INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Maternal smoking during pregnancy is an avoidable risk factor for a number of adverse outcomes in infancy and later childhood, including low birthweight, preterm delivery, and sudden death in infancy (1, 2). The risk of lower respiratory illness, wheezing and asthma is increased among young children whose mothers smoke (3), and it has been suggested that these effects are mediated predominantly by prenatal effects on lung development (4, 5) and may operate by diminishing airway or alveolar growth, which is particularly rapid during fetal life.

Impaired airway function has been reported in infants of mothers who smoke (6), but most studies do not allow separation of the effects of prenatal and postnatal exposure. Although altered expiratory flow patterns during tidal breathing have been reported recently among term newborns (5, 9), the issue of how early in pregnancy these adverse effects may occur has not been addressed.

We now report a study examining the influence of maternal smoking during pregnancy on respiratory function during early life in preterm infants without known respiratory disease. The aim was to determine whether the adverse effects of maternal smoking on respiratory function are limited to the last weeks of pregnancy when placental function is likely to be most compromised.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Preterm infants (less than 36 completed weeks of gestation) without major congenital abnormalities or neonatal respiratory disease, whose parents were either both Northern European (white) or both Afro-Caribbean (black), were eligible for recruitment from the Special Care Baby Unit at the Homerton Hospital, London. Infants who, after delivery, required assisted ventilation for more than 6 h or supplemental oxygen for more than 24 h were ineligible, as were those whose parents did not speak English. The study was approved by the City and Hackney Local Research Ethics Committee. Written informed consent was obtained from the parents, who were usually present during respiratory measurements.

Information on maternal smoking during pregnancy was obtained retrospectively from maternal report at recruitment, as prospective documentation was not feasible in this study. Mothers were asked whether they had smoked at any time during pregnancy and, if so, the gestational age at stopping or starting smoking. Mothers were classified as smokers if they had smoked at any time after 4 wk of gestation, i.e., once they were aware that they might be pregnant. Indirect biochemical validation of maternal report of smoking was also performed by cotinine analysis (10) of urine obtained, whenever possible, from each infant within 24 h of delivery, and of maternal saliva, collected at infant respiratory function testing. Additional background information was obtained by the research midwife from mothers at recruitment, including paternal and maternal occupation, maternal age on leaving full-time education, and first degree family history of asthma.

Infant respiratory function was measured prior to discharge from hospital, with the infant supine and naturally asleep, and with ambient temperature maintained at 23 to 25° C. Data were collected as described previously (11) during periods of behaviorally determined quiet sleep. The main outcome variables were maximal forced expiratory flow at functional residual capacity (max_FRC) (13, 14), an index of peripheral airway function, and time to peak tidal expiratory flow as a proportion of total expiratory time (TPTEF:TE) (12), an index of tidal expiratory flow patterns. The latter reflects the degree to which expiratory flow and timing are modulated in order to slow lung emptying during expiration. A reduction in TPTEF:TE is associated with diminished airway function (15). max_FRC was measured using the rapid thoraco-abdominal compression technique and reported as the mean of the three highest technically satisfactory measurements obtained (13, 14). TPTEF:TE, together with respiratory rate and tidal volume, were measured via a face mask and flowmeter and reported as the mean of 60 consecutive regular breaths collected over two or three separate epochs (12). Passive respiratory compliance (Crs) was measured using the multiple occlusion test (11, 14), so that any observed differences in peripheral airway function might be interpreted with respect to the elastic recoil of the respiratory system.

Statistical Analysis

Background characteristics were examined in relation to maternal report of smoking during pregnancy (Minitab for Windows, Release 11), and the association between maternal smoking during pregnancy and max_FRC or TPTEF:TE examined using multiple regression analyses (SPSS for Windows, Release 7). The sample size was sufficient to detect a 1 standard deviation difference in max_FRC or TPTEF:TE between infants born to smoking or to nonsmoking mothers, with 90% power at the 5% level after adjustment for sex, ethnic group, body size, postnatal age, and socioeconomic status (16).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

TPTEF:TE was successfully measured in 108 infants, and max_FRC and Crs in 89 and 101 infants, respectively. Failures were largely due to insufficient quiet sleep or technically unacceptable measurements (11). Of the entire group, 49 (45%) were male and 65 (60%) were white (Table 1). Nine infants had received low concentrations of oxygen for less than 24 h (inspired oxygen concentration < 0.30), five had been ventilated briefly, but none had been given surfactant. Forty (37%) infants were born to mothers who had smoked during pregnancy. The proportion of infants whose mothers smoked was similar in those with and without measurements of max_FRC, as were all other background characteristics (data not shown). A similar proportion of mothers in the smoking and nonsmoking groups had received antenatal steroids just prior to delivery (18 and 19%, respectively).

Of the 68 mothers who reported not smoking during pregnancy, one smoked up to but not after the third week of pregnancy. Two of 40 mothers classified as smokers had ceased smoking at 12 and 20 wk of gestation, respectively. Mothers reported smoking between 1 and 40 cigarettes a day (median, 10). Maternal salivary cotinine, available for 81 mothers, discriminated well between self-reported smokers and nonsmokers: median (range) in smokers being 246.8 ng/ml (0 to 673.3) compared with 0.2 ng/ml (0 to 3.0) in nonsmokers (p < 0.0001) (Table 1). Similarly, median (range) urinary cotinine within 48 h of birth was 69.2 ng/ml (1.5 to 458.0) in infants of smoking mothers and 0.4 ng/ml (0 to 3.6) in infants born to nonsmokers (p < 0.0001) (Table 1).

Infants of smoking mothers were significantly more likely than those of nonsmoking mothers to be white (75% versus 51%), to have a father with a manual occupation (66% versus 43%), and to have a mother who left full-time education at a younger age (median [range], 16 yr (14 to 23) versus 17 yr [15 to 31]) (Table 1). They were equally likely to have a first degree relative with asthma (28% in each group). Infants of smoking mothers tended to be of lower birthweight (mean [SD]; 1,873 g [391] versus 1,959 g [457]) (p < 0.10), and at the time of respiratory function testing were significantly lighter (2,241 g [309] versus 2,363 g [356]) (p < 0.05) and shorter (44.9 cm [2.1] versus 45.9 cm [2.5]) (p < 0.05) compared with infants of non-smokers (Table 2).

max_FRC and TPTEF:TE were diminished in infants exposed to tobacco in utero. Mean (SD) TPTEF:TE was 0.369 (0.109) and 0.426 (0.135), respectively, in infants of mothers who did and did not smoke during pregnancy (p  0.02), and mean (SD) max_FRC was 85.2 (41.7) ml/s and 103.8 (49.7) ml/s (p = 0.07; Table 3). Compliance, tidal volume, respiratory rate, and total expiratory time did not differ according to maternal smoking in pregnancy when these variables were compared in absolute terms (Table 3). Similarly, when corrected for body weight, there was no significant difference between the groups for either Crs (11.5 versus 11.2 ml/kg/kPa for smoking versus nonsmoking) or tidal volume (7.2 versus 7.0 ml/kg, respectively).

There were no sex differences in ethnic group or paternal social class, however, significantly more girls than boys had a positive family history of asthma (32 and 14%, respectively; p  0.05), and although of similar postnatal age, body weight, and length, the boys had lower max_FRC, and TPTEF:TE was significantly diminished. In boys and girls, respectively, mean (SD) max_FRC was 87.0 (41.2) ml/s and 105.4 (51.4) ml/s (95% CI, boys-girls: 38.2 to 1.3 ml/s; p = 0.07) and TPTEF:TE was 0.371 (0.125) and 0.433 (0.126) (95% CI, boys-girls: 0.110 to 0.013; p  0.01). There were otherwise no significant sex differences in respiratory function.

More white than black mothers reported smoking during pregnancy (Table 1). max_FRC and TPTEF:TE were significantly lower in white infants. In white and black infants, respectively, mean (SD) max_FRC was 87.3 (49.2) ml/s and 109.3 (42.3) ml/s (95% CI, white-black: 41.8 to 2.3 ml/s; p = 0.03), and mean (SD) TPTEF:TE was 0.373 (0.108) and 0.452 (0.143) (95% CI, white-black: 0.127 to 0.031; p = 0.002). Weight-corrected compliance was higher (11.7 versus 10.7 ml/ kPa/kg; 95% CI, 0.02 to 2.00), respiratory rate lower (60 versus 66 breaths/min; 95% CI, 11.0, 0.4), and expiratory time was longer (0.596 versus 0.549 s; 95% CI, 0.015, 0.110) in white infants. Although maternal smoking during pregnancy was more prevalent in infants whose fathers were in manual occupations (Table 1), there were no social class differences in size at birth or at respiratory function testing, or in any of the respiratory function parameters measured except TPTEF:TE.

Multiple regression analyses were undertaken before and after adjusting for those variables identified in univariate analyses as being significantly associated with maternal smoking and either max_FRC or TPTEF:TE. Within this population, postnatal age, body weight, and length at test were not significantly associated with TPTEF:TE (p = 0.33, 0.93, and 0.76, respectively) and were therefore not included in the model, whereas ethnic group, sex and social class were. On univariate analysis, max_FRC was not associated with weight at test (p = 0.21), or social class (p = 0.63), but there was a significant relationship with ethnic group, postnatal age, sex, and crown-heel length. After adjustment for these variables, there was no longer an association between maternal smoking and max_FRC (Table 4). By contrast, after adjustment for all appropriate factors, TPTEF:TE remained significantly diminished in infants whose mothers reported smoking during pregnancy, being a mean (95% CI) 0.046 (0.092 to 0.001) lower than in those whose mothers did not smoke (p < 0.0) (Table 4).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the United Kingdom, approximately one-third of mothers smoke during pregnancy, and this is more common among white women and those of lower socioeconomic status (3, 17). Although it is well recognized that mothers who smoke during pregnancy are at increased risk of low birthweight and preterm delivery, uncertainty remains regarding the contribution and timing of prenatal tobacco exposure to impaired respiratory function and respiratory morbidity in early childhood. By measuring respiratory function in preterm infants before hospital discharge, we were able to examine the effect of maternal smoking during pregnancy separately from that of postnatal exposure to environmental tobacco smoke. In addition, as infants in this study were born prematurely, we were able to examine whether adverse effects were limited to the last weeks of pregnancy.

In a group of otherwise healthy infants who were delivered on average 7 wk early, we found that respiratory function was impaired in those whose mothers had smoked during pregnancy. Both max_FRC, a measure of peripheral airway function, and TPTEF:TE, which reflects the degree to which expiratory flow and timing are modulated, were diminished. Maternal smoking habits were associated with paternal social class and ethnic origin of the parents, and max_FRC and TPTEF:TE with sex, ethnic origin, and, for max_FRC alone, postnatal age and body length. After adjustment for these variables, TPTEF:TE, but not max_FRC, remained significantly reduced in infants whose mothers smoked.

A number of published studies have examined the association between maternal smoking and infant respiratory function (5, 18) but, with few exceptions (5, 9), have been unable to distinguish the effects of prenatal and postnatal exposure. Stick and colleagues (5) found a significant reduction in TPTEF:TE at a postnatal age of 3 d among term infants exposed to maternal smoking in utero but not postnatally, with similar results being reported in awake full-term newborns by Lodrup and colleagues (9). Beyond the neonatal period, marked reductions in max_FRC and/or increased bronchial responsiveness have been reported among infants whose mothers smoked in some (6, 7), but not all (18), studies. The current study is the first to have evaluated the impact of maternal smoking on respiratory function in preterm infants, and from our findings it would appear that there is a small but demonstrable effect on respiratory function by at least 33 wk of gestation.

Biochemical validation of maternally reported smoking habits (based on infants' urinary and maternal salivary cotinine) suggests that misclassification of smoking exposure is unlikely to account for the results of this study. Cotinine is a sensitive measure of recent smoke exposure and it discriminates well between active, passive, and nonsmokers (3, 10). As urine was collected any time up to 24 h postnatal age, and many mothers refrain from smoking during labour, infant urinary cotinine levels on the first day of life may underestimate in utero exposure. Despite this, values as great as 458 ng/ml were observed among some infants whose mothers reported smoking, similar to levels reported in actively smoking adults (10), suggesting marked in utero exposure. By contrast, among the non-smoking group, infant urinary and maternal salivary cotinine values were very low or undetectable, the former not exceeding 3 ng/ml, below the range of values reported in passive smokers (10). Because women who smoke during pregnancy usually continue to do so after the birth of their child, maternal salivary cotinine provided a proxy measure of smoking during pregnancy.

Because neonatal respiratory disease and its treatment may affect subsequent respiratory function (21), infants requiring such treatment were excluded from the study. This could potentially introduce bias because of a "healthy infant" effect if preterm infants with impaired respiratory function caused by maternal smoking were more likely to need ventilatory assistance at birth. Ideally, we would have compared the characteristics of infants recruited to the study with those of all age-eligible infants over the study period (with respect to sex, ethnic group, maternal smoking, size at birth, etc.). Regrettably, a systematic audit with reasons for exclusion was not maintained prospectively, and it was not feasible to collect such data reliably retrospectively. However, had any such bias occurred, it would have tended to strengthen rather than to diminish our findings. Although an association between maternal smoking and preterm delivery has been reported (3), an increased risk of neonatal ventilation has not. Indeed, maturation of lung surfactant may be enhanced in infants exposed to tobacco in utero, possibly through stress-related intrauterine steroid release (6, 22). In infants delivered prematurely, this may partially compensate for the adverse effects of in utero tobacco exposure on the lung, at least during the immediate neonatal period.

Two measures of outcome were selected for this study. max_FRC, which provides a measure of peripheral airway function that is relatively independent of upper airway and nasal resistance (11, 14), is determined largely by the geometry and resistive pressure losses along the small airways under conditions of dynamic compression. This measure has been widely used to characterise the normal growth and development of the airway during infancy, and the pulmonary abnormalities associated with acute and chronic lung disorders during early childhood (14). However, the caliber of the small intrathoracic airways is determined not only by their anatomic dimensions but by the structure and compliance of the airway wall and the distending pressures surrounding them, with the latter being influenced by the lung volume at which measurements are made. We did not consider lung volume measurements to be feasible in this study of small unsedated preterm infants, but we conclude that a significant reduction in lung volume is unlikely since respiratory rate, tidal volume, and respiratory compliance did not vary according to maternal smoking habits. Despite the narrow age range studied, there was a positive correlation between max_FRC and length (but not weight). Once corrected for length, the association between max_FRC and maternal smoking disappeared (Table 4). However, it is debatable whether it is appropriate to adjust for body size since in both the current and previous (6) studies, infants of smoking mothers have been shown to be significantly shorter at birth and at testing.

max_FRC was diminished in boys and white infants, and was inversely related to postnatal age at testing. The observed sex differences are consistent with previously published reports of diminished airway function in boys during infancy and childhood (8, 13, 18, 20, 23) and may contribute to the higher prevalence of wheezing and asthma reported in boys at all ages to puberty (23, 24). We have previously reported that max_FRC tends to be higher in black than in white preterm infants (13). This may be a transient difference reflecting the measurement of flows at a slightly higher end-expiratory lung volume among the black babies because of their different pattern of breathing, with more marked expiratory braking (longer TPTEF:TE) during the neonatal period. A similar mechanism may explain the relatively high flows previously reported in newborns (19) and the negative correlation of max_FRC with postnatal age noted in this study (Table 4).

In the East Boston cohort study (6, 8, 25), max_FRC was reduced in infants of mothers who smoked during pregnancy to a much greater extent than in this study. This may reflect the fact that infants in the current study were delivered before the full impact of prenatal exposure to maternal smoking on somatic and airway growth had occurred, and they had not been exposed to any tobacco smoke postnatally. Furthermore, in the current study, but not in the East Boston study, the adjusted model included body length.

We found that TPTEF:TE was significantly lower in preterm infants whose mothers had smoked, a reduction of similar magnitude to that reported in full-term newborns (5, 9). The ratio was also reduced in boys, as has been previously reported (20), in white infants (13), and in those whose fathers had a manual occupation. After allowing for these variables, TPTEF:TE remained significantly diminished in infants of mothers who smoked, suggesting an underlying alteration in respiratory mechanics or control of breathing that had occurred several weeks before the infants were due to be delivered.

TPTEF:TE reflects the degree to which expiration is actively modulated by the upper airways and diaphragm, and it was first noted to be diminished in adults with chronic obstructive airways disease (15). Martinez and colleagues (20, 26) reported that a reduced ratio among infant boys preceded and predicted an increased risk of wheezing during the first 3 yr of life. In the presence of severe airflow limitation, the tidal breathing curve is defined by intrathoracic airway mechanics. However in healthy subjects, it represents an integrated response of the entire respiratory system, includes neural, laryngeal, and intrathoracic components, and is strongly influenced by respiratory rate and postnatal age (12, 27).

The potential effects of smoking during pregnancy on lung and airway development may include structural alterations (28) as well as interference with the control of respiration (29, 30) and the developing immune system (4). A marked reduction in fetal movements, indicative of exposure to hypoxia, has been reported for at least an hour after the mother has smoked (31). Prenatal or postnatal exposure to nicotine can induce changes in central nervous system sensitivity to blood gases, resulting in impaired control of breathing (29, 30). In addition, nicotine may act as a vasoconstrictor, resulting in reduced placental blood flow, reduced supply of nutrients and oxygen to the fetus, and growth retardation. In the rat model, maternal smoking during pregnancy causes not only fetal growth retardation but marked structural changes to the developing lung, even when identical calorific intake is taken by control and experimental animals (28). Thus, the observed reduction in TPTEF:TE among infants exposed prenatally to tobacco may reflect altered maturation of the lung and control of breathing as much as structural changes in airway development (27, 32).

In conclusion, the findings of this study suggest that, in infants of mothers who smoke during pregnancy, changes in respiratory function are evident by at least 7 wk prior to the expected date of delivery. At the time of testing, none of these infants had been exposed to tobacco smoke postnatally. However, many had been heavily exposed in utero as reflected by infant urinary cotinine levels at birth of a magnitude comparable to that reported in active smokers. Thus, prenatal exposure to tobacco has adverse effects that are evident some 2 mo before most babies are born. Because diminished respiratory function in very early life has been shown to precede and predict wheezing lower respiratory illnesses in early childhood (20), it is likely that these infants are also at increased risk of such illnesses. This would suggest that the incidence and severity of respiratory illnesses in preterm infants might be diminished by programs to reduce the prevalence of maternal smoking in pregnancy. These findings highlight the need to identify effective strategies to help women stop smoking.