Passive Cigarette Smoking and Reduced HDL Cholesterol Levels in Children With High-Risk Lipid Profiles
Background HDL cholesterol levels are known to be lower in smokers than in nonsmokers. Previous studies have demonstrated an association of decreased HDL cholesterol with passive smoking in children but have not adjusted for potential confounding factors.
Methods and Results In a cross-sectional, pilot-scale study, we examined the relationship of HDL cholesterol levels to passive smoking in children and adolescents referred to a tertiary hyperlipidemia clinic. Eligibility criteria included (1) first visit to a lipid clinic, (2) LDL cholesterol >95th percentile for age or HDL cholesterol <5th percentile, (3) age between 2 and 18 years, and (4) absence of secondary causes of hyperlipidemia. Sociodemographic information, diet record, medical history, and fasting lipid profiles were obtained. Of 109 eligible patients, 103 (94%) were studied. Twenty-seven percent came from households with cigarette smokers. HDL cholesterol levels were 38.7±1.2 mg/dL (mean±SEM) in passive smokers versus 43.6±1.2 mg/dL in children without smoke exposure (P=.005). Smoking exposure was not significantly associated with other lipid values. The effect of smoking on HDL cholesterol was minimally affected by potential confounders. In multivariate regression adjusting for body mass index, age, sex, exercise, and dietary fat intake, passive smoking remained a significant risk factor for decreased HDL cholesterol (P=.012).
Conclusions Mean HDL cholesterol levels are lower in dyslipidemic children from households with smokers than in those without household smoke exposure. Passive smoking may worsen the risk profile for later atherosclerosis among high-risk young persons.
Exposure to environmental tobacco smoke, or “passive smoking,” has been shown to have important medical consequences in adults and children, including increased risk of cancer and respiratory disease.1 2 3 Active smoking is associated with reduced HDL cholesterol levels in adults4 5 and young persons.6 In adults, passive smoking has been reported to have a similar effect.7 Previous studies in children and adolescents have documented a relationship between lower HDL cholesterol levels and passive smoking8 9 but have not adjusted for potential confounding factors, such as diet, obesity, or sociodemographic factors. No studies of passive smoking in childhood have examined subjects at high risk for future cardiovascular disease.
The purpose of our cross-sectional study was to examine the relationship of HDL cholesterol levels to passive smoking in children with abnormal lipid profiles. Our study population was composed of children and adolescents newly referred to a tertiary hyperlipidemia clinic. Most patients had a positive family history of early heart disease, and approximately one third came from households with smokers. We recorded factors that were potentially associated with low HDL cholesterol levels, including sociodemographic variables, dietary intake, exercise, and fasting lipid profiles.
Subjects were drawn from among 161 new patients referred to the lipid program at Children’s Hospital, Boston, over an 18-month period. Entry criteria for the study included (1) first visit to a lipid clinic, (2) dyslipidemia (LDL cholesterol >95th percentile for age or HDL cholesterol <5th percentile for age), (3) age between 2 and 18 years, (4) absence of secondary causes of hyperlipidemia or medications that might affect the lipid profile, and (5) informed consent of parents. The study was approved by the Committee on Clinical Investigation at Children’s Hospital.
This was a cross-sectional study of children and adolescents referred for hyperlipidemia. Before the first lipid clinic visit, we mailed a survey of demographic information and 3-day diet record to families of referred patients. Upon arrival at the clinic, a fund-of-knowledge questionnaire and a survey of attitudes about lipid disorders and therapy were administered. A nutritionist reviewed dietary records with families, and the physician’s evaluation included a medical history and physical examination. Subjects and their families were asked to estimate the number of hours of television that the child watched daily and the number of hours of aerobic exercise performed weekly (giving examples of sports, cycling, walking, running around, etc).
Assessment of Smoking Status
Each subject was asked if he or she smoked cigarettes; all denied active smoking. Adolescent patients (≥13 years old) were questioned about smoking in confidence, with their parents out of the room. We classified subjects as passive smokers if they lived in a household in which at least one individual smoked. This information was obtained at the initial visit. To further quantify passive smoke exposure, we conducted a telephone survey after the initial clinic visit. We asked parents to describe the number of smokers in the household, the number of packs per day smoked by each smoker, and the approximate number of hours per week that each smoker was in the household with the patient. From these data, we estimated the total number of packs per day to which each patient was exposed. These estimates were inexact because of variation in such factors as the percentage of time children spent in the homes of divorced parents, the packs per day smoked by household members, and the time the child spent outside the home. As a result, we dichotomized the passive smoking group into children whose passive smoke exposure was ≥1 pack/d and those exposed to <1 pack/d.
During this retrospective follow-up, we discovered one subject, classified as coming from a nonsmoking home, whose mother was a smoker. She never smoked in the home, did not want her children to know she smoked, and so deliberately misanswered the question regarding smokers in the household. Because she did not smoke at home, we continued to classify this child as coming from a nonsmoking household, although the relationship between passive smoking and HDL cholesterol was significant whether he was included or excluded.
The Minnesota Nutrition Evaluation System (Nutrition Coordinating Center, University of Minnesota, Minneapolis) was used to quantify fat and cholesterol intake. Diet record review and diet histories were obtained from the parents and children together (regardless of patient age) by the lipid clinic nutritionist. The RISCC score10 11 (ratio of ingested saturated fat and cholesterol to ingested calories) was calculated by the formula [1.01×saturated fat intake (g)±0.05×cholesterol intake (mg)]/ingested kcal/1000.
Assessment of Obesity
Weight and height were measured with shoes removed, with a single clinical balance and scale. Body mass index was calculated as weight (kg)/[height (m)]2.
Quantification of Exercise and Television Viewing
Families were asked by the physician to estimate the number of hours of aerobic exercise per week and the amount of television viewing per day. This information was gained in the examination room with parents present.
Survey of Attitudes and Fund-of-Knowledge Questionnaire
Three questions were used to probe attitudes about therapy for lipid disorders (Table 1⇓). The score on each question ranged from 1 to 5, with 5 being the most positive. Analyses involving attitude were performed on the summary score of these three questions. Similarly, seven true-false questions (Table 1⇓) were used to explore fund of knowledge, with a score range of zero to seven questions correct. Questionnaires were self-administered to parents, regardless of the age of the subjects. The questions were designed in our clinic for use in this study and were not validated in other settings.
Serum samples were obtained on the morning of the clinic visit after a 12-hour fast. Triglyceride and total cholesterol levels were determined with an Ektachem analyzer by slide enzymatic methods. HDL cholesterol was measured after precipitation of LDL cholesterol and VLDL cholesterol with dextran sulfate. LDL cholesterol was estimated by the method of Friedewald et al.12 The instrument was calibrated daily with Kodak standards. Our institution’s laboratory participates in the Centers for Disease Control/National Heart, Lung, and Blood Institute lipid certification program.
We used two-sided t tests to compare continuous variables. χ2 tests were used to compare categorical variables. We used ANCOVA to compare mean HDL levels in nonsmokers and passive smokers, adjusting for potential confounders including demographic characteristics, dietary intake, and other lipid variables. We estimated the effect of passive smoking on HDL, adjusting for each potential confounder separately and then adjusting simultaneously for those potential confounders known to affect HDL levels. ANCOVA with interaction terms was used to investigate effect modification by potential confounders. None of the variables examined were statistically significant effect modifiers, but the small number of patients limited our power to detect significant effects. Results were considered statistically significant at a value of P<.05.
Among 161 patients referred to our lipid clinic, 109 met the entry criteria and 103 (94%) were enrolled. Of the remaining 6 children, 1 was excluded post hoc because he had homozygous familial hypercholesterolemia and extreme outlying lipid values. Families of the other 5 eligible children declined participation. Twenty-eight of the ineligible patients had normal LDL cholesterol (<95th percentile) at the time of evaluation. None of the patients evaluated during this period had HDL <5th percentile and normal LDL cholesterol. Patient characteristics are shown in Table 2⇓. Twenty-eight subjects (27%) came from households with smokers, constituting the passive smoking group. Children classified as nonsmokers and those classified as passive smokers did not differ significantly in mean age, race, proportion of boys (Table 2⇓), or proportion of adolescents (19% versus 24%, respectively, were >12 years old). More than 70% of patients in each group had a family history of premature heart disease, and nearly all patients had a family history of hypercholesterolemia, reflecting the initial reasons for evaluation of cholesterol in childhood and referral to a lipid clinic in particular.
Mothers of children in the nonsmoking households, compared with those in smoking households, had more education, with a greater percentage achieving a graduate degree (24% versus 4%, respectively) and fewer having only a high-school education (12% versus 36%, respectively; Table 2⇑). Although these difference did not achieve statistical significance in our small sample, this trend is consistent with the greater preponderance of smokers among persons of lower socioeconomic and education status.
Children in the nonsmoking and passive smoking groups did not differ significantly in dietary fat intake, degree of obesity, self-reported hours of exercise and television, maternal education, or fund of knowledge/attitudes (Table 2⇑). The groups ingested a similar number of calories for body weight. Subjects in both nonsmoking and passive smoking groups were generally on low-fat diets at the time of their presentation to the lipid clinic. Mean fat intake in each group was <28% of total calories, and mean saturated fat intake, <10% of total calories. The ratio of ingested saturated fat and cholesterol to calories was also similar in nonsmokers and passive smokers (14.9±0.5 versus 14.6±1.0, respectively). The majority of subjects in each group reported >7 hours of aerobic exercise weekly and <3 hours of television daily.
Mean serum HDL cholesterol levels were significantly lower among passive smokers than among children without smoke exposure (38.7±1.2 versus 43.6±1.2 mg/dL, P=.005) (Table 3⇓). For both groups, these values were lower than the expected population mean for age (ranging from 54 to 57 mg/dL before puberty, with values in boys dropping to a mean of 47 mg/dL during adolescence). In contrast, mean levels of LDL cholesterol and triglycerides were similar in the two groups. Mean LDL cholesterol levels were >160 mg/dL in each group, with the expected population 95th percentile during childhood and adolescence ranging from 133 to 136 mg/dL in boys and 140 to 144 mg/dL in girls.13 Mean triglyceride levels were borderline elevated, without significant difference between the nonsmoking and passive smoking groups.
We explored whether HDL cholesterol was related to the estimated intensity of smoke exposure ascertained in a retrospective telephone survey. Estimates were possible in 24 of the 28 patients who reported exposure. When patients were classified into groups exposed to at least 1 pack/d (n=10), exposed to <1 pack/d (n=14), or not exposed (n=75), the relationship between smoking dose and HDL cholesterol did not reach statistical significance (ANOVA, P=.07). However, the small number of patients in the smoking subgroups and the inexact quantification of smoke exposure that could be obtained from a retrospective survey provided insufficient statistical power to exclude such a relationship.
Relation of HDL Cholesterol to Other Variables
We explored the correlation of HDL cholesterol level to variables other than smoking status. As expected, levels were inversely related to body mass index (r=−.21, P=.034) and plasma triglyceride level (r=−.46, P<.001). Higher HDL cholesterol levels also were associated with greater dietary fat intake, assessed either as RISCC score (r=.19, P=.055), total fat intake as a fraction of total calories (r=.21, P=.032), or percentage of calories as saturated fat (r=.21, P=.037). The HDL cholesterol levels of the subjects were not significantly associated with sex, race, level of parental education, self-reported amount of exercise and time spent watching television, or measures of fund of knowledge or attitude. There was not sufficient statistical power to determine the effect of smoking according to the sex of the member of the household who smoked.
Exclusion of Confounding
We explored whether the effect of passive smoking on HDL cholesterol could be explained by potential confounders, including other lipid values (eg, triglyceride levels), dietary factors, fund of knowledge, hours of aerobic exercise, measures of obesity, or sociodemographic variables, including maternal education. Household cigarette smoking was not significantly associated with any of these variables. Passive smoking remained a statistically significant predictor of HDL cholesterol level when we adjusted individually for each potential confounding variable. Furthermore, the effect estimates for difference in HDL cholesterol levels between nonsmoking and passive smoking children were minimally affected by adjustment for confounders (Table 4⇓). Children in nonsmoking versus smoking households had a mean difference in serum HDL cholesterol levels of 4.9 mg/dL when unadjusted, of 4.7 mg/dL when adjusted for dietary fat expressed as percent of calories, and of 4.1 mg/dL when adjusted for serum triglyceride level. In multivariate regression adjusting simultaneously for body mass index, age, sex, RISCC score, triglyceride level, hours of self-reported exercise, and percent dietary fat, passive smoking remained a significant factor for decreased HDL cholesterol (adjusted mean difference in HDL cholesterol, 3.7 mg/dL, P=.012).
We have found that mean HDL cholesterol levels were significantly lower in hyperlipidemic children who came from households with smokers compared with those from nonsmoking households. The effect of passive smoking in the present study was not attributable to demographic characteristics of the smoking households, the knowledge about lipids or attitudes of the parents, or physiological factors including age, body mass index, exercise, diet, or serum triglyceride levels. To the best of our knowledge, this is the first report of the effect of passive smoking in children with significant dyslipidemia and family history of early heart disease. Such children constitute a group of particular importance, because currently available data suggest that their long-term risk will be greater than that of the general population.
Two previous reports examined the relation between passive cigarette smoke exposure and HDL cholesterol levels in children and adolescents. Moskowitz et al8 studied 216 pairs of twins as part of the Medical College of Virginia twin study. One hundred five pairs of twins had at least one smoking parent. In these normocholesterolemic children, HDL cholesterol and the HDL2 subfraction were significantly lower in children exposed to passive cigarette smoke. The effect of passive smoking on HDL2 was significant only in boys and was related to thiocyanate levels. Feldman et al9 studied 391 teenagers. They found that HDL cholesterol levels were a mean of 6.8% lower in teens with plasma cotinine levels ≥2.5 mg/mL compared with those with lower cotinine levels. Our study differs from these previous reports in several important regards. First, the population studied is at greater risk of early heart disease, on the basis of family history of premature atherosclerosis and lipid profile abnormalities. Second, the subjects in our study are younger by several years than the high school students studied by Feldman et al; the great majority were too young to have been smokers themselves. Children of smokers are more likely to smoke themselves, and false denial of personal cigarette use by adolescents during a physician visit would greatly confuse interpretation of the role of passive smoke. Most importantly, the design of the present study allowed us to exclude confounding factors such as dietary fat intake, reported exercise, obesity, and triglyceride levels in the effect of smoking on HDL cholesterol. The present study extends those of Moskowitz8 and Feldman9 and suggests that a trial of active smoking cessation counseling for lipid clinic family members could be an important therapeutic intervention. Smoking cessation is an effective means of increasing HDL cholesterol in adults.14 15 16 It is interesting to compare the present results with those of a large study of passive smoking exposure in Chinese adults, in which passive smokers had lipid abnormalities approximating those of light smokers.7
Interpretation of these data is constrained by a few limitations. We did not use pharmacological tests of smoking exposure such as thiocyanate or cotinine levels or perform HDL subclass analyses. The small number of patients analyzed precludes potentially interesting analyses of subgroups, including analysis of the relationship of passive smoking dose (assessed by parental reports of smoking) to HDL cholesterol levels. We studied a restricted population with abnormal lipid profiles, and no subjects had pure low HDL phenotypes. Therefore, one cannot necessarily generalize these conclusions to the population at large. Nevertheless, there are nearly 3 million American children and adolescents in the age range studied with LDL cholesterol levels >95th percentile for age and who would be expected to be at the highest risk from lowered levels of HDL cholesterol. Finally, it should be noted that our study design did not permit us to prove that cessation of smoke exposure in our passive smokers would increase HDL cholesterol to levels in controls.
In summary, the present data demonstrate that HDL cholesterol levels are lower in dyslipidemic children with a history of exposure to cigarette smoke in the household than in children not exposed to cigarette smoke at home. The magnitude of the potential improvement (11.2%) for HDL cholesterol with cessation of smoke exposure is as great as that observed for almost any other intervention. Whereas intensive individual dietary intervention has been shown in the multisite DISC Study to lower LDL cholesterol in moderately hyperlipidemic school-age children,17 this type of intervention did not improve the HDL cholesterol.17 Adjustment for potential confounders, including dietary fat intake, serum triglyceride level, and exercise, only minimally changed the magnitude of difference in HDL cholesterol between passive smokers and nonsmokers. The results of this small-scale study suggest that interventions resulting in decreased cigarette exposure may substantially increase HDL cholesterol levels in this group of patients at higher risk for premature cardiovascular disease. This hypothesis should be the focus of future investigations.
This work was supported in part by Grant-in-Aid 13-515-901 from the American Heart Association, Massachusetts Affiliate (Dr Mietus-Snyder) and by the Kobren Fund. Dr Neufeld is a fellow of the Lucille P. Markey Foundation. We gratefully acknowledge the assistance of Isabel Vasquez, RD, in performing the diet analyses.
Reprint requests to Dr Ellis J. Neufeld, Division of Hematology, Children’s Hospital, 300 Longwood Ave, Boston, MA 02115.
- Received March 5, 1997.
- Revision received April 24, 1997.
- Accepted April 27, 1997.
- Copyright © 1997 by American Heart Association
Committee on Atherosclerosis and Hypertension in Children, Council on Cardiovascular Disease in the Young, American Heart Association. Active and passive tobacco exposure: a serious pediatric health problem. Circulation. 1994;90:2581-2590.
Environmental Protection Agency. Respiratory Health Effects of Passive Smoking: Lung Cancer and Other Disorders. Washington, DC: Office of Research and Development, Office of Air and Radiation; EPA/600/6-90-006F.
Criqui MH, Wallace RB, Heiss G, Mishkel M, Schonfeld G, Jones GT. Cigarette smoking and plasma high-density lipoprotein cholesterol: the Lipid Research Clinics Program Prevalence Study. Circulation. 1980;62(pt 2):IV-70-IV-76.
Wilson PW, Garrison RJ, Abbott RD, Castelli WP. Factors associated with lipoprotein cholesterol levels: the Framingham study. Arteriosclerosis. 1983;3:273-281.
Craig WY, Palomaki GE, Johnson AM, Haddow JE. Cigarette smoking-associated changes in blood lipid and lipoprotein levels in the 8- to 19-year-old age group: a meta-analysis. Pediatrics. 1990;85:155-158.
Zhang Y. Influence of smoking on cholesterol concentrations in serum lipoprotein of healthy subjects [in Chinese]. Chung-Hua Liu Hsing Ping Hsueh Tsa Chih. 1992;13:97-100.
Moskowitz WB, Mosteller M, Schieken RM, Bossano R, Hewitt JK, Bodurtha JN, Segrest JP. Lipoprotein and oxygen transport alterations in passive smoking preadolescent children: the MCV Twin Study. Circulation. 1990;81:586-592.
Feldman J, Shenker IR, Etzel RA, Spierto FW, Lilienfield DE, Nussbaum M, Jacobson MS. Passive smoking alters lipid profiles in adolescents. Pediatrics. 1991;88:259-264.
Hunninghake DB, Stein EA, Dujovne CA, Harris WS, Feldman EB, Miller VT, Tobert JA, Laskarzewski PM, Quiter E, Held J, Taylor AM, Hopper S, Leonard SB, Brewer BK. The efficacy of intensive dietary therapy alone or combined with lovastatin in outpatients with hypercholesterolemia. N Engl J Med. 1993;328:1213-1219.
Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972;18:499-502.
Lipid Research Clinics. Population Studies Data Book, vol I: The Prevalence Study. Bethesda, Md: National Institutes of Health; 1980.
Stubbe I, Eskilsson J, Nilsson-Ehle P. High-density lipoprotein concentrations increase after stopping smoking. Br Med J. 1982;284:1511-1513.
Fortmann SP, Haskell WL, Williams PT. Changes in plasma high density lipoprotein cholesterol after changes in cigarette use. Am J Epidemiol. 1986;124:706-710.
Lauer RM, Obarzanek E, Kwiterovich PO, Kimm SYS, Hunsberger SA, Barton BA, Vanhorn L, Stevens VJ, Lasser NL, Robson AM, Franklin FA, Simons Morton DG. Efficacy and safety of lowering dietary intake of fat and cholesterol in children with elevated low-density lipoprotein cholesterol: the Dietary Intervention Study in Children (DISC). JAMA. 1995;273:1429-1435.