suffer more severe diabetic complications
as the number of cigarettes smoked per day mul- ［10, 11］, and smoking adversely affects blood-
tiplied by the number of years of smoking for the glucose levels in these patients［10-13］.
index of cumulative number of cigarettes smoked. A consensus regarding the relationship between
Multivariate logistic-model analysis was conducted cigarette smoking and diabetes risk has not been
to determine the relationship between BI and the reached. The MONICA Project reported that com-
diabetes risk. We documented 26 new cases during pared to non-smokers, odds ratios（ORs）for the
the observation period. Multivariate-adjusted odds development of type 2 diabetes in male smokers
ratio（OR）for diabetes risk compared with BI of who smoked 15-19 cigarettes per day was 1.78, and
0 was 3.52（95% CI: 0.84-14.71）for BIs of 1-600 2.43 for those who smoked more than 20 cigarettes
and 10.19（95% CI: 2.38-43.64）for BI ≥ 601（p- per day［14］. Several studies in the US, Europe,
trend=0.002）. We found that the BI is an effective and Japan have shown that smokers display higher
marker for predicting future risk for diabetes, and risk and hazard ratios for impaired glucose toler-
indicated a BI of 600 as a useful cutoff value. ance（IGT）than non-smokers［7, 15-20］. In the
Framingham studies, in contrast, no relationship be-
Key words: Brinkman index, diabetes mellitus, glu- tween cigarette smoking or the development of dia-
cose intolerance, smoking, longitudinal study betes was observed［21］. Moreover, Hu et al. also
reported that current smokers were not more likely
to suffer from impaired glucose regulation［22］.
INTRODUCTION
We propose that these discrepancies arise because
Cigarette smoking has been linked to cardiovas- smoking status was not estimated accurately or
cular disease, cerebrovascular disease, diabetic reti- consistently across studies, and that the cumulative
nopathy, diabetic nephropathy, dyslipidemia, obesity, number of cigarettes smoked is a useful measure
cancer, and tooth loss among others［1-9］. Komiya for predicting diabetes risk. The Brinkman Index
et al. reported significantly increased visceral fat （BI）is such a measure, and has already been used
levels in obese heavy smokers［5］. Additionally, to relate smoking to poor health［23-26］. Further,
Anan et al. reported that BI values have a dose-
Correspondence: Tsuyoshi Tanabe, MD, Ph. D, response relationship with insulin resistance［27］.
Department of Public Health, Shimane University Faculty of The purpose of this study was to estimate a cutoff
Medicine level for BI values critical for diabetes prevention.
Enya-cho 89-1, Izumo, 693-8501, Japan
Tel: +81-853-20-2162
Fax: +81-853-20-2160
E-mail: ttanabe@med.shimane-u.ac.jp

71
72 Kakamu et al.

HbA1c to assess diabetes risk. Subjects were diag-

METHODS
nosed with diabetes risk if they were classified as
Study population “in need of observation”（5.5% ≤ HbA1c < 6.0%
Health checks were conducted on 273 male resi- or 100 mg/dl ≤ BS < 126 mg/dl）or “in need of
dents aged 20 years or older in a single town in medical care”（HbA1c ≥ 6.0%, BS ≥ 126 mg/dl,
Shimane Prefecture, Japan, in 1998. Fifty-seven or under medication）. If diabetes-management clas-
subjects either suspected of being at risk for diabe- sification data could not be obtained, those with
tes or who had medical histories related to diabetes HbA1c levels ≥ 5.5% were said to have diabetes
were excluded, and thus the original cohort group risk.
consisted of 216 men. This group was tracked from
2002 to 2005, and 121 subjects provided written Smoking status
consent to participate in the study. The data from Smoking status was determined by questionnaire
1998 were used as baseline, and the development of during the health checks between 2002 and 2005.
diabetes risk was determined using the results from Current smokers and former smokers were asked
check-ups conducted between 2002 and 2005. how many cigarettes they smoked per day and the
This study was approved by the ethics committee number of years spent smoking. Brinkman Index
of Shimane University Faculty of Medicine in 2003 （BI）values were calculated as the number of ciga-
（Notification no. 85）. rettes smoked per day multiplied by the number of
years of smoking［23］.
Health checks
Baseline health checks were conducted in 1998 Statistical analysis
based on the Health and Medical Service Law for Mean and standard deviations for each value ob-
the Aged. After fasting overnight, the subject’s tained from the 1998 health check were calculated,
height, weight, and blood pressure were measured, and the diabetes-risk group and the no-risk group
and a blood profile and urine analysis were per- were compared using Student’s t-tests. The me-
formed. A medical history interview was conducted dian, 25% quartile, and 75% quartile scores were
by a public health nurse, a questionnaire for health- calculated for TG and compared using the Mann-
related lifestyle issues（smoking, alcohol consump- Whitney U test.
tion, and physical activity）was given, and patients Subjects were grouped by smoking category.
were examined by a physician. The blood profile Multivariate logistic-model analysis was conducted
included total cholesterol（TC）, HDL-cholesterol for the development of diabetes risk, with odds
（HDL-C）, triglycerides（TG）, AST, ALT, γ-GTP, ratios（ORs）and 95% confidence intervals（CIs）
blood glucose（BS）, hemoglobin A1c（HbA1c）, calculated and p-trends derived with logistic mod-
and creatinine. The urine profile included protein, els. Confounding factors were determined from the
occult blood, and urinary glucose. The HbA1c value results of the 1998 health checkup. The respective
was determined by Japan Diabetes Society（JDS） cutoff values were defined as laboratory test val-
method which is 0.4% lower than National Glyco- ues that increase the risk of cardiovascular disease
hemoglobin Standardization Program（NGSP）value or a major complication of diabetes［30, 31］, and
［28］. Body-mass index（BMI）was calculated as laboratory test values used as a standard in diabe-
weight/height2（kg/m2）. All health checks were car- tes screening［32］. Finally, we selected age, BMI,
ried out at the same hospital, by the same organiza- blood pressure, HDL-C, TG, alcohol consumption,
tion. physical exercise, and follow-up year as confound-
ing factors.
Definition of diabetes risk ORs for the development of diabetes risk were
We used the diabetes-management classifica- obtained by age-adjusted logistic regression analysis
tions［29］found in the Law of Health and Medi- for each confounding factor. Multivariate-adjusted
cal Services for the Aged, and observed levels of logistic analysis was conducted with the confound-
 Cumulative number of cigarettes predicts diabetes 73

ing factors of age, BMI ≥ 25 kg/m2, hypertension period. Cumulative incidence rate was 21.5%. At
（systolic blood pressure ≥ 140 mmHg and/or dia- baseline level results obtained in 1998, HbA1c was
stolic blood pressure ≥ 90 mmHg）, HDLC < 40 significantly higher in those who later developed
mg/dl, TG ≥ 150 mg/dl, current drinking, exercise diabetes risk（p<0.001, Table 1）. Among the 121
less than three times/week, and the follow-up year, subjects, there were 53 BIs of 0,36 ranged from
and ORs and p values were also obtained. The cut- 1-600, and 32 were > 600. Of these, 19 were for-
off level of BI was changed from 400 to 800 and mer smokers, with BIs ranging from 1-600 in 13
the ORs were again compared. and > 600 in 6. Table 2 shows the diabetes-risk
The SAS software version 9.0（SAS Institute, status for each confounding factor. The group with
Inc., Cary, NC）was used to perform all statistical BMI ≥ 25 showed a significant relationship with
analyses. diabetes-risk development（p=0.018）, with an age-
adjusted OR of 4.08（95% CI: 1.27-13.10）. BI val-
ues and diabetes-risk status are shown in Table 3.
RESULTS
Smokers had higher incidence rates for developing
Of the 216 people in the cohort group established IGT than nonsmokers at every BI grade.
in 1998, 121 were followed from 2002 to 2005 and Adjusted odds ratios with cutoffs of BI values
gave consent to participate in the study, giving a for the development of diabetes risk are presented
follow-up rate of 56.0%. in Table 4. A significant dose-response relationship
The mean age（SD）was 61.9（13.6）years, was found for each cutoff value. Heavy smokers
ranging from 21 to 88 years. We documented 26 showed a significantly higher OR until the cutoff
new cases of diabetes risk during the observation BI-value reaches 600, but after the cutoff value ex-

Table 1. Baseline data（case vs. control）

case (n = 26) control (n = 95) P-value*

Age (years) † 66.1 (8.9) 60.8 (14.5) 0.08

2
Body mass index (kg/m ) † 23.0 (3.2) 22.1 (2.5) 0.21
Systolic blood pressure (mm Hg) † 124.3 (19.6) 127.2 (16.0) 0.47
Diastolic blood pressure (mm Hg) † 76.2 (10.3) 77.7 (10.7) 0.50
HDL-cholesterol (mg/dl) † 48.7 (13.8) 49.7 (13.6) 0.72
Blood sugar (mg/dl) † 101.9 (11.4) 97.2 (9.6) 0.06
Hemoglobin A1c (%)† 5.1 (0.2) 4.8 (0.2) < 0.01
Triglyceride (mg/dl) ‡ 124 (90 - 169) 112 (7 - 153) 0.29
Alcohol consumption §
Non-drinker 15 (57.6) 51 (53.7)
Ex-drinker 1 (3.9) 1 (1.1)
Current drinker 10 (38.5) 43 (45.2) 0.54
Brinkman Index§
0 7 (26.9) 46 (48.4)
1–600 8 (30.8) 28 (29.5)
601+ 11 (42.3) 21 (22.1) 0.07
*P-values are from student t-tests or Mann-Whitney U test or χ 2 (chi-square) tests comparing
cases against Without IGT
† Mean (SD)
‡ Median (25% quartile – 75% quartile)
§ n (%)
74 Kakamu et al.

Table 2. Adjusted odds ratio for diabetes risk

case (%) control (%) Age-adjusted OR*

(95% CI†)
Body mass index
< 25 kg/m2 19 (18.5) 84 (81.5) 1.0 (reference)
2
≥ 25 kg/m 7 (38.9) 11 (61.1) 4.08 (1.27–13.10)
Blood pressure (BP)
Systolic BP < 140 mm Hg
and diastolic BP < 90 mm Hg 21 (22.6) 72 (77.4) 1.0 (reference)
Systolic BP > 140 mm Hg
and/or diastolic BP ≥ 90 mm Hg 5 (17.9) 23 (82.1) 0.54 (0.17–1.69)
HDL cholesterol
≥ 40 mg/dl 19 (21.6) 69 (78.4) 1.0 (reference)
< 40 mg/dl 7 (21.2) 26 (78.7) 0.97 (0.39–2.41)
Triglyceride
< 150 mg/dl 17 (19.5) 70 (80.5) 1.0 (reference)
≥ 150 mg/dl 9 (26.5) 25 (73.5) 1.87 (0.70–5.01)
Alcohol consumption
non-drinker or ex-drinker 11 (20.0) 44 (80.0) 1.0 (reference)
current drinker 15 (22.7) 51 (77.3) 1.34 (0.53–3.31)
Physical exercise
≥ 3 times/week 9 (25.7) 26 (74.3) 1.0 (reference)
< 3 times/week 17 (19.8) 69 (80.2) 0.80 (0.31–2.05)

*OR: odds ratio

†CI: confidence interval

Table 3. Relationship between Brinkman Index significant predictors of diabetes risk when a BI
and diabetes risk
cutoff-value of 600 was used.
Brinkman Total (n) case (%) control (%)
Index
DISCUSSION
0 53 7 (13.2) 46 (86.8)
1– 400 24 5 (20.8) 19 (79.2) In this analysis of the relationship between dia-
401– 500 5 1 (20.0) 4 (80.0) betes risk and cigarette smoking, we found that the
501– 600 7 2 (28.6) 5 (71.4) BI is an effective marker for predicting diabetes
601– 700 7 3 (42.9) 4 (57.1) risk. A BI value > 600 indicated a significant risk
701– 800 5 1 (20.0) 4 (80.0) for diabetes. To our knowledge, this is the first
801– 20 7 (35.0) 13 (65.0) study to estimate successfully the risk of developing
diabetes by using BI as an indicator of the cumula-
ceeded 600, the OR for heavy smokers and other tive number of cigarettes smoked.
smokers was reversed. Multivariate-adjusted OR for Our study revealed that cigarette smoking increas-
diabetes risk compared with a BI of 0 was 3.52 es the risk of developing diabetes in all smokers.
（95% CI: 0.84-14.71）for BIs of 1-600 and 10.19 Although previous studies have related smoking to
（95% CI: 2.38-43.64）for BI ≥ 601（p-trend=0.002）. diabetes risk［6, 14-19］, our study provides a use-
For confounding factors, only BMI produced a sig- ful quantification of smoking history, and a cutoff
nificant OR. Thus, only BI and BMI remained as value above which the risk of developing diabetes
 Cumulative number of cigarettes predicts diabetes 75

Table 4. Adjusted odds ratio of the Brinkman Index for the development of diabetes risk

Brinkman n Diabetes Adjusted OR* (95% CI†)

Index risk (%) Age-adjusted p-trend multivariate-adjusted‡ p-trend
0 53 7 (13.2) 1.0 (reference) 1.0 (reference)
1– 400 24 5 (20.8) 4.07 (0.92–17.97) 3.24 (0.64–16.36)
401– 44 14 (31.8) 4.45 (1.47–13.50) 0.010 7.92 (2.01–31.22) 0.003

0 53 7 (13.2) 1.0 (reference) 1.0 (reference)

1– 500 29 6 (20.7) 3.53 (0.89–13.95) 3.30 (0.73–15.04)
501– 39 13 (33.3) 4.77 (1.54–14.76) 0.007 8.76 (2.17–35.43) 0.002

0 53 7 (13.2) 1.0 (reference) 1.0 (reference)

1– 600 36 8 (22.2) 3.68 (1.02–13.27) 3.52 (0.84–14.71)
601– 32 11 (34.4) 4.90 (1.53–15.74) 0.008 10.19 (2.38–43.64) 0.002

0 53 7 (13.2) 1.0 (reference) 1.0 (reference)

1– 700 43 11 (25.6) 4.76 (1.37–16.61) 5.34 (1.37–20.83)
701– 25 8 (32.0) 4.06 (1.20–13.76) 0.020 6.55(1.54–27.84) 0.009

0 53 7 (13.2) 1.0 (reference) 1.0 (reference)

1– 800 48 12 (25.0) 4.29 (1.29–14.34) 4.97(1.30–18.97)
801– 20 7 (35.0) 4.49 (1.26–16.01) 0.014 7.65 (1.72–34.09) 0.005
*OR: odds ratio
† CI: confidence interval
‡ adjusted for age, BMI ≥ 25 kg/m 2 , hypertension (systolic blood pressure ≥ 140 mmHg
and/or diastolic blood pressure ≥ 90 mmHg), HDL cholesterol < 40 mg/dl, triglyceride ≥ 150
mg/dl, current drinker, physical exercise (< 3 times/week), and follow-up year

sharply rises. This can be helpful when educating Further, heavy smoking is related to both the devel-
smokers about the dangers of cigarettes. opment of IGT and type-2 diabetes［14, 17, 20］.
The BI, calculated as the number of cigarettes This study predicts that smoking cessation before
smoked per day multiplied by the number of years BI reaches 600 may reduce the risk of developing
of smoking, is a useful index for estimating the diabetes.
cumulative number of cigarettes smoked over an Results also indicate that increased BMI is in-
entire lifetime. In previous studies, the BI was pri- dependently related to diabetes risk. Other factors
marily used as a marker of respiratory syndromes such as age, blood pressure, dyslipidemia, and lack
such as pneumonia and lung cancer［23-26］. Brink- of exercise are linked to type-2 diabetes［21, 22,
man determined that individuals with BIs > 600 are 34-36］. Indeed, obese people display elevated in-
heavy smokers［23］. However, this determination sulin resistance, which is factor in the development
varies depending on the specific disease, with heavy of diabetes［3, 27］. Some studies have reported
smokers having BIs > 400 for respiratory disease, that many young people smoke for the purpose of
and 500 for arteriosclerosis［24-26］. Our present losing weight［37, 38］, although other studies indi-
data suggest that a BI of 600 is the appropriate cate that smoking is likely a risk factor for future
cutoff value for heavy smoking with regard to dia- obesity［5］. Peltzer et al. reported that smoking
betes. This supports similar conclusions reached by history is significantly related to high weight and
Anan et al. , who indicate that a BI of 600 is the obesity among school children［39］. Our data pro-
cutoff level for increased insulin resistance［27］. vide a clear and useful example of health risks that
76 Kakamu et al.

increase with smoking, and can be used to educate sample size and a BI calculated from data at the
younger people. Further evidence that smoking leads follow-up health check, results revealed a significant
to a risk for diabetes is required. cut-off value at which the cumulative number of
Cigarette smoking causes increases in insulin re- cigarettes smoked can predict elevated blood-glucose
sistance［10, 11, 40-44］and postprandial blood- levels. The meaning and effectiveness of this cut-
glucose levels［5］. Our results suggest that BI val- off value should be further investigated. Further,
ues may predict diabetes-risk development, and that the number of confounding factors analyzed was
early guidance to quit smoking can be an important relatively high. Past studies have also used more
factor in preventing diabetes risk. We consider that than five confounding factors to estimate the rela-
this result is particularly useful for adolescents who tionship between cigarette smoking and diabetes,
have not yet reached high BI values. and experts think that developing diabetes results
This study shows the usefulness of the BI as a from a combination of many factors［4, 13, 17, 19,
means to quantify smoking consistently. Currently, 20］. Therefore, even with small sample sizes, test-
different health checks assess smoking status with ing many confounding factors is necessary. We de-
different questions. Some merely ask whether or not termined risk for diabetes using data from medical
a subject smokes, others ask about the daily number histories and HbA1c levels, information commonly
of cigarettes that the subject currently smokes, and available through health checks in Japan. Although
still others ask about both the number of cigarettes both blood-glucose and HbA1c levels are commonly
smoked daily and number of years spent smoking. used in Japan［29］, because the Ministry of Health,
Other alternatives include asking subjects to select Labor, and Welfare allows blood tests a short time
from categories relating to smoking status（as in after eating, HbA1c data are more accurate［45］.
the present study）, or asking subjects about the ac- Here, we found that the BI is an effective marker
tual number of cigarettes smoked. In health checks for predicting future risk for diabetes, and indicated
performed in Japan since April 2008, all subjects a BI of 600 as a useful cutoff value. These find-
have been asked whether or not they smoke, but ings may indicate the effectiveness of anti-smoking
confirming the number of cigarettes smoked per day measures as one way of preventing diabetes, and
or the number of years spent smoking has not been the importance of guiding young smokers to quit
necessary［29］. The present results indicate that smoking as soon as possible.
this practice may not be sufficient, and that ascer-
taining the cumulative number of cigarettes smoked
ACKNOWLEDGEMENTS
will more likely accurately estimate the risk of de-
veloping lifestyle-related diseases. The authors would like to thank the staffs of
Several methodological limitations to the present Izumo City Hall Taki Branch Office Welfare and
study warrant mention. First, this study was targeted Education Division for their cooperation.
for male, second, many confounding factors are This study was supported in part by the Japanese
used for the small subjects, finally, we determined Ministry of Education, Culture, Sports, Science, and
the diabetes risk from the health check. Although Technology through a Grant-in-Aid for Scientific
a Ministry of Health, Labour and Welfare survey Research B-15390201.
in 2007 showed that 38.2% of men and 33.7% of
women potentially have diabetes, our study only
REFERENCES
surveyed males. Because diabetes is a common dis-
ease found in Japanese men, with a 2010 survey １）Seidell JC, Cigolini M, Deslypere JP, Charzewska J,
showing that men smoke more than women（smok- Ellsinger BM and Cruz A（1991）Body fat distribu-
ing rate: males, 32.2%; women, 8.4%）, we deemed tion in relation to physical activity and smoking hab-
it worthwhile to limit our efforts to examining the its in 38-year-old European men. The European Fat
association between smoking and glucose tolerance Distribution Study. Am J Epidemiol 133: 257-265.
specifically in men. Additionally, despite a small ２）Jonas MA, Oates JA, Ockene JK and Hennekens
 Cumulative number of cigarettes predicts diabetes 77

CH（1992）Statement on smoking and cardiovas- 13）Targher G, Alberiche M, Zenere MB, Bo-

cular disease for health care professionals. Ameri- nadonna RC, Muggeo M and Bonora E（1997）
can Heart Association. Circulation 86: 1664-1669. Cigarette smoking and insulin resistance in pa-
３）Ichinohe M, Mita R, Saito K, Shinkawa H, Na- tients with noninsulin-dependent diabetes mellitus.
kaji S, Coombs M, et al.（2005）The prevalence J Clin Endocrinol Metab 82: 3619-3624.
of obesity and its relationship with lifestyle fac- 14）Meisinger C, Doring A, Thorand B and Lowel
tors in Jamaica. Tohoku J Exp Med 207: 21-32. H（2006）Association of cigarette smoking and
４）Hanioka T, Ojima M, Tanaka K and Aoyama tar and nicotine intake with development of type
H（2007）Relationship between smoking status 2 diabetes mellitus in men and women from the
and tooth loss: findings from national databases general population: the MONICA/KORA Augs-
in Japan. J Epidemiol 17: 125-132. burg Cohort Study. Diabetologia 49: 1770-1776.
５）Komiya H, Mori Y, Yokose T and Tajima N 15）Kawakami N, Takatsuka N, Shimizu H and
（2006）Smoking as a risk factor for visceral fat Ishibashi H（1997）Effects of smoking on the in-
accumulation in Japanese men. Tohoku J Exp cidence of non-insulin-dependent diabetes mellitus:
Med 208: 123-132. Replication and extension in a Japanese cohort of
６）Patja K, Jousilahti P, Hu G, Valle T, Qiao Q male employees. Am J Epidemiol 145: 103-109.
and Tuomilehto J（2005）Effects of smoking, 16）Hu FB, Manson JE, Stampfer MJ, Colditz G,
obesity and physical activity on the risk of type Liu S, Solomon CG, et al.（2001）Diet, life-
2 diabetes in middle-aged Finnish men and wom- style, and the risk of type 2 diabetes mellitus in
en. J Intern Med 258: 356-362. women. N Engl J Med 345: 790-797.
７）Unnikrishnan RI, Rema M, Pradeepa R, Deepa 17）Nakanishi N, Nakamura K, Matsuo Y, Suzuki
M, Shanthirani CS, Deepa R, et al.（2007）Prev- K and Tatara K（2000）Cigarette smoking and
alence and risk factors of diabetic nephropathy in risk for impaired fasting glucose and type 2 dia-
an urban South Indian population: the Chennai betes in middle-aged Japanese men. Ann Intern
Urban Rural Epidemiology Study（CURES 45）. Med 133: 183-191.
Diabetes Care 30: 2019-2024. 18）Sairenchi T, Iso H, Nishimura A, Hosoda T,
８）Woodward M, Zhang X, Barzi F, Pan W, Irie F, Saito Y, et al.（2004）Cigarette smoking
Ueshima H, Rodgers A, et al.（2003）The ef- and risk of type 2 diabetes mellitus among mid-
fects of diabetes on the risks of major cardio- dle-aged and elderly Japanese men and women.
vascular diseases and death in the Asia-Pacific Am J Epidemiol 160: 158-162.
region. Diabetes Care 26: 360-366. 19）Rimm EB, Manson JE, Stampfer MJ, Colditz
９）Valdez R, Narayan KM, Geiss LS and Engel- GA, Willett WC, Rosner B, et al.（1993）Ciga-
gau MM（1999）Impact of diabetes mellitus on rette smoking and the risk of diabetes in women.
mortality associated with pneumonia and influenza Am J Public Health 83: 211-214.
among non-Hispanic black and white US adults. 20）Willi C, Bodenmann P, Ghali WA, Faris PD
Am J Public Health 89: 1715-1721. and Cornuz J（2007）Active smoking and the
10）Paetkau ME, Boyd TA, Winship B and Grace risk of type 2 diabetes: a systematic review and
M（1977）Cigarette smoking and diabetic reti- meta-analysis. JAMA 298: 2654-2664.
nopaty. Deabetes 26: 46-49. 21） Wilson PW, Anderson KM and Kannel WB （1986）
11）Muhlhauser I, Sawicki P and Berger M（1986） Epidemiology of diabetes mellitus in the elderly.
Cigarette-smoking as a risk factor for macroprotein- The Framingham Study. Am J Med 80: 3-9.
uria and proliferative retinopathy in type 1（insulin- 22）Hu G, Lindstrom J, Valle TT, Eriksson JG,
dependent）diabetes. Diabetologia 29: 500-502. Jousilahti P, Silventoinen K, et al.（2004）Physi-
12）Lundman BM, Asplund K and Norberg A （1990） cal activity, body mass index, and risk of type 2
Smoking and metabolic control in patients with diabetes in patients with normal or impaired glu-
insulin-dependent diabetes mellitus. J Intern Med cose regulation. Arch Intern Med 164: 892-896.
227: 101-106. 23）Brinkman GL and Coates EO, Jr（1963）The
78 Kakamu et al.

effect of bronchitis, smoking, and occupation on Lakka TA, Lynch JW, Kaplan GA, et al.（1998）
ventilation. Am Rev Respir Dis 87: 684-693. Weight gain and the risk of developing insulin
24）Kojima S, Sakakibara H, Motani S, Hirose K, resistance syndrome. Diabetes Care 21: 1637-
Mizuno F, Ochiai M and Hashimoto S（2007）In- 1643.
cidence of chronic obstructive pulmonary disease, 35）Kahn SE, Zinman B, Haffner SM, O'Neill MC,
and the relationship between age and smoking in a Kravitz BG, Yu D, et al.（2006）Obesity is a
Japanese population. J Epidemiol 17: 54-60. major determinant of the association of C-reactive
25）Hu L, Sekine M, Gaina A, Wang H and Kaga- protein levels and the metabolic syndrome in type
mimori S（2007）Nested case-control study on 2 diabetes. Diabetes 55: 2357-2364.
associations between lung function, smoking and 36）Tanaka K, Ikeda J, Higashi A, Nakazawa A,
mortality in Japanese population. Environ Health Nakatani M, Irie H, et al.（2005）Lifestyle Char-
Prev Med 12: 265-271. acteristics of Underweight and Normal Weight
26）Kubozono T, Miyata M, Ueyama K, Hamasaki 20-29-Year-Old Females on Weight Loss Control.
S, Kusano K, Kubozono O, et al.（2011）Acute Jpn J Nutr Diet 3: 67-74.（Eng Abstr）
and chronic effects of smoking on arterial stiff- 37）Delnevo CD, Hrywna M, Abatemarco DJ and
ness. Circ J 75: 698-702. Lewis MJ（2003）Relationships between cigarette
27）Anan F, Takahashi N, Shinohara T, Nakagawa smoking and weight control in young women.
M, Masaki T, Katsuragi I, et al.（2006）Smok- Fam Community Health 26: 140-146.
ing is associated with insulin resistance and car- 38）Peltzer K and Pengpid S（2011）Overweight
diovascular autonomic dysfunction in type 2 dia- and obesity and associated factors among school-
betic patients. Eur J Clin Invest 36: 459-465. aged adolescents in Ghana and Uganda. Int J En-
28）Kashiwagi A, Kasuga M, Araki E, Oka Y, viron Res Public Health 8: 3859-3870.
Hanafusa T and Ito H（2012）International clini- 39）Kong C, Nimmo L, Elatrozy T, Anyaoku V,
cal harmonization of glycated hemoglobin in Hughes C, Robinson S, et al.（2001）Smoking is
Japan: From Japan Diabetes Society to National associated with increased hepatic lipase activity,
Glycohemoglobin Standardization Program values. insulin resistance, dyslipidaemia and early athero-
J Diabetes Invest 3:39-40Tahara Y and Shima K sclerosis in Type 2 diabetes. Atherosclerosis 156:
（2005）HbA1c, Glycoalbumin. Nippon Rinsho 63 373-378.
（Suppl 2）: 382-385.（in Japanese） 40）Facchini FS, Hollenbeck CB, Jeppesen J, Chen
29）Yoshida H and Yanai H（2006）Hypertriglyc- YD and Reaven GM（1992）Insulin resistance
eridemia. Nippon Rinsho 64 （Suppl 9）: 378-384. （in and cigarette smoking. Lancet 339: 1128-1130.
Japanese） 41）Hautanen A and Adlercreutz H（1993）Hyper-
30）Bujo H（2006）Hypoalphalipoproteinemia. Nip- insulinaemia, dyslipidaemia and exaggerated ad-
pon Rinsho 64（Suppl 9） : 385-390.（in Japanese） renal androgen response to adrenocorticotropin in
31）Cecil RL, Goldman L and Ausiello DA（2004）In: male smokers. Diabetologia 36: 1275-1281.
Cecil textbook of medicine. Saunders, Philadelphia. 42）Bott S, Shafagoj YA, Sawicki PT and Heise
32）Imatoh T, Sugie T, Miyazaki M, Tanihara S, T（2005）Impact of smoking on the metabolic
Baba M, Momose Y, Uryu Y and Une H（2009） action of subcutaneous regular insulin in type 2
Is heat shock protein 60 associated with type 2 diabetic patients. Horm Metab Res 37: 445-449.
diabetes mellitus? Diabetes Res Clin Pract 85: 43）Hashimoto Y and Futamura A（2007）Effects
208-212. of smoking on HbA1c and fasting plasma glucose.
33）Bagdade JD, Bierman EL and Porte Jr D（1967） J Japan Diab Soc 50: 373-377.（Eng Abstr）
The significance of basal insulin levels in the 44）Ministry of Health,Labour,and Welfare,Japan
evaluation of the insulin response to glucose in （2007）Standard program of health check-
diabetic and nondiabetic subjects. J Clin Invest up and health guidance. Tokyo: Ministry of
46: 1549-1557. Health,Labour,and Welfare,Japan.（in Japanese）
34）Everson SA, Goldberg DE, Helmrich SP,