Acta Obstet Gynecol Scand 2000; 79: 485 489 Copyright C Acta Obstet Gynecol Scand 2000 Printed in Denmark All rights reserved Acta Obstetricia et Gynecologica Scandinavica ISSN 0001-6349 ORIGINAL ARTICLE Maternal smoking and birthweight: effect modification of period, maternal age and paternal smoking KJELL HAUG 1, LORENTZ M. IRGENS 1,2, ROLV SKJÆRVEN 1,2, TROND MARKESTAD 3, VALBORG BASTE 2 AND PATRICIA SCHREUDER 2 From the 1 Department of Public Health and Primary Health Care and the 2 Medical Birth Registry of Norway, Locus of Registrybased Epidemiology, University of Bergen, and the 3 Department of Pediatrics, Haukeland University Hospital, Bergen, Norway Acta Obstet Gynecol Scand 2000; 79: 485 489. C Acta Obstet Gynecol Scand 2000 Objective. To study the effect on birthweight of maternal smoking, and its modification by study period, maternal age and paternal smoking. Design. A retrospective questionnaire based national survey comprising a random sample (nω34,799) of all mothers giving birth in Norway 1970 91. Variables studied were parental smoking during pregnancy, birthweight, maternal age and infant s year of birth. Results. The overall difference in mean birthweight between non-smoking and smoking mothers was 197 g. The difference in birthweight between non-smoking and smoking mothers increased with maternal age from 182 g ( 20 years of age) to 232 g (35π years of age). There was no significant effect of paternal smoking on birthweight when the mother was a nonsmoker. When the mother was a smoker and the father was a non-smoker, the birthweight, adjusted for maternal age, was reduced by 153 g (p 0.005). However, when both parents smoked, the birthweight, adjusted for maternal age, was reduced by 201 g (p 0.0005). Even though the prevalence of paternal smoking decreased by 38% during the study period, there was no significant increase in overall mean birthweight. Implication and relevance of results. The negative effect of maternal smoking on birthweight appears to increase with maternal age. For a non-smoking pregnant woman to live with a smoking partner has little, if any, effect on birthweight. The negative effect of paternal smoking was only observed when the mother was smoking and might reflect two possible mechanisms: (1) that a smoking mother has a greater cigarette consumption when the partner also smokes, and (2) that a smoking mother is less concerned about passive smoking than a nonsmoking mother. Key words: birthweight; maternal age; Norway; parental smoking; secular trends Since Simpson in 1957 described a possible association between maternal smoking and reduced birthweight (1), a reduction in mean birthweight of approximately 200 g with a range between 90 and 300 g has been reported in a series of studies (2), in some with a dose-response relationship (3, 4). Infants of smoking mothers have also been found to be smaller than those of non-smokers in length, as well as circumference of head, chest and shoulder (5). In 1986, Martin and Bracken demonstrated that non-smoking women exposed to passive smoking for at least two hours per day delivered infants who were on average 24 g lighter, even when adjusting for confounding factors such as gestational age, parity and maternal age (6). A dose-response relationship has also been demonstrated for paternal smoking (7, 8). In 1997, Horta et al. demonstrated a reduction in mean birthweight of 30 g if the partner smoked, after adjusting for a series of confounders, including social class, education, inter-pregnancy interval, prior low birthweight and maternal height (9). Some of these results were based on low num-
486 K. Haug et al. bers. Furthermore, attitude to passive smoking exposure has changed during the last years, and the effects described may have been modified over time. Thus, the objective of the present study was to assess secular trends in the effect of maternal smoking on birthweight and its modification of maternal age and paternal smoking in a large sample of randomly selected women in Norway 1970 91. to test for differences in birthweight, with and without adjustment for maternal age, with 5 categories dummy variables. Variance analysis was also used to test for linear increase in birthweight with increasing maternal age, and to test if the linear trend differed between non-smokers and smokers. Data analysis was conducted by the SPSS/PCπ V7.5.1. The study was approved by the Regional Committee on Medical Research Ethics. Material and methods Smoking habits of pregnant women and their partners, as well as birthweight of the children, were assessed in a retrospective study for the years 1970 through 1991, which primarily addressed the effect of sleeping position on sudden infant death syndrome (10). The study was carried out in November-December 1992 based on a one page postal questionnaire, which was sent to a sample of mothers who had given birth in the 20 largest maternity institutions in Norway in the survey years 1970, 1975, 1980, 1985, 1989, 1990 and 1991. For each survey year and institution, 250 mothers were randomly sampled from The Medical Birth Registry of Norway among all mothers of infants weighing 2500 g or more at birth, and without any reported birth defects, who had survived the first year of life; altogether 34,799 mothers. The parents were asked about smoking habits during pregnancy. In addition, the questionnaire contained information on the infant s date of birth, the birthweight of the child as well as the year of birth of the parents. The questionnaire contained no identifying information, precluding a selective reminder to the study population. Of the 34,799 mothers who received the questionnaire, 24,438 (70.2%) responded and, among these, 22,883 (65.8% of the study population) answered all the questions of interest in the present study. The effect of smoking on birthweight was defined as the difference in birthweight between nonsmokers and smokers. Variance analysis was used Results The prevalence of smoking during pregnancy declined from 32% in 1970 to 27% in 1991 among mothers and from 59% to 36% among fathers. The smoking prevalence of mothers under 20 and more than 34 years of age were 49% and 19%, respectively. The mean birthweight was 3602 g (s.d. 483). The overall difference in mean birthweight between non-smoking and smoking women was 197 g. Unadjusted for maternal smoking, the difference in birthweight between infants of non-smoking and smoking fathers was 88 g. The mean maternal age increased from 26.2 to 28.5 years during the study period. The difference in birthweight between infants of non-smoking and smoking mothers increased with maternal age (Table I) (p 0.05). If the mother did not smoke, the mean birthweight increased steadily by maternal age (p 0.001), irrespective of whether the father was a non-smoker or smoker. When both parents smoked, the mean birthweight did not increase when the mother was more than 24 years of age. If only the mother smoked, a plateau was reached five years later. Adjusted for maternal age and paternal smoking, maternal smoking led to a reduction in birthweight of 153 g (Table II). Adjusted for maternal age, the effect of paternal smoking was a reduction of 1 g if mother did not smoke (n.s.) and 48 g if mother smoked (p 0.01). If both parents smoked, the birthweight, adjusted for maternal age, was re- Table I. Birthweight by maternal smoking habits and maternal age in Norway 1970 91. Mean and standard error of the mean in non-smokers and smokers with difference in grams and 95% confidence intervals Non-smokers Smokers Difference Maternal age n Mean birthweight (g) SE of the mean n Mean birthweight (g) SE of the mean (95% CI) 20 467 3590 22.2 459 3408 21.2 182 (123 243) 20 24 3829 3622 7.6 2302 3451 9.3 171 (147 195) 25 29 6419 3651 5.9 2408 3468 9.3 183 (161 205) 30 34 4175 3686 7.5 1284 3478 13.2 208 (178 238) 35π 1659 3706 12.4 408 3474 24.6 232 (177 287)
Birthweight and parental smoking 487 Table II. Mean birthweight by parental smoking habits in Norway 1970 91. Effects of smoking expressed as differences in mean birthweight, with 95% confidence intervals, unadjusted and adjusted for maternal age Birthweight Effect of smoking on birthweight (g) Parental smoking habits Unad- Adjusted Mother Father N Mean (g) SE mean justed p-value (95% CI) p-value Non-S* Non-S 11496 3661 4.47 ref ref Non-S Smoker 4934 3652 6.93 º7 0.13 º1 (º17-π15) 0.89 Smoker Non-S 1538 3501 12.05 º160 0.0005 º153 (º178-º128) 0.0005 Smoker Smoker 4870 3447 6.51 º214 0.0005 º201 (º218-º185) 0.0005 Non-S *Ωnon-smoker. duced by 201 g compared to non-smoking parents (p 0.0005). The difference in birthweight between nonsmokers and smokers was most evident in 1975 (221 g) while the smallest difference (173 g) was found in 1985 (Fig. 1). From 1980 onwards, less reduction in birthweight was observed when only the mother was smoking compared to both parents smoking. Mean birthweight was also investigated in three different age strata ( 25, 25 29, 30π) during the study period for non-smokers and smokers. At each time during the study period, the age dependent pattern was stable. Fig. 1. Mean birthweight by parental smoking habits and year of birth of the child, Norway 1970 91. Abbreviations: nonsmoking mother (Mª), smoking mother (Mπ), non-smoking father (Fª), smoking father (Fπ). Discussion The difference in birthweight between non-smoking and smoking mothers amounted to 197 g, which is in accordance with the reduction in birthweight observed repeatedly during the last 30 years (2). Dependent on the smoking habits of the mother, we found a difference in the effect of paternal smoking. If the mother did not smoke, paternal smoking resulted in a moderate 7 g reduction in mean birthweight, or a reduction of 1 g when adjusted for maternal age. This pattern was observed throughout the observation period. Due to the attitude towards passive smoking, a smoking father might not expose a non-smoking mother to tobacco smoke. Results indicate that passive smoking among non-smoking, pregnant women is a minor problem in Norway. On the other hand, if the mother did smoke, the effect of paternal smoking was a reduction of 54 g, or 48 g adjusted for maternal age. There might be at least two possible mechanisms which can explain this phenomenon: (A) a woman who lives with a smoking partner, has a higher cigarette consumption than a woman who lives with a non-smoking partner, and (B) when both parents smoke, it is likely that they smoke while they are together in their leisure time, in other words passive smoking in addition. However, we have neither information about the number of cigarettes smoked per day, nor environmental smoking. Smoking prevalence among fathers decreased from 59% in 1970 to 36% in 1991. If environmental smoking does have an effect on birthweight, we would have expected an increase in overall mean birthweight during the period. However, no significant increase was observed, either in our study population, or in the whole newborn population of Norway (11, 12). The small observed secular increase in overall mean birthweight may be ex-
488 K. Haug et al. plained by the small decrease in smoking prevalence found among pregnant women (from 32% in 1970 to 27% in 1991), and the increase in mean age at birth among the mothers during the period. This finding supports the hypothesis that paternal smoking has only a marginal, if any, effect on birthweight. The predominant cause of the reduction in mean birthweight attributable to smoking in pregnant women would seem to be their own cigarette smoking. Focusing on environmental and paternal smoking with regard to birthweight could, in fact, be a blind alley. However, it is easier for a smoking woman living with a non-smoking partner to stop smoking than for a woman living with a smoking partner (13). Therefore, smoking habits of the partner are still of importance. We have found a linear increase in birthweight with increasing age among non-smoking women. This increase was less evident among smokers. The difference in birthweight between non-smokers and smokers therefore increased with increasing age of the mothers, which is in accordance with the results in two other studies (3, 14). We have excluded children with a birthweight less than 2500 g. In 1994, the proportion of children in Norway with a birthweight less than 2500 g was 5.4% (11). This is a very heterogeneous group and 82% of the small-for-gestational age (SGA) children are born with low birthweight for reasons other than smoking (3). The response rate to the questionnaire increased during the study period and was 70% on average (15). Some women might have had problems recalling details 20 years back in time. However, a selection bias related to smoking is not very likely since the study was designed to explore possible risk factors for SIDS and did not focus especially on smoking and birthweight. This assumption is supported by the high smoking prevalence found in the study population, comparable with contemporary surveys of the smoking prevalence in the general population in the study period (16). Validity of self-reported smoking habits is of vital importance in smoking related studies. In other studies of pregnant women, the questionnaire responses have been an accurate source of information on smoking habits, internally consistent and highly reliable (17, 18). Mothers with normal births have also been found to be an adequate reference group for obtaining information on lifestyle (19). Since SIDS was a health problem of great concern among parents in Norway, due to the high mortality rate, we also think that the responders took part in the study with more obligation and care than in most other questionnaire surveys. There is no reason to believe that the information on birthweight in our study is less reliable (15). In conclusion, we have found a difference of 197 g in mean birthweight between infants of nonsmoking and smoking mothers. The negative effect of smoking on birthweight increased with increasing maternal age. When the mother and father smoked, there was a significant reduction of 48 g in birthweight, compared with a smoking mother and a non-smoking father. No significant effect of paternal smoking was observed when the mother was a non-smoker. Due to the attitude of nonsmoking pregnant women in Norway towards avoiding environmental smoking, we do not believe passive smoking is a serious problem with regard to birthweight. References 1. Simpson WJ. A preliminary report of cigarette smoking and the incidence of prematurity. Am J Obstet Gynecol 1957; 73: 808 15. 2. Lumley J, Astbury J. Smoking. In: Chalmers I, Enkin M, Keirse MJNC eds: Effective care in pregnancy and childbirth. Oxford: Oxford University Press, 1989: 242 7. 3. Backe B. Maternal smoking and age. Effects on birthweight and risk for small-for-gestational age births. Acta Obstet Gynecol Scand 1993; 72: 172 6. 4. Brooke OG, Anderson HR, Bland JM, Peacock JL, Stewart CM. Effects on birth weight of smoking, alcohol, caffeine socioeconomic factors, and psychosocial stress. BMJ 1989; 298: 795 801. 5. The health consequences of smoking for women: a report of the surgeon general. Rockville, Md.: US Department of Health and Human Services, 1980. 6. Martin TR, Bracken MB. Association of low birthweight with passive smoke exposure in pregnancy. Am J Epidemiol 1986; 124: 633 42. 7. Rubin DH, Krasilnikoff PA, Leventhal JM, Weile B, Berget A. Effect of passive smoking on birthweight. Lancet 1986; 2: 415 17. 8. Martinez FD, Wright AL, Taussig LM et al. The effect of paternal smoking on the birthweight of newborns whose mothers did not smoke. Am J Public Health 1994; 84: 1489 91. 9. Horta BL, Victora CG, Menezes AM, Halpern R, Barros FC. Low birthweight, preterm births and intrauterine growth retardation in relation to maternal smoking. Paediatr Perinat Epidemiol 1997; 11: 140 51. 10. Irgens LM, Markestad T, Baste V, Schreuder P, Skjærven R, Øyen N. Sleeping position and sudden infant death syndrome in Norway 1967 91. Arch Dis Child 1995; 72: 478 82. 11. Birth in Norway through 30 years. Medical Birth Registry of Norway. Bergen: University of Bergen/National Institute of Health, 1997. 12. Daltveit AK, Vollset SE, Skjærven R, Irgens LM. The impact of multiple births and elective deliveries on the trends in low birthweight in Norway 1967 95. Am J Epidemiol 1999; 149: 1128 33. 13. Haug K, Aarø LE, Fugelli P. Smoking habits in early pregnancy and attitudes towards smoking cessation among pregnant women and their partners. Fam Pract 1992; 9: 494 9. 14. Fox SH, Koepsell TD, Daling JR. Birth weight and smok-
Birthweight and parental smoking 489 ing during pregnancy effect modification by maternal age. Am J Epidemiol 1994; 139: 1008 15. 15. Haug K, Irgens LM, Baste V, Markestad T, Skjærven R, Schreuder P. Secular trends in breastfeeding and parental smoking. Acta Paed 1998; 87: 1023 7. 16. Tobacco consumption and attitudes to smoking in Norway 1973 95 (Tobakksbruk og holdninger i Norge utviklingen 1973 95, in Norwegian, English summary). Oslo: The Norwegian Council on Tobacco and Health, 1996. 17. Fox NL, Sexton M, Hebel JR, Thomson B. The reliability of self-reports of smoking and alcohol consumption by pregnant women. Addict Behav 1989; 14: 187 95. 18. Petitti DB, Friedman GD, Kahn W. Accuracy of information on smoking habits provided on self-administered research questionnaires. Am J Public Health 1981; 71: 308 11. 19. Delgado-Rodriguez M, Gomez-Olmedo M, Bueno-Cavanillas A, Garcia-Martin M, Galvez-Vargas R. Recall bias in a case-control study of low birth weight. J Clin Epidemiol 1995; 48: 1133 40. Address for correspondence: Kjell Haug, M.D., Ph.D. Department of Public Health and Primary Health Care Ulriksdal 8c N-5009 Bergen Norway