Cigarette Smoking. Findings from the 1989 National Health. Interview Survey. EARL S. FORD, MD. ANN M. MALARCHER, PHD. WILLIAM H. HERMAN, MD.
N A L
A R T I C L E
Diabetes Mellitus and Cigarette Smoking Findings from the 1989 National Health Interview Survey EARL S. FORD, MD ANN M. MALARCHER, PHD
WILLIAM H. HERMAN, MD RON E. AUBERT, PHD
OBJECTIVE — To compare the prevalence of current smoking in the U.S. diabetic population with that of the nondiabetic population. RESEARCH DESIGN A N D M E T H O D S — Using data from the 1989 National Health Interview Survey—a nationally representative sample—we calculated the prevalence of current smoking for 2,405 people with self-reported diabetes and 20,131 people without this condition. RESULTS— Overall, the age-adjusted prevalence of smoking was 27.3% among people with diabetes and 25.9% among people without diabetes. The prevalence of smoking did not differ significantly between participants with and without diabetes when they were stratified by age, sex, race, or education. Black and Hispanic men with diabetes had a higher prevalence of smoking than did white men with diabetes and black and Hispanic men without diabetes, but none of these differences were statistically significant. Among people with diabetes, age, race, sex, and educational status were independent predictors of current smoking in a multiple-logistic regression model. Duration of diabetes was not related to smoking. CONCLUSIONS — These data again emphasize the need to prevent and reduce smoking in the diabetic population. Smoking cessation programs should particularly target people with diabetes who are ^ 4 4 years of age. Black and Hispanic men are also prime targets for intervention efforts.
From the Division of Environmental Hazards and Health Effects (E.S.F.), National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, Georgia; and the Division of Diabetes Translation (A.M.M., W.H.H., R.E.A.), National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia. Address correspondence and reprint requests to Earl Ford, MD, MPH, Division of Environmental Hazards and Health Effects, National Center for Environmental Health, Centers for Disease Control and Prevention, 4770 Buford Highway, N.E., Mailstop F35, Atlanta, GA 30341. Received for publication 4 October 1993 and accepted in revised form 17 February 1994. BRFSS, Behavioral Risk Factor Surveillance System; NHIS, National Health Interview Survey; IDDM, insulin-dependent diabetes mellitus; NIDDM, non-insulin-dependent diabetes mellitus.
688
T
he adverse health consequences of cigarette smoking have been well documented. Because people with diabetes are at an increased risk for a variety of macrovascular and microvascular complications that are exacerbated by smoking, one might hope that the prevalence of smoking in this group would be less than that in the nondiabetic population. Some recent publications have indicated that this, unfortunately, is not the case. An analysis of the 1988 Behavioral Risk Factor Surveillance System (BRFSS) suggested that people with diabetes who were between 18 and 34 years of age, who had not graduated from high school, and who were African-American men had higher rates of smoking than did comparable people without diabetes (1). The 1989 National Health Interview Survey (NHIS), which included a diabetes supplement, offered another opportunity to compare smoking patterns between people with and without diabetes. This data offers several advantages, including a nationally representative sample, a large sample size, and more detailed information about diabetes. Furthermore, in contrast to studies based on telephone interviews in which telephone coverage may be an issue, NHIS interviews are conducted in person. In addition to examining the effect of several demographic variables on smoking patterns, we also examined the association of the type of diabetes and duration of the disease with smoking status among people with diabetes.
RESEARCH DESIGN A N D METHODS — Since 1957, a sample of U.S. households has participated in the NHIS each year. Household respondents complete a core questionnaire on basic health and demographic items and one or more supplements on current health topics that change from year to year. The methods of the 1989 NHIS have been published (2,3). In 1989, a total of 45,711 households were contacted. All people who were identified by the house-
DIABETES CARE, VOLUME 17,
NUMBER 7, JULY
1994
Ford and Associates
Table 1—Basic characteristics of people with and without diabetes, National Health Interview Survey, 1989 People without diabetes
People with diabetes n Age (years) 18-44 45-54 55-64 >65 Sex Men Women Race Black White Hispanic Other Education (years)
13 Duration of diabetes (years) 20 Diabetes type 1DDM N1DDM Current cigarette smoker Yes No
%
n
%
357 393 586 1,069
15.4 15.7 24.4 44.6
11,499 2,564 2,278 3,790
59.4 13.7 11.6 15.3
980 1,425
42.3 57.7
8,523 11,608
47.6 52.4
599 1,576 166 60
19.4 70.5 7.3 2.8
2,599 15,615 1,342 537
10.6 79.1 7.3 3.0
644 484 737 523
26.1 19.6 31.6 22.6
1,995 2,448 7,497 8,147
9.1 12.1 38.5 40.3
101 735 475 596 392
4.5 31.8 21.3 26.2 16.1
123 2,260
5.6 94.4
480 1,906
20.2 78.0
5,361 14,737
26.1 74.0
n, Unweighted sample size.
hold respondent as having diabetes and who confirmed this information were asked to complete the Diabetes Supplement. This supplement queried respondents about various aspects of their disease and about diabetes risk factors. A sample of respondents was also asked to complete the Diabetes Risk Factor Supplement, which contained questions that were similar to those on the Diabetes Supplement. The overall nonresponse rate was 5.1% for the core interview, 3.7% for the Diabetes Supplement, and 10.0% for sample individuals of the Diabetes Risk Factor Supplement.
DIABETES CARE, VOLUME 17,
NUMBER 7, JULY
1994
A respondent was classified as having diabetes if that respondent answered the question "Are you now a diabetic?" affirmatively. A respondent who responded affirmatively to the question "Do you now smoke cigarettes?" was classified as a current smoker. People were defined as having insulin-dependent diabetes mellitus (IDDM) if the onset of disease occurred when they were ^ 2 9 years of age, if insulin use started within 1 year of the onset of their disease, and if body mass index was 18 years of age.
age increased, the prevalence of current smoking among people with diabetes became less than those without diabetes. Smoking was somewhat more prevalent among men with diabetes than among men without diabetes and slightly less prevalent among women with diabetes than among women without diabetes. The prevalence of smoking did not differ significantly between participants with and without diabetes when they were stratified by race or education. Among people with diabetes, the prevalence of smoking decreased as the duration of the disease increased. In addition, the prevalence of smoking among people with IDDM was similar to that among those with NIDDM. No differences in the age-adjusted prevalence of smoking was detected when
690
people with diabetes who had various complications (retinal disease, cataracts, glaucoma, hypertension, angina, heart disease, stroke, kidney disease, and lower extremity amputation) were compared with people with diabetes who did not have complications (data not shown). Black and Hispanic men with diabetes had a higher prevalence of smoking than did white men with diabetes and black and Hispanic men without diabetes (Table 3). Hispanic women with diabetes had the lowest prevalence of smoking and were the only group in which the smoking prevalence was substantially lower among people with diabetes than among their counterparts without diabetes. None of these differences were statisti-
cally significant because of the large standard errors associated with the prevalence estimates. Age and education patterns in Table 2 did not change greatly in the race-stratified analysis. When predictors of current cigarette smoking among people with diabetes were examined in a multiple-logistic regression model, age, race, sex, and college education emerged as independent predictors (Table 4). People with diabetes were less likely to smoke as their age increased. Compared with white women with diabetes, black men with diabetes were 97% more likely to smoke, and Hispanic women with diabetes were 43% less likely to smoke. People with diabetes who had a college education were ~29% less likely to smoke than were those with diabetes who had not graduated from high school. The type of diabetes and the duration of the disease were not associated with the probability of current smoking.
C O N C L U S I O N S — The results from this analysis indicate that the age-adjusted prevalence of current smoking is similar among people with and without diabetes. In addition, the prevalence of smoking among people with diabetes may be especially high in certain subgroups, such as younger people and black men. These results are in agreement with previously published results from the BRFSS (1). Unlike the BRFSS, the NHIS data did not show greater smoking prevalence among less educated people with diabetes than that among less educated people without diabetes. However, in a multivariate model, education was an important predictor of smoking among people with diabetes. Some of the differences between the two studies may be due to differences in methodology: information for the BRFSS is collected through telephone interviews, and information for the NHIS is collected through personal interviews. In addition, blacks were oversampled in the NHIS to improve the stability of the esti-
DIABETES CARE, VOLUME 17,
NUMBER 7, JULY
1994
Ford and Associates
Table 3—Age-specific and age-adjusted prevalence of current cigarette smoking by selected characteristics, by race, and diabetes status, National Health Interview Survey, 1989 People with diabetes Black
Age (years) 18-44 45-54 55-64 >65 Sex Men Women Education (years) 13 Duration of diabetes (years) 20 Diabetes type 1DDM N1DDM
People without diabetes
White
n
%
SE
n
79 109 168 239
35.7 34.6 28.0 9.5
5.0 5.0 4.9 1.8
213 382
40.5 22.1
361 132 97
Black
Hispanic
White
Hispanic
%
SE
n
%
SE
n
%
SE
n
%
SE
n
%
SE
233 242 339 750
33.0 25.6 20.8 12.6
3.3 3.1 2.5 1.4
25 27 59 53
32.0 15.5 16.9 14.9
8.9 6.8 5.7 4.7
1,578 337 282 397
30.7 36.8 27.3 14.9
1.4 3.2 2.8 1.8
8,549 1,999 1,836 3,207
29.5 28.0 25.5 13.7
0.6 1.1 1.2 0.7
952 154 110 124
24.1 23.4 18.7 14.1
2.1 3.5 4.6 4.0
5.1 3.8
661 903
30.7 24.4
2.7 2.6
68 96
36.6 17.6
10.8 5.0
907 1,687
33.8 24.5
1.9 1.3
6,761 8,830
27.4 25.3
0.6 0.6
578 762
23.8 19.8
2.9 1.5
34.2 24.7 32.1
7.0 4.9 6.0
622 546 388
41.5 29.4 20.7
4.6 3.3 2.5
116 29 18
32.5 18.1 22.1
8.5 7.9 11.6
877 967 737
39.6 25.3 22.2
2.2 1.7 1.9
2,913 5,978 6,677
45.7 29.4 18.5
1.3 0.7 0.6
535 407 396
25.9 19.0 18.7
2.8 1.8 2.7
21 188 120 134 89
26.9 33.6 31.1 26.7 10.9
10.5 5.6 7.5 7.7 2.4
66 484 303 391 272
29.4 27.7 26.9 27.4 23.2
8.0 3.3 4.9 3.9 4.2
8 42 33 50 22
29.3 18.1 30.7 46.7 29.8
13.7 7.2 11.9 17.9 20.0
7 579
27.4 30.0
17.1 3.4
108 1,445
30.5 29.0
4.5 2.5
1 162
— 24.0
— 5.5
n, Unweighted sample size; age, age-specific estimates. Sex, education, duration of diabetes, and diabetes type are age-adjusted estimates, adjusted to 1980 U.S. population >18 years of age.
mates. Nevertheless, even with the relatively large sample size, the standard errors associated with the estimates are fairly large, especially when the data are stratified. In some instances, the relative standard error exceeds 30%, which may indicate that the corresponding estimate is unstable. Surprisingly, the duration of diabetes was not an independent predictor of current smoking, nor did the presence of a complication of the disease appear to be related to a lower prevalence of smoking. These unfortunate findings, if true, should serve to motivate the medical community to increase its efforts to prevent smoking among people with diabetes and to help people with diabetes who smoke to stop smoking.
DIABETES CARE, VOLUME 17,
NUMBER 7, JULY
1994
The effect of smoking on the health of people with either IDDM or NIDDM has been demonstrated by several studies in which diabetes and smoking interact to produce excess macrovascular and microvascular morbidity and mortality (6-8). These data suggest that the decrease in the prevalence of smoking with advancing age may be due, in part, to excess mortality among people with diabetes who smoke. The data from the BRFSS and the NHIS emphasize the need to reduce the prevalence of smoking in the diabetic population, in which smoking exacts a heavy toll in terms of morbidity and mortality. Subgroups of people with diabetes that should be targeted for intensive inter-
vention programs include those ^ 4 4 years of age, black men, and those with lower levels of education. The challenge of sucessfully intervening in the diabetic population is illustrated by a recent publication from Australia, in which only 50% of diabetic smokers agreed to participate in a smoking cessation program and 40% of those dropped out by 6 months (9). Nevertheless, physicians, volunteer organizations, and the public health community all have significant roles to play in preventing people with diabetes, especially young people, from starting to smoke and in helping people with diabetes who smoke to cease smoking. In this way, they will contribute to the preservation of health of the diabetic population.
691
Smoking and diabetes
Table 4—Multiple-logistic regression model of predictors of cigarette smoking among participants with diabetes, National Health Interview Survey, 1989 P Age (years) Race and sex White women (reference) Black women Hispanic women White men Black men Hispanic men Education (years) : £ l l (reference) 12 >12 Duration of diabetes (years) Type of diabetes N1DDM (reference) 1DDM n = 2,171.
692
-0.0356
-0.3031 -0.5566 0.0932 0.6778 0.2838
-0.2363 -0.3340 -0.0050
-0.3020
