Tobacco Documents Online

616 shown that the perinatal mortality is not significantly higher in the better-e8ucated U.S. whites, or in the upper social classes in Britain (social classes 1 an3 2). On the other hand, smoking is associated with higher perinatal mortality in only certain poorer groups such as the less educ3ted whites and blacks in the TJ.S.A. , an3 the manual and unskilled workers in Britain. This lack of consistency of the so-called "effect" of smoking i s agains t the causa l hypothesis. Fetal growth retardation is not specific to the babies of mothers who smoke in pregnancy. Other associations such as maternal stature, maternal nutrition and socio-economic status are already shown to be of causal significance. The strength of the reported statistical association bet- ween maternal smoking and perinatal mortality is much Less than that of other factors such as previous history of perinatal loss or socio-economic status. For example social class 5 (unskilled workers) in Britain shows an excess of perinatal mortality o f 400% ove r that o f social clas s 1. "Ihi s exces s i s more than ten times that which is associated with smoking (about 358)• Such a comparatively weak statistical association is not in favour of the caus3l hypothesis. The claimed dose-response relationship is often quoted as evidence f or the c obviously important different doses and ficult except by st~ co-variance. Most response relationshi other factors. How have done this and smoking, with publ: loss showing mucY mortality. In a s 50,000) by Naeye (2: affect birthweight M the mean birthweight -smokers at 37-41 M different. At 39 an subjects, at each ws wa s only 2 grams and results of this wel subjects should be v, response relationship The temporal rel dence of disease and has been cited as ev:

617 ity is not significantly higher ites, or in the upper social sses 1 and 2). On the other z higher perinatal mortality in as the less educate-i whites and anual and unskilled workers in icy of the so-called "effect" of 3othe si s - not specific to the babies of Other associations such as ition and socio-economic status 1 significance. sd statistical association bet- :al mortality is much less than previous history of perinatal For example social class 5 shows an excess of perinatal ocial class 1. This excess is h is associated with smoking :y wea k statistical association pothesi s. lationship is ofte n quoted as evidence for the causal hypothesis. In such studi,~s, it is obviously imoortant to keep other factors constant, while the different d oses and their response are compared. This is dif- ficult except by statistical methods such as by the analysis of co-variance. Most of the studies which have reported a dose response relationship have not corrected for the influence of other factors. However, a study by Mayer and associates (21) have done this and claimed a small "independent" effect of smoking, with public hospital status and previous pregnancy loss showing much stronger associations with perinatal mortality. In a study of a large number of subjects (some 50,000) by Naeye (22) where all the factors that are known to affect birthweight were taken into account, the difference in the mean birthweight o; babies of light smokers and of heavy -smokers at 37-41 weeks of gestation was not significantly different. At 39 and 40 weeks of gestation with 8454 and 10300 subjects, at each week the difference in the mean birthweight was only 2 grams and 1 gram, respectively (See Figure 1). The results of this w ell c ontrolled s tudy using large numbers of subjects should b e very reliable, but do not indicate a dose- response relationship. The t emporal relationship between the change in the inci- dence of disease and the amount of smoking in the population has been cited as evidence in the controversy regarding smoking -12 -

618 and certain diseases e.g., lung cancer. In the case of smoking and perinata 1 mortality, ther e i s a n invers e tempora 1 relation- ship. Smoking has marke3ly increased in the last thirty _vears in women, whereas perinatal mortality has markedly decreased during this pericd in most Western countries. - - . - In view of the foregoing, it is necessary to consider some alternative hypotheses to explain the higher incidence of fetal growth retardation and perinatal mortality reported in some groups of mothers who smoke during pregnancy. ' "Nutritional"hynothesis and the role of weight gain in oreg- nancv. It is well known that in poorer countries with nutritional deficiencies, the weight of the mother as well as that of the baby are lower than that of those in prosperous countris. Nutritional supplementation is known to incresse the average birthweight in these poor countries. -It is generally assutned that in prosperous western countries, such as the U.S.A. and Britain, there are no nutritional deficiencies. However, several studies have shown that it is not true for all sections of the population in these countries. Proteins, vitamins and trace elements which are essential for the normal development of the fetus are the most expensive of foods and could be defi- cient in the diet of th '_n a study in the U.S.A. , 1 ow-i ncome g roup a te 1 e income group (12) : = In third of the mothers (5 calorie intake and had a birthweight babies (13). trace metal, naanely zinc, pooulation in the U.S.A undernutrition as a ca. reflected in the higher the poorer sections of ,•K. For example,-the : 3^d 2) have an incidence the l ower s ocial c lass es litv per 1000 births (3' whereas for the lower so is 35. In the U.S.A., pe rinatal mortality i s h and the less-educated wh 'cetter-educated whites. amerges from many of th, =hat the excess of per, "'oke is not found in tl ?'»rer sections of the '~.:-077 O-x2--40

=ancer. In the case of smoking 3 a n inverse tempora 1 relation- -eased in the last thirty years -tality`has markedly decreased g, it is necessary to consider :xplain the higher incidence of ?rinatal mortality reported in during pregnancy. )orer countries with nutritional mother as well as that of the those in prosperous countries. known to increase the average tries. It is generally assumed ttries, such as the U.S.A. and tional deficiencies. However, it is not true for all sections intries. Proteins, vitamins and :ial for the normal developm_nt sive of foods and could be defi- 13- cient in the diet of the mothers from 1ow-income grouns (11) . In a study in the U.S.A. , it has been shown that mothers from a low-income group ate less protein than mothers from a high income group (12). In amther study in the U.S.A. nearly a third of the mothers (53 out of 182) had a low protein-low calorie intake and had a significantly higher incidence of low birthweight babies (13). Deficiency of an extremely important trace metal, namely zinc, can occur in certain sections of the population in the U.S.A. (14). The possible importance of undernutrition as a cause of fetal growth retardation is reflected in the higher incidence of low birthweight babies in the poorer sections of the population in the U.S.A. and the U.K. For example, the upper social groups (social classes 1 and 2) have an incidence'of low birthweight of 4.5% whereas for the lower social clam es it is S.2$ (15). The perinatal morta- lity per 1000 births (3) for the upper social classes is 9.5 whereas f or the lower social c lasses (social c lass 4 and 5) i t is 35. In the U.S.A. , the incidence of low birthweight and perinatal mortality is higher for t-he babies of black mothers and the less-educate3 whites than it is for the babies of the better-educated whites. An extremely important fact that emerges from many of the studies on smoking and pregnancy is that the excess of perinatal mortality found in mothers who smoke is not found in the prosperous groups, but only in the Poorer sections of the community in the U.S.A. and the U.K. 95-077 0-82-40 -14-

620 (See Table 2) . It can be seen that perinatal mortality ratios of smokers' babies to those of non-smokers' babies, varies between 1.02 to 1.13 in seven studies in which the mothers were from the upper social groups whereas for the lower social groups the mortality ratio varied from 1.39 to 2.16. 'Shese results make it abundantly clear that the soci o-economi c status is very important in determining the incidence of perinatal mortality. A large number of studies have reported a higher incidence of low birthweight babies for mothers wh o smok e during preg- nancy since Simpson in 1957 first made this observation (See ref. 1 for further references). As mentione d before, most of these have not taken into aceount factors which affect birth- -weight such as social class. In a few studies which have taken these factors into account there is no significant difference in birthweight betwee n smokers' and non-smokers' babies (See Table 3) . Th e possible r ole of nutrition is highlighted in a recent study conducted on mothers from Bellshill Maternity Fbspital (16) . The birthweight was adjusted for the gestational parity of the mother and the sex of the infant (4). age, mothers were classified according to the social class of the husband. It was found that the mean birthweight of the upper -15- Q social classes 1, 2 z not show a signific s:nokers whereas this cant in the lower sc unskilled.) The ("SFGA" ) infant s wa s and non-smokers in 3reate r i n socia 1 clz 4). If smoking durir retardation, why doe mothers and not in o; so-called "effects" o 3nd pick on only the F The results from now the same data ca: `or or against the "c :nfants for all soci smokers and 15.2% for SFGA infants between significant (p < 0.0 _iese results that incidence of SFGA inf. other s tudies . It w _his claim with thi

: that perinatal mortality ratios of non-smokers' babies, varies :tudies in which the mothers were whereas for the lower social ried from 1.39 to 2.16. 'Shese r tha t th e soci o-economi c status ning the incidenee of perinatal have reported a higher incidence mothers who smoke during preg- `irst made this observatirn (See ). As mentioned before, most of bunt factors which affect birth- In a few studies which have taken iere i s no s ignif icant d i fference s' and non-smokers' babies (See ition is highlighted in a recent rom Bellshill Maternity Ebspital djusted for the gestational age, ie sex of the infant (4). ".he rding to the social class of the .he mean birthweight of the upper 621 social classes 1, 2 and also 3(pro_°essional an:'. technical) did not show a significant difference between smokers and non- smokers w hereas this difference w as large and highly signifi- cant in the lower social classes namely 4 and 5 (manual and unskilled.) The incids nce of sma11-`.or-gestational-age ("SFGA" ) infants was significantly different between smokers and non-smokers in only the lower social classes and was greater in social class 5 than in social class 4. (See Table 4). If smoking during pregnancy is the cause of fetal growth retardation, why does it occur only in certain groups of mothers and not in others? It can scarcely be argued that the so-called "effects" of smoking respect the upper social classes and pick on only the poor and underprivileged mothers. • The results from this study give an excellent example of how the same data can be interpreted in different ways, either for or against the "causal" hypothesis. The incidence of SFGA infants for all social classes as a whole was 6.4% for non- smokers and 15.2% for smokers. This difference in incidence of SFGA infants b etween smokers a nd non-smokers is statistically significant (p < 0.01). It would be quite easy to claim from these results that smoking was the cause of the higher incidence of SFGA infants in smokers, as has been done in many other s tudies . It would a lso be quite d i fficult to d isprove this claim with this experimental design. However, if an -16 -

622 attempt is made to make smoking the only variable between smokers and non-smokers by correcting, as far as possible,- for factors that are k nown to affect birthweight, the results are no longer in favour of the causal hypothesis, and are in fact against -the causal hypothesis. Thus, if smokers were classified according to the social class of the mothers' husband, entirely different incidences of Low birthweight are seen in the different social classes for the same amount of smoking, a nd smoking mothers from the upper socio-ecoromic groups do not show an excess of incidence of SFGA infants. If smoking d oes n ot cause fetal growth retardation in one social group, it canno t obviously b e th e cause in another ethnically same social group. _Therefore, the associat~on between the higher incidence of SFGA infants and smoking in the poorer social groups c an not b e considered t o have a causal relation- ship. A causal relationship might be claime3 if smoking were the only variable between smokers and non-smokers. Most studies on smoking and pregnancy assume that the smoking habit is the only difference between the life-style of smokers and non-smokers. z Contrary to this assumption, there is much evidence that smokers and non-smokers differ in their personality, their attitude to life and their eating and drinking habits (23). A. significant difference in the drinking habits alone could -17- account for the hic groups of smoking mo fie3 to claim that retar9ation in any c non-s moking mothers a "_he gradient in t n socio-economic str that some factor sucr associated with the tively related and °etal growth retardat nutr itiona l statu s o f cause of the higher the poore r socia 1 gro .ri11 lend itself to recently verified on Hosoital. Maternal nutritic the protei n intake i n _^.others who had a low portionately higher ir those with a normal p '-others who smoked, t

the only variable between ng, as. far as possible, for, 3irthweight, the results are hypothesis, and are in fact Thus, if smokers were ial class of the mothers' tnces of low birthweight are ses for the same amount of )m the upper socio-economic cidence of SFGA infants.,_If th retardation in one social caus e in another-ethnically =he_association.between the and smoking in the poorer i to have a- causal, relation- z b e claimed if smoking were ers and non-s mokers. - Most ssume that the smoking ha3it ze l:ife-s_tyle of smokers and ..-. . • .._ > -. . -. _ there is much evidence that in their, personality. their and drinking habits (23). A :rinking. habits alone could 623 account for the higher incidence of SFGA infants in certain groups of smoking mothers. There`.ore, it is entirely unjusti- fie3 to claim that smoking M%as the cause of fetal growth retardation in any group of smokers, unless the smoking and non-smoking mothers are exactly comaarable in every other wav. The gradient in the incidence of SFGA infants with decrease in socio-economic status (Fig. 4) of smoking mothers indicates that some factor such as family income, which is qu3ntitatively associated with the social class gradient, is also quantita- tively related and possibly bears a causal relationship to fetal growth retardation. qince family income could affect the nutritional status of the mother, maternal nutrition may be the cause of the higher incidence of fetal gro.:th retardation in the poorer social groups. This view has the advantage that it will lend itself to experimental verification and has been recently verified on a grou p of mothers in Bellshill Maternity HosDital. Maternal nutritional status was investigated by measuring the protein intake in the third trimester o`_ pregnancy. 'Ihose mothers who had a low protein intake in pregnancy had a dispro- portionately higher incidence o f SFGA infants (438 ) comnared to those with a normal protein intake (3.4$). See Table 6. ?imong mothers who smoked, those who had a normal protein intake had -18-

624 the same lcw incidence of SFGA infants as non-smokers. This indicates that nutritional deficiency, but not smoking, is a pre-requisite for the development of fetal growth retardation. Since the majority of mothers who were delivered of SFGA infants (65 out of 95, 68%) had a low protein intake, correc- tion of this and other associated nutritional deficiencies is likely to result in a striking reduction in the incidence of low birthweight, irrespective of their smoking habits. Further evidence in favour of the nutritional hypothe- sis comes from studies on maternal weight gain during preg- nancy. It has been suggested that smoking mediated its so- called "effect" through maternal weight gain; Rush (17) and Davies and associates (18) have shown t~hat the mean weekly -weight gains were reflected in the infant's birthweight. Davies et al. reported that their results are consistent with the view that mothers' own weight gain has a greater effect on fetal growth with smoking having a very small "independent" ef fect . Other studies (19, 20) investigating the role of weight gain during pregnancy have reported an independent "effect" of smoking when weight gain was held Eiowever, when an analysis of variance was carried constant. out (21) , previous pregnancy loss and hospital status (private or public) -19 - showed a much ' ratio 97.4 and 4 and associates ( gai n i n feta l grc non-smokers and • who gained less trimesters of F retardation 'was with low weight g mothers with grea times as high amo with greater weic delivered infants of mothers wh o wer -were significantly 1070 mothers who retardation. Further evi9er pregnancy is provi: the weight gain in non-smokers and twc the first or the : who smoked during o infants in the pre

M 625 infants as non-smokers. .,This _ency, but not smoking, is a : of fetal growth retardation. who were delivered of S?GA a low protein intake, correc- d nutritional deficiencies is reduction in the incidence of :heir smoking habits. ,ur of the nutritional hypothe- nal weight gain during preg- :hat smoking mediated its so- : weight gain; a shown Rush (17) and that the mean weekly .n the infant's birthweight. ir results are consistent with t gain has a greater effect on ng a very small "independent" ) investigating the role of have reported an independent ght gain was heli constant. ariance was carried out (21). ,ital status (private or public) showed a much higher degree of statistical correlati~n (F. ratio 97.4 and 44.2 respectivel_v) tha n smoking (8.4). !iiller and associates (9) have shed some light on the role of weight gain in fetal growth retardation. They founfl in a study of 688 non-smokers and 424 smokers that there were 62 and 67 mothers who gained less than 0.5 pound per week in the last two trimesters of pregnancy. The incidence of fetal growth retardation was five times higher among non-smoking mothers with low weight gain (less than 0.5 pound'per week) than among mothers with greater weight gain. For smokers, it was three times as high among those with low weight gain than among those with greater weight gain. In th e group of 42 mothers who delivered infants with fetr~l growth retardation, the incidence of mothers who were in the lower social classes and who smoked -were significantly higher than the respective incidences a mong 1070 mothers who did not have infants with fetal growth retardation. Further e vidence for the importance of weight gain in pregnancy is provided by the work of Naeye (22). He studied the weight gain in successive pregnancies in groups of smokers, non-smokers and two groups of wois?n who smoked only in either the first or the secon-i pregnancy. Naeye found that mothers who smoked during one pregnancy but not in another had smaller infants in the pregra ncy in which they smoked irrespective of -20- 9-