Abstract
Aim
The relations between smoking and glycaemic parameters are not well explored. We compare HbA1c, fasting plasma glucose (FPG) and 2-hour plasma glucose (2H-PG) in current-, ex- and never-smokers.
Methods
This meta-analysis used individual data from 16 886 men and 18 539 women without known diabetes, in 12 DETECT-2 consortium studies and in the French D.E.S.I.R. and TELECOM studies. Means of the three glycaemic parameters in current-, ex- and never-smokers were modelled by linear regression, with study as a random factor. The I2 statistic evaluated heterogeneity among studies.
Results
HbA1c was 0.10 (95%CI:0.08,0.12) % [1.1 (0.9,1.3) mmol/mol] higher in current-smokers and 0.03 (0.01,0.05) % [0.3 (0.1,0.5) mmol/l] higher in ex-smokers, compared with never-smokers. For FPG, there was no significant difference between current- and never-smokers: −0.004 (−0.03,0.02) mmol/l but FPG was higher in ex-smokers: 0.12 (0.09,0.14) mmol/l. In comparison to never-smokers, 2H-PG was lower: −0.44 (−0.52,−0.37) mmol/l in current-smokers, with no difference for ex-smokers: 0.02 (−0.06,0.09) mmol/l. There was a large and unexplained heterogeneity among studies, with I2 always higher than 50%: after stratification by sex and adjustment for age and BMI, I2 changed little. In this study population, current-smokers had a prevalence of diabetes as screened by HbA1c, 1.30% higher and that screened by 2H-PG, 0.52% lower than in comparison to never-smokers.
Conclusion
Current-smokers had a higher HbA1c and a lower 2H-PG than never-smokers, across this heterogeneous group of studies; this will effect the chances of smokers being diagnosed with diabetes.
Keywords: HbA1c, FPG, 2H-PG, smoking
Introduction
The prevalence of Type 2 diabetes is still increasing. Providing current and future estimations of diabetes prevalence is useful for public health planning. It is also important to be aware of factors associated with screened-detected diabetes, to interpret these prevalences, and to provide appropriate prevention programs.
Smoking is a well recognised risk factor for cancers, cardiovascular disease and respiratory function (1), less so for diabetes. However, in a meta-analysis, Willi et al. reported a pooled adjusted relative risk (95% CI) for incident diabetes of 1.44 (1.31, 1.58) for current-smokers compared to non-smokers, and they concluded that the risk in active smokers was now established, and future studies should concentrate on whether the relation is causal (2). However, this is not easy to show, as there are a number of potential confounding factors, such as socioeconomic status, education level, adiposity, ethnicity. Further, an individual’s phenotype may change the chances that he is screen-detected as having diabetes according to whether fasting plasma glucose (FPG) or 2 hour plasma glucose (2H-PG) following an oral glucose tolerance test (OGTT) is used (3,4); the same may be the case for HbA1c.
The association between smoking status and glycaemic parameters HbA1c, FPG and 2H-PG has been studied, but few have studied all three parameters in the same population. A higher HbA1c was found in smokers in six studies [5-10]; three studies showed no difference in FPG between smokers and never smokers [9, 11, 12], one study showed a lower FPG [13]; 2H-PG was lower in smokers in two studies [9, 11] and higher in one study [12].
The aim of our study was to assess and to quantify the influence of smoking status on mean values of HbA1c, FPG and 2H-PG in populations of adults worldwide and so describe the potential impact of smoking in the context of screening for diabetes.
Methods
Study population
The DETECT-2 study (the Evaluation of Screening and Early Detection Strategies for Type 2 Diabetes and Impaired Glucose Tolerance) aims to study the detection of type 2 diabetes using existing epidemiological studies around the world, with data on FPG and 2H-PG. In our analysis, we included 12 studies from the DETECT-2 consortium that also had information on HbA1c and agreed to participate, as well as two additional French studies. The studies included from DETECT-2 were: the Australian Diabetes, Obesity and Lifestyle study (AusDiab), Australia [14], Japanese-Brazilian Diabetes study, Brazil [15], a population study in Qingdao, China [16], The Inter99 study, Denmark [17], Diabetes in Egypt Study, Egypt [18], the Finnish cohort of the Seven Countries Study, Finland [19], Chennai Urban Rural Epidemiology Study (CURES), India [20], the Kenyan Diabetes Study, Kenya [21], the Hoorn Study, the Netherlands [22], a Tonga population based study, Tonga [23], Seattle Japanese American Community Diabetes Study, from greater Seattle, USA [24], and the National Health and Nutrition Examination Survey (NHANES III), USA [25]. Two additional French studies included in the analyses were: the Data on an Epidemiological Study on the Insulin Resistance Syndrome (D.E.S.I.R.) study in a general population [26] and employees in the TELECOM Company in the Ile de France region [5].
All studies were approved by local ethics committees
BMI was calculated in all studies as (weight in kilogram)/(height in metres)2.
For the studies in Egypt and Kenya and in the French TELECOM study, the questionnaires did not distinguish between ex- and never-smokers, and ex-smokers have been grouped with never-smokers – to give two groups, current-smokers and non-smokers. We do not have data on the time since quitting smoking to be able to define ex-smokers precisely.
Assay methods for HbA1c and for glucose are detailed in the on-line Supplementary Material Table 1.
The original data of participants from each study were available for analysis. We excluded participants with known diabetes and those with missing data on gender, age, BMI, smoking status, FPG, HbA1c or 2H-PG, with one exception: the D.E.S.I.R. study has no data on 2H-PG. The final study population included 16 886 men and 18 539 women (35 425 participants).
Statistical analyses
The populations studied are described by their mean ± SD and by n (%). The mean differences (95% CI) in HbA1c, FPG and 2H-PG for ex- and current-smokers in comparison with never-smokers are presented, by study, in forest plots; all studies were combined, and the mean differences were evaluated using a mixed model, with study included as a random effect. The I2 statistic was calculated to evaluate the heterogeneity between studies [27]. We tested for interactions between smoking status with gender and with age and BMI classes, on the entire study population. The differences in the glycaemic parameters for ex- and current-smokers in comparison with never-smokers are shown, stratified by these covariates, as some of the interactions were significant. Further, we evaluated the sensitivity of the results by (i) stratifying on gender and adjusting for age and BMI; (ii) restricting analyses to the seven studies from Western populations with similar exposures to smoking; (iii) deleting the three studies that combined never- and ex- smokers; (iv) combining ex- and never-smokers in analyses.
From the pooled data we estimated for never-, ex- and current-smokers, the prevalence of screened diabetes, as defined by the three criteria: HbA1c≥ 6.5% [47.5 mmol/mol], FPG≥ 7.0 mmol/l and 2H-PG≥ 11.1 mmol/l. To quantify the effect of smoking on these prevalences, we calculated the mean differences between ex- and never-smokers and between current- and never-smokers, for each of the three glycaemic parameters. For the ex- and current-smokers, we then adjusted the participants’ values for the three glycaemic parameters, by subtracting these mean differences, to estimate the values of the glycaemic parameters had each individual been a non-smoker. Thus, the mean values of the glycaemic parameters were adjusted so that they were equal for never-, ex- and non-smokers. As an example, the mean FPG among ex-smokers was 0.12 mmol/l higher than for never-smokers (Supplementary Material Table 2); 0.12 mmol/L was subtracted from the FPG value of each ex-smoker, to give an adjusted FPG value, and diabetes prevalence was calculated on these adjusted FPG values.
SAS version 9.3 was used in all statistical analyses.
Results
Characteristics differed significantly among cohorts (p<0.0001): mean age varied between 37 ± 10 (Kenyan women) and 76±4 years (Finnish men) and the highest mean BMI was in Tongan women 35.2±6.3 kg/m2. The means of glycaemic parameters (HbA1c, FPG, 2H-PG) differed among cohorts: the highest mean HbA1c (6.4±1.0%, [46.4±10.9 mol/mol]) was in the Japanese-American men, and the highest mean FPG (7.7±1.5 mmol/l) and 2H-PG (11.5±5.1 mmol/l) were in Japanese-Brazilian men (Table 1).
Table 1.
Characteristics (mean ± SD and n (%)) of men (n=16 886), by study
| Country, Study [reference] |
Years of Study |
n | Age (years) | BMI (kg/m2) | HbA1c (%) |
HbA1c (mmol/mol) |
FPG (mmol/l) |
2H-PG (mmol/l) |
Never smoked |
Ex-smoker | Current smoker |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Australia, AusDiab [14] | 1999-2000 | 4 542 | 51±14 | 27.1 ±4.0 | 5.2±0.4 | 33.3±4.4 | 5.6±0.8 | 6.2±2.3 | 2147 (47%) |
1575 (35%) |
820 (18%) |
|
Brazil, Japanese- Brazilians [15] |
2000 | 145 | 59±12 | 26.5 ±4.1 | 6.1±1.0 | 43.2±10.9 | 7.7±1.5 | 11.5±5.1 | 74 (51%) |
56 (39%) |
15 (10%) |
| China, Qingdao [16] | 2001-2002 | 97 | 55±10 | 26.7 ±3.0 | 5.2±0.6 | 33.3±6.6 | 5.3±1.3 | 6.9±3.4 | 57 (59%) |
8 (8%) |
32 (33%) |
| Denmark, Inter99 [17] | 1999-2001 | 3 055 | 46±8 | 27.0 ±4.0 | 5.9±0.5 | 41.0±5.5 | 5.7±0.8 | 6.2±2.3 | 1114 (36%) |
806 (26%) |
1135 (37%) |
|
Egypt, Diabetes in Egypt Project [18] |
1991-1993 | 372 | 46±16 | 27.0±5.7 | 5.6±1.5 | 37.7±16.4 | 5.5±2.4 | 6.6±4.0 | 201 (54%) |
- | 171 (46%) |
|
Finland, Seven Countries Study [19] |
1989 | 314 | 76±4 | 26.3±3.7 | 5.5±0.6 | 36.6±6.6 | 5.7±0.8 | 7.6±2.7 | 83 (26%) |
186 (59%) |
45 (14%) |
| France-D, D.E.S.I.R. [26] | 1994-1996 | 2 479 | 47±10 | 25.4±2.9 | 5.5±0.5 | 36.6±5.5 | 5.5±0.7 | - | 846 (34%) |
924 (37%) |
709 (29%) |
|
France-T, TELECOM [5] |
1985-1987 | 1 869 | 39±12 | 24.2±2.9 | 5.0±0.6 | 31.1±6.6 | 5.2±0.6 | 5.5±1.6 | 1211 (65%) |
- | 658 (35%) |
| India, CURES [20] | 2004-2006 | 1 002 | 40±13 | 22.8±3.9 | 5.9±1.2 | 41.0±13.1 | 5.2±1.7 | 7.0±3.8 | 601 (60%) |
97 (10%) |
304 (30%) |
| Kenya, Kenya [21] | 2005-2006 | 130 | 38±11 | 22.4±5.0 | 4.9±0.5 | 30.1±5.5 | 4.4±0.7 | 5.4±2.0 | 109 (84%) |
- | 21 (16%) |
| Netherlands, Hoorn [22] | 1989 | 1100 | 61±7 | 26.1±3.0 | 5.4±0.7 | 35.5±7.7 | 5.7±1.1 | 5.9±3.0 | 141 (13%) |
506 (46%) |
453 (41%) |
| Tonga, Tonga [23] | 1998, 2000 | 191 | 45±15 | 30.6±5.6 | 5.7±0.9 | 38.8±9.8 | 5.6±1.3 | 7.0±3.2 | 37 (19%) |
42 (22%) |
112 (59%) |
|
USA-J, Japanese- Americans [24] |
1983-1985 | 187 | 48±11 | 25.1±3.0 | 6.4±1.0 | 46.4±10.9 | 5.3±0.7 | 7.2±2.2 | 25 (13%) |
131 (70%) |
31 (17%) |
|
USA-N, NHANES III [25] |
1988-1992 | 1 403 | 56±11 | 27.3 ±4.5 | 5.6±0.8 | 37.7±8.7 | 5.8±1.4 | 7.2±3.5 | 390 (28%) |
573 (41%) |
440 (31%) |
D.E.S.I.R.: Data from an Epidemiological Study on the Insulin Resistance syndrome
CURES: Chennai Urban Rural Epidemiology Study
The mean HbA1c was higher in current-compared with never-smokers in ten of the 14 studies, significantly higher in seven studies; in the Egyptian study, the mean HbA1c was 0.33% lower (significantly lower) in current-smokers in comparison to never-smokers (Figure 1a, Supplementary Material Table 2). In most studies, the mean HbA1c in never- and ex-smokers were not significantly different. Combining all studies, and adjusting for study as a random factor, HbA1c was 0.10 (95% CI: 0.08,0.12) % [1.1 (0.9,1.3) mmol/l] higher in current-smokers and 0.03 (0.01, 0.05) % [0.3 (0.2,0.5) mmol/mol] higher in ex-smokers compared with never-smokers. This relation was consistent across gender, age and BMI strata for current-smokers, but not among the ex-smokers (Table 3).
Fig. 1.
(a) Difference in mean (95% CI) HbA1c between current-smokers and never-smokers is 0.10 (0.08,0.12) % [1.1 (0.9,1.3) mmol/mol]; between ex-smokers and never-smokers 0.03 (0.01,0.05) %.[0.3 (0.1,0.5)]. The I2 heterogeneity statistic was 93% and 55% respectively.
(b) Difference in mean (95% CI) FPG between current-smokers and never-smokers is −0.004 (−0.06,0.02) mmol/l; between ex-smokers and never-smokers 0.12 (0.09,0.14) mmol/l. The I2 heterogeneity statistic was 71% and 89% respectively.
(c) Difference in mean (95% CI) 2H-PG (c) between current-smokers and never-smokers is −0.44 (−0.52,−0.37) mmol/l; between ex-smokers and never-smokers 0.02 (−0.06,0.09) mmol/l. The I2 heterogeneity statistic was 88% and 62% respectively.
Table 3.
Differences of mean (95% CI) for HbA1c, FPG, 2H-PG for ex- and current-smokers in comparison with never-smokers, adjusted for study centre, as a random factor
| Gender | Age (years) | BMI (kg/m2) | ||||||
|---|---|---|---|---|---|---|---|---|
| Men | Women | <40 | 40-60 | >60 | <25 | 25-30 | >30 | |
| HbA1c (%) | ||||||||
| Ex-smoker | 0.08 [0.05,0.11] |
−0.06 [−0.08,−0.03] |
0.02 [−0.01,0.06] |
0.02 [0.00,0.05] |
0.03 [−0.004,0.06] |
0.001 [−0.02,0.03] |
0.03 [0.01,0.06] |
0.04 [0.01,0.10] |
| Current-smoker | 0.13 [0.11,0.15] |
0.05 [0.02,0.07] |
0.11 [0.09,0.14] |
0.15 [0.12,0.17] |
0.07 [0.03,0.12] |
0.11 [0.09,0.13] |
0.13 [0.10,0.15] |
0.11 [0.06,0.17] |
| pinte < 0.0001 | pint = 0.03 | pint = 0.57 | ||||||
| HbA1c (mmol/mol) | ||||||||
| Ex-smoker |
0.9
[0.5,1.2] |
−0.7
[−0.9,−0.3] |
0.2
[−0.1,0.7] |
0.2
[0.0,0.5] |
0.3
[−0.0,0.7] |
0.0
[−0.2,0.3] |
0.3
[0.1,0.7] |
0.4
[0.1,1.1] |
| Current-smoker |
1.4
[1.2,1.6] |
0.5
[0.2,0.8] |
1.2
[1.0,1.5] |
1.6
[1.3,1.9] |
0.08
[0.3,1.3] |
1.2
[1.0,1.4] |
1.4
[1.1,1.6] |
1.2
[0.7,1.9] |
| pint < 0.0001 | pint = 0.03 | pint 0.57 | ||||||
| FPG (mmol/l) | ||||||||
| Ex-smoker | 0.13 [0.09,0.17] |
−0.03 [−0.07,0.01] |
0.06 [0.01,0.11] |
0.10 [0.07,0.14] |
0.14 [0.08,0.19] |
0.07 [0.03,0.10] |
0.10 [0.06,0.15] |
0.15 [0.07,0.24] |
| Current-smoker | −0.03 [−0.06,0.01] |
−0.06 [−0.09,−0.02] |
0.002 [−0.03,0.04] |
0.05 [0.01,0.08] |
−0.01 [−0.08,0.07] |
0.01 [−0.02,0.04] |
0.04 [−0.002,0.08] |
0.06 [−0.03,0.15] |
| pint < 0.0001 | pint = 0.33 | pint 0.22 | ||||||
| 2H-PG (mmol/l) | ||||||||
| Ex-smokers | 0.29 [0.18,0.41] |
−0.25 [−0.35,−0.14] |
−0.02 [−0.16,0.12] |
0.02 [−0.12,0.08] |
−0.09 [−0.25,0.08] |
−0.10 [−0.21,0.08] |
0.04 [−0.08,0.15] |
0.03 [−0.17,0.24] |
| Current-smokers | −0.35 [−0.46.−0.24] |
−0.45 [−0.56.−0.35] |
−0.29 [−0.38.−0.19] |
−0.34 [−0.44,−0.24] |
−0.54 [−0.75,−0.33] |
−0.31 [−0.40,−0.22] |
−0.37 [−0.50,−0.25] |
−0.46 [−0.70,−0.23] |
| pint < 0.0001 | pint = 0.12 | pint = 0.16 | ||||||
pint indicates the p-value of the interaction between smoking status and the variable shown in the stratification
In the study of Japanese-Brazilians, the mean FPG was significantly higher by 0.66 mmol/l in current-smokers than in never-smokers, and in the Egyptian and French Telecom studies, it was significantly lower by 0.55 mmol/l and 0.06 mmol/l, respectively. However, in all other studies, the mean FPG did not differ between current- and never-smokers (Figure 1b, Supplementary Material Table 2). Combining studies, there was no significant difference in FPG between current- and never-smokers; ex-smokers had a 0.12 (0.09, 0.14) mmol/l higher FPG than never-smokers, and this relation was seen in all three age and BMI strata, and was consistent across age and BMI classes (Table 3).
The mean 2H-PG of current-smokers was lower than that of never-smokers in all studies, except for the Indian CURES study. When the studies were combined, 2H-PG in current-smokers was lower by −0.44 (−0.52, −0.37) mmol/l in comparison to never-smokers (Figure 1c, Supplementary Material Table 2). These relations were consistent over gender, age and BMI strata (Table 3). In contrast, 2H-PG did not differ among ex-smokers and never-smokers.
When we stratified on gender, and adjusted on age and BMI, the differences in the glycaemic parameters between current- and never-smokers were consistent with the unadjusted differences, and were also consistent in the seven studies from ‘western’ populations (Supplementary Material Table 2). The comparisons between ex- and never-smokers were less consistent (Supplementary Material Table 2).
As might be expected from the figures (Figure 1a, 1b, 1c), we found heterogeneity among studies with I2=93% for differences in mean HbA1c for current-smokers and 55% for ex-smokers in comparison with never-smokers; for FPG I2=71% for current-smokers and 89% for ex-smokers; for 2H-PG, I2=88% for current-smokers and 62% for ex-smokers. We stratified by gender, and adjusted for age and BMI to determine whether the resulting heterogeneity would be reduced (Supplementary Material Figures 1, 2, 3); this was the case for some of the comparisons; for the ex-smokers, the within study heterogeneity was high in comparison to the between study heterogeneity (Supplementary Material Table 3), but there were still some I2 values over 90%.
In a sensitivity analysis, we excluded the three studies where ex- and never-smokers were combined, but there was still a high heterogeneity among studies. Further, when we combined never- and ex-smokers as non-smokers, current-smokers compared with non-smokers had a significantly higher HbA1c: 0.09 (0.07, 0.11) %, [1.0 (0.8,1.2] mmol/mol] but significantly lower FPG: −0.04 (−0.06, −0.01) mmol/l and 2H-PG: −0.45 (−0.51, −0.40) mmol/l.
As gender, age and BMI are correlated with the three glycemic parameters, we tested the interactions among smoking status and strata of these variables. Interactions with sex were significant, but those with age and BMI strata were not significant except for smoking with age for HbA1c (Table 3).
For current-smokers compared with never-smokers, HbA1c showed a significantly greater difference in men than in women, 0.13 (0.11, 0.15) % and 0.05 (0.02, 0.07) % [1.4 (1.2,1.6) and 0.5 (0.2,0.8) mmol/mol] respectively (Table 3). In contrast, while men who were ex-smokers had a higher HbA1c than never-smokers, women had a lower HbA1c. For the three age strata, the differences were all significant comparing current- and never-smokers; the differences were all positive but smaller for the ex-smokers. For the three BMI strata, the HbA1c differences were very similar for both current- and ex-smokers.
For FPG when analysed according to strata, it was mainly the ex-smokers for whom the FPG was significantly higher than for the never-smokers, whereas for current-smokers, women had a significantly lower FPG.
For 2H-PG and current-smokers, all strata had significantly lower values than the never-smokers. For ex-smokers the only significant difference was for gender (Table 3), and these results paralleled those for HbA1c: for ex-smokers, men had a higher and women had a lower 2H-PG than the never-smokers.
For the population included in this study, we show the prevalences of screened diabetes according to each of the three glycaemic parameters (Table 4), using the original crude data and then the adjusted prevalences for ex- and current-smokers, after subtracting constants from the concentrations for ex- and current-smokers, so that the mean values were identical to those of never-smokers. In current-smokers the prevalence of diabetes screened by HbA1c would change from 5.32% to 4.02%, a 1.30% lower prevalence of diabetes after adjustment; there was no change for FPG; for 2H-PG the crude prevalence was 3.36%, the adjusted prevalence 3.88%, thus the prevalence was 0.52% higher after adjustment. For ex-smokers, all prevalences were increased after adjustment: by 0.71% for HbA1c, by 0.87% for FPG and by 0.16% for 2H-PG.
Table 4.
Prevalences of screened diabetes according to the definitions for the three glycaemic parameters, before and after equalizing the means of the glycaemic parameters for never-, ex- and current-smokers, as described in the Methods section. The differences in diabetes prevalence are also shown.
| Definitions of diabetes |
Never-smokers | Ex-smokers | Current-smokers | |
|---|---|---|---|---|
| HbA1c ≥ 6.5 % (47.5 mmol/mol) |
Original data | 3.77% | 4.02% | 5.32% |
| Adjusted to never-smokers |
3.77% | 3.31% | 4.02% | |
| Difference in prevalence | - | 0.71% | 1.30% | |
| Fasting plasma glucose ≥ 7.0 mmol/l |
Original data | 3.49% | 4.23% | 3.20% |
| Adjusted to never-smokers |
3.49% | 3.41% | 3.20% | |
| Difference in prevalence |
- | 0.82% | 0.00% | |
| 2 hour plasma glucose ≥ 11.1 mmol/l |
Original data | 4.43% | 4.67% | 3.36% |
| Adjusted to never-smokers |
4.43% | 4.51% | 3.88% | |
| Difference in prevalence |
- | 0.16% | −0.52% |
Discussion
Current-smokers compared with never-smokers had a 0.10% higher HbA1c, no difference in FPG and a 0.44 mmol/l lower 2H-PG. These results remained consistent when never- and ex-smokers were combined. In most of the individual studies in our analyses, the mean HbA1c was significantly higher in current-smokers than in never-smokers; for FPG, the difference between smokers and never-smokers was not significant in 11 of the 14 studies and for 2H-PG the means were significantly lower in current-smokers than in never-smokers in 9 of the 14 studies. The differences in mean HbA1c, FPG and 2H-PG between never- and ex-smokers was not significant in about half of the studies.
One of the strengths of our work is the large number of participants from 14 studies from countries around the world. However, there will be measurement error in both the exposure to smoking and in the glycaemic outcome measures (HbA1c, FPG and 2H-PG) within each of the studies.
One of the limitations of our study is the high heterogeneity between studies, with I2 statistics, that measure the difference between individual studies, being higher than 50% for the three parameters (HbA1c, FPG and 2H-PG) for both current-smokers and ex-smokers. The heterogeneity statistic I2 was lower for some parameters after stratifying on gender and adjusting for age and BMI, but remained high for others; similar results were obtained when analyses were restricted to the more comparable Western populations, and when those studies that combined ex- and never-smokers were not included in analyses. As we study differences in the glycaemic measures between smoking status groups for each study, measurement error in the outcomes (differences) is less likely to be a reason for high heterogeneity, than smoking status, as reported in questionnaires, and as perceived by the participants who answered the questions on smoking.
Further, tobacco consumption was not coded in the same way for all studies: in the TELECOM study, those who had been ex-smokers for less than 6 months were considered to be current-smokers and only those who had stopped smoking for more than 6 months were considered non-smokers; in the studies in Kenya and Egypt ex-smokers were not coded. While the definition of never-smokers may be unambiguous, the time since ex-smokers had quit smoking may differ between studies.
There were also some differences in the outcome measures among studies: in Tonga, the OGTT was only performed in participants with fasting capillary glucose between 5.0 and 11.1 mmol/l or in those with fasting capillary glucose ≤ 5 mmol/l and HbA1c > 6.0% [42.1 mmol/mol], as well as in one in five participants with fasting glucose ≤5 mmol/l [23]. Mean HbA1c differed among studies, more than that of glucose, as we have shown in other analyses [28, 29]. Indeed, in the Japanese-Americans study [24], HbA1c had a much higher mean than in other studies. HbA1c was assayed over a long time period, and the techniques used were certainly not aligned with the recent International Federation of Clinical Chemistry standard [30], although the assays for some studies were aligned with the National Glycohemoglobin Standardization Program (NGSP) that was used for the DCCT and the UKPDS [31, 32]. However, as commented above, for the main analyses shown on the forest plots, we are dealing with differences within centres in comparison to never-smokers.
Finally, adding to the heterogeneity between studies are differences in socioeconomic levels and characteristics of participants among studies – for example, age, BMI.
Studies that have examined these glycaemic measurements as a function of tobacco consumption have in the main, found similar results to ours, despite being analysed differently, and adjusted for different factors.
In the meta-analysis by Willi et al. [2], smoking was convincingly shown to be associated with an increased risk of developing type 2 diabetes, even after taking into account different confounding factors and despite the fact that diabetes was not defined homogeneously over studies; however, a causal relation cannot be assumed. Various authors have hypothesised mechanisms that might generate changes in glucose metabolism, leading to diabetes [32, 33]. Several studies have shown that smoking reduces insulin sensitivity [33, 34], and this may be due to nicotine as it stimulates the secretion of insulin-antagonizing hormones such as cortisol, catecholamines and growth hormone which impair the action of insulin [33]. However Wareham et al. [11] concluded that a causal relation is unlikely, as it was attenuated after adjusting for age and BMI. Alterations in insulin secretion have not been shown to be associated with smoking [33]. It is also probable that smokers have other unhealthy behaviours and dietary habits as well as a low socio-economic status, and these are well recognised risk factors for diabetes [2].
It is of note that we (and others) have found associations between smoking and HbA1c and 2H-PG in opposing directions. One can speculate that the HbA1c assay techniques were affected by a metabolite of tobacco. A study in smokers and non-smokers using the International Federation of Clinical Chemistry reference assay which analyses only one molecular species of glycated A1c, would confirm or refute this hypothesis. Higgins et al. [35] have hypothesised that smoking may induce an increased permeability of the erythrocyte membrane to the passage of glucose, thus increasing HbA1c. Smoking has been associated with both lower arterial oxygen saturation and higher red cell 2,3-diphosphorglycerate concentrations. In vitro studies have demonstrated that deoxyhaemoglobin is glycated more rapidly than oxyhaemoglobin [36], and the rate of HbA1c formation is increased with elevated 2,3-diphosphorglycerate concentrations. Another possibility is the effect of tobacco smoking on red blood cell life span. Carbon monoxide, administered to two patients with sickle-cell disease was found to increase red blood cell life span [37]. Further studies are needed to assess the role of these factors.
Some authors reported that smoking inhibits gastric motor activity [38, 39], and may therefore affect the rate at which a meal (or the glucose in an OGTT) is absorbed. Other authors have studied the effect of tobacco on solid and liquid meals separately and found that cigarette smoking accelerates the emptying of the liquid component of meals [40]. In agreement with this report, Janzon et al. [41] showed a dose response relation between glucose and habitual cigarette consumption at 40 min following an OGTT, and at 120 min, the inverse relation, indicating a quicker absorption of glucose. This is corroborated in the European RISC Study [42]: age, gender and BMI adjusted glucose concentrations were identical at fasting in current- and never-smokers, significantly higher in current-smokers at 30 and 60 min, no difference at 90 min, and significantly lower in current-smokers at 120 min (unpublished data, B Balkau); smokers appear to have an accelerated gastric emptying during an OGTT. Further, the area under the glucose curve during the OGTT was significantly higher in the current-smokers, and this would presumably lead to a higher HbA1c.
If HbA1c is used to screen for diabetes, rather than the formerly considered gold standard 2H-PG, overall in our study population, using data from our study, for smokers the prevalence of screened diabetes would be 1.8% higher, due to the higher HbA1c and lower 2H-PG in current-smokers. Using HbA1c as a screening test should have the beneficial effect of preferentially identifying current-smokers with high HbA1c, and so provide early preventive measures – with an encouragement to stop smoking. For the ex-smokers, both HbA1c and 2H-PG were higher than in never-smokers, with a net difference of 0.55% in prevalence.
In conclusion, previous publications have shown that smoking is associated with an increased risk of diabetes. In our meta-analysis, we found that HbA1c was higher in smokers than in non-smokers, this was not the case for FPG and indeed 2H-PG was lower. The mechanisms explaining the higher level of HbA1c and the lower level of 2H-PG in smokers need to be further explored.
Supplementary Material
Table 2.
Characteristics (mean ± SD and n (%)) of women studied (n=18 539), by study.
| Country [reference] |
Years of Study |
n | Age (years) |
BMI (kg/m2) |
HbA1c (%) |
HbA1c (mmol/mol) |
FPG (mmol/l) |
2H-PG (mmol/l) |
Never smoked |
Ex-smoker | Current smoker |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Australia, AusDiab [14] | 1999-2000 | 5564 | 51±14 | 26.6±5.4 | 5.1±0.3 | 32.2±3.3 | 5.3±0.6 | 6.3±2.1 | 3459 (62%) |
1299 (23%) |
806 (14%) |
|
Brazil, Japanese- Brazilians [15] |
2000 | 151 | 59±11 | 25.9±3.9 | 6.0±0.8 | 42.1±8.7 | 7.2±1.2 | 10.5±3.9 | 135 (89%) |
6 (4%) |
10 (7%) |
| China, Qingdao [16] | 2001-2002 | 196 | 55±10 | 26.1±3.7 | 5.2±1.0 | 33.3±10.9 | 5.8±2.1 | 7.0±4.2 | 195 (99%) |
0 (0%) |
1 (0.5%) |
| Denmark, Inter99 [17] | 1999-2001 | 3058 | 46±8 | 25.7±4.9 | 5.7±0.4 | 38.8±4.4 | 5.3±0.6 | 6.2±1.9 | 1251 (41%) |
738 (24%) |
1069 (35%) |
|
Egypt, Diabetes in Egypt Project [18] |
1991-1993 | 521 | 44±13 | 31.6±7.3 | 5.7±1.3 | 38.8±14.2 | 5.9±2.2 | 7.5±4.0 | 458 (88%) |
- | 63 (12%) |
|
Finland, Seven Countries Study [19] |
1989 | ||||||||||
|
France-D, D.E.S.I.R. [26] |
1994-1996 | 2558 | 47±10 | 24.0±4.1 | 5.4±0.4 | 35.5±4.4 | 5.2±0.6 | - | 1752 (68%) |
405 (16%) |
401 (16%) |
|
France-T, TELECOM [5] |
1985-1987 | 1893 | 42±12 | 23.0±3.6 | 5.0±0.6 | 31.1±6.6 | 4.9±0.5 | 5.7±1.4 | 1454 (77%) |
- | 439 (23%) |
| India, CURES [20] | 2004-2006 | 1179 | 38±12 | 23.2 ±4.1 | 5.8±1.1 | 39.9±12.0 | 5.1±1.7 | 7.0±3.3 | 1179 (100%) |
0 (0%) |
0 (0%) |
| Kenya, Kenya [21] | 2005-2006 | 153 | 37±10 | 22.1±4.4 | 5.1±0.5 | 32.2±5.5 | 4.6±0.6 | 5.8±1.4 | 143 (93%) |
- | 10 (7%) |
| Netherlands, Hoorn [22] | 1989 | 1274 | 62±7 | 26.7±3.9 | 5.4±0.7 | 35.5±7.7 | 5.5±1.1 | 6.2±2.9 | 633 (50%) |
289 (23%) |
352 (27%) |
| Tonga, Tonga [23] | 1998, 2000 | 264 | 43±13 | 35.2±6.3 | 5.7±0.8 | 38.8±8.7 | 5.5±1.0 | 7.6±2.8 | 209 (79%) |
24 (9%) |
31 (12%) |
|
USA-J, Japanese- Americans [24] |
1983-1985 | 263 | 53±12 | 23.1±3.1 | 6.0±0.9 | 42.1±9.8 | 5.1±0.9 | 7.7±2.7 | 114 (43%) |
122 (46%) |
27 (10%) |
| USA-N, NHANES III [25] | 1988-1992 | 1465 | 55±10 | 28.4±6.3 | 5.5±0.8 | 36.6±8.7 | 5.6±1.5 | 7.2±3.4 | 816 (56%) |
330 (23%) |
319 (22%) |
D.E.S.I.R.: Data from an Epidemiological Study on the Insulin Resistance syndrome
CURES: Chennai Urban Rural Epidemiology Study
Acknowledgements
The DETECT-2 project was undertaken on the initiative of the World Health Organization and the International Diabetes Federation. DETECT-2: Task Group. S Colagiuri and CMY Lee (Australia), K Borch-Johnsen and D Vistisen (Denmark), B Balkau (France), JM Dekker (The Netherlands). CMY Lee was supported by an Australian National Health and Medical Research Council Training Fellowship. S Soulimane was supported by a PhD grant, from Paris-Sud University, France. J Shaw was supported by an Australian National Health and Medical Research Council Senior Research Fellowship. D.L. Christensen was supported by a DANIDA grant and by a grant from the Cluster of International Health, University of Copenhagen.
Abbreviations
- CURES
Chennai Urban Rural Epidemiology Study
- D.E.S.I.R.
Data from an Epidemiological Study on the Insulin Resistance syndrome
- FPG
fasting plasma glucose
- OGTT
oral glucose tolerance test
- 2H-PG
2-hour plasma glucose
Footnotes
Duality of interest Dorte Vistisen is employed by Steno Diabetes Center A/S, a research hospital working in the Danish National Health Service and owned by Novo Nordisk A/S. Dorte Vistisen, Knut Borch-Johnsen own shares in Novo Nordisk A/S. The other authors declare no other conflicts of interest.
Contribution statement DV and BB conceived this study; SS and BB analysed and interpreted the data; SS and BB drafted the article. Authors DS, CL, WHH, CL, SC, JES, PZZ, DM, SRSF, YD, LZ, TJ, JT, VM, DLC, LK, JMD, GN, CDAS, OL, WYF, DLL, MJMcN, KB-J, EJB, DV, BB have conceived and designed their original individual studies, contributed to the data collection in their individual studies; CMYL and DS were responsible for co-ordinating the data in the DETECT-2 Study. All authors have critically reviewed various drafts of the manuscript and all authors have approved the final version.
Appendix. Supplementary data
Supplementary data associated with this article can be found, in the on-line Appendix
References
- 1.Beaglehole R, Bonita R, Horton R, et al. Priority actions for the non-communicable disease crisis. Lancet. 2011;377:1438–1447. doi: 10.1016/S0140-6736(11)60393-0. [DOI] [PubMed] [Google Scholar]
- 2.Willi C, Bodenmann P, Ghali WA, Faris PD, Cornuz J. Active smoking and the risk of type 2 diabetes: a systematic review and meta-analysis. JAMA. 2007;298:2654–2664. doi: 10.1001/jama.298.22.2654. [DOI] [PubMed] [Google Scholar]
- 3.Sicree RA, Zimmet PZ, Dunstan DW, Cameron AJ, Welborn TA, Shaw JE. Differences in height explain gender differences in the response to the oral glucose tolerance test- the AusDiab study. Diabet Med. 1988;25:296–302. doi: 10.1111/j.1464-5491.2007.02362.x. [DOI] [PubMed] [Google Scholar]
- 4.DECODE Study Group on behalf of the European Diabetes Epidemiology Study Group Will new diagnostic criteria for diabetes mellitus change phenotype of patients with diabetes? Reanalysis of European epidemiological data. BMJ. 1998;317:371–375. doi: 10.1136/bmj.317.7155.371. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Simon D, Senan C, Garnier P, Saint-Paul M, Papoz L. Epidemiological features of glycated haemoglobin A1c-distribution in a healthy population. The Telecom Study. Diabetologia. 1989;32:864–869. doi: 10.1007/BF00297451. [DOI] [PubMed] [Google Scholar]
- 6.Yudkin JS, Forrest RD, Jackson CA, Ryle AJ, Davie S, Gould BJ. Unexplained variability of glycated haemoglobin in non-diabetic subjects not related to glycaemia. Diabetologia. 1990;33:208–215. doi: 10.1007/BF00404798. [DOI] [PubMed] [Google Scholar]
- 7.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;44:557–561. doi: 10.1016/0026-0495(95)90110-8. [DOI] [PubMed] [Google Scholar]
- 8.Sargeant LA, Khaw KT, Bingham S, et al. Cigarette smoking and glycaemia: the EPIC-Norfolk Study. European Prospective Investigation into Cancer. Int J Epidemiol. 2001;30:547–554. doi: 10.1093/ije/30.3.547. [DOI] [PubMed] [Google Scholar]
- 9.Cho NH, Chan JC, Jang HC, Lim S, Kim HL, Choi SH. Cigarette smoking is an independent risk factor for type 2 diabetes: a four-year community-based prospective study. Clin Endocrinol (Oxf) 2009;71:679–685. doi: 10.1111/j.1365-2265.2009.03586.x. [DOI] [PubMed] [Google Scholar]
- 10.Clair C, Bitton A, Meigs JB, Rigotti NA. Relationships of cotinine and self-reported cigarette smoking with hemoglobin A1c in the U.S.: results from the National Health and Nutrition Examination Survey, 1999-2008. Diabetes Care. 2011;34:2250–2255. doi: 10.2337/dc11-0710. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.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;45:1551–1556. doi: 10.1016/s0026-0495(96)90187-0. [DOI] [PubMed] [Google Scholar]
- 12.Frati AC, Iniestra F, Ariza CR. Acute effect of cigarette smoking on glucose tolerance and other cardiovascular risk factors. Diabetes Care. 1996;19:112–118. doi: 10.2337/diacare.19.2.112. [DOI] [PubMed] [Google Scholar]
- 13.Masulli M, Riccardi G, Galasso R, Vaccaro O. Relationship between smoking habits and the features of the metabolic syndrome in a non-diabetic population. Nutr Metab Cardiovasc Dis. 2006;16:364–370. doi: 10.1016/j.numecd.2005.06.005. [DOI] [PubMed] [Google Scholar]
- 14.Dunstan DW, Zimmet PZ, Welborn TA, et al. The Australian Diabetes, Obesity and Lifestyle Study (AusDiab)--methods and response rates. Diabetes Res Clin Pract. 2002;57:119–129. doi: 10.1016/s0168-8227(02)00025-6. [DOI] [PubMed] [Google Scholar]
- 15.Gimeno SG, Ferreira SR, Franco LJ, Hirai AT, Matsumura L, Moises RS. Prevalence and 7-year incidence of Type II diabetes mellitus in a Japanese-Brazilian population: an alarming public health problem. Diabetologia. 2002;45:1635–1638. doi: 10.1007/s00125-002-0963-x. [DOI] [PubMed] [Google Scholar]
- 16.Zhou X, Pang Z, Gao W, et al. Fresh vegetable intake and prevalence of diabetes in a Chinese population in Qingdao. Diabetes Res Clin Pract. 2011;92:137–142. doi: 10.1016/j.diabres.2010.12.034. [DOI] [PubMed] [Google Scholar]
- 17.Glumer C, Jorgensen T, Borch-Johnsen K. Prevalences of diabetes and impaired glucose regulation in a Danish population: the Inter99 study. Diabetes Care. 2003;26:2335–2340. doi: 10.2337/diacare.26.8.2335. [DOI] [PubMed] [Google Scholar]
- 18.Herman WH, Ali MA, Aubert RE, et al. Diabetes mellitus in Egypt: risk factors and prevalence. Diabet Med. 1995;12:1126–1131. doi: 10.1111/j.1464-5491.1995.tb00432.x. [DOI] [PubMed] [Google Scholar]
- 19.Stengard JH, Tuomilehto J, Pekkanen J, et al. Diabetes mellitus, impaired glucose tolerance and mortality among elderly men: the Finnish cohorts of the Seven Countries Study. Diabetologia. 1992;35:760–765. doi: 10.1007/BF00429097. [DOI] [PubMed] [Google Scholar]
- 20.Deepa M, Pradeepa R, Rema M, et al. The Chennai Urban Rural Epidemiology Study (CURES)--study design and methodology (urban component) (CURES-I) J Assoc Physicians India. 2003;51:863–870. [PubMed] [Google Scholar]
- 21.Christensen DL, Friis H, Mwaniki DL, et al. Prevalence of glucose intolerance and associated risk factors in rural and urban populations of different ethnic groups in Kenya. Diabetes Res Clin Pract. 2009;84:303–310. doi: 10.1016/j.diabres.2009.03.007. [DOI] [PubMed] [Google Scholar]
- 22.de Vegt F, Dekker JM, Ruhe HG, et al. Hyperglycaemia is associated with all-cause and cardiovascular mortality in the Hoorn population: the Hoorn Study. Diabetologia. 1999;42:926–931. doi: 10.1007/s001250051249. [DOI] [PubMed] [Google Scholar]
- 23.Colagiuri S, Colagiuri R, Na’ati S, Muimuiheata S, Hussain Z, Palu T. The prevalence of diabetes in the kingdom of Tonga. Diabetes Care. 2002;25:1378–1383. doi: 10.2337/diacare.25.8.1378. [DOI] [PubMed] [Google Scholar]
- 24.Fujimoto WY, Leonetti DL, Kinyoun JL, et al. Prevalence of diabetes mellitus and impaired glucose tolerance among second-generation Japanese-American men. Diabetes. 1987;36:721–729. doi: 10.2337/diab.36.6.721. [DOI] [PubMed] [Google Scholar]
- 25.Alexander CM, Landsman PB, Teutsch SM, Haffner SM. NCEP-defined metabolic syndrome, diabetes, and prevalence of coronary heart disease among NHANES III participants age 50 years and older. Diabetes. 2003;52:1210–1214. doi: 10.2337/diabetes.52.5.1210. [DOI] [PubMed] [Google Scholar]
- 26.Balkau B, Lange C, Fezeu L, et al. Predicting diabetes: clinical, biological, and genetic approaches: data from the Epidemiological Study on the Insulin Resistance Syndrome (DESIR) Diabetes Care. 2008;31:2056–2061. doi: 10.2337/dc08-0368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analysis. BMJ. 2003;327:557–560. doi: 10.1136/bmj.327.7414.557. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Soulimane S, Simon D, Shaw JE, et al. Comparing incident diabetes as defined by fasting plasma glucose or by HbA(1c). The AusDiab, Inter99 and DESIR studies. Diabet Med. 2011;28:1311–1318. doi: 10.1111/j.1464-5491.2011.03403.x. [DOI] [PubMed] [Google Scholar]
- 29.Soulimane S, Simon D, Shaw J, et al. HbA1c, fasting plasma glucose and the prediction of diabetes: Inter99, AusDiab and D.E.S.I.R. Diabetes Res Clin Pract. 2012;96:392–399. doi: 10.1016/j.diabres.2011.06.003. [DOI] [PubMed] [Google Scholar]
- 30.Consensus statement on the worldwide standardization of the hemoglobin A1C measurement. American Diabetes Association, European Association for the Study of Diabetes, International Federation of Clinical Chemistry and Laboratory Medicine, and the International Diabetes Federation Diabetes Care. 2007;30:2399–2400. doi: 10.2337/dc07-9925. [DOI] [PubMed] [Google Scholar]
- 31.Group DCCT The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group. N Engl J Med. 1993;329:977–986. doi: 10.1056/NEJM199309303291401. [DOI] [PubMed] [Google Scholar]
- 32.Manley S. Haemoglobin A1c--a marker for complications of type 2 diabetes: the experience from the UK Prospective Diabetes Study (UKPDS) Clin Chem Lab Med. 2003;41:1182–1190. doi: 10.1515/CCLM.2003.182. [DOI] [PubMed] [Google Scholar]
- 33.Eliasson B. Cigarette smoking and diabetes. Prog Cardiovasc Dis. 2003;45:405–413. doi: 10.1053/pcad.2003.00103. [DOI] [PubMed] [Google Scholar]
- 34.Facchini FS, Hollenbeck CB, Jeppesen J, Chen YD, Reaven GM. Insulin resistance and cigarette smoking. Lancet. 1992;339:1128–1130. doi: 10.1016/0140-6736(92)90730-q. [DOI] [PubMed] [Google Scholar]
- 35.Higgins T, Cembrowski G, Tran D, Lim E, Chan J. Influence of variables on hemoglobin A1c values and nonheterogeneity of hemoglobin A1c reference ranges. J Diabetes Sci Technol. 2009;3:644–648. doi: 10.1177/193229680900300404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Smith RJ, Koenig RJ, Binnerts A, Soeldner JS, Aoki TT. Regulation of hemoglobin AIc formation in human erythrocytes in vitro. Effects of physiologic factors other than glucose. J Clin Invest. 1982;69:1164–1168. doi: 10.1172/JCI110552. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Beutler E. The effect of carbon monoxide on red cell life span in sickle cell disease. Blood. 1975;46:253–259. [PubMed] [Google Scholar]
- 38.Kohagen KR, Kim MS, McDonnell WM, Chey WD, Owyang C, Hasler WL. Nicotine effects on prostaglandin-dependent gastric slow wave rhythmicity and antral motility in nonsmokers and smokers. Gastroenterology. 1996;110:3–11. doi: 10.1053/gast.1996.v110.pm8536873. [DOI] [PubMed] [Google Scholar]
- 39.Johnson RD, Horowitz M, Maddox AF, Wishart JM, Shearman DJ. Cigarette smoking and rate of gastric emptying: effect on alcohol absorption. BMJ. 1991;302:20–23. doi: 10.1136/bmj.302.6767.20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Grimes DS, Goddard J. Effect of cigarette smoking on gastric emptying. BMJ. 1978;2:460–461. doi: 10.1136/bmj.2.6135.460. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Janzon L, Berntorp K, Hanson M, Lindell SE, Trell E. Glucose tolerance and smoking: a population study of oral and intravenous glucose tolerance tests in middle-aged men. Diabetologia. 1983;25:86–88. doi: 10.1007/BF00250893. [DOI] [PubMed] [Google Scholar]
- 42.Hills SA, Balkau B, Coppack SW, et al. The EGIR-RISC STUDY (The European group for the study of insulin resistance: relationship between insulin sensitivity and cardiovascular disease risk): I. Methodology and objectives. Diabetologia. 2004;47:566–557. doi: 10.1007/s00125-004-1335-5. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.