5
A Prospective Study of Cigarette Smoking and the Incidence of Diabetes Mellitus among US Male Physicians JoAnn E. Manson, MD, DrPH, Umed A. Ajani, MBBS, MPH, Simin Liu, MD, ScD, David M. Nathan, MD, Charles H. Hennekens, MD, DrPH PURPOSE: To determine the association between cigarette smoking and the incidence of type 2 diabetes mellitus. SUBJECTS AND METHODS: We studied 21,068 US male physicians aged 40 to 84 years in the Physicians’ Health Study who were initially free of diagnosed diabetes mellitus, cardio- vascular disease, and cancer. Information about cigarette smok- ing and other risk indicators was obtained at baseline. The pri- mary outcome was reported diagnosis of type 2 diabetes melli- tus. RESULTS: During 255,830 person-years of follow-up, 770 new cases of type 2 diabetes mellitus were identified. Smokers had a dose-dependent increased risk of developing type 2 diabetes mellitus: compared with never smokers, the age-adjusted rela- tive risk was 2.1 (95% confidence interval [CI]: 1.7 to 2.6) for current smokers of $20 cigarettes per day, 1.4 (95% CI: 1.0 to 2.0) for current smokers of ,20 cigarettes per day, and 1.2 (95% CI: 1.0 to 1.4) for past smokers. After multivariate adjustment for body mass index, physical activity, and other risk factors, the relative risks were 1.7 (95% CI: 1.3 to 2.3) for current smokers of $20 cigarettes per day, 1.5 (95% CI: 1.0 to 2.2) for current smokers of ,20 cigarettes per day, and 1.1 (95% CI: 1.0 to 1.4) for past smokers. Total pack-years of cigarette smoking was also associated with the risk of type 2 diabetes mellitus (P for trend ,0.001). CONCLUSIONS: These prospective data support the hypoth- esis that cigarette smoking is an independent and modifiable determinant of type 2 diabetes mellitus. Am J Med. 2000;109: 538 –542. q2000 by Excerpta Medica, Inc. T ype 2 (noninsulin-dependent) diabetes mellitus is one of the most prevalent chronic diseases world- wide and affects more than 15 million people in the United States (1). The morbidity associated with dia- betes is substantial, including markedly increased risks of cardiovascular disease, renal failure, and blindness; eco- nomic costs have been estimated to approach $100 billion annually in the United States (2). Four large-scale pro- spective studies have raised the possibility that cigarette smoking may increase the risk of type 2 diabetes mellitus (3– 6), while three others found no association between smoking and diabetes (7–9). A causal association between cigarette smoking and type 2 diabetes mellitus is biologically plausible for sev- eral reasons. Smoking increases blood glucose levels after an oral glucose challenge (10 –12) and may impair insulin sensitivity (12). Smokers are more resistant than non- smokers to insulin-mediated glucose uptake and are more hyperinsulinemic in response to an oral glucose load. This insulin resistance associated with smoking may account, in part, for the dyslipidemia (decreased high- density lipoprotein [HDL] cholesterol and increased tri- glyceride concentrations) and increased risk of coronary heart disease among smokers (13). Although smokers tend to be thinner, cigarette smoking has also been linked to increased abdominal fat distribution and greater waist- to-hip ratio (14,15), which may affect glucose tolerance (16). Vascular changes and reduced blood flow to skeletal muscles in smokers may contribute to the insulin resis- tance (13). Further, smoking increases free radical oxida- tive damage and oxidative stress (17), factors that have been implicated in causing diabetes (18). Finally, nico- tine, carbon monoxide, or other chemical components of tobacco may have direct toxic effects on the pancreas, beta- cell function, and insulin receptor sensitivity (19 –21). In a large prospective cohort study of US male physicians, we investigated the relation between cigarette smoking and the incidence of type 2 diabetes mellitus. Participants in the Physicians’ Health Study were 40 to 84 years of age at entry and were observed for an average of 12 years. METHODS Study Sample The Physicians’ Health Study (22,23) was a randomized, double-blind, placebo-controlled trial designed to test whether low-dose aspirin and beta-carotene reduce the From the Division of Preventive Medicine (JEM, UAA, SL), Depart- ment of Medicine, Brigham and Women’s Hospital; the Diabetes Cen- ter and Department of Medicine (DMN), Massachusetts General Hos- pital; Harvard Medical School, the Department of Epidemiology (JEM), Harvard School of Public Health, Boston, Massachusetts, and the De- partments of Medicine, Epidemiology, and Public Health, University of Miami School of Medicine (CHH), Miami, Florida. Supported by research grants DK36798 and CA40360 from the Na- tional Institutes of Health. Requests for reprints should be addressed to JoAnn E. Manson, MD, DrPH, 900 Commonwealth Avenue East, Boston, Massachusetts 02215-1204. Manuscript submitted November 17, 1999, and accepted in revised form July 7, 2000. 538 q2000 by Excerpta Medica, Inc. 0002-9343/00/$–see front matter All rights reserved. PII S0002-9343(00)00568-4

A prospective study of cigarette smoking and the incidence of diabetes mellitus among us male physicians

Embed Size (px)

Citation preview

A Prospective Study of Cigarette Smokingand the Incidence of Diabetes Mellitus among

US Male Physicians

JoAnn E. Manson, MD, DrPH, Umed A. Ajani, MBBS, MPH, Simin Liu, MD, ScD,David M. Nathan, MD, Charles H. Hennekens, MD, DrPH

PURPOSE: To determine the association between cigarettesmoking and the incidence of type 2 diabetes mellitus.SUBJECTS AND METHODS: We studied 21,068 US malephysicians aged 40 to 84 years in the Physicians’ Health Studywho were initially free of diagnosed diabetes mellitus, cardio-vascular disease, and cancer. Information about cigarette smok-ing and other risk indicators was obtained at baseline. The pri-mary outcome was reported diagnosis of type 2 diabetes melli-tus.RESULTS: During 255,830 person-years of follow-up, 770 newcases of type 2 diabetes mellitus were identified. Smokers had adose-dependent increased risk of developing type 2 diabetesmellitus: compared with never smokers, the age-adjusted rela-tive risk was 2.1 (95% confidence interval [CI]: 1.7 to 2.6) for

current smokers of $20 cigarettes per day, 1.4 (95% CI: 1.0 to2.0) for current smokers of ,20 cigarettes per day, and 1.2 (95%CI: 1.0 to 1.4) for past smokers. After multivariate adjustmentfor body mass index, physical activity, and other risk factors, therelative risks were 1.7 (95% CI: 1.3 to 2.3) for current smokers of$20 cigarettes per day, 1.5 (95% CI: 1.0 to 2.2) for currentsmokers of ,20 cigarettes per day, and 1.1 (95% CI: 1.0 to 1.4)for past smokers. Total pack-years of cigarette smoking was alsoassociated with the risk of type 2 diabetes mellitus (P for trend,0.001).CONCLUSIONS: These prospective data support the hypoth-esis that cigarette smoking is an independent and modifiabledeterminant of type 2 diabetes mellitus. Am J Med. 2000;109:538 –542. q2000 by Excerpta Medica, Inc.

Type 2 (noninsulin-dependent) diabetes mellitus isone of the most prevalent chronic diseases world-wide and affects more than 15 million people in

the United States (1). The morbidity associated with dia-betes is substantial, including markedly increased risks ofcardiovascular disease, renal failure, and blindness; eco-nomic costs have been estimated to approach $100 billionannually in the United States (2). Four large-scale pro-spective studies have raised the possibility that cigarettesmoking may increase the risk of type 2 diabetes mellitus(3– 6), while three others found no association betweensmoking and diabetes (7–9).

A causal association between cigarette smoking andtype 2 diabetes mellitus is biologically plausible for sev-eral reasons. Smoking increases blood glucose levels afteran oral glucose challenge (10 –12) and may impair insulinsensitivity (12). Smokers are more resistant than non-

smokers to insulin-mediated glucose uptake and aremore hyperinsulinemic in response to an oral glucoseload. This insulin resistance associated with smoking mayaccount, in part, for the dyslipidemia (decreased high-density lipoprotein [HDL] cholesterol and increased tri-glyceride concentrations) and increased risk of coronaryheart disease among smokers (13). Although smokerstend to be thinner, cigarette smoking has also been linkedto increased abdominal fat distribution and greater waist-to-hip ratio (14,15), which may affect glucose tolerance(16). Vascular changes and reduced blood flow to skeletalmuscles in smokers may contribute to the insulin resis-tance (13). Further, smoking increases free radical oxida-tive damage and oxidative stress (17), factors that havebeen implicated in causing diabetes (18). Finally, nico-tine, carbon monoxide, or other chemical components oftobacco may have direct toxic effects on the pancreas, beta-cell function, and insulin receptor sensitivity (19–21).

In a large prospective cohort study of US male physicians,we investigated the relation between cigarette smoking andthe incidence of type 2 diabetes mellitus. Participants in thePhysicians’ Health Study were 40 to 84 years of age at entryand were observed for an average of 12 years.

METHODS

Study SampleThe Physicians’ Health Study (22,23) was a randomized,double-blind, placebo-controlled trial designed to testwhether low-dose aspirin and beta-carotene reduce the

From the Division of Preventive Medicine (JEM, UAA, SL), Depart-ment of Medicine, Brigham and Women’s Hospital; the Diabetes Cen-ter and Department of Medicine (DMN), Massachusetts General Hos-pital; Harvard Medical School, the Department of Epidemiology (JEM),Harvard School of Public Health, Boston, Massachusetts, and the De-partments of Medicine, Epidemiology, and Public Health, University ofMiami School of Medicine (CHH), Miami, Florida.

Supported by research grants DK36798 and CA40360 from the Na-tional Institutes of Health.

Requests for reprints should be addressed to JoAnn E. Manson, MD,DrPH, 900 Commonwealth Avenue East, Boston, Massachusetts02215-1204.

Manuscript submitted November 17, 1999, and accepted in revisedform July 7, 2000.

538 q2000 by Excerpta Medica, Inc. 0002-9343/00/$–see front matterAll rights reserved. PII S0002-9343(00)00568-4

risks of cardiovascular disease and cancer. Briefly, 22,071US male physicians 40 to 84 years of age at entry in 1982and free from prior myocardial infarction, stroke, andtransient cerebral ischemia were assigned at random us-ing a 2 3 2 factorial design to aspirin, beta carotene, bothdrugs, or placebo. A total of 1,003 men who reportedhaving diabetes mellitus, coronary heart disease, cerebro-vascular disease, or cancer before entry were excludedfrom the analyses.

QuestionnairesInformation was collected at baseline by mailed question-naires about previously diagnosed medical conditions,including diabetes mellitus, as well as about cigarettesmoking, height and weight, history of hypertension andhigh cholesterol levels, parental history of myocardial in-farction, frequency of vigorous exercise, alcohol use, andother health habits. Information about family history ofdiabetes was not ascertained. Study participants providedinformation about whether they had ever smoked ciga-rettes regularly (never, past only, or current) and, if cur-rently smoking, how many cigarettes per day on averagethey smoked.

Follow-upEvery 6 months for the first year and annually thereafter,participants were mailed brief questionnaires askingabout their compliance with the randomized treatmentassignment and any new medical diagnoses, includingdiabetes (with the approximate date of diagnosis). Be-cause the participants are physicians, medical recordswere not requested to confirm self-reports of diagnoseddiabetes. Owing to the age structure of the sample, allincident cases of diabetes were diagnosed after the age of40 years and were therefore classified as type 2 diabetesmellitus. For these analyses, follow-up was continued un-til October 1995. Vital status was known for all physi-cians, and follow-up information on morbidity was99.7% complete.

Statistical AnalysesParticipants were classified at baseline into one of thefollowing categories: never smoker, past smoker, or cur-rent smoker (subclassified into $20 cigarettes per dayand ,20 cigarettes per day). Incidence rates for type 2diabetes mellitus were obtained by dividing incident casesby person-years in each category of cigarette smoking,after adjustment for age. Age-adjusted relative risks werecomputed as the rate of occurrence of type 2 diabetesmellitus in a specific category of cigarette smoking di-vided by the corresponding rate among never smokers,after adjustment for age (1-year categories) and random-ized treatment assignment. Proportional hazards regres-sion models were used to adjust for age, treatment assign-ment, body mass index (defined as weight in kilogramsdivided by height in meters squared), history of hyperten-

sion, history of high serum cholesterol, parental historyof myocardial infarction before age 60 years, episodes ofvigorous exercise per week (,1, 1, 2 to 4, and 5 or moretimes per week), and alcohol consumption (daily, weekly,none). As the presence of coronary risk factors may leadto increased medical surveillance, these covariates wereincluded in the model to minimize confounding by thesevariables. The Mantel-extension test was used to assessthe overall trend between greater levels of smoking andthe risk of diabetes (24). Multivariate analyses of bodymass index, physical activity, and history of hypertensionas predictors of type 2 diabetes mellitus were also per-formed. In all multivariate models, we included availablecovariates that have been shown to be important risk fac-tors for type 2 diabetes mellitus. The population attribut-able-risk of diabetes from cigarette smoking was calcu-lated as the difference between the incidence rate for type2 diabetes mellitus in the total sample and that in neversmokers, divided by the incidence rate in the total sample(3100%). We calculated the 95% confidence intervals(CI) for each relative risk, and all P values were two-sided.

RESULTS

The prevalence of current smoking in the cohort was11%. Never smokers tended to be slightly younger, some-what thinner, more active physically, and less likely toconsume alcohol than either past or current smokers (Ta-ble 1). Other risk factors did not differ appreciably bysmoking status.

During 12 years of follow-up (255,830 person-years), atotal of 770 incident cases of diabetes were reported.Compared with never smokers, men who currentlysmoked $20 cigarettes per day were about twice as likelyto develop diabetes as never smokers; current smokers of,20 cigarettes per day were also at increased risk (Table2). A linear test for trend for increasing risk of diabeteswith increasing category of cigarettes was highly signifi-cant (P for trend ,0.001). In multivariate analyses, therelative risks of type 2 diabetes mellitus among currentsmokers remained substantially elevated. Past smokershad only a modest and nonsignificantly elevated risk (Ta-ble 2). Further control for body mass index in finer cate-gories (as a continuous variable) did not appreciably alterthese relative risk estimates.

Total pack-years of smoking was a strong predictor ofthe risk of diabetes (Table 3). However, regardless of totalpack-years of smoking, men who had quit smoking atleast 10 years before study entry did not have an increasedrisk of type 2 diabetes mellitus, with relative risks of 1.3(95% CI: 0.9 to 2.0) for those who quit smoking ,5 yearsbefore baseline, 1.3 (95% CI: 0.9 to 1.8) for those whoquit 5 to 10 years before baseline, and 1.1 (95% CI: 0.9 to1.3) for those who quit .10 years before baseline, as com-pared with never smokers.

Cigarette Smoking and Diabetes Mellitus among Male Physicians/Manson et al

November 2000 THE AMERICAN JOURNAL OF MEDICINEt Volume 109 539

The multivariate-adjusted relative risk of diabetes as-sociated with current smoking ($20 cigarettes daily) was1.7 (95% CI: 1.3 to 2.3), which was similar to the relativerisk associated with mild overweight (body mass index24.4 to 26.4 kg/m2), which had a relative risk of 2.1 (95%

CI: 1.5 to 2.8), and history of hypertension, which had arelative risk of 1.7 (95% CI: 1.4 to 2.0).

Because of the possibility that physicians who smokemay have been more likely to be diagnosed with (asymp-tomatic) diabetes mellitus as the result of an incidental

Table 1. Age-Adjusted Distribution of Baseline Variables by Cigarette Smoking Status in a Cohort of US Male Physicians 40 to 84Years of Age in 1982

Never Smokers(n 5 10,511)

Past Smokers(n 5 8,258)

Current Smokers

,20 Cigarettes/Day(n 5 811)

$20 Cigarettes/Day(n 5 1,488)

Percent or Mean 6 SD

Age group40–49 years 47.6 35.9 43.5 39.950–59 years 31.6 36.3 35.3 36.860–69 years 15.6 20.5 14.8 19.270–84 years 5.3 7.3 6.4 4.0

Age (years) 51.8 6 9.3 54.2 6 9.4 52.4 6 9.3 52.9 6 8.6Body mass index (kg/m2)

,23.0 26.7 23.6 23.5 22.923.0 to ,24.4 26.0 25.5 26.5 23.324.4 to ,26.4 24.3 24.8 27.5 24.6$26.4 23.0 26.1 22.5 29.2

Body mass index (kg/m2) 24.8 6 2.9 25.0 6 3.0 24.9 6 3.2 25.2 6 3.2History of hypertension* 12.4 13.7 12.6 13.8History of high cholesterol† 6.2 7.1 6.9 5.8Parental history of myocardial

infarction before age 60 years13.0 12.6 12.6 14.6

Vigorous exercise‡

Less than once a week 26.2 26.3 29.8 39.2At least weekly 73.8 73.7 70.2 60.8

Alcohol useDaily 18.1 30.1 31.2 38.0Weekly 49.4 51.9 50.0 39.7

* Defined as reported systolic blood pressure of 160 mm Hg or greater, diastolic blood pressure of 95 mm Hg or greater, or history of treatment forhigh blood pressure.† Defined as reported high cholesterol, reported blood cholesterol level of 260 mg/dL (6.8 mmol/L) or greater, or history of treatment with choles-terol-lowering medication.‡ Defined as physical activity long enough to work up a sweat.

Table 2. Cigarette Smoking and Risk of Type 2 Diabetes Mellitus during 12 Years of Follow-up

Cigarette SmokingCategory Person-Years Cases

Incidence per 1000Person-Years Age-Adjusted* Multivariate Adjusted†

Never smokers 129,090 323 2.5 1.0 (Referent) 1.0 (Referent)Past smokers 99,827 320 3.2 1.2 (1.0–1.4) 1.1 (1.0–1.4)Current smokers

,20 cigarettes/day 9,685 35 3.6 1.4 (1.0–2.0) 1.5 (1.0–2.2)$20 cigarettes/day 17,228 92 5.3 2.1 (1.7–2.6) 1.7 (1.3–2.3)

P for trend‡ ,0.001 ,0.001

* Adjusted for age (1-year categories) and treatment assignment (aspirin and beta-carotene).† Adjusted for age (1-year categories), body mass index (quartiles), physical activity (,1, 1, 2– 4, and 5 or more times per week), history ofhypertension, history of high cholesterol, parental history of myocardial infarction at age ,60, alcohol consumption (,monthly, monthly, weekly,daily), and treatment assignment.‡ Trend by smoking categories as listed in the Table.

Cigarette Smoking and Diabetes Mellitus among Male Physicians/Manson et al

540 November 2000 THE AMERICAN JOURNAL OF MEDICINEt Volume 109

laboratory test for a smoking-related problem than werephysicians who did not smoke, we repeated our analysisafter excluding cases of diabetes that occurred during thefirst 2 years of follow-up. We found essentially the samerelative risks: 1.8 (95% CI: 1.3 to 2.4) for current smokersof $20 cigarettes per day, 1.2 (95% CI: 0.8 to 1.8) forcurrent smokers of ,20 cigarettes per day, and 1.1 (95%CI: 0.9 to 1.4) for past smokers.

DISCUSSION

In this prospective study of US male physicians, currentcigarette smoking was associated with a substantial in-crease in the incidence of type 2 diabetes mellitus. A dose-response gradient between increased smoking and therisk of diabetes was observed. These increased risks per-sisted after adjustment for body mass index and othermeasured risk factors and health habits. The risk eleva-tion associated with cigarette smoking was similar to thatattributable to mild overweight or hypertension. Assum-ing the association between cigarette smoking and type 2diabetes mellitus is causal, our results suggest that in thegeneral US population—in which approximately 25% ofadults smoke (25)—about 10% (95 CI: 6% to 14%) of theincidence of type 2 diabetes mellitus may be attributableto cigarette smoking. For populations in which smokingis (or was) more common, greater population-attribut-able risks are likely. Cigarette smoking, therefore, maycontribute to the increased risk of type 2 diabetes mellitusobserved with the adoption of habits that are, or oncewere, common in developed countries.

Some but not all prospective studies have observed apositive association between smoking and risk of type 2diabetes. In a 25-year prospective study among 841 mid-dle-aged Dutch men, the relative risk of diabetes was 3.3(95% CI: 1.4 to 7.9) among men who smoked .20 ciga-

rettes daily compared with never smokers, after adjustingfor age, subscapular skinfold thickness, and other riskindicators (3). Among 114,247 women participating inthe Nurses’ Health Study, those who smoked at least 25cigarettes per day had a relative risk of type 2 diabetesmellitus of 1.4 (95% CI: 1.2 to 1.7; dose-response P fortrend ,0.001) compared with never smokers, after ad-justing for age, body mass index, and other risk factors(4). Among 41,810 men participating in the Health Pro-fessionals’ Follow-up Study, the relative risk of type 2diabetes mellitus was 1.9 (95% CI: 1.3 to 3.0) among menwho smoked 25 or more cigarettes daily, compared withnonsmokers (5). In the most recent prospective study, therelative risk of type 2 diabetes among 6,250 Japanese menaged 35 to 60 years was 1.5 (95% CI: 1.1 to 1.9) amongcurrent smokers compared with nonsmokers during anaverage 10-year follow-up (6).

In contrast, three previous studies showed no associa-tion between cigarette smoking and type 2 diabetes. In a10-year follow-up study of 241 men and women withimpaired glucose tolerance, cigarette smoking was not anindependent predictor of progression to diabetes (8). Inthe Framingham cohort, smoking did not predict the de-velopment of diabetes among 3,100 men and womenduring an 8-year follow-up period (9). A similar lack ofassociation was observed in a prospective study of 10,000middle-aged Israeli workers followed for 5 years (7). Ineach of these studies, the number of incident cases ofdiabetes was relatively small, which may have limited thepower to detect modest-to-moderate associations.

The prospective design of our study precludes the pos-sibility that the reporting of cigarette smoking was biasedby the diagnosis of diabetes. It is possible, however, thatmen at high risk of diabetes due to family history of thedisorder, subclinical chemical glucose intolerance, orcoronary risk factors may have quit smoking (or neveradopted the habit) to reduce their risks of subsequentdisease. This would have led, however, to an underesti-mation of the risks of current smoking in relation to dia-betes. The follow-up rate in the sample was extremelyhigh (.99%) and similar among categories of cigarettesmoking; thus these results cannot be biased by losses tofollow-up.

Some limitations of this study deserve comment. Our“nondiabetic” cohort was not screened for glucose intol-erance, and nearly one half of diabetes cases in the UnitedStates may be undiagnosed (1). However, the prevalenceof undiagnosed diabetes is likely to be considerably lowerin this cohort of physicians, and the age-standardizedrates of diabetes in our sample were similar to those in ascreened population of a similar ethnic composition (9).Although medical records were not obtained, diabeteswould be expected to be reported reliably by the physi-cian participants in this study; a validation study of self-reported diabetes in registered nurses suggested a high

Table 3. Association between Pack-Years of Cigarette Smokingand Risk of Type 2 Diabetes Mellitus

Pack-Years ofSmoking Cases

Age-Adjusted*

MultivariateAdjusted†

0 (never smoker) 323 1.0 (Referent) 1.0 (Referent)1 to 19.9 148 1.0 (0.8–1.2) 1.0 (0.8–1.3)20 to 39.9 116 1.4 (1.1–1.7) 1.3 (1.0–1.6)$40 122 2.1 (1.7–2.5) 1.6 (1.3–2.1)P for trend‡ ,0.001 ,0.001

* Adjusted for age (1-year categories) and treatment assignment (aspi-rin and beta-carotene).† Adjusted for age (1-year categories), body mass index (quartiles),physical activity (,1, 1, 2– 4, and 5 or more times per week), history ofhypertension, history of high cholesterol, parental history of myocardialinfarction at age ,60, alcohol consumption (,monthly, monthly,weekly, daily), and treatment assignment.‡ Trend by smoking categories as listed in the Table.

Cigarette Smoking and Diabetes Mellitus among Male Physicians/Manson et al

November 2000 THE AMERICAN JOURNAL OF MEDICINEt Volume 109 541

rate of corroboration by medical record review (26). In-formation about family history of diabetes was not avail-able, but adjusting for this variable in the Nurses’ HealthStudy analyses did not materially alter the results (4).

Increased medical surveillance of smokers remains a po-tential explanation for their greater risk of being diagnosedwith diabetes as an incidental finding. We addressed thisconcern in part by repeating the analysis after excluding par-ticipants who developed diabetes during the first 2 years offollow-up—which gave asymptomatic participants moretime to develop symptomatic diabetes—and the results ofour analyses were similar. In addition, we previously foundthat smoking was a risk factor for symptomatic diabetes inthe Nurses’ Health Study (4). However, the strictest inter-pretation of our results is that, in our cohort, smoking wasassociated with the clinical diagnosis of diabetes, rather thanthe development of diabetes.

In conclusion, these prospective data from the Physi-cians’ Health Study support the hypothesis that cigarettesmoking is an independent and modifiable determinantof type 2 diabetes mellitus. The biologic plausibility ofthis association lends further credence to a causal inter-pretation, and we believe that type 2 diabetes can beadded to the list of major adverse health outcomes linkedto smoking. Strategies to prevent the adoption of ciga-rette smoking or to facilitate smoking cessation may alsoreduce the incidence of diabetes and its complications.Populations at high risk of type 2 diabetes mellitus shouldbe considered for special targeted smoking interventions.Guidelines for physicians (27) to ask about tobacco useand to assist smokers to quit could help reduce the bur-den of diabetes as well as the even larger burdens of car-diovascular disease and cancer.

ACKNOWLEDGMENTWe are grateful to the 22,071 dedicated and conscientious par-ticipants in the Physicians’ Health Study and to P. J. Skerrett forexpert assistance.

REFERENCES1. Harris MI, Flegal KM, Cowie CC, et al. Prevalence of diabetes, im-

paired fasting glucose, and impaired glucose tolerance in U.S.adults. The Third National Health and Nutrition Examination Sur-vey, 1988 –1994. Diabetes Care. 1998;21:518 –524.

2. American Diabetes Association. Diabetes Facts and Figures. Avail-able at: http://www.diabetes.org/ada/facts.asp. Accessed on 15 May2000.

3. Feskens EJ, Kromhout D. Cardiovascular risk factors and the 25-year incidence of diabetes mellitus in middle-aged men. The Zut-phen Study. Am J Epidemiol. 1989;130:1101–1108.

4. Rimm EB, Manson JE, Stampfer MJ, et al. Cigarette smoking andthe risk of diabetes in women. Am J Public Health. 1993;83:211–214.

5. Rimm EB, Chan J, Stampfer MJ, et al. Prospective study of cigarettesmoking, alcohol use, and the risk of diabetes in men. BMJ. 1995;310:555–559.

6. Uchimoto S, Tsumura K, Hayashi T, et al. Impact of cigarettesmoking on the incidence of type 2 diabetes mellitus in middle-aged Japanese men: the Osaka Health Survey. Diabet Med. 1999;16:951–955.

7. Medalie JH, Papier CM, Goldbourt U, et al. Major factors in thedevelopment of diabetes mellitus in 10,000 men. Arch Intern Med.1975;135:811– 817.

8. Keen H, Jarrett RJ, McCartney P. The ten-year follow-up of theBedford survey (1962–1972): glucose tolerance and diabetes. Dia-betologia. 1982;22:73–78.

9. Wilson PW, Anderson KM, Kannel WB. Epidemiology of diabetesmellitus in the elderly. The Framingham Study. Am J Med. 1986;80(suppl 5A):3–9.

10. Janzon L, Berntorp K, Hanson M, et al. Glucose tolerance andsmoking: a population study of oral and intravenous glucose toler-ance tests in middle-aged men. Diabetologia. 1983;25:86 – 88.

11. Grunberg NE, Popp KA, Bowen DJ, et al. Effects of chronic nicotineadministration on insulin, glucose, epinephrine, and norepineph-rine. Life Sci. 1988;42:161–170.

12. Frati AC, Iniestra F, Ariza CR. Acute effect of cigarette smoking onglucose tolerance and other cardiovascular risk factors. DiabetesCare. 1996;19:112–118.

13. Facchini FS, Hollenbeck CB, Jeppesen J, et al. Insulin resistance andcigarette smoking. Lancet. 1992;339:1128 –1130.

14. Seidell JC, Cigolini M, Deslypere JP, et al. Body fat distribution inrelation to physical activity and smoking habits in 38-year-old Eu-ropean men. The European Fat Distribution Study. Am J Epidemiol.1991;133:257–265.

15. Shimokata H, Muller DC, Andres R. Studies in the distribution of bodyfat. III. Effects of cigarette smoking. JAMA. 1989;261:1169–1173.

16. Ohlsson S, Tufvesson B, Polling A, et al. Distribution of the secre-tory leucocyte proteinase inhibitor in human articular cartilage.Biol Chem. 1997;378:1055–1058.

17. Blackman BC, White P, Tsou W, et al. Peroxidation of plasma andplatelet lipids in chronic smokers and insulin-dependent diabetics.Ann NY Acad Sci. 1984;435:385–387.

18. Paolisso G, D’Amore A, Di Maro G, et al. Evidence for a relation-ship between free radicals and insulin action in the elderly. Metab-olism. 1993;42:659 – 663.

19. Borisova AM, Georgieva R. Tobacco smoking and beta-cell secre-tion. Vutr Boles. 1991;30:58 – 60.

20. Chowdhury P, Rayford PL, Chang LW. Pathophysiological effectsof nicotine on the pancreas. Proc Soc Exp Biol Med. 1998;218:168 –173.

21. Madsbad S, McNair P, Christensen MS, et al. Influence of smokingon insulin requirement and metabolic status in diabetes mellitus.Diabetes Care. 1980;3:41– 43.

22. Steering Committee of the Physicians Health Study ResearchGroup. Findings from the aspirin component of the ongoing Phy-sicians’ Health Study. N Engl J Med. 1988;318:262–264.

23. Steering Committee of the Physicians’ Health Study ResearchGroup. Final report on the aspirin component of the ongoing Phy-sicians’ Health Study. N Engl J Med. 1989;321:129 –135.

24. Kleinbaum DG, Kupper LL, Morgenstern H. EpidemiologicResearch: Principles and Quantitative Methods. Belmont, Calif: Life-time Learning; 1982.

25. Cigarette smoking among adults–United States, 1997. MMWR.1999;48:993–996.

26. Manson JE, Rimm EB, Stampfer MJ, et al. Physical activity andincidence of non-insulin-dependent diabetes mellitus in women.Lancet. 1991;338:774 –778.

27. Helping Smokers Quit. A Guide for Primary Care Clinicians. Rock-ville, Md: Agency for Health Care Policy and Research; 1996. Avail-able at: http://www.ahcpr.gov/clinic/smokepcc.htm. Accessed onMay 8, 2000.

Cigarette Smoking and Diabetes Mellitus among Male Physicians/Manson et al

542 November 2000 THE AMERICAN JOURNAL OF MEDICINEt Volume 109