International Journal of Obesity (2003) 27, 1219–1226 & 2003 Nature Publishing Group All rights reserved 0307-0565/03 $25.00 www.nature.com/ijo PAPER is associated with energy balance in premenopausal African-American adult women differently than in similarly aged white women

LH Clemens1*, RC Klesges2, DL Slawson2 and AJ Bush3

1Department of Consumer Science and Education, The University of Memphis, Memphis, TN, USA; 2Center for Community Health, The University of Memphis, Memphis, TN, USA; and 3Department of Preventive Medicine, College of Medicine, The University of Tennessee Center for the Health Sciences, Memphis, TN, USA

OBJECTIVE: To investigate the differential association of cigarette smoking with energy balance in African-American and white premenopausal women. DESIGN: Cross-sectional study of energy balance, weight, and smoking in women. SUBJECT: A total of 374 women: 191 African-American (mean age ¼ 29.876.5 y) and 183 white women (mean age ¼ 28.977.1 y). MEASUREMENTS: Weight, cigarette smoking habits, resting energy expenditure, dietary intake, and physical activity. RESULTS: There were no significant differences in dietary intake by race or smoking status. The model for physical activity was significant (P ¼ 0.0004), with body mass index (BMI) having the largest effect on activity (Po0.001). Smoking status was related to activity, with the heaviest smokers reporting more activity than nonsmokers (P ¼ 0.008) or light smokers (P ¼ 0.028). The model for resting energy expenditure (REE) was significant (Po0.0001), with the largest again being BMI (Po0.001). There was also an interaction between ethnicity and smoking status (Po0.0001) such that African-American nonsmokers and light smokers tended to have lower REE than several other groups, most often the African-American moderate heavy smokers. The model for BMI was significant (Po0.0001) with an interaction for ethnicity and smoking status (P ¼ 0.0009). African-American nonsmokers and light smokers had significantly higher BMIs than most of the other groups. CONCLUSION: African-American women who were the heaviest smokers had a lower adjusted BMI than the heaviest smoking white women. This effect, at least partially, may be related to an increased REE in the African-American smoking women. While energy intake did not appear to be important in this relationship, energy expended in physical activity appeared to be increased with smoking, as was REE. International Journal of Obesity (2003) 27, 1219–1226. doi:10.1038/sj.ijo.0802400

Keywords: smoking; body weight; energy balance; REE; physical activity

Introduction occur. Cigarette smoking has long been thought to alter this Energy intake (kilocalories ingested) must equal energy equation, given that smokers often weigh less than non- output (kilocalories expended via resting energy expenditure smokers;1 but the results of a rather large body of literature (REE) plus energy expenditure due to physical activity and examining the effect of smoking on components of the thermic effect of food) in order for weight maintenance to energy balance equation are equivocal. For example, an earlier cross-sectional analysis of NHANES II data2 found that African-American women were heavier than the white *Correspondence: Dr LH Clemens, Department of Consumer Science and women as a group, and that white women who smoked Education, The University of Memphis, Manning 410A, Memphis, TN 38152, USA. had lower body weights than white nonsmoking women E-mail: [email protected] while African-American women who smoked heavily were This work was supported by funds from the National, Heart, Lung, and heavier than African-American nonsmoking women. This Blood Institute HL50946 awarded to the first author and HL463521 suggests that smoking does not confer the same awarded to the second author. Received 30 August 2002; revised 14 February 2003; weight ‘benefit’ to African-American women that it does to accepted 2 March 2003 white women. However, a recent longitudinal analysis of Smoking and energy balance LH Clemens et al 1220 weight gain across 7 y in a biracial sample of young adults3 menstrual cycles. Thus, a total of two laboratory sessions found that continued smoking suppressed weight gain to a were held, with the first occurring between days 2 and 4 of small degree in African American but not in white subjects, the subjects’ cycles, and the second between days 7 and 9. with no effect for gender. Although the 1998 study included Participants kept logs to track the onset and offset of their a number of covariates, including age, energy and alcohol menstrual flow. If a subject was still bleeding by day 9, she intake, and level of physical fitness, the analysis did not was rescheduled for the next month. All procedures were include energy expenditure. The objective of the current completed at each laboratory session, all subjects completed study is to investigate the relationship between ethnicity and the laboratory session in the same order, and measures were smoking status and energy balance in African-American and averaged across the two laboratory sessions for statistical white women, using the energy balance equation (energy analyses. intake, REE, and energy expenditure due to physical activity). Thermic effect of food was not considered as it has been shown to represent a very small portion of total Measures energy expenditure. Smoking status. Subjects were queried on current smoking status and smoking history. Subjects who were current or previous smokers were asked about the total number of years Methods of smoking, the number of cigarettes per day, the age of Subjects smoking onset, and, if quit, the age at cessation. Pack-years Subjects were premenopausal women participating in a was calculated as cigarettes per day divided by 20 and 6 study of the impact of smoking and ethnicity on body multiplied by the number of years smoked. For analysis weight. Subjects were recruited via public service announce- purposes, past smokers were considered to be nonsmokers as ments on local radio and television stations, flyers posted in we felt there was no reason to think that past smoking would establishments frequented by women (eg libraries, beauty have an effect on the energy balance variables or on current shops, community centers, grocery stores), a message on the BMI. Subjects who were previous smokers had to be quit for ‘hold’ line on the university’s telephone system, and at least 6 months prior to their admission into the study. The advertisements placed in community newspapers. Potential following categories were used for smoking status: nonsmo- subjects were invited to telephone for more information. At kers ¼ all current nonsmokers; light smokers ¼ current smo- that time, their questions were answered and a brief screen kers up to 5 pack-years; moderate smokers ¼ current smokers was administered if they agreed and were still interested in between 5 and 10 pack-years; heavy smokers ¼ greater than participation. That screen covered the following criteria for 10 pack-years. inclusion: age 18–40 y, weight less than 275 pounds, regular menstrual cycles, not pregnant or nursing in the past 6 Dietary intake. Intake was assessed via diet records, con- months, no major health problems, not fasting on a regular sidered to provide a more reliable estimate of dietary intake basis, not on a special or weight-reducing diet, and weight than other measurements,7 and the measure by which other stable for the previous 6 months. These criteria were chosen methods of quantifying food intake are compared.8 In order because of the effect each could have on some aspect of to insure more accurate portion estimation and more energy balance. Women who met the criteria were scheduled complete, detailed records, multiple training sessions were for an orientation session where they were consented, conducted. The initial training session was carried out in scheduled for participation, and trained for some proce- groups of up to 15 participants after patients completed the dures. Recruiting was targeted to result in a sample that was orientation session. Models of common foods, measuring ethnically representative of the area population. A total of cups and spoons, and templates for estimating sizes of 374 women (191 African-American (mean age ¼ 29.897 common foods (eg pizza, cake, cookies, bagels) were used in 6.52 y) and 183 white women (mean age ¼ 28.8977.07 y) training. This session lasted about an hour and participants were recruited and completed the study. An individual was practiced measuring foods, as well as recorded that day’s considered to be African American if she reported being born intake and practiced the level of detail needed. Dietary in America and having at least three grandparents of African records were checked for adequate level of detail when heritage. Similarly, a classification of white was made if a subjects returned for their first laboratory session (between subject was American-born and reported having at least days 2 and 4). If detail was not sufficient, subjects were three grandparents of European background. reminded of the detail needed. All tools used in the group training were available for use, if needed. The length of this session depended on the quality of the records already Procedure recorded but tended to be less than 15 min. All assessments As dietary intake, physical activity, and metabolic rate have of dietary intake were made under the supervision of a been shown to vary as a function of the menstrual cycle,4,5 registered dietitian. all assessments were taken twice, once during the menses Subjects completed seven consecutive days of diet records. stage and once during the follicular stage of the subjects’ In order to minimize the effects of reactivity to the process of

International Journal of Obesity Smoking and energy balance LH Clemens et al 1221 recording intake, the first day of records was not coded. in a thermoneutral (231C) and humidity-controlled, quiet, Further, in order to guard against the possibility of decreased semidarkened environment. The calorimeter was calibrated diligence in recording as subjects ended their recording, the prior to each testing session using gas mixtures of precisely final day of records was not utilized. Therefore, days two to known O2 and CO2 concentrations. Validation was verified six were selected for analysis. Records were coded and on a weekly basis using the Medical Graphicss Exchange analyzed using the DINE Nutrient Analysis System.9 System Validator. During this period, steady state was reached with RQ readings below 0.79. If initial RQs were Physical activity. A short, self-administered questionnaire above 0.81, subjects were rescheduled for another visit. Data developed by Baecke et al10 was used to assess levels of from the initial 5 min of measurement, along with periods of physical activity. The instrument examines three compo- movement by the participant, were eliminated from the nents of physical activity participation as identified by factor calculation of mean REE. Breath-by-breath measures of O2 analysis: work, sport, leisure excluding sport, as well as total consumption, CO2 production, and respiratory rate were activity.11 Estimation of total activity was calculated accord- obtained. Estimates of REE per 24 h were calculated by ing to the methods described by the authors. The ques- computer from the O2 and CO2 gas exchange using the tionnaire has been found to have a high correlation (r = 0.69) abbreviated Weir formula.15 The coefficient of variation for with doubly labelled water,12 and be significantly correlated the sample was 14.4%. 11 13 with VO2 peak (r ¼ 0.49) as well VO2 max (r ¼ 0.54). Test– retest coefficients on the three subscales have ranged between 0.74 and 0.93 in two previous investigations.10,13 Analysis plan Analysis of covariance (ANCOVA) was carried out for each of Anthropometry. Anthropometric measurements were taken the main outcomes: total energy intake, REE, and energy using standardized methods, with subjects dressed only in expenditure via physical activity. Subjects’ ethnicity, smok- lightweight clothing, with shoes removed. Body weight was ing status, and the interaction thereof served as independent measured with the subject standing on a leveled platform variables. Covariates for each analysis were age and BMI. See scale with a beam and movable weights. Height was Table 1 for means and standard deviations for age and BMI. measured to the nearest 0.25 in by using a wall-mounted ANCOVA was also employed to investigate the impacts of stadiometer. Body mass index (BMI) was calculated as ethnicity and smoking status on BMI while controlling for body weight in kilograms divided by height (in meters the following variables: total reported energy intake, REE, squared). energy expenditure via physical activity, and age. The Tukey– Kramer procedure was used to follow up all significant factor Resting energy expenditure. Subjects were instructed to fast effects. This procedure permits making all possible pairwise and abstain from exercise, food intake, and smoking for at comparisons while holding the collective probability of type least 10 h before each laboratory visit. All assessments were 1 error to 0.05. scheduled between 0600 and 0900. Upon arriving at the laboratory, participants sat for a period of approximately 20 min while self-report measures were reviewed. Partici- Results pants then were measured at rest in a supine position for a Covariates period of at least 25 min, including a 5-min acclimation An a priori decision was made to control for age and BMI period.14 REE was estimated through indirect calorimetry consistently in all models given their well-established using an open-circuit canopy collection system (Medical relationship with one or more of the main outcomes. Highly Graphics, St Paul, MN, USA). Measurements were obtained significant relationships between BMI and energy expendi-

Table 1 Subject demographics (n ¼ 374) by smoking status and ethnicity (mean7s.d.)

NS LS MS HS Total Characteristic n ¼ 120 n ¼ 25 n ¼ 12 n ¼ 26 n ¼ 183

White subjects Age (y) 29.12 (77.30) 22.96 (74.65) 27.50 (75.23) 33.65 (74.22) 28.89 (77.07) BMI (kg/m2) 23.6 (74.45) 23.32 (74.68) 22.44 (75.32) 26.31 (76.16) 23.85 (74.88)

n ¼ 123 n ¼ 20 n ¼ 27 n ¼ 21 n ¼ 191 African-American subjects Age (y) 28.18 (76.53) 28.55 (76.45) 33.93 (74.14) 35.09 (73.48) 29.79 (76.52) BMI (kg/m2) 27.54 (76.25) 27.71 (76.44) 27.70 (74.97) 26.98 (74.11) 27.52 (75.87)

NS ¼ nonsmoker (0 pack-years), LS ¼ light smoker (0 to less than 5 pack-years), MS ¼ moderate smoker (5–10 pack-years), HS ¼ heavy smoker (over 10 pack-years).

International Journal of Obesity Smoking and energy balance LH Clemens et al 1222 ture from physical activity (Po0.0001) and REE (Po0.001) and between age and REE (P ¼ 0.02) in our models supported this decision. 6109.22 (1013.85) 3230.46 (971.19) Energy intake 7417.22 (2334.03)

The effects of ethnicity and smoking status on average w **,

reported daily energy intake were investigated while con- 191 ¼ Total trolling for age and BMI. When adjusted for all variables n (1423.83) 6764.87 3613.31 (1111.09) within the model, the ANCOVA model was not significant 7248.14 (3059.21)

(RSQ ¼ 0.021; F ¼ 0.88; P ¼ 0.54). Raw (unadjusted) and y *,

model (adjusted) means, along with standard deviations for 21 HS ¼

the raw means and ranges for the raw means are presented in n (941.01) 6664.41 3435.41 (909.24) Table 2. African-American subjects reported slightly lower 7283.01 (2390.71) y ,

energy intakes overall (model means: 7291.1 vs 7784.4 kJ/day z , w for the white subjects). 27 ¼ MS n 5942.32 (1111.84) 3141.26 (865.25) 7418.76 (2529.56) Energy expenditure from physical activity

The ANCOVA model for activity-based energy expenditure *,**,*** 20 LS

was shown to be significant (RSQ ¼ 0.350;. F ¼ 21.75; ¼ P ¼ 0.0004). Raw (unadjusted) and model (adjusted) means, n 5902.59 (835.15) 3134.60 (963.66) along with standard deviations for the raw means and ranges 7475.28 (2173.25) for the raw means are presented in Table 2. Here again, the impact of ethnicity and smoking status was investigated

while controlling for age and BMI. When adjusted for all 123 NS ¼

variables within the model, the covariate BMI had the n greatest impact on reported physical activity expenditure 6095.65 (976.97) 2858.20 (962.28) 7867.59 (2208.53)

(PAE) (F ¼ 145.70; Po0.0001). As to the independent factors for all subjects a in the model (ethnicity and smoking status), smoking status was related to PAE (F ¼ 3.58; P ¼ 0.0142). The heaviest 183 ¼ Total n 6295.41 (1139.55) 3393.17 (1469.58) smokers reported more activity than did either the non- 8209.96 (2830.45) smokers (P ¼ 0.008) or the light smokers (P ¼ 0.0238). Figures

1 and 2 present the results of the ANCOVA model with all 26 HS ¼

smokers collapsed into one category, for African-American n 5810.46 (868.22) 2574.89 (861.23) and white women, respectively. 7450.08 (2257.72) 0.0591. 12 ¼ MS ¼ P y n 5765.38 (693.37) 2621.02 (693.03) Resting energy expenditure 7342.75 (2116.61)

The ANCOVA model for REE was significant (RSQ ¼ 0.357; z 0.0424, ***, F ¼ 22.5; Po0.0001). Raw (unadjusted) and model (adjusted) White subjects African-American subjects ¼ 25 P LS z means, along with standard deviations for the raw means ¼ n (988.73) (844.53) and ranges for the raw means are presented in Table 2. Of the (2072.45) covariates in the model (age and BMI), BMI had a significant 0.0131,

impact on REE (F ¼ 152.80; Po0.0001). The interaction of ¼ P w

the independent factors, ethnicity and smoking status, was 120 NS significant (F ¼ 7.71; P 0.0001). African-American nonsmo- ¼

o n

kers had lower average REEs (model mean ¼ 5724.7) than did 0.0001, RangeModel mean 6361.67Raw mean (s.d.) 2820.15 6004.73 6134.16 6235.51 6184.02 5724.65 5746.96 6467.62 6638.83 4013.37–9665.47 6144.50 4219.49–9621.52 Raw mean (s.d.) 6149.69 RangeModel mean 3026.06 2855.44 2889.76 3333.60 3026.23 2962.22 2951.84 3242.85 3489.07 1508.43–8129.51 3161.48 1646.40–7708.52 RangeModel mean 7984.42 7489.26 7536.06 8127.79 7784.37 7463.26 7397.58 7172.63 7130.68 2609.85–14 562.51 7291.05 2727.47–17 528.58 o African-American moderate smokers (model mean ¼ 6467.6; P *** P ¼ 0.008), African-American heavy smokers (model mean ¼ 6638.8; P ¼ 0.001), and white nonsmokers (model 0.001,

mean ¼ 6361.7; Po0.0001). Additionally, African-American ¼ P

light smokers had lower average REEs (model mean ¼ 5747.0) ** Energy intake and expenditure raw means (s.d.) and model means by smoking status than African-American heavy smokers (model

mean ¼ 6638.8; P ¼ 0.0131) and white nonsmokers (model 0.008, ¼ P Table 2 * Energy Variables Energy expenditure from physical activity (kJ/day) Energy intake (kJ/day) Raw mean (s.d.) 7944.48 mean ¼ 6361.7; P ¼ 0.0424). African-American light smokers Resting energy expenditure (kJ/day)

International Journal of Obesity Smoking and energy balance LH Clemens et al 1223 Physical Activity Energy Expenditure vs. BMI by Smoking Status for Resting Energy Expenditure vs. BMI by Smoking Status for African African American Women American Women 9000 10000 8000 9000 7000 6000 8000 5000 4000 7000

PA-EE (kJ) 3000 6000

2000 REE (kJ) 1000 5000 0 10 15 20 25 30 35 40 45 50 4000 BMI

Nonsmokers Smokers 3000 Linear (Nonsmokers) Linear (Smokers) 10 15 20 25 30 35 40 45 50 BMI

Figure 1 Physical activity energy expenditure vs BMI by smoking status for Nonsmokers Smokers African-American women. Linear (Nonsmokers) Linear (Smokers)

Figure 3 Resting energy expenditure vs BMI by smoking status for African- Physical Activity Energy Expenditure vs. BMI by Smoking Status American women. for White Women 9000 8000 Resting Energy Expenditure vs. BMI by Smoking Status for White 7000 Women 10000 6000

5000 9000 4000

PA-EE (kJ) 8000 3000

2000 7000 1000

REE (kJ) 6000 0 10 15 20 25 30 35 40 45 50 BMI 5000 Nonsmokers Smokers Linear (Nonsmokers) Linear (Smokers) 4000

Figure 2 Physical activity energy expenditure vs BMI by smoking status for 3000 10 15 20 25 30 35 40 45 50 White women. BMI

Nonsmokers Smokers Linear (Nonsmokers) Linear (Smokers) also had marginally lower average REEs (model mean ¼ 5747.0) than African-American moderate smokers Figure 4 Resting energy expenditure vs BMI by smoking status for White (model mean ¼ 6467.6; P ¼ 0.0591). Figures 3 and 4 present women. the results of the ANCOVA model with all smokers collapsed into one category, for African-American and white women, mean ¼ 27.91) than did African-American heavy smokers respectively. (model mean ¼ 23.63; P ¼ 0.0227) and white nonsmokers (model mean ¼ 24.03; P ¼ 0.0024). Figure 5 presents the model means for BMI, as well as significant differences. Body mass index The impacts of ethnicity and smoking status on BMI were assessed while controlling for the following variables: total reported energy intake, REE, physical activity energy ex- Discussion penditure, and age. The ANCOVA model was significant In the present sample, African-American women who were (RSQ ¼ 0.514; F ¼ 34.81; Po0001). Again, the interaction of the heaviest smokers had a lower adjusted BMI than the ethnicity and smoking status was significant (P ¼ 0.0009). heaviest smoking white women (Figure 5). This effect, at African-American nonsmokers had higher BMIs (model least partially, appears to be related to an increased REE in mean ¼ 27.88) than did African-American moderate smokers the African-American smoking women. While energy intake (model mean ¼ 25.09; P ¼ 0.0395), African-American heavy did not appear to be important in this relationship, energy smokers (model mean ¼ 23.63; P ¼ 0.0007), as well as white expended in physical activity appeared to be increased with nonsmokers (model mean ¼ 24.03; Po0.0001) and white smoking, as was REE. heavy smokers (model mean ¼ 24.83; P ¼ 0.0161). Further, There was an a priori decision to control for age and BMI in African-American light smokers had higher BMIs (model the ANCOVA models to test the relationship between

International Journal of Obesity Smoking and energy balance LH Clemens et al 1224 weights than do white women at similar levels of energy Heavy Smoker intake, it is also likely that inaccuracies in reporting dietary intake (in both groups) may have led to null findings in the current study. It is well-established that dietary intake data Moderate Smoker are subject to a number of problems and that under- reporting of energy intake is widespread.22

Light As expected, the covariate BMI was significantly related to * Smoking Status Smoker energy expenditure from physical activity. However, finding that smoking status was also significantly related to energy Non-smoker * * expenditure from physical activity was not expected. In this sample, heavy smokers reported more activity than any of 21 22 23 24 25 26 27 28 29 the other groups. Since there were no questions regarding Body Mass Index the reasons for physical activity, it is impossible to know why White African-American this occurred. One possibility is that heavier smokers engaged in more physical activity (or perceived themselves * African-American high smokers had higher BMIs than white non-smokers (p=0.0024) and AA heavy smokers (p=0.0227) to be engaging in more physical activity) in an effort to * * African-American non-smokers had higher BMIs than white non-smokers (p<0.0001); counteract the harm they felt they were doing by smoking white heavy smokers (p=0.0161); AA moderate smokers (p=0.0306); and AA heavy smokers (p=0.0007) cigarettes. Future studies are needed to validate the finding Figure 5 Mean body mass index for African-American and White women by of more physical activity and should seek reasons for this smoking status. increase. The covariate BMI was also significantly related to REE, a finding that is well documented.23 Our finding that smoking smoking status and ethnicity and the components of energy appears to significantly impact REE in African-American balance, that is, energy intake, REE, and energy expenditure women but not in white women has not been previously from physical activity. Both age and BMI (or other measures noted. In a study of 152 healthy young women,24 smoking of weight status) have repeatedly been shown to indepen- was found to have an overall effect for increasing REE dently affect dietary intake, REE, and energy expenditure (283 kJ/day; P ¼ 0.03) but did not affect the REE of the from physical activity. African-American women significantly more than the white The association between smoking and body weight has women. However, that study did not account for level of been the subject of numerous investigations. While there is smoking; thus, light and moderate smokers may have some consensus that this relationship is U-shaped in all diluted any effect of smoking on REE. There were also a groups16,17 or in certain groups (eg African-American very small number of African-American smokers (six of 76 women2), others have concluded that smokers generally African-American women) in the Kimm et al study. Another weigh less than nonsmokers.6,18 Making things even less study25 found significantly lower REE in African-American clear, in a large screening study (n ¼ 2096 women) Istvan and white smokers but did not analyze for gender differ- et al19 found that cigarettes/day was positively correlated ences. Since that study found no ethnic differences in the with BMI while serum cotinine was negatively correlated change in REE following , it is possible with BMI. Clearly, the relationship between smoking and that baseline REE differences were due to ethnic REE body weight is complex and consideration of the factors that differences previously noted in women26–31 and also in affect body weight is essential. This is demonstrated by a men.32,33 The Vander Weg et al25 also did not consider 1998 study of young adult, African-American and white men smoking rate or carbon monoxide level as they found neither and women.3 While the Klesges et al3 study found no effect related to baseline REE. Thus, the current study was for smoking on body weight before adjustment for con- conducted and analyzed in a different manner than earlier founders, after adjusting for age, total energy intake, alcohol studies. The pattern in the current sample is very clear with intake, and physical fitness, smokers weighed less than African-American heavy smokers having a significantly nonsmokers. That finding would indicate that smoking is higher adjusted REE than African–American nonsmokers related to at least one of the covariates. and light smokers as well as white light smokers. African- Although the African-American women in the current American nonsmokers and light smokers had a lower sample reported a slightly lower total energy intake, after adjusted REE than all other groups save white light smokers. controlling for the covariates, smoking status and ethnicity This is especially interesting in the light of the fact that had no significant relationship with dietary intake. There is numerous recent reports have found that African-American nothing in the previous literature to indicate that there are women have lower REEs than white women.26–31 Most energy intake differences between African-American and recently, this difference has also been noted in smoking white women,20 even though there is a higher prevalence of women.25 Smoking may, in some way, work to increase a obesity in African-American women.21 While it is plausible depressed REE to a higher level in African-American women. that African-American women maintain higher body If the women in the current sample are similar to those in

International Journal of Obesity Smoking and energy balance LH Clemens et al 1225 previous studies, this would appear to be what has 6 Albanes D, Jones DY, Micozzi MS, Mattson ME. Associations happened. between smoking and body weight in the US population. Analysis Am J Publ Health The literature regarding how smoking affects REE is of NHANES II. 1987; 77: 439–444. 7 Drewnoski A. Diet image: a new perspective on the food- conflicting but the general consensus appears to be that frequency questionnaire. Nutr Rev 2001; 59: 370–372. smoking has no effect on chronic metabolic rate.34 Although 8 Block G, Hartman AM. Issues in reproducibility and validity of there do not appear to be any studies that have examined the dietary studies. Am J Clin Nutr 1989; 50: 1133–1138. relationship between smoking and metabolic rate in African- 9 Dennison DD, Dennison KF, Addman M, Groeger LA. The DINE system: nutrient analysis, diet improvement software for health American women, it has been suggested that the conflicting care settings. In: Dayhoff R (ed). Proceedings of the 11th annual results regarding smoking and REE overall may be due to symposium on computer applications in medical care, November 1–4, moderators (such as ethnicity and gender) that have not 1983. 10 Baecke JAH, Burema J, Frijters JER. A short questionnaire for the been included in these previous investigations.35 Given that measurement of habitual physical activity in epidemiological there are clearly untoward variations in metabolic rate in studies. Am J Clin Nutr 1982; 36: 936–942. African-American women, it may not be unusual that this 11 Montoye HJ, Kemper HCG, Saris WHM, Washburn RA. Measuring relationship was noted in the current sample. physical activity and energy expenditure. Human Kinetics: Cham- paign, IL; 1996. If, as it appears in this sample, the African-American 12 Philippaerts RM, Westerterp KR, Lefevre J. Comparison of two smoking women are deriving a weight benefit from smoking, questionnaires with a tri-axial accelerometer to assess physical this is especially important given that African-American activity patterns. Int J Sports Med 2001; 22: 34–39. smokers have been reported to be less successful in cessation 13 Jacobs Jr DR, Ainsworth BE, Hartman TJ, Leon AA. A simulta- neous evaluation of ten commonly used physical activity attempts than are white smokers36 and that African-Amer- questionnaires. Med Sci Sports Exerc 1993; 25: 81–91. ican women gain significantly more weight than white 14 Isbell TR, Klesges RC, Meyers AW, Klesges LM. Measurement women when they quit smoking.37 Combined with the reliability and reactivity using repeated measurements of resting finding that African-American smokers also may derive a energy expenditure with a face mask, mouthpiece, and ventilated 38 canopy. J Parenter Enteral Nutr 1991; 15: 165–168. different metabolic effect from smoking, it may be that the 15 Weir JB de V. New methods for calculating metabolic rate metabolic effects of smoking differ by ethnicity. This would with special reference to protein metabolism. J Physiol 1949; be especially important in cessation interventions aimed at 109: 1–9. African-American women. 16 Istvan JA, Cunningham TW, Garfinkel L. Cigarette smoking and body weight in the Cancer Prevention Study I. Int J Epidemiol 1992; 21: 849–953. 17 Rasky E, Stronegger W-J, Freidl W. The relationship between body Summary weight and patterns of smoking in women and men. Int J Epidemiol 1996; 25: 1208–1212. The current study utilized a large sample of African- 18 Klesges RC, Zbikowski SM, Lando HA, Haddock CK, Talcott GW, American and white women, including both smokers and Robinson LA. The relationship between smoking and body nonsmokers, in an analyses of the relationship of ethnicity weight in a population of young military personnel. Health and current smoking status with energy balance and BMI. Psychol 1998; 17: 454–458. 19 Istvan JA, Nides MA, Buist AS, Greene P, Voelker H. Salivary This pattern of results is consistent with the hypothesis that cotinine, frequency of cigarette smoking, and body mass index: the African-American women in the current study might be findings at baseline in the Lung Health Study. Am J Epidemiol receiving more weight suppression benefit from smoking, 1994; 139: 628–636. and that those benefits might be due to increases in REE. 20 Chitwood LF, Brown SP, Lundy MJ, Dupper MA. Metabolic propensity toward obesity in black vs white females: responses Physical activity was increased in all smokers. during rest, exercise and recovery. Int J Obes Relate Metab Disord 1996; 20: 455–462. 21 Kumanyika SK. Understanding ethnic differences in energy balance: can we get there from here. Am J Clin Nutr 1999; 70: References 1–2. 1 U.S. Department of Health and human Services. Women and 22 Clemens LH, Klesges RC, Slawson DL, Kumanyika SK, smoking. A report of the Surgeon General F 2001. U.S. Department Murray DM, Beech BM. The relationships between personal and of Health and Public health Service, Center for Disease Control, psychosocial characteristics and underreporting of energy intake Office on smoking and Health: Rockville, MD; 2001. [accessed at: in African-American and white women (manuscript under www.cdc.gov/tobacco/sgr_forwomen.htm; accessed on 7/25/01] revision). 2 Klesges RC, Klesges LM. The relationship between body mass and 23 Blackburn GL, Kanders BS (eds). Obesity, pathophysiology, psychol- cigarette smoking using a biochemical index of smoking ogy, and treatment. Chapman & Hall Series in Clinical Nutrition. exposure. Intl J Obes Relate Metab Disord. 1993; 17: 585–591. Chapman & Hall: New York; 1994. 3 Klesges RC, Ward KD, Ray JW, Cutter G, Jacobs DR, Wagenknecht 24 Kimm SYS, Glynn NW, Aston CE, Poehlman ET, Daniels SR. LE. The prospective relationships between smoking and weight in Effects of race, cigarette smoking, and use of contraceptive a young, biracial chohort: The Coronary Artery Risk Develop- medications on resting energy expenditure in young women. Am ment in Young Adults Study. J Consult Clin Psychol 1998; 66: J Epidemiol 2001; 154: 718–724. 987–993. 25 Vander Weg MW, Klesges RC, Ward KD. Differences in resting 4 DeBon M, Klesges RC, Klesges LM. When should you measure energy expenditure between black and white smokers: implica- metabolic rate? Ann Behav Med 1993; 15: S67. tions for postcessation weight gain. Eur J Clin Nutr 2000; 54: 5 Eck LH, DeBon M, Klesges RC, Preacher-Ryan H, Klesges LM. 895–899. Gathering dietary intake during the menstrual cycle. Ann Behav 26 Forman JN, Miller WC, Szymanski LM, Fernhall B. Differences in Med 1994; 16: S165. resting metabolic rates of inactive obese African-American and

International Journal of Obesity Smoking and energy balance LH Clemens et al 1226 Caucasian women. Int J Obes Relat Metab Disord 1998; 22: 33 Sharp TA, Bell ML, Grunwald GK, Schmitz KH, Sidney S, Lewis 215–221. CE, Tolan K, Hill JO. Differences in resting metabolic rate 27 Foster GD, Wadden TA, Vogt RA. Resting energy expenditure in between white and African-American young adults. Obes Res obese African American and Caucasian women. Obes Res 1997; 5: 2002; 10: 726–732. 1–8. 34 Warwick PM, Baines J. Energy expenditure in free-living smokers 28 Jakicic JM, Wing RR. Differences in resting energy expenditure in and nonsmokers: comparison between factorial, intake-balance, African-American vs Caucasian overweight females. Int J Obes and doubly labeled water measures. Am J Clin Nutr 1996; 63: Relat Metab Disord 1998; 22: 236–242. 15–21. 29 Nicklas BJ, Berman DM, Davis DC, Dobrovolny CL, Dennis KE. 35 Ward KD, Klesges RC, Vander Weg MW. Cessation of smoking and Racial differences in metabolic predictors of obesity among body weight. In Bjorntorp P (ed). International textbook of obesity. postmenopausal women. Obes Res 1999; 7: 463–468. John Wiley & Sons, Ltd: London; 2001. pp 323–336. 30 Weyer C, Snitker S, Bogardus C, Ravussin E. Energy metabolism in 36 Fiore MC, Novotny TE, Pierce JP, Hatziandreu EJ, Patel KM, David African Americans: potential risk factors for obesity. Am J Clin RM. Trends in cigarette smoking in the United States: Nutr 1999; 70: 13–20. the changing influence of gender and race. JAMA 1989; 261: 31 Weinsier RL, Hunter GR, Zuckerman PA, Redden DT, Darnell BE, 49–55. Larson DE, Newcomer BR, Goran MI. Energy expenditure 37 Williamson DF, Madans J, Anda RF, Kleinman JC, Giovino GA, and free-living physical activity in black and white women: Byers T. Smoking cessation and severity of weight gain in a comparison before and after weight loss. Am J Clin Nutr 2000; 71: national cohort. N Engl J Med 1991; 324: 739–745. 1138–1146. 38 Wagenknecht LE, Cutter GR, Haley NJ, Sidney S, Manolio TA, 32 Carpenter WH, Fonong T, Toth MJ, Ades PA, Calles-Escandon J, Hughes GH, Jacobs DR. Racial differences in serum Walst JD, Poehlman ET. Total daily energy expenditure in free- cotinine levels among smokers in the Coronary Artery Risk living older African-Americans and Caucasians. Am J Physiol Development in (Young) Adults Study. Am J Public Health 1990; 1998; 274(1 Part 1): E96E101. 80: 1053–1056.

International Journal of Obesity