a Department of Public Health and Primary Care, University of Cambridge, Institute of Public Health, Robinson Way, Cambridge CB2 2SR, UK.
b MRC Dunn Human Nutrition Unit, Hills Road, Cambridge CB2 2DH, UK.
Dr Nick Wareham, Department of Public Health and Primary Care, University of Cambridge, Institute of Public Health, Robinson Way, Cambridge CB2 2SR, UK. E-mail: njw1004{at}medschl.cam.ac.uk
Abstract
Background Previous prospective studies have suggested that cigarette smoking may be associated with an increased risk of type 2 diabetes, but the possibility of confounding, particularly by dietary factors has not been fully examined.
Methods Cross-sectional analysis of the association between cigarette smoking and HbA1C, a marker of long-term glucose homeostasis in 2704 men and 3385 women, aged 4574 years who were recruited to a population-based study of diet and chronic disease.
Results Twelve per cent of men and 11% of women reported being current smokers. Mean HbA1C was lowest in never smokers, intermediate in former smokers and highest in current smokers. There was a dose-response relationship between HbA1C levels and number of cigarettes smoked per day and a positive association with total smoking exposure as measured by pack-years. The unadjusted increase in HbA1C for 20 pack-years of smoking was 0.12% (95% CI : 0.090.16) in men and 0.12% (95% CI : 0.080.17) in women. After adjustment for possible confounders including dietary variables, the values were 0.08% (95% CI : 0.04 0.12) and 0.07% (95% CI : 0.020.12) for men and women, respectively. Mean HbA1C was inversely related to time since quitting smoking in men.
Conclusions These results add support to the hypothesis that smoking has long-term effects on glucose homeostasis, an association that cannot be explained by confounding by dietary factors as measured in this study.
KEY MESSAGES
Keywords Smoking, glycosylated haemoglobin, diabetes mellitus, epidemiology
Accepted 17 November 2000
Cigarette smoking is known to cause transient elevations in blood glucose concentration1,2 and may also influence insulin sensitivity.36 Smokers tend to have lower body mass index (BMI) than non-smokers7,8 but are also more likely to have increased central adiposity.9,10 Several studies have found that current smokers have higher glycosylated haemoglobin concentrations compared to non-smokers.1114
These data suggest that smoking may be a risk factor for diabetes and prospective studies in both men1518 and women19 support the association. However, other prospective studies did not find an independent association between smoking and diabetes.2022 The role of confounding by dietary factors has not been investigated. Smokers have different dietary patterns compared to non-smokers2325 and the finding of increased risk among smokers could be due to diet rather than smoking per se.
Glycosylated haemoglobin (HbA1C) is a marker of long-term glucose homeostasis reflecting average blood glucose concentrations in the past 23 months. Microvascular complications of diabetes are associated with the concentration of HbA1C26 and HbA1C may predict cardiovascular disease.27 Investigating the association between smoking and HbA1C may clarify the role smoking plays in the risk of diabetes and its complications.
We, therefore, examined the cross-sectional relationship between cigarette smoking and glycated haemoglobin in a large population-based study of men and women, controlling for possible confounding by dietary factors.
Methods
Subjects and measurements
Subjects in this study were participants of the East Anglian component of the European Prospective Investigation into Cancer (EPIC-Norfolk), a multicentre international cohort designed to investigate the relationship between diet, cancer and chronic disease. The detailed design and operation of the study have been previously described.28,29 The intention of the Norfolk EPIC study was to recruit a cohort of 25 000 men and women aged 4574 years from the general population in a geographically circumscribed area which has relatively little outward migration in this age group. The primary objective was to create a cohort for prospective analysis. At baseline survey between 1993 and 1998, 77 630 men and women aged 4574 years were identified from general practice age-sex registers in Norfolk and invited to participate in the study. In all, 30 447 agreed to participate and provided informed consent and 25 633 volunteers attended for a health check which included a detailed health and lifestyle questionnaire. In November 1995, midway through the recruitment of this cohort, we introduced measurement of HbA1C. The sub-cohort that has been selected for this analysis consists of all individuals who had HbA1C measurement and on whom all data had been processed by July 1998 (n = 6089).
Smoking history was derived from yes/no responses to the questions Have you ever smoked as much as one cigarette a day for as long as a year? and Do you smoke cigarettes now? Current smokers recorded the number of cigarettes smoked each day. Subjects were asked to record the age at which they started to smoke and those who stopped smoking the age at which they gave up. The number of cigarettes smoked at age 20, 30, 40, 50 years was also recorded. Pack-years of cigarette consumption were calculated from these data assuming that smoking patterns indicated at each age applied to that decade of life. A pack-year was defined as 20 cigarettes a day for a year.
Participants were also asked to record their average diet over the past year by means of a food frequency questionnaire (FFQ) that listed food items and frequency categories. Nutrient intakes were calculated by multiplying the frequency of food consumption by standard portion weights to obtain weight of food consumed per day; these were then converted to nutrient intakes using food tables.29,30 Individuals who reported no alcohol consumption over the past year were considered teetotallers. Tertiles of alcohol consumption were created based on FFQ data.
Subjects were classified as vegetarians if they gave a positive response to the vegetarian option of the question Do you follow any particular diets? and as supplement takers if they answered yes to the question Have you taken any vitamins, minerals or other food supplements regularly during the past year (such as vitamin C, vitamin D, iron, calcium, fish oils, primrose oil, beta carotene, etc.)?
The question Did you have any further education at college or university after you left school? identified those who had tertiary education. Subjects were asked to choose among four options to describe the type and amount of physical activity involved in their work. These options were sedentary (most of time sitting), standing (most time standing or walking but no intense physical activity), physical work (handling heavy objects and use of tools) or heavy manual work (very vigorous physical activity). Subjects also recorded the hours spent each week on leisure-time physical activity during summer and winter.31 Hormone replacement therapy (HRT) use in women was derived from yes/no responses to the questions Have you ever received any hormone replacement therapy? and If yes, are you currently taking this treatment?
Subjects were asked about personal illness by the question Has the doctor ever told you that you have any of the following? Positive responses to the following options were used for analysis: high blood pressure (hypertension) requiring treatment with drugs, high blood cholesterol (hyperlipidaemia), angina, heart attack (myocardial infarction), stroke, other vascular disease (peripheral vascular disease), diabetes (excluding gestational diabetes) and cancer.
The group of individuals with known diabetes were defined as those who reported having been told by a doctor that they had diabetes, or by responding positively to the diabetes option of the question Have you modified your diet in the past year (give reasons)? Subjects who reported diabetes in parents or siblings were classified as having a positive family history.
Following completion of the questionnaire, participants were invited to attend the general practice surgery where research nurses performed a health check. Height and weight were measured with subjects in light clothing and with their shoes removed. Height was measured to the nearest 0.1 cm using a stadiometer. Weight was measured to the nearest 100 g using Salter scales. These were used to calculate the body mass index (BMI) as weight (kg)/height2 (m2). Waist circumference was measured at the smallest circumference between the ribs and iliac crest to the nearest 0.1 cm and hip circumference as the maximum circumference between the iliac crest and the crotch to the nearest 0.1 cm. These measurements were used to calculate the waist-to-hip ratio (WHR).
Ninety-five per cent of the cohort provided a non-fasting blood sample. Plasma vitamin C level was measured in citrated plasma, stored overnight in a dark box at 47°C, and then spun at 2100 g for 15 min at 4°C. Plasma was stabilized in a standardized volume of metaphosphoric acid then stored at 70°C. Plasma vitamin C concentration was estimated using a fluorometric assay within one week of sampling.32 The coefficient of variation was 5.6% at lower end of the range (mean = 33.2 µmol/l) and 4.6% at the upper end (mean = 102.3 mol/l). HbA1C was assayed using HPLC on a Biorad Diomat.33 The coefficient of variation was 3.6% at the lower end of the range (mean = 4.94%) and 3.0% at the upper end (mean = 9.76%).
Statistical analysis
Subjects who had completed the health and lifestyle questionnaire, FFQ, health check, had given blood for HbA1C and vitamin C measurement and had complete data entry by July, 1998 formed the study population. Individuals with self-reported diabetes (n = 174) were excluded from analysis since they may have changed their behaviour due to the diagnosis. Subjects were classified as never smokers, former smokers or current smokers with the latter group further subdivided by the number of cigarettes smoked daily. Statistical analysis was performed using SAS Version 6.12 (SAS Institute, Cary, NC). Significance testing was done using analysis of variance (ANOVA) for means and the 2 test for proportions. A value of P < 0.05 was used for statistical significance.
Results
The cohort defined for this analysis were the 2704 men and 3385 women subjects who had baseline HbA1C measurements and were recruited between 1995 and 1998 to the EPIC-Norfolk study. The prevalence (n) of self-reported diabetes in men by category of smoking history was 2.2% (21) in never smokers, 4.9% (75) in former smokers, 3.3% (4) in current smokers of 114 cigarettes per day and 1.0% (2) in current smokers of 15 cigarettes per day. Corresponding figures in women were 2.0% (38), 2.5% (29), 1.0% (2) and 1.7% (3), respectively. In the subsequent analyses, we excluded these 174 participants with self-reported diabetes, since their report of diet and lifestyle could have been modified by the diagnostic label.
Tables 1 and 2 show characteristics by smoking status in non-diabetic men and women, respectively. Mean HbA1C among former smokers was greater than among never smokers. Among current smokers, mean HbA1C increased with smoking exposure. Smokers were generally younger and thinner compared to non-smokers. Women who were current smokers had higher mean WHR compared to non-smokers. This pattern was not seen in men where former smokers had the greatest abdominal girths. Smokers consumed more saturated fat, less fibre and less vitamin C compared to non-smokers.
|
|
|
|
|
In this large population-based study, cigarette smoking was independently associated with higher HbA1C concentrations in both men and women. There was evidence of a dose-response relationship with number of cigarettes smoked in current smokers and pack-years of cigarette smoking in ever smokers. Among male former smokers there was also an inverse association between years since smoking cessation and HbA1C.
These results are consistent with those of other large epidemiological studies11,12 and smaller metabolic investigations13,14 that have found higher glycated haemoglobin concentrations in current smokers compared to non-smokers. However, none of these studies examined the effect of former smoking on glycated haemoglobin or whether there was a dose-dependent relationship. In addition, the role of confounding was not addressed.
Our results are unlikely to be due to chance but several potential biases need to be considered. As we have previously reported,29 the EPIC-Norfolk cohort has a lower smoking prevalence than nationally representative samples.34 The proportion of men who currently smoke was 14.8%, 11.0% and 9.8% in the age groups 4554, 5564 and 6574 years, respectively, compared to national figures of 28%, 25% and 20%. In women in the same age ranges, the figures for the EPIC-Norfolk cohort were 14.2%, 10.7% and 7.8% compared to 27%, 25% and 18% nationally. However, this comparison to the whole country data from the Health Survey for England 1993 may accentuate the difference in current smoking pre-valence in our sample as the East Region has a relatively low smoking prevalence in this age group compared to other regions.35 Even though there may be some degree of selection, bias is an unlikely explanation for our results since participants were unaware of their HbA1C status prior to the study. There is no reason to believe that smokers with high HbA1C concentrations would have differentially participated in the study. Any under- or over-reporting of smoking exposures would most likely be non-differential with respect to HbA1C status. In our analysis, we excluded individuals with self-reported diabetes who might have altered their smoking habits and lifestyle as a result of the diagnosis. It is possible that smoking exposure could have been influenced by the presence of major illnesses especially if these are smoking-related. However, the association was present when these individuals were excluded and remained significant after adjustment for major illness in multivariate models.
The possibility of confounding in this study was diminished as the association was independent of age, BMI, WHR, family history of diabetes, alcohol consumption, physical activity, tertiary education, major illness, any supplement use, vegetarianism, dietary saturated fat, dietary fibre, dietary vitamin C, plasma vitamin C status and HRT use in women. It is possible that adjustment for factors such as BMI, the degree of central adiposity and vitamin C may be over-adjustment as these could be on the causal pathway linking cigarette smoking and hyperglycaemia. One limitation of this study is that at the time of this analysis, we did not have data on social class and used tertiary education as a proxy measure. Social class may be associated both with cigarette smoking and glycated haemoglobin as a true confounding factor. Alternatively smoking may be part of the explanation for an observed association between social class and glycated haemoglobin. The unravelling of these relationships will be an important topic for future analyses. Some of the confounding factors considered in this study, such as dietary factors and physical activity, are measured with error and, therefore, we cannot exclude residual confounding as an explanation for these results.
Despite these limitations, the results support a causal relationship between smoking and HbA1C. Although we cannot be sure of the direction of causation in cross-sectional analysis, it is reasonable to assume, given the prospective data,1519 that smoking influences HbA1C rather than vice versa. The presence of a dose-response relationship and the inverse relationship since cessation of smoking also strengthen the inference about causality. There was a 0.20.3% absolute or about 5% relative difference in HbA1C between smokers and non-smokers. The effect was not large but was consistent and surprising given only a single measurement of HbA1C and a truncation of the distribution of HbA1C by the exclusion of people with self-reported diabetes.
The link between cigarette smoking and abnormalities of glucose homeostasis is biologically plausible as several studies have suggested that smoking may directly impair insulin sensitivity,36 one of the key determinants of glucose tolerance.36 This observation is not consistent in all studies,13,37 and at least part of the variation in findings between studies is attributable to study design and the extent to which confounding is removed. Previous studies have shown, as we have in this study, that smoking reduces overall obesity but accentuates its central deposition.9,10,38 Thus, the inconsistency of results relating smoking to measures of insulin sensitivity could be due to differences in how confounding by obesity is considered. An alternative explanation for an apparent effect of cigarette smoking on glucose tolerance would be through increased oxidative stress. This is known to be increased in cigarette smoking,39,40 and experimental evidence suggests that increased oxidative stress may impair insulin action.41,42
It is also difficult from previously published data to determine whether the effect of smoking is acute or chronic. Experimental data indicate that smoking causes only transient perturbations in glucose homeostasis1,3,13 but these data may underestimate the cumulative effects of cigarette smoke. The association of smoking with HbA1C suggests long-term effects that may lead to increased risk of diabetes and diabetic complications including cardiovascular disease.
Acknowledgments
The EPIC-Norfolk study is supported by grant funding from the Cancer Research Campaign, the Medical Research Council, the Stroke Association, the British Heart Foundation, the Department of Health, the Europe Against Cancer Programme Commission of the European Union and the Ministry of Agriculture, Fisheries and Food. Dr Wareham is a MRC Clinician Scientist Fellow. We thank the staff of EPIC for their invaluable contributions and Terry Elsey and colleagues of the University of Cambridge Department of Clinical Biochemistry who performed blood assays. We are indebted to the general practitioners who allowed us to approach people on their lists and to the people of Norfolk who took part in this study.
References
1 Janzon L, Berntorp K, Hanson M, Lindell SE, Trell E. Glucose tolerance and smoking: a population study of oral and intravenous glucose tolerance test in middle-aged men. Diabetologia 1983;25:8688.[ISI][Medline]
2 Sandberg H, Roman L, Zavodnick J, Kupers N. The effect of smoking on serum somatotropin, immunoreactive insulin and blood glucose levels of young adult males. J Pharmacol Exp Ther 1973;184:78791.[ISI][Medline]
3 Attvall S, Fowelin J, Lager I, Von-Schenck H, Smith U. Smoking induces insulin resistancea potential link with the insulin resistance syndrome. J Intern Med 1993;233:32732.[ISI][Medline]
4 Eliasson B, Attvall S, Taskinen MR, Smith U. The insulin resistance syndrome in smokers is related to smoking habits. Arterioscler Thromb 1994;14:194650.[Abstract]
5 Facchini FS, Hollenbeck CB, Jeppesen J, Chen YD, Reaven GM. Insulin resistance and cigarette smoking. Lancet 1992;339:112830.[ISI][Medline]
6 Ronnemaa T, Ronnemaa EM, Puukka P, Pyorala K, Laakso M. Smoking is independently associated with high plasma insulin levels in nondiabetic men. Diabetes Care 1996;19:122932.[Abstract]
7 Albanes D, Jones DY, Micozzi MS, Mattson ME. Associations between smoking and body weight in the US population: analysis of NHANES II. Am J Public Health 1987;77:43944.[Abstract]
8 Eisen SA, Lyons MJ, Goldberg J, True WR. The impact of cigarette and alcohol consumption on weight and obesity. An analysis of 1911 monozygotic male twin pairs. Arch Intern Med 1993;153:245763.[Abstract]
9 Shimokata H, Muller DC, Andres R. Studies in the distribution of body fat. III. Effects of cigarette smoking. JAMA 1989;261:116973.[Abstract]
10 Barrett-Connor E, Khaw KT. Cigarette smoking and increased central adiposity. Ann Intern Med 1989;111:78387.[ISI][Medline]
11 Simon D, Senan C, Garnier P, Saint-Paul M, Papoz L. Epidemiological features of glycated haemoglobin A1cdistribution in a healthy population. The Telecom study. Diabetologia 1989;32:86469.[ISI][Medline]
12 Modan M, Meytes D, Rozeman P et al. Significance of high HbA1 levels in normal glucose tolerance. Diabetes Care 1988;11:42228.[Abstract]
13 Nilsson PM, Lind L, Pollare T, Berne C, Lithell HO. Increased level of hemoglobin A1c, but not impaired insulin sensitivity, found in hypertensive and normotensive smokers. Metabolism 1995;44:55761.[ISI][Medline]
14 Urberg M, Shammas R, Rajdev K. The effects of cigarette smoking on glycosylated hemoglobin in nondiabetic individuals. J Fam Pract 1989;28:52931.[ISI][Medline]
15 Feskens EJ, Kromhout D. Cardiovascular risk factors and the 25-year incidence of diabetes mellitus in middle-aged men. The Zutphen Study. Am J Epidemiol 1989;130:110108.[Abstract]
16 Uchimoto S, Tsumura K, Hayashi T et al. Impact of cigarette smoking on the incidence of Type 2 diabetes mellitus in middle-aged Japanese men: the Osaka Health Survey. Diabet Med 1999;16:95155.[ISI][Medline]
17 Kawakami N, Takatsuka N, Shimizu H, Ishibashi H. Effects of smoking on the incidence of non-insulin-dependent diabetes mellitus. Replication and extension in a Japanese cohort of male employees. Am J Epidemiol 1997;145:10309.[Abstract]
18
Rimm EB, Chan J, Stampfer MJ, Colditz GA, Willett WC. Prospective study of cigarette smoking, alcohol use, and the risk of diabetes in men. BMJ 1995;310:55559.
19 Rimm EB, Manson JE, Stampfer MJ et al. Cigarette smoking and the risk of diabetes in women. Am J Public Health 1993;83:21114.[Abstract]
20 Medalie JH, Papier CM, Goldbourt U, Herman JB. Major factors in the development of diabetes mellitus in 10 000 men. Arch Intern Med 1975;135:81117.[Abstract]
21 Wilson PW, Anderson KM, Kannel WB. Epidemiology of diabetes mellitus in the elderly: the Framingham study. Am J Med 1986; 80(Suppl.5A):39.[ISI][Medline]
22
Perry IJ, Wannamethee SG, Walker MK, Thomson AG, Whincup PH, Shaper AG. Prospective study of risk factors for development of non-insulin dependent diabetes in middle aged British men. BMJ 1995;310:56064.
23 Walmsley CM, Bates CJ, Prentice A, Cole TJ. Relationship between cigarette smoking and nutrient intakes and blood status indices of older people living in the UK: further analysis of data from the National Diet and Nutrition Survey of people aged 65 years and over, 1994/95. Public Health Nutr 1999;2:199208.[Medline]
24
Ma J, Hampl JS, Betts NM. Antioxidant intakes and smoking status: data from the continuing survey of food intakes by individuals 1994 1996. Am J Clin Nutr 2000;71:77480.
25 Cade JE, Margetts BM. Relationship between diet and smoking is the diet of smokers different? J Epidemiol Community Health 1991; 45:27072.[Abstract]
26 Engelgau MM, Thompson TJ, Herman WH et al. Comparison of fasting and 2-hour glucose and HbA1c levels for diagnosing diabetes. Diabetes Care 1997;20:78591.[Abstract]
27 Park S, Barrett CE, Wingard DL, Shan J, Edelstein S. GHb is a better predictor of cardiovascular disease than fasting or postchallenge plasma glucose in women without diabetes. The Rancho Bernardo Study. Diabetes Care 1996;19:45056.[Abstract]
28
Riboli E, Kaaks R. The EPIC Project: rationale and study design. European Prospective Investigation into Cancer and Nutrition. Int J Epidemiol 1997;26(Suppl.1):S614.
29 Day N, Oakes S, Luben R et al. EPIC-Norfolk: study design and characteristics of the cohort. European Prospective Investigation of Cancer. Br J Cancer 1999;80(Suppl.1):95103.[ISI][Medline]
30
Bingham SA, Gill C, Welch A et al. Validation of dietary assessment methods in the UK arm of EPIC using weighed record, and 24-hour urinary nitrogen and potassium and serum vitamin C and caretenoids as biomarkers. Int J Epidemiol 1997;26(Suppl.1):S13751.
31
Pols MA, Peeters PH, Ocke MC, Slimani N, Bueno-de-Mesquita HB, Collette HJ. Estimation of reproducibility and relative validity of the questions included in the EPIC Physical Activity Questionnaire. Int J Epidemiol 1997;26(Suppl.1):S18189.
32 Vuilleumier JP, Keck E. Fluorometric assay of vitamin C in biological materials using a centrifugal analyser with fluorescence attachment. J Micronutrient Anal 1989;5:2534.[ISI]
33 Standing SJ, Taylor RP. Glycated haemoglobin: an assessment of high capacity liquid chromatographic and immunoassay methods. Ann Clin Biochem 1992;29:494505.[ISI][Medline]
34 Bennett N, Dodd J, Flatley J, Freet S, Boiling K. Health Survey for England, 1993. London: HMSO, 1995.
35 Health Survey for England, 1996. Available at http://www.official-documents.co.uk/document/doh/survey96/ehtitle.htm
36 Reaven GM. Role of insulin resistance in human disease. Diabetes 1988;37:1595607.[Abstract]
37 Wareham NJ, Ness EM, Byrne CD, Cox BD, Day NE, Hales CN. Cigarette smoking is not associated with hyperinsulinemia: evidence against a causal relationship between smoking and insulin resistance. Metabolism 1996;45:155156.[ISI][Medline]
38 Slattery ML, McDonald A, Bild DE et al. Associations of body fat and its distribution with dietary intake, physical activity, alcohol, and smoking in blacks and whites. Am J Clin Nutr 1992;55:94349.[Abstract]
39
Morrow JD, Frei B, Longmire AW et al. Increase in circulating products of lipid peroxidation (F2-isoprostanes) in smokers. Smoking as a cause of oxidative damage. N Engl J Med 1995;332:1198203.
40 Rahman I, Morrison D, Donaldson K, MacNee W. Systemic oxidative stress in asthma, COPD, and smokers. Am J Respir Crit Care Med 1996; 154:105560.[Abstract]
41 Paolisso G, D'Amore A, Volpe C et al. Evidence for a relationship between oxidative stress and insulin action in non-insulin-dependent (type II) diabetic patients. Metabolism 1994;43:142629.[ISI][Medline]
42 Paolisso G, Giugliano D. Oxidative stress and insulin action: is there a relationship?. Diabetologia 1996;39:35763.[ISI][Medline]