INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION FUTURE DIRECTIONS REFERENCES

Keywords: forced expiratory flow; infant; reference values; respiratory function tests; sex

Despite the introduction of the raised volume technique for assessing full forced expiratory maneuvers in infants (1-5), the measurement of maximal flow at functional residual capacity (max_FRC) by the tidal rapid thoracoabdominal compression technique remains the most popular method of assessing small airway function in infants and young children (4, 6-8). One of the advantages of this technique is that it has been used in a large number of physiologic, clinical, and epidemiological investigations to assess the early determinants of airway function (9-14); investigate airway responsiveness in infants (15- 20); examine diseases such as cystic fibrosis (18, 21-23), bronchiolitis (24, 25), and bronchopulmonary dysplasia (26); and evaluate the impact of treatment strategies (22, 27-30) or antecedent respiratory illness on subsequent airway function (11, 31-34).

Nonetheless, interpretation of max_FRC measurements in both clinical practice and research has been limited by a lack of appropriate reference data that are applicable outside the centers that developed them. Indeed, despite considerable advances in standardizing equipment and techniques for assessing lung function in infants (35, 36), reliable normative reference data remain limited. This can be ascribed to the time-consuming nature of these tests and the need for sedation in all but the youngest infants. Although there have been several attempts to report normative data for max_FRC (14, 37-40), the relatively small size of most of these studies makes characterization of even the most basic influences on airway growth, such as age, sex, and size, difficult to describe reliably (41, 42). Difficulties in generalizing such results are compounded when they are used elsewhere to interpret results from populations with different background characteristics and/or where measurements have been obtained by using different instruments or protocols. A further problem is the use of "percent predicted" to express results, as this does not take into account the natural variability of max_FRC between subjects of similar age or body size (43).

The current investigation attempts to address this deficiency of normative reference data by combining max_FRC results from four studies performed in three centers on two continents. It includes data from four published studies (9, 14, 38, 44).

The aim of this study was to combine existing data for max_FRC in healthy infants measured during the first 2 years of life, to develop more robust and generalizable reference data and prediction equations that could be expressed as z scores after taking sex, age, and body size into account.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION FUTURE DIRECTIONS REFERENCES

max_FRC and background details including sex, age, weight, and length at test from all individual infants were collated from the three participating centers (London, Indianapolis, and Boston) (9, 10, 14, 38, 44). These centers were known to have used similar equipment and methodology and to have applied similar criteria regarding technical acceptability of results. The London center (Institute of Child Health) provided data from two groups of infants, one recruited as part of a study investigating the effects of maternal smoking (9) and ethnicity (10) on lung function in preterm infants (hereafter referred to as London preterm infants), and the other recruited to a study of the influence of low birth weight for gestational age on airway function in full-term infants (London term infants) (44). As summarized in Table 1, ethnic composition varied by center. Of the 92 London preterm infants, 48 were white (46% boys) and 38 were black (53% boys), whereas all the London term infants were white, of whom 46% were boys. Of the 153 infants in the Boston study, 62 were white (45% boys) and 83 were Hispanic (43% boys). Only limited individual data were stored long term from the Indianapolis study, but details in the original publication and communication with the authors confirmed that ~ 85% of these infants were white and 56% were boys (14).

Cross-sectional max_FRC data were obtained from the London preterm infants at 3 weeks corrected postnatal age (i.e., age after the expected rather than actual date of delivery, where expected dates were based on ultrasonic assessments before 20 weeks of gestation), whereas the initial measurements of the London term infants were performed between 1 and 3 months corrected postnatal age, with measurements being repeated in a subset of the sample at 710 months of age. Of the 110 London term infants, repeat measurements were available on two occasions for 28 infants and on three occasions for 3 infants. Both American studies were based primarily on term infants. Data from Indianapolis were from a cross-sectional study aiming to extend the age range of normative data for max_FRC and lung volume in infants (14), whereas data from Boston were obtained as part of a longitudinal investigation of the impact of prenatal cigarette smoke exposure on lung function in infancy (13, 31). Additional max_FRC values from 50 Boston infants and six Indianapolis infants that had not been included in previous publications were available for the current collated data set. In the Boston study, lung function tests were performed during four discrete age intervals after birth, namely 26 weeks, and then 46, 1012, and 1518 months of age. Seventy-eight of the 153 Boston infants were studied twice, 46 were studied three times, and 28 were studied on four occasions; the remaining infants were studied only once.

Details of study design and methods of measuring max_FRC within each center have been reported previously (9, 10, 14, 38, 44) and are summarized in Table 2. The mean max_FRC, derived from the three technically satisfactory maneuvers with the highest forced expiratory flows at FRC, was used to calculate the prediction equations. Results from these three maneuvers were generally within 10% of each other (for details see online data supplement).

Description of the Representative Data Set

max_FRC values for 514 infants measured on 719 occasions were compiled from the four studies. To ensure that the final data set was representative of normal infants and that sufficient infants were included from multiple centers across the entire age/length range studied, two major exclusions were necessary. First, only 14 infants had been measured beyond 20 months of age, and all of these infants had been studied in Indianapolis. Results from these tests were excluded. Second, the London term study had been designed to compare airway function in term babies of appropriate birth weight with those born small for gestational age. As such, babies born small for gestational age were by design over-represented in the original cohort. To create a sample that would be representative of the normal population, that is, one in which only 10% of infants would have birth weights less than or equal to the 10th centile for gestational age, all but 13 of the original 54 small for gestational age London term infants (four of whom had been tested on two occasions) were randomly excluded.

Background characteristics of the remaining 459 infants (654 test occasions) who formed the representative population are summarized in Table 1, with infant characteristics at the time of first test summarized in Table 3. Weight and crown-heel length at time of test were converted to z scores, correcting for prematurity, by use of age after the expected rather than actual delivery date, where appropriate (45).

Statistical Analysis

The prediction equations were developed using all the available data, including repeated measures for many infants. The smallest interval between repeat measurements in any one child was 8 weeks, with a mean (SD) of 23 (8) weeks between repeat measurements. Multiple regression analysis was used to develop prediction equations, adjusting for the repeated measures by using generalized estimating equations assuming an exchangeable correlation matrix. The two sexes were analyzed separately as equations combining them required significant sex interactions with age or length (p < 0.01). To adjust for heteroscedasticity, the outcome measure of max_FRC was transformed to its square root. This transformation was identified graphically to minimize both skewness and heteroscedasticity. The square root transformation achieved this for both sexes. (max_FRC)^0.5 was adjusted for body size in two ways: using length (Model 1; Table 4), and using age plus length z score (Model 2; Table 4). Developing both equations allows for situations in which the child's size is thought to be more relevant than their age. Fractional polynomials of age and length were used to adjust for curvature (46), and dummy variables for center, ethnicity, and/or prematurity effects were added as required. Residuals were checked for normality and homoscedasticity.

Individual values of max_FRC were converted to z scores (a z score being the number of standard deviations by which a value deviates from the mean), adjusting for age and/or length (and other variables) using the formula z score = ([max_FRC]^0.5  LP)/RSD, where max_FRC is the infant's measurement (i.e., mean max_FRC), LP is the linear predictor for the child's sex, age, and length (and other variables), and RSD is the residual standard deviation of (max_FRC)^0.5. Analyses were done with Stata 6.0 (Stata, College Station, TX) and Data Desk 6.1.1 (Data Description, Ithaca, NY).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION FUTURE DIRECTIONS REFERENCES

Scatter plots with infants identified by test center demonstrated a positive relationship between mean max_FRC and both age (Figure 1) and length (Figure 2). A scatter plot distinguishing infants by sex (data not shown) suggested that the relationship between max_FRC and length differed by sex, with flows being lower in boys than girls during early infancy but subsequently increasing more rapidly. This was confirmed by the presence of significant interactions of sex with age or length (p < 0.01) in both univariate and multivariate models. For reference equations relating (max_FRC)^0.5 to length, the optimal powers were length for girls and length² for boys. After adjusting for length, (max_FRC)^0.5 variance was not further explained by adding age to the model. For the prediction equations using age and length z score, age itself was optimal for boys whereas the square root of age was optimal for girls, with 10 weeks added to age to deal with the negative corrected postnatal ages in those tested before 40 weeks postconceptional age. Length z score was added to both age models even though it was not significant for boys.

View larger version (31K): [in this window] [in a new window]   Figure 1.   max FRC plotted against postnatal age at time of test. London preterm infants (asterisks); London term infants (open triangles); Indianapolis (open circles); Boston (pluses). Note: postnatal age has been corrected for prematurity in all but the Indianapolis group.

View larger version (33K): [in this window] [in a new window]   Figure 2.   max FRC plotted against crown-heel length at time of test. Symbols are as for Figure 1.

Table 4 gives details of the four regression equations. The regressions and generalized estimating equation analyses gave similar results and those based on an exchangeable correlation matrix, which took into account any association between repeated measures, are presented. With the exception of length z score for boys, all the coefficients were highly significant. Each equation accounted for about half the variation in max_FRC, with the age model slightly better for boys and the length model slightly better for girls. The equations for the two sexes were quite different, with more curvature in boys for both age and length as shown by the differing transformations. These differences are summarized in Tables 5 and 6, which show predicted values for boys and girls at specified lengths and ages. It can be seen that mean max_FRC remains higher in girls to at least 75 cm when based on length, and 12 months when predicted on age. Figure 3 shows max_FRC predicted by length for 2, 0, and +2 z scores in boys and girls. These correspond roughly to the 2nd, 50th, and 98th centiles. The skewness in the centiles is apparent, with the curves closer together below the median than above it. Figure 4 shows the effect of length when predicted values of max_FRC are based on age in girls. The three sets of curves represent predictions by age of 2, 0, and +2 z scores for max_FRC, and within each cluster the three curves are for 2, 0, and +2 length z scores, to take into account girls who are long or short for their age. The effect of length on age-predicted values of max_FRC is far smaller in boys as reflected by the nonsignificant length coefficient (Model 2; Table 4).

View larger version (18K): [in this window] [in a new window]   Figure 3.   Predicted max FRC z scores (0 ± 2) for boys (dashed lines) and girls (solid lines) plotted against length.

View larger version (27K): [in this window] [in a new window]   Figure 4.   The effect of length z score on predicted max FRC based on age in girls.

Within each center, the mean z score of max_FRC was close to 0 (1) when using either the length- or age-based equations, suggesting that there were no important differences either between preterm and term infants or between centers (Table E1; see online data supplement). The effect of prematurity was further investigated by fitting a dummy variable just for the London preterm group. In no case was the effect significant (p > 0.15), showing that after adjustment for age and length, max_FRC in the preterm infants was close to prediction. Omitting the preterm data entirely from the model led to the same conclusion. Ethnicity effects were tested for by using dummy variables for black subjects and Hispanic subjects, with white subjects as the baseline. Neither variable was significant in boys but black girls had significantly greater max_FRC than white girls (+0.65 z score adjusted for height, p = 0.02; +0.60 z score adjusted for age, p = 0.02).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION FUTURE DIRECTIONS REFERENCES

In this investigation, we have presented sex-specific prediction equations for max_FRC during the first 20 months of life. By combining data from three centers, these prediction equations are based on a larger sample than reported previously and are relatively evenly distributed throughout the first 20 months of life. They are thus likely to provide more robust estimates of normal peripheral airway function during infancy than have been available in the past.

Sex Differences in Airway Function

These results strongly support previous evidence that length-adjusted max_FRC is significantly higher in girls than boys during the first year of life. Airway structure has been shown to differ in male and female infants, and it has been suggested that the greater amount of smooth muscle and thicker inner airway wall in boys may provide part of the explanation for sex differences in airway function and susceptibility to respiratory disease in early life (47). The observed sex differences in flow may also be attributed to differences in airway tone (48) and/or lung mechanical properties (49) between boys and girls.

The magnitude of this sex difference was greatest during the first 9 months of life (approximately 20% in infants 50 to 70 cm in length), and decreased with advancing age (Table 6). Although several studies have reported that boys have lower max_FRC values than girls (9, 37, 38, 40), others have observed only a trend (14) or no significant difference. These discrepancies may well be related to relatively small sample size in some of the studies and/or the age distribution of subjects. The current data further demonstrate that the rate of increase in max_FRC, which presumably reflects airway growth, proceeds more slowly in boys than in girls from birth to 6-9 months of age, then accelerates faster in boys than in girls, so that when predicted on the basis of age, flows are similar by 15 months and 10% greater in boys by 18 months (Table 6). However, when based on length, max_FRC remained higher in girls until at least 75 cm (Figure 3), beyond which length relatively small numbers and wide confidence intervals limit interpretation of potential sex differences. More recently, when using the raised volume rapid thoracoabdominal compression technique, Jones and coworkers (5) reported significantly lower FEF_75% values (forced expiratory flow rate when 75% of the vital capacity has been expired) for boys than for girls up to 100 cm in length and 34 months of age. Pulmonary function results from 8- to 15-year-old children have suggested different growth patterns of airways and parenchyma during childhood and adolescence (50). In boys, lung growth appears to be proportional to growth of the small airways whereas in girls, the small airways grow faster than lung volume such that, in contrast to boys, FEF_50% and FEF_75% corrected for total lung capacity (TLC) are positively correlated to body height in girls (50). Similarly, values of lung size-corrected max_FRC have been found to be significantly higher for 4- to 6-year-old girls than for boys (51). These observed differences in small airway and lung growth between the sexes may contribute to the higher prevalence and severity of lower respiratory tract illness in boys during infancy and childhood. Sex differences in lung size-corrected airway function appear to persist into adulthood, with women having higher values of FEV₁/FVC than men (52). Regrettably, it was not feasible to examine sex-specific lung size-corrected forced expiratory flows in this current analysis, because lung volumes were not routinely measured in all the studies. In addition, it is extremely difficult to measure resting lung volume simultaneously with forced expired flow in infants. The variability of the end-expiratory level in infants and the differences in measured values of lung volume according to the technique used (53) further complicate expression and interpretation of "volume-corrected flows" during infancy. Nevertheless, the current analysis provides clear evidence of sex differences in max_FRC that must be accounted for when interpreting results.

It is important to stress that prediction equations in this study represent the cross-sectional, not longitudinal, relationship between max_FRC and other variables (40, 54). This may be problematic for clinicians who wish to monitor change over time in individual patients as, to date, longitudinal data are lacking and will be difficult to obtain. Models based on longitudinal data are unlikely to change the predicted mean and z scores, but may help quantify the extent of centile crossing observed in healthy infants and hence serve to identify individuals with a significant decline or improvement in airway function.

In this study, modeling was performed using the mean of the three highest technically acceptable max_FRC values. In contrast to the rationale for reporting the "highest" values of forced flows and volumes in conventional spirometry or when using the raised volume technique, we recommend reporting the mean of the best three max_FRC values, because the highest may simply reflect a temporary increase in the extent to which end-expiratory level is being dynamically elevated. The relationship between mean and best max_FRC was investigated on the basis of the London data, for which both were available, and found to be as follows: best max_FRC = 1.07 × mean max_FRC (r² = 0.99). (Additional details are available in the online data supplement.)

Choice of Predictor

In this study, body size and age were found to be equally powerful predictors of peripheral airway function, accounting for about half the variance across the age range studied. In physiological terms this makes sense as lung growth is related to growth in length, which is in turn related to age. Length and age are highly correlated but age has the potential advantage over length of being less prone to measurement error and less dependent on socioeconomic status. Nevertheless, if age is to be used as the primary predictor of max_FRC, it is essential that gestational age be recorded accurately and any prematurity corrected for by adjusting estimates of postnatal age for expected rather than actual date of delivery, especially in the first few months of life.

In clinical settings, reference data may be used to interpret individual measurements in infants with lung disease such as cystic fibrosis or congenital heart disease. As such infants may be short and light for their age, use of reference data modeled primarily on age, which assumes normal growth patterns, may mean that the predicted values far exceed those that could be reasonably expected for such a child. In these circumstances, the clinician may wish to report values predicted from length, hence the reason for reporting both in this article.

Issues and Limitations in the Presentation of Normative Reference Values

These collated data represent the largest multicenter study of max_FRC in healthy infants. Nonetheless, several issues could impact on the generalizability of these findings and therefore demand that these data continue to be interpreted and applied with caution. Moreover, the difficulties in recruiting and completing these tests with healthy infants makes the number of observations even in this "large" data set far smaller than those serving as reference standards for lung function in adults.

Center Differences

Although there was no evidence of any significant "center" influence on the results presented (Table E1; see online data supplement), differences in population, equipment and methods, and recruitment procedure will inevitably occur in any multicenter study. Thus, whereas the London and Indianapolis studies were conducted in well-equipped infant pulmonary function laboratories within university tertiary care pediatric hospitals, the Boston infant laboratory was specifically set up in 1986 in an urban community outpatient health center for the purpose of the study (38). Similarly, although development of commercially available equipment that meets American Thoracic Society/European Respiratory Society specifications (36) means that identical equipment will be available for future multicenter studies, this was not the case previously. Nevertheless, data collated for this study were collected from three centers using similar methods, equipment, and criteria for quality control as ascertained by interlaboratory visits. Reassuringly, despite some inevitable differences both in populations recruited and methods used, comparison of z scores between the different centers revealed no important differences (Table E1; see online data supplement).

Participation/Selection Bias

A potential source of bias that could influence the generalizability of these data would be that parents of infants who participated in this study had some real or perceived concern about the respiratory health of their infant. Recruitment prenatally in the Boston study and shortly after birth in the London and Indianapolis studies diminished the likelihood that this was an important or likely influence. Similarly, although concern over existing familial respiratory conditions could have influenced parental decisions, reports from studies that have compared the characteristics of parents who do or do not give consent for testing suggest that this may not be a major factor determining parental participation (55).

Normal Infants and Lung Function Reference Data: Inclusion Criteria

The purpose of this investigation was to provide normative sex-specific reference data for max_FRC during the first 20 months of life. One important issue is to define what constitutes a "normal" infant. For infants with known chronic pulmonary conditions such as cystic fibrosis or bronchopulmonary dysplasia, exclusion justification is readily apparent. However, in the case of infants with chronic conditions that are not known to impact on pulmonary function, these decisions are not so straightforward. Similarly, whereas infants born to mothers who smoke during pregnancy are generally "normal," several studies have suggested that their pulmonary function may be adversely influenced by this exposure (9, 13, 31, 55, 56). In the current investigation, ~ 30% of the "healthy" infants were born to mothers who smoked during pregnancy, which is consistent with that reported for antenatal populations (57-59). The influence of smoking on airway function is the subject of a separate report (60).

Prior Respiratory Illness

Diminished airway function has been reported to precede and follow lower respiratory tract illness (LRI) in infancy (11, 40, 61). Infants with LRI before the first test were excluded from all studies, but this is unlikely to have produced any population bias, especially as infants included in this investigation were not selected according to recognized risk factors for LRI such as family history of atopy or maternal smoking (40, 55). Whether infants who had been affected by an LRI should be excluded from reference pulmonary function data once completely recovered remains debatable. Guiding our decisions with respect to this issue was an attempt to include common conditions that are widely represented in the general population, so that the reference standards developed would represent part of the broad spectrum of "normal." Thus, whereas infants with known chronic cardiopulmonary conditions were excluded, neither exposure to environmental tobacco smoke (an experience shared by up to 50% of children) (62, 63) nor an LRI after the first test occasion (up to 30% of infants wheeze in the first year of life) (55, 61) was a reason for exclusion from this study. All infants were, however, required to be free of any respiratory symptoms for at least 1 week (3 weeks in London and Indianapolis) before testing was undertaken. Use of the less stringent criteria in Boston did not appear to introduce any systematic bias, but needs further validation before it could be recommended, as discussed in the online data supplement.

Prematurity

The decision to include data from otherwise "healthy" preterm infants in the current investigation was based on examination of (1) scatter plots of the preterm max_FRC data, which showed a similar relationship to both age and length as in normal term and older infants, and (2) prediction equations formulated without the preterm infant data, which demonstrated no substantial differences from those with these infants included (Table E1; see online data supplement). We therefore consider these reference standards appropriate for use in preterm infants who fall within the gestational and postconceptional age range of the population reported here (i.e., 2936 weeks gestational age, and 3241 weeks postconceptional age) provided that postnatal age is adjusted for expected rather than actual date of birth. Whether this is also true for preterm infants studied during later infancy has yet to be ascertained, because premature delivery could have an adverse effect on subsequent airway growth (64, 65).

Ethnic Group

Potential ethnic influences on the prediction equations were evaluated by repeat analyses with separate terms for Hispanic and black infants. These two terms failed to show differences by ethnic group for male infants, whereas higher flows (~ +0.6 z score) were observed for black versus white infant girls when adjusted for age and height. This result should be viewed as preliminary, as the majority of black girls were studied during the first few weeks of life (Table 3), when ethnic differences in lung maturation and breathing pattern may influence the extent to which lung volume is dynamically elevated (10). Ethnic representation in the current investigation was inadequate to allow a comprehensive analysis of the possible impact of ethnic differences on infant airway growth, and it is therefore inappropriate to publish separate regression equations at this time. This underscores the need to continue recruiting and studying airway function in normal infants, and to document not only sex, age, and body size but any other potentially important influences such as ethnicity, history of respiratory illness, environmental exposures, and family history of lung disease.

Comparison with Previously Published Data

Although there have been several attempts to report normative data for max_FRC (14, 38-40), the use of currently available prediction equations could lead to different interpretation of results, as summarized in Tables 7 and 8.

As can be seen, whether based on length or age, much lower predicted values were published from early applications of this technique in Tucson (37) (on which were also subsequently based the collated data of Hampton and coworkers [39]). This has been attributed at least partially to failure to achieve flow limitation in every infant during some of the pioneering work. The reason for similarly low values in data published from Perth, Australia (40) is less easily explained, and emphasizes the difficulties in using reference data developed in different laboratories. Thus, whereas Young and coworkers (40) also reported significantly lower max_FRC in boys than in girls, flows in both sexes were 1525% lower than those reported in either the independent Indianapolis and Boston studies (14, 38) or the currently collated populations. As mentioned above, there were no important center differences observed in this study and any discrepancies in predicted values between the current and "original" equations published by Tepper and Reister (14) and Hanrahan and coworkers (38) primarily reflect the different age distribution of infants, which is much more evenly balanced in the collated data set. Furthermore, neither Hanrahan and coworkers nor Tepper and Reister presented sex-specific equations. This could preclude detection of clinically significant changes in girls, and lead to false reports of impaired airway function in boys. The current analysis has two further advantages: First, the data have been presented as z scores, which provide a true indication of the natural variability among healthy infants. As can be seen from Figure 3 and Tables 5 and 6, the wide range of max_FRC values during early life among "normal" infants limits the extent to which any young infant can be labeled as "abnormal" with respect to such flows. Appropriate use of z scores, which indicate a "normal" range (0 ± 2 z scores) for max_FRC during infancy should improve interpretation of both clinical and research studies. Second, by extending the range of age and length to these prediction equations by including preterm infants, more reliable estimates of the immediate effects of neonatal lung disease should be possible. Extrapolation of any of the previously available equations (which should be strongly discouraged) would have resulted in marked underestimation of expected flows at time of discharge from the neonatal unit (Table 7).

	FUTURE DIRECTIONS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION FUTURE DIRECTIONS REFERENCES

It is possible that other techniques for measuring forced expiration in infants will supplant max_FRC as a measure of airway function in the future. Indeed, several centers are now using raised volume forced expiratory maneuvers. This technique has the theoretic appeal of sampling flows across the entire volume range and is thus more akin to a fuller voluntary forced expiratory maneuver in adults, a well-established standard for decades. Nonetheless, normative standards for any new technique will need to be developed before it can replace max_FRC, as will a body of clinical and epidemiological data to match that currently available for max_FRC, and which has made it such a useful measure of the influence of environmental exposures, subsequent respiratory symptoms, or illness.

Conclusion

In conclusion, we have presented data on max_FRC from a combined cohort of 459 infants and 654 test sessions, and have established sex-specific prediction equations for this measure in the first 20 months of life. We conclude that normal infant girls have higher flows up to at least 12 months of age and 75 cm in length. The observed sex differences in flows during infancy may reflect differences in airway structure and tone, and rate of growth of the peripheral airways and lung parenchyma between boys and girls. These differences appear to persist in childhood and adolescence.

This study underscores the need for all infant lung function centers to embrace consensus standard methodologies and for multicenter data sharing. In addition, it emphasizes the importance of thorough, standardized characterization of demographic features, environmental exposures, illness, and family history of all infants undergoing tests. Only by careful scrutiny of all factors that could impact on airway growth, and a willingness on the part of investigators to share and combine data to enable studies with adequate power to address these issues, can a clearer understanding of the influences on infant lung function emerge.