E TABLE 3.-Continued Study Design and sample Major results Moderator variables Limitations Haworth et al. t1990) 536 women (2.74 nonsmokers, 302 smokers) interviewed last prenatal visit (18%) or within day after delivery (82%1 No smoker/nonsmoker pregnancy weight gain difference Hickey and 150 men 1124 smokers); Bmonth. Mulcahy 2.year followups after myocardial (1973) infarction Quitter, reducer, continuing smoker differences not signilicant Smoking self-report: pregnancy weight gain data only Smoking self-report; postmyocardial infarction may motivate heahhy behavior Iiolme et al. (1985) 16,202 Oslo men, aged 4049, screening program; 1,232 (elevated cholesterol or upper quartile coronary risk score) randomly assigned diet/smoking intervention or control: &year fOlLOWID 17% controls, 24% intervention quit; l- to 2.yearquitter weight increased more than controls, then decreased to below prequit level Smoking self-report; confounded by high cardiovascular disease risk health intervention; weights not reported Howell (1971) Retrospective, 1,121 men, aged 40-54; 15 to 20year weight gain examinations Light smokers (~20 cigarettes/day) gained 1.9 lb less than heavy smokers, 3.1 lb less than ex-smokers, 3.6 lb less than never smokers Smoking rate: lower rate related to less weight gain Retrospective report Hughes and Hutchinson (1963) 37 smokers and 19 ex-smokers with pulmonary emphysema followed 23 years Smokers lost 0.32 lb/yr, ex- smokers gained 1.17 lb/yr; significant difference Smoking self-report; pulmonary emphysema population TABLE 3.-Continued Study Design and sample Major results Moderator variables Limitations Jenkins et al. ( 1973) 2,318 men (546 never smokers, 359 previous quitters, 547 light smokers, 666 heavy smokers), aged 39-49, 11 California corporations in Western Collaborative Group Study; changes asses& since age 25; 1960-1969 studv Weight loss more likely for light and heavy smokers than never smokers and quitters Smoking self-report; weights not presented Kramer ( 1982) 175 subjects, commercial cessation prcgram (41 nonparticipants or nonlocated, 59 quitters. 75 continuing smokers) 2 l-year followup 76% nonsmokers, 56% smokers gained weight: these smokers mean gain 1.7 lb, these nonsmokers mean gain 3.0 lb All date self-report; high attrition. data loss; presentation incomplete Lund-Larsen and Tretli (1992, 12,329 men and women, aged 2049. cardiovascular disease project; 2 screenings 3 years apart Stpokers mean and relative weight less than nonsmokers; female quitters gained 5.95 lb, male quitters 7.64 lb; smoking- starter men lost 1.96 lb, women 5.5 lb; smokers and nonsmokers little/no change Sex: men. women weight change/smoking cessation and initiation similar Self-report Manley and Roland (1963) 39 male, 55 female smokers, cessation program; randomly assigned, 1 of 3 4-week treatments or attention placebo control; Bmonth followup; CC verification 31% abstinent at followup: abstainers averaged 10.93 lb gain, relapsers 6.92 lb Relapser definition unclear TABLE 3.-Continued Study Design and sample M.?+ar zwults Moderator variables Limitations Noppa and Rengtsmn (1980) 1,362 Swedish women, aged 38-W Current smokers leaner than nonsmokers; At 6 years, quitters gained 7.72 lb; smoking-starters lost 1.54 lb, nonchangers gained 3.09 lb Smoking self-report Pincherle (19711 222 upper-class male quitters; followup 2 1 year after first visit 28% gained weight; 22% last Smoking self-report; limited population; incomplete report; no weights presented Powell and McCann (1981) 29 women, 22 men, 5day cessation project; 2- and 6month followup At 2 months, 54% gained weight, range 3-20 lb. mean 8.96 lb; all subjects mean 4.69 lb Smoking self-report; no separate abstainer, smoker data; small sample sire Puddey et al. (1985) 66 cessation program volunteers, pair-matched by age, sex, body mass index; randomly assigned experimental, control groups; 2- week baseline, &weak treatment, 6week followup; thiocyanate, Co verification 14 quitters gained 3.97 lb; controls 0.44 lb Small sample sire Rabkin (1964) 40 male, 67 female smokers, assigned to 3 cessation groups; followup 3 weeks w&completion; 67.3% gained weight, average 1.76 lb; skinfold increase 6.6 mm bicchet&l verification No age. age at smoking start, rate. relative weight, anxiety correlation to male or female weight change Small sample size; weight self-report TABLE 3.-Continued Study Design and sample Major results Moderator variables Limitations Rantakallio and 12.068 pregnant women. n. No smoking/nonsmoking Pregnant women only; Hartikainen-Sorrin Finland, 1966, 15% smokers pregnancy weight gain difference smoking self-report; (1981) (smoked after 2 months pregnancy weight gain pregnant); nonsmoking controls data only matched for age. parity, place of residence, marital status Rush 162 low-income urban pregnant Mean pregnancy weight gain Smoking rate: higher rate Pregnant women only; (1974) women, no known medical lower for smokers (0.73 lb/wk) related to lower pregnancy smoking self-report; problems, < 140 lb preconception than nonsmokers (0.90 lb/wk) weight gain pregnancy weight gain weight; had borne low data only birthweight infant; randomized controlled nutritional supplementation trial Schoenenberger 4,421 male MRFIT volunteers, With MRFIT intervention, Smoking self-report; (1982) aged 35-57, good health but significant body weight decrease confounded by risk upper lO-15% coronary risk in smokers (mean 4.6 lb). factor reduction factor score; randomly assigned nonsmokers (mean 5.6 lb), program participation; to intervention or control groups; reducers (mean 3.75 lb); quitters restricted population followup 3 annual visits average weight change minimal (mean 0.55 lb) Seltzer 794 adult white male veterans, At admission, ex-smokem 5.9 lb White veterans; (1974) average age 45, Normative Aging heavier than nonsmokers, 8.1 lb smoking self-report Study; screened for "high" health heavier than currant smokers; at level, geographic stability; 214 5 years, quitters gained 8.0 lb, screened at 5 years continuing smokers 3.5 lb TABLE 3.-Continued Study Design and sample Major results Moderator variables Limitations Stamford et al. (1966) 13 sedentary women, 48day su-ful quitters; l-year follow"p At 48 days, weight increased 4.85 lb; at 1 year, quitters increased 18.07 lb; 3 relapsers reduced weight to baseline levels; per hydrostatic weighing, gain was 96% fat Small female sample; smoking self-report Tuomilehto et al. (198.5) 10,940 cardiovascular disease prevention program participants. aged 25-59, random sample. e. Finland; selectees with high blood pressure or hypertensive medicine asses& 5 years apart; smoking data from 2,264 Quitters body nmss increased 2.31 lb/m*; starting smokers decreased 1.46 lb/m' Smoking self-report; hypertensives Vandenbroucke et al. (1984) 3,091 Netherlands civil servants, spouses (1,583 men. 1,508 women), aged 40-65, general health exam: 25year followup 76.6% lean, 65.1% obese men smoked; 22.1% lean, 11.3% obese women smoked Smoking self-report; restricted population Gottenborg 1981; Khosla and Lowe 1971) documented increasing weight differences between smokers and nonsmokers with advancing age. Typically, aging smokers failed to gain as much weight as aging nonsmokers. Three evaluations systematically compared males with females (Bjelke 1971; Kopczynski 1972; Sutherland et al. 1980). Two of the three (Bjelke 1971; Sutherland et al. 1980) reported the differences in body weight between smokers and nonsmokers to be greater in females than in males. Longitudinal Evaluations of Smoking and Body Weight Table 3 presents the results of 43 longitudinal evaluations of the effects of smoking on body weight. Consistent with the cross-section- al evaluations, the overwhelming majority (86 percent, 37 of 43) present evidence that smokers who quit smoking gain weight, that people who quit smoking gain more weight than nonsmokers, and that people who initiate smoking lose weight relative to nonsmokers. Of the six studies that did not find these relationships, three limited their examination to smoking and weight changes in pregnant women (Gormican, Valentine, Satter 1980; Haworth et al. 1980; Rantakallio and Hartikainen-Sorri 1981). two relied on participants making broad cardiovascular risk factor reduction efforts in subjects at high risk for cardiovascular disease (Hickey and Mulcahy 1973; Holme et al. 19851, and the remaining study supplied incomplete reports of the data (Kramer 1982). Of those studies on the effects of smoking cessation on weight, the length of followup ranged from 4 days to 7 years. According to these investigations, those who quit smoking gained an average of 6.16 lb (range: 1.76 to 18.07) during the year after cessation. Daily cigarette consumption was the only moderator variable that received sufficient attention in this group of studies reaching specific conclusions. Seven of nine studies (78 percent) (Blitzer, Rimm, Giefer 1977; Bosse, Garvey, Costa 1980; Comstock and Stone 1972; Fried- man and Siegelaub 1980; Hall, Ginsberg, Jones 1986; Howell 1971; Rush 1974) reported a positive relationship between cigarette consumption and weight change; that is, as pretest cigarette consumption increased, postcessation weight gains also increased. Two studies (Carney and Goldberg 1984; Rabkin 1984) did not find a relationship between cigarette consumption and postcessation weight gain. In summary, there is substantial evidence of an inverse relation- ship between cigarette smoking and body weight. Of 71 studies reported since 1970, 62 (87 percent) collectively indicate that smokers weigh less than nonsmokers and that people who quit smoking gain weight. Older smokers, females, and those smoking approximately one pack of cigarettes/day may experience the 431 largest weight control effects of cigarette smoking. Smokers who smoke heavily tend to gain the most weight following smoking cessation. These generalizations are consistent with reviews based on other studies reported since 1880 (Grunberg 1986a). Not all smokers who quit smoking gain weight. Further, for ex-smokers who do gain weight, the amount of weight infrequently poses a serious health risk. The Role of Nicotine Animal studies indicate that nicotine administration results in weight loss or decreased weight gains and that cessation of nicotine results in body weight gains greater than those of controls (Bowen, Eury, Grunberg 1986; Grunberg 1982, 1985, 198613; Grunberg, Bowen, Morse 1984; Grunberg, Bowen, Winders 1986; Grunberg, Winders, Popp 1987; McNair and Bryson 1983; Morgan and Ellison 1987; Schechter and Cook 1976; Wager-Srdar et al. 1984; Wellman et al. 1986). Most of these studies report inverse dose-response relation- ships between nicotine and body weight. Becent research on nicotine polacrilex gum with humans corrobo- rates the role of nicotine in body weight effects. Fagerstriim (1987) reported that subjects who quit smoking were much less likely to gain weight when they consistently used nicotine polacrilex gum. Abstinent subjects who regularly used the gum gained less than 2 lb at a 6month followup. In contrast, the infrequent gum users gained almost 7 lb (p< 0.05). Emont and Cummings (1987) reported a significant negative relationship (r=-O.37) between the number of pieces of nicotine polacrilex gum chewed per day and weight gain for heavy smokers ( > 26 cigarettes/day). No such relationship between gum use and weight gain was observed for lighter smokers (< 26 cigarettes/day). Mechanisms Underlying The Relationship Between Smoking and Rody Weight The inverse relationship between smoking and body weight may result from changes in energy intake, changes in energy expendi- ture, or both. Energy intake involves dietary intake. Energy expenditure is affected by behavioral factors (physical activity1 and biological factors (e.g., metabolism). These potential mechanisms are examined below. Dietary Intake Several prospective investigations have evaluated dietary intake changes following smoking cessation in humans. Hatsukami and coworkers (1984) hospitalized 27 smokers for a 7day period. After a 3day baseline, 20 of the subjects were deprived of smoking for 4 days 432 while the remaining 7 served as a control group. During this I-day period of abstinence, caloric intake increased significantly (from 1,397 to 1,651 kcal), which corresponded with a significant 1.76lb increase in weight. In the most comprehensive study to date, Stamford and coworkers (1986) evaluated changes in dietary intake, physical activity, and resting metabolic rate in 13 sedentary females who quit smoking for a 48day period. Following smoking cessation, mean daily caloric consumption increased by 227 kcal, which accounted for 69 percent of the variance in postcessation weight gain (4.85 lb). Robinson and York (1986) followed 11 smokers who quit for 7 days. Mean dietary intake significantly increased, but changes in resting metabolic rate were not observed. Dallosso and James (1984) followed 10 subjects for 6 weeks after they participated in a stop smoking clinic. There was a 4percent drop in resting metabolic rate in smokers who quit, a drop which was reliable when the data were expressed per kilogram of body weight. The average dietary intake increased by 6.5 percent, but this difference did not reach statistical significance. Preliminary results of a recent investigation indicate gender differences in the effects of short-term smoking cessation on body weight and food intake (Klesges, Meyers et al. 1987). Female smokers who quit for 1 week increased their body weight and dietary intake significantly more than male smokers who quit. This sex difference is consistent with animal studies (Grunberg, Bowen, Winders 1986; Grunberg, Winders, Popp 1987). Given females' marked concerns regarding postcessation weight gain (Klesges and Klesges, in press), future studies will need to investigate possible gender differences in response to smoking cessation. Several studies indicate that smokers may differ from nonsmokers in their intake of sweet-tasting simple carbohydrates (sugar) in particular. In a human laboratory study, Grunberg (1982) observed that smokers who were allowed to smoke ate less sweet food than smokers who were not allowed to smoke or nonsmokers. Smokers not allowed to smoke also reported the greatest preference for sweet foods. There were no differences among the three subject groups in consumption of other types of foods. Rodin (1987) conducted a prospective study in which food intake after smoking cessation was carefully evaluated. Smokers who gained weight after quitting smoking increased their sugar consumption in particular. Further, smokers increase consumption of sweet snack foods when they are deprived of cigarette smoking (Duffy and Hall, in press; Perlick 1977). On the other hand, two early investigations (Bennett, Doll, Howell 1970; Richardson 1972) found higher sugar consumption in smokers relative to nonsmokers. However, Richardson (1972) found that this difference was because of low-sugar intake in ex-smokers, while Bennett, Doll, and Howell (1970) argued that the differences 433 were largely due to increased added sugar intake because of hot beverage consumption. These two studies, which are inconsistent with the more recent studies, did not carefully measure all food intake and did not assess intentional changes in food intake to control body weight. Several animal experiments have documented that food intake decreases during nicotine administration and increases after admin- istration has ceased and that these changes in food intake corre- spond with changes in body weight (Bowen, Eury, Grunberg 1986; Grunberg 1982; Grunberg, Bowen, Winders 1986; Levin et al. 1987; McNair and Bryson 1983; Wager-Srdar et al. 1984). Consumption of sweet foods by male rats is particulary affected by nicotine (Grun- berg 1982; Grunberg et al. 1985). However, nicotine also reduces bland food intake in female rats and has a greater effect on body weight of female rats than of male rats (Grunberg, Winders, Popp 1987; Grunberg, Bowen, Winders 1986; Levin et al. 1987). Several investigations have reported that changes in body weight in animals also occur without observing decreases in food intake as the result of nicotine administration (Grunberg, Bowen, Morse 1984; Schechter and Cook 1976; Wellman et al. 1986). In one investigation, chronic exposure to cigarette smoke reduced body weight and food intake in rata; however, hamsters exposed to cigarette smoke decreased body weight without reducing food intake (Wager-&&r et al. 1984). Several methodological factors complicate these results, including the use of different strains of animals, different routes of administration and dosages of nicotine, and whether acute versus chronic effects of nicotine were reported. However, these results indicate that more than the mechanism of food intake was involved in producing nicotine- and smoking-related weight changes. Data from short-term human studies and several animal experi- ments indicate that dietary intake is involved with smoking-related energy imbalance. Based on self-reported cross-sectional surveys, it has been reported that smokers' dietary intake is the same as (Albanes et al. 1987; Fehily, Phillips, Yarnell 1984; Fisher and Gordon 1985; Matsuya 1982) or significantly higher than (Picone et al. 1982; Stamford et al. 1984a,b) that of nonsmokers while the smokers simultaneously maintained a lower body weight. Assuming that smokers are not consistently biased in their reports of dietary intake, it appears that either differences in physical activity or metabolic rate are maintaining the body weight differences between smokers and nonsmokers. Physical Activity The data available from cross-sectional investigations, short-term prospective studies, and animal investigations seem to indicate that changes in physical activity do not play a role in either differences in 434 body weight between smokers and nonsmokers or the weight gain associated with smoking cessation. Some cross-sectional investiga- tions have found that smokers have lower levels of physical activity compared with nonsmokers (Karmas 1981). Others have not found differences in physical activity and physical fitness between smokers and nonsmokers (Gyntelberg and Meyer 1974; Stamford et al. 1984b; Stephens and Pederson 1983). A recent review (Blair, Jacobs, Powell 1985) that addressed the relationships among exercise, physical activity, and smoking concluded that smoking and physical activity are negatively associated; however, the relationship was extremely weak and variable. Animal studies on the relationship between nicotine and physical activity have generally found that physical activity plays a small role or fails to correspond to decreases in weight during nicotine administration (Bowen, Eury, Grunberg 1986; Cronan, Conrad, Bryson 1985; Grunberg and Bowen 1985b). One study found that decreases in physical activity after cessation of nicotine appeared to contribute to postdrug body weight increases (Grunberg and Bowen 1985b), but this effect was quite small and occurred only in males. A few prospective human investigations have evaluated physical activity changes following smoking cessation (Hatsukami et al. 1984; Hofstetter et al. 1986; Klesges, Brown et al. 1987; Rodin 1987; Stamford et al. 1986). These investigations found no changes in physical activity as a result of smoking cessation. Metabolic Rate Metabolic rate is an important consideration in energy imbalances associated with smoking cessation because approximately 75 percent of total energy expenditure is in the form of metabolism (Bernstein et al. 1983; Ravussin et al. 1982). Metabolism increases as the result of acute nicotine administration and immediate effects of smoking (Ghanem 1973; Ilebekk, Miller, Mjos 1975; Robinson and York 1986; Schievelbein et al. 1978; Wennmalm 1982). The major question, however, is whether these effects persist long enough to have a direct impact on body weight. Given that (1) smokers do not have higher levels of physical activity compared with nonsmokers (Blair, Jacobs, Powell 1985). (2) some studies report smokers' dietary intakes are the same as or higher than those of nonsmokers (Picone et al. 1982; Stamford et al. 1984a,b), and (3) smokers maintain lower body weights than nonsmokers, it is reasonable to postulate that changes in metabolism contribute to the relationship between smoking and body weight. Additionally, there are several reports in the literature on animals that have documented nicotine-induced reductions in body weight without a concomitant reduction in food intake (Grun- berg, Bowen, Morse 1984; Schechter and Cook 1976; Wellman et al. 1986). 435 Direct evidence supporting a chronic metabolic mechanism that modulates the smoking/body weight relationship is beginning to emerge. Metabolic rate was chronically measured in a study of rat and hamster exposure to cigarette smoke (Wager-Srdar et al. 1984). Higher resting metabolic rates were observed on only one of the test days compared with the pretest in the rat investigation, while no significant differences were observed in the hamster study. Another recent investigation (Wellman et. al. 1986) evaluated brown adipose tissue (BAT) thermogenesis at different levels of nicotine and caffeine injections. No differences in BAT thermogenesis were observed in response to either nicotine or caffeine. The group that received a combination of caffeine and nicotine showed a 63 percent increase in BAT thermogenesis. The few studies that have evaluated metabolic rate changes in response to smoking cessation in humans have produced inconclu- sive results. Three investigations found metabolic changes after cessation in human smokers. An early report (Glauser et al. 1970) found decreases in oxygen consumption for seven male subjects who quit smoking for 1 month (neither food intake nor physical activity was monitored). A more recent investigation found a 4-percent drop in metabolic rate (reliable when data were expressed per kilogram of body weight) and no significant increase in dietary intake for 10 subjects who quit smoking for 6 weeks (Dallosso and James 1984). In the only study that used a respiration chamber, Hofstetter and others (1986) reported that total energy expenditure was 10 percent higher during a 24-hr period of smoking versus a 24-hr period of abstinence in eight smokers. No changes were observed in physical activity or mean basal (sleeping) metabolic rate (dietary intake was held constant). However, this difference in energy expenditure disappeared after 24 hr. Three investigations did not find a change in metabolic rate as the result of smoking cessation. Burse and associates (1982, 1975) did not observe changes in resting metabolism in a sample of four smokers who quit for 3 weeks. This investigation did find reliable increases in desire for food, however. In another study, 11 smokers were studied after a `I-day period of smoking abstinence (Robinson and York 1986). Total energy expenditure following a meal did not change during the cessation period. Stamford and colleagues (1986) failed to fmd changes in oxygen consumption in 13 subjects who quit smoking for 48 days. This investigation did find marked dietary intake changes that accounted for 69 percent of the variance of postcessation weight gain. There are several possible explanations for the inconsistency observed in the literature on metabolic rate. Different investigators have used different criteria (e.g., resting oxygen consumption, BAT thermogenesis) for operationalizing metabolism. It is possible that 436 previous dieting history (Brownell et al. 1986) and the use of nicotine polacrilex gum (Fagerstrom 1987) may directly impact the metabolic response to smoking cessation. It is not clear what the metabolic response to nicotine with added agents is likely to be. For example, one study found that while neither nicotine nor caffeine alone produced a change in BAT thermogenesis, the two combined increased thermogenesis by 63 percent (Wellman et al. 1986). This finding is particularly interesting given that smokers may be more likely to drink caffeinated beverages than nonsmokers (Blair et al. 1980). Finally, the available literature on human studies used very small subject groups, making it impossible to detect subtle but potentially meaningful changes in resting metabolic rate. The small sample sizes do not allow for an evaluation of variables that may potentially moderate the metabolic response to smoking cessation. Summary of Mechanisms Literature Changes in dietary intake appear to be involved in weight gains after cessation of smoking or cessation of nicotine administration. Physical activity plays little or no role in the relationship between smoking and body weight. The data on metabolic contributions to postcessation weight gain are suggestive, but further research is needed. Unfortunately, much of the relevant human research literature is characterized by small sample sizes, short followup evaluations, and inadequate evaluations of energy balance following smoking cessation. To date, only one investigation has comprehen- sively evaluated (i.e., simultaneous assessment of dietary intake, physical activity, and metabolic rate) energy balance changes as the result of smoking cessation. This was a sample of 13 sedentary females followed for 48 days (Stamford et al. 1986). Comprehensive, prospective evaluations of energy balance changes in response to smoking cessation are needed. Additionally, no study has evaluated possible long-term changes in dietary intake, physical activity, and metabolic rate as a result of smoking cessation. The longest followup period reported in the literature to date is 2 months (Dallosso and James 1984). Finally, evaluation of potential moderator variables of dietary intake, physical activity, and metabolic rate as the result of cessation is needed. Gender (Grunberg, Winders, Popp 1987; Klesges, Meyers et al. 1987), previous dieting history (Brownell et al. 1986; Hall, Ginsberg, Jones 1986). pretest levels of lipoprotein lipase (Carney and Goldberg 1984), and the use of nicotine polacrilex gum (Fagerstrom 1987) appear to be important variables influencing weight gain and need further investigation. Does the Relationship Between Smoking and Weight Promote Either the Initiation or Maintenance of Smoking Behavior? Some research attention has been given to body weight as a potential moderator of smoking initiation, maintenance, and cessa- tion. Unfortunately, many investigations do not report weight- related issues (Borkon, Baird, Siff 1983; Eiser et al. 1985; Pederson and Lefcoe 1976; Perri, Richards, Schultheis 1977). The investiga- tions that have evaluated these issues consistently report relation- ships between body weight and smoking initiation (Charlton 1984a) and maintenance (Klesges and Klesges, in press). A survey of 16,000 school children (Charlton 1984a) in England found that the heaviest regular smokers were the most likely to agree that smoking controls weight (42.2 percent) compared with those students who never smoked (16.6 percent). Agreement in- creased with increased levels of smoking. More girls than boys agreed with this statement, and girls were also more likely to be regular smokers. Charlton (1984b) also reported that among the perceived effects of smoking, smokers viewed "calming the nerves" as the most popular reason (72 percent) followed by "smoking keeps your weight down" (39 percent). Other investigations are consistent with the Charlton (1984a,b) report. In a recent study of 1,000 adolescents in Canada (Feldman, Hodgson, Corber 1985), significantly more girls than boys were concerned about becoming overweight (36 vs. 14 percent, p