more weight to large versus small samples. Table 2 provides more detailed information
(e.9.. Y-percent confidence intervals for weight gain and relative risk) regarding each
of these investigation<.
As indicated in Tables I and 2. the average sample sire of these investigations was
I .34X (range=ZX-9.539). The followup period ranged from I month to 5 years. with a
median followup period of 7 years. Consistent w'ith previous reviews of the smoking
and body weight literature (Klesges et al. 19X9: US DHHS I9XXa). the adjusted average
weight gain among smokers u ho quit was approximately 5 pounds (mean=4.6:
range= I .&I I .2 pounds). The weight gain among smokers who quit was considerably
greater than the adjusted average gain of 0.X pounds observed among subjects who
continued to smoke (range=0 to +3.S pounds). Thus. although variability of wseight
gain is quite marked (Tables I and 2). smoking cessation produces approximately a
l-pound greater weight gain thail that associated with continued smoking.
A commonly reported, but erroneous. estimate regarding postcessation weight gain
is that one-third of smokers gain weight after smoking cessation. one-third maintain
body weight. and one-third lose weight after cessation (US DHEW 1977). In the five
investigations providing detailed information regarding changes in body weight. the
actual percentage of quitters gaining weight appears to be much greater than previously
estimated. Considering the results of all five studies and adjusting for sample size, 79
percent of those who quit smohing experienced a weight gain (range=%-87 percent).
Over the same followup period, an adjusted average of 56 percent of continuing smokers
experienced an increase in body weight (range=33-62 percent) and. as presented above.
the average amount of weight gain was less among continuing smokers.
Data allowing computation of a relative rish estimate of weight gain after smohing
cessation w'ere available from five investigations. This relative risk estimate compares
the likelihood of weight gain in quitters versus continuing smokers. That is. a higher
relative risk ratio indicates that the percentage ofquitters who ssined wright was higher
compared with that of corresponding continuing smokers. Overall. the risk of n,eight
gain after cessation was 45 percent greater for quitters (mean=l.35. ranFe=I .3 l-l .7S)
than for continuing smokers. This increased rish of weight gain was consistent across
differing follow up periods. appearin,      _
                    (7 as earlv as 6 weeks (Rodin 19X7: relative rish
(RR)=I.75) and lasting up to 6 bears after smohinf cessation (hoppa and Bengtsson
19x0: RR=I 3 I ). Additionall) _ one investigation found the relative risk ofgaining more
than 1 pounds after smohin, (7 cessation to he I .3X (Bosst!. Carve!. Costa 19X0). In
another investigation. the rish of gaining more than IO pounds was XX percent higher
for quitters than for continuin g smohers (RR= I .XX) (Friedman and Sieselaub 19X0).
Although the rish of yining more than IO pounds appears to be almost 90 percent
greater among quitter\ than con1lnuin, `7 smohcrs (Friedman and Siegelaub IYXO). actual
occurrence of I ()-pound L\ eight fains uas reluti\el\ Iou (20.3 ~4. 10.X percent amonp
quitters and conlinuin, `7 smohers. respccti\el\ 1. Frkman and Siegelaub (19x0). with
a large sample of quitters (`1:=7.7.3X) xid continuin g smohers (N=6.X() I ). presented the
percentage\ of those gaining 20 pounds or mow o\ t`r ;I median I X-month follow up.
Among males. 3.7 percent of those i\ ho quit smohin, 17 gained more than 20 pounds
compared 14 ith 0.9 percent of those u ho continued to smohe. Amon females, 3. I



`I'AHLE: 2.--Continued

Kclallvc IWh ol ~;IIIlIn~ >I0 Ih=I.XX
`)5'/; (`I ( t .70-1.0X)


TABLE 2.--Continued



TABLE 2.--Continued

Quit
period

Average gain
+SD (lb)

Re\ult\

Rwl~n t tYX7)

Sctl/er t lY7-t)

Quitters: 3.1+3.Y     (?.I-S.7)
Males: S.Y+4. I     (3.X-X. I )
Fcmalrs: 3.2f3.0   (2.04.h)

Continuing
>mohera: 0.753.7    (60.2-1.5)
M&x 7 `+3 1
      A.___.     (O.Y-3.5)
Femalw 4.4k3.0 t-1.34).4)

Quitters: 3. IX

(`ontinumg
smohrrs: 0.30

Quitters: 7.Y

No chance or Io\t
41.7'/;(10)
66.7% ( I?)

ucight=l.75


TABLE 2.-Continued


percent of those who quit smoking gained more than 20 pounds compared with 1.6
percent of those who continued to smoke.

In summary. while approximately four-fifths of smokers who quit will gain weight
after cessation. average weight gain is approximately 4 pounds greater than that
expected among continuing smokers. The risk of weight gain after cessation is 45
percent greater than the risk associated with continued smoking. although individual
weight gains of 20 pounds or more are rare.

Although weight gain is common after cessation. little is known concerning the types
of individuals at risk for substantial increases in body weight. Researchers have
concluded that women. moderate smokers. and older smokers have the greatest weight
control effect from smoking (US DHHS 1988a). although the tremendous variability
in body weight changes after cessation has yet to be explained. That is. while the
average weight gain after smoking cessation is approximately 5 pounds. individual
responses range from weight loss to a weight gain exceeding 20 pounds. Studies are
needed that focus carefully on individuals at risk of excessive weight gain after smoking
cessation and the differences between these individuals and those who do not gain
weight.

Additionally. investigators hypothesize that the relationship between smoking and
body weight is attenuated by other health behaviors (Marti et al. 1989). Although the
effects of smoking to reduce body weight are acknowledged, individuals who smoke
are more likely than nonsmokers to have unhealthy lifestyles associated with increased
body weight (e.g.. lower levelsof physical activity and higher dietary intakes)(Klesges.
Eck et al. 1990: Chapter I I ).

CAUSES OF POSTCESSATION WEIGHT GAIN

Cross-sectional and longitudinal studies clearly indicate the inverse relationship
between smoking and body weight in humans and between nicotine and body weight
in animals (Grunberg 1986: Klesges et al. 1989: US DHHS 1988a: Winders and
Grunberg 1989). However, no study has included a simultaneous evaluation of the
long-term changes in all of the variables that may account for this relationship. including
food intake. physical activity. and energy expenditure. Of the currently published
investigations, the longest followup period evaluating all three aspects of the energy
balance equation has been 8 weeks (Stamford et al. 1986). A recent study evaluated
food intake and physical activity changes over a 26-week followup but did not include
metabolic measures (Hall et al. 1989). Short-term evaluations do not allow for an
adequate determination of predictors of weight gain. This review focuses on those
studies that have directly evaluated either food intake, physical activity, and/or meta-
bolic rate as a function of smoking cessation. nicotine administration. or nicotine
deprivation. The available data on changes in the energy balance equation that result
from smoking cessation are summarized below.


Food Intake

Most short-term evaluations (e.g.. 3 days or less) found that food intake, particularly
the consumption of sweet foods and simple carbohydrates, increases after smoking
cessation. For example in a I -day experiment. Grunberg ( 1982a) reported that smokers
who were allowed to smoke ate fewer sweet foods, but consumed similar amounts of
non-sweet foods. compared with nonsmokers and smokers not allowed to smoke. This
between-subjects laboratory study was short term and did not measure body weight
changes. In another short-term study. Hatsukami and colleagues (1984) hospitalized
27 smokers for 7 days. After a 3-day baseline, 20 of the subjects were deprived of
smoking for 4 days while the remaining 7 served as a control group. During this 4-day
abstinence. caloric intake increased significantly in the abstinence group and was
accompanied by a I .76-pound increase in weight compared with baseline. Recently,
Duffy and Hall ( 1988) assessed smokers who differed in degree of eating disinhibition.
defined as eating that occurs in situations in which self-control behaviors are disrupted
(e.g.. binge eating). Smokers who were allowed to smoke before eating ice cream did
not show food consumption differences as a function of level of disinhibition. How-
ever, results for smokers who had abstained from smoking for 24 hours showed a
different pattern. Abstaining smokers who scored high on eating disinhibition ate more
than three times (273.6 g) as much ice cream as those who scored low (86.4 g) on eating
disinhibition. The results from this investigation indicate that dietary changes follow-
ing smoking cessation may vary as a function of dieting history. use of cigarettes to
curb appetite. and other wei_pht history variables.

Some prospective investigations have qualitatively ashed participants who quit
smoking if they believed that their dietary intake had changed. These studies also
reported that food intake increases after cessation. For example. Manley and Boland
(19X3) examined the side effects experienced by 94 subjects quitting smoking and
whether these side effects varied as a function of relapse. On a withdrawal rating
system. those whoquit smoking rated themselvesas furthest from"optimal"at followup
on general appetite and overeating. On a separate rating scale. abstainers also ga1.e
higher ratings than relapsers at followupon "eating more." In astudy of 53 self-quitters.
Black and coworker\ ( IYXX) found that of those reporting that they ate more. average
weight gain uas 6.9 pounds. In contrast. of those reporting that they ate the same or
less. average Mcifht gain ua'r I .-I pounds.

Unfortunately. there are few prospective human investigations that have attempted
to quantify carefully food intahe changes over time among subjects after quitting
smohing. The\e 5tudies penerally indicate that food intahe increases after cessation;
however, result\ v'ary greatly across investigation\. Of eight studies to date. t&o
reported clear increases in food consumption after cessation (Leischow and Stitzer
IYXY: Stamford et al. 19X6). four provided qualified support for increased food
consumption after cessation (Hall et al. 1989: Kles,_ges et al.. in press: Perkins. Epstein.
Pastor I YYO: Rodin 19x7). and tv+ o reported no changes in food intake after cessation
(Dallosso and James 19x3: DiLorenTo et al. 1988).

In what may be the most comprehensive evaluation to date of change in energry
balance. Stamford and colleagues ( 1986) analy.zed changes in food intake. physical


activity, and resting metabolic rate in 13 sedentary females who quit smoking for 48
days. Mean daily food intake increased by 177 kcal and explained 69 percent of the
variance in changes in weight (3.85 pounds). No changes in physical activ,ity or resting
metabolic rate were observed.
To evaluate dietary changes after cessation. Leischow and Stitzer ( 1989) assigned
subjects, in an inpatient setting, to either smoke-ad-libidum (N=6) or quit-smoking
(N=9) conditions for at least I4 days after a 4-day baseline period. Results revealed a
significant difference in weight gain (p<O.OS) between smokers and those who quit
smoking (2.0 vs. 4.7 pounds. respectively). The weight gain in those who quit smoking
was associated with a significant increase in food intahe over time compared with
continuing smokers.
Four investigations have provided qualified support for dietary changes after cessa-
tion. Perkins, Epstein, and Pastor (1990) evaluated caloric intahe, resting energy
expenditure (REE). and physical activity in seven female smokers for 3 weeks. vvhich
included normal smoking (week I ). smoking cessation (week 3). and resumption of
smoking (week 3). Total caloric intake did not increase during the week of cessation.
However, once smokers resumed smoking during week 3. caloric intake decreased
significantly. Caloric intake from alcohol. however. rose from 3 I9 kcal per day in the
first week to 432 kcal per day during the week of abstinence. When subjects resumed
smoking during the third week, alcohol intake dropped to I19 kcal per day. During the
cessation week, REE did not decrease compared with baseline. However. a significant
increase in REE was observed when subjects resumed smoking compared with the week
of abstinence (p<O.OOl ). No changes in physical activity were observed.
Rodin (1987) evaluated changes in food intake and physical activity in 24 subjects
who quit smoking and I8 smokers who failed to quit smoking. Subjects w'ho quit
smoking gained an average of 3.2 pounds over the R-week study. Consistent with the
literature concerning animals as subjects and some studies using humans (Grunberg
1986; Winders and Grunberg 1989). smokers who gained weight after stopping smok-
ing increased their carbohydrate consumption. particularly sugar. This increase was
accompanied by decreased protein consumption.  However. these subjects did not
increase their total food intake nor did they decrease their levels of physical activity.
Levels of physical activity generally increased.
Hall and coworkers (1989) assessed changes in food intake and physical activity
among 9.5 subjects who enrolled in a stop-smoking program. In contrast to all other
investigations reviewed in this Section, Hall and coworkers ( 1989) evaluated long-term
changes in food intake and physical activity (for a 6-month followup). Caloric intake
increased significantly in one group and marginally in another group during the first 8
weeks of abstinence. Both sugar and total fat increases were noted in the group that
significantly increased energy intake. Total dietary intake increased approximately 200
kcal per day over the 8-week period. In assessing 6-month changes, Hall and coworkers
(1989) reported a gender difference in caloric intake with time. Among men who quit.
mean daily caloric intake decreased by almost 1,000 kcal from a mean of 3.014 kcal
during week I to 2,035 kcal at week 26. Among women. caloric intake remained stable
(mean= I .84 I kcal at week I: mean=1 .X67 kcal at week 16). However. vveight con-
tinued to increase for both groups. From the 13.week to the 6month followup. men


increased their weight 3.56 pounds (X.65 pounds total). and women increased their
weight by 4.53 pounds ( 10.34 pounds total). NO changes in physical activity were
observed. Weight continued to increase despite no changes from baseline in dietary
intake and physical activity in female ex-smokers and despite decreases in dietary intake
and no physical activity changes in male ex-smokers.

Klesges and coworkers (in press) reported gender differences in response to smoking
cessation. In this study, the food intake and physical activity of 68 smokers and
nonsmokers were evaluated during a 2- week period. At the end of the first week, the
smokers were paid to quit smoking, and 36 percent were successful at remaining
abstinent for the entire week (confirmed by carbon monoxide (CO) readings). Non-
smokers continued to monitor their food intake and physical activity. At the end of the
second week, subjects were allowed to return to smoking. In this investigation, female
smokers who quit smoking increased their body weight in comparison with non-
smokers. Smokers who quit increased their consumption of mono- and polyunsaturated
fats and decreased their intake of fiber. In contrast, males who quit smoking did not
change either their weight or dietary intake compared with males in the other groups.
No changes in physical activity were detected in any of the groups.

Dallosso and James (1984) reported on IO subjects who quit smoking and were
observed for 6 weeks after they participated in a stop-smoking clinic. Resting metabolic
rate dropped by 4 percent in smokers who quit, a drop which was significant only when
the data were expressed as per kilogram of body weight. The average food intake
increased by 6.5 percent, but this difference was not statistically significant.

DiLorenzo and colleagues (1988) evaluated changes in body weight and caloric
consumption in I6 subjects who quit smoking for 5 weeks compared with I I subjects
who continued to smoke and I6 nonsmokers studied over the same time period.
Subjects who quit smoking gained an average of 5 pounds over the 5 weeks; the smoking
and nonsmoking control groups did not change body weight significantly (p<O.OOOl ).
This weight gain was not associated with changes in dietary intake.

Physical Activity

In contrast to the findings on dietary intake and smoking cessation. the available data
indicate that change in physical activity does not play a role in either differences in
body weight between smokers and nonsmokers or the weight gain associated with
smoking cessation. The small number of prospective investigations has generally
reported unchanged physical activity after smoking cessation (Hall et al. 1989; Hat-
sukami et al. 19X-l: Klesges et al.. in press: Perkins. Epstein. Pastor 1990; Stamford et
al. 1986). and those that found a change in activity reported an increase in physical
activity after smoking cessation (Leischou and Stitzer 1989: Rodin 1987). The
literature consistently indicates that reduced physical activity after cessation cannot
account for postcessation weight gain.


Energy Expenditure

An important and often overlooked variable in energy imbalance leading to weight
gain is REE. Approximately 75 percent of total energy expenditure is in the form of
metabolism (Ravussin et al. 19X2). Ample indirect evidence supports the hypothesis
of increased energy expenditure in smohers. That is. given that smohers do not have
higher levels of physical activity compared with nonsmohers. the only known
mechanism remaining to explain the energy imbalance is \orne aspect of metabolism
(Blair. Jacobs. Powell 19X5): smokers' dietary intahes may be the same or higher than
those of nonsmokers (Picone et al. 19X2; Stamford. Matter. Fell. Sady. Cresanta et al.
19X4: Stamford. Matter. Fell. Sady. Papanek et al. I9X-4): mohers maintain lower body
weights than do nonsmokers (Klesges et al. 19X9: US DHHS I9XXa): and weight gain
has been reported in individuals quitting smokin g without any dietary and physical
activity changes (DiLorenro et al. 1988: Hall et al. 19X9). Additionally. several reports
document nicotine-induced reductions in body weight in laboratory animals without a
concomitant reduction in food intake (Grunber g. Bouen. Morse 19X1: Schechter and
Cook 1976; Wellman et al. 19X6). However. those few studies that have evaluated
metabolic changes in response to smoking cessation among humans have produced
inconclusive and equivocal results.

Eight studies have reported either acute changes in REE following smoking or
nicotine administration or have reported decreases in REE after smoking cessation. An
early study (Glauser et al. 1970) reported decreases in oxygen consumption for seven
male subjects who quit smoking for I month. Food intake and physical activity were
not monitored. Reanalysis of these data (Klesges et al. 19X9) revealed that the changes
in metabolic rate reported by Glauser and cowtorkers ( 1970) waere significant only with
improper methods of statistical analysis. In the only study that utilized an indirect
calorimetry respiration chamber. Hofstetter and coworkers ( 1986) reported a I O-percent
difference in total energy expenditure during a 24-hour period of smoking compared
with a 24-hour period of abstinence among eight smokers. However. this difference in
energy expenditure disappeared after 24 hours. No changes were observed in mean
basal (sleeping) metabolic rate. Diet was held constant.

Perkins and colleagues have conducted a series of studies evaluating the effects of
nicotine, in the form of nicotine nasal spray, on changes in REE. In a study of nicotine
administration in I8 male smokers, Perkins and colleagues (Perkins. Epstein, Stiller.
Marks et al. 1989) reported REE changes that were 6 percent above baseline after
nicotine administration, which was significantly greater than the 3-percent increase
after placebo administration. Another investigation (Perkins et al. 1989a) sought to
determine if nicotine-induced increases in metabolic rate observed at rest were also
present during physical activity. Ten male smokers were administered nicotine and
were then compared with IO male smokers who were administered placebo. Metabolic
rates increased both at rest and during light exercise. Although the percent change in
REE due to nicotine was equivalent both at rest and during activity. the excess energy
expenditure (in kilocalories) attributable to nicotine was more than twice as great during
exercise. A third study using nicotine nasal spray assessed the combined effects of
nicotine and consumption of a meal on REE (Perkins. Epstein. Stiller. Sexton et al.

4x7


1989). Eight male smokers were assessed using a repeated measures design. These
individuals were given a caloric load (vs. water) and nicotine (vs. placebo). Both the
caloric load and nicotine increased REE significantly. However, no interaction be-
tween these factors emerged, and the effects were slightly less than additive when
combined. Nicotine alone increased REE by 4.95 kcal per hour. food alone increased
metabolic rate by 14.30 kcal per hour, but nicotine plus food increased metabolic rate
by 17.00 kcal per hour. Finally. in a study of the effects of changes in energy balance
as a function of smoking cessation, Perkins, Epstein, and Pastor (1990) evaluated REE
in seven female smokers across 3 weeks: normal smoking (week I), smoking cessation
(week 2), and resumption of smoking (week 3). REE did not drop during the week of
abstinence compared with baseline. However. a significant increase in REE was
observed when subjects resumed smoking compared with the week that they were
abstinent.
The effects of smoking and coffee consumption on REE were recently evaluated by
Klesges. Brown. and colleagues (1990). Of 45 regular cigarette smokers and coffee
drinkers, IS were randomly assigned to smoke 2 cigarettes, I5 were assigned to drink
two standardized cups of coffee, and I5 were assigned to smoke cigarettes and drink
coffee. All three groups had acute increases in REE with a similar pattern of response
in each group.
In the largest study to date of all-day changes in metabolic rate, Klesges, Coday, and
coworkers (1990) evaluated changes in REE among 39 individuals over a IO-hour
period using multiple assessments of REE. Of the 30 smokers. 20 were assigned
randomly to continuous, regular smoking and IO were assigned to a no-smoking group.
A nonsmoking control group of nine subjects was also evaluated over the same time
period. The increase in REE among nonsmokers was not significant. In marked
contrast, smokers who did not smoke decreased REE over the course of the day.
Additionally, there were two distinct patterns of results among smokers who smoked
over time. Of the 20 smokers. I3 (70 percent) markedly increased their REE over time,
but 6 smokers (30 percent) decreased REE over time (similar to the pattern of smokers
who did not smoke). Closer inspection of the minute-by-minute metabolic changes of
those subjects who increased metabolic rate indicated an acute metabolic increase
followed by a return to baseline early in the day. or an acute metabolic increase followed
by a reduction. but to a level higher than baseline later in the day. In contrast, subjects
who had a mean decrease in REE also had an acute metabolic increase followed by a
drop below baseline early in the day, or an acute metabolic increase followed by a return
to baseline later in the day. Subjects who responded with decreases in REE smoked
more (as measured by expired CO) than those who responded with a cumulative
increase in energy expenditure. These results are consistent with recent observations
of a U-shaped relationship between daily cigarette consumption and body weight. with
moderate smokers weighing less than nonsmokers but heavy smokers approximating
the body weights of nonsmohcr\ (Albanes et al. 1987).
Four studies found no relationship between smoking and metabolic rate. Burse and
coworhers ( 1982) did not observe chronic changes in resting metabolism in a sample
of three makers who quit for 3 vveeks.  However, the small sample size in this
investigation limits interpretation of the results. Although Robinson and York (1986)


reported an elevated metabolic response to food intake (i.e.. thermic effect of food).
chronic REE did not change as a function of smoking and total energy expenditure after
a meal during the cessation period. Stamford and colleagues (1986) did not find
changes in oxygen consumption in 13 subjects who quit smoking for 38 days. These
investigators did find marked food intake changes that accounted for 69 percent of the
variance of postcessation weight gain. In a study of the chronic effects of smoking
status on REE, Perkins and coworkers ( 1989b) assessed 20 male smoker, and IO male
nonsmokers after overnight abstinence from food and caffeine in both groups and after
overnight abstinence from smoking in the smohing group. No differences in REE were
observed.
Two recent studies evaluating the acute effects of cigarette smoking on REE have
provided equivocal findings. In a sample of five occasional and five regular smoker4
(Warwick. Chapple. Thomson 1987). REE did not increase after smoking. even during
the first IS to 30 minutes after smoking. Additionally. the thermic effect of food was
slightly, but not significantly. lower with smoking than without smoking. Dallosso and
James ( 1984) evaluated short- and long-term metabolic changes associated with smok-
ing. The thermogenic (metabolic) response for I hour after smoking 1 cigarette wa\
not significant, although an acute increase w/as observed during the first 30 minutes.
However, variability of responses was marked, ranging from a 4.5percent decrease in
metabolic rate to a 9.0-percent increase. No consistent long-term changes in metabolic
rate were observed. Rather, the metabolic rate of four smokers clearly decreased after
cessation; the rate stayed the same in two smokers and increased in two others.
The literature generally indicates that both dietary and metabolic changes are respon-
sible for weight gain after smoking cessation, but these changes probably occur through
complex mechanisms. Physical activity does not appear to be related to postcessation
weight gain. Although the pattern generally indicates that both dietary and metabolic
factors are involved. there is inconsistency both within and between studies indicating
tremendous individual differences in subjects' dietary and metabolic changes after
smoking cessation.
Investigators need to try to determine carefully the potential moderator variables of
dietary and metabolic changes after smoking cessation. Factors such as gender, age,
race, weight history, and concerns about postcessation weight gain may all play a role
in predicting dietary changes after cessation. Some individuals, for example, may
respond to smoking cessation by dramatically increasing their dietary intake (Duffy and
Hall l988), whereas others may impose dietary restrictions in an attempt to avoid
postcessation weight gain (Klesges et al.. in press).
There also appears to be tremendous individual variation in the metabolic response
to smoking and smoking cessation. Overall, evaluations of short-term, acute responses
to smoking generally report increases in metabolic rate as a function of nicotine
administration and smoking (Hofstetter et al. 1986; Perkins et al. 1989a; Klesges.
Brown et al. 1990), although long-term (overnight or longer) studies generally do not
indicate changes in metabolic rate as a function of smoking cessation (Stamford et al.
1986). However, some investigators have reported that the acute effects of smoking
have not produced a change in REE (Warwick, Chapple. Thomson 1987).


Research needs to focus on a number of potential moderators of smoking and
metabolic rate. Levels of plasma nicotine vary greatly even for the same level of
cigarette consumption and for the same nicotine content of cigarettes (US DHHS
I98Xa). The relationship between nicotine, as well as other constituents of tobacco
smoke. and metabolic rate needs to be evaluated carefully. It is also possible that
heavier, chronic smokers may habituate to the effects of nicotine over time (US DHHS
1988a) and their metabolic responses may become blunted (Klesges, Coday et al. 1990).
Other important moderators. such as years smoked. gender. and relative weight. should
also be carefully evaluated in future investigations.

RELATIONSHIP BETWEEN OVERWEIGHT AND ADVERSE MEDICAL
           AND PSYCHOSOCIAL OUTCOMES

Obesity refers to excess body fat. whereas overweight refers to excess body weight
relative to height compared with gender-specific norms (Powers 1980). Obesity and
overweight are highly correlated across the population. although some individuals are
overweight but not obese (e.g., bodybuilders). and others are obese but not overweight
(e.g., a normal weight "couch potato") (Grunberg 1982b). In the context of this
Chapter. the relevant data are those that are related to health risks. The most commonly
used methods to measure or estimate body fat in studies of health consequences of body
size are measures of height and weight in comparison with gender-specific norms
(which actually determine overweight) and measurement of subcutaneous fat by
skinfold thickness at one or more sites (which determines obesity). Therefore. the data
cited in this Chapter are sometimes based on estimates of obesity and sometimes based
on estimates of overweight; both terms appear in the text. Normative values for these
anthropometric measures have generally been derived in one of two ways: either by
averaging the values found in populations of healthy persons or by tabulating values
reported to be associated with greatest longevity in population-based studies. Inclusion
of data based on these various standard measures provides the most complete informa-
tion available. Although the volume of research related to obesity and health risk
precludes comprehensive review here. a summary of this literature is a useful starting
point for examining the health risks of weight gain following smoking cessation.

Large amount\ of epidemiologic and clinical data clearly indicate a positive associa-
tion between excess body weight and medical ri5k. Cross-sectional. longitudinal.
ecologic. and case<ontrol studies indicate that there is a graded relationship between
weight and various diseases and disease risk factors. Positive associations have been
reported between body weight and glucose intolerance and type II diabetes (Kannell.
Gordon. Castelli 1979: Rimm et al. 1972: West and Kalblleisch I Y7 I: Negri et al. 1988:
Hadden and Harris 1987): elevated blood pressure and hypertension (MacMahon et al.
1987: Chiang. Perlman. Epstein lY6Y: MacMahon et al. lY83: Blackburn and Prineas
1983: Pan et al. 1986): elevated total blood cholesterol and lowered high-density
lipoprotein cholesterol (HDL-C) (Jooste et al. 1988: Garrison et al. 1980: Nanas et al.
IYX7): gout (Larsson. Bjorntorp. Tibblin IYXI ): kidney stones (Larc\on. Bjorntorp.
Tibblin 1981 ): gall bladder disease (Rimm et al. 1972): cardiovascular disease (CVD)
(Rabkin. Mathewson. Hsu 1977: Noppa et al. 1980: Garrison and Castelli 1985):

490


cancers of the endometrium and colon (Garfinkel 1985: Graham et al. 1988: Verreault
et al. 1989): arthritis (Anderson and Felson 1988: Felson 1988): and varicose veins.
Obese women are more likely than lean women to experience menstrual abnormalities
(Hartz et al. 1979) and complications in pregnancy (Abrams and Parker 1988). Obese
individuals require more medical care (Tsai. Lucas, Bernacki 1988). experience more
complications during and following surgical procedures (Schwartz 1955). and report
greater limitations in performing tasks ofeveryday living (Stewart. Brook. Kane 1980).

The strength and consistency of the data and the understanding of causal mechanisms
underlying obesitydisease associations vary from end-point to end-point. Neverthe-
less, there is little doubt that obesity represents an important health risk that may reduce
both the quality and duration of life. The overall evidence linking overweight todisease
has led to recommendations from numerous health organizations for individuals in the
general population to control their weight as a means of preventing future illness
(National Institutes of Health Consensus Development Conference Statement lY8S;
Subcommittee on Nonpharmacological Therapy of the I984 Joint National Committee
on Detection. Evaluation. and Treatment of High Blood Pressure 1986: US DHHS
1988b).

Despite convincing data linking obesity to ill health. several issues in the area remain
controversial. A key issue that is particularly germane to smoking cessation-induced
weight gain is the extent to which modest degrees of overweight represent a health
hazard. The most commonly recognized standards for acceptable body weights are
those developed by the life insurance industry based on followup studies of policy
holders conducted in 1959 and 1979 (Metropolitan Life Insurance Company 1960:
Society of Actuaries and Association of Life Insurance Medical Directors of America
1980).

Each of these studies evaluated the mortality of approximately 4,000,000 life in-
surance policy holders. "Ideal" weight standards that were developed from these
studies and widely used in subsequent research represent the gender- and height-
specific weights associated with lowest mortality. Overall. a J-shaped relationship is
observed between weight and mortality. Lowest premature mortality is associated with
body weights that are about IO percent below the population average. Excess premature
mortality is associated with extremely low weights (i.e., body weights more than IO
percent below the standards), and premature mortality increases incrementally for
increasing weights above the standard. In the range of weights that encompasses the
vast majority of the population (i.e., relative weights of I.0 to 1.3). the relationship
between weight and mortality was approximately linear with each l-percent increase
in weight associated with about a l-percent increase in premature mortality. Above
relative weights of about 1.3, the curve rises even more steeply so that premature
mortality may double at relative weights of I .5 or more (Manson et al. 1987).

The overall relationship between weight and mortality has been confirmed in several
other large scale prospective studies.  For example. the American Cancer Society
followup study of 750,000 men and women from the general U.S. population provides
confirmatory data with specific detail on various causes of death (Lew and Garfinkel
1979). Table 3 presents mortality ratios for this study group by weight status for
selected causes. Table 4 presents mortality ratios by weight and smoking status. Most

491


of the deaths associated with leanness occur among smokers, and although the shape
of the weight-mortality curve\ are similar among smokers and never smokers. smokers
have nearly twice the mortality rate compared with never smokers over much of the
weight distribution. A recent IO-year followup study of 1.700.000 Norwegians con-
firms these findings in a non-U.S. population with regard to the shape of the weight
mortality association and the causes of death at both ends of the distribution (Waaler
1988).

The reported relationship with age further complicates the relationship between body
weight and health (Andres et al. 1985). For example. the strongest relationship between
body weight and premature mortality holds for younger age groups (i.e.. under 40 years
of age). In older adults. the relationship between weight and mortality is weak over
much of the weight distribution. and in the oldest groups studied (i.e.. over 60 years of
age), mortality appears inversely related to weight. Indeed. many prospective studies
of middle-aged adults have observed little or no prognostic significance of body weight
for either total premature mortality or major disease endpoints except at the extremes
of the body weight distribution. These findings have led some researchers to argue that
concerns about weight and overall health for most individuals have been exaggerated
(Keys 1981; Barrett-Connor 1985). In contrast, other investigators have noted that
cigarette smoking has not been statistically controlled in many of these analyses, and
in addition, pathophysiologic effects of obesity. such as hypertension and hyper-
glycemia. have been inappropriately adjusted (Manson et al. 1987). Therefore. the
health risks of obesity may have been underestimated.

Another issue to consider in the relationship between body weight and health is that
all forms of overweight may not pose the same health risks. In particular. health risk
may depend on weight status at different times in an individual's life. A study by
Abraham, Collins, and Nordsieck ( I97 I ). for example. studied 1.087 white males for
whom height and weight data were available at ages 9 to I3 and after a period of
approximately 30 years. By cross-classifying respondents by childhood and adult
weight status, these researchers found that individuals who were at the low end of the
weight distribution as children. but who gained weight to reach the high end of the
weight distribution as adults, were at significantly higher risk of hypertensive vascular
disease and cardiovascular renal disease than were individuals who had high weights
both as children and as adults. Similarly, in a report based on the Normative Aging
Study. Borkan and colleagues ( IYXh) found age by weight gain interactions. relating
weight gain to health risk. Weight gain had a stronger positive association with change
in fasting glucose levels for older men compared with younger men: however. weight
gain was more strongly related to change in uric acid (positive) and forced vital capacity
(negative) in younger men (Borkan et al. 1986).

The importance oftiming issue\ in the relationship between body weight and disease
is also apparent in weight cycling. Weight cycling refers to gaining and losing weight
repeatedly over time. Such weight fluctuations might occur in individuals who
repeatedly diet but are unable to maintain weight losses. Weight cycling might be
caused by recurrent illnesses or major fluctuations in lifestyle. Such fluctuations might
conceivably also occur among smokers who quit but relapse to smoking on multiple
occasions. Several recent reports suggest that weight cycling may be associated with

497


TABLE 3.-Mortality ratios for all ages combined in relation to the death rate of those 90-109% of average weight


                                          Weight index"
CXM uf        7th rev
  death         ICD       Gendrr        <x0        X&X9      YO- IO')       I IO-I I')      120-12')     I.%-ii')     >I40

Total dtxiths                MA          I .2s        I .os        I .oo         I.15         I 27        I .4b       I .x7
                       F~lll~k        1.1')       0.Yc-J        I .oo         1.17         I .2Y        I .Jh       I .XY

CIiD          420         Male         o.xx       O.YO        I .oo         I .73         1.32        I .ss       I .YS
                       Femalr        I .o I       0.x9        I .oo         I .2i         LAY        I.54       1.07

CXKW.         I4tk20.5      Mdr         I .33        I.13        I .oo         I .07         I .OY        I.13       I ..11
011 \ite\                    Frm:ilr        0.96       0.02        I .oo         I.10         I I 0        I .23       I.55


Diabetes        260         Male         o.xx       0.x4        I .oo         I .hS         2.X-l        3,s I       S.IY
                       Female        O.hS       O.hl        I .oo         I .Y2         3.3-l        2.7x       7.00
Dlgative        S4ObS42      Male          I.3Y        1.2x        I .oo         I .-Is         .XX        7.X')       3.w
tliw~w~         570-57x


             5X4-5X6      Femalr        1.5X       O.Y2        I .lK)         I .hh          h I        7.1s       7.3

C`errhrovawulnr    330-3.34      Mb          I.21        I .OY        I .oo         I.15         .I7        I .SJ       1.17
diwaw                  F~lll~k        1.33       O.YX        I .oo         I .OY         I.10        1.40       I.52


Nol`E. (`llD=co~ w.q he;w d,\~il\~!
"C`;il~ul.~trd hy ~hvdin~ a t"r\,m`\ .~ctuid wc,gt,t hy the ~orrc\,,w,d,n~ ;,vcr.,g' welghl t,,r Ihc .~,,,mqm.w wx-IIICII 01 thaghl-5.!r ;I~C ~"`L,,`. n~d~,t~t,cd II> 11Kl
SOI XC'E: 1.w id (;artinkl ( lY7Y I.


TABLE 4.-Mortality ratios for all ages combined according to smoking status in relation to those 90-1099'~ of average age

Weight index"

Never vwkd       o.xx
20 cig/dny        I .6X
Other            I.77
Ncvcr \n~kxi       1.10
x0 q/day        I .YH
Other            I..53

0.75
I .40
1.01
o.xx
1.5')
I.13

0.75
I.34
O.Y3
O.Y3
1.64
I.12

0.98
I .76
1.1s
I .20
2.22
I .42

1.16       I .6Y
2.00      2.21
I.29       I .66
1.37       I .74
2.30       2.73
I .62       2.04

Coronq
artery dwax
(ICD 420)

Never wwked
t20 clg/day
Other
Never smohed
X0 cig/duy
Other

0.72
I .Oh
0.`) t
O.Y3
t 3 I
1.54

0.66        0.76
I.13        1.33
0.90       0.93
ox        0.92
I .70        2.12
I.14        I.IX

0.96
t .66
1.12
I.10
2.20
I .xx

I .04
I.81
1.1')
I .2Y
3.4x
I .44

1.24
2.1 I
t 37
t .3Y
3.79
2.01

1.73
2.1 I
1.x4
I.86
4.74
2 .3 3

Cancer,
at1 slteh
( ICD I-K-205)

Malt'




Fcmule

Never mohrd
220 cydday
Other
Never welled
220 cigiday
Other

0.60
2.07
I.20
0.x.5
t .4Y
I.1 I

0.60
I.71
I .03
0.x5
t .x1
O.YX

0.66
I .43
I .YO
O.Yh
I.34
I .03

0.69
I .46
0.89
t .Oh
t .x1
I .Oh

0.7Y
t 5s
t .05
I.th
I.34
I.16

0.90
I.71
0.87
1.1')
I .70
I.1 I

0.76
2.00
I .22
t .so
I .4Y
I.60


elevated premature mortality compared with maintaining a more stable weight ov'er
time. In a study by Hamm. Shekelle. and Stamler ( IYXY). for example. CVD and cancer
mortality and total mortality were compared among individuals who reported either
having gained significant weight (N=l33). having remained at the same weight
(N=l7X), or both having gained and lost significant weight (N=YX). Both gainers and
cyclers had significantly elevated total mortality experience. relative risks of 1 .S and
I .4. respectively. compared with individuals whose weights remained constant. Three
recently published abstracts (Lissner et al. l9XY: Lissner. Collins et al. IYXX: Lissner.
Odell et al. 19X8) have reported even Freater health risks of weight cycling. Using
prospective data from the Multiple Risk Factor Intervention Trial (MRFIT) (Lissner.
Collins et al. IYXX). two prospective studies from Goteborg. Sw,eden (Lissner et al.
1987). and the Framingham Study (Lissner. Odell et al. 19Xx). weight cycling vvas
defined as the variability of vveights recorded at repeat examinations. Controlling for
a variety of possible confounding variables, weight cycling was independently predic-
tive of total premature mortality and CVD mortality. In the analyses based on MRFIT.
premature mortality among men with the most variable weights was 36 to X9 percent
higher than among men with the most stable weights.
An additional issue to consider in the relationship between body weight and health
is the distribution of body fat. Individuals differ in the location of stored adipose tissue.
Research data show that individuals who store greater amounts of body fnt in the
abdominal region rather than in the hips or limbs have elevated cardiovascular risk
factors (Gillum 19X7; Selby. Friedman. Quesenberry l9XY ). CVD. and diabetes rates
(Freedman and Rimm 19X9: Lapidus and Bengtsson 198X) as well as reproductive
system cancers among women (Bjomtorp 198X ).
Usually measured by the ratio of abdominal circumference to hip circumference or
the ratio of trunk versus peripheral skinfolds. a central body fat distribution is positively
correlated with absolute body weight. However. in several studies, the centrality of fat
distribution has proven to be a much stronger predictor of disease than body weight. A
landmark study in this area was conducted by Larsson and colleagues (1984) who
reported on I3 years of followup for 792 Swedish men aged 54 years at the time of first
observation. Outcome measures were stroke. ischemic heart disease. and all-cause
mortality. None of these health outcomes was significantly related to measures of
adiposity (body mass index weight/height', the sum of several skinfold measurements.
and body circumferences). However. the ratio of waist to hip circumference (WHR)
was significantly and positively related to all three measures of illness and death. The
relevance of this finding for ex-smokers, as discussed below, is that smoking is
positively related to WHR and that smoking cessation is associated with a reduced
WHR (Shimokata. Muller. Andres 1989).
Compared with pathophysiologic health risks, social and psychological pathologies
associated with overweight are not as well established. This situation may reflect the
relative absence of research in this area, but it may also indicate the absence of a strong
relationship. Obesity is strongly disapproved of and discriminated against in this
society (Allon 1973: Grunberg 19X2b; Wadden and Stunkard 19X5). Overweight
individuals are falsely stereotyped as having a variety of undesirable characteristics.
including self-indulgence, laziness. lack of self-control, and lack of intelligence.


The perception in this culture of obesity as unattractive has been documented in
various populations. For example Richardson (1971). in a study of IO- and I l-year-
olds' perception of the likableness of children with a variety of handicaps. found that
obese children were judged less attractive than were children with amputations and
facial disfigurement or children confined to wheelchairs. Similar biased impressions
have been documented among adults and among physicians and medical students
(Allon 1973; Maddox and Liederman 1969). Canning and Mayer ( 1966) found that the
prevalence of obese students in college was less than the prevalence of obese students
in high school despite no difference in academic performance in high school or in
college application rates. A survey of employers indicates that many profess not to hire
obese individuals (Roe and Eickwort 1976). and at least one survey of business
executives suggests an inverse association between obesity and salary (Indrrst~:\ We&
1974). In a survey of college students, Kallen and Doughty ( 19X4) found lower rates
of reported dating in overweight subjects. although no less satisfaction with intimate
relationships.
Although it is obvious that many overweight individuals are dissatisfied with their
personal appearance. desire to lose weight. and frequently make efforts to lose weight
(Wadden et al. 19X9: Polivy, Gamer. Garfinkel 1986: Adams 19X0: Guggenheim.
Poznanski. Kaufmann 1973: Dwyer. Feldman, Mayer 1975: Dwyer and Mayer 1970:
Stewart and Brook 1983; Jeffery et al. 19X4), evidence for severe psychological or
social impairment in all but the most severe cases of obesity is generally lacking.
Moore. Stunkard. and Srole ( 1962), reporting data from the Midtown Manhattan Study.
found higher scores on three measures of psychological disability in the obese compared
with the nonobese.
Data from the Rand Health Study and a Dutch population-based study indicated that
obese individuals report that their weight imposes some restrictions on their everyday
activities and causes them more pain and worry compared with the nonobese (Stew*art.
Brook, Kane 19X0: Stewart and Brook 19X.1: Seidell et al. 19X6). However. Stewart
and Brook ( 19X3) also reported that obese persons are less depressed than normal-
woeight persons. a finding corroborated in a study of British citizens by Crisp and
McGuiness ( 1976). These mixed and inconsistent findings from studies of obese adults
also have characterized studies of obese children (Wadden et al. 1989: Wadden et al.
19X4). In extremely obese individuals presenting themselves for treatment (i.e.. those
75 percent or more overweight). higher levels of psychological disturbance have been
reported (Halmi et al. 1980: Atkinson and Ringuette 1967). Even here. it has been
questioned whether such pathology is greater than that oh\erved in normal-weight
individuals pre\enting for medical or surgical procedures (Wise and Fernandez 1979:
Swenson. Pearson. Osborne 1973). It has been su,,
                           CToested that unwarranted concerns
about vveight gain mav contribute to eating disorders such as anorexia and bulimia
(Wooley and Wooley 19X-l). Data supportin, (7 this idea. however. are largely anecdotal
(Wadden and Stunkard 19X5 ).
Prospective studies on the effects of weight gain on psychosocial functioning have
not yet been reported. Studies of psychological changes accompanying weight loss
generally show positive effects. even when weight loss is modest and not well main-
tained (Wing et al. 19X-t). Therefore. consistent with intuition. many people feel better


about themselves when they lose weight. However. the extrapolation of these findings
to weight gain lacks empirical support.

In summary. although adverse psychological and social consequences of overweight
have been much discussed in both lay and professional circles. \uch effect\ have not
been well documented. Moreover. to the extent that associations have been reported.
the direction of causation is unclear. More research in this area is warranted. particu-
larly because the available research is not extensive and much of it is methodologically
weak. At this time. data suggest that only the most extreme forms of obesity, the upper
I or 2 percent of the weight distribution in this domain. pose significant hazards.
However. it is important to emphasize that these conclusions reflect the lack of evidence
for serious psychosocial problems resulting from modest weight pains. Nevertheless.
many persons want to lose weight. many persons seek ways to lose height. and many
persons feel better about themselves when they lose weight.

CHANGE IIV WEIGHT-RELATED HEALTH RISKS
     AFTER SMOKING CESSATION

As documented earlier in this Chapter. smoking cessation is associated with weight
gain. An important question is the extent to which this weight gain might lead to
elevations in blood pressure. cholesterol. glucose intolerance, or other factors that
would offset the benefits of smoking cessation discussed in detail throughout this
Report.

Relatively few studies have specifically examined the effect of smoking cessation on
weight-related health risks. Seven studies were reviewed for this Report. Gordon and
coworkers ( 1975) reported changes over an I S-year period in weight and related risk
characteristics among individuals in the Framingham Study. At entry into the study,
61 percent of men and 40 percent of women smoked cigarettes; at the I S-year followup.
37 percent of men and 3 I percent of women continued to smoke. Analyses of changes
were restricted to men because of the small numbers of women who quit smoking in
this sample. Male quitters were similar to those who continued to smoke in baseline
characteristics except that the former group contained more diabetics. The authors
interpret this finding as suggesting that ill health is an incentive to stop smoking.

Short-term effects of smoking cessation, defined as the change between the last
examination at which smoking was reported and the first examination at which
nonsmoking was reported (I-year intervals). included a weight gain of 3.8 pounds. an
increase in systolic blood pressure of I .6 mm Hg. and an increase in serum cholesterol
of0.2 mg/dL. Continuing smokers had an average weight gain of 0.3 pound. increased
systolic blood pressure of 0.7 mm Hp. and decreased serum cholesterol of 0.2 mgjdl.
For the same time period, nonsmokers had an average weight gain of 0.5 pound.
increased systolic blood pressure of 0.7 mm Hg, and increased serum cholesterol of 0.3
mg/dL. Differences among groups in blood pressure and cholesterol changes were not
statistically significant. Long-term changes associated with smohing cessation were
evaluated by comparing changes between the fourth and the tenth examination. a period
of 11 years. among continuing smokers. nonsmokers. and individuals smoking at entry
but not smoking from the fourth to the tenth examination. Trends in weight. blood


pressure, serum cholesterol. and blood glucose did not differ significantly among these
three groups.
Schoenenberger (19x2) reported the relationship between smoking cessation and
changes in body weight. blood pressure, and serum cholesterol over 3 years among men
in the special intervention group in MRFIT. All men in the study were at high risk for
heart disease and were being counseled throughout the study in smoking cessation and
dietary changes to effect cholesterol reduction. When necessary. the men were also
treated pharmacologically for elevated blood pressure. Results indicated significantly
less weight loss in quitters (-0.6 pounds. i.e.. a gain of 0.6 pounds) compared with
nonsmokers and continuing smokers (5.7 and 3.6 pounds, respectively), no differences
in blood pressure change (-9.6, -8.7. and -9.4 mm Hg, respectively, for systolic blood
pressure among men not on medication). and greater reductions in serum cholesterol
among quitters (-I 3.4 mg/dL) than in the other two groups (-10.0 and -8. I mg/dL).
The latter effect was interpreted as possibly reflecting a higher level of generalized
motivation to reduce risk in the quitting group.
In a S-year followup study of 2.383 persons with mild hypertension in eastern Finland,
Tuomilehto and colleagues (1986) found that 26 percent of men and 35 percent of
women who smoked at the time of the initial examination had quit. Among men,
smoking cessation was associated with a 7.9-pound weight gain compared with 0.2-
pound and 2.2~pound weight gains among nonsmokers and continuing smokers. respec-
tively. Among women. weight loss after smoking cessation averaged 0.7 pound
compared with gains of 0. I pound and 2.2 pounds among nonsmokers and continuing
smokers, respectively. Smoking cessation was not associated with a significant in-
crease in blood pressure or serum cholesterol compared with continuing smokers or
nonsmokers. Mean arterial pressure fell by 5.0 and 13.1 mm Hg in male and female
quitters. respectively. compared with decreases of 6.9 and 8.7 mm Hg among non-
smokers and of 7.0 and 9.6 mm Hg among continuing smokers. Serum cholesterol fell
between 0.63 and 0.66 mmol/L across the various subgroups.
Two papers relating smoking cessation to weight-related risks have been published
based on data from the Normative Aging Study. The first report examined change over
5 years among 2 14 continuing smokers and 103 quitters (Garvey. Bosse, Seltzer 1973).
An average weight gain of 4.2 pounds. which was accompanied by a 3.6 mm Hg
increase in diastolic blood pressure. was observed among quitters compared with
continuing smokers. The second report examined the relationship between smoking
and body fat distribution. both cross-sectionally and longitudinally between examina-
tion visits scheduled 7 years apart (Shimokata, Muller. Andres 1989). Central body fat
distribution. which poses increased health risks. as assessed by WHR was positively
associated u ith making. Moreover. among smokers. daily cigarette consumption was
positively associated with central adiposity. Smoking cessation was associated with
increased body weight. However. despite the weight gain. the change in WHR among
ex-smohers was small and. in hct, decreased slightly because hip circumference
increased. Therefore. based on WHR data only. smoking rather than smoking cessation
may pose a wjeight-related health risk.
Stamford and coworkers ( 19X6) studied the short-tern1 effects of smoking cessation
on lipoprotein fractions. Amon I3 women who successfully quit smoking for a period


of 48 days, these investigators observed a weight increase of 4.9 pounds. This weight
change was accompanied by a nonsignificant increase in total cholesterol of 9 mg/dL
and a significant increase in HDL-C of 7 mg/dL. Over the subsequent year. these
favorable HDL-C changes were maintained in three individuals continuing to abstain
from smoking, but were lost in nine individuals who returned to smoking.
One randomized trial of smoking cessation and weight-related health risks was
located for this review. Rabkin (1984a) randomized I07 smokers to smoking cessation
and 33 to continued smoking in a comparative study of smoking cessation strategies.
A battery of physiologic measures was obtained at baseline and repeated 2 to 3 months
following randomization. No differences were found in cessation rates among the
different quitting strategies. Physiologic changes observed in the smoking cessation
group as a whole (i.e., all those randomized) included a significant increase in weight
(I .8 pounds) and skinfold thickness (6.6 mm) compared with the control group (0.4
pound and -7.0 mm). but no significant change in lipid profiles, fasting glucose, or
blood pressure. Only 35 subjects in the cessation groups were successful in quitting
smoking. Successful quitters gained significant amounts of weight compared with
individuals who did not quit (4.4 vs. 0.7 pounds, respectively). Successful quitters also
experienced significant increases in HDL-C compared with nonquitters (4.2 vs. 0.1
mg/dL). Changes in other weight-related risk factors did not differ among groups.
The studies reviewed above are consistent in their findings. Individuals who quit
smoking andgain weight appearto experience relatively small changes in health-related
risk factors such as blood pressure. serum cholesterol, and blood glucose. Moreover.
some of the potentially adverse effects of weight gain on health risks are mitigated by
changes in lipid profiles and in body fat distribution in a direction predictive of
improved health outcomes. It seems likely that only those smokers who have large
weight gains after smoking cessation would experience important changes in weight-
related risk factors.
The characteristics of individuals most likely to gain harmfully large amounts of
weight after smoking cessation merit additional investigation. Bosse, Garvey. and
Costa (1980) have reported relevant findings from the Normative Aging Study. Over
a S-year period these investigators found that factors most predictive of weight gain
among recent quitters were younger age, leanness of body build. and greater amounts
of smoking. The latter finding is confirmed by other studies (Blitzer. Rimm. Giefer
1977; Gordon et al. 1975). There are no data available on specific predictors of
excessive weight gain among ex-smokers.  Research on predictors of weight gain
suggest that those persons most likely to gain weight after smoking cessation may be
those who can best afford it because they are relatively lean. They also may be those
who need smoking cessation most because they smoke the most.
Quantitatively estimating the extent of health risk associated with weight gain after
smoking cessation is a complex process. The health risks of obesity vary with age, the
temporal patterning of weight changes. type of obesity, and other risk factors.
Moreover, smoking cessation itself appears to have independent effects on some
weight-related risk factors that may actually be beneficial.
It has been estimated that the health risks posed by regular smoking double overall
mortality rates compared with never smoking (US DHHS 1989). Moreover. as detailed


elsewhere in this Report. there are clear health benefits associated with smoking
cessation. The amount of excess body weight that would have to occur to offset the
benefits of smoking cessation would have to be considerable. Yet. average weight gains
after smoking cessation are only about 5 pounds. bringing most individuals to a weight
level similar to that of their nonsmoking peers. As discussed in this Chapter. the
proportion of ex-smokers who are likely to gain large amounts of weight (e.g.. more
than 20 pounds) is small. Therefore. although some individuals may experience these
large weight gains. the number of individuals likely to gain enough weight to offset the
benefits of smoking cessation is negligible. Also, the likelihood of adverse psychoso-
cial consequences because of small weight gain seems remote for most people.
Although further research in this area is w)arranted. there is little reason to expect weight
gain to pose a substantive medical or psychosocial hazard to the vast majority of
smokers who are quitting. For those persons wsho do pain excessive amounts of weight
after smoking cessation. the health benefits of cessation still exist. and weight control
programs rather than smoking relapse should be implemented. In conclusion, the clear
reduction in health risks that results from smoking cessation overshadows any health
risks that may result from smoking cessation-induced body weight gain.

STRATEGIES TO CONTROL POSTCESSATION WEIGHT CAIN

Because weight gain after smoking cessation commonly occurs and because many
people, particularly young women, report smoking to control weight gain (Klesges and
Klesges I9XX: US DHHS 1990). strategies that successfully moderate postcessation
weight gain ma) encourage weight-conscious smohers to attempt cessation and ma)
facilitate the efforts of successful quitters to remain abstinent. Only a few controlled
investigations have examined interventions for reducing wjeight gain after smoking
cessation. Currcntl! existing behavioral and pharmacologic intervention\ are sum
marized belo\

Behavioral Methods for Reducing Postcessation Weight Gain

Smoking cessation programs that include a M eight control component have not
successfull! increased \mohing cessation. In one study. 79 women Here randomly
assigned to a 7-h eek smohing cc\sation program either u ith or u ithout @eight control
information (Mermrlsrein 19x7 ). At po\ttreatmt'nt and at thllowup. there Mere no
significant differences in \mohing ct`\\ation rate\ bet\rren the tuo groups. Participant\
in both groups gained Meight during treatment: ho\\tc\,er. the weight increase for the
smohing-ce\~;ltic,n-plLi~-~\ei~ht-colitr~)l - c'roup ~3s \ignit'icantl) le\s than the increase
for the \rnokin~-cc~~ati(~~i-~)~ll~ group ( I .f \ s. 2.4 pound\).

Several \reight control \tratcgie\. as adjunct\ to smohing cessation. were evaluated
b> Grinstead ( 19X I ). Fort>-fi\,e \ub.jects were randomly a\\ifned to ;I I-weeh smohing
a\,er\ion pr+ram LI ith one of three height control intt'r\,cntions. No difference\ in
smohing cc\sation rate\ wcrc observed. and there were no \\cight change differences
among the group\. Suh,ject\ in all groups gained iheight during treatment.

500